{
  "best_global_step": 6784,
  "best_metric": 0.5017365217208862,
  "best_model_checkpoint": "saves_multiple/prefix-tuning/llama-3-8b-instruct/train_codealpacapy_123_1762496345/checkpoint-6784",
  "epoch": 20.0,
  "eval_steps": 3392,
  "global_step": 33920,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00294811320754717,
      "grad_norm": 64.7282485961914,
      "learning_rate": 1.179245283018868e-08,
      "loss": 6.5898,
      "num_input_tokens_seen": 2880,
      "step": 5
    },
    {
      "epoch": 0.00589622641509434,
      "grad_norm": 49.12639236450195,
      "learning_rate": 2.6533018867924528e-08,
      "loss": 6.6231,
      "num_input_tokens_seen": 5728,
      "step": 10
    },
    {
      "epoch": 0.00884433962264151,
      "grad_norm": 51.32429885864258,
      "learning_rate": 4.127358490566038e-08,
      "loss": 6.5842,
      "num_input_tokens_seen": 8736,
      "step": 15
    },
    {
      "epoch": 0.01179245283018868,
      "grad_norm": 60.82346725463867,
      "learning_rate": 5.601415094339623e-08,
      "loss": 6.5648,
      "num_input_tokens_seen": 11648,
      "step": 20
    },
    {
      "epoch": 0.01474056603773585,
      "grad_norm": 37.56840133666992,
      "learning_rate": 7.075471698113208e-08,
      "loss": 6.4989,
      "num_input_tokens_seen": 16224,
      "step": 25
    },
    {
      "epoch": 0.01768867924528302,
      "grad_norm": 61.440269470214844,
      "learning_rate": 8.549528301886793e-08,
      "loss": 6.7535,
      "num_input_tokens_seen": 19136,
      "step": 30
    },
    {
      "epoch": 0.020636792452830188,
      "grad_norm": 58.689029693603516,
      "learning_rate": 1.0023584905660378e-07,
      "loss": 6.3632,
      "num_input_tokens_seen": 22208,
      "step": 35
    },
    {
      "epoch": 0.02358490566037736,
      "grad_norm": 68.44844055175781,
      "learning_rate": 1.1497641509433962e-07,
      "loss": 6.7811,
      "num_input_tokens_seen": 24576,
      "step": 40
    },
    {
      "epoch": 0.02653301886792453,
      "grad_norm": 55.26619338989258,
      "learning_rate": 1.297169811320755e-07,
      "loss": 6.83,
      "num_input_tokens_seen": 28800,
      "step": 45
    },
    {
      "epoch": 0.0294811320754717,
      "grad_norm": 40.49510955810547,
      "learning_rate": 1.4445754716981135e-07,
      "loss": 6.4312,
      "num_input_tokens_seen": 31776,
      "step": 50
    },
    {
      "epoch": 0.03242924528301887,
      "grad_norm": 41.20639419555664,
      "learning_rate": 1.591981132075472e-07,
      "loss": 6.1927,
      "num_input_tokens_seen": 35648,
      "step": 55
    },
    {
      "epoch": 0.03537735849056604,
      "grad_norm": 59.88456726074219,
      "learning_rate": 1.7393867924528304e-07,
      "loss": 6.5442,
      "num_input_tokens_seen": 38624,
      "step": 60
    },
    {
      "epoch": 0.038325471698113206,
      "grad_norm": 55.380611419677734,
      "learning_rate": 1.886792452830189e-07,
      "loss": 6.2018,
      "num_input_tokens_seen": 41728,
      "step": 65
    },
    {
      "epoch": 0.041273584905660375,
      "grad_norm": 42.346500396728516,
      "learning_rate": 2.0341981132075473e-07,
      "loss": 6.0513,
      "num_input_tokens_seen": 44672,
      "step": 70
    },
    {
      "epoch": 0.044221698113207544,
      "grad_norm": 40.66829299926758,
      "learning_rate": 2.1816037735849058e-07,
      "loss": 6.3024,
      "num_input_tokens_seen": 48160,
      "step": 75
    },
    {
      "epoch": 0.04716981132075472,
      "grad_norm": 50.525733947753906,
      "learning_rate": 2.3290094339622643e-07,
      "loss": 6.0573,
      "num_input_tokens_seen": 51776,
      "step": 80
    },
    {
      "epoch": 0.05011792452830189,
      "grad_norm": 52.028682708740234,
      "learning_rate": 2.476415094339623e-07,
      "loss": 6.4386,
      "num_input_tokens_seen": 54528,
      "step": 85
    },
    {
      "epoch": 0.05306603773584906,
      "grad_norm": 42.27737045288086,
      "learning_rate": 2.6238207547169815e-07,
      "loss": 6.3051,
      "num_input_tokens_seen": 57568,
      "step": 90
    },
    {
      "epoch": 0.05601415094339623,
      "grad_norm": 52.775020599365234,
      "learning_rate": 2.7712264150943397e-07,
      "loss": 5.7796,
      "num_input_tokens_seen": 60256,
      "step": 95
    },
    {
      "epoch": 0.0589622641509434,
      "grad_norm": 68.52125549316406,
      "learning_rate": 2.9186320754716984e-07,
      "loss": 5.4755,
      "num_input_tokens_seen": 62688,
      "step": 100
    },
    {
      "epoch": 0.061910377358490566,
      "grad_norm": 51.297088623046875,
      "learning_rate": 3.0660377358490567e-07,
      "loss": 5.6559,
      "num_input_tokens_seen": 65344,
      "step": 105
    },
    {
      "epoch": 0.06485849056603774,
      "grad_norm": 34.314109802246094,
      "learning_rate": 3.213443396226416e-07,
      "loss": 5.1255,
      "num_input_tokens_seen": 68544,
      "step": 110
    },
    {
      "epoch": 0.06780660377358491,
      "grad_norm": 34.0240592956543,
      "learning_rate": 3.3608490566037736e-07,
      "loss": 5.5851,
      "num_input_tokens_seen": 71328,
      "step": 115
    },
    {
      "epoch": 0.07075471698113207,
      "grad_norm": 33.414390563964844,
      "learning_rate": 3.508254716981133e-07,
      "loss": 5.1278,
      "num_input_tokens_seen": 74240,
      "step": 120
    },
    {
      "epoch": 0.07370283018867925,
      "grad_norm": 46.9061393737793,
      "learning_rate": 3.6556603773584905e-07,
      "loss": 5.1234,
      "num_input_tokens_seen": 77248,
      "step": 125
    },
    {
      "epoch": 0.07665094339622641,
      "grad_norm": 29.65032386779785,
      "learning_rate": 3.80306603773585e-07,
      "loss": 5.103,
      "num_input_tokens_seen": 80704,
      "step": 130
    },
    {
      "epoch": 0.07959905660377359,
      "grad_norm": 38.87522888183594,
      "learning_rate": 3.9504716981132075e-07,
      "loss": 4.762,
      "num_input_tokens_seen": 83840,
      "step": 135
    },
    {
      "epoch": 0.08254716981132075,
      "grad_norm": 41.98786926269531,
      "learning_rate": 4.097877358490567e-07,
      "loss": 5.2388,
      "num_input_tokens_seen": 86560,
      "step": 140
    },
    {
      "epoch": 0.08549528301886793,
      "grad_norm": 38.62727355957031,
      "learning_rate": 4.2452830188679244e-07,
      "loss": 5.1523,
      "num_input_tokens_seen": 89376,
      "step": 145
    },
    {
      "epoch": 0.08844339622641509,
      "grad_norm": 35.342803955078125,
      "learning_rate": 4.3926886792452837e-07,
      "loss": 4.4728,
      "num_input_tokens_seen": 95904,
      "step": 150
    },
    {
      "epoch": 0.09139150943396226,
      "grad_norm": 32.85224914550781,
      "learning_rate": 4.5400943396226414e-07,
      "loss": 4.9182,
      "num_input_tokens_seen": 99520,
      "step": 155
    },
    {
      "epoch": 0.09433962264150944,
      "grad_norm": 31.7283878326416,
      "learning_rate": 4.6875000000000006e-07,
      "loss": 4.8899,
      "num_input_tokens_seen": 102656,
      "step": 160
    },
    {
      "epoch": 0.0972877358490566,
      "grad_norm": 31.581472396850586,
      "learning_rate": 4.834905660377359e-07,
      "loss": 4.7307,
      "num_input_tokens_seen": 105312,
      "step": 165
    },
    {
      "epoch": 0.10023584905660378,
      "grad_norm": 27.684341430664062,
      "learning_rate": 4.982311320754717e-07,
      "loss": 4.9404,
      "num_input_tokens_seen": 108160,
      "step": 170
    },
    {
      "epoch": 0.10318396226415094,
      "grad_norm": 30.44116973876953,
      "learning_rate": 5.129716981132076e-07,
      "loss": 4.2148,
      "num_input_tokens_seen": 111680,
      "step": 175
    },
    {
      "epoch": 0.10613207547169812,
      "grad_norm": 28.67049789428711,
      "learning_rate": 5.277122641509435e-07,
      "loss": 4.8944,
      "num_input_tokens_seen": 114912,
      "step": 180
    },
    {
      "epoch": 0.10908018867924528,
      "grad_norm": 30.86630630493164,
      "learning_rate": 5.424528301886793e-07,
      "loss": 4.2355,
      "num_input_tokens_seen": 117376,
      "step": 185
    },
    {
      "epoch": 0.11202830188679246,
      "grad_norm": 38.3621940612793,
      "learning_rate": 5.571933962264151e-07,
      "loss": 4.059,
      "num_input_tokens_seen": 120416,
      "step": 190
    },
    {
      "epoch": 0.11497641509433962,
      "grad_norm": 25.196552276611328,
      "learning_rate": 5.71933962264151e-07,
      "loss": 4.4158,
      "num_input_tokens_seen": 123424,
      "step": 195
    },
    {
      "epoch": 0.1179245283018868,
      "grad_norm": 23.164745330810547,
      "learning_rate": 5.866745283018868e-07,
      "loss": 4.3497,
      "num_input_tokens_seen": 127136,
      "step": 200
    },
    {
      "epoch": 0.12087264150943396,
      "grad_norm": 25.149311065673828,
      "learning_rate": 6.014150943396227e-07,
      "loss": 4.2567,
      "num_input_tokens_seen": 131840,
      "step": 205
    },
    {
      "epoch": 0.12382075471698113,
      "grad_norm": 21.28417205810547,
      "learning_rate": 6.161556603773585e-07,
      "loss": 3.9834,
      "num_input_tokens_seen": 135360,
      "step": 210
    },
    {
      "epoch": 0.1267688679245283,
      "grad_norm": 29.37235450744629,
      "learning_rate": 6.308962264150945e-07,
      "loss": 4.1029,
      "num_input_tokens_seen": 139328,
      "step": 215
    },
    {
      "epoch": 0.12971698113207547,
      "grad_norm": 28.774032592773438,
      "learning_rate": 6.456367924528302e-07,
      "loss": 4.156,
      "num_input_tokens_seen": 142976,
      "step": 220
    },
    {
      "epoch": 0.13266509433962265,
      "grad_norm": 23.721956253051758,
      "learning_rate": 6.603773584905661e-07,
      "loss": 3.4806,
      "num_input_tokens_seen": 146528,
      "step": 225
    },
    {
      "epoch": 0.13561320754716982,
      "grad_norm": 19.850208282470703,
      "learning_rate": 6.75117924528302e-07,
      "loss": 3.7368,
      "num_input_tokens_seen": 152192,
      "step": 230
    },
    {
      "epoch": 0.13856132075471697,
      "grad_norm": 34.27486801147461,
      "learning_rate": 6.898584905660379e-07,
      "loss": 4.0799,
      "num_input_tokens_seen": 154624,
      "step": 235
    },
    {
      "epoch": 0.14150943396226415,
      "grad_norm": 19.940319061279297,
      "learning_rate": 7.045990566037736e-07,
      "loss": 4.2881,
      "num_input_tokens_seen": 157920,
      "step": 240
    },
    {
      "epoch": 0.14445754716981132,
      "grad_norm": 25.46551513671875,
      "learning_rate": 7.193396226415095e-07,
      "loss": 3.5618,
      "num_input_tokens_seen": 160640,
      "step": 245
    },
    {
      "epoch": 0.1474056603773585,
      "grad_norm": 27.28227996826172,
      "learning_rate": 7.340801886792454e-07,
      "loss": 3.6146,
      "num_input_tokens_seen": 164448,
      "step": 250
    },
    {
      "epoch": 0.15035377358490565,
      "grad_norm": 33.54536056518555,
      "learning_rate": 7.488207547169812e-07,
      "loss": 3.4774,
      "num_input_tokens_seen": 168192,
      "step": 255
    },
    {
      "epoch": 0.15330188679245282,
      "grad_norm": 30.10123634338379,
      "learning_rate": 7.63561320754717e-07,
      "loss": 3.5721,
      "num_input_tokens_seen": 172192,
      "step": 260
    },
    {
      "epoch": 0.15625,
      "grad_norm": 24.84919548034668,
      "learning_rate": 7.783018867924529e-07,
      "loss": 3.4567,
      "num_input_tokens_seen": 175008,
      "step": 265
    },
    {
      "epoch": 0.15919811320754718,
      "grad_norm": 24.56399917602539,
      "learning_rate": 7.930424528301888e-07,
      "loss": 3.2578,
      "num_input_tokens_seen": 177888,
      "step": 270
    },
    {
      "epoch": 0.16214622641509435,
      "grad_norm": 22.678579330444336,
      "learning_rate": 8.077830188679246e-07,
      "loss": 3.0722,
      "num_input_tokens_seen": 181472,
      "step": 275
    },
    {
      "epoch": 0.1650943396226415,
      "grad_norm": 22.689022064208984,
      "learning_rate": 8.225235849056605e-07,
      "loss": 3.1394,
      "num_input_tokens_seen": 187296,
      "step": 280
    },
    {
      "epoch": 0.16804245283018868,
      "grad_norm": 26.912845611572266,
      "learning_rate": 8.372641509433963e-07,
      "loss": 3.2233,
      "num_input_tokens_seen": 189600,
      "step": 285
    },
    {
      "epoch": 0.17099056603773585,
      "grad_norm": 19.499067306518555,
      "learning_rate": 8.520047169811321e-07,
      "loss": 3.3117,
      "num_input_tokens_seen": 192256,
      "step": 290
    },
    {
      "epoch": 0.17393867924528303,
      "grad_norm": 18.910808563232422,
      "learning_rate": 8.66745283018868e-07,
      "loss": 3.458,
      "num_input_tokens_seen": 195168,
      "step": 295
    },
    {
      "epoch": 0.17688679245283018,
      "grad_norm": 19.73526954650879,
      "learning_rate": 8.814858490566039e-07,
      "loss": 3.4695,
      "num_input_tokens_seen": 198368,
      "step": 300
    },
    {
      "epoch": 0.17983490566037735,
      "grad_norm": 27.31045913696289,
      "learning_rate": 8.962264150943397e-07,
      "loss": 3.3337,
      "num_input_tokens_seen": 201184,
      "step": 305
    },
    {
      "epoch": 0.18278301886792453,
      "grad_norm": 17.952632904052734,
      "learning_rate": 9.109669811320755e-07,
      "loss": 3.3536,
      "num_input_tokens_seen": 204000,
      "step": 310
    },
    {
      "epoch": 0.1857311320754717,
      "grad_norm": 36.304290771484375,
      "learning_rate": 9.257075471698114e-07,
      "loss": 3.1915,
      "num_input_tokens_seen": 207136,
      "step": 315
    },
    {
      "epoch": 0.18867924528301888,
      "grad_norm": 24.410093307495117,
      "learning_rate": 9.404481132075473e-07,
      "loss": 2.6377,
      "num_input_tokens_seen": 210112,
      "step": 320
    },
    {
      "epoch": 0.19162735849056603,
      "grad_norm": 20.085615158081055,
      "learning_rate": 9.551886792452833e-07,
      "loss": 2.676,
      "num_input_tokens_seen": 212736,
      "step": 325
    },
    {
      "epoch": 0.1945754716981132,
      "grad_norm": 24.513856887817383,
      "learning_rate": 9.699292452830188e-07,
      "loss": 2.634,
      "num_input_tokens_seen": 215584,
      "step": 330
    },
    {
      "epoch": 0.19752358490566038,
      "grad_norm": 17.04359245300293,
      "learning_rate": 9.846698113207548e-07,
      "loss": 3.4525,
      "num_input_tokens_seen": 218912,
      "step": 335
    },
    {
      "epoch": 0.20047169811320756,
      "grad_norm": 16.813438415527344,
      "learning_rate": 9.994103773584906e-07,
      "loss": 2.8842,
      "num_input_tokens_seen": 222432,
      "step": 340
    },
    {
      "epoch": 0.2034198113207547,
      "grad_norm": 18.97490119934082,
      "learning_rate": 1.0141509433962265e-06,
      "loss": 2.9732,
      "num_input_tokens_seen": 226240,
      "step": 345
    },
    {
      "epoch": 0.20636792452830188,
      "grad_norm": 18.00664710998535,
      "learning_rate": 1.0288915094339623e-06,
      "loss": 3.066,
      "num_input_tokens_seen": 229888,
      "step": 350
    },
    {
      "epoch": 0.20931603773584906,
      "grad_norm": 43.31511688232422,
      "learning_rate": 1.043632075471698e-06,
      "loss": 2.8984,
      "num_input_tokens_seen": 234176,
      "step": 355
    },
    {
      "epoch": 0.21226415094339623,
      "grad_norm": 19.27219009399414,
      "learning_rate": 1.058372641509434e-06,
      "loss": 3.0836,
      "num_input_tokens_seen": 237696,
      "step": 360
    },
    {
      "epoch": 0.21521226415094338,
      "grad_norm": 16.996538162231445,
      "learning_rate": 1.07311320754717e-06,
      "loss": 2.8502,
      "num_input_tokens_seen": 241568,
      "step": 365
    },
    {
      "epoch": 0.21816037735849056,
      "grad_norm": 19.156700134277344,
      "learning_rate": 1.0878537735849056e-06,
      "loss": 2.7408,
      "num_input_tokens_seen": 243968,
      "step": 370
    },
    {
      "epoch": 0.22110849056603774,
      "grad_norm": 24.887527465820312,
      "learning_rate": 1.1025943396226416e-06,
      "loss": 2.5705,
      "num_input_tokens_seen": 246656,
      "step": 375
    },
    {
      "epoch": 0.2240566037735849,
      "grad_norm": 22.936687469482422,
      "learning_rate": 1.1173349056603773e-06,
      "loss": 2.7849,
      "num_input_tokens_seen": 250528,
      "step": 380
    },
    {
      "epoch": 0.2270047169811321,
      "grad_norm": 21.138553619384766,
      "learning_rate": 1.1320754716981133e-06,
      "loss": 2.6605,
      "num_input_tokens_seen": 253280,
      "step": 385
    },
    {
      "epoch": 0.22995283018867924,
      "grad_norm": 29.792118072509766,
      "learning_rate": 1.1468160377358493e-06,
      "loss": 2.659,
      "num_input_tokens_seen": 256448,
      "step": 390
    },
    {
      "epoch": 0.2329009433962264,
      "grad_norm": 24.42670249938965,
      "learning_rate": 1.1615566037735849e-06,
      "loss": 2.8954,
      "num_input_tokens_seen": 259392,
      "step": 395
    },
    {
      "epoch": 0.2358490566037736,
      "grad_norm": 14.09592056274414,
      "learning_rate": 1.1762971698113208e-06,
      "loss": 2.6005,
      "num_input_tokens_seen": 262432,
      "step": 400
    },
    {
      "epoch": 0.23879716981132076,
      "grad_norm": 18.65346908569336,
      "learning_rate": 1.1910377358490568e-06,
      "loss": 2.4863,
      "num_input_tokens_seen": 265088,
      "step": 405
    },
    {
      "epoch": 0.2417452830188679,
      "grad_norm": 21.989351272583008,
      "learning_rate": 1.2057783018867926e-06,
      "loss": 2.882,
      "num_input_tokens_seen": 268480,
      "step": 410
    },
    {
      "epoch": 0.2446933962264151,
      "grad_norm": 13.782401084899902,
      "learning_rate": 1.2205188679245284e-06,
      "loss": 2.4369,
      "num_input_tokens_seen": 272448,
      "step": 415
    },
    {
      "epoch": 0.24764150943396226,
      "grad_norm": 15.481195449829102,
      "learning_rate": 1.2352594339622641e-06,
      "loss": 2.3918,
      "num_input_tokens_seen": 275168,
      "step": 420
    },
    {
      "epoch": 0.2505896226415094,
      "grad_norm": 19.427631378173828,
      "learning_rate": 1.25e-06,
      "loss": 2.3237,
      "num_input_tokens_seen": 277888,
      "step": 425
    },
    {
      "epoch": 0.2535377358490566,
      "grad_norm": 27.305288314819336,
      "learning_rate": 1.264740566037736e-06,
      "loss": 2.6138,
      "num_input_tokens_seen": 281280,
      "step": 430
    },
    {
      "epoch": 0.25648584905660377,
      "grad_norm": 31.20403289794922,
      "learning_rate": 1.2794811320754718e-06,
      "loss": 2.642,
      "num_input_tokens_seen": 283904,
      "step": 435
    },
    {
      "epoch": 0.25943396226415094,
      "grad_norm": 25.38358497619629,
      "learning_rate": 1.2942216981132078e-06,
      "loss": 2.586,
      "num_input_tokens_seen": 286528,
      "step": 440
    },
    {
      "epoch": 0.2623820754716981,
      "grad_norm": 18.536182403564453,
      "learning_rate": 1.3089622641509436e-06,
      "loss": 2.555,
      "num_input_tokens_seen": 290048,
      "step": 445
    },
    {
      "epoch": 0.2653301886792453,
      "grad_norm": 17.035594940185547,
      "learning_rate": 1.3237028301886792e-06,
      "loss": 2.4563,
      "num_input_tokens_seen": 294080,
      "step": 450
    },
    {
      "epoch": 0.26827830188679247,
      "grad_norm": 28.207172393798828,
      "learning_rate": 1.3384433962264151e-06,
      "loss": 2.7682,
      "num_input_tokens_seen": 296608,
      "step": 455
    },
    {
      "epoch": 0.27122641509433965,
      "grad_norm": 18.966447830200195,
      "learning_rate": 1.353183962264151e-06,
      "loss": 2.4682,
      "num_input_tokens_seen": 299136,
      "step": 460
    },
    {
      "epoch": 0.27417452830188677,
      "grad_norm": 23.712392807006836,
      "learning_rate": 1.3679245283018869e-06,
      "loss": 2.3755,
      "num_input_tokens_seen": 302240,
      "step": 465
    },
    {
      "epoch": 0.27712264150943394,
      "grad_norm": 20.29737663269043,
      "learning_rate": 1.3826650943396229e-06,
      "loss": 2.2079,
      "num_input_tokens_seen": 305888,
      "step": 470
    },
    {
      "epoch": 0.2800707547169811,
      "grad_norm": 10.936732292175293,
      "learning_rate": 1.3974056603773586e-06,
      "loss": 2.7853,
      "num_input_tokens_seen": 308992,
      "step": 475
    },
    {
      "epoch": 0.2830188679245283,
      "grad_norm": 18.892715454101562,
      "learning_rate": 1.4121462264150946e-06,
      "loss": 2.7675,
      "num_input_tokens_seen": 311360,
      "step": 480
    },
    {
      "epoch": 0.28596698113207547,
      "grad_norm": 15.853782653808594,
      "learning_rate": 1.4268867924528304e-06,
      "loss": 2.2163,
      "num_input_tokens_seen": 313824,
      "step": 485
    },
    {
      "epoch": 0.28891509433962265,
      "grad_norm": 24.81093978881836,
      "learning_rate": 1.4416273584905664e-06,
      "loss": 2.5636,
      "num_input_tokens_seen": 316640,
      "step": 490
    },
    {
      "epoch": 0.2918632075471698,
      "grad_norm": 17.189212799072266,
      "learning_rate": 1.456367924528302e-06,
      "loss": 2.1607,
      "num_input_tokens_seen": 320576,
      "step": 495
    },
    {
      "epoch": 0.294811320754717,
      "grad_norm": 15.070216178894043,
      "learning_rate": 1.4711084905660377e-06,
      "loss": 2.4592,
      "num_input_tokens_seen": 324288,
      "step": 500
    },
    {
      "epoch": 0.2977594339622642,
      "grad_norm": 17.137371063232422,
      "learning_rate": 1.4858490566037737e-06,
      "loss": 2.2579,
      "num_input_tokens_seen": 327072,
      "step": 505
    },
    {
      "epoch": 0.3007075471698113,
      "grad_norm": 19.067598342895508,
      "learning_rate": 1.5005896226415096e-06,
      "loss": 2.2751,
      "num_input_tokens_seen": 329408,
      "step": 510
    },
    {
      "epoch": 0.30365566037735847,
      "grad_norm": 14.832257270812988,
      "learning_rate": 1.5153301886792454e-06,
      "loss": 2.1318,
      "num_input_tokens_seen": 332704,
      "step": 515
    },
    {
      "epoch": 0.30660377358490565,
      "grad_norm": 23.307889938354492,
      "learning_rate": 1.5300707547169814e-06,
      "loss": 2.4837,
      "num_input_tokens_seen": 335552,
      "step": 520
    },
    {
      "epoch": 0.3095518867924528,
      "grad_norm": 17.52256202697754,
      "learning_rate": 1.5448113207547172e-06,
      "loss": 2.3355,
      "num_input_tokens_seen": 338560,
      "step": 525
    },
    {
      "epoch": 0.3125,
      "grad_norm": 23.74030303955078,
      "learning_rate": 1.5595518867924531e-06,
      "loss": 2.1778,
      "num_input_tokens_seen": 342592,
      "step": 530
    },
    {
      "epoch": 0.3154481132075472,
      "grad_norm": 13.900066375732422,
      "learning_rate": 1.574292452830189e-06,
      "loss": 1.9473,
      "num_input_tokens_seen": 346176,
      "step": 535
    },
    {
      "epoch": 0.31839622641509435,
      "grad_norm": 18.985767364501953,
      "learning_rate": 1.5890330188679245e-06,
      "loss": 1.8161,
      "num_input_tokens_seen": 350464,
      "step": 540
    },
    {
      "epoch": 0.32134433962264153,
      "grad_norm": 11.604238510131836,
      "learning_rate": 1.6037735849056604e-06,
      "loss": 2.2977,
      "num_input_tokens_seen": 353856,
      "step": 545
    },
    {
      "epoch": 0.3242924528301887,
      "grad_norm": 16.48134422302246,
      "learning_rate": 1.6185141509433964e-06,
      "loss": 1.9008,
      "num_input_tokens_seen": 356608,
      "step": 550
    },
    {
      "epoch": 0.3272405660377358,
      "grad_norm": 21.324331283569336,
      "learning_rate": 1.6332547169811322e-06,
      "loss": 2.6972,
      "num_input_tokens_seen": 359520,
      "step": 555
    },
    {
      "epoch": 0.330188679245283,
      "grad_norm": 14.078737258911133,
      "learning_rate": 1.6479952830188682e-06,
      "loss": 2.1941,
      "num_input_tokens_seen": 363616,
      "step": 560
    },
    {
      "epoch": 0.3331367924528302,
      "grad_norm": 33.28125762939453,
      "learning_rate": 1.662735849056604e-06,
      "loss": 2.5141,
      "num_input_tokens_seen": 366336,
      "step": 565
    },
    {
      "epoch": 0.33608490566037735,
      "grad_norm": 13.483412742614746,
      "learning_rate": 1.67747641509434e-06,
      "loss": 2.0076,
      "num_input_tokens_seen": 369632,
      "step": 570
    },
    {
      "epoch": 0.33903301886792453,
      "grad_norm": 13.411723136901855,
      "learning_rate": 1.6922169811320757e-06,
      "loss": 1.975,
      "num_input_tokens_seen": 373280,
      "step": 575
    },
    {
      "epoch": 0.3419811320754717,
      "grad_norm": 22.47404670715332,
      "learning_rate": 1.7069575471698112e-06,
      "loss": 2.4221,
      "num_input_tokens_seen": 378176,
      "step": 580
    },
    {
      "epoch": 0.3449292452830189,
      "grad_norm": 30.219545364379883,
      "learning_rate": 1.7216981132075472e-06,
      "loss": 1.9371,
      "num_input_tokens_seen": 380992,
      "step": 585
    },
    {
      "epoch": 0.34787735849056606,
      "grad_norm": 20.832317352294922,
      "learning_rate": 1.736438679245283e-06,
      "loss": 2.1914,
      "num_input_tokens_seen": 383840,
      "step": 590
    },
    {
      "epoch": 0.35082547169811323,
      "grad_norm": 12.889510154724121,
      "learning_rate": 1.751179245283019e-06,
      "loss": 1.7578,
      "num_input_tokens_seen": 386720,
      "step": 595
    },
    {
      "epoch": 0.35377358490566035,
      "grad_norm": 26.015750885009766,
      "learning_rate": 1.765919811320755e-06,
      "loss": 2.049,
      "num_input_tokens_seen": 389664,
      "step": 600
    },
    {
      "epoch": 0.35672169811320753,
      "grad_norm": 11.565692901611328,
      "learning_rate": 1.7806603773584907e-06,
      "loss": 2.5405,
      "num_input_tokens_seen": 393024,
      "step": 605
    },
    {
      "epoch": 0.3596698113207547,
      "grad_norm": 22.778697967529297,
      "learning_rate": 1.7954009433962267e-06,
      "loss": 1.9969,
      "num_input_tokens_seen": 396576,
      "step": 610
    },
    {
      "epoch": 0.3626179245283019,
      "grad_norm": 16.982954025268555,
      "learning_rate": 1.8101415094339625e-06,
      "loss": 2.3094,
      "num_input_tokens_seen": 399488,
      "step": 615
    },
    {
      "epoch": 0.36556603773584906,
      "grad_norm": 23.14407730102539,
      "learning_rate": 1.8248820754716984e-06,
      "loss": 1.9377,
      "num_input_tokens_seen": 402496,
      "step": 620
    },
    {
      "epoch": 0.36851415094339623,
      "grad_norm": 22.094654083251953,
      "learning_rate": 1.839622641509434e-06,
      "loss": 1.7228,
      "num_input_tokens_seen": 405568,
      "step": 625
    },
    {
      "epoch": 0.3714622641509434,
      "grad_norm": 27.078800201416016,
      "learning_rate": 1.8543632075471698e-06,
      "loss": 1.8129,
      "num_input_tokens_seen": 408416,
      "step": 630
    },
    {
      "epoch": 0.3744103773584906,
      "grad_norm": 26.680383682250977,
      "learning_rate": 1.8691037735849057e-06,
      "loss": 1.7833,
      "num_input_tokens_seen": 411360,
      "step": 635
    },
    {
      "epoch": 0.37735849056603776,
      "grad_norm": 13.581195831298828,
      "learning_rate": 1.8838443396226417e-06,
      "loss": 1.6358,
      "num_input_tokens_seen": 414112,
      "step": 640
    },
    {
      "epoch": 0.3803066037735849,
      "grad_norm": 23.067106246948242,
      "learning_rate": 1.8985849056603775e-06,
      "loss": 1.7158,
      "num_input_tokens_seen": 417408,
      "step": 645
    },
    {
      "epoch": 0.38325471698113206,
      "grad_norm": 21.167978286743164,
      "learning_rate": 1.9133254716981133e-06,
      "loss": 1.8702,
      "num_input_tokens_seen": 420640,
      "step": 650
    },
    {
      "epoch": 0.38620283018867924,
      "grad_norm": 24.932361602783203,
      "learning_rate": 1.9280660377358494e-06,
      "loss": 1.9492,
      "num_input_tokens_seen": 422688,
      "step": 655
    },
    {
      "epoch": 0.3891509433962264,
      "grad_norm": 14.881708145141602,
      "learning_rate": 1.9428066037735852e-06,
      "loss": 2.083,
      "num_input_tokens_seen": 427136,
      "step": 660
    },
    {
      "epoch": 0.3920990566037736,
      "grad_norm": 24.323211669921875,
      "learning_rate": 1.957547169811321e-06,
      "loss": 1.7583,
      "num_input_tokens_seen": 430400,
      "step": 665
    },
    {
      "epoch": 0.39504716981132076,
      "grad_norm": 12.40609359741211,
      "learning_rate": 1.9722877358490568e-06,
      "loss": 1.7232,
      "num_input_tokens_seen": 433632,
      "step": 670
    },
    {
      "epoch": 0.39799528301886794,
      "grad_norm": 16.017995834350586,
      "learning_rate": 1.9870283018867925e-06,
      "loss": 1.8499,
      "num_input_tokens_seen": 436960,
      "step": 675
    },
    {
      "epoch": 0.4009433962264151,
      "grad_norm": 14.273941040039062,
      "learning_rate": 2.0017688679245283e-06,
      "loss": 1.5063,
      "num_input_tokens_seen": 439968,
      "step": 680
    },
    {
      "epoch": 0.40389150943396224,
      "grad_norm": 20.851783752441406,
      "learning_rate": 2.0165094339622645e-06,
      "loss": 1.5491,
      "num_input_tokens_seen": 443168,
      "step": 685
    },
    {
      "epoch": 0.4068396226415094,
      "grad_norm": 14.981064796447754,
      "learning_rate": 2.0312500000000002e-06,
      "loss": 1.5964,
      "num_input_tokens_seen": 445984,
      "step": 690
    },
    {
      "epoch": 0.4097877358490566,
      "grad_norm": 22.555187225341797,
      "learning_rate": 2.045990566037736e-06,
      "loss": 1.7867,
      "num_input_tokens_seen": 449280,
      "step": 695
    },
    {
      "epoch": 0.41273584905660377,
      "grad_norm": 18.690513610839844,
      "learning_rate": 2.0607311320754718e-06,
      "loss": 1.5458,
      "num_input_tokens_seen": 452832,
      "step": 700
    },
    {
      "epoch": 0.41568396226415094,
      "grad_norm": 16.948097229003906,
      "learning_rate": 2.075471698113208e-06,
      "loss": 1.3588,
      "num_input_tokens_seen": 455360,
      "step": 705
    },
    {
      "epoch": 0.4186320754716981,
      "grad_norm": 11.862958908081055,
      "learning_rate": 2.0902122641509437e-06,
      "loss": 1.3306,
      "num_input_tokens_seen": 458944,
      "step": 710
    },
    {
      "epoch": 0.4215801886792453,
      "grad_norm": 13.978707313537598,
      "learning_rate": 2.1049528301886795e-06,
      "loss": 1.4604,
      "num_input_tokens_seen": 462080,
      "step": 715
    },
    {
      "epoch": 0.42452830188679247,
      "grad_norm": 21.152074813842773,
      "learning_rate": 2.1196933962264153e-06,
      "loss": 1.3827,
      "num_input_tokens_seen": 465888,
      "step": 720
    },
    {
      "epoch": 0.42747641509433965,
      "grad_norm": 36.478572845458984,
      "learning_rate": 2.134433962264151e-06,
      "loss": 1.6747,
      "num_input_tokens_seen": 468928,
      "step": 725
    },
    {
      "epoch": 0.43042452830188677,
      "grad_norm": 15.41918659210205,
      "learning_rate": 2.149174528301887e-06,
      "loss": 0.9826,
      "num_input_tokens_seen": 472032,
      "step": 730
    },
    {
      "epoch": 0.43337264150943394,
      "grad_norm": 12.585308074951172,
      "learning_rate": 2.163915094339623e-06,
      "loss": 1.2454,
      "num_input_tokens_seen": 474816,
      "step": 735
    },
    {
      "epoch": 0.4363207547169811,
      "grad_norm": 15.616990089416504,
      "learning_rate": 2.1786556603773588e-06,
      "loss": 1.61,
      "num_input_tokens_seen": 477344,
      "step": 740
    },
    {
      "epoch": 0.4392688679245283,
      "grad_norm": 14.420464515686035,
      "learning_rate": 2.1933962264150945e-06,
      "loss": 1.2651,
      "num_input_tokens_seen": 480096,
      "step": 745
    },
    {
      "epoch": 0.44221698113207547,
      "grad_norm": 9.266770362854004,
      "learning_rate": 2.2081367924528303e-06,
      "loss": 0.9204,
      "num_input_tokens_seen": 483712,
      "step": 750
    },
    {
      "epoch": 0.44516509433962265,
      "grad_norm": 14.800943374633789,
      "learning_rate": 2.2228773584905665e-06,
      "loss": 0.9592,
      "num_input_tokens_seen": 487072,
      "step": 755
    },
    {
      "epoch": 0.4481132075471698,
      "grad_norm": 12.020878791809082,
      "learning_rate": 2.237617924528302e-06,
      "loss": 1.185,
      "num_input_tokens_seen": 490432,
      "step": 760
    },
    {
      "epoch": 0.451061320754717,
      "grad_norm": 13.600262641906738,
      "learning_rate": 2.252358490566038e-06,
      "loss": 0.8052,
      "num_input_tokens_seen": 493440,
      "step": 765
    },
    {
      "epoch": 0.4540094339622642,
      "grad_norm": 31.355731964111328,
      "learning_rate": 2.267099056603774e-06,
      "loss": 0.947,
      "num_input_tokens_seen": 496000,
      "step": 770
    },
    {
      "epoch": 0.4569575471698113,
      "grad_norm": 13.489664077758789,
      "learning_rate": 2.2818396226415096e-06,
      "loss": 0.9103,
      "num_input_tokens_seen": 499904,
      "step": 775
    },
    {
      "epoch": 0.45990566037735847,
      "grad_norm": 13.685105323791504,
      "learning_rate": 2.2965801886792453e-06,
      "loss": 0.9831,
      "num_input_tokens_seen": 504256,
      "step": 780
    },
    {
      "epoch": 0.46285377358490565,
      "grad_norm": 10.982120513916016,
      "learning_rate": 2.3113207547169815e-06,
      "loss": 0.7673,
      "num_input_tokens_seen": 507616,
      "step": 785
    },
    {
      "epoch": 0.4658018867924528,
      "grad_norm": 7.945463180541992,
      "learning_rate": 2.3260613207547173e-06,
      "loss": 1.0915,
      "num_input_tokens_seen": 510720,
      "step": 790
    },
    {
      "epoch": 0.46875,
      "grad_norm": 10.44337272644043,
      "learning_rate": 2.340801886792453e-06,
      "loss": 0.8194,
      "num_input_tokens_seen": 514016,
      "step": 795
    },
    {
      "epoch": 0.4716981132075472,
      "grad_norm": 6.346359729766846,
      "learning_rate": 2.355542452830189e-06,
      "loss": 0.8714,
      "num_input_tokens_seen": 517024,
      "step": 800
    },
    {
      "epoch": 0.47464622641509435,
      "grad_norm": 7.394376277923584,
      "learning_rate": 2.3702830188679246e-06,
      "loss": 1.102,
      "num_input_tokens_seen": 520032,
      "step": 805
    },
    {
      "epoch": 0.47759433962264153,
      "grad_norm": 9.357361793518066,
      "learning_rate": 2.3850235849056604e-06,
      "loss": 0.9178,
      "num_input_tokens_seen": 523232,
      "step": 810
    },
    {
      "epoch": 0.4805424528301887,
      "grad_norm": 9.115985870361328,
      "learning_rate": 2.3997641509433966e-06,
      "loss": 0.8341,
      "num_input_tokens_seen": 526080,
      "step": 815
    },
    {
      "epoch": 0.4834905660377358,
      "grad_norm": 13.859272956848145,
      "learning_rate": 2.4145047169811323e-06,
      "loss": 0.8807,
      "num_input_tokens_seen": 528384,
      "step": 820
    },
    {
      "epoch": 0.486438679245283,
      "grad_norm": 9.01256275177002,
      "learning_rate": 2.429245283018868e-06,
      "loss": 0.699,
      "num_input_tokens_seen": 531808,
      "step": 825
    },
    {
      "epoch": 0.4893867924528302,
      "grad_norm": 11.935478210449219,
      "learning_rate": 2.443985849056604e-06,
      "loss": 0.9971,
      "num_input_tokens_seen": 535392,
      "step": 830
    },
    {
      "epoch": 0.49233490566037735,
      "grad_norm": 11.69451904296875,
      "learning_rate": 2.45872641509434e-06,
      "loss": 0.9152,
      "num_input_tokens_seen": 538752,
      "step": 835
    },
    {
      "epoch": 0.49528301886792453,
      "grad_norm": 11.001533508300781,
      "learning_rate": 2.473466981132076e-06,
      "loss": 0.8088,
      "num_input_tokens_seen": 541920,
      "step": 840
    },
    {
      "epoch": 0.4982311320754717,
      "grad_norm": 8.332868576049805,
      "learning_rate": 2.4882075471698116e-06,
      "loss": 0.856,
      "num_input_tokens_seen": 545024,
      "step": 845
    },
    {
      "epoch": 0.5011792452830188,
      "grad_norm": 9.973461151123047,
      "learning_rate": 2.5029481132075474e-06,
      "loss": 0.8352,
      "num_input_tokens_seen": 548384,
      "step": 850
    },
    {
      "epoch": 0.504127358490566,
      "grad_norm": 8.795652389526367,
      "learning_rate": 2.517688679245283e-06,
      "loss": 0.8574,
      "num_input_tokens_seen": 551712,
      "step": 855
    },
    {
      "epoch": 0.5070754716981132,
      "grad_norm": 5.081294536590576,
      "learning_rate": 2.532429245283019e-06,
      "loss": 0.7831,
      "num_input_tokens_seen": 554688,
      "step": 860
    },
    {
      "epoch": 0.5100235849056604,
      "grad_norm": 11.282505989074707,
      "learning_rate": 2.547169811320755e-06,
      "loss": 1.0198,
      "num_input_tokens_seen": 557888,
      "step": 865
    },
    {
      "epoch": 0.5129716981132075,
      "grad_norm": 5.791973114013672,
      "learning_rate": 2.561910377358491e-06,
      "loss": 0.7339,
      "num_input_tokens_seen": 561856,
      "step": 870
    },
    {
      "epoch": 0.5159198113207547,
      "grad_norm": 5.20530891418457,
      "learning_rate": 2.5766509433962266e-06,
      "loss": 0.711,
      "num_input_tokens_seen": 565024,
      "step": 875
    },
    {
      "epoch": 0.5188679245283019,
      "grad_norm": 5.657922744750977,
      "learning_rate": 2.5913915094339624e-06,
      "loss": 0.9388,
      "num_input_tokens_seen": 568480,
      "step": 880
    },
    {
      "epoch": 0.5218160377358491,
      "grad_norm": 12.070975303649902,
      "learning_rate": 2.6061320754716986e-06,
      "loss": 1.1166,
      "num_input_tokens_seen": 572096,
      "step": 885
    },
    {
      "epoch": 0.5247641509433962,
      "grad_norm": 6.184645652770996,
      "learning_rate": 2.6208726415094343e-06,
      "loss": 0.9136,
      "num_input_tokens_seen": 575520,
      "step": 890
    },
    {
      "epoch": 0.5277122641509434,
      "grad_norm": 7.415177822113037,
      "learning_rate": 2.63561320754717e-06,
      "loss": 0.748,
      "num_input_tokens_seen": 578848,
      "step": 895
    },
    {
      "epoch": 0.5306603773584906,
      "grad_norm": 7.983150005340576,
      "learning_rate": 2.650353773584906e-06,
      "loss": 0.6337,
      "num_input_tokens_seen": 583008,
      "step": 900
    },
    {
      "epoch": 0.5336084905660378,
      "grad_norm": 16.439682006835938,
      "learning_rate": 2.665094339622642e-06,
      "loss": 0.7095,
      "num_input_tokens_seen": 586432,
      "step": 905
    },
    {
      "epoch": 0.5365566037735849,
      "grad_norm": 7.634627819061279,
      "learning_rate": 2.679834905660378e-06,
      "loss": 0.877,
      "num_input_tokens_seen": 589568,
      "step": 910
    },
    {
      "epoch": 0.5395047169811321,
      "grad_norm": 7.967339038848877,
      "learning_rate": 2.694575471698113e-06,
      "loss": 0.7737,
      "num_input_tokens_seen": 592416,
      "step": 915
    },
    {
      "epoch": 0.5424528301886793,
      "grad_norm": 7.440727233886719,
      "learning_rate": 2.709316037735849e-06,
      "loss": 0.4487,
      "num_input_tokens_seen": 595552,
      "step": 920
    },
    {
      "epoch": 0.5454009433962265,
      "grad_norm": 5.393298625946045,
      "learning_rate": 2.724056603773585e-06,
      "loss": 0.5329,
      "num_input_tokens_seen": 599104,
      "step": 925
    },
    {
      "epoch": 0.5483490566037735,
      "grad_norm": 5.736482620239258,
      "learning_rate": 2.738797169811321e-06,
      "loss": 0.5568,
      "num_input_tokens_seen": 602304,
      "step": 930
    },
    {
      "epoch": 0.5512971698113207,
      "grad_norm": 18.674306869506836,
      "learning_rate": 2.7535377358490567e-06,
      "loss": 0.947,
      "num_input_tokens_seen": 605376,
      "step": 935
    },
    {
      "epoch": 0.5542452830188679,
      "grad_norm": 6.868896961212158,
      "learning_rate": 2.7682783018867925e-06,
      "loss": 0.9636,
      "num_input_tokens_seen": 607648,
      "step": 940
    },
    {
      "epoch": 0.5571933962264151,
      "grad_norm": 3.72993803024292,
      "learning_rate": 2.7830188679245286e-06,
      "loss": 0.7103,
      "num_input_tokens_seen": 611232,
      "step": 945
    },
    {
      "epoch": 0.5601415094339622,
      "grad_norm": 6.169341087341309,
      "learning_rate": 2.7977594339622644e-06,
      "loss": 0.6143,
      "num_input_tokens_seen": 615776,
      "step": 950
    },
    {
      "epoch": 0.5630896226415094,
      "grad_norm": 18.689990997314453,
      "learning_rate": 2.8125e-06,
      "loss": 0.842,
      "num_input_tokens_seen": 618656,
      "step": 955
    },
    {
      "epoch": 0.5660377358490566,
      "grad_norm": 6.143091678619385,
      "learning_rate": 2.827240566037736e-06,
      "loss": 0.6467,
      "num_input_tokens_seen": 622528,
      "step": 960
    },
    {
      "epoch": 0.5689858490566038,
      "grad_norm": 6.399939060211182,
      "learning_rate": 2.841981132075472e-06,
      "loss": 0.6852,
      "num_input_tokens_seen": 625920,
      "step": 965
    },
    {
      "epoch": 0.5719339622641509,
      "grad_norm": 5.964336395263672,
      "learning_rate": 2.856721698113208e-06,
      "loss": 0.6648,
      "num_input_tokens_seen": 628544,
      "step": 970
    },
    {
      "epoch": 0.5748820754716981,
      "grad_norm": 7.366810321807861,
      "learning_rate": 2.8714622641509437e-06,
      "loss": 0.7258,
      "num_input_tokens_seen": 631072,
      "step": 975
    },
    {
      "epoch": 0.5778301886792453,
      "grad_norm": 5.7462663650512695,
      "learning_rate": 2.8862028301886794e-06,
      "loss": 0.6933,
      "num_input_tokens_seen": 633504,
      "step": 980
    },
    {
      "epoch": 0.5807783018867925,
      "grad_norm": 10.469639778137207,
      "learning_rate": 2.9009433962264156e-06,
      "loss": 0.7531,
      "num_input_tokens_seen": 636800,
      "step": 985
    },
    {
      "epoch": 0.5837264150943396,
      "grad_norm": 4.1671223640441895,
      "learning_rate": 2.9156839622641514e-06,
      "loss": 0.5618,
      "num_input_tokens_seen": 640160,
      "step": 990
    },
    {
      "epoch": 0.5866745283018868,
      "grad_norm": 6.803442478179932,
      "learning_rate": 2.930424528301887e-06,
      "loss": 0.7435,
      "num_input_tokens_seen": 643936,
      "step": 995
    },
    {
      "epoch": 0.589622641509434,
      "grad_norm": 5.059441566467285,
      "learning_rate": 2.9451650943396225e-06,
      "loss": 0.5652,
      "num_input_tokens_seen": 648384,
      "step": 1000
    },
    {
      "epoch": 0.5925707547169812,
      "grad_norm": 5.3123650550842285,
      "learning_rate": 2.9599056603773587e-06,
      "loss": 0.6299,
      "num_input_tokens_seen": 651168,
      "step": 1005
    },
    {
      "epoch": 0.5955188679245284,
      "grad_norm": 3.5172204971313477,
      "learning_rate": 2.9746462264150945e-06,
      "loss": 0.7942,
      "num_input_tokens_seen": 654272,
      "step": 1010
    },
    {
      "epoch": 0.5984669811320755,
      "grad_norm": 4.174629211425781,
      "learning_rate": 2.9893867924528302e-06,
      "loss": 0.5136,
      "num_input_tokens_seen": 657824,
      "step": 1015
    },
    {
      "epoch": 0.6014150943396226,
      "grad_norm": 4.112761497497559,
      "learning_rate": 3.004127358490566e-06,
      "loss": 0.6794,
      "num_input_tokens_seen": 661184,
      "step": 1020
    },
    {
      "epoch": 0.6043632075471698,
      "grad_norm": 5.572396755218506,
      "learning_rate": 3.018867924528302e-06,
      "loss": 0.6936,
      "num_input_tokens_seen": 665184,
      "step": 1025
    },
    {
      "epoch": 0.6073113207547169,
      "grad_norm": 4.812779426574707,
      "learning_rate": 3.033608490566038e-06,
      "loss": 0.5847,
      "num_input_tokens_seen": 667968,
      "step": 1030
    },
    {
      "epoch": 0.6102594339622641,
      "grad_norm": 4.473247528076172,
      "learning_rate": 3.0483490566037737e-06,
      "loss": 0.5575,
      "num_input_tokens_seen": 671360,
      "step": 1035
    },
    {
      "epoch": 0.6132075471698113,
      "grad_norm": 5.123810768127441,
      "learning_rate": 3.0630896226415095e-06,
      "loss": 0.5811,
      "num_input_tokens_seen": 674336,
      "step": 1040
    },
    {
      "epoch": 0.6161556603773585,
      "grad_norm": 5.766445159912109,
      "learning_rate": 3.0778301886792457e-06,
      "loss": 0.5332,
      "num_input_tokens_seen": 677504,
      "step": 1045
    },
    {
      "epoch": 0.6191037735849056,
      "grad_norm": 9.049609184265137,
      "learning_rate": 3.0925707547169815e-06,
      "loss": 0.6762,
      "num_input_tokens_seen": 680864,
      "step": 1050
    },
    {
      "epoch": 0.6220518867924528,
      "grad_norm": 5.498325347900391,
      "learning_rate": 3.1073113207547172e-06,
      "loss": 0.6861,
      "num_input_tokens_seen": 683776,
      "step": 1055
    },
    {
      "epoch": 0.625,
      "grad_norm": 4.329188346862793,
      "learning_rate": 3.122051886792453e-06,
      "loss": 0.6801,
      "num_input_tokens_seen": 686784,
      "step": 1060
    },
    {
      "epoch": 0.6279481132075472,
      "grad_norm": 4.751805782318115,
      "learning_rate": 3.136792452830189e-06,
      "loss": 0.5865,
      "num_input_tokens_seen": 690208,
      "step": 1065
    },
    {
      "epoch": 0.6308962264150944,
      "grad_norm": 7.6678385734558105,
      "learning_rate": 3.151533018867925e-06,
      "loss": 0.7595,
      "num_input_tokens_seen": 693664,
      "step": 1070
    },
    {
      "epoch": 0.6338443396226415,
      "grad_norm": 8.4994478225708,
      "learning_rate": 3.1662735849056607e-06,
      "loss": 0.8149,
      "num_input_tokens_seen": 697696,
      "step": 1075
    },
    {
      "epoch": 0.6367924528301887,
      "grad_norm": 4.134837627410889,
      "learning_rate": 3.181014150943397e-06,
      "loss": 0.6523,
      "num_input_tokens_seen": 700480,
      "step": 1080
    },
    {
      "epoch": 0.6397405660377359,
      "grad_norm": 7.377701759338379,
      "learning_rate": 3.1957547169811327e-06,
      "loss": 0.6243,
      "num_input_tokens_seen": 704096,
      "step": 1085
    },
    {
      "epoch": 0.6426886792452831,
      "grad_norm": 8.720990180969238,
      "learning_rate": 3.210495283018868e-06,
      "loss": 0.6165,
      "num_input_tokens_seen": 706720,
      "step": 1090
    },
    {
      "epoch": 0.6456367924528302,
      "grad_norm": 10.208901405334473,
      "learning_rate": 3.225235849056604e-06,
      "loss": 0.7694,
      "num_input_tokens_seen": 709536,
      "step": 1095
    },
    {
      "epoch": 0.6485849056603774,
      "grad_norm": 5.441288948059082,
      "learning_rate": 3.2399764150943396e-06,
      "loss": 0.7433,
      "num_input_tokens_seen": 713152,
      "step": 1100
    },
    {
      "epoch": 0.6515330188679245,
      "grad_norm": 8.408998489379883,
      "learning_rate": 3.2547169811320758e-06,
      "loss": 0.7226,
      "num_input_tokens_seen": 716160,
      "step": 1105
    },
    {
      "epoch": 0.6544811320754716,
      "grad_norm": 8.885174751281738,
      "learning_rate": 3.2694575471698115e-06,
      "loss": 0.6081,
      "num_input_tokens_seen": 719168,
      "step": 1110
    },
    {
      "epoch": 0.6574292452830188,
      "grad_norm": 4.572237014770508,
      "learning_rate": 3.2841981132075473e-06,
      "loss": 0.5957,
      "num_input_tokens_seen": 722944,
      "step": 1115
    },
    {
      "epoch": 0.660377358490566,
      "grad_norm": 13.368962287902832,
      "learning_rate": 3.298938679245283e-06,
      "loss": 0.7109,
      "num_input_tokens_seen": 726912,
      "step": 1120
    },
    {
      "epoch": 0.6633254716981132,
      "grad_norm": 11.320699691772461,
      "learning_rate": 3.3136792452830192e-06,
      "loss": 0.6351,
      "num_input_tokens_seen": 730688,
      "step": 1125
    },
    {
      "epoch": 0.6662735849056604,
      "grad_norm": 3.7924575805664062,
      "learning_rate": 3.328419811320755e-06,
      "loss": 0.7068,
      "num_input_tokens_seen": 734656,
      "step": 1130
    },
    {
      "epoch": 0.6692216981132075,
      "grad_norm": 12.168998718261719,
      "learning_rate": 3.3431603773584908e-06,
      "loss": 0.7861,
      "num_input_tokens_seen": 737408,
      "step": 1135
    },
    {
      "epoch": 0.6721698113207547,
      "grad_norm": 4.969852447509766,
      "learning_rate": 3.3579009433962266e-06,
      "loss": 0.5272,
      "num_input_tokens_seen": 740224,
      "step": 1140
    },
    {
      "epoch": 0.6751179245283019,
      "grad_norm": 3.422187089920044,
      "learning_rate": 3.3726415094339627e-06,
      "loss": 0.5752,
      "num_input_tokens_seen": 742688,
      "step": 1145
    },
    {
      "epoch": 0.6780660377358491,
      "grad_norm": 5.729838848114014,
      "learning_rate": 3.3873820754716985e-06,
      "loss": 0.6491,
      "num_input_tokens_seen": 745728,
      "step": 1150
    },
    {
      "epoch": 0.6810141509433962,
      "grad_norm": 4.363771438598633,
      "learning_rate": 3.4021226415094343e-06,
      "loss": 0.586,
      "num_input_tokens_seen": 749984,
      "step": 1155
    },
    {
      "epoch": 0.6839622641509434,
      "grad_norm": 2.963905096054077,
      "learning_rate": 3.4168632075471705e-06,
      "loss": 0.6162,
      "num_input_tokens_seen": 753408,
      "step": 1160
    },
    {
      "epoch": 0.6869103773584906,
      "grad_norm": 11.451322555541992,
      "learning_rate": 3.4316037735849062e-06,
      "loss": 0.6239,
      "num_input_tokens_seen": 755936,
      "step": 1165
    },
    {
      "epoch": 0.6898584905660378,
      "grad_norm": 4.222641468048096,
      "learning_rate": 3.446344339622642e-06,
      "loss": 0.6231,
      "num_input_tokens_seen": 759264,
      "step": 1170
    },
    {
      "epoch": 0.6928066037735849,
      "grad_norm": 3.43054461479187,
      "learning_rate": 3.4610849056603778e-06,
      "loss": 0.5357,
      "num_input_tokens_seen": 762976,
      "step": 1175
    },
    {
      "epoch": 0.6957547169811321,
      "grad_norm": 6.6390790939331055,
      "learning_rate": 3.475825471698113e-06,
      "loss": 0.5106,
      "num_input_tokens_seen": 765888,
      "step": 1180
    },
    {
      "epoch": 0.6987028301886793,
      "grad_norm": 2.2927517890930176,
      "learning_rate": 3.4905660377358493e-06,
      "loss": 0.6496,
      "num_input_tokens_seen": 769120,
      "step": 1185
    },
    {
      "epoch": 0.7016509433962265,
      "grad_norm": 2.9012715816497803,
      "learning_rate": 3.505306603773585e-06,
      "loss": 0.6433,
      "num_input_tokens_seen": 772704,
      "step": 1190
    },
    {
      "epoch": 0.7045990566037735,
      "grad_norm": 5.627623081207275,
      "learning_rate": 3.520047169811321e-06,
      "loss": 0.6652,
      "num_input_tokens_seen": 775328,
      "step": 1195
    },
    {
      "epoch": 0.7075471698113207,
      "grad_norm": 13.428121566772461,
      "learning_rate": 3.5347877358490566e-06,
      "loss": 0.6089,
      "num_input_tokens_seen": 778048,
      "step": 1200
    },
    {
      "epoch": 0.7104952830188679,
      "grad_norm": 4.02173376083374,
      "learning_rate": 3.549528301886793e-06,
      "loss": 0.6877,
      "num_input_tokens_seen": 781792,
      "step": 1205
    },
    {
      "epoch": 0.7134433962264151,
      "grad_norm": 9.602651596069336,
      "learning_rate": 3.5642688679245286e-06,
      "loss": 0.6718,
      "num_input_tokens_seen": 784000,
      "step": 1210
    },
    {
      "epoch": 0.7163915094339622,
      "grad_norm": 4.908822059631348,
      "learning_rate": 3.5790094339622643e-06,
      "loss": 0.7184,
      "num_input_tokens_seen": 788256,
      "step": 1215
    },
    {
      "epoch": 0.7193396226415094,
      "grad_norm": 4.420402526855469,
      "learning_rate": 3.59375e-06,
      "loss": 0.6878,
      "num_input_tokens_seen": 791648,
      "step": 1220
    },
    {
      "epoch": 0.7222877358490566,
      "grad_norm": 5.150833606719971,
      "learning_rate": 3.6084905660377363e-06,
      "loss": 0.5438,
      "num_input_tokens_seen": 795872,
      "step": 1225
    },
    {
      "epoch": 0.7252358490566038,
      "grad_norm": 7.396393299102783,
      "learning_rate": 3.623231132075472e-06,
      "loss": 0.6473,
      "num_input_tokens_seen": 798752,
      "step": 1230
    },
    {
      "epoch": 0.7281839622641509,
      "grad_norm": 4.693133354187012,
      "learning_rate": 3.637971698113208e-06,
      "loss": 0.6027,
      "num_input_tokens_seen": 802048,
      "step": 1235
    },
    {
      "epoch": 0.7311320754716981,
      "grad_norm": 5.008869647979736,
      "learning_rate": 3.652712264150944e-06,
      "loss": 0.65,
      "num_input_tokens_seen": 804768,
      "step": 1240
    },
    {
      "epoch": 0.7340801886792453,
      "grad_norm": 3.0386149883270264,
      "learning_rate": 3.66745283018868e-06,
      "loss": 0.6283,
      "num_input_tokens_seen": 808896,
      "step": 1245
    },
    {
      "epoch": 0.7370283018867925,
      "grad_norm": 5.325741291046143,
      "learning_rate": 3.6821933962264156e-06,
      "loss": 0.6092,
      "num_input_tokens_seen": 812288,
      "step": 1250
    },
    {
      "epoch": 0.7399764150943396,
      "grad_norm": 2.8650918006896973,
      "learning_rate": 3.6969339622641513e-06,
      "loss": 0.6638,
      "num_input_tokens_seen": 815840,
      "step": 1255
    },
    {
      "epoch": 0.7429245283018868,
      "grad_norm": 5.373302936553955,
      "learning_rate": 3.7116745283018875e-06,
      "loss": 0.5759,
      "num_input_tokens_seen": 819168,
      "step": 1260
    },
    {
      "epoch": 0.745872641509434,
      "grad_norm": 2.8748016357421875,
      "learning_rate": 3.726415094339623e-06,
      "loss": 0.4967,
      "num_input_tokens_seen": 822944,
      "step": 1265
    },
    {
      "epoch": 0.7488207547169812,
      "grad_norm": 4.558932781219482,
      "learning_rate": 3.7411556603773586e-06,
      "loss": 0.6567,
      "num_input_tokens_seen": 826272,
      "step": 1270
    },
    {
      "epoch": 0.7517688679245284,
      "grad_norm": 3.5304794311523438,
      "learning_rate": 3.7558962264150944e-06,
      "loss": 0.585,
      "num_input_tokens_seen": 830144,
      "step": 1275
    },
    {
      "epoch": 0.7547169811320755,
      "grad_norm": 8.17846393585205,
      "learning_rate": 3.77063679245283e-06,
      "loss": 0.6279,
      "num_input_tokens_seen": 832960,
      "step": 1280
    },
    {
      "epoch": 0.7576650943396226,
      "grad_norm": 9.533529281616211,
      "learning_rate": 3.7853773584905664e-06,
      "loss": 0.6032,
      "num_input_tokens_seen": 835392,
      "step": 1285
    },
    {
      "epoch": 0.7606132075471698,
      "grad_norm": 2.678600311279297,
      "learning_rate": 3.800117924528302e-06,
      "loss": 0.5209,
      "num_input_tokens_seen": 838592,
      "step": 1290
    },
    {
      "epoch": 0.7635613207547169,
      "grad_norm": 6.290689945220947,
      "learning_rate": 3.814858490566038e-06,
      "loss": 0.4959,
      "num_input_tokens_seen": 841824,
      "step": 1295
    },
    {
      "epoch": 0.7665094339622641,
      "grad_norm": 4.903485298156738,
      "learning_rate": 3.829599056603774e-06,
      "loss": 0.5725,
      "num_input_tokens_seen": 845760,
      "step": 1300
    },
    {
      "epoch": 0.7694575471698113,
      "grad_norm": 5.686278820037842,
      "learning_rate": 3.8443396226415094e-06,
      "loss": 0.8529,
      "num_input_tokens_seen": 852288,
      "step": 1305
    },
    {
      "epoch": 0.7724056603773585,
      "grad_norm": 8.209455490112305,
      "learning_rate": 3.859080188679246e-06,
      "loss": 0.553,
      "num_input_tokens_seen": 855008,
      "step": 1310
    },
    {
      "epoch": 0.7753537735849056,
      "grad_norm": 5.806557655334473,
      "learning_rate": 3.873820754716982e-06,
      "loss": 0.5266,
      "num_input_tokens_seen": 857728,
      "step": 1315
    },
    {
      "epoch": 0.7783018867924528,
      "grad_norm": 4.764500141143799,
      "learning_rate": 3.888561320754717e-06,
      "loss": 0.5665,
      "num_input_tokens_seen": 861472,
      "step": 1320
    },
    {
      "epoch": 0.78125,
      "grad_norm": 11.363874435424805,
      "learning_rate": 3.903301886792453e-06,
      "loss": 0.6286,
      "num_input_tokens_seen": 865568,
      "step": 1325
    },
    {
      "epoch": 0.7841981132075472,
      "grad_norm": 8.491943359375,
      "learning_rate": 3.9180424528301895e-06,
      "loss": 0.7196,
      "num_input_tokens_seen": 868928,
      "step": 1330
    },
    {
      "epoch": 0.7871462264150944,
      "grad_norm": 5.015326023101807,
      "learning_rate": 3.932783018867925e-06,
      "loss": 0.4917,
      "num_input_tokens_seen": 872096,
      "step": 1335
    },
    {
      "epoch": 0.7900943396226415,
      "grad_norm": 2.4574317932128906,
      "learning_rate": 3.947523584905661e-06,
      "loss": 0.434,
      "num_input_tokens_seen": 875296,
      "step": 1340
    },
    {
      "epoch": 0.7930424528301887,
      "grad_norm": 25.127607345581055,
      "learning_rate": 3.962264150943396e-06,
      "loss": 0.5318,
      "num_input_tokens_seen": 878144,
      "step": 1345
    },
    {
      "epoch": 0.7959905660377359,
      "grad_norm": 4.2509918212890625,
      "learning_rate": 3.977004716981133e-06,
      "loss": 0.7072,
      "num_input_tokens_seen": 880928,
      "step": 1350
    },
    {
      "epoch": 0.7989386792452831,
      "grad_norm": 8.605904579162598,
      "learning_rate": 3.991745283018868e-06,
      "loss": 0.5436,
      "num_input_tokens_seen": 883680,
      "step": 1355
    },
    {
      "epoch": 0.8018867924528302,
      "grad_norm": 6.206994533538818,
      "learning_rate": 4.006485849056604e-06,
      "loss": 0.7083,
      "num_input_tokens_seen": 886368,
      "step": 1360
    },
    {
      "epoch": 0.8048349056603774,
      "grad_norm": 4.296226978302002,
      "learning_rate": 4.0212264150943395e-06,
      "loss": 0.7309,
      "num_input_tokens_seen": 889728,
      "step": 1365
    },
    {
      "epoch": 0.8077830188679245,
      "grad_norm": 7.924245834350586,
      "learning_rate": 4.035966981132076e-06,
      "loss": 0.5954,
      "num_input_tokens_seen": 892992,
      "step": 1370
    },
    {
      "epoch": 0.8107311320754716,
      "grad_norm": 5.801962375640869,
      "learning_rate": 4.050707547169812e-06,
      "loss": 0.6435,
      "num_input_tokens_seen": 895712,
      "step": 1375
    },
    {
      "epoch": 0.8136792452830188,
      "grad_norm": 6.71798038482666,
      "learning_rate": 4.065448113207547e-06,
      "loss": 0.6819,
      "num_input_tokens_seen": 898400,
      "step": 1380
    },
    {
      "epoch": 0.816627358490566,
      "grad_norm": 4.68912410736084,
      "learning_rate": 4.080188679245283e-06,
      "loss": 0.5789,
      "num_input_tokens_seen": 901440,
      "step": 1385
    },
    {
      "epoch": 0.8195754716981132,
      "grad_norm": 11.836541175842285,
      "learning_rate": 4.094929245283019e-06,
      "loss": 0.6054,
      "num_input_tokens_seen": 904288,
      "step": 1390
    },
    {
      "epoch": 0.8225235849056604,
      "grad_norm": 5.365652084350586,
      "learning_rate": 4.109669811320755e-06,
      "loss": 0.5652,
      "num_input_tokens_seen": 907776,
      "step": 1395
    },
    {
      "epoch": 0.8254716981132075,
      "grad_norm": 6.0620293617248535,
      "learning_rate": 4.124410377358491e-06,
      "loss": 0.6495,
      "num_input_tokens_seen": 911584,
      "step": 1400
    },
    {
      "epoch": 0.8284198113207547,
      "grad_norm": 6.532092571258545,
      "learning_rate": 4.1391509433962265e-06,
      "loss": 0.5128,
      "num_input_tokens_seen": 915232,
      "step": 1405
    },
    {
      "epoch": 0.8313679245283019,
      "grad_norm": 5.426935195922852,
      "learning_rate": 4.153891509433963e-06,
      "loss": 0.5075,
      "num_input_tokens_seen": 917696,
      "step": 1410
    },
    {
      "epoch": 0.8343160377358491,
      "grad_norm": 12.016154289245605,
      "learning_rate": 4.168632075471699e-06,
      "loss": 0.7284,
      "num_input_tokens_seen": 922656,
      "step": 1415
    },
    {
      "epoch": 0.8372641509433962,
      "grad_norm": 7.776079177856445,
      "learning_rate": 4.183372641509434e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 925856,
      "step": 1420
    },
    {
      "epoch": 0.8402122641509434,
      "grad_norm": 3.587447166442871,
      "learning_rate": 4.19811320754717e-06,
      "loss": 0.661,
      "num_input_tokens_seen": 930176,
      "step": 1425
    },
    {
      "epoch": 0.8431603773584906,
      "grad_norm": 3.0893397331237793,
      "learning_rate": 4.212853773584907e-06,
      "loss": 0.6749,
      "num_input_tokens_seen": 933472,
      "step": 1430
    },
    {
      "epoch": 0.8461084905660378,
      "grad_norm": 8.176351547241211,
      "learning_rate": 4.227594339622642e-06,
      "loss": 0.5572,
      "num_input_tokens_seen": 936416,
      "step": 1435
    },
    {
      "epoch": 0.8490566037735849,
      "grad_norm": 3.689838171005249,
      "learning_rate": 4.242334905660378e-06,
      "loss": 0.6833,
      "num_input_tokens_seen": 939136,
      "step": 1440
    },
    {
      "epoch": 0.8520047169811321,
      "grad_norm": 5.29396390914917,
      "learning_rate": 4.2570754716981135e-06,
      "loss": 0.6223,
      "num_input_tokens_seen": 943168,
      "step": 1445
    },
    {
      "epoch": 0.8549528301886793,
      "grad_norm": 14.127381324768066,
      "learning_rate": 4.271816037735849e-06,
      "loss": 0.5195,
      "num_input_tokens_seen": 945568,
      "step": 1450
    },
    {
      "epoch": 0.8579009433962265,
      "grad_norm": 2.6478066444396973,
      "learning_rate": 4.286556603773585e-06,
      "loss": 0.5574,
      "num_input_tokens_seen": 948416,
      "step": 1455
    },
    {
      "epoch": 0.8608490566037735,
      "grad_norm": 2.951313018798828,
      "learning_rate": 4.301297169811321e-06,
      "loss": 0.5297,
      "num_input_tokens_seen": 950880,
      "step": 1460
    },
    {
      "epoch": 0.8637971698113207,
      "grad_norm": 7.67824125289917,
      "learning_rate": 4.3160377358490565e-06,
      "loss": 0.5256,
      "num_input_tokens_seen": 953632,
      "step": 1465
    },
    {
      "epoch": 0.8667452830188679,
      "grad_norm": 6.527291774749756,
      "learning_rate": 4.330778301886793e-06,
      "loss": 0.6593,
      "num_input_tokens_seen": 956320,
      "step": 1470
    },
    {
      "epoch": 0.8696933962264151,
      "grad_norm": 3.0069620609283447,
      "learning_rate": 4.345518867924529e-06,
      "loss": 0.6287,
      "num_input_tokens_seen": 959712,
      "step": 1475
    },
    {
      "epoch": 0.8726415094339622,
      "grad_norm": 3.1576664447784424,
      "learning_rate": 4.360259433962264e-06,
      "loss": 0.5644,
      "num_input_tokens_seen": 963264,
      "step": 1480
    },
    {
      "epoch": 0.8755896226415094,
      "grad_norm": 3.573331594467163,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 0.5753,
      "num_input_tokens_seen": 967424,
      "step": 1485
    },
    {
      "epoch": 0.8785377358490566,
      "grad_norm": 9.45398235321045,
      "learning_rate": 4.389740566037737e-06,
      "loss": 0.5454,
      "num_input_tokens_seen": 974592,
      "step": 1490
    },
    {
      "epoch": 0.8814858490566038,
      "grad_norm": 4.0027875900268555,
      "learning_rate": 4.404481132075472e-06,
      "loss": 0.6659,
      "num_input_tokens_seen": 978784,
      "step": 1495
    },
    {
      "epoch": 0.8844339622641509,
      "grad_norm": 7.140857219696045,
      "learning_rate": 4.419221698113208e-06,
      "loss": 0.5154,
      "num_input_tokens_seen": 981696,
      "step": 1500
    },
    {
      "epoch": 0.8873820754716981,
      "grad_norm": 11.862122535705566,
      "learning_rate": 4.4339622641509435e-06,
      "loss": 0.6376,
      "num_input_tokens_seen": 985408,
      "step": 1505
    },
    {
      "epoch": 0.8903301886792453,
      "grad_norm": 6.621129512786865,
      "learning_rate": 4.44870283018868e-06,
      "loss": 0.6862,
      "num_input_tokens_seen": 988576,
      "step": 1510
    },
    {
      "epoch": 0.8932783018867925,
      "grad_norm": 10.653403282165527,
      "learning_rate": 4.463443396226416e-06,
      "loss": 0.59,
      "num_input_tokens_seen": 991840,
      "step": 1515
    },
    {
      "epoch": 0.8962264150943396,
      "grad_norm": 5.599077224731445,
      "learning_rate": 4.478183962264151e-06,
      "loss": 0.6499,
      "num_input_tokens_seen": 995200,
      "step": 1520
    },
    {
      "epoch": 0.8991745283018868,
      "grad_norm": 4.309756278991699,
      "learning_rate": 4.4929245283018875e-06,
      "loss": 0.6689,
      "num_input_tokens_seen": 998912,
      "step": 1525
    },
    {
      "epoch": 0.902122641509434,
      "grad_norm": 6.143583297729492,
      "learning_rate": 4.507665094339623e-06,
      "loss": 0.5739,
      "num_input_tokens_seen": 1002848,
      "step": 1530
    },
    {
      "epoch": 0.9050707547169812,
      "grad_norm": 4.078437805175781,
      "learning_rate": 4.522405660377359e-06,
      "loss": 0.6967,
      "num_input_tokens_seen": 1005312,
      "step": 1535
    },
    {
      "epoch": 0.9080188679245284,
      "grad_norm": 3.9049553871154785,
      "learning_rate": 4.537146226415094e-06,
      "loss": 0.5949,
      "num_input_tokens_seen": 1008576,
      "step": 1540
    },
    {
      "epoch": 0.9109669811320755,
      "grad_norm": 2.814183473587036,
      "learning_rate": 4.5518867924528305e-06,
      "loss": 0.6176,
      "num_input_tokens_seen": 1012736,
      "step": 1545
    },
    {
      "epoch": 0.9139150943396226,
      "grad_norm": 3.7803328037261963,
      "learning_rate": 4.566627358490566e-06,
      "loss": 0.5534,
      "num_input_tokens_seen": 1016320,
      "step": 1550
    },
    {
      "epoch": 0.9168632075471698,
      "grad_norm": 4.066215515136719,
      "learning_rate": 4.581367924528302e-06,
      "loss": 0.6508,
      "num_input_tokens_seen": 1019392,
      "step": 1555
    },
    {
      "epoch": 0.9198113207547169,
      "grad_norm": 4.001386642456055,
      "learning_rate": 4.596108490566038e-06,
      "loss": 0.4678,
      "num_input_tokens_seen": 1022080,
      "step": 1560
    },
    {
      "epoch": 0.9227594339622641,
      "grad_norm": 4.080419063568115,
      "learning_rate": 4.610849056603774e-06,
      "loss": 0.5864,
      "num_input_tokens_seen": 1024960,
      "step": 1565
    },
    {
      "epoch": 0.9257075471698113,
      "grad_norm": 3.7366015911102295,
      "learning_rate": 4.62558962264151e-06,
      "loss": 0.6301,
      "num_input_tokens_seen": 1028480,
      "step": 1570
    },
    {
      "epoch": 0.9286556603773585,
      "grad_norm": 5.948261260986328,
      "learning_rate": 4.640330188679246e-06,
      "loss": 0.5089,
      "num_input_tokens_seen": 1031328,
      "step": 1575
    },
    {
      "epoch": 0.9316037735849056,
      "grad_norm": 4.513754367828369,
      "learning_rate": 4.655070754716981e-06,
      "loss": 0.4194,
      "num_input_tokens_seen": 1034048,
      "step": 1580
    },
    {
      "epoch": 0.9345518867924528,
      "grad_norm": 3.3506526947021484,
      "learning_rate": 4.6698113207547175e-06,
      "loss": 0.5023,
      "num_input_tokens_seen": 1037024,
      "step": 1585
    },
    {
      "epoch": 0.9375,
      "grad_norm": 4.534147262573242,
      "learning_rate": 4.684551886792454e-06,
      "loss": 0.5958,
      "num_input_tokens_seen": 1040128,
      "step": 1590
    },
    {
      "epoch": 0.9404481132075472,
      "grad_norm": 4.203045845031738,
      "learning_rate": 4.699292452830189e-06,
      "loss": 0.5879,
      "num_input_tokens_seen": 1043328,
      "step": 1595
    },
    {
      "epoch": 0.9433962264150944,
      "grad_norm": 4.055345058441162,
      "learning_rate": 4.714033018867925e-06,
      "loss": 0.5938,
      "num_input_tokens_seen": 1046176,
      "step": 1600
    },
    {
      "epoch": 0.9463443396226415,
      "grad_norm": 2.8795461654663086,
      "learning_rate": 4.728773584905661e-06,
      "loss": 0.6332,
      "num_input_tokens_seen": 1048896,
      "step": 1605
    },
    {
      "epoch": 0.9492924528301887,
      "grad_norm": 2.4212327003479004,
      "learning_rate": 4.743514150943397e-06,
      "loss": 0.6347,
      "num_input_tokens_seen": 1052768,
      "step": 1610
    },
    {
      "epoch": 0.9522405660377359,
      "grad_norm": 10.593609809875488,
      "learning_rate": 4.758254716981133e-06,
      "loss": 0.5877,
      "num_input_tokens_seen": 1055616,
      "step": 1615
    },
    {
      "epoch": 0.9551886792452831,
      "grad_norm": 2.602233648300171,
      "learning_rate": 4.772995283018868e-06,
      "loss": 0.5421,
      "num_input_tokens_seen": 1059072,
      "step": 1620
    },
    {
      "epoch": 0.9581367924528302,
      "grad_norm": 4.015645503997803,
      "learning_rate": 4.787735849056604e-06,
      "loss": 0.7592,
      "num_input_tokens_seen": 1061920,
      "step": 1625
    },
    {
      "epoch": 0.9610849056603774,
      "grad_norm": 7.811851501464844,
      "learning_rate": 4.80247641509434e-06,
      "loss": 0.581,
      "num_input_tokens_seen": 1064896,
      "step": 1630
    },
    {
      "epoch": 0.9640330188679245,
      "grad_norm": 3.367180585861206,
      "learning_rate": 4.817216981132076e-06,
      "loss": 0.6342,
      "num_input_tokens_seen": 1067744,
      "step": 1635
    },
    {
      "epoch": 0.9669811320754716,
      "grad_norm": 2.3358185291290283,
      "learning_rate": 4.831957547169811e-06,
      "loss": 0.5076,
      "num_input_tokens_seen": 1070944,
      "step": 1640
    },
    {
      "epoch": 0.9699292452830188,
      "grad_norm": 5.156381607055664,
      "learning_rate": 4.8466981132075476e-06,
      "loss": 0.5585,
      "num_input_tokens_seen": 1073888,
      "step": 1645
    },
    {
      "epoch": 0.972877358490566,
      "grad_norm": 9.319182395935059,
      "learning_rate": 4.861438679245283e-06,
      "loss": 0.8427,
      "num_input_tokens_seen": 1078592,
      "step": 1650
    },
    {
      "epoch": 0.9758254716981132,
      "grad_norm": 4.825729846954346,
      "learning_rate": 4.876179245283019e-06,
      "loss": 0.5229,
      "num_input_tokens_seen": 1081920,
      "step": 1655
    },
    {
      "epoch": 0.9787735849056604,
      "grad_norm": 5.914137363433838,
      "learning_rate": 4.890919811320755e-06,
      "loss": 0.6266,
      "num_input_tokens_seen": 1084800,
      "step": 1660
    },
    {
      "epoch": 0.9817216981132075,
      "grad_norm": 4.3888139724731445,
      "learning_rate": 4.905660377358491e-06,
      "loss": 0.5046,
      "num_input_tokens_seen": 1088832,
      "step": 1665
    },
    {
      "epoch": 0.9846698113207547,
      "grad_norm": 3.9114036560058594,
      "learning_rate": 4.920400943396227e-06,
      "loss": 0.8285,
      "num_input_tokens_seen": 1092608,
      "step": 1670
    },
    {
      "epoch": 0.9876179245283019,
      "grad_norm": 3.129378318786621,
      "learning_rate": 4.935141509433963e-06,
      "loss": 0.5754,
      "num_input_tokens_seen": 1095776,
      "step": 1675
    },
    {
      "epoch": 0.9905660377358491,
      "grad_norm": 1.8735584020614624,
      "learning_rate": 4.949882075471698e-06,
      "loss": 0.4921,
      "num_input_tokens_seen": 1098816,
      "step": 1680
    },
    {
      "epoch": 0.9935141509433962,
      "grad_norm": 2.893059253692627,
      "learning_rate": 4.9646226415094346e-06,
      "loss": 0.5883,
      "num_input_tokens_seen": 1101664,
      "step": 1685
    },
    {
      "epoch": 0.9964622641509434,
      "grad_norm": 4.612443923950195,
      "learning_rate": 4.979363207547171e-06,
      "loss": 0.5248,
      "num_input_tokens_seen": 1105504,
      "step": 1690
    },
    {
      "epoch": 0.9994103773584906,
      "grad_norm": 2.739424467086792,
      "learning_rate": 4.994103773584906e-06,
      "loss": 0.6374,
      "num_input_tokens_seen": 1109088,
      "step": 1695
    },
    {
      "epoch": 1.0023584905660377,
      "grad_norm": 2.9989612102508545,
      "learning_rate": 5.0088443396226414e-06,
      "loss": 0.4929,
      "num_input_tokens_seen": 1111592,
      "step": 1700
    },
    {
      "epoch": 1.005306603773585,
      "grad_norm": 3.7178943157196045,
      "learning_rate": 5.023584905660378e-06,
      "loss": 0.6107,
      "num_input_tokens_seen": 1115016,
      "step": 1705
    },
    {
      "epoch": 1.008254716981132,
      "grad_norm": 6.758875370025635,
      "learning_rate": 5.038325471698113e-06,
      "loss": 0.5915,
      "num_input_tokens_seen": 1118248,
      "step": 1710
    },
    {
      "epoch": 1.0112028301886793,
      "grad_norm": 12.016427993774414,
      "learning_rate": 5.05306603773585e-06,
      "loss": 0.5639,
      "num_input_tokens_seen": 1122312,
      "step": 1715
    },
    {
      "epoch": 1.0141509433962264,
      "grad_norm": 3.3121728897094727,
      "learning_rate": 5.067806603773585e-06,
      "loss": 0.5197,
      "num_input_tokens_seen": 1125736,
      "step": 1720
    },
    {
      "epoch": 1.0170990566037736,
      "grad_norm": 6.567498683929443,
      "learning_rate": 5.0825471698113216e-06,
      "loss": 0.4926,
      "num_input_tokens_seen": 1128040,
      "step": 1725
    },
    {
      "epoch": 1.0200471698113207,
      "grad_norm": 3.436720609664917,
      "learning_rate": 5.097287735849057e-06,
      "loss": 0.4927,
      "num_input_tokens_seen": 1131176,
      "step": 1730
    },
    {
      "epoch": 1.022995283018868,
      "grad_norm": 2.521578550338745,
      "learning_rate": 5.112028301886793e-06,
      "loss": 0.5068,
      "num_input_tokens_seen": 1134504,
      "step": 1735
    },
    {
      "epoch": 1.025943396226415,
      "grad_norm": 3.6427536010742188,
      "learning_rate": 5.1267688679245284e-06,
      "loss": 0.4646,
      "num_input_tokens_seen": 1137224,
      "step": 1740
    },
    {
      "epoch": 1.0288915094339623,
      "grad_norm": 5.974146366119385,
      "learning_rate": 5.1415094339622655e-06,
      "loss": 0.5454,
      "num_input_tokens_seen": 1141160,
      "step": 1745
    },
    {
      "epoch": 1.0318396226415094,
      "grad_norm": 2.740487813949585,
      "learning_rate": 5.156250000000001e-06,
      "loss": 0.4068,
      "num_input_tokens_seen": 1144840,
      "step": 1750
    },
    {
      "epoch": 1.0347877358490567,
      "grad_norm": 11.60596752166748,
      "learning_rate": 5.170990566037736e-06,
      "loss": 0.7379,
      "num_input_tokens_seen": 1147688,
      "step": 1755
    },
    {
      "epoch": 1.0377358490566038,
      "grad_norm": 3.635998010635376,
      "learning_rate": 5.185731132075472e-06,
      "loss": 0.6728,
      "num_input_tokens_seen": 1151144,
      "step": 1760
    },
    {
      "epoch": 1.040683962264151,
      "grad_norm": 4.0117034912109375,
      "learning_rate": 5.200471698113208e-06,
      "loss": 0.5028,
      "num_input_tokens_seen": 1153576,
      "step": 1765
    },
    {
      "epoch": 1.0436320754716981,
      "grad_norm": 5.33103084564209,
      "learning_rate": 5.215212264150944e-06,
      "loss": 0.606,
      "num_input_tokens_seen": 1156520,
      "step": 1770
    },
    {
      "epoch": 1.0465801886792452,
      "grad_norm": 6.539937973022461,
      "learning_rate": 5.229952830188679e-06,
      "loss": 0.727,
      "num_input_tokens_seen": 1159112,
      "step": 1775
    },
    {
      "epoch": 1.0495283018867925,
      "grad_norm": 4.715179920196533,
      "learning_rate": 5.2446933962264154e-06,
      "loss": 0.6011,
      "num_input_tokens_seen": 1162696,
      "step": 1780
    },
    {
      "epoch": 1.0524764150943395,
      "grad_norm": 3.8241524696350098,
      "learning_rate": 5.259433962264151e-06,
      "loss": 0.4691,
      "num_input_tokens_seen": 1166216,
      "step": 1785
    },
    {
      "epoch": 1.0554245283018868,
      "grad_norm": 3.14703631401062,
      "learning_rate": 5.274174528301888e-06,
      "loss": 0.6155,
      "num_input_tokens_seen": 1169704,
      "step": 1790
    },
    {
      "epoch": 1.0583726415094339,
      "grad_norm": 4.975287437438965,
      "learning_rate": 5.288915094339623e-06,
      "loss": 0.5914,
      "num_input_tokens_seen": 1172968,
      "step": 1795
    },
    {
      "epoch": 1.0613207547169812,
      "grad_norm": 4.610471248626709,
      "learning_rate": 5.303655660377359e-06,
      "loss": 0.6862,
      "num_input_tokens_seen": 1176872,
      "step": 1800
    },
    {
      "epoch": 1.0642688679245282,
      "grad_norm": 6.072854042053223,
      "learning_rate": 5.318396226415095e-06,
      "loss": 0.4426,
      "num_input_tokens_seen": 1179304,
      "step": 1805
    },
    {
      "epoch": 1.0672169811320755,
      "grad_norm": 3.9580442905426025,
      "learning_rate": 5.333136792452831e-06,
      "loss": 0.4884,
      "num_input_tokens_seen": 1182408,
      "step": 1810
    },
    {
      "epoch": 1.0701650943396226,
      "grad_norm": 5.04810905456543,
      "learning_rate": 5.347877358490566e-06,
      "loss": 0.4798,
      "num_input_tokens_seen": 1185384,
      "step": 1815
    },
    {
      "epoch": 1.0731132075471699,
      "grad_norm": 3.018094778060913,
      "learning_rate": 5.362617924528302e-06,
      "loss": 0.6003,
      "num_input_tokens_seen": 1188392,
      "step": 1820
    },
    {
      "epoch": 1.076061320754717,
      "grad_norm": 4.425589561462402,
      "learning_rate": 5.377358490566038e-06,
      "loss": 0.5613,
      "num_input_tokens_seen": 1192264,
      "step": 1825
    },
    {
      "epoch": 1.0790094339622642,
      "grad_norm": 3.30791974067688,
      "learning_rate": 5.392099056603775e-06,
      "loss": 0.5039,
      "num_input_tokens_seen": 1195592,
      "step": 1830
    },
    {
      "epoch": 1.0819575471698113,
      "grad_norm": 5.349789142608643,
      "learning_rate": 5.40683962264151e-06,
      "loss": 0.7268,
      "num_input_tokens_seen": 1198216,
      "step": 1835
    },
    {
      "epoch": 1.0849056603773586,
      "grad_norm": 4.947377681732178,
      "learning_rate": 5.4215801886792455e-06,
      "loss": 0.491,
      "num_input_tokens_seen": 1201224,
      "step": 1840
    },
    {
      "epoch": 1.0878537735849056,
      "grad_norm": 3.6661951541900635,
      "learning_rate": 5.436320754716982e-06,
      "loss": 0.5731,
      "num_input_tokens_seen": 1204232,
      "step": 1845
    },
    {
      "epoch": 1.0908018867924527,
      "grad_norm": 5.3298258781433105,
      "learning_rate": 5.451061320754717e-06,
      "loss": 0.611,
      "num_input_tokens_seen": 1207624,
      "step": 1850
    },
    {
      "epoch": 1.09375,
      "grad_norm": 5.922006130218506,
      "learning_rate": 5.465801886792453e-06,
      "loss": 0.5949,
      "num_input_tokens_seen": 1210664,
      "step": 1855
    },
    {
      "epoch": 1.0966981132075473,
      "grad_norm": 2.9588658809661865,
      "learning_rate": 5.4805424528301886e-06,
      "loss": 0.71,
      "num_input_tokens_seen": 1214024,
      "step": 1860
    },
    {
      "epoch": 1.0996462264150944,
      "grad_norm": 2.568260431289673,
      "learning_rate": 5.495283018867925e-06,
      "loss": 0.3621,
      "num_input_tokens_seen": 1217192,
      "step": 1865
    },
    {
      "epoch": 1.1025943396226414,
      "grad_norm": 2.226083517074585,
      "learning_rate": 5.51002358490566e-06,
      "loss": 0.5362,
      "num_input_tokens_seen": 1219912,
      "step": 1870
    },
    {
      "epoch": 1.1055424528301887,
      "grad_norm": 4.561788082122803,
      "learning_rate": 5.524764150943397e-06,
      "loss": 0.5,
      "num_input_tokens_seen": 1222184,
      "step": 1875
    },
    {
      "epoch": 1.1084905660377358,
      "grad_norm": 7.513849258422852,
      "learning_rate": 5.5395047169811325e-06,
      "loss": 0.7646,
      "num_input_tokens_seen": 1224680,
      "step": 1880
    },
    {
      "epoch": 1.111438679245283,
      "grad_norm": 4.5633978843688965,
      "learning_rate": 5.554245283018869e-06,
      "loss": 0.7365,
      "num_input_tokens_seen": 1228104,
      "step": 1885
    },
    {
      "epoch": 1.1143867924528301,
      "grad_norm": 4.611073970794678,
      "learning_rate": 5.568985849056604e-06,
      "loss": 0.5355,
      "num_input_tokens_seen": 1231528,
      "step": 1890
    },
    {
      "epoch": 1.1173349056603774,
      "grad_norm": 6.0877604484558105,
      "learning_rate": 5.58372641509434e-06,
      "loss": 0.494,
      "num_input_tokens_seen": 1234728,
      "step": 1895
    },
    {
      "epoch": 1.1202830188679245,
      "grad_norm": 3.8698697090148926,
      "learning_rate": 5.5984669811320755e-06,
      "loss": 0.5302,
      "num_input_tokens_seen": 1237800,
      "step": 1900
    },
    {
      "epoch": 1.1232311320754718,
      "grad_norm": 15.551494598388672,
      "learning_rate": 5.613207547169813e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 1240744,
      "step": 1905
    },
    {
      "epoch": 1.1261792452830188,
      "grad_norm": 2.462249279022217,
      "learning_rate": 5.627948113207548e-06,
      "loss": 0.5518,
      "num_input_tokens_seen": 1245608,
      "step": 1910
    },
    {
      "epoch": 1.1291273584905661,
      "grad_norm": 3.139821767807007,
      "learning_rate": 5.642688679245284e-06,
      "loss": 0.4392,
      "num_input_tokens_seen": 1248456,
      "step": 1915
    },
    {
      "epoch": 1.1320754716981132,
      "grad_norm": 2.583340644836426,
      "learning_rate": 5.6574292452830195e-06,
      "loss": 0.5152,
      "num_input_tokens_seen": 1252232,
      "step": 1920
    },
    {
      "epoch": 1.1350235849056605,
      "grad_norm": 4.024645805358887,
      "learning_rate": 5.672169811320756e-06,
      "loss": 0.5063,
      "num_input_tokens_seen": 1255048,
      "step": 1925
    },
    {
      "epoch": 1.1379716981132075,
      "grad_norm": 3.1879117488861084,
      "learning_rate": 5.686910377358491e-06,
      "loss": 0.5155,
      "num_input_tokens_seen": 1258664,
      "step": 1930
    },
    {
      "epoch": 1.1409198113207548,
      "grad_norm": 3.5429654121398926,
      "learning_rate": 5.701650943396226e-06,
      "loss": 0.6073,
      "num_input_tokens_seen": 1261576,
      "step": 1935
    },
    {
      "epoch": 1.1438679245283019,
      "grad_norm": 2.05853009223938,
      "learning_rate": 5.7163915094339625e-06,
      "loss": 0.4653,
      "num_input_tokens_seen": 1264392,
      "step": 1940
    },
    {
      "epoch": 1.146816037735849,
      "grad_norm": 2.876725435256958,
      "learning_rate": 5.731132075471698e-06,
      "loss": 0.5307,
      "num_input_tokens_seen": 1267720,
      "step": 1945
    },
    {
      "epoch": 1.1497641509433962,
      "grad_norm": 3.2926602363586426,
      "learning_rate": 5.745872641509435e-06,
      "loss": 0.4903,
      "num_input_tokens_seen": 1270600,
      "step": 1950
    },
    {
      "epoch": 1.1527122641509433,
      "grad_norm": 4.320738315582275,
      "learning_rate": 5.76061320754717e-06,
      "loss": 0.7938,
      "num_input_tokens_seen": 1274312,
      "step": 1955
    },
    {
      "epoch": 1.1556603773584906,
      "grad_norm": 3.644739866256714,
      "learning_rate": 5.7753537735849065e-06,
      "loss": 0.577,
      "num_input_tokens_seen": 1278088,
      "step": 1960
    },
    {
      "epoch": 1.1586084905660377,
      "grad_norm": 2.6592800617218018,
      "learning_rate": 5.790094339622642e-06,
      "loss": 0.5244,
      "num_input_tokens_seen": 1281160,
      "step": 1965
    },
    {
      "epoch": 1.161556603773585,
      "grad_norm": 2.3588733673095703,
      "learning_rate": 5.804834905660378e-06,
      "loss": 0.5503,
      "num_input_tokens_seen": 1284840,
      "step": 1970
    },
    {
      "epoch": 1.164504716981132,
      "grad_norm": 3.7015442848205566,
      "learning_rate": 5.819575471698113e-06,
      "loss": 0.5648,
      "num_input_tokens_seen": 1288232,
      "step": 1975
    },
    {
      "epoch": 1.1674528301886793,
      "grad_norm": 3.35829758644104,
      "learning_rate": 5.8343160377358495e-06,
      "loss": 0.6383,
      "num_input_tokens_seen": 1291464,
      "step": 1980
    },
    {
      "epoch": 1.1704009433962264,
      "grad_norm": 7.999370574951172,
      "learning_rate": 5.849056603773585e-06,
      "loss": 0.5669,
      "num_input_tokens_seen": 1295496,
      "step": 1985
    },
    {
      "epoch": 1.1733490566037736,
      "grad_norm": 9.482475280761719,
      "learning_rate": 5.863797169811322e-06,
      "loss": 0.5102,
      "num_input_tokens_seen": 1298248,
      "step": 1990
    },
    {
      "epoch": 1.1762971698113207,
      "grad_norm": 16.183162689208984,
      "learning_rate": 5.878537735849057e-06,
      "loss": 0.555,
      "num_input_tokens_seen": 1301416,
      "step": 1995
    },
    {
      "epoch": 1.179245283018868,
      "grad_norm": 4.1765522956848145,
      "learning_rate": 5.8932783018867934e-06,
      "loss": 0.5001,
      "num_input_tokens_seen": 1304168,
      "step": 2000
    },
    {
      "epoch": 1.182193396226415,
      "grad_norm": 2.824888229370117,
      "learning_rate": 5.908018867924529e-06,
      "loss": 0.4784,
      "num_input_tokens_seen": 1309928,
      "step": 2005
    },
    {
      "epoch": 1.1851415094339623,
      "grad_norm": 3.2439990043640137,
      "learning_rate": 5.922759433962265e-06,
      "loss": 0.3776,
      "num_input_tokens_seen": 1312424,
      "step": 2010
    },
    {
      "epoch": 1.1880896226415094,
      "grad_norm": 4.059011936187744,
      "learning_rate": 5.9375e-06,
      "loss": 0.6335,
      "num_input_tokens_seen": 1315272,
      "step": 2015
    },
    {
      "epoch": 1.1910377358490567,
      "grad_norm": 3.7251877784729004,
      "learning_rate": 5.952240566037736e-06,
      "loss": 0.4161,
      "num_input_tokens_seen": 1318632,
      "step": 2020
    },
    {
      "epoch": 1.1939858490566038,
      "grad_norm": 4.338551044464111,
      "learning_rate": 5.966981132075472e-06,
      "loss": 0.6086,
      "num_input_tokens_seen": 1322504,
      "step": 2025
    },
    {
      "epoch": 1.196933962264151,
      "grad_norm": 23.651634216308594,
      "learning_rate": 5.981721698113207e-06,
      "loss": 0.6115,
      "num_input_tokens_seen": 1326184,
      "step": 2030
    },
    {
      "epoch": 1.1998820754716981,
      "grad_norm": 4.28742790222168,
      "learning_rate": 5.996462264150944e-06,
      "loss": 0.5803,
      "num_input_tokens_seen": 1329128,
      "step": 2035
    },
    {
      "epoch": 1.2028301886792452,
      "grad_norm": 4.192485809326172,
      "learning_rate": 6.01120283018868e-06,
      "loss": 0.3544,
      "num_input_tokens_seen": 1334568,
      "step": 2040
    },
    {
      "epoch": 1.2057783018867925,
      "grad_norm": 3.5532679557800293,
      "learning_rate": 6.025943396226416e-06,
      "loss": 0.6456,
      "num_input_tokens_seen": 1338024,
      "step": 2045
    },
    {
      "epoch": 1.2087264150943395,
      "grad_norm": 2.531177520751953,
      "learning_rate": 6.040683962264151e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 1341672,
      "step": 2050
    },
    {
      "epoch": 1.2116745283018868,
      "grad_norm": 8.4788236618042,
      "learning_rate": 6.055424528301887e-06,
      "loss": 0.5664,
      "num_input_tokens_seen": 1344072,
      "step": 2055
    },
    {
      "epoch": 1.2146226415094339,
      "grad_norm": 3.6985864639282227,
      "learning_rate": 6.070165094339623e-06,
      "loss": 0.5289,
      "num_input_tokens_seen": 1349256,
      "step": 2060
    },
    {
      "epoch": 1.2175707547169812,
      "grad_norm": 18.42817497253418,
      "learning_rate": 6.08490566037736e-06,
      "loss": 0.5231,
      "num_input_tokens_seen": 1351464,
      "step": 2065
    },
    {
      "epoch": 1.2205188679245282,
      "grad_norm": 2.877884864807129,
      "learning_rate": 6.099646226415095e-06,
      "loss": 0.6018,
      "num_input_tokens_seen": 1354664,
      "step": 2070
    },
    {
      "epoch": 1.2234669811320755,
      "grad_norm": 4.898430824279785,
      "learning_rate": 6.114386792452831e-06,
      "loss": 0.5997,
      "num_input_tokens_seen": 1358856,
      "step": 2075
    },
    {
      "epoch": 1.2264150943396226,
      "grad_norm": 4.283647060394287,
      "learning_rate": 6.129127358490567e-06,
      "loss": 0.4314,
      "num_input_tokens_seen": 1361320,
      "step": 2080
    },
    {
      "epoch": 1.2293632075471699,
      "grad_norm": 4.192919731140137,
      "learning_rate": 6.143867924528303e-06,
      "loss": 0.6109,
      "num_input_tokens_seen": 1364808,
      "step": 2085
    },
    {
      "epoch": 1.232311320754717,
      "grad_norm": 4.79982852935791,
      "learning_rate": 6.158608490566038e-06,
      "loss": 0.6597,
      "num_input_tokens_seen": 1367912,
      "step": 2090
    },
    {
      "epoch": 1.2352594339622642,
      "grad_norm": 3.2680094242095947,
      "learning_rate": 6.173349056603774e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 1371432,
      "step": 2095
    },
    {
      "epoch": 1.2382075471698113,
      "grad_norm": 4.653439998626709,
      "learning_rate": 6.18808962264151e-06,
      "loss": 0.7139,
      "num_input_tokens_seen": 1373928,
      "step": 2100
    },
    {
      "epoch": 1.2411556603773586,
      "grad_norm": 6.5873517990112305,
      "learning_rate": 6.202830188679245e-06,
      "loss": 0.5577,
      "num_input_tokens_seen": 1377800,
      "step": 2105
    },
    {
      "epoch": 1.2441037735849056,
      "grad_norm": 3.0361716747283936,
      "learning_rate": 6.217570754716982e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 1381064,
      "step": 2110
    },
    {
      "epoch": 1.2470518867924527,
      "grad_norm": 3.097989559173584,
      "learning_rate": 6.232311320754717e-06,
      "loss": 0.6226,
      "num_input_tokens_seen": 1384232,
      "step": 2115
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.255221366882324,
      "learning_rate": 6.2470518867924536e-06,
      "loss": 0.518,
      "num_input_tokens_seen": 1387240,
      "step": 2120
    },
    {
      "epoch": 1.2529481132075473,
      "grad_norm": 3.4855329990386963,
      "learning_rate": 6.261792452830189e-06,
      "loss": 0.6236,
      "num_input_tokens_seen": 1390248,
      "step": 2125
    },
    {
      "epoch": 1.2558962264150944,
      "grad_norm": 2.9747116565704346,
      "learning_rate": 6.276533018867925e-06,
      "loss": 0.4841,
      "num_input_tokens_seen": 1393768,
      "step": 2130
    },
    {
      "epoch": 1.2588443396226414,
      "grad_norm": 2.1630825996398926,
      "learning_rate": 6.2912735849056604e-06,
      "loss": 0.5816,
      "num_input_tokens_seen": 1396808,
      "step": 2135
    },
    {
      "epoch": 1.2617924528301887,
      "grad_norm": 2.1857645511627197,
      "learning_rate": 6.306014150943397e-06,
      "loss": 0.5023,
      "num_input_tokens_seen": 1399656,
      "step": 2140
    },
    {
      "epoch": 1.2647405660377358,
      "grad_norm": 3.070129156112671,
      "learning_rate": 6.320754716981132e-06,
      "loss": 0.506,
      "num_input_tokens_seen": 1402792,
      "step": 2145
    },
    {
      "epoch": 1.267688679245283,
      "grad_norm": 3.632359266281128,
      "learning_rate": 6.335495283018869e-06,
      "loss": 0.5521,
      "num_input_tokens_seen": 1405160,
      "step": 2150
    },
    {
      "epoch": 1.2706367924528301,
      "grad_norm": 5.355973243713379,
      "learning_rate": 6.350235849056604e-06,
      "loss": 0.5534,
      "num_input_tokens_seen": 1409512,
      "step": 2155
    },
    {
      "epoch": 1.2735849056603774,
      "grad_norm": 3.930882692337036,
      "learning_rate": 6.3649764150943406e-06,
      "loss": 0.4787,
      "num_input_tokens_seen": 1413672,
      "step": 2160
    },
    {
      "epoch": 1.2765330188679245,
      "grad_norm": 5.208805084228516,
      "learning_rate": 6.379716981132076e-06,
      "loss": 0.6297,
      "num_input_tokens_seen": 1416968,
      "step": 2165
    },
    {
      "epoch": 1.2794811320754718,
      "grad_norm": 9.071327209472656,
      "learning_rate": 6.394457547169812e-06,
      "loss": 0.5866,
      "num_input_tokens_seen": 1419208,
      "step": 2170
    },
    {
      "epoch": 1.2824292452830188,
      "grad_norm": 4.71073579788208,
      "learning_rate": 6.4091981132075474e-06,
      "loss": 0.5419,
      "num_input_tokens_seen": 1422728,
      "step": 2175
    },
    {
      "epoch": 1.2853773584905661,
      "grad_norm": 2.9785704612731934,
      "learning_rate": 6.423938679245284e-06,
      "loss": 0.5076,
      "num_input_tokens_seen": 1426568,
      "step": 2180
    },
    {
      "epoch": 1.2883254716981132,
      "grad_norm": 3.893916130065918,
      "learning_rate": 6.438679245283019e-06,
      "loss": 0.6449,
      "num_input_tokens_seen": 1430088,
      "step": 2185
    },
    {
      "epoch": 1.2912735849056602,
      "grad_norm": 4.394443035125732,
      "learning_rate": 6.453419811320756e-06,
      "loss": 0.6354,
      "num_input_tokens_seen": 1432424,
      "step": 2190
    },
    {
      "epoch": 1.2942216981132075,
      "grad_norm": 2.975356340408325,
      "learning_rate": 6.468160377358491e-06,
      "loss": 0.5598,
      "num_input_tokens_seen": 1436488,
      "step": 2195
    },
    {
      "epoch": 1.2971698113207548,
      "grad_norm": 12.325146675109863,
      "learning_rate": 6.482900943396227e-06,
      "loss": 0.4876,
      "num_input_tokens_seen": 1439848,
      "step": 2200
    },
    {
      "epoch": 1.3001179245283019,
      "grad_norm": 3.521345615386963,
      "learning_rate": 6.497641509433963e-06,
      "loss": 0.463,
      "num_input_tokens_seen": 1443176,
      "step": 2205
    },
    {
      "epoch": 1.303066037735849,
      "grad_norm": 4.2001872062683105,
      "learning_rate": 6.512382075471698e-06,
      "loss": 0.5403,
      "num_input_tokens_seen": 1451400,
      "step": 2210
    },
    {
      "epoch": 1.3060141509433962,
      "grad_norm": 10.69773006439209,
      "learning_rate": 6.5271226415094344e-06,
      "loss": 0.7305,
      "num_input_tokens_seen": 1454280,
      "step": 2215
    },
    {
      "epoch": 1.3089622641509435,
      "grad_norm": 3.4736387729644775,
      "learning_rate": 6.54186320754717e-06,
      "loss": 0.5723,
      "num_input_tokens_seen": 1456904,
      "step": 2220
    },
    {
      "epoch": 1.3119103773584906,
      "grad_norm": 3.9165291786193848,
      "learning_rate": 6.556603773584907e-06,
      "loss": 0.4412,
      "num_input_tokens_seen": 1460712,
      "step": 2225
    },
    {
      "epoch": 1.3148584905660377,
      "grad_norm": 3.57822847366333,
      "learning_rate": 6.571344339622641e-06,
      "loss": 0.525,
      "num_input_tokens_seen": 1464616,
      "step": 2230
    },
    {
      "epoch": 1.317806603773585,
      "grad_norm": 2.1331441402435303,
      "learning_rate": 6.586084905660378e-06,
      "loss": 0.4586,
      "num_input_tokens_seen": 1468584,
      "step": 2235
    },
    {
      "epoch": 1.320754716981132,
      "grad_norm": 2.6206865310668945,
      "learning_rate": 6.600825471698114e-06,
      "loss": 0.4612,
      "num_input_tokens_seen": 1472072,
      "step": 2240
    },
    {
      "epoch": 1.3237028301886793,
      "grad_norm": 3.5069103240966797,
      "learning_rate": 6.61556603773585e-06,
      "loss": 0.5793,
      "num_input_tokens_seen": 1477928,
      "step": 2245
    },
    {
      "epoch": 1.3266509433962264,
      "grad_norm": 2.6483612060546875,
      "learning_rate": 6.630306603773585e-06,
      "loss": 0.5162,
      "num_input_tokens_seen": 1481032,
      "step": 2250
    },
    {
      "epoch": 1.3295990566037736,
      "grad_norm": 3.1596710681915283,
      "learning_rate": 6.645047169811321e-06,
      "loss": 0.7315,
      "num_input_tokens_seen": 1484232,
      "step": 2255
    },
    {
      "epoch": 1.3325471698113207,
      "grad_norm": 5.666794300079346,
      "learning_rate": 6.659787735849057e-06,
      "loss": 0.5708,
      "num_input_tokens_seen": 1487592,
      "step": 2260
    },
    {
      "epoch": 1.335495283018868,
      "grad_norm": 3.1301307678222656,
      "learning_rate": 6.674528301886794e-06,
      "loss": 0.4436,
      "num_input_tokens_seen": 1490344,
      "step": 2265
    },
    {
      "epoch": 1.338443396226415,
      "grad_norm": 3.248842239379883,
      "learning_rate": 6.689268867924529e-06,
      "loss": 0.3889,
      "num_input_tokens_seen": 1493288,
      "step": 2270
    },
    {
      "epoch": 1.3413915094339623,
      "grad_norm": 5.361477851867676,
      "learning_rate": 6.704009433962265e-06,
      "loss": 0.608,
      "num_input_tokens_seen": 1496008,
      "step": 2275
    },
    {
      "epoch": 1.3443396226415094,
      "grad_norm": 2.7907636165618896,
      "learning_rate": 6.718750000000001e-06,
      "loss": 0.6057,
      "num_input_tokens_seen": 1498856,
      "step": 2280
    },
    {
      "epoch": 1.3472877358490565,
      "grad_norm": 2.6450541019439697,
      "learning_rate": 6.733490566037736e-06,
      "loss": 0.5424,
      "num_input_tokens_seen": 1501800,
      "step": 2285
    },
    {
      "epoch": 1.3502358490566038,
      "grad_norm": 3.6860363483428955,
      "learning_rate": 6.748231132075472e-06,
      "loss": 0.5354,
      "num_input_tokens_seen": 1504872,
      "step": 2290
    },
    {
      "epoch": 1.353183962264151,
      "grad_norm": 2.0752086639404297,
      "learning_rate": 6.7629716981132076e-06,
      "loss": 0.5135,
      "num_input_tokens_seen": 1508520,
      "step": 2295
    },
    {
      "epoch": 1.3561320754716981,
      "grad_norm": 3.1735289096832275,
      "learning_rate": 6.777712264150944e-06,
      "loss": 0.4051,
      "num_input_tokens_seen": 1511528,
      "step": 2300
    },
    {
      "epoch": 1.3590801886792452,
      "grad_norm": 4.095285415649414,
      "learning_rate": 6.792452830188679e-06,
      "loss": 0.5546,
      "num_input_tokens_seen": 1514376,
      "step": 2305
    },
    {
      "epoch": 1.3620283018867925,
      "grad_norm": 4.507347106933594,
      "learning_rate": 6.807193396226416e-06,
      "loss": 0.4569,
      "num_input_tokens_seen": 1517896,
      "step": 2310
    },
    {
      "epoch": 1.3649764150943398,
      "grad_norm": 2.3421244621276855,
      "learning_rate": 6.8219339622641515e-06,
      "loss": 0.419,
      "num_input_tokens_seen": 1521224,
      "step": 2315
    },
    {
      "epoch": 1.3679245283018868,
      "grad_norm": 4.5384111404418945,
      "learning_rate": 6.836674528301888e-06,
      "loss": 0.5165,
      "num_input_tokens_seen": 1523848,
      "step": 2320
    },
    {
      "epoch": 1.3708726415094339,
      "grad_norm": 2.6614108085632324,
      "learning_rate": 6.851415094339623e-06,
      "loss": 0.3965,
      "num_input_tokens_seen": 1527912,
      "step": 2325
    },
    {
      "epoch": 1.3738207547169812,
      "grad_norm": 8.495092391967773,
      "learning_rate": 6.866155660377359e-06,
      "loss": 0.5324,
      "num_input_tokens_seen": 1530920,
      "step": 2330
    },
    {
      "epoch": 1.3767688679245282,
      "grad_norm": 8.671310424804688,
      "learning_rate": 6.8808962264150946e-06,
      "loss": 0.5339,
      "num_input_tokens_seen": 1534760,
      "step": 2335
    },
    {
      "epoch": 1.3797169811320755,
      "grad_norm": 5.221708297729492,
      "learning_rate": 6.895636792452831e-06,
      "loss": 0.6572,
      "num_input_tokens_seen": 1538920,
      "step": 2340
    },
    {
      "epoch": 1.3826650943396226,
      "grad_norm": 2.072862148284912,
      "learning_rate": 6.910377358490566e-06,
      "loss": 0.5419,
      "num_input_tokens_seen": 1541928,
      "step": 2345
    },
    {
      "epoch": 1.3856132075471699,
      "grad_norm": 2.0162837505340576,
      "learning_rate": 6.925117924528303e-06,
      "loss": 0.6152,
      "num_input_tokens_seen": 1545672,
      "step": 2350
    },
    {
      "epoch": 1.388561320754717,
      "grad_norm": 3.287925958633423,
      "learning_rate": 6.9398584905660385e-06,
      "loss": 0.4749,
      "num_input_tokens_seen": 1549256,
      "step": 2355
    },
    {
      "epoch": 1.3915094339622642,
      "grad_norm": 2.5606045722961426,
      "learning_rate": 6.954599056603775e-06,
      "loss": 0.3977,
      "num_input_tokens_seen": 1551912,
      "step": 2360
    },
    {
      "epoch": 1.3944575471698113,
      "grad_norm": 3.1936581134796143,
      "learning_rate": 6.96933962264151e-06,
      "loss": 0.6615,
      "num_input_tokens_seen": 1555848,
      "step": 2365
    },
    {
      "epoch": 1.3974056603773586,
      "grad_norm": 3.9231581687927246,
      "learning_rate": 6.984080188679245e-06,
      "loss": 0.7967,
      "num_input_tokens_seen": 1559080,
      "step": 2370
    },
    {
      "epoch": 1.4003537735849056,
      "grad_norm": 7.183620929718018,
      "learning_rate": 6.9988207547169815e-06,
      "loss": 0.5446,
      "num_input_tokens_seen": 1562216,
      "step": 2375
    },
    {
      "epoch": 1.4033018867924527,
      "grad_norm": 2.132411003112793,
      "learning_rate": 7.013561320754717e-06,
      "loss": 0.5031,
      "num_input_tokens_seen": 1566568,
      "step": 2380
    },
    {
      "epoch": 1.40625,
      "grad_norm": 4.0367937088012695,
      "learning_rate": 7.028301886792454e-06,
      "loss": 0.6425,
      "num_input_tokens_seen": 1569608,
      "step": 2385
    },
    {
      "epoch": 1.4091981132075473,
      "grad_norm": 2.6854450702667236,
      "learning_rate": 7.043042452830188e-06,
      "loss": 0.5996,
      "num_input_tokens_seen": 1572968,
      "step": 2390
    },
    {
      "epoch": 1.4121462264150944,
      "grad_norm": 5.15828800201416,
      "learning_rate": 7.0577830188679255e-06,
      "loss": 0.6181,
      "num_input_tokens_seen": 1575848,
      "step": 2395
    },
    {
      "epoch": 1.4150943396226414,
      "grad_norm": 2.7500927448272705,
      "learning_rate": 7.072523584905661e-06,
      "loss": 0.4485,
      "num_input_tokens_seen": 1579048,
      "step": 2400
    },
    {
      "epoch": 1.4180424528301887,
      "grad_norm": 7.795233726501465,
      "learning_rate": 7.087264150943397e-06,
      "loss": 0.6,
      "num_input_tokens_seen": 1582408,
      "step": 2405
    },
    {
      "epoch": 1.4209905660377358,
      "grad_norm": 4.185930252075195,
      "learning_rate": 7.102004716981132e-06,
      "loss": 0.4986,
      "num_input_tokens_seen": 1585320,
      "step": 2410
    },
    {
      "epoch": 1.423938679245283,
      "grad_norm": 4.134632110595703,
      "learning_rate": 7.1167452830188685e-06,
      "loss": 0.545,
      "num_input_tokens_seen": 1588584,
      "step": 2415
    },
    {
      "epoch": 1.4268867924528301,
      "grad_norm": 3.4842369556427,
      "learning_rate": 7.131485849056604e-06,
      "loss": 0.5655,
      "num_input_tokens_seen": 1591240,
      "step": 2420
    },
    {
      "epoch": 1.4298349056603774,
      "grad_norm": 2.5747876167297363,
      "learning_rate": 7.146226415094341e-06,
      "loss": 0.4412,
      "num_input_tokens_seen": 1595048,
      "step": 2425
    },
    {
      "epoch": 1.4327830188679245,
      "grad_norm": 2.4336013793945312,
      "learning_rate": 7.160966981132076e-06,
      "loss": 0.5324,
      "num_input_tokens_seen": 1598664,
      "step": 2430
    },
    {
      "epoch": 1.4357311320754718,
      "grad_norm": 6.856317043304443,
      "learning_rate": 7.1757075471698125e-06,
      "loss": 0.5495,
      "num_input_tokens_seen": 1601288,
      "step": 2435
    },
    {
      "epoch": 1.4386792452830188,
      "grad_norm": 2.9326436519622803,
      "learning_rate": 7.190448113207548e-06,
      "loss": 0.5499,
      "num_input_tokens_seen": 1604552,
      "step": 2440
    },
    {
      "epoch": 1.4416273584905661,
      "grad_norm": 3.4927523136138916,
      "learning_rate": 7.205188679245284e-06,
      "loss": 0.5121,
      "num_input_tokens_seen": 1608168,
      "step": 2445
    },
    {
      "epoch": 1.4445754716981132,
      "grad_norm": 3.1840388774871826,
      "learning_rate": 7.219929245283019e-06,
      "loss": 0.542,
      "num_input_tokens_seen": 1611336,
      "step": 2450
    },
    {
      "epoch": 1.4475235849056602,
      "grad_norm": 9.476007461547852,
      "learning_rate": 7.2346698113207555e-06,
      "loss": 0.5779,
      "num_input_tokens_seen": 1614760,
      "step": 2455
    },
    {
      "epoch": 1.4504716981132075,
      "grad_norm": 3.77824068069458,
      "learning_rate": 7.249410377358491e-06,
      "loss": 0.5957,
      "num_input_tokens_seen": 1618376,
      "step": 2460
    },
    {
      "epoch": 1.4534198113207548,
      "grad_norm": 11.594400405883789,
      "learning_rate": 7.264150943396226e-06,
      "loss": 0.5047,
      "num_input_tokens_seen": 1621256,
      "step": 2465
    },
    {
      "epoch": 1.4563679245283019,
      "grad_norm": 1.5943620204925537,
      "learning_rate": 7.278891509433963e-06,
      "loss": 0.4867,
      "num_input_tokens_seen": 1624808,
      "step": 2470
    },
    {
      "epoch": 1.459316037735849,
      "grad_norm": 3.769818067550659,
      "learning_rate": 7.293632075471699e-06,
      "loss": 0.6964,
      "num_input_tokens_seen": 1627688,
      "step": 2475
    },
    {
      "epoch": 1.4622641509433962,
      "grad_norm": 1.8063501119613647,
      "learning_rate": 7.308372641509435e-06,
      "loss": 0.4521,
      "num_input_tokens_seen": 1630216,
      "step": 2480
    },
    {
      "epoch": 1.4652122641509435,
      "grad_norm": 2.5699081420898438,
      "learning_rate": 7.32311320754717e-06,
      "loss": 0.6018,
      "num_input_tokens_seen": 1633800,
      "step": 2485
    },
    {
      "epoch": 1.4681603773584906,
      "grad_norm": 3.077378749847412,
      "learning_rate": 7.337853773584906e-06,
      "loss": 0.4558,
      "num_input_tokens_seen": 1636168,
      "step": 2490
    },
    {
      "epoch": 1.4711084905660377,
      "grad_norm": 3.187596082687378,
      "learning_rate": 7.352594339622642e-06,
      "loss": 0.4537,
      "num_input_tokens_seen": 1638792,
      "step": 2495
    },
    {
      "epoch": 1.474056603773585,
      "grad_norm": 2.2626333236694336,
      "learning_rate": 7.367334905660378e-06,
      "loss": 0.6826,
      "num_input_tokens_seen": 1642088,
      "step": 2500
    },
    {
      "epoch": 1.477004716981132,
      "grad_norm": 2.3543529510498047,
      "learning_rate": 7.382075471698113e-06,
      "loss": 0.4919,
      "num_input_tokens_seen": 1644808,
      "step": 2505
    },
    {
      "epoch": 1.4799528301886793,
      "grad_norm": 4.24591064453125,
      "learning_rate": 7.39681603773585e-06,
      "loss": 0.5272,
      "num_input_tokens_seen": 1647496,
      "step": 2510
    },
    {
      "epoch": 1.4829009433962264,
      "grad_norm": 1.8173989057540894,
      "learning_rate": 7.411556603773586e-06,
      "loss": 0.4371,
      "num_input_tokens_seen": 1650760,
      "step": 2515
    },
    {
      "epoch": 1.4858490566037736,
      "grad_norm": 2.048732042312622,
      "learning_rate": 7.426297169811322e-06,
      "loss": 0.5664,
      "num_input_tokens_seen": 1653864,
      "step": 2520
    },
    {
      "epoch": 1.4887971698113207,
      "grad_norm": 2.793461322784424,
      "learning_rate": 7.441037735849057e-06,
      "loss": 0.5064,
      "num_input_tokens_seen": 1656616,
      "step": 2525
    },
    {
      "epoch": 1.491745283018868,
      "grad_norm": 1.835691213607788,
      "learning_rate": 7.455778301886793e-06,
      "loss": 0.54,
      "num_input_tokens_seen": 1659656,
      "step": 2530
    },
    {
      "epoch": 1.494693396226415,
      "grad_norm": 3.5888102054595947,
      "learning_rate": 7.470518867924529e-06,
      "loss": 0.6386,
      "num_input_tokens_seen": 1662920,
      "step": 2535
    },
    {
      "epoch": 1.4976415094339623,
      "grad_norm": 3.981990098953247,
      "learning_rate": 7.485259433962266e-06,
      "loss": 0.6062,
      "num_input_tokens_seen": 1666472,
      "step": 2540
    },
    {
      "epoch": 1.5005896226415094,
      "grad_norm": 5.066642761230469,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.4959,
      "num_input_tokens_seen": 1670760,
      "step": 2545
    },
    {
      "epoch": 1.5035377358490565,
      "grad_norm": 2.1030285358428955,
      "learning_rate": 7.5147405660377355e-06,
      "loss": 0.5104,
      "num_input_tokens_seen": 1674120,
      "step": 2550
    },
    {
      "epoch": 1.5064858490566038,
      "grad_norm": 2.4119467735290527,
      "learning_rate": 7.5294811320754726e-06,
      "loss": 0.4651,
      "num_input_tokens_seen": 1677960,
      "step": 2555
    },
    {
      "epoch": 1.509433962264151,
      "grad_norm": 2.145543336868286,
      "learning_rate": 7.544221698113208e-06,
      "loss": 0.5482,
      "num_input_tokens_seen": 1681672,
      "step": 2560
    },
    {
      "epoch": 1.5123820754716981,
      "grad_norm": 7.2220025062561035,
      "learning_rate": 7.558962264150944e-06,
      "loss": 0.7174,
      "num_input_tokens_seen": 1685160,
      "step": 2565
    },
    {
      "epoch": 1.5153301886792452,
      "grad_norm": 2.2348978519439697,
      "learning_rate": 7.5737028301886795e-06,
      "loss": 0.4796,
      "num_input_tokens_seen": 1689032,
      "step": 2570
    },
    {
      "epoch": 1.5182783018867925,
      "grad_norm": 2.266817569732666,
      "learning_rate": 7.588443396226416e-06,
      "loss": 0.5485,
      "num_input_tokens_seen": 1691816,
      "step": 2575
    },
    {
      "epoch": 1.5212264150943398,
      "grad_norm": 2.5918333530426025,
      "learning_rate": 7.603183962264151e-06,
      "loss": 0.6361,
      "num_input_tokens_seen": 1695016,
      "step": 2580
    },
    {
      "epoch": 1.5241745283018868,
      "grad_norm": 2.9727864265441895,
      "learning_rate": 7.617924528301888e-06,
      "loss": 0.5028,
      "num_input_tokens_seen": 1697704,
      "step": 2585
    },
    {
      "epoch": 1.5271226415094339,
      "grad_norm": 5.12477445602417,
      "learning_rate": 7.632665094339623e-06,
      "loss": 0.5944,
      "num_input_tokens_seen": 1700232,
      "step": 2590
    },
    {
      "epoch": 1.5300707547169812,
      "grad_norm": 5.668395519256592,
      "learning_rate": 7.64740566037736e-06,
      "loss": 0.5677,
      "num_input_tokens_seen": 1703144,
      "step": 2595
    },
    {
      "epoch": 1.5330188679245285,
      "grad_norm": 2.8164591789245605,
      "learning_rate": 7.662146226415095e-06,
      "loss": 0.4764,
      "num_input_tokens_seen": 1706600,
      "step": 2600
    },
    {
      "epoch": 1.5359669811320755,
      "grad_norm": 2.616431713104248,
      "learning_rate": 7.676886792452832e-06,
      "loss": 0.4812,
      "num_input_tokens_seen": 1709864,
      "step": 2605
    },
    {
      "epoch": 1.5389150943396226,
      "grad_norm": 4.092056751251221,
      "learning_rate": 7.691627358490567e-06,
      "loss": 0.6062,
      "num_input_tokens_seen": 1712776,
      "step": 2610
    },
    {
      "epoch": 1.5418632075471699,
      "grad_norm": 3.0650835037231445,
      "learning_rate": 7.706367924528303e-06,
      "loss": 0.3853,
      "num_input_tokens_seen": 1715688,
      "step": 2615
    },
    {
      "epoch": 1.544811320754717,
      "grad_norm": 2.5430006980895996,
      "learning_rate": 7.721108490566038e-06,
      "loss": 0.5598,
      "num_input_tokens_seen": 1718632,
      "step": 2620
    },
    {
      "epoch": 1.547759433962264,
      "grad_norm": 2.7769598960876465,
      "learning_rate": 7.735849056603775e-06,
      "loss": 0.4801,
      "num_input_tokens_seen": 1721832,
      "step": 2625
    },
    {
      "epoch": 1.5507075471698113,
      "grad_norm": 4.4878106117248535,
      "learning_rate": 7.75058962264151e-06,
      "loss": 0.431,
      "num_input_tokens_seen": 1724744,
      "step": 2630
    },
    {
      "epoch": 1.5536556603773586,
      "grad_norm": 3.1825881004333496,
      "learning_rate": 7.765330188679246e-06,
      "loss": 0.6774,
      "num_input_tokens_seen": 1727848,
      "step": 2635
    },
    {
      "epoch": 1.5566037735849056,
      "grad_norm": 2.2764179706573486,
      "learning_rate": 7.780070754716981e-06,
      "loss": 0.3957,
      "num_input_tokens_seen": 1730408,
      "step": 2640
    },
    {
      "epoch": 1.5595518867924527,
      "grad_norm": 2.125385284423828,
      "learning_rate": 7.794811320754716e-06,
      "loss": 0.5313,
      "num_input_tokens_seen": 1732552,
      "step": 2645
    },
    {
      "epoch": 1.5625,
      "grad_norm": 6.923036098480225,
      "learning_rate": 7.809551886792453e-06,
      "loss": 0.4426,
      "num_input_tokens_seen": 1735080,
      "step": 2650
    },
    {
      "epoch": 1.5654481132075473,
      "grad_norm": 4.45912504196167,
      "learning_rate": 7.824292452830189e-06,
      "loss": 0.549,
      "num_input_tokens_seen": 1737672,
      "step": 2655
    },
    {
      "epoch": 1.5683962264150944,
      "grad_norm": 1.6451839208602905,
      "learning_rate": 7.839033018867926e-06,
      "loss": 0.7074,
      "num_input_tokens_seen": 1741352,
      "step": 2660
    },
    {
      "epoch": 1.5713443396226414,
      "grad_norm": 4.598958492279053,
      "learning_rate": 7.853773584905661e-06,
      "loss": 0.5314,
      "num_input_tokens_seen": 1744360,
      "step": 2665
    },
    {
      "epoch": 1.5742924528301887,
      "grad_norm": 2.721264600753784,
      "learning_rate": 7.868514150943397e-06,
      "loss": 0.4833,
      "num_input_tokens_seen": 1748136,
      "step": 2670
    },
    {
      "epoch": 1.577240566037736,
      "grad_norm": 2.762397050857544,
      "learning_rate": 7.883254716981132e-06,
      "loss": 0.5651,
      "num_input_tokens_seen": 1751912,
      "step": 2675
    },
    {
      "epoch": 1.580188679245283,
      "grad_norm": 4.709754467010498,
      "learning_rate": 7.897995283018869e-06,
      "loss": 0.7078,
      "num_input_tokens_seen": 1755624,
      "step": 2680
    },
    {
      "epoch": 1.5831367924528301,
      "grad_norm": 1.4719289541244507,
      "learning_rate": 7.912735849056604e-06,
      "loss": 0.6381,
      "num_input_tokens_seen": 1759272,
      "step": 2685
    },
    {
      "epoch": 1.5860849056603774,
      "grad_norm": 1.7858352661132812,
      "learning_rate": 7.927476415094341e-06,
      "loss": 0.3713,
      "num_input_tokens_seen": 1761960,
      "step": 2690
    },
    {
      "epoch": 1.5890330188679245,
      "grad_norm": 2.4074413776397705,
      "learning_rate": 7.942216981132077e-06,
      "loss": 0.6918,
      "num_input_tokens_seen": 1764808,
      "step": 2695
    },
    {
      "epoch": 1.5919811320754715,
      "grad_norm": 4.9043731689453125,
      "learning_rate": 7.956957547169812e-06,
      "loss": 0.5839,
      "num_input_tokens_seen": 1768104,
      "step": 2700
    },
    {
      "epoch": 1.5949292452830188,
      "grad_norm": 7.370609283447266,
      "learning_rate": 7.971698113207547e-06,
      "loss": 0.6301,
      "num_input_tokens_seen": 1770856,
      "step": 2705
    },
    {
      "epoch": 1.5978773584905661,
      "grad_norm": 2.049870252609253,
      "learning_rate": 7.986438679245284e-06,
      "loss": 0.6001,
      "num_input_tokens_seen": 1774376,
      "step": 2710
    },
    {
      "epoch": 1.6008254716981132,
      "grad_norm": 1.861102819442749,
      "learning_rate": 8.00117924528302e-06,
      "loss": 0.5317,
      "num_input_tokens_seen": 1778408,
      "step": 2715
    },
    {
      "epoch": 1.6037735849056602,
      "grad_norm": 2.4521374702453613,
      "learning_rate": 8.015919811320757e-06,
      "loss": 0.5523,
      "num_input_tokens_seen": 1782984,
      "step": 2720
    },
    {
      "epoch": 1.6067216981132075,
      "grad_norm": 2.4826347827911377,
      "learning_rate": 8.030660377358492e-06,
      "loss": 0.6567,
      "num_input_tokens_seen": 1786216,
      "step": 2725
    },
    {
      "epoch": 1.6096698113207548,
      "grad_norm": 6.061221122741699,
      "learning_rate": 8.045400943396227e-06,
      "loss": 0.6095,
      "num_input_tokens_seen": 1788872,
      "step": 2730
    },
    {
      "epoch": 1.6126179245283019,
      "grad_norm": 3.01009202003479,
      "learning_rate": 8.060141509433963e-06,
      "loss": 0.453,
      "num_input_tokens_seen": 1793288,
      "step": 2735
    },
    {
      "epoch": 1.615566037735849,
      "grad_norm": 2.6989073753356934,
      "learning_rate": 8.074882075471698e-06,
      "loss": 0.6921,
      "num_input_tokens_seen": 1797800,
      "step": 2740
    },
    {
      "epoch": 1.6185141509433962,
      "grad_norm": 1.6631203889846802,
      "learning_rate": 8.089622641509435e-06,
      "loss": 0.5082,
      "num_input_tokens_seen": 1800936,
      "step": 2745
    },
    {
      "epoch": 1.6214622641509435,
      "grad_norm": 4.207683086395264,
      "learning_rate": 8.10436320754717e-06,
      "loss": 0.6135,
      "num_input_tokens_seen": 1803560,
      "step": 2750
    },
    {
      "epoch": 1.6244103773584906,
      "grad_norm": 6.315492630004883,
      "learning_rate": 8.119103773584906e-06,
      "loss": 0.5156,
      "num_input_tokens_seen": 1806280,
      "step": 2755
    },
    {
      "epoch": 1.6273584905660377,
      "grad_norm": 3.7335734367370605,
      "learning_rate": 8.133844339622641e-06,
      "loss": 0.6498,
      "num_input_tokens_seen": 1808904,
      "step": 2760
    },
    {
      "epoch": 1.630306603773585,
      "grad_norm": 2.51202392578125,
      "learning_rate": 8.148584905660378e-06,
      "loss": 0.5332,
      "num_input_tokens_seen": 1811752,
      "step": 2765
    },
    {
      "epoch": 1.6332547169811322,
      "grad_norm": 2.711686134338379,
      "learning_rate": 8.163325471698114e-06,
      "loss": 0.5591,
      "num_input_tokens_seen": 1815784,
      "step": 2770
    },
    {
      "epoch": 1.6362028301886793,
      "grad_norm": 2.3494319915771484,
      "learning_rate": 8.17806603773585e-06,
      "loss": 0.483,
      "num_input_tokens_seen": 1818568,
      "step": 2775
    },
    {
      "epoch": 1.6391509433962264,
      "grad_norm": 4.432181358337402,
      "learning_rate": 8.192806603773586e-06,
      "loss": 0.5385,
      "num_input_tokens_seen": 1821288,
      "step": 2780
    },
    {
      "epoch": 1.6420990566037736,
      "grad_norm": 3.001502275466919,
      "learning_rate": 8.207547169811321e-06,
      "loss": 0.4553,
      "num_input_tokens_seen": 1823624,
      "step": 2785
    },
    {
      "epoch": 1.6450471698113207,
      "grad_norm": 11.780852317810059,
      "learning_rate": 8.222287735849057e-06,
      "loss": 0.717,
      "num_input_tokens_seen": 1830280,
      "step": 2790
    },
    {
      "epoch": 1.6479952830188678,
      "grad_norm": 9.597511291503906,
      "learning_rate": 8.237028301886794e-06,
      "loss": 0.5962,
      "num_input_tokens_seen": 1834280,
      "step": 2795
    },
    {
      "epoch": 1.650943396226415,
      "grad_norm": 2.1464192867279053,
      "learning_rate": 8.251768867924529e-06,
      "loss": 0.5397,
      "num_input_tokens_seen": 1838408,
      "step": 2800
    },
    {
      "epoch": 1.6538915094339623,
      "grad_norm": 3.3558614253997803,
      "learning_rate": 8.266509433962266e-06,
      "loss": 0.4121,
      "num_input_tokens_seen": 1842280,
      "step": 2805
    },
    {
      "epoch": 1.6568396226415094,
      "grad_norm": 5.738358497619629,
      "learning_rate": 8.281250000000001e-06,
      "loss": 0.5273,
      "num_input_tokens_seen": 1846504,
      "step": 2810
    },
    {
      "epoch": 1.6597877358490565,
      "grad_norm": 2.227666139602661,
      "learning_rate": 8.295990566037737e-06,
      "loss": 0.4514,
      "num_input_tokens_seen": 1849864,
      "step": 2815
    },
    {
      "epoch": 1.6627358490566038,
      "grad_norm": 3.857785701751709,
      "learning_rate": 8.310731132075472e-06,
      "loss": 0.4724,
      "num_input_tokens_seen": 1852648,
      "step": 2820
    },
    {
      "epoch": 1.665683962264151,
      "grad_norm": 2.4147236347198486,
      "learning_rate": 8.325471698113207e-06,
      "loss": 0.4499,
      "num_input_tokens_seen": 1855784,
      "step": 2825
    },
    {
      "epoch": 1.6686320754716981,
      "grad_norm": 3.537865400314331,
      "learning_rate": 8.340212264150944e-06,
      "loss": 0.5938,
      "num_input_tokens_seen": 1858888,
      "step": 2830
    },
    {
      "epoch": 1.6715801886792452,
      "grad_norm": 3.631096601486206,
      "learning_rate": 8.35495283018868e-06,
      "loss": 0.5335,
      "num_input_tokens_seen": 1861544,
      "step": 2835
    },
    {
      "epoch": 1.6745283018867925,
      "grad_norm": 4.720225811004639,
      "learning_rate": 8.369693396226415e-06,
      "loss": 0.4885,
      "num_input_tokens_seen": 1864712,
      "step": 2840
    },
    {
      "epoch": 1.6774764150943398,
      "grad_norm": 1.8945636749267578,
      "learning_rate": 8.38443396226415e-06,
      "loss": 0.4812,
      "num_input_tokens_seen": 1868232,
      "step": 2845
    },
    {
      "epoch": 1.6804245283018868,
      "grad_norm": 1.3173551559448242,
      "learning_rate": 8.399174528301888e-06,
      "loss": 0.4645,
      "num_input_tokens_seen": 1871880,
      "step": 2850
    },
    {
      "epoch": 1.6833726415094339,
      "grad_norm": 4.247873306274414,
      "learning_rate": 8.413915094339623e-06,
      "loss": 0.5878,
      "num_input_tokens_seen": 1874856,
      "step": 2855
    },
    {
      "epoch": 1.6863207547169812,
      "grad_norm": 4.463372707366943,
      "learning_rate": 8.42865566037736e-06,
      "loss": 0.5682,
      "num_input_tokens_seen": 1878120,
      "step": 2860
    },
    {
      "epoch": 1.6892688679245285,
      "grad_norm": 2.1462230682373047,
      "learning_rate": 8.443396226415095e-06,
      "loss": 0.5031,
      "num_input_tokens_seen": 1881064,
      "step": 2865
    },
    {
      "epoch": 1.6922169811320755,
      "grad_norm": 3.8219377994537354,
      "learning_rate": 8.45813679245283e-06,
      "loss": 0.4777,
      "num_input_tokens_seen": 1884808,
      "step": 2870
    },
    {
      "epoch": 1.6951650943396226,
      "grad_norm": 3.8765056133270264,
      "learning_rate": 8.472877358490566e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 1888168,
      "step": 2875
    },
    {
      "epoch": 1.6981132075471699,
      "grad_norm": 6.2940144538879395,
      "learning_rate": 8.487617924528303e-06,
      "loss": 0.5026,
      "num_input_tokens_seen": 1890792,
      "step": 2880
    },
    {
      "epoch": 1.701061320754717,
      "grad_norm": 2.945626974105835,
      "learning_rate": 8.502358490566038e-06,
      "loss": 0.537,
      "num_input_tokens_seen": 1894408,
      "step": 2885
    },
    {
      "epoch": 1.704009433962264,
      "grad_norm": 2.455336809158325,
      "learning_rate": 8.517099056603775e-06,
      "loss": 0.6238,
      "num_input_tokens_seen": 1897480,
      "step": 2890
    },
    {
      "epoch": 1.7069575471698113,
      "grad_norm": 3.1407828330993652,
      "learning_rate": 8.53183962264151e-06,
      "loss": 0.4434,
      "num_input_tokens_seen": 1901032,
      "step": 2895
    },
    {
      "epoch": 1.7099056603773586,
      "grad_norm": 1.9645917415618896,
      "learning_rate": 8.546580188679246e-06,
      "loss": 0.548,
      "num_input_tokens_seen": 1904200,
      "step": 2900
    },
    {
      "epoch": 1.7128537735849056,
      "grad_norm": 3.248854637145996,
      "learning_rate": 8.561320754716981e-06,
      "loss": 0.6487,
      "num_input_tokens_seen": 1906984,
      "step": 2905
    },
    {
      "epoch": 1.7158018867924527,
      "grad_norm": 1.8698830604553223,
      "learning_rate": 8.576061320754717e-06,
      "loss": 0.5328,
      "num_input_tokens_seen": 1910664,
      "step": 2910
    },
    {
      "epoch": 1.71875,
      "grad_norm": 2.6611201763153076,
      "learning_rate": 8.590801886792454e-06,
      "loss": 0.6134,
      "num_input_tokens_seen": 1913736,
      "step": 2915
    },
    {
      "epoch": 1.7216981132075473,
      "grad_norm": 2.473524570465088,
      "learning_rate": 8.605542452830189e-06,
      "loss": 0.5109,
      "num_input_tokens_seen": 1916648,
      "step": 2920
    },
    {
      "epoch": 1.7246462264150944,
      "grad_norm": 3.032504081726074,
      "learning_rate": 8.620283018867926e-06,
      "loss": 0.7577,
      "num_input_tokens_seen": 1919272,
      "step": 2925
    },
    {
      "epoch": 1.7275943396226414,
      "grad_norm": 3.726055383682251,
      "learning_rate": 8.635023584905662e-06,
      "loss": 0.5543,
      "num_input_tokens_seen": 1922664,
      "step": 2930
    },
    {
      "epoch": 1.7305424528301887,
      "grad_norm": 7.119819641113281,
      "learning_rate": 8.649764150943397e-06,
      "loss": 0.4303,
      "num_input_tokens_seen": 1925096,
      "step": 2935
    },
    {
      "epoch": 1.733490566037736,
      "grad_norm": 2.822646379470825,
      "learning_rate": 8.664504716981132e-06,
      "loss": 0.4655,
      "num_input_tokens_seen": 1928168,
      "step": 2940
    },
    {
      "epoch": 1.736438679245283,
      "grad_norm": 1.768086552619934,
      "learning_rate": 8.67924528301887e-06,
      "loss": 0.5053,
      "num_input_tokens_seen": 1932808,
      "step": 2945
    },
    {
      "epoch": 1.7393867924528301,
      "grad_norm": 2.935765504837036,
      "learning_rate": 8.693985849056605e-06,
      "loss": 0.4913,
      "num_input_tokens_seen": 1935624,
      "step": 2950
    },
    {
      "epoch": 1.7423349056603774,
      "grad_norm": 3.623666524887085,
      "learning_rate": 8.70872641509434e-06,
      "loss": 0.4896,
      "num_input_tokens_seen": 1938824,
      "step": 2955
    },
    {
      "epoch": 1.7452830188679245,
      "grad_norm": 2.279567241668701,
      "learning_rate": 8.723466981132075e-06,
      "loss": 0.4742,
      "num_input_tokens_seen": 1941864,
      "step": 2960
    },
    {
      "epoch": 1.7482311320754715,
      "grad_norm": 3.118412733078003,
      "learning_rate": 8.738207547169812e-06,
      "loss": 0.7312,
      "num_input_tokens_seen": 1944840,
      "step": 2965
    },
    {
      "epoch": 1.7511792452830188,
      "grad_norm": 2.315941095352173,
      "learning_rate": 8.752948113207548e-06,
      "loss": 0.3845,
      "num_input_tokens_seen": 1947208,
      "step": 2970
    },
    {
      "epoch": 1.7541273584905661,
      "grad_norm": 2.0478763580322266,
      "learning_rate": 8.767688679245285e-06,
      "loss": 0.4721,
      "num_input_tokens_seen": 1949672,
      "step": 2975
    },
    {
      "epoch": 1.7570754716981132,
      "grad_norm": 1.9229283332824707,
      "learning_rate": 8.78242924528302e-06,
      "loss": 0.577,
      "num_input_tokens_seen": 1952616,
      "step": 2980
    },
    {
      "epoch": 1.7600235849056602,
      "grad_norm": 3.9421474933624268,
      "learning_rate": 8.797169811320755e-06,
      "loss": 0.5033,
      "num_input_tokens_seen": 1955016,
      "step": 2985
    },
    {
      "epoch": 1.7629716981132075,
      "grad_norm": 4.925143241882324,
      "learning_rate": 8.81191037735849e-06,
      "loss": 0.5216,
      "num_input_tokens_seen": 1958088,
      "step": 2990
    },
    {
      "epoch": 1.7659198113207548,
      "grad_norm": 3.3835883140563965,
      "learning_rate": 8.826650943396226e-06,
      "loss": 0.4932,
      "num_input_tokens_seen": 1960712,
      "step": 2995
    },
    {
      "epoch": 1.7688679245283019,
      "grad_norm": 2.604309558868408,
      "learning_rate": 8.841391509433963e-06,
      "loss": 0.6008,
      "num_input_tokens_seen": 1964488,
      "step": 3000
    },
    {
      "epoch": 1.771816037735849,
      "grad_norm": 2.102077007293701,
      "learning_rate": 8.856132075471698e-06,
      "loss": 0.5004,
      "num_input_tokens_seen": 1967304,
      "step": 3005
    },
    {
      "epoch": 1.7747641509433962,
      "grad_norm": 2.1938815116882324,
      "learning_rate": 8.870872641509435e-06,
      "loss": 0.444,
      "num_input_tokens_seen": 1970888,
      "step": 3010
    },
    {
      "epoch": 1.7777122641509435,
      "grad_norm": 4.659045219421387,
      "learning_rate": 8.88561320754717e-06,
      "loss": 0.6787,
      "num_input_tokens_seen": 1974568,
      "step": 3015
    },
    {
      "epoch": 1.7806603773584906,
      "grad_norm": 5.26933479309082,
      "learning_rate": 8.900353773584906e-06,
      "loss": 0.5124,
      "num_input_tokens_seen": 1977384,
      "step": 3020
    },
    {
      "epoch": 1.7836084905660377,
      "grad_norm": 1.8948862552642822,
      "learning_rate": 8.915094339622642e-06,
      "loss": 0.5914,
      "num_input_tokens_seen": 1980712,
      "step": 3025
    },
    {
      "epoch": 1.786556603773585,
      "grad_norm": 1.7375677824020386,
      "learning_rate": 8.929834905660379e-06,
      "loss": 0.4686,
      "num_input_tokens_seen": 1983496,
      "step": 3030
    },
    {
      "epoch": 1.7895047169811322,
      "grad_norm": 2.4515326023101807,
      "learning_rate": 8.944575471698114e-06,
      "loss": 0.4699,
      "num_input_tokens_seen": 1986632,
      "step": 3035
    },
    {
      "epoch": 1.7924528301886793,
      "grad_norm": 2.278822898864746,
      "learning_rate": 8.959316037735851e-06,
      "loss": 0.3655,
      "num_input_tokens_seen": 1990664,
      "step": 3040
    },
    {
      "epoch": 1.7954009433962264,
      "grad_norm": 2.8040771484375,
      "learning_rate": 8.974056603773586e-06,
      "loss": 0.5044,
      "num_input_tokens_seen": 1993608,
      "step": 3045
    },
    {
      "epoch": 1.7983490566037736,
      "grad_norm": 2.134744644165039,
      "learning_rate": 8.988797169811322e-06,
      "loss": 0.492,
      "num_input_tokens_seen": 1997192,
      "step": 3050
    },
    {
      "epoch": 1.8012971698113207,
      "grad_norm": 2.776709794998169,
      "learning_rate": 9.003537735849057e-06,
      "loss": 0.5813,
      "num_input_tokens_seen": 2001224,
      "step": 3055
    },
    {
      "epoch": 1.8042452830188678,
      "grad_norm": 2.366702079772949,
      "learning_rate": 9.018278301886794e-06,
      "loss": 0.5031,
      "num_input_tokens_seen": 2003848,
      "step": 3060
    },
    {
      "epoch": 1.807193396226415,
      "grad_norm": 5.882743835449219,
      "learning_rate": 9.03301886792453e-06,
      "loss": 0.6091,
      "num_input_tokens_seen": 2006536,
      "step": 3065
    },
    {
      "epoch": 1.8101415094339623,
      "grad_norm": 5.8054327964782715,
      "learning_rate": 9.047759433962265e-06,
      "loss": 0.5772,
      "num_input_tokens_seen": 2010472,
      "step": 3070
    },
    {
      "epoch": 1.8130896226415094,
      "grad_norm": 1.5873346328735352,
      "learning_rate": 9.0625e-06,
      "loss": 0.457,
      "num_input_tokens_seen": 2013672,
      "step": 3075
    },
    {
      "epoch": 1.8160377358490565,
      "grad_norm": 6.886104106903076,
      "learning_rate": 9.077240566037735e-06,
      "loss": 0.7132,
      "num_input_tokens_seen": 2016232,
      "step": 3080
    },
    {
      "epoch": 1.8189858490566038,
      "grad_norm": 1.863874912261963,
      "learning_rate": 9.091981132075472e-06,
      "loss": 0.5605,
      "num_input_tokens_seen": 2019080,
      "step": 3085
    },
    {
      "epoch": 1.821933962264151,
      "grad_norm": 2.7448742389678955,
      "learning_rate": 9.106721698113208e-06,
      "loss": 0.5342,
      "num_input_tokens_seen": 2021928,
      "step": 3090
    },
    {
      "epoch": 1.8248820754716981,
      "grad_norm": 3.2121334075927734,
      "learning_rate": 9.121462264150945e-06,
      "loss": 0.569,
      "num_input_tokens_seen": 2024584,
      "step": 3095
    },
    {
      "epoch": 1.8278301886792452,
      "grad_norm": 2.3998215198516846,
      "learning_rate": 9.13620283018868e-06,
      "loss": 0.5409,
      "num_input_tokens_seen": 2028072,
      "step": 3100
    },
    {
      "epoch": 1.8307783018867925,
      "grad_norm": 2.167696237564087,
      "learning_rate": 9.150943396226416e-06,
      "loss": 0.4276,
      "num_input_tokens_seen": 2031048,
      "step": 3105
    },
    {
      "epoch": 1.8337264150943398,
      "grad_norm": 2.7847344875335693,
      "learning_rate": 9.165683962264151e-06,
      "loss": 0.4951,
      "num_input_tokens_seen": 2033896,
      "step": 3110
    },
    {
      "epoch": 1.8366745283018868,
      "grad_norm": 4.738274097442627,
      "learning_rate": 9.180424528301888e-06,
      "loss": 0.5943,
      "num_input_tokens_seen": 2036232,
      "step": 3115
    },
    {
      "epoch": 1.8396226415094339,
      "grad_norm": 10.097674369812012,
      "learning_rate": 9.195165094339623e-06,
      "loss": 0.5577,
      "num_input_tokens_seen": 2039208,
      "step": 3120
    },
    {
      "epoch": 1.8425707547169812,
      "grad_norm": 2.300649404525757,
      "learning_rate": 9.20990566037736e-06,
      "loss": 0.3829,
      "num_input_tokens_seen": 2041992,
      "step": 3125
    },
    {
      "epoch": 1.8455188679245285,
      "grad_norm": 1.9927616119384766,
      "learning_rate": 9.224646226415096e-06,
      "loss": 0.4404,
      "num_input_tokens_seen": 2045480,
      "step": 3130
    },
    {
      "epoch": 1.8484669811320755,
      "grad_norm": 1.9734218120574951,
      "learning_rate": 9.239386792452831e-06,
      "loss": 0.4147,
      "num_input_tokens_seen": 2049224,
      "step": 3135
    },
    {
      "epoch": 1.8514150943396226,
      "grad_norm": 3.009843111038208,
      "learning_rate": 9.254127358490566e-06,
      "loss": 0.7019,
      "num_input_tokens_seen": 2052680,
      "step": 3140
    },
    {
      "epoch": 1.8543632075471699,
      "grad_norm": 4.120747089385986,
      "learning_rate": 9.268867924528303e-06,
      "loss": 0.4869,
      "num_input_tokens_seen": 2055912,
      "step": 3145
    },
    {
      "epoch": 1.857311320754717,
      "grad_norm": 1.9137389659881592,
      "learning_rate": 9.283608490566039e-06,
      "loss": 0.5925,
      "num_input_tokens_seen": 2059496,
      "step": 3150
    },
    {
      "epoch": 1.860259433962264,
      "grad_norm": 2.596644878387451,
      "learning_rate": 9.298349056603774e-06,
      "loss": 0.5065,
      "num_input_tokens_seen": 2062600,
      "step": 3155
    },
    {
      "epoch": 1.8632075471698113,
      "grad_norm": 2.587611198425293,
      "learning_rate": 9.31308962264151e-06,
      "loss": 0.4166,
      "num_input_tokens_seen": 2065704,
      "step": 3160
    },
    {
      "epoch": 1.8661556603773586,
      "grad_norm": 3.6842384338378906,
      "learning_rate": 9.327830188679245e-06,
      "loss": 0.4984,
      "num_input_tokens_seen": 2068104,
      "step": 3165
    },
    {
      "epoch": 1.8691037735849056,
      "grad_norm": 8.193092346191406,
      "learning_rate": 9.342570754716982e-06,
      "loss": 0.6733,
      "num_input_tokens_seen": 2071240,
      "step": 3170
    },
    {
      "epoch": 1.8720518867924527,
      "grad_norm": 1.2968326807022095,
      "learning_rate": 9.357311320754717e-06,
      "loss": 0.4775,
      "num_input_tokens_seen": 2076456,
      "step": 3175
    },
    {
      "epoch": 1.875,
      "grad_norm": 1.3121097087860107,
      "learning_rate": 9.372051886792454e-06,
      "loss": 0.4993,
      "num_input_tokens_seen": 2079720,
      "step": 3180
    },
    {
      "epoch": 1.8779481132075473,
      "grad_norm": 2.8920905590057373,
      "learning_rate": 9.38679245283019e-06,
      "loss": 0.7159,
      "num_input_tokens_seen": 2083528,
      "step": 3185
    },
    {
      "epoch": 1.8808962264150944,
      "grad_norm": 1.498838186264038,
      "learning_rate": 9.401533018867925e-06,
      "loss": 0.5269,
      "num_input_tokens_seen": 2087400,
      "step": 3190
    },
    {
      "epoch": 1.8838443396226414,
      "grad_norm": 2.6425116062164307,
      "learning_rate": 9.41627358490566e-06,
      "loss": 0.4166,
      "num_input_tokens_seen": 2091176,
      "step": 3195
    },
    {
      "epoch": 1.8867924528301887,
      "grad_norm": 2.202831983566284,
      "learning_rate": 9.431014150943397e-06,
      "loss": 0.4323,
      "num_input_tokens_seen": 2093928,
      "step": 3200
    },
    {
      "epoch": 1.889740566037736,
      "grad_norm": 2.5115792751312256,
      "learning_rate": 9.445754716981133e-06,
      "loss": 0.4202,
      "num_input_tokens_seen": 2097992,
      "step": 3205
    },
    {
      "epoch": 1.892688679245283,
      "grad_norm": 1.6988894939422607,
      "learning_rate": 9.46049528301887e-06,
      "loss": 0.5423,
      "num_input_tokens_seen": 2101256,
      "step": 3210
    },
    {
      "epoch": 1.8956367924528301,
      "grad_norm": 3.428349018096924,
      "learning_rate": 9.475235849056605e-06,
      "loss": 0.4868,
      "num_input_tokens_seen": 2104488,
      "step": 3215
    },
    {
      "epoch": 1.8985849056603774,
      "grad_norm": 6.883533954620361,
      "learning_rate": 9.48997641509434e-06,
      "loss": 0.475,
      "num_input_tokens_seen": 2107464,
      "step": 3220
    },
    {
      "epoch": 1.9015330188679245,
      "grad_norm": 3.8597731590270996,
      "learning_rate": 9.504716981132076e-06,
      "loss": 0.7006,
      "num_input_tokens_seen": 2110280,
      "step": 3225
    },
    {
      "epoch": 1.9044811320754715,
      "grad_norm": 4.466874599456787,
      "learning_rate": 9.519457547169813e-06,
      "loss": 0.5718,
      "num_input_tokens_seen": 2113384,
      "step": 3230
    },
    {
      "epoch": 1.9074292452830188,
      "grad_norm": 1.9670289754867554,
      "learning_rate": 9.534198113207548e-06,
      "loss": 0.5183,
      "num_input_tokens_seen": 2116520,
      "step": 3235
    },
    {
      "epoch": 1.9103773584905661,
      "grad_norm": 3.1198627948760986,
      "learning_rate": 9.548938679245285e-06,
      "loss": 0.5145,
      "num_input_tokens_seen": 2119528,
      "step": 3240
    },
    {
      "epoch": 1.9133254716981132,
      "grad_norm": 2.5630526542663574,
      "learning_rate": 9.56367924528302e-06,
      "loss": 0.4883,
      "num_input_tokens_seen": 2122440,
      "step": 3245
    },
    {
      "epoch": 1.9162735849056602,
      "grad_norm": 1.8640820980072021,
      "learning_rate": 9.578419811320756e-06,
      "loss": 0.5062,
      "num_input_tokens_seen": 2125352,
      "step": 3250
    },
    {
      "epoch": 1.9192216981132075,
      "grad_norm": 2.2215869426727295,
      "learning_rate": 9.593160377358491e-06,
      "loss": 0.5346,
      "num_input_tokens_seen": 2128680,
      "step": 3255
    },
    {
      "epoch": 1.9221698113207548,
      "grad_norm": 3.1649980545043945,
      "learning_rate": 9.607900943396226e-06,
      "loss": 0.3889,
      "num_input_tokens_seen": 2132040,
      "step": 3260
    },
    {
      "epoch": 1.9251179245283019,
      "grad_norm": 4.021702766418457,
      "learning_rate": 9.622641509433963e-06,
      "loss": 0.4944,
      "num_input_tokens_seen": 2135592,
      "step": 3265
    },
    {
      "epoch": 1.928066037735849,
      "grad_norm": 2.178122043609619,
      "learning_rate": 9.637382075471699e-06,
      "loss": 0.5111,
      "num_input_tokens_seen": 2138312,
      "step": 3270
    },
    {
      "epoch": 1.9310141509433962,
      "grad_norm": 3.385652780532837,
      "learning_rate": 9.652122641509434e-06,
      "loss": 0.5849,
      "num_input_tokens_seen": 2141512,
      "step": 3275
    },
    {
      "epoch": 1.9339622641509435,
      "grad_norm": 2.067254066467285,
      "learning_rate": 9.66686320754717e-06,
      "loss": 0.4269,
      "num_input_tokens_seen": 2143528,
      "step": 3280
    },
    {
      "epoch": 1.9369103773584906,
      "grad_norm": 1.750783920288086,
      "learning_rate": 9.681603773584907e-06,
      "loss": 0.5524,
      "num_input_tokens_seen": 2147432,
      "step": 3285
    },
    {
      "epoch": 1.9398584905660377,
      "grad_norm": 3.428699254989624,
      "learning_rate": 9.696344339622642e-06,
      "loss": 0.5061,
      "num_input_tokens_seen": 2150024,
      "step": 3290
    },
    {
      "epoch": 1.942806603773585,
      "grad_norm": 1.5365263223648071,
      "learning_rate": 9.711084905660379e-06,
      "loss": 0.4664,
      "num_input_tokens_seen": 2153160,
      "step": 3295
    },
    {
      "epoch": 1.9457547169811322,
      "grad_norm": 2.592742681503296,
      "learning_rate": 9.725825471698114e-06,
      "loss": 0.5878,
      "num_input_tokens_seen": 2155944,
      "step": 3300
    },
    {
      "epoch": 1.9487028301886793,
      "grad_norm": 2.3278636932373047,
      "learning_rate": 9.74056603773585e-06,
      "loss": 0.5118,
      "num_input_tokens_seen": 2160360,
      "step": 3305
    },
    {
      "epoch": 1.9516509433962264,
      "grad_norm": 3.8329479694366455,
      "learning_rate": 9.755306603773585e-06,
      "loss": 0.5316,
      "num_input_tokens_seen": 2163080,
      "step": 3310
    },
    {
      "epoch": 1.9545990566037736,
      "grad_norm": 1.9475082159042358,
      "learning_rate": 9.770047169811322e-06,
      "loss": 0.4617,
      "num_input_tokens_seen": 2167176,
      "step": 3315
    },
    {
      "epoch": 1.9575471698113207,
      "grad_norm": 3.7722434997558594,
      "learning_rate": 9.784787735849057e-06,
      "loss": 0.5588,
      "num_input_tokens_seen": 2169928,
      "step": 3320
    },
    {
      "epoch": 1.9604952830188678,
      "grad_norm": 3.540411949157715,
      "learning_rate": 9.799528301886794e-06,
      "loss": 0.7276,
      "num_input_tokens_seen": 2172872,
      "step": 3325
    },
    {
      "epoch": 1.963443396226415,
      "grad_norm": 2.510077714920044,
      "learning_rate": 9.81426886792453e-06,
      "loss": 0.5249,
      "num_input_tokens_seen": 2175752,
      "step": 3330
    },
    {
      "epoch": 1.9663915094339623,
      "grad_norm": 3.3462636470794678,
      "learning_rate": 9.829009433962265e-06,
      "loss": 0.5895,
      "num_input_tokens_seen": 2179528,
      "step": 3335
    },
    {
      "epoch": 1.9693396226415094,
      "grad_norm": 1.5495750904083252,
      "learning_rate": 9.84375e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 2182824,
      "step": 3340
    },
    {
      "epoch": 1.9722877358490565,
      "grad_norm": 1.9227246046066284,
      "learning_rate": 9.858490566037736e-06,
      "loss": 0.5313,
      "num_input_tokens_seen": 2185960,
      "step": 3345
    },
    {
      "epoch": 1.9752358490566038,
      "grad_norm": 5.357235431671143,
      "learning_rate": 9.873231132075473e-06,
      "loss": 0.4859,
      "num_input_tokens_seen": 2189512,
      "step": 3350
    },
    {
      "epoch": 1.978183962264151,
      "grad_norm": 2.080068588256836,
      "learning_rate": 9.887971698113208e-06,
      "loss": 0.6377,
      "num_input_tokens_seen": 2193064,
      "step": 3355
    },
    {
      "epoch": 1.9811320754716981,
      "grad_norm": 2.2908098697662354,
      "learning_rate": 9.902712264150945e-06,
      "loss": 0.5751,
      "num_input_tokens_seen": 2195560,
      "step": 3360
    },
    {
      "epoch": 1.9840801886792452,
      "grad_norm": 2.213653326034546,
      "learning_rate": 9.917452830188679e-06,
      "loss": 0.6579,
      "num_input_tokens_seen": 2200616,
      "step": 3365
    },
    {
      "epoch": 1.9870283018867925,
      "grad_norm": 2.739194393157959,
      "learning_rate": 9.932193396226416e-06,
      "loss": 0.6471,
      "num_input_tokens_seen": 2203880,
      "step": 3370
    },
    {
      "epoch": 1.9899764150943398,
      "grad_norm": 3.510056257247925,
      "learning_rate": 9.946933962264151e-06,
      "loss": 0.5257,
      "num_input_tokens_seen": 2206664,
      "step": 3375
    },
    {
      "epoch": 1.9929245283018868,
      "grad_norm": 2.35871958732605,
      "learning_rate": 9.961674528301888e-06,
      "loss": 0.5255,
      "num_input_tokens_seen": 2208808,
      "step": 3380
    },
    {
      "epoch": 1.9958726415094339,
      "grad_norm": 2.699148416519165,
      "learning_rate": 9.976415094339624e-06,
      "loss": 0.6001,
      "num_input_tokens_seen": 2212200,
      "step": 3385
    },
    {
      "epoch": 1.9988207547169812,
      "grad_norm": 12.069649696350098,
      "learning_rate": 9.991155660377359e-06,
      "loss": 0.6065,
      "num_input_tokens_seen": 2215656,
      "step": 3390
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.5241208672523499,
      "eval_runtime": 19.0435,
      "eval_samples_per_second": 89.059,
      "eval_steps_per_second": 22.265,
      "num_input_tokens_seen": 2216560,
      "step": 3392
    },
    {
      "epoch": 2.0017688679245285,
      "grad_norm": 2.4409494400024414,
      "learning_rate": 9.999999894098275e-06,
      "loss": 0.4593,
      "num_input_tokens_seen": 2218416,
      "step": 3395
    },
    {
      "epoch": 2.0047169811320753,
      "grad_norm": 2.269171953201294,
      "learning_rate": 9.99999870270391e-06,
      "loss": 0.4976,
      "num_input_tokens_seen": 2222480,
      "step": 3400
    },
    {
      "epoch": 2.0076650943396226,
      "grad_norm": 3.3023276329040527,
      "learning_rate": 9.999996187538341e-06,
      "loss": 0.4705,
      "num_input_tokens_seen": 2224752,
      "step": 3405
    },
    {
      "epoch": 2.01061320754717,
      "grad_norm": 4.064953327178955,
      "learning_rate": 9.999992348602233e-06,
      "loss": 0.4812,
      "num_input_tokens_seen": 2227216,
      "step": 3410
    },
    {
      "epoch": 2.013561320754717,
      "grad_norm": 2.483228921890259,
      "learning_rate": 9.999987185896598e-06,
      "loss": 0.6215,
      "num_input_tokens_seen": 2230224,
      "step": 3415
    },
    {
      "epoch": 2.016509433962264,
      "grad_norm": 2.4251623153686523,
      "learning_rate": 9.99998069942281e-06,
      "loss": 0.5158,
      "num_input_tokens_seen": 2234288,
      "step": 3420
    },
    {
      "epoch": 2.0194575471698113,
      "grad_norm": 2.1537842750549316,
      "learning_rate": 9.999972889182583e-06,
      "loss": 0.5134,
      "num_input_tokens_seen": 2238224,
      "step": 3425
    },
    {
      "epoch": 2.0224056603773586,
      "grad_norm": 2.692521572113037,
      "learning_rate": 9.999963755177984e-06,
      "loss": 0.6871,
      "num_input_tokens_seen": 2240624,
      "step": 3430
    },
    {
      "epoch": 2.025353773584906,
      "grad_norm": 2.8487043380737305,
      "learning_rate": 9.999953297411434e-06,
      "loss": 0.6166,
      "num_input_tokens_seen": 2243504,
      "step": 3435
    },
    {
      "epoch": 2.0283018867924527,
      "grad_norm": 1.4369525909423828,
      "learning_rate": 9.999941515885699e-06,
      "loss": 0.5816,
      "num_input_tokens_seen": 2247600,
      "step": 3440
    },
    {
      "epoch": 2.03125,
      "grad_norm": 2.5074074268341064,
      "learning_rate": 9.999928410603897e-06,
      "loss": 0.5001,
      "num_input_tokens_seen": 2250544,
      "step": 3445
    },
    {
      "epoch": 2.0341981132075473,
      "grad_norm": 3.1471071243286133,
      "learning_rate": 9.999913981569502e-06,
      "loss": 0.5998,
      "num_input_tokens_seen": 2253552,
      "step": 3450
    },
    {
      "epoch": 2.037146226415094,
      "grad_norm": 14.052713394165039,
      "learning_rate": 9.999898228786332e-06,
      "loss": 0.5376,
      "num_input_tokens_seen": 2256208,
      "step": 3455
    },
    {
      "epoch": 2.0400943396226414,
      "grad_norm": 3.779376983642578,
      "learning_rate": 9.999881152258557e-06,
      "loss": 0.5176,
      "num_input_tokens_seen": 2259536,
      "step": 3460
    },
    {
      "epoch": 2.0430424528301887,
      "grad_norm": 2.664102077484131,
      "learning_rate": 9.999862751990697e-06,
      "loss": 0.497,
      "num_input_tokens_seen": 2262448,
      "step": 3465
    },
    {
      "epoch": 2.045990566037736,
      "grad_norm": 2.0035548210144043,
      "learning_rate": 9.999843027987628e-06,
      "loss": 0.455,
      "num_input_tokens_seen": 2266160,
      "step": 3470
    },
    {
      "epoch": 2.048938679245283,
      "grad_norm": 3.569155216217041,
      "learning_rate": 9.999821980254567e-06,
      "loss": 0.5841,
      "num_input_tokens_seen": 2269232,
      "step": 3475
    },
    {
      "epoch": 2.05188679245283,
      "grad_norm": 1.706276774406433,
      "learning_rate": 9.99979960879709e-06,
      "loss": 0.4524,
      "num_input_tokens_seen": 2272720,
      "step": 3480
    },
    {
      "epoch": 2.0548349056603774,
      "grad_norm": 3.2420685291290283,
      "learning_rate": 9.99977591362112e-06,
      "loss": 0.4712,
      "num_input_tokens_seen": 2275632,
      "step": 3485
    },
    {
      "epoch": 2.0577830188679247,
      "grad_norm": 6.191486835479736,
      "learning_rate": 9.999750894732927e-06,
      "loss": 0.3858,
      "num_input_tokens_seen": 2277936,
      "step": 3490
    },
    {
      "epoch": 2.0607311320754715,
      "grad_norm": 2.5046162605285645,
      "learning_rate": 9.999724552139136e-06,
      "loss": 0.6106,
      "num_input_tokens_seen": 2280624,
      "step": 3495
    },
    {
      "epoch": 2.063679245283019,
      "grad_norm": 2.8309781551361084,
      "learning_rate": 9.999696885846724e-06,
      "loss": 0.5104,
      "num_input_tokens_seen": 2287312,
      "step": 3500
    },
    {
      "epoch": 2.066627358490566,
      "grad_norm": 1.7976757287979126,
      "learning_rate": 9.999667895863012e-06,
      "loss": 0.4961,
      "num_input_tokens_seen": 2289904,
      "step": 3505
    },
    {
      "epoch": 2.0695754716981134,
      "grad_norm": 2.188403606414795,
      "learning_rate": 9.99963758219568e-06,
      "loss": 0.4808,
      "num_input_tokens_seen": 2293264,
      "step": 3510
    },
    {
      "epoch": 2.0725235849056602,
      "grad_norm": 1.96734619140625,
      "learning_rate": 9.999605944852749e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 2296048,
      "step": 3515
    },
    {
      "epoch": 2.0754716981132075,
      "grad_norm": 2.087714195251465,
      "learning_rate": 9.999572983842599e-06,
      "loss": 0.4187,
      "num_input_tokens_seen": 2299280,
      "step": 3520
    },
    {
      "epoch": 2.078419811320755,
      "grad_norm": 1.04439115524292,
      "learning_rate": 9.999538699173951e-06,
      "loss": 0.4511,
      "num_input_tokens_seen": 2302128,
      "step": 3525
    },
    {
      "epoch": 2.081367924528302,
      "grad_norm": 3.2082581520080566,
      "learning_rate": 9.99950309085589e-06,
      "loss": 0.4407,
      "num_input_tokens_seen": 2306896,
      "step": 3530
    },
    {
      "epoch": 2.084316037735849,
      "grad_norm": 10.509344100952148,
      "learning_rate": 9.999466158897835e-06,
      "loss": 0.4563,
      "num_input_tokens_seen": 2309808,
      "step": 3535
    },
    {
      "epoch": 2.0872641509433962,
      "grad_norm": 2.3370521068573,
      "learning_rate": 9.999427903309569e-06,
      "loss": 0.566,
      "num_input_tokens_seen": 2313104,
      "step": 3540
    },
    {
      "epoch": 2.0902122641509435,
      "grad_norm": 1.587203025817871,
      "learning_rate": 9.99938832410122e-06,
      "loss": 0.5894,
      "num_input_tokens_seen": 2316368,
      "step": 3545
    },
    {
      "epoch": 2.0931603773584904,
      "grad_norm": 1.8656679391860962,
      "learning_rate": 9.999347421283267e-06,
      "loss": 0.459,
      "num_input_tokens_seen": 2319184,
      "step": 3550
    },
    {
      "epoch": 2.0961084905660377,
      "grad_norm": 2.6150450706481934,
      "learning_rate": 9.999305194866538e-06,
      "loss": 0.6886,
      "num_input_tokens_seen": 2322896,
      "step": 3555
    },
    {
      "epoch": 2.099056603773585,
      "grad_norm": 1.2895225286483765,
      "learning_rate": 9.99926164486221e-06,
      "loss": 0.4123,
      "num_input_tokens_seen": 2326224,
      "step": 3560
    },
    {
      "epoch": 2.1020047169811322,
      "grad_norm": 3.896390676498413,
      "learning_rate": 9.99921677128182e-06,
      "loss": 0.5663,
      "num_input_tokens_seen": 2329552,
      "step": 3565
    },
    {
      "epoch": 2.104952830188679,
      "grad_norm": 1.6260031461715698,
      "learning_rate": 9.99917057413724e-06,
      "loss": 0.6796,
      "num_input_tokens_seen": 2334288,
      "step": 3570
    },
    {
      "epoch": 2.1079009433962264,
      "grad_norm": 2.6255950927734375,
      "learning_rate": 9.99912305344071e-06,
      "loss": 0.5253,
      "num_input_tokens_seen": 2337968,
      "step": 3575
    },
    {
      "epoch": 2.1108490566037736,
      "grad_norm": 1.1682586669921875,
      "learning_rate": 9.999074209204803e-06,
      "loss": 0.4651,
      "num_input_tokens_seen": 2342160,
      "step": 3580
    },
    {
      "epoch": 2.113797169811321,
      "grad_norm": 2.3923587799072266,
      "learning_rate": 9.999024041442455e-06,
      "loss": 0.4624,
      "num_input_tokens_seen": 2346000,
      "step": 3585
    },
    {
      "epoch": 2.1167452830188678,
      "grad_norm": 1.3662807941436768,
      "learning_rate": 9.998972550166948e-06,
      "loss": 0.4804,
      "num_input_tokens_seen": 2348656,
      "step": 3590
    },
    {
      "epoch": 2.119693396226415,
      "grad_norm": 2.483086585998535,
      "learning_rate": 9.998919735391915e-06,
      "loss": 0.427,
      "num_input_tokens_seen": 2351344,
      "step": 3595
    },
    {
      "epoch": 2.1226415094339623,
      "grad_norm": 2.371692419052124,
      "learning_rate": 9.998865597131336e-06,
      "loss": 0.5055,
      "num_input_tokens_seen": 2354160,
      "step": 3600
    },
    {
      "epoch": 2.1255896226415096,
      "grad_norm": 1.2813798189163208,
      "learning_rate": 9.998810135399545e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 2357744,
      "step": 3605
    },
    {
      "epoch": 2.1285377358490565,
      "grad_norm": 1.744266152381897,
      "learning_rate": 9.99875335021123e-06,
      "loss": 0.4861,
      "num_input_tokens_seen": 2361264,
      "step": 3610
    },
    {
      "epoch": 2.1314858490566038,
      "grad_norm": 2.644942283630371,
      "learning_rate": 9.998695241581423e-06,
      "loss": 0.394,
      "num_input_tokens_seen": 2365072,
      "step": 3615
    },
    {
      "epoch": 2.134433962264151,
      "grad_norm": 1.478508710861206,
      "learning_rate": 9.998635809525504e-06,
      "loss": 0.5237,
      "num_input_tokens_seen": 2367856,
      "step": 3620
    },
    {
      "epoch": 2.137382075471698,
      "grad_norm": 2.8589885234832764,
      "learning_rate": 9.998575054059212e-06,
      "loss": 0.4819,
      "num_input_tokens_seen": 2370576,
      "step": 3625
    },
    {
      "epoch": 2.140330188679245,
      "grad_norm": 7.645646095275879,
      "learning_rate": 9.998512975198633e-06,
      "loss": 0.4702,
      "num_input_tokens_seen": 2372912,
      "step": 3630
    },
    {
      "epoch": 2.1432783018867925,
      "grad_norm": 1.6633960008621216,
      "learning_rate": 9.998449572960202e-06,
      "loss": 0.4821,
      "num_input_tokens_seen": 2376688,
      "step": 3635
    },
    {
      "epoch": 2.1462264150943398,
      "grad_norm": 3.9250335693359375,
      "learning_rate": 9.998384847360705e-06,
      "loss": 0.8291,
      "num_input_tokens_seen": 2379568,
      "step": 3640
    },
    {
      "epoch": 2.1491745283018866,
      "grad_norm": 1.8899283409118652,
      "learning_rate": 9.998318798417276e-06,
      "loss": 0.5631,
      "num_input_tokens_seen": 2383056,
      "step": 3645
    },
    {
      "epoch": 2.152122641509434,
      "grad_norm": 6.645150184631348,
      "learning_rate": 9.998251426147403e-06,
      "loss": 0.4873,
      "num_input_tokens_seen": 2385872,
      "step": 3650
    },
    {
      "epoch": 2.155070754716981,
      "grad_norm": 2.8679258823394775,
      "learning_rate": 9.998182730568927e-06,
      "loss": 0.5898,
      "num_input_tokens_seen": 2388560,
      "step": 3655
    },
    {
      "epoch": 2.1580188679245285,
      "grad_norm": 2.6323959827423096,
      "learning_rate": 9.998112711700028e-06,
      "loss": 0.5721,
      "num_input_tokens_seen": 2392016,
      "step": 3660
    },
    {
      "epoch": 2.1609669811320753,
      "grad_norm": 2.7856907844543457,
      "learning_rate": 9.99804136955925e-06,
      "loss": 0.6504,
      "num_input_tokens_seen": 2394576,
      "step": 3665
    },
    {
      "epoch": 2.1639150943396226,
      "grad_norm": 3.2992379665374756,
      "learning_rate": 9.99796870416548e-06,
      "loss": 0.5721,
      "num_input_tokens_seen": 2397424,
      "step": 3670
    },
    {
      "epoch": 2.16686320754717,
      "grad_norm": 2.566983461380005,
      "learning_rate": 9.997894715537953e-06,
      "loss": 0.4476,
      "num_input_tokens_seen": 2400848,
      "step": 3675
    },
    {
      "epoch": 2.169811320754717,
      "grad_norm": 1.0927411317825317,
      "learning_rate": 9.997819403696263e-06,
      "loss": 0.4068,
      "num_input_tokens_seen": 2404688,
      "step": 3680
    },
    {
      "epoch": 2.172759433962264,
      "grad_norm": 1.9017906188964844,
      "learning_rate": 9.997742768660345e-06,
      "loss": 0.6369,
      "num_input_tokens_seen": 2407632,
      "step": 3685
    },
    {
      "epoch": 2.1757075471698113,
      "grad_norm": 2.437448024749756,
      "learning_rate": 9.99766481045049e-06,
      "loss": 0.456,
      "num_input_tokens_seen": 2412464,
      "step": 3690
    },
    {
      "epoch": 2.1786556603773586,
      "grad_norm": 3.0124521255493164,
      "learning_rate": 9.997585529087338e-06,
      "loss": 0.3509,
      "num_input_tokens_seen": 2415440,
      "step": 3695
    },
    {
      "epoch": 2.1816037735849054,
      "grad_norm": 1.8364503383636475,
      "learning_rate": 9.997504924591878e-06,
      "loss": 0.6574,
      "num_input_tokens_seen": 2418928,
      "step": 3700
    },
    {
      "epoch": 2.1845518867924527,
      "grad_norm": 1.6308305263519287,
      "learning_rate": 9.99742299698545e-06,
      "loss": 0.4771,
      "num_input_tokens_seen": 2422160,
      "step": 3705
    },
    {
      "epoch": 2.1875,
      "grad_norm": 2.1061747074127197,
      "learning_rate": 9.997339746289749e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 2426352,
      "step": 3710
    },
    {
      "epoch": 2.1904481132075473,
      "grad_norm": 1.671177625656128,
      "learning_rate": 9.997255172526812e-06,
      "loss": 0.4523,
      "num_input_tokens_seen": 2430544,
      "step": 3715
    },
    {
      "epoch": 2.1933962264150946,
      "grad_norm": 1.9334760904312134,
      "learning_rate": 9.99716927571903e-06,
      "loss": 0.5889,
      "num_input_tokens_seen": 2433744,
      "step": 3720
    },
    {
      "epoch": 2.1963443396226414,
      "grad_norm": 2.145174741744995,
      "learning_rate": 9.997082055889147e-06,
      "loss": 0.5652,
      "num_input_tokens_seen": 2436624,
      "step": 3725
    },
    {
      "epoch": 2.1992924528301887,
      "grad_norm": 2.908642292022705,
      "learning_rate": 9.996993513060252e-06,
      "loss": 0.4681,
      "num_input_tokens_seen": 2438672,
      "step": 3730
    },
    {
      "epoch": 2.202240566037736,
      "grad_norm": 1.6083283424377441,
      "learning_rate": 9.996903647255789e-06,
      "loss": 0.3877,
      "num_input_tokens_seen": 2442256,
      "step": 3735
    },
    {
      "epoch": 2.205188679245283,
      "grad_norm": 3.1025285720825195,
      "learning_rate": 9.99681245849955e-06,
      "loss": 0.5571,
      "num_input_tokens_seen": 2445264,
      "step": 3740
    },
    {
      "epoch": 2.20813679245283,
      "grad_norm": 20.3349666595459,
      "learning_rate": 9.996719946815679e-06,
      "loss": 0.5638,
      "num_input_tokens_seen": 2448816,
      "step": 3745
    },
    {
      "epoch": 2.2110849056603774,
      "grad_norm": 4.318816661834717,
      "learning_rate": 9.996626112228665e-06,
      "loss": 0.5967,
      "num_input_tokens_seen": 2451344,
      "step": 3750
    },
    {
      "epoch": 2.2140330188679247,
      "grad_norm": 1.7857407331466675,
      "learning_rate": 9.996530954763355e-06,
      "loss": 0.3934,
      "num_input_tokens_seen": 2453808,
      "step": 3755
    },
    {
      "epoch": 2.2169811320754715,
      "grad_norm": 2.3087985515594482,
      "learning_rate": 9.99643447444494e-06,
      "loss": 0.3961,
      "num_input_tokens_seen": 2456816,
      "step": 3760
    },
    {
      "epoch": 2.219929245283019,
      "grad_norm": 1.8218660354614258,
      "learning_rate": 9.996336671298965e-06,
      "loss": 0.5955,
      "num_input_tokens_seen": 2460816,
      "step": 3765
    },
    {
      "epoch": 2.222877358490566,
      "grad_norm": 10.457060813903809,
      "learning_rate": 9.996237545351323e-06,
      "loss": 0.6665,
      "num_input_tokens_seen": 2464016,
      "step": 3770
    },
    {
      "epoch": 2.2258254716981134,
      "grad_norm": 2.463376760482788,
      "learning_rate": 9.996137096628259e-06,
      "loss": 0.4216,
      "num_input_tokens_seen": 2467536,
      "step": 3775
    },
    {
      "epoch": 2.2287735849056602,
      "grad_norm": 2.1375865936279297,
      "learning_rate": 9.996035325156366e-06,
      "loss": 0.4539,
      "num_input_tokens_seen": 2470928,
      "step": 3780
    },
    {
      "epoch": 2.2317216981132075,
      "grad_norm": 1.808367371559143,
      "learning_rate": 9.995932230962589e-06,
      "loss": 0.3544,
      "num_input_tokens_seen": 2474928,
      "step": 3785
    },
    {
      "epoch": 2.234669811320755,
      "grad_norm": 1.7464582920074463,
      "learning_rate": 9.995827814074223e-06,
      "loss": 0.7117,
      "num_input_tokens_seen": 2477456,
      "step": 3790
    },
    {
      "epoch": 2.237617924528302,
      "grad_norm": 2.25754976272583,
      "learning_rate": 9.995722074518913e-06,
      "loss": 0.6492,
      "num_input_tokens_seen": 2479888,
      "step": 3795
    },
    {
      "epoch": 2.240566037735849,
      "grad_norm": 1.4202998876571655,
      "learning_rate": 9.99561501232465e-06,
      "loss": 0.2765,
      "num_input_tokens_seen": 2483344,
      "step": 3800
    },
    {
      "epoch": 2.2435141509433962,
      "grad_norm": 1.5416580438613892,
      "learning_rate": 9.995506627519786e-06,
      "loss": 0.4672,
      "num_input_tokens_seen": 2486448,
      "step": 3805
    },
    {
      "epoch": 2.2464622641509435,
      "grad_norm": 2.608919620513916,
      "learning_rate": 9.995396920133012e-06,
      "loss": 0.7162,
      "num_input_tokens_seen": 2489392,
      "step": 3810
    },
    {
      "epoch": 2.2494103773584904,
      "grad_norm": 1.4983595609664917,
      "learning_rate": 9.995285890193373e-06,
      "loss": 0.4994,
      "num_input_tokens_seen": 2492400,
      "step": 3815
    },
    {
      "epoch": 2.2523584905660377,
      "grad_norm": 2.3803086280822754,
      "learning_rate": 9.995173537730267e-06,
      "loss": 0.4635,
      "num_input_tokens_seen": 2495088,
      "step": 3820
    },
    {
      "epoch": 2.255306603773585,
      "grad_norm": 1.5899994373321533,
      "learning_rate": 9.99505986277344e-06,
      "loss": 0.4547,
      "num_input_tokens_seen": 2498640,
      "step": 3825
    },
    {
      "epoch": 2.2582547169811322,
      "grad_norm": 2.9083821773529053,
      "learning_rate": 9.994944865352986e-06,
      "loss": 0.5333,
      "num_input_tokens_seen": 2501424,
      "step": 3830
    },
    {
      "epoch": 2.261202830188679,
      "grad_norm": 1.6034116744995117,
      "learning_rate": 9.994828545499351e-06,
      "loss": 0.461,
      "num_input_tokens_seen": 2504816,
      "step": 3835
    },
    {
      "epoch": 2.2641509433962264,
      "grad_norm": 1.83975350856781,
      "learning_rate": 9.994710903243334e-06,
      "loss": 0.6171,
      "num_input_tokens_seen": 2508272,
      "step": 3840
    },
    {
      "epoch": 2.2670990566037736,
      "grad_norm": 1.3657933473587036,
      "learning_rate": 9.994591938616079e-06,
      "loss": 0.5321,
      "num_input_tokens_seen": 2511664,
      "step": 3845
    },
    {
      "epoch": 2.270047169811321,
      "grad_norm": 3.2497334480285645,
      "learning_rate": 9.994471651649082e-06,
      "loss": 0.646,
      "num_input_tokens_seen": 2514384,
      "step": 3850
    },
    {
      "epoch": 2.2729952830188678,
      "grad_norm": 4.231931209564209,
      "learning_rate": 9.99435004237419e-06,
      "loss": 0.4872,
      "num_input_tokens_seen": 2517136,
      "step": 3855
    },
    {
      "epoch": 2.275943396226415,
      "grad_norm": 1.5508261919021606,
      "learning_rate": 9.9942271108236e-06,
      "loss": 0.3894,
      "num_input_tokens_seen": 2520464,
      "step": 3860
    },
    {
      "epoch": 2.2788915094339623,
      "grad_norm": 2.5334084033966064,
      "learning_rate": 9.994102857029859e-06,
      "loss": 0.5706,
      "num_input_tokens_seen": 2523568,
      "step": 3865
    },
    {
      "epoch": 2.2818396226415096,
      "grad_norm": 2.0650854110717773,
      "learning_rate": 9.993977281025862e-06,
      "loss": 0.55,
      "num_input_tokens_seen": 2526928,
      "step": 3870
    },
    {
      "epoch": 2.2847877358490565,
      "grad_norm": 2.080620288848877,
      "learning_rate": 9.993850382844858e-06,
      "loss": 0.53,
      "num_input_tokens_seen": 2530864,
      "step": 3875
    },
    {
      "epoch": 2.2877358490566038,
      "grad_norm": 1.3480745553970337,
      "learning_rate": 9.993722162520443e-06,
      "loss": 0.504,
      "num_input_tokens_seen": 2534992,
      "step": 3880
    },
    {
      "epoch": 2.290683962264151,
      "grad_norm": 2.177626848220825,
      "learning_rate": 9.993592620086564e-06,
      "loss": 0.6044,
      "num_input_tokens_seen": 2538032,
      "step": 3885
    },
    {
      "epoch": 2.293632075471698,
      "grad_norm": 1.4110749959945679,
      "learning_rate": 9.993461755577518e-06,
      "loss": 0.5696,
      "num_input_tokens_seen": 2540912,
      "step": 3890
    },
    {
      "epoch": 2.296580188679245,
      "grad_norm": 1.6280128955841064,
      "learning_rate": 9.99332956902795e-06,
      "loss": 0.3586,
      "num_input_tokens_seen": 2543600,
      "step": 3895
    },
    {
      "epoch": 2.2995283018867925,
      "grad_norm": 2.3977067470550537,
      "learning_rate": 9.993196060472859e-06,
      "loss": 0.5448,
      "num_input_tokens_seen": 2547408,
      "step": 3900
    },
    {
      "epoch": 2.3024764150943398,
      "grad_norm": 1.3785814046859741,
      "learning_rate": 9.993061229947591e-06,
      "loss": 0.4536,
      "num_input_tokens_seen": 2550352,
      "step": 3905
    },
    {
      "epoch": 2.3054245283018866,
      "grad_norm": 2.375364065170288,
      "learning_rate": 9.992925077487845e-06,
      "loss": 0.571,
      "num_input_tokens_seen": 2553456,
      "step": 3910
    },
    {
      "epoch": 2.308372641509434,
      "grad_norm": 2.570561170578003,
      "learning_rate": 9.992787603129666e-06,
      "loss": 0.5683,
      "num_input_tokens_seen": 2556560,
      "step": 3915
    },
    {
      "epoch": 2.311320754716981,
      "grad_norm": 3.5251641273498535,
      "learning_rate": 9.99264880690945e-06,
      "loss": 0.5614,
      "num_input_tokens_seen": 2560208,
      "step": 3920
    },
    {
      "epoch": 2.3142688679245285,
      "grad_norm": 4.972915172576904,
      "learning_rate": 9.992508688863947e-06,
      "loss": 0.5344,
      "num_input_tokens_seen": 2562768,
      "step": 3925
    },
    {
      "epoch": 2.3172169811320753,
      "grad_norm": 2.352879047393799,
      "learning_rate": 9.99236724903025e-06,
      "loss": 0.4333,
      "num_input_tokens_seen": 2566416,
      "step": 3930
    },
    {
      "epoch": 2.3201650943396226,
      "grad_norm": 13.549699783325195,
      "learning_rate": 9.992224487445809e-06,
      "loss": 0.5047,
      "num_input_tokens_seen": 2569616,
      "step": 3935
    },
    {
      "epoch": 2.32311320754717,
      "grad_norm": 3.3971519470214844,
      "learning_rate": 9.99208040414842e-06,
      "loss": 0.4573,
      "num_input_tokens_seen": 2572560,
      "step": 3940
    },
    {
      "epoch": 2.326061320754717,
      "grad_norm": 1.5874186754226685,
      "learning_rate": 9.99193499917623e-06,
      "loss": 0.4473,
      "num_input_tokens_seen": 2575312,
      "step": 3945
    },
    {
      "epoch": 2.329009433962264,
      "grad_norm": 1.1310614347457886,
      "learning_rate": 9.991788272567735e-06,
      "loss": 0.4401,
      "num_input_tokens_seen": 2578448,
      "step": 3950
    },
    {
      "epoch": 2.3319575471698113,
      "grad_norm": 1.1150015592575073,
      "learning_rate": 9.991640224361781e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 2582384,
      "step": 3955
    },
    {
      "epoch": 2.3349056603773586,
      "grad_norm": 1.9409315586090088,
      "learning_rate": 9.991490854597563e-06,
      "loss": 0.3575,
      "num_input_tokens_seen": 2585456,
      "step": 3960
    },
    {
      "epoch": 2.3378537735849054,
      "grad_norm": 2.227268934249878,
      "learning_rate": 9.991340163314632e-06,
      "loss": 0.4573,
      "num_input_tokens_seen": 2590448,
      "step": 3965
    },
    {
      "epoch": 2.3408018867924527,
      "grad_norm": 1.3155293464660645,
      "learning_rate": 9.991188150552878e-06,
      "loss": 0.5791,
      "num_input_tokens_seen": 2593872,
      "step": 3970
    },
    {
      "epoch": 2.34375,
      "grad_norm": 2.829507827758789,
      "learning_rate": 9.991034816352553e-06,
      "loss": 0.5234,
      "num_input_tokens_seen": 2597648,
      "step": 3975
    },
    {
      "epoch": 2.3466981132075473,
      "grad_norm": 1.6845955848693848,
      "learning_rate": 9.990880160754248e-06,
      "loss": 0.4496,
      "num_input_tokens_seen": 2600048,
      "step": 3980
    },
    {
      "epoch": 2.3496462264150946,
      "grad_norm": 2.4369006156921387,
      "learning_rate": 9.990724183798914e-06,
      "loss": 0.4534,
      "num_input_tokens_seen": 2603632,
      "step": 3985
    },
    {
      "epoch": 2.3525943396226414,
      "grad_norm": 1.6276899576187134,
      "learning_rate": 9.990566885527841e-06,
      "loss": 0.4845,
      "num_input_tokens_seen": 2607216,
      "step": 3990
    },
    {
      "epoch": 2.3555424528301887,
      "grad_norm": 1.5323408842086792,
      "learning_rate": 9.99040826598268e-06,
      "loss": 0.5368,
      "num_input_tokens_seen": 2610256,
      "step": 3995
    },
    {
      "epoch": 2.358490566037736,
      "grad_norm": 2.649141311645508,
      "learning_rate": 9.990248325205418e-06,
      "loss": 0.4007,
      "num_input_tokens_seen": 2613616,
      "step": 4000
    },
    {
      "epoch": 2.361438679245283,
      "grad_norm": 2.233700752258301,
      "learning_rate": 9.990087063238408e-06,
      "loss": 0.5299,
      "num_input_tokens_seen": 2617104,
      "step": 4005
    },
    {
      "epoch": 2.36438679245283,
      "grad_norm": 9.76878547668457,
      "learning_rate": 9.989924480124342e-06,
      "loss": 0.5633,
      "num_input_tokens_seen": 2619856,
      "step": 4010
    },
    {
      "epoch": 2.3673349056603774,
      "grad_norm": 1.9927561283111572,
      "learning_rate": 9.989760575906264e-06,
      "loss": 0.5369,
      "num_input_tokens_seen": 2623344,
      "step": 4015
    },
    {
      "epoch": 2.3702830188679247,
      "grad_norm": 3.241832733154297,
      "learning_rate": 9.98959535062757e-06,
      "loss": 0.5063,
      "num_input_tokens_seen": 2627024,
      "step": 4020
    },
    {
      "epoch": 2.3732311320754715,
      "grad_norm": 1.706433653831482,
      "learning_rate": 9.989428804332002e-06,
      "loss": 0.4324,
      "num_input_tokens_seen": 2629968,
      "step": 4025
    },
    {
      "epoch": 2.376179245283019,
      "grad_norm": 4.104156970977783,
      "learning_rate": 9.989260937063656e-06,
      "loss": 0.524,
      "num_input_tokens_seen": 2633328,
      "step": 4030
    },
    {
      "epoch": 2.379127358490566,
      "grad_norm": 1.3404372930526733,
      "learning_rate": 9.989091748866973e-06,
      "loss": 0.4958,
      "num_input_tokens_seen": 2637360,
      "step": 4035
    },
    {
      "epoch": 2.3820754716981134,
      "grad_norm": 2.513782024383545,
      "learning_rate": 9.988921239786748e-06,
      "loss": 0.436,
      "num_input_tokens_seen": 2640496,
      "step": 4040
    },
    {
      "epoch": 2.3850235849056602,
      "grad_norm": 2.727499485015869,
      "learning_rate": 9.988749409868124e-06,
      "loss": 0.5028,
      "num_input_tokens_seen": 2643280,
      "step": 4045
    },
    {
      "epoch": 2.3879716981132075,
      "grad_norm": 1.549669861793518,
      "learning_rate": 9.988576259156593e-06,
      "loss": 0.5083,
      "num_input_tokens_seen": 2646000,
      "step": 4050
    },
    {
      "epoch": 2.390919811320755,
      "grad_norm": 1.8130030632019043,
      "learning_rate": 9.988401787697996e-06,
      "loss": 0.4772,
      "num_input_tokens_seen": 2649520,
      "step": 4055
    },
    {
      "epoch": 2.393867924528302,
      "grad_norm": 2.0780134201049805,
      "learning_rate": 9.98822599553853e-06,
      "loss": 0.5954,
      "num_input_tokens_seen": 2652528,
      "step": 4060
    },
    {
      "epoch": 2.396816037735849,
      "grad_norm": 2.7208192348480225,
      "learning_rate": 9.988048882724732e-06,
      "loss": 0.519,
      "num_input_tokens_seen": 2655184,
      "step": 4065
    },
    {
      "epoch": 2.3997641509433962,
      "grad_norm": 3.3026835918426514,
      "learning_rate": 9.987870449303497e-06,
      "loss": 0.4735,
      "num_input_tokens_seen": 2658192,
      "step": 4070
    },
    {
      "epoch": 2.4027122641509435,
      "grad_norm": 2.592559814453125,
      "learning_rate": 9.98769069532206e-06,
      "loss": 0.4967,
      "num_input_tokens_seen": 2660944,
      "step": 4075
    },
    {
      "epoch": 2.4056603773584904,
      "grad_norm": 2.1517956256866455,
      "learning_rate": 9.98750962082802e-06,
      "loss": 0.5323,
      "num_input_tokens_seen": 2663280,
      "step": 4080
    },
    {
      "epoch": 2.4086084905660377,
      "grad_norm": 1.9323025941848755,
      "learning_rate": 9.987327225869312e-06,
      "loss": 0.6117,
      "num_input_tokens_seen": 2666576,
      "step": 4085
    },
    {
      "epoch": 2.411556603773585,
      "grad_norm": 1.5806341171264648,
      "learning_rate": 9.987143510494225e-06,
      "loss": 0.5078,
      "num_input_tokens_seen": 2670064,
      "step": 4090
    },
    {
      "epoch": 2.4145047169811322,
      "grad_norm": 1.1306326389312744,
      "learning_rate": 9.9869584747514e-06,
      "loss": 0.4714,
      "num_input_tokens_seen": 2673680,
      "step": 4095
    },
    {
      "epoch": 2.417452830188679,
      "grad_norm": 1.7992746829986572,
      "learning_rate": 9.986772118689828e-06,
      "loss": 0.535,
      "num_input_tokens_seen": 2676688,
      "step": 4100
    },
    {
      "epoch": 2.4204009433962264,
      "grad_norm": 1.7710566520690918,
      "learning_rate": 9.986584442358845e-06,
      "loss": 0.6292,
      "num_input_tokens_seen": 2681200,
      "step": 4105
    },
    {
      "epoch": 2.4233490566037736,
      "grad_norm": 1.5686014890670776,
      "learning_rate": 9.98639544580814e-06,
      "loss": 0.4655,
      "num_input_tokens_seen": 2684336,
      "step": 4110
    },
    {
      "epoch": 2.426297169811321,
      "grad_norm": 1.2640758752822876,
      "learning_rate": 9.98620512908775e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 2687344,
      "step": 4115
    },
    {
      "epoch": 2.4292452830188678,
      "grad_norm": 3.3218111991882324,
      "learning_rate": 9.986013492248064e-06,
      "loss": 0.3735,
      "num_input_tokens_seen": 2690224,
      "step": 4120
    },
    {
      "epoch": 2.432193396226415,
      "grad_norm": 1.7268750667572021,
      "learning_rate": 9.985820535339817e-06,
      "loss": 0.4501,
      "num_input_tokens_seen": 2693104,
      "step": 4125
    },
    {
      "epoch": 2.4351415094339623,
      "grad_norm": 1.0716211795806885,
      "learning_rate": 9.985626258414093e-06,
      "loss": 0.4762,
      "num_input_tokens_seen": 2695248,
      "step": 4130
    },
    {
      "epoch": 2.4380896226415096,
      "grad_norm": 1.8963731527328491,
      "learning_rate": 9.985430661522333e-06,
      "loss": 0.3703,
      "num_input_tokens_seen": 2698672,
      "step": 4135
    },
    {
      "epoch": 2.4410377358490565,
      "grad_norm": 3.09753155708313,
      "learning_rate": 9.98523374471632e-06,
      "loss": 0.5363,
      "num_input_tokens_seen": 2701264,
      "step": 4140
    },
    {
      "epoch": 2.4439858490566038,
      "grad_norm": 1.3349759578704834,
      "learning_rate": 9.985035508048186e-06,
      "loss": 0.5639,
      "num_input_tokens_seen": 2704464,
      "step": 4145
    },
    {
      "epoch": 2.446933962264151,
      "grad_norm": 2.703191041946411,
      "learning_rate": 9.984835951570417e-06,
      "loss": 0.509,
      "num_input_tokens_seen": 2708080,
      "step": 4150
    },
    {
      "epoch": 2.449882075471698,
      "grad_norm": 2.7819161415100098,
      "learning_rate": 9.984635075335847e-06,
      "loss": 0.5347,
      "num_input_tokens_seen": 2710832,
      "step": 4155
    },
    {
      "epoch": 2.452830188679245,
      "grad_norm": 1.8129849433898926,
      "learning_rate": 9.984432879397659e-06,
      "loss": 0.5451,
      "num_input_tokens_seen": 2713712,
      "step": 4160
    },
    {
      "epoch": 2.4557783018867925,
      "grad_norm": 2.087080717086792,
      "learning_rate": 9.984229363809383e-06,
      "loss": 0.4389,
      "num_input_tokens_seen": 2716688,
      "step": 4165
    },
    {
      "epoch": 2.4587264150943398,
      "grad_norm": 1.6445637941360474,
      "learning_rate": 9.984024528624904e-06,
      "loss": 0.3589,
      "num_input_tokens_seen": 2719440,
      "step": 4170
    },
    {
      "epoch": 2.4616745283018866,
      "grad_norm": 3.5430915355682373,
      "learning_rate": 9.98381837389845e-06,
      "loss": 0.4865,
      "num_input_tokens_seen": 2721840,
      "step": 4175
    },
    {
      "epoch": 2.464622641509434,
      "grad_norm": 1.6669857501983643,
      "learning_rate": 9.983610899684601e-06,
      "loss": 0.5398,
      "num_input_tokens_seen": 2726192,
      "step": 4180
    },
    {
      "epoch": 2.467570754716981,
      "grad_norm": 3.279939889907837,
      "learning_rate": 9.983402106038292e-06,
      "loss": 0.5636,
      "num_input_tokens_seen": 2728816,
      "step": 4185
    },
    {
      "epoch": 2.4705188679245285,
      "grad_norm": 7.45900821685791,
      "learning_rate": 9.983191993014793e-06,
      "loss": 0.5628,
      "num_input_tokens_seen": 2731344,
      "step": 4190
    },
    {
      "epoch": 2.4734669811320753,
      "grad_norm": 2.8235793113708496,
      "learning_rate": 9.982980560669742e-06,
      "loss": 0.4141,
      "num_input_tokens_seen": 2734096,
      "step": 4195
    },
    {
      "epoch": 2.4764150943396226,
      "grad_norm": 2.873528480529785,
      "learning_rate": 9.98276780905911e-06,
      "loss": 0.4932,
      "num_input_tokens_seen": 2737040,
      "step": 4200
    },
    {
      "epoch": 2.47936320754717,
      "grad_norm": 1.5973680019378662,
      "learning_rate": 9.982553738239225e-06,
      "loss": 0.4105,
      "num_input_tokens_seen": 2740528,
      "step": 4205
    },
    {
      "epoch": 2.482311320754717,
      "grad_norm": 2.9403560161590576,
      "learning_rate": 9.982338348266766e-06,
      "loss": 0.4165,
      "num_input_tokens_seen": 2743728,
      "step": 4210
    },
    {
      "epoch": 2.485259433962264,
      "grad_norm": 2.4394748210906982,
      "learning_rate": 9.982121639198756e-06,
      "loss": 0.5364,
      "num_input_tokens_seen": 2748464,
      "step": 4215
    },
    {
      "epoch": 2.4882075471698113,
      "grad_norm": 2.896411895751953,
      "learning_rate": 9.98190361109257e-06,
      "loss": 0.5718,
      "num_input_tokens_seen": 2752784,
      "step": 4220
    },
    {
      "epoch": 2.4911556603773586,
      "grad_norm": 1.8368418216705322,
      "learning_rate": 9.981684264005934e-06,
      "loss": 0.7493,
      "num_input_tokens_seen": 2755728,
      "step": 4225
    },
    {
      "epoch": 2.4941037735849054,
      "grad_norm": 1.5333503484725952,
      "learning_rate": 9.981463597996917e-06,
      "loss": 0.4883,
      "num_input_tokens_seen": 2758960,
      "step": 4230
    },
    {
      "epoch": 2.4970518867924527,
      "grad_norm": 3.050692319869995,
      "learning_rate": 9.981241613123944e-06,
      "loss": 0.532,
      "num_input_tokens_seen": 2762480,
      "step": 4235
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6425899267196655,
      "learning_rate": 9.981018309445785e-06,
      "loss": 0.406,
      "num_input_tokens_seen": 2765808,
      "step": 4240
    },
    {
      "epoch": 2.5029481132075473,
      "grad_norm": 6.3773112297058105,
      "learning_rate": 9.980793687021564e-06,
      "loss": 0.4636,
      "num_input_tokens_seen": 2768720,
      "step": 4245
    },
    {
      "epoch": 2.5058962264150946,
      "grad_norm": 2.54063081741333,
      "learning_rate": 9.980567745910746e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 2772624,
      "step": 4250
    },
    {
      "epoch": 2.5088443396226414,
      "grad_norm": 1.9294198751449585,
      "learning_rate": 9.980340486173155e-06,
      "loss": 0.76,
      "num_input_tokens_seen": 2776048,
      "step": 4255
    },
    {
      "epoch": 2.5117924528301887,
      "grad_norm": 2.7724790573120117,
      "learning_rate": 9.980111907868954e-06,
      "loss": 0.5102,
      "num_input_tokens_seen": 2778576,
      "step": 4260
    },
    {
      "epoch": 2.514740566037736,
      "grad_norm": 2.284367799758911,
      "learning_rate": 9.979882011058662e-06,
      "loss": 0.4261,
      "num_input_tokens_seen": 2781328,
      "step": 4265
    },
    {
      "epoch": 2.517688679245283,
      "grad_norm": 1.9525171518325806,
      "learning_rate": 9.979650795803146e-06,
      "loss": 0.5924,
      "num_input_tokens_seen": 2784432,
      "step": 4270
    },
    {
      "epoch": 2.52063679245283,
      "grad_norm": 1.3154902458190918,
      "learning_rate": 9.979418262163621e-06,
      "loss": 0.3881,
      "num_input_tokens_seen": 2787088,
      "step": 4275
    },
    {
      "epoch": 2.5235849056603774,
      "grad_norm": 2.033742904663086,
      "learning_rate": 9.979184410201652e-06,
      "loss": 0.4902,
      "num_input_tokens_seen": 2790000,
      "step": 4280
    },
    {
      "epoch": 2.5265330188679247,
      "grad_norm": 1.8792797327041626,
      "learning_rate": 9.97894923997915e-06,
      "loss": 0.4323,
      "num_input_tokens_seen": 2792688,
      "step": 4285
    },
    {
      "epoch": 2.5294811320754715,
      "grad_norm": 2.144305467605591,
      "learning_rate": 9.97871275155838e-06,
      "loss": 0.5608,
      "num_input_tokens_seen": 2796400,
      "step": 4290
    },
    {
      "epoch": 2.532429245283019,
      "grad_norm": 1.8108692169189453,
      "learning_rate": 9.978474945001949e-06,
      "loss": 0.5732,
      "num_input_tokens_seen": 2799280,
      "step": 4295
    },
    {
      "epoch": 2.535377358490566,
      "grad_norm": 1.6497257947921753,
      "learning_rate": 9.978235820372822e-06,
      "loss": 0.3648,
      "num_input_tokens_seen": 2802672,
      "step": 4300
    },
    {
      "epoch": 2.538325471698113,
      "grad_norm": 1.7847671508789062,
      "learning_rate": 9.977995377734307e-06,
      "loss": 0.5498,
      "num_input_tokens_seen": 2805360,
      "step": 4305
    },
    {
      "epoch": 2.5412735849056602,
      "grad_norm": 3.381925106048584,
      "learning_rate": 9.977753617150061e-06,
      "loss": 0.3648,
      "num_input_tokens_seen": 2809040,
      "step": 4310
    },
    {
      "epoch": 2.5442216981132075,
      "grad_norm": 2.541572093963623,
      "learning_rate": 9.977510538684094e-06,
      "loss": 0.5923,
      "num_input_tokens_seen": 2812080,
      "step": 4315
    },
    {
      "epoch": 2.547169811320755,
      "grad_norm": 3.1444692611694336,
      "learning_rate": 9.977266142400757e-06,
      "loss": 0.5526,
      "num_input_tokens_seen": 2816400,
      "step": 4320
    },
    {
      "epoch": 2.550117924528302,
      "grad_norm": 3.7041258811950684,
      "learning_rate": 9.977020428364759e-06,
      "loss": 0.5813,
      "num_input_tokens_seen": 2820144,
      "step": 4325
    },
    {
      "epoch": 2.553066037735849,
      "grad_norm": 2.6744472980499268,
      "learning_rate": 9.976773396641154e-06,
      "loss": 0.4183,
      "num_input_tokens_seen": 2823184,
      "step": 4330
    },
    {
      "epoch": 2.5560141509433962,
      "grad_norm": 2.6345441341400146,
      "learning_rate": 9.976525047295342e-06,
      "loss": 0.3819,
      "num_input_tokens_seen": 2826320,
      "step": 4335
    },
    {
      "epoch": 2.5589622641509435,
      "grad_norm": 1.8418720960617065,
      "learning_rate": 9.976275380393077e-06,
      "loss": 0.4535,
      "num_input_tokens_seen": 2830928,
      "step": 4340
    },
    {
      "epoch": 2.5619103773584904,
      "grad_norm": 3.202420711517334,
      "learning_rate": 9.976024396000459e-06,
      "loss": 0.5383,
      "num_input_tokens_seen": 2833968,
      "step": 4345
    },
    {
      "epoch": 2.5648584905660377,
      "grad_norm": 1.9077577590942383,
      "learning_rate": 9.975772094183935e-06,
      "loss": 0.5048,
      "num_input_tokens_seen": 2837392,
      "step": 4350
    },
    {
      "epoch": 2.567806603773585,
      "grad_norm": 2.1330316066741943,
      "learning_rate": 9.975518475010306e-06,
      "loss": 0.5276,
      "num_input_tokens_seen": 2840048,
      "step": 4355
    },
    {
      "epoch": 2.5707547169811322,
      "grad_norm": 2.8477582931518555,
      "learning_rate": 9.975263538546717e-06,
      "loss": 0.4091,
      "num_input_tokens_seen": 2842768,
      "step": 4360
    },
    {
      "epoch": 2.5737028301886795,
      "grad_norm": 1.9988538026809692,
      "learning_rate": 9.975007284860664e-06,
      "loss": 0.531,
      "num_input_tokens_seen": 2846512,
      "step": 4365
    },
    {
      "epoch": 2.5766509433962264,
      "grad_norm": 1.7546887397766113,
      "learning_rate": 9.974749714019993e-06,
      "loss": 0.6445,
      "num_input_tokens_seen": 2849808,
      "step": 4370
    },
    {
      "epoch": 2.5795990566037736,
      "grad_norm": 5.490547180175781,
      "learning_rate": 9.974490826092894e-06,
      "loss": 0.4806,
      "num_input_tokens_seen": 2853360,
      "step": 4375
    },
    {
      "epoch": 2.5825471698113205,
      "grad_norm": 1.0776605606079102,
      "learning_rate": 9.974230621147907e-06,
      "loss": 0.4182,
      "num_input_tokens_seen": 2857296,
      "step": 4380
    },
    {
      "epoch": 2.5854952830188678,
      "grad_norm": 2.1588449478149414,
      "learning_rate": 9.973969099253928e-06,
      "loss": 0.5549,
      "num_input_tokens_seen": 2861136,
      "step": 4385
    },
    {
      "epoch": 2.588443396226415,
      "grad_norm": 2.1320228576660156,
      "learning_rate": 9.973706260480194e-06,
      "loss": 0.4759,
      "num_input_tokens_seen": 2863888,
      "step": 4390
    },
    {
      "epoch": 2.5913915094339623,
      "grad_norm": 4.187750339508057,
      "learning_rate": 9.97344210489629e-06,
      "loss": 0.5552,
      "num_input_tokens_seen": 2867504,
      "step": 4395
    },
    {
      "epoch": 2.5943396226415096,
      "grad_norm": 1.5328081846237183,
      "learning_rate": 9.973176632572158e-06,
      "loss": 0.47,
      "num_input_tokens_seen": 2871056,
      "step": 4400
    },
    {
      "epoch": 2.5972877358490565,
      "grad_norm": 1.2889806032180786,
      "learning_rate": 9.972909843578076e-06,
      "loss": 0.5567,
      "num_input_tokens_seen": 2873712,
      "step": 4405
    },
    {
      "epoch": 2.6002358490566038,
      "grad_norm": 1.62148916721344,
      "learning_rate": 9.972641737984681e-06,
      "loss": 0.6168,
      "num_input_tokens_seen": 2876720,
      "step": 4410
    },
    {
      "epoch": 2.603183962264151,
      "grad_norm": 1.920217752456665,
      "learning_rate": 9.972372315862956e-06,
      "loss": 0.478,
      "num_input_tokens_seen": 2879664,
      "step": 4415
    },
    {
      "epoch": 2.606132075471698,
      "grad_norm": 2.1170432567596436,
      "learning_rate": 9.97210157728423e-06,
      "loss": 0.4594,
      "num_input_tokens_seen": 2882704,
      "step": 4420
    },
    {
      "epoch": 2.609080188679245,
      "grad_norm": 6.475421905517578,
      "learning_rate": 9.971829522320185e-06,
      "loss": 0.3238,
      "num_input_tokens_seen": 2890448,
      "step": 4425
    },
    {
      "epoch": 2.6120283018867925,
      "grad_norm": 2.1703567504882812,
      "learning_rate": 9.971556151042843e-06,
      "loss": 0.4571,
      "num_input_tokens_seen": 2893424,
      "step": 4430
    },
    {
      "epoch": 2.6149764150943398,
      "grad_norm": 1.6484168767929077,
      "learning_rate": 9.971281463524588e-06,
      "loss": 0.5833,
      "num_input_tokens_seen": 2899600,
      "step": 4435
    },
    {
      "epoch": 2.617924528301887,
      "grad_norm": 1.962673544883728,
      "learning_rate": 9.971005459838136e-06,
      "loss": 0.4556,
      "num_input_tokens_seen": 2903600,
      "step": 4440
    },
    {
      "epoch": 2.620872641509434,
      "grad_norm": 2.691267490386963,
      "learning_rate": 9.970728140056567e-06,
      "loss": 0.5396,
      "num_input_tokens_seen": 2907088,
      "step": 4445
    },
    {
      "epoch": 2.623820754716981,
      "grad_norm": 1.9610198736190796,
      "learning_rate": 9.9704495042533e-06,
      "loss": 0.4049,
      "num_input_tokens_seen": 2910672,
      "step": 4450
    },
    {
      "epoch": 2.6267688679245285,
      "grad_norm": 2.563339948654175,
      "learning_rate": 9.970169552502105e-06,
      "loss": 0.5402,
      "num_input_tokens_seen": 2913008,
      "step": 4455
    },
    {
      "epoch": 2.6297169811320753,
      "grad_norm": 2.2070553302764893,
      "learning_rate": 9.969888284877102e-06,
      "loss": 0.5486,
      "num_input_tokens_seen": 2916240,
      "step": 4460
    },
    {
      "epoch": 2.6326650943396226,
      "grad_norm": 2.5004208087921143,
      "learning_rate": 9.969605701452757e-06,
      "loss": 0.3449,
      "num_input_tokens_seen": 2919408,
      "step": 4465
    },
    {
      "epoch": 2.63561320754717,
      "grad_norm": 3.3195559978485107,
      "learning_rate": 9.969321802303882e-06,
      "loss": 0.4772,
      "num_input_tokens_seen": 2922224,
      "step": 4470
    },
    {
      "epoch": 2.638561320754717,
      "grad_norm": 2.4619383811950684,
      "learning_rate": 9.969036587505644e-06,
      "loss": 0.4397,
      "num_input_tokens_seen": 2925616,
      "step": 4475
    },
    {
      "epoch": 2.641509433962264,
      "grad_norm": 5.624481678009033,
      "learning_rate": 9.968750057133555e-06,
      "loss": 0.3676,
      "num_input_tokens_seen": 2928304,
      "step": 4480
    },
    {
      "epoch": 2.6444575471698113,
      "grad_norm": 3.099343776702881,
      "learning_rate": 9.968462211263474e-06,
      "loss": 0.5841,
      "num_input_tokens_seen": 2931440,
      "step": 4485
    },
    {
      "epoch": 2.6474056603773586,
      "grad_norm": 1.4520219564437866,
      "learning_rate": 9.96817304997161e-06,
      "loss": 0.3536,
      "num_input_tokens_seen": 2934416,
      "step": 4490
    },
    {
      "epoch": 2.6503537735849054,
      "grad_norm": 6.3251776695251465,
      "learning_rate": 9.967882573334519e-06,
      "loss": 0.5362,
      "num_input_tokens_seen": 2936912,
      "step": 4495
    },
    {
      "epoch": 2.6533018867924527,
      "grad_norm": 1.255062460899353,
      "learning_rate": 9.967590781429106e-06,
      "loss": 0.4731,
      "num_input_tokens_seen": 2940912,
      "step": 4500
    },
    {
      "epoch": 2.65625,
      "grad_norm": 5.443777084350586,
      "learning_rate": 9.967297674332625e-06,
      "loss": 0.5464,
      "num_input_tokens_seen": 2944336,
      "step": 4505
    },
    {
      "epoch": 2.6591981132075473,
      "grad_norm": 1.4844576120376587,
      "learning_rate": 9.967003252122675e-06,
      "loss": 0.4632,
      "num_input_tokens_seen": 2948208,
      "step": 4510
    },
    {
      "epoch": 2.6621462264150946,
      "grad_norm": 2.8074753284454346,
      "learning_rate": 9.96670751487721e-06,
      "loss": 0.5545,
      "num_input_tokens_seen": 2951088,
      "step": 4515
    },
    {
      "epoch": 2.6650943396226414,
      "grad_norm": 2.77970290184021,
      "learning_rate": 9.966410462674525e-06,
      "loss": 0.4656,
      "num_input_tokens_seen": 2954000,
      "step": 4520
    },
    {
      "epoch": 2.6680424528301887,
      "grad_norm": 2.996859550476074,
      "learning_rate": 9.966112095593264e-06,
      "loss": 0.4189,
      "num_input_tokens_seen": 2956624,
      "step": 4525
    },
    {
      "epoch": 2.670990566037736,
      "grad_norm": 2.449484348297119,
      "learning_rate": 9.965812413712425e-06,
      "loss": 0.5658,
      "num_input_tokens_seen": 2959568,
      "step": 4530
    },
    {
      "epoch": 2.673938679245283,
      "grad_norm": 1.4806125164031982,
      "learning_rate": 9.965511417111346e-06,
      "loss": 0.5916,
      "num_input_tokens_seen": 2962928,
      "step": 4535
    },
    {
      "epoch": 2.67688679245283,
      "grad_norm": 4.137581825256348,
      "learning_rate": 9.96520910586972e-06,
      "loss": 0.509,
      "num_input_tokens_seen": 2966352,
      "step": 4540
    },
    {
      "epoch": 2.6798349056603774,
      "grad_norm": 0.9742993116378784,
      "learning_rate": 9.964905480067585e-06,
      "loss": 0.3752,
      "num_input_tokens_seen": 2969584,
      "step": 4545
    },
    {
      "epoch": 2.6827830188679247,
      "grad_norm": 3.860938310623169,
      "learning_rate": 9.964600539785328e-06,
      "loss": 0.3916,
      "num_input_tokens_seen": 2972016,
      "step": 4550
    },
    {
      "epoch": 2.6857311320754715,
      "grad_norm": 3.1197290420532227,
      "learning_rate": 9.96429428510368e-06,
      "loss": 0.6724,
      "num_input_tokens_seen": 2974800,
      "step": 4555
    },
    {
      "epoch": 2.688679245283019,
      "grad_norm": 2.4082212448120117,
      "learning_rate": 9.963986716103724e-06,
      "loss": 0.4719,
      "num_input_tokens_seen": 2977136,
      "step": 4560
    },
    {
      "epoch": 2.691627358490566,
      "grad_norm": 2.8378469944000244,
      "learning_rate": 9.963677832866893e-06,
      "loss": 0.4485,
      "num_input_tokens_seen": 2979888,
      "step": 4565
    },
    {
      "epoch": 2.694575471698113,
      "grad_norm": 2.55816388130188,
      "learning_rate": 9.963367635474962e-06,
      "loss": 0.572,
      "num_input_tokens_seen": 2983248,
      "step": 4570
    },
    {
      "epoch": 2.6975235849056602,
      "grad_norm": 2.482215642929077,
      "learning_rate": 9.96305612401006e-06,
      "loss": 0.5488,
      "num_input_tokens_seen": 2986320,
      "step": 4575
    },
    {
      "epoch": 2.7004716981132075,
      "grad_norm": 6.601240158081055,
      "learning_rate": 9.96274329855466e-06,
      "loss": 0.6084,
      "num_input_tokens_seen": 2989392,
      "step": 4580
    },
    {
      "epoch": 2.703419811320755,
      "grad_norm": 3.119270086288452,
      "learning_rate": 9.962429159191583e-06,
      "loss": 0.5299,
      "num_input_tokens_seen": 2992400,
      "step": 4585
    },
    {
      "epoch": 2.706367924528302,
      "grad_norm": 4.138728618621826,
      "learning_rate": 9.962113706003997e-06,
      "loss": 0.4753,
      "num_input_tokens_seen": 2995056,
      "step": 4590
    },
    {
      "epoch": 2.709316037735849,
      "grad_norm": 2.0090062618255615,
      "learning_rate": 9.961796939075424e-06,
      "loss": 0.5708,
      "num_input_tokens_seen": 2998352,
      "step": 4595
    },
    {
      "epoch": 2.7122641509433962,
      "grad_norm": 2.013165235519409,
      "learning_rate": 9.961478858489728e-06,
      "loss": 0.4354,
      "num_input_tokens_seen": 3002256,
      "step": 4600
    },
    {
      "epoch": 2.7152122641509435,
      "grad_norm": 1.9014936685562134,
      "learning_rate": 9.961159464331119e-06,
      "loss": 0.5572,
      "num_input_tokens_seen": 3005008,
      "step": 4605
    },
    {
      "epoch": 2.7181603773584904,
      "grad_norm": 2.1144134998321533,
      "learning_rate": 9.960838756684161e-06,
      "loss": 0.4895,
      "num_input_tokens_seen": 3008272,
      "step": 4610
    },
    {
      "epoch": 2.7211084905660377,
      "grad_norm": 1.9594963788986206,
      "learning_rate": 9.960516735633764e-06,
      "loss": 0.5084,
      "num_input_tokens_seen": 3010864,
      "step": 4615
    },
    {
      "epoch": 2.724056603773585,
      "grad_norm": 2.1598689556121826,
      "learning_rate": 9.960193401265181e-06,
      "loss": 0.5067,
      "num_input_tokens_seen": 3013872,
      "step": 4620
    },
    {
      "epoch": 2.7270047169811322,
      "grad_norm": 2.671921968460083,
      "learning_rate": 9.959868753664018e-06,
      "loss": 0.6774,
      "num_input_tokens_seen": 3016912,
      "step": 4625
    },
    {
      "epoch": 2.7299528301886795,
      "grad_norm": 1.183402419090271,
      "learning_rate": 9.959542792916227e-06,
      "loss": 0.5242,
      "num_input_tokens_seen": 3020272,
      "step": 4630
    },
    {
      "epoch": 2.7329009433962264,
      "grad_norm": 1.7900513410568237,
      "learning_rate": 9.959215519108108e-06,
      "loss": 0.3996,
      "num_input_tokens_seen": 3024208,
      "step": 4635
    },
    {
      "epoch": 2.7358490566037736,
      "grad_norm": 2.524648666381836,
      "learning_rate": 9.958886932326306e-06,
      "loss": 0.5381,
      "num_input_tokens_seen": 3026960,
      "step": 4640
    },
    {
      "epoch": 2.7387971698113205,
      "grad_norm": 2.692333698272705,
      "learning_rate": 9.958557032657817e-06,
      "loss": 0.6327,
      "num_input_tokens_seen": 3029424,
      "step": 4645
    },
    {
      "epoch": 2.7417452830188678,
      "grad_norm": 1.5379713773727417,
      "learning_rate": 9.958225820189984e-06,
      "loss": 0.5178,
      "num_input_tokens_seen": 3033424,
      "step": 4650
    },
    {
      "epoch": 2.744693396226415,
      "grad_norm": 1.6538691520690918,
      "learning_rate": 9.957893295010495e-06,
      "loss": 0.4539,
      "num_input_tokens_seen": 3036112,
      "step": 4655
    },
    {
      "epoch": 2.7476415094339623,
      "grad_norm": 1.4671686887741089,
      "learning_rate": 9.957559457207391e-06,
      "loss": 0.4823,
      "num_input_tokens_seen": 3041872,
      "step": 4660
    },
    {
      "epoch": 2.7505896226415096,
      "grad_norm": 1.641699194908142,
      "learning_rate": 9.957224306869053e-06,
      "loss": 0.5121,
      "num_input_tokens_seen": 3044720,
      "step": 4665
    },
    {
      "epoch": 2.7535377358490565,
      "grad_norm": 7.5341620445251465,
      "learning_rate": 9.956887844084216e-06,
      "loss": 0.4737,
      "num_input_tokens_seen": 3047440,
      "step": 4670
    },
    {
      "epoch": 2.7564858490566038,
      "grad_norm": 1.8347375392913818,
      "learning_rate": 9.956550068941958e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 3050256,
      "step": 4675
    },
    {
      "epoch": 2.759433962264151,
      "grad_norm": 2.062375545501709,
      "learning_rate": 9.95621098153171e-06,
      "loss": 0.4572,
      "num_input_tokens_seen": 3053552,
      "step": 4680
    },
    {
      "epoch": 2.762382075471698,
      "grad_norm": 1.9609460830688477,
      "learning_rate": 9.955870581943243e-06,
      "loss": 0.4695,
      "num_input_tokens_seen": 3056304,
      "step": 4685
    },
    {
      "epoch": 2.765330188679245,
      "grad_norm": 6.357059478759766,
      "learning_rate": 9.955528870266681e-06,
      "loss": 0.5201,
      "num_input_tokens_seen": 3058960,
      "step": 4690
    },
    {
      "epoch": 2.7682783018867925,
      "grad_norm": 2.26557993888855,
      "learning_rate": 9.955185846592495e-06,
      "loss": 0.324,
      "num_input_tokens_seen": 3061616,
      "step": 4695
    },
    {
      "epoch": 2.7712264150943398,
      "grad_norm": 1.3830561637878418,
      "learning_rate": 9.9548415110115e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 3064880,
      "step": 4700
    },
    {
      "epoch": 2.774174528301887,
      "grad_norm": 1.7953895330429077,
      "learning_rate": 9.95449586361486e-06,
      "loss": 0.4798,
      "num_input_tokens_seen": 3067664,
      "step": 4705
    },
    {
      "epoch": 2.777122641509434,
      "grad_norm": 3.8955676555633545,
      "learning_rate": 9.954148904494085e-06,
      "loss": 0.5824,
      "num_input_tokens_seen": 3070064,
      "step": 4710
    },
    {
      "epoch": 2.780070754716981,
      "grad_norm": 1.786041259765625,
      "learning_rate": 9.95380063374104e-06,
      "loss": 0.5195,
      "num_input_tokens_seen": 3072848,
      "step": 4715
    },
    {
      "epoch": 2.7830188679245285,
      "grad_norm": 1.5143654346466064,
      "learning_rate": 9.953451051447927e-06,
      "loss": 0.5824,
      "num_input_tokens_seen": 3077168,
      "step": 4720
    },
    {
      "epoch": 2.7859669811320753,
      "grad_norm": 1.824815273284912,
      "learning_rate": 9.953100157707299e-06,
      "loss": 0.5513,
      "num_input_tokens_seen": 3080496,
      "step": 4725
    },
    {
      "epoch": 2.7889150943396226,
      "grad_norm": 4.569231033325195,
      "learning_rate": 9.952747952612056e-06,
      "loss": 0.6045,
      "num_input_tokens_seen": 3082608,
      "step": 4730
    },
    {
      "epoch": 2.79186320754717,
      "grad_norm": 2.402606248855591,
      "learning_rate": 9.952394436255451e-06,
      "loss": 0.6115,
      "num_input_tokens_seen": 3086128,
      "step": 4735
    },
    {
      "epoch": 2.794811320754717,
      "grad_norm": 1.223108172416687,
      "learning_rate": 9.952039608731072e-06,
      "loss": 0.5221,
      "num_input_tokens_seen": 3089424,
      "step": 4740
    },
    {
      "epoch": 2.797759433962264,
      "grad_norm": 1.2881407737731934,
      "learning_rate": 9.951683470132868e-06,
      "loss": 0.4511,
      "num_input_tokens_seen": 3092528,
      "step": 4745
    },
    {
      "epoch": 2.8007075471698113,
      "grad_norm": 1.4854148626327515,
      "learning_rate": 9.951326020555122e-06,
      "loss": 0.5285,
      "num_input_tokens_seen": 3095696,
      "step": 4750
    },
    {
      "epoch": 2.8036556603773586,
      "grad_norm": 3.1883795261383057,
      "learning_rate": 9.950967260092473e-06,
      "loss": 0.5104,
      "num_input_tokens_seen": 3098928,
      "step": 4755
    },
    {
      "epoch": 2.8066037735849054,
      "grad_norm": 2.4417901039123535,
      "learning_rate": 9.950607188839905e-06,
      "loss": 0.538,
      "num_input_tokens_seen": 3103024,
      "step": 4760
    },
    {
      "epoch": 2.8095518867924527,
      "grad_norm": 1.0449641942977905,
      "learning_rate": 9.950245806892749e-06,
      "loss": 0.4575,
      "num_input_tokens_seen": 3106480,
      "step": 4765
    },
    {
      "epoch": 2.8125,
      "grad_norm": 1.4901535511016846,
      "learning_rate": 9.94988311434668e-06,
      "loss": 0.4051,
      "num_input_tokens_seen": 3109840,
      "step": 4770
    },
    {
      "epoch": 2.8154481132075473,
      "grad_norm": 2.4330759048461914,
      "learning_rate": 9.949519111297723e-06,
      "loss": 0.4717,
      "num_input_tokens_seen": 3113072,
      "step": 4775
    },
    {
      "epoch": 2.8183962264150946,
      "grad_norm": 1.079999327659607,
      "learning_rate": 9.949153797842252e-06,
      "loss": 0.4592,
      "num_input_tokens_seen": 3116304,
      "step": 4780
    },
    {
      "epoch": 2.8213443396226414,
      "grad_norm": 1.1438630819320679,
      "learning_rate": 9.948787174076982e-06,
      "loss": 0.4717,
      "num_input_tokens_seen": 3119152,
      "step": 4785
    },
    {
      "epoch": 2.8242924528301887,
      "grad_norm": 1.6952239274978638,
      "learning_rate": 9.948419240098978e-06,
      "loss": 0.4545,
      "num_input_tokens_seen": 3122160,
      "step": 4790
    },
    {
      "epoch": 2.827240566037736,
      "grad_norm": 1.4051077365875244,
      "learning_rate": 9.948049996005657e-06,
      "loss": 0.3353,
      "num_input_tokens_seen": 3125296,
      "step": 4795
    },
    {
      "epoch": 2.830188679245283,
      "grad_norm": 1.6469001770019531,
      "learning_rate": 9.947679441894773e-06,
      "loss": 0.4559,
      "num_input_tokens_seen": 3128496,
      "step": 4800
    },
    {
      "epoch": 2.83313679245283,
      "grad_norm": 3.1621716022491455,
      "learning_rate": 9.947307577864433e-06,
      "loss": 0.5985,
      "num_input_tokens_seen": 3132272,
      "step": 4805
    },
    {
      "epoch": 2.8360849056603774,
      "grad_norm": 2.071216344833374,
      "learning_rate": 9.94693440401309e-06,
      "loss": 0.5325,
      "num_input_tokens_seen": 3135280,
      "step": 4810
    },
    {
      "epoch": 2.8390330188679247,
      "grad_norm": 2.6229159832000732,
      "learning_rate": 9.946559920439545e-06,
      "loss": 0.4734,
      "num_input_tokens_seen": 3138352,
      "step": 4815
    },
    {
      "epoch": 2.8419811320754715,
      "grad_norm": 2.319192886352539,
      "learning_rate": 9.946184127242942e-06,
      "loss": 0.4258,
      "num_input_tokens_seen": 3141520,
      "step": 4820
    },
    {
      "epoch": 2.844929245283019,
      "grad_norm": 1.5826201438903809,
      "learning_rate": 9.945807024522774e-06,
      "loss": 0.3801,
      "num_input_tokens_seen": 3143952,
      "step": 4825
    },
    {
      "epoch": 2.847877358490566,
      "grad_norm": 1.9659805297851562,
      "learning_rate": 9.945428612378881e-06,
      "loss": 0.5233,
      "num_input_tokens_seen": 3146736,
      "step": 4830
    },
    {
      "epoch": 2.850825471698113,
      "grad_norm": 3.3256261348724365,
      "learning_rate": 9.94504889091145e-06,
      "loss": 0.6951,
      "num_input_tokens_seen": 3148944,
      "step": 4835
    },
    {
      "epoch": 2.8537735849056602,
      "grad_norm": 7.192554473876953,
      "learning_rate": 9.944667860221013e-06,
      "loss": 0.4822,
      "num_input_tokens_seen": 3152432,
      "step": 4840
    },
    {
      "epoch": 2.8567216981132075,
      "grad_norm": 2.0407843589782715,
      "learning_rate": 9.944285520408448e-06,
      "loss": 0.4719,
      "num_input_tokens_seen": 3156688,
      "step": 4845
    },
    {
      "epoch": 2.859669811320755,
      "grad_norm": 1.4957295656204224,
      "learning_rate": 9.943901871574984e-06,
      "loss": 0.4947,
      "num_input_tokens_seen": 3159408,
      "step": 4850
    },
    {
      "epoch": 2.862617924528302,
      "grad_norm": 3.536105155944824,
      "learning_rate": 9.943516913822192e-06,
      "loss": 0.5738,
      "num_input_tokens_seen": 3163248,
      "step": 4855
    },
    {
      "epoch": 2.865566037735849,
      "grad_norm": 1.4576832056045532,
      "learning_rate": 9.943130647251994e-06,
      "loss": 0.4385,
      "num_input_tokens_seen": 3165584,
      "step": 4860
    },
    {
      "epoch": 2.8685141509433962,
      "grad_norm": 2.1835362911224365,
      "learning_rate": 9.94274307196665e-06,
      "loss": 0.3143,
      "num_input_tokens_seen": 3168656,
      "step": 4865
    },
    {
      "epoch": 2.8714622641509435,
      "grad_norm": 3.3358004093170166,
      "learning_rate": 9.942354188068778e-06,
      "loss": 0.363,
      "num_input_tokens_seen": 3172048,
      "step": 4870
    },
    {
      "epoch": 2.8744103773584904,
      "grad_norm": 2.5637922286987305,
      "learning_rate": 9.941963995661333e-06,
      "loss": 0.6071,
      "num_input_tokens_seen": 3174896,
      "step": 4875
    },
    {
      "epoch": 2.8773584905660377,
      "grad_norm": 1.916135311126709,
      "learning_rate": 9.941572494847622e-06,
      "loss": 0.5665,
      "num_input_tokens_seen": 3177488,
      "step": 4880
    },
    {
      "epoch": 2.880306603773585,
      "grad_norm": 3.585164785385132,
      "learning_rate": 9.941179685731297e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 3179440,
      "step": 4885
    },
    {
      "epoch": 2.8832547169811322,
      "grad_norm": 1.5545594692230225,
      "learning_rate": 9.940785568416354e-06,
      "loss": 0.4224,
      "num_input_tokens_seen": 3182896,
      "step": 4890
    },
    {
      "epoch": 2.8862028301886795,
      "grad_norm": 3.028679847717285,
      "learning_rate": 9.940390143007137e-06,
      "loss": 0.4948,
      "num_input_tokens_seen": 3186320,
      "step": 4895
    },
    {
      "epoch": 2.8891509433962264,
      "grad_norm": 1.2530959844589233,
      "learning_rate": 9.939993409608339e-06,
      "loss": 0.5518,
      "num_input_tokens_seen": 3189200,
      "step": 4900
    },
    {
      "epoch": 2.8920990566037736,
      "grad_norm": 1.295553207397461,
      "learning_rate": 9.939595368324996e-06,
      "loss": 0.474,
      "num_input_tokens_seen": 3193104,
      "step": 4905
    },
    {
      "epoch": 2.8950471698113205,
      "grad_norm": 1.7203960418701172,
      "learning_rate": 9.93919601926249e-06,
      "loss": 0.3859,
      "num_input_tokens_seen": 3195920,
      "step": 4910
    },
    {
      "epoch": 2.8979952830188678,
      "grad_norm": 1.2748231887817383,
      "learning_rate": 9.938795362526552e-06,
      "loss": 0.4464,
      "num_input_tokens_seen": 3200848,
      "step": 4915
    },
    {
      "epoch": 2.900943396226415,
      "grad_norm": 4.31773567199707,
      "learning_rate": 9.938393398223255e-06,
      "loss": 0.474,
      "num_input_tokens_seen": 3203984,
      "step": 4920
    },
    {
      "epoch": 2.9038915094339623,
      "grad_norm": 1.6798877716064453,
      "learning_rate": 9.937990126459024e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 3207824,
      "step": 4925
    },
    {
      "epoch": 2.9068396226415096,
      "grad_norm": 2.577207326889038,
      "learning_rate": 9.937585547340624e-06,
      "loss": 0.5225,
      "num_input_tokens_seen": 3210992,
      "step": 4930
    },
    {
      "epoch": 2.9097877358490565,
      "grad_norm": 1.8575819730758667,
      "learning_rate": 9.937179660975174e-06,
      "loss": 0.3623,
      "num_input_tokens_seen": 3214544,
      "step": 4935
    },
    {
      "epoch": 2.9127358490566038,
      "grad_norm": 2.627897024154663,
      "learning_rate": 9.936772467470127e-06,
      "loss": 0.4696,
      "num_input_tokens_seen": 3217424,
      "step": 4940
    },
    {
      "epoch": 2.915683962264151,
      "grad_norm": 1.8516827821731567,
      "learning_rate": 9.936363966933294e-06,
      "loss": 0.6843,
      "num_input_tokens_seen": 3220400,
      "step": 4945
    },
    {
      "epoch": 2.918632075471698,
      "grad_norm": 1.6161060333251953,
      "learning_rate": 9.935954159472828e-06,
      "loss": 0.375,
      "num_input_tokens_seen": 3223408,
      "step": 4950
    },
    {
      "epoch": 2.921580188679245,
      "grad_norm": 1.6432689428329468,
      "learning_rate": 9.935543045197222e-06,
      "loss": 0.418,
      "num_input_tokens_seen": 3226288,
      "step": 4955
    },
    {
      "epoch": 2.9245283018867925,
      "grad_norm": 2.416006565093994,
      "learning_rate": 9.935130624215326e-06,
      "loss": 0.5549,
      "num_input_tokens_seen": 3229296,
      "step": 4960
    },
    {
      "epoch": 2.9274764150943398,
      "grad_norm": 1.7391762733459473,
      "learning_rate": 9.934716896636329e-06,
      "loss": 0.5579,
      "num_input_tokens_seen": 3232048,
      "step": 4965
    },
    {
      "epoch": 2.930424528301887,
      "grad_norm": 3.599201202392578,
      "learning_rate": 9.934301862569764e-06,
      "loss": 0.6727,
      "num_input_tokens_seen": 3234832,
      "step": 4970
    },
    {
      "epoch": 2.933372641509434,
      "grad_norm": 8.448949813842773,
      "learning_rate": 9.933885522125517e-06,
      "loss": 0.6517,
      "num_input_tokens_seen": 3237648,
      "step": 4975
    },
    {
      "epoch": 2.936320754716981,
      "grad_norm": 1.8905072212219238,
      "learning_rate": 9.933467875413813e-06,
      "loss": 0.6168,
      "num_input_tokens_seen": 3240944,
      "step": 4980
    },
    {
      "epoch": 2.9392688679245285,
      "grad_norm": 0.8193274140357971,
      "learning_rate": 9.933048922545227e-06,
      "loss": 0.5366,
      "num_input_tokens_seen": 3244016,
      "step": 4985
    },
    {
      "epoch": 2.9422169811320753,
      "grad_norm": 1.5942493677139282,
      "learning_rate": 9.932628663630679e-06,
      "loss": 0.4305,
      "num_input_tokens_seen": 3246576,
      "step": 4990
    },
    {
      "epoch": 2.9451650943396226,
      "grad_norm": 3.5463485717773438,
      "learning_rate": 9.932207098781432e-06,
      "loss": 0.6666,
      "num_input_tokens_seen": 3250160,
      "step": 4995
    },
    {
      "epoch": 2.94811320754717,
      "grad_norm": 1.8341648578643799,
      "learning_rate": 9.931784228109102e-06,
      "loss": 0.5147,
      "num_input_tokens_seen": 3253296,
      "step": 5000
    },
    {
      "epoch": 2.951061320754717,
      "grad_norm": 2.0751190185546875,
      "learning_rate": 9.93136005172564e-06,
      "loss": 0.459,
      "num_input_tokens_seen": 3256496,
      "step": 5005
    },
    {
      "epoch": 2.954009433962264,
      "grad_norm": 1.4762810468673706,
      "learning_rate": 9.930934569743354e-06,
      "loss": 0.4601,
      "num_input_tokens_seen": 3259280,
      "step": 5010
    },
    {
      "epoch": 2.9569575471698113,
      "grad_norm": 1.3667514324188232,
      "learning_rate": 9.930507782274888e-06,
      "loss": 0.5588,
      "num_input_tokens_seen": 3264144,
      "step": 5015
    },
    {
      "epoch": 2.9599056603773586,
      "grad_norm": 2.6102676391601562,
      "learning_rate": 9.930079689433236e-06,
      "loss": 0.427,
      "num_input_tokens_seen": 3267440,
      "step": 5020
    },
    {
      "epoch": 2.9628537735849054,
      "grad_norm": 1.4854927062988281,
      "learning_rate": 9.92965029133174e-06,
      "loss": 0.5439,
      "num_input_tokens_seen": 3271760,
      "step": 5025
    },
    {
      "epoch": 2.9658018867924527,
      "grad_norm": 2.2654364109039307,
      "learning_rate": 9.929219588084084e-06,
      "loss": 0.4858,
      "num_input_tokens_seen": 3275216,
      "step": 5030
    },
    {
      "epoch": 2.96875,
      "grad_norm": 1.452226161956787,
      "learning_rate": 9.9287875798043e-06,
      "loss": 0.4602,
      "num_input_tokens_seen": 3278992,
      "step": 5035
    },
    {
      "epoch": 2.9716981132075473,
      "grad_norm": 4.526086807250977,
      "learning_rate": 9.92835426660676e-06,
      "loss": 0.5581,
      "num_input_tokens_seen": 3281776,
      "step": 5040
    },
    {
      "epoch": 2.9746462264150946,
      "grad_norm": 1.8233877420425415,
      "learning_rate": 9.927919648606188e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 3284304,
      "step": 5045
    },
    {
      "epoch": 2.9775943396226414,
      "grad_norm": 1.2043710947036743,
      "learning_rate": 9.927483725917652e-06,
      "loss": 0.3332,
      "num_input_tokens_seen": 3286896,
      "step": 5050
    },
    {
      "epoch": 2.9805424528301887,
      "grad_norm": 6.421375751495361,
      "learning_rate": 9.927046498656562e-06,
      "loss": 0.5722,
      "num_input_tokens_seen": 3290128,
      "step": 5055
    },
    {
      "epoch": 2.983490566037736,
      "grad_norm": 2.4058752059936523,
      "learning_rate": 9.926607966938679e-06,
      "loss": 0.597,
      "num_input_tokens_seen": 3293520,
      "step": 5060
    },
    {
      "epoch": 2.986438679245283,
      "grad_norm": 1.6264110803604126,
      "learning_rate": 9.926168130880103e-06,
      "loss": 0.4408,
      "num_input_tokens_seen": 3296624,
      "step": 5065
    },
    {
      "epoch": 2.98938679245283,
      "grad_norm": 2.234410524368286,
      "learning_rate": 9.925726990597283e-06,
      "loss": 0.5116,
      "num_input_tokens_seen": 3300080,
      "step": 5070
    },
    {
      "epoch": 2.9923349056603774,
      "grad_norm": 2.243040084838867,
      "learning_rate": 9.925284546207015e-06,
      "loss": 0.5711,
      "num_input_tokens_seen": 3303280,
      "step": 5075
    },
    {
      "epoch": 2.9952830188679247,
      "grad_norm": 4.0567731857299805,
      "learning_rate": 9.924840797826436e-06,
      "loss": 0.524,
      "num_input_tokens_seen": 3306128,
      "step": 5080
    },
    {
      "epoch": 2.9982311320754715,
      "grad_norm": 1.8744608163833618,
      "learning_rate": 9.924395745573029e-06,
      "loss": 0.644,
      "num_input_tokens_seen": 3309264,
      "step": 5085
    },
    {
      "epoch": 3.001179245283019,
      "grad_norm": 1.2985773086547852,
      "learning_rate": 9.923949389564629e-06,
      "loss": 0.572,
      "num_input_tokens_seen": 3312280,
      "step": 5090
    },
    {
      "epoch": 3.004127358490566,
      "grad_norm": 2.911729574203491,
      "learning_rate": 9.923501729919404e-06,
      "loss": 0.5164,
      "num_input_tokens_seen": 3314968,
      "step": 5095
    },
    {
      "epoch": 3.0070754716981134,
      "grad_norm": 1.4047752618789673,
      "learning_rate": 9.923052766755878e-06,
      "loss": 0.4092,
      "num_input_tokens_seen": 3318392,
      "step": 5100
    },
    {
      "epoch": 3.0100235849056602,
      "grad_norm": 1.6989785432815552,
      "learning_rate": 9.922602500192914e-06,
      "loss": 0.5324,
      "num_input_tokens_seen": 3321144,
      "step": 5105
    },
    {
      "epoch": 3.0129716981132075,
      "grad_norm": 1.2788058519363403,
      "learning_rate": 9.922150930349725e-06,
      "loss": 0.4799,
      "num_input_tokens_seen": 3323672,
      "step": 5110
    },
    {
      "epoch": 3.015919811320755,
      "grad_norm": 1.490939974784851,
      "learning_rate": 9.921698057345863e-06,
      "loss": 0.3751,
      "num_input_tokens_seen": 3327064,
      "step": 5115
    },
    {
      "epoch": 3.018867924528302,
      "grad_norm": 1.3321319818496704,
      "learning_rate": 9.921243881301229e-06,
      "loss": 0.499,
      "num_input_tokens_seen": 3330840,
      "step": 5120
    },
    {
      "epoch": 3.021816037735849,
      "grad_norm": 2.367957353591919,
      "learning_rate": 9.920788402336068e-06,
      "loss": 0.3978,
      "num_input_tokens_seen": 3334616,
      "step": 5125
    },
    {
      "epoch": 3.0247641509433962,
      "grad_norm": 1.01689612865448,
      "learning_rate": 9.92033162057097e-06,
      "loss": 0.4086,
      "num_input_tokens_seen": 3337944,
      "step": 5130
    },
    {
      "epoch": 3.0277122641509435,
      "grad_norm": 1.487201452255249,
      "learning_rate": 9.919873536126869e-06,
      "loss": 0.4396,
      "num_input_tokens_seen": 3341272,
      "step": 5135
    },
    {
      "epoch": 3.0306603773584904,
      "grad_norm": 1.277281641960144,
      "learning_rate": 9.919414149125046e-06,
      "loss": 0.4916,
      "num_input_tokens_seen": 3344344,
      "step": 5140
    },
    {
      "epoch": 3.0336084905660377,
      "grad_norm": 1.537855863571167,
      "learning_rate": 9.918953459687126e-06,
      "loss": 0.4229,
      "num_input_tokens_seen": 3347192,
      "step": 5145
    },
    {
      "epoch": 3.036556603773585,
      "grad_norm": 1.5782262086868286,
      "learning_rate": 9.918491467935078e-06,
      "loss": 0.4529,
      "num_input_tokens_seen": 3351224,
      "step": 5150
    },
    {
      "epoch": 3.0395047169811322,
      "grad_norm": 2.3634634017944336,
      "learning_rate": 9.918028173991218e-06,
      "loss": 0.4174,
      "num_input_tokens_seen": 3354072,
      "step": 5155
    },
    {
      "epoch": 3.042452830188679,
      "grad_norm": 1.8541532754898071,
      "learning_rate": 9.917563577978202e-06,
      "loss": 0.4613,
      "num_input_tokens_seen": 3356856,
      "step": 5160
    },
    {
      "epoch": 3.0454009433962264,
      "grad_norm": 0.9669644832611084,
      "learning_rate": 9.917097680019035e-06,
      "loss": 0.4816,
      "num_input_tokens_seen": 3360280,
      "step": 5165
    },
    {
      "epoch": 3.0483490566037736,
      "grad_norm": 2.462339401245117,
      "learning_rate": 9.916630480237066e-06,
      "loss": 0.4723,
      "num_input_tokens_seen": 3362968,
      "step": 5170
    },
    {
      "epoch": 3.051297169811321,
      "grad_norm": 1.8991142511367798,
      "learning_rate": 9.916161978755988e-06,
      "loss": 0.4736,
      "num_input_tokens_seen": 3366648,
      "step": 5175
    },
    {
      "epoch": 3.0542452830188678,
      "grad_norm": 2.372828722000122,
      "learning_rate": 9.915692175699838e-06,
      "loss": 0.4842,
      "num_input_tokens_seen": 3369304,
      "step": 5180
    },
    {
      "epoch": 3.057193396226415,
      "grad_norm": 2.6835107803344727,
      "learning_rate": 9.915221071193e-06,
      "loss": 0.5507,
      "num_input_tokens_seen": 3372504,
      "step": 5185
    },
    {
      "epoch": 3.0601415094339623,
      "grad_norm": 2.030510663986206,
      "learning_rate": 9.914748665360199e-06,
      "loss": 0.4882,
      "num_input_tokens_seen": 3375096,
      "step": 5190
    },
    {
      "epoch": 3.0630896226415096,
      "grad_norm": 10.450382232666016,
      "learning_rate": 9.914274958326507e-06,
      "loss": 0.6181,
      "num_input_tokens_seen": 3377592,
      "step": 5195
    },
    {
      "epoch": 3.0660377358490565,
      "grad_norm": 1.533773422241211,
      "learning_rate": 9.913799950217341e-06,
      "loss": 0.5641,
      "num_input_tokens_seen": 3380728,
      "step": 5200
    },
    {
      "epoch": 3.0689858490566038,
      "grad_norm": 3.3019399642944336,
      "learning_rate": 9.91332364115846e-06,
      "loss": 0.4957,
      "num_input_tokens_seen": 3383416,
      "step": 5205
    },
    {
      "epoch": 3.071933962264151,
      "grad_norm": 1.7606898546218872,
      "learning_rate": 9.912846031275972e-06,
      "loss": 0.5678,
      "num_input_tokens_seen": 3386904,
      "step": 5210
    },
    {
      "epoch": 3.074882075471698,
      "grad_norm": 2.5911142826080322,
      "learning_rate": 9.912367120696322e-06,
      "loss": 0.403,
      "num_input_tokens_seen": 3389496,
      "step": 5215
    },
    {
      "epoch": 3.077830188679245,
      "grad_norm": 2.262333631515503,
      "learning_rate": 9.911886909546307e-06,
      "loss": 0.4631,
      "num_input_tokens_seen": 3392856,
      "step": 5220
    },
    {
      "epoch": 3.0807783018867925,
      "grad_norm": 1.7548032999038696,
      "learning_rate": 9.911405397953063e-06,
      "loss": 0.4369,
      "num_input_tokens_seen": 3396280,
      "step": 5225
    },
    {
      "epoch": 3.0837264150943398,
      "grad_norm": 1.9484949111938477,
      "learning_rate": 9.910922586044073e-06,
      "loss": 0.4041,
      "num_input_tokens_seen": 3400216,
      "step": 5230
    },
    {
      "epoch": 3.0866745283018866,
      "grad_norm": 4.740269660949707,
      "learning_rate": 9.910438473947163e-06,
      "loss": 0.4477,
      "num_input_tokens_seen": 3402744,
      "step": 5235
    },
    {
      "epoch": 3.089622641509434,
      "grad_norm": 1.7629338502883911,
      "learning_rate": 9.909953061790506e-06,
      "loss": 0.6051,
      "num_input_tokens_seen": 3405816,
      "step": 5240
    },
    {
      "epoch": 3.092570754716981,
      "grad_norm": 2.18304705619812,
      "learning_rate": 9.909466349702613e-06,
      "loss": 0.5483,
      "num_input_tokens_seen": 3408824,
      "step": 5245
    },
    {
      "epoch": 3.0955188679245285,
      "grad_norm": 2.186913013458252,
      "learning_rate": 9.908978337812348e-06,
      "loss": 0.4303,
      "num_input_tokens_seen": 3412504,
      "step": 5250
    },
    {
      "epoch": 3.0984669811320753,
      "grad_norm": 1.807883620262146,
      "learning_rate": 9.908489026248909e-06,
      "loss": 0.6035,
      "num_input_tokens_seen": 3415192,
      "step": 5255
    },
    {
      "epoch": 3.1014150943396226,
      "grad_norm": 2.9743282794952393,
      "learning_rate": 9.907998415141846e-06,
      "loss": 0.4859,
      "num_input_tokens_seen": 3418456,
      "step": 5260
    },
    {
      "epoch": 3.10436320754717,
      "grad_norm": 1.8896187543869019,
      "learning_rate": 9.907506504621052e-06,
      "loss": 0.4454,
      "num_input_tokens_seen": 3421560,
      "step": 5265
    },
    {
      "epoch": 3.107311320754717,
      "grad_norm": 6.483303546905518,
      "learning_rate": 9.907013294816759e-06,
      "loss": 0.5021,
      "num_input_tokens_seen": 3425272,
      "step": 5270
    },
    {
      "epoch": 3.110259433962264,
      "grad_norm": 1.8346004486083984,
      "learning_rate": 9.906518785859548e-06,
      "loss": 0.4863,
      "num_input_tokens_seen": 3428216,
      "step": 5275
    },
    {
      "epoch": 3.1132075471698113,
      "grad_norm": 1.776559591293335,
      "learning_rate": 9.906022977880344e-06,
      "loss": 0.3763,
      "num_input_tokens_seen": 3431032,
      "step": 5280
    },
    {
      "epoch": 3.1161556603773586,
      "grad_norm": 1.3047335147857666,
      "learning_rate": 9.905525871010412e-06,
      "loss": 0.3967,
      "num_input_tokens_seen": 3434104,
      "step": 5285
    },
    {
      "epoch": 3.119103773584906,
      "grad_norm": 2.0682482719421387,
      "learning_rate": 9.905027465381363e-06,
      "loss": 0.4336,
      "num_input_tokens_seen": 3436824,
      "step": 5290
    },
    {
      "epoch": 3.1220518867924527,
      "grad_norm": 4.566401481628418,
      "learning_rate": 9.904527761125155e-06,
      "loss": 0.6082,
      "num_input_tokens_seen": 3439352,
      "step": 5295
    },
    {
      "epoch": 3.125,
      "grad_norm": 1.373267650604248,
      "learning_rate": 9.904026758374083e-06,
      "loss": 0.381,
      "num_input_tokens_seen": 3442680,
      "step": 5300
    },
    {
      "epoch": 3.1279481132075473,
      "grad_norm": 3.013223648071289,
      "learning_rate": 9.903524457260794e-06,
      "loss": 0.4136,
      "num_input_tokens_seen": 3445944,
      "step": 5305
    },
    {
      "epoch": 3.1308962264150946,
      "grad_norm": 1.5326875448226929,
      "learning_rate": 9.90302085791827e-06,
      "loss": 0.4896,
      "num_input_tokens_seen": 3449080,
      "step": 5310
    },
    {
      "epoch": 3.1338443396226414,
      "grad_norm": 1.798431158065796,
      "learning_rate": 9.902515960479844e-06,
      "loss": 0.5089,
      "num_input_tokens_seen": 3453208,
      "step": 5315
    },
    {
      "epoch": 3.1367924528301887,
      "grad_norm": 2.233085870742798,
      "learning_rate": 9.902009765079188e-06,
      "loss": 0.5379,
      "num_input_tokens_seen": 3456856,
      "step": 5320
    },
    {
      "epoch": 3.139740566037736,
      "grad_norm": 2.2206714153289795,
      "learning_rate": 9.90150227185032e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 3459672,
      "step": 5325
    },
    {
      "epoch": 3.142688679245283,
      "grad_norm": 2.3948395252227783,
      "learning_rate": 9.900993480927603e-06,
      "loss": 0.5942,
      "num_input_tokens_seen": 3462040,
      "step": 5330
    },
    {
      "epoch": 3.14563679245283,
      "grad_norm": 2.5827414989471436,
      "learning_rate": 9.90048339244574e-06,
      "loss": 0.5012,
      "num_input_tokens_seen": 3464856,
      "step": 5335
    },
    {
      "epoch": 3.1485849056603774,
      "grad_norm": 2.1674084663391113,
      "learning_rate": 9.899972006539776e-06,
      "loss": 0.4121,
      "num_input_tokens_seen": 3467928,
      "step": 5340
    },
    {
      "epoch": 3.1515330188679247,
      "grad_norm": 2.314133882522583,
      "learning_rate": 9.899459323345106e-06,
      "loss": 0.5219,
      "num_input_tokens_seen": 3471096,
      "step": 5345
    },
    {
      "epoch": 3.1544811320754715,
      "grad_norm": 2.3987629413604736,
      "learning_rate": 9.898945342997467e-06,
      "loss": 0.4654,
      "num_input_tokens_seen": 3474808,
      "step": 5350
    },
    {
      "epoch": 3.157429245283019,
      "grad_norm": 2.3828718662261963,
      "learning_rate": 9.898430065632933e-06,
      "loss": 0.3936,
      "num_input_tokens_seen": 3478296,
      "step": 5355
    },
    {
      "epoch": 3.160377358490566,
      "grad_norm": 1.3809982538223267,
      "learning_rate": 9.897913491387929e-06,
      "loss": 0.4716,
      "num_input_tokens_seen": 3481304,
      "step": 5360
    },
    {
      "epoch": 3.1633254716981134,
      "grad_norm": 1.3537770509719849,
      "learning_rate": 9.897395620399219e-06,
      "loss": 0.5651,
      "num_input_tokens_seen": 3486200,
      "step": 5365
    },
    {
      "epoch": 3.1662735849056602,
      "grad_norm": 2.0660784244537354,
      "learning_rate": 9.896876452803913e-06,
      "loss": 0.5685,
      "num_input_tokens_seen": 3489784,
      "step": 5370
    },
    {
      "epoch": 3.1692216981132075,
      "grad_norm": 1.3299871683120728,
      "learning_rate": 9.896355988739461e-06,
      "loss": 0.4432,
      "num_input_tokens_seen": 3492824,
      "step": 5375
    },
    {
      "epoch": 3.172169811320755,
      "grad_norm": 1.5161328315734863,
      "learning_rate": 9.895834228343658e-06,
      "loss": 0.5175,
      "num_input_tokens_seen": 3495736,
      "step": 5380
    },
    {
      "epoch": 3.175117924528302,
      "grad_norm": 2.922492027282715,
      "learning_rate": 9.895311171754644e-06,
      "loss": 0.5193,
      "num_input_tokens_seen": 3498616,
      "step": 5385
    },
    {
      "epoch": 3.178066037735849,
      "grad_norm": 2.018202066421509,
      "learning_rate": 9.8947868191109e-06,
      "loss": 0.5631,
      "num_input_tokens_seen": 3501976,
      "step": 5390
    },
    {
      "epoch": 3.1810141509433962,
      "grad_norm": 3.446112632751465,
      "learning_rate": 9.894261170551249e-06,
      "loss": 0.4348,
      "num_input_tokens_seen": 3507096,
      "step": 5395
    },
    {
      "epoch": 3.1839622641509435,
      "grad_norm": 1.0192890167236328,
      "learning_rate": 9.893734226214861e-06,
      "loss": 0.4042,
      "num_input_tokens_seen": 3510776,
      "step": 5400
    },
    {
      "epoch": 3.1869103773584904,
      "grad_norm": 8.99704647064209,
      "learning_rate": 9.893205986241246e-06,
      "loss": 0.4858,
      "num_input_tokens_seen": 3514712,
      "step": 5405
    },
    {
      "epoch": 3.1898584905660377,
      "grad_norm": 0.9235435724258423,
      "learning_rate": 9.892676450770257e-06,
      "loss": 0.3711,
      "num_input_tokens_seen": 3517688,
      "step": 5410
    },
    {
      "epoch": 3.192806603773585,
      "grad_norm": 1.7647525072097778,
      "learning_rate": 9.892145619942092e-06,
      "loss": 0.3671,
      "num_input_tokens_seen": 3520760,
      "step": 5415
    },
    {
      "epoch": 3.1957547169811322,
      "grad_norm": 2.5886619091033936,
      "learning_rate": 9.891613493897289e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 3523672,
      "step": 5420
    },
    {
      "epoch": 3.198702830188679,
      "grad_norm": 2.938159704208374,
      "learning_rate": 9.891080072776733e-06,
      "loss": 0.6848,
      "num_input_tokens_seen": 3526552,
      "step": 5425
    },
    {
      "epoch": 3.2016509433962264,
      "grad_norm": 1.8357021808624268,
      "learning_rate": 9.890545356721649e-06,
      "loss": 0.5061,
      "num_input_tokens_seen": 3530552,
      "step": 5430
    },
    {
      "epoch": 3.2045990566037736,
      "grad_norm": 3.6945109367370605,
      "learning_rate": 9.890009345873603e-06,
      "loss": 0.4507,
      "num_input_tokens_seen": 3533656,
      "step": 5435
    },
    {
      "epoch": 3.207547169811321,
      "grad_norm": 1.415722131729126,
      "learning_rate": 9.889472040374509e-06,
      "loss": 0.3722,
      "num_input_tokens_seen": 3536632,
      "step": 5440
    },
    {
      "epoch": 3.2104952830188678,
      "grad_norm": 2.3702099323272705,
      "learning_rate": 9.88893344036662e-06,
      "loss": 0.4766,
      "num_input_tokens_seen": 3539256,
      "step": 5445
    },
    {
      "epoch": 3.213443396226415,
      "grad_norm": 4.88722038269043,
      "learning_rate": 9.888393545992531e-06,
      "loss": 0.5999,
      "num_input_tokens_seen": 3541624,
      "step": 5450
    },
    {
      "epoch": 3.2163915094339623,
      "grad_norm": 0.7700355648994446,
      "learning_rate": 9.887852357395184e-06,
      "loss": 0.3474,
      "num_input_tokens_seen": 3545272,
      "step": 5455
    },
    {
      "epoch": 3.2193396226415096,
      "grad_norm": 2.5676627159118652,
      "learning_rate": 9.88730987471786e-06,
      "loss": 0.4934,
      "num_input_tokens_seen": 3548760,
      "step": 5460
    },
    {
      "epoch": 3.2222877358490565,
      "grad_norm": 1.616018295288086,
      "learning_rate": 9.886766098104183e-06,
      "loss": 0.5045,
      "num_input_tokens_seen": 3553208,
      "step": 5465
    },
    {
      "epoch": 3.2252358490566038,
      "grad_norm": 1.3842121362686157,
      "learning_rate": 9.886221027698122e-06,
      "loss": 0.4052,
      "num_input_tokens_seen": 3555800,
      "step": 5470
    },
    {
      "epoch": 3.228183962264151,
      "grad_norm": 1.9846680164337158,
      "learning_rate": 9.885674663643983e-06,
      "loss": 0.6302,
      "num_input_tokens_seen": 3558616,
      "step": 5475
    },
    {
      "epoch": 3.231132075471698,
      "grad_norm": 1.8711097240447998,
      "learning_rate": 9.885127006086423e-06,
      "loss": 0.734,
      "num_input_tokens_seen": 3562264,
      "step": 5480
    },
    {
      "epoch": 3.234080188679245,
      "grad_norm": 2.813077449798584,
      "learning_rate": 9.884578055170434e-06,
      "loss": 0.5638,
      "num_input_tokens_seen": 3565912,
      "step": 5485
    },
    {
      "epoch": 3.2370283018867925,
      "grad_norm": 1.1934574842453003,
      "learning_rate": 9.884027811041353e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 3570200,
      "step": 5490
    },
    {
      "epoch": 3.2399764150943398,
      "grad_norm": 2.346193313598633,
      "learning_rate": 9.883476273844861e-06,
      "loss": 0.5261,
      "num_input_tokens_seen": 3572952,
      "step": 5495
    },
    {
      "epoch": 3.2429245283018866,
      "grad_norm": 1.7948797941207886,
      "learning_rate": 9.882923443726977e-06,
      "loss": 0.3781,
      "num_input_tokens_seen": 3575800,
      "step": 5500
    },
    {
      "epoch": 3.245872641509434,
      "grad_norm": 1.5508087873458862,
      "learning_rate": 9.882369320834068e-06,
      "loss": 0.4212,
      "num_input_tokens_seen": 3579256,
      "step": 5505
    },
    {
      "epoch": 3.248820754716981,
      "grad_norm": 5.076117515563965,
      "learning_rate": 9.88181390531284e-06,
      "loss": 0.4401,
      "num_input_tokens_seen": 3583160,
      "step": 5510
    },
    {
      "epoch": 3.2517688679245285,
      "grad_norm": 1.8209553956985474,
      "learning_rate": 9.88125719731034e-06,
      "loss": 0.4397,
      "num_input_tokens_seen": 3587448,
      "step": 5515
    },
    {
      "epoch": 3.2547169811320753,
      "grad_norm": 1.63091242313385,
      "learning_rate": 9.880699196973962e-06,
      "loss": 0.4377,
      "num_input_tokens_seen": 3591160,
      "step": 5520
    },
    {
      "epoch": 3.2576650943396226,
      "grad_norm": 1.5736274719238281,
      "learning_rate": 9.880139904451436e-06,
      "loss": 0.4794,
      "num_input_tokens_seen": 3593880,
      "step": 5525
    },
    {
      "epoch": 3.26061320754717,
      "grad_norm": 1.3822582960128784,
      "learning_rate": 9.879579319890838e-06,
      "loss": 0.4034,
      "num_input_tokens_seen": 3596792,
      "step": 5530
    },
    {
      "epoch": 3.263561320754717,
      "grad_norm": 2.072382688522339,
      "learning_rate": 9.879017443440584e-06,
      "loss": 0.4078,
      "num_input_tokens_seen": 3599928,
      "step": 5535
    },
    {
      "epoch": 3.266509433962264,
      "grad_norm": 0.9391538500785828,
      "learning_rate": 9.878454275249436e-06,
      "loss": 0.4507,
      "num_input_tokens_seen": 3602872,
      "step": 5540
    },
    {
      "epoch": 3.2694575471698113,
      "grad_norm": 4.76472282409668,
      "learning_rate": 9.877889815466493e-06,
      "loss": 0.4563,
      "num_input_tokens_seen": 3606328,
      "step": 5545
    },
    {
      "epoch": 3.2724056603773586,
      "grad_norm": 1.3194336891174316,
      "learning_rate": 9.877324064241198e-06,
      "loss": 0.4057,
      "num_input_tokens_seen": 3609400,
      "step": 5550
    },
    {
      "epoch": 3.2753537735849054,
      "grad_norm": 3.9348933696746826,
      "learning_rate": 9.876757021723338e-06,
      "loss": 0.5476,
      "num_input_tokens_seen": 3612216,
      "step": 5555
    },
    {
      "epoch": 3.2783018867924527,
      "grad_norm": 1.4248347282409668,
      "learning_rate": 9.876188688063038e-06,
      "loss": 0.3909,
      "num_input_tokens_seen": 3616408,
      "step": 5560
    },
    {
      "epoch": 3.28125,
      "grad_norm": 3.1484227180480957,
      "learning_rate": 9.875619063410768e-06,
      "loss": 0.4765,
      "num_input_tokens_seen": 3620216,
      "step": 5565
    },
    {
      "epoch": 3.2841981132075473,
      "grad_norm": 1.2532635927200317,
      "learning_rate": 9.875048147917339e-06,
      "loss": 0.6603,
      "num_input_tokens_seen": 3623384,
      "step": 5570
    },
    {
      "epoch": 3.2871462264150946,
      "grad_norm": 1.9907788038253784,
      "learning_rate": 9.874475941733902e-06,
      "loss": 0.3378,
      "num_input_tokens_seen": 3626712,
      "step": 5575
    },
    {
      "epoch": 3.2900943396226414,
      "grad_norm": 1.4493025541305542,
      "learning_rate": 9.873902445011952e-06,
      "loss": 0.4546,
      "num_input_tokens_seen": 3629752,
      "step": 5580
    },
    {
      "epoch": 3.2930424528301887,
      "grad_norm": 2.23189640045166,
      "learning_rate": 9.873327657903324e-06,
      "loss": 0.5013,
      "num_input_tokens_seen": 3632536,
      "step": 5585
    },
    {
      "epoch": 3.295990566037736,
      "grad_norm": 2.6468918323516846,
      "learning_rate": 9.872751580560194e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 3635768,
      "step": 5590
    },
    {
      "epoch": 3.298938679245283,
      "grad_norm": 1.497018575668335,
      "learning_rate": 9.872174213135084e-06,
      "loss": 0.3771,
      "num_input_tokens_seen": 3638424,
      "step": 5595
    },
    {
      "epoch": 3.30188679245283,
      "grad_norm": 1.6334091424942017,
      "learning_rate": 9.871595555780855e-06,
      "loss": 0.4962,
      "num_input_tokens_seen": 3641912,
      "step": 5600
    },
    {
      "epoch": 3.3048349056603774,
      "grad_norm": 5.743154525756836,
      "learning_rate": 9.871015608650705e-06,
      "loss": 0.4433,
      "num_input_tokens_seen": 3644856,
      "step": 5605
    },
    {
      "epoch": 3.3077830188679247,
      "grad_norm": 0.9836852550506592,
      "learning_rate": 9.870434371898182e-06,
      "loss": 0.5282,
      "num_input_tokens_seen": 3647960,
      "step": 5610
    },
    {
      "epoch": 3.3107311320754715,
      "grad_norm": 1.7344549894332886,
      "learning_rate": 9.869851845677165e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 3651416,
      "step": 5615
    },
    {
      "epoch": 3.313679245283019,
      "grad_norm": 2.3151936531066895,
      "learning_rate": 9.869268030141886e-06,
      "loss": 0.4013,
      "num_input_tokens_seen": 3655192,
      "step": 5620
    },
    {
      "epoch": 3.316627358490566,
      "grad_norm": 3.1920158863067627,
      "learning_rate": 9.86868292544691e-06,
      "loss": 0.4207,
      "num_input_tokens_seen": 3658200,
      "step": 5625
    },
    {
      "epoch": 3.3195754716981134,
      "grad_norm": 2.742058753967285,
      "learning_rate": 9.868096531747149e-06,
      "loss": 0.5041,
      "num_input_tokens_seen": 3661720,
      "step": 5630
    },
    {
      "epoch": 3.3225235849056602,
      "grad_norm": 1.7880358695983887,
      "learning_rate": 9.867508849197848e-06,
      "loss": 0.4718,
      "num_input_tokens_seen": 3664216,
      "step": 5635
    },
    {
      "epoch": 3.3254716981132075,
      "grad_norm": 1.2627674341201782,
      "learning_rate": 9.866919877954602e-06,
      "loss": 0.5067,
      "num_input_tokens_seen": 3667800,
      "step": 5640
    },
    {
      "epoch": 3.328419811320755,
      "grad_norm": 3.6054415702819824,
      "learning_rate": 9.866329618173344e-06,
      "loss": 0.485,
      "num_input_tokens_seen": 3671800,
      "step": 5645
    },
    {
      "epoch": 3.331367924528302,
      "grad_norm": 2.8440802097320557,
      "learning_rate": 9.865738070010346e-06,
      "loss": 0.4444,
      "num_input_tokens_seen": 3674424,
      "step": 5650
    },
    {
      "epoch": 3.334316037735849,
      "grad_norm": 1.8752282857894897,
      "learning_rate": 9.865145233622223e-06,
      "loss": 0.5183,
      "num_input_tokens_seen": 3677144,
      "step": 5655
    },
    {
      "epoch": 3.3372641509433962,
      "grad_norm": 3.0664122104644775,
      "learning_rate": 9.864551109165935e-06,
      "loss": 0.6434,
      "num_input_tokens_seen": 3679768,
      "step": 5660
    },
    {
      "epoch": 3.3402122641509435,
      "grad_norm": 1.25641930103302,
      "learning_rate": 9.863955696798773e-06,
      "loss": 0.4124,
      "num_input_tokens_seen": 3687576,
      "step": 5665
    },
    {
      "epoch": 3.3431603773584904,
      "grad_norm": 1.6079306602478027,
      "learning_rate": 9.863358996678378e-06,
      "loss": 0.4306,
      "num_input_tokens_seen": 3690936,
      "step": 5670
    },
    {
      "epoch": 3.3461084905660377,
      "grad_norm": 1.7526575326919556,
      "learning_rate": 9.86276100896273e-06,
      "loss": 0.3543,
      "num_input_tokens_seen": 3694616,
      "step": 5675
    },
    {
      "epoch": 3.349056603773585,
      "grad_norm": 4.281879901885986,
      "learning_rate": 9.862161733810147e-06,
      "loss": 0.5386,
      "num_input_tokens_seen": 3697720,
      "step": 5680
    },
    {
      "epoch": 3.3520047169811322,
      "grad_norm": 1.758099913597107,
      "learning_rate": 9.86156117137929e-06,
      "loss": 0.4007,
      "num_input_tokens_seen": 3700120,
      "step": 5685
    },
    {
      "epoch": 3.354952830188679,
      "grad_norm": 2.017382860183716,
      "learning_rate": 9.860959321829159e-06,
      "loss": 0.4636,
      "num_input_tokens_seen": 3703768,
      "step": 5690
    },
    {
      "epoch": 3.3579009433962264,
      "grad_norm": 2.042393445968628,
      "learning_rate": 9.860356185319102e-06,
      "loss": 0.514,
      "num_input_tokens_seen": 3707448,
      "step": 5695
    },
    {
      "epoch": 3.3608490566037736,
      "grad_norm": 4.5664873123168945,
      "learning_rate": 9.859751762008796e-06,
      "loss": 0.4787,
      "num_input_tokens_seen": 3709976,
      "step": 5700
    },
    {
      "epoch": 3.363797169811321,
      "grad_norm": 1.3283536434173584,
      "learning_rate": 9.859146052058266e-06,
      "loss": 0.6716,
      "num_input_tokens_seen": 3713208,
      "step": 5705
    },
    {
      "epoch": 3.3667452830188678,
      "grad_norm": 1.2281264066696167,
      "learning_rate": 9.858539055627876e-06,
      "loss": 0.3992,
      "num_input_tokens_seen": 3716504,
      "step": 5710
    },
    {
      "epoch": 3.369693396226415,
      "grad_norm": 2.228283166885376,
      "learning_rate": 9.857930772878333e-06,
      "loss": 0.6256,
      "num_input_tokens_seen": 3718744,
      "step": 5715
    },
    {
      "epoch": 3.3726415094339623,
      "grad_norm": 2.2391412258148193,
      "learning_rate": 9.857321203970682e-06,
      "loss": 0.4481,
      "num_input_tokens_seen": 3721112,
      "step": 5720
    },
    {
      "epoch": 3.3755896226415096,
      "grad_norm": 2.3149242401123047,
      "learning_rate": 9.856710349066307e-06,
      "loss": 0.5411,
      "num_input_tokens_seen": 3725048,
      "step": 5725
    },
    {
      "epoch": 3.3785377358490565,
      "grad_norm": 1.328795313835144,
      "learning_rate": 9.856098208326937e-06,
      "loss": 0.4255,
      "num_input_tokens_seen": 3728504,
      "step": 5730
    },
    {
      "epoch": 3.3814858490566038,
      "grad_norm": 2.1709048748016357,
      "learning_rate": 9.855484781914639e-06,
      "loss": 0.5363,
      "num_input_tokens_seen": 3731448,
      "step": 5735
    },
    {
      "epoch": 3.384433962264151,
      "grad_norm": 2.2954556941986084,
      "learning_rate": 9.854870069991817e-06,
      "loss": 0.5811,
      "num_input_tokens_seen": 3734808,
      "step": 5740
    },
    {
      "epoch": 3.387382075471698,
      "grad_norm": 3.190495252609253,
      "learning_rate": 9.854254072721222e-06,
      "loss": 0.5007,
      "num_input_tokens_seen": 3737688,
      "step": 5745
    },
    {
      "epoch": 3.390330188679245,
      "grad_norm": 1.5982915163040161,
      "learning_rate": 9.853636790265938e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 3740888,
      "step": 5750
    },
    {
      "epoch": 3.3932783018867925,
      "grad_norm": 1.603284478187561,
      "learning_rate": 9.853018222789397e-06,
      "loss": 0.3697,
      "num_input_tokens_seen": 3744184,
      "step": 5755
    },
    {
      "epoch": 3.3962264150943398,
      "grad_norm": 1.6270633935928345,
      "learning_rate": 9.852398370455367e-06,
      "loss": 0.4767,
      "num_input_tokens_seen": 3747384,
      "step": 5760
    },
    {
      "epoch": 3.3991745283018866,
      "grad_norm": 4.308558464050293,
      "learning_rate": 9.851777233427955e-06,
      "loss": 0.5057,
      "num_input_tokens_seen": 3750296,
      "step": 5765
    },
    {
      "epoch": 3.402122641509434,
      "grad_norm": 2.298288345336914,
      "learning_rate": 9.85115481187161e-06,
      "loss": 0.3903,
      "num_input_tokens_seen": 3753624,
      "step": 5770
    },
    {
      "epoch": 3.405070754716981,
      "grad_norm": 1.2613229751586914,
      "learning_rate": 9.850531105951123e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 3757400,
      "step": 5775
    },
    {
      "epoch": 3.4080188679245285,
      "grad_norm": 2.876115083694458,
      "learning_rate": 9.84990611583162e-06,
      "loss": 0.4246,
      "num_input_tokens_seen": 3760344,
      "step": 5780
    },
    {
      "epoch": 3.4109669811320753,
      "grad_norm": 1.5410088300704956,
      "learning_rate": 9.849279841678572e-06,
      "loss": 0.4753,
      "num_input_tokens_seen": 3763512,
      "step": 5785
    },
    {
      "epoch": 3.4139150943396226,
      "grad_norm": 1.900124192237854,
      "learning_rate": 9.848652283657785e-06,
      "loss": 0.4299,
      "num_input_tokens_seen": 3766232,
      "step": 5790
    },
    {
      "epoch": 3.41686320754717,
      "grad_norm": 1.909995198249817,
      "learning_rate": 9.848023441935411e-06,
      "loss": 0.3983,
      "num_input_tokens_seen": 3770616,
      "step": 5795
    },
    {
      "epoch": 3.419811320754717,
      "grad_norm": 1.2678277492523193,
      "learning_rate": 9.847393316677935e-06,
      "loss": 0.3599,
      "num_input_tokens_seen": 3773464,
      "step": 5800
    },
    {
      "epoch": 3.422759433962264,
      "grad_norm": 1.2165656089782715,
      "learning_rate": 9.846761908052188e-06,
      "loss": 0.4248,
      "num_input_tokens_seen": 3776472,
      "step": 5805
    },
    {
      "epoch": 3.4257075471698113,
      "grad_norm": 1.8495151996612549,
      "learning_rate": 9.846129216225338e-06,
      "loss": 0.4617,
      "num_input_tokens_seen": 3780056,
      "step": 5810
    },
    {
      "epoch": 3.4286556603773586,
      "grad_norm": 3.7332139015197754,
      "learning_rate": 9.845495241364892e-06,
      "loss": 0.4772,
      "num_input_tokens_seen": 3783800,
      "step": 5815
    },
    {
      "epoch": 3.4316037735849054,
      "grad_norm": 2.291036367416382,
      "learning_rate": 9.844859983638696e-06,
      "loss": 0.4921,
      "num_input_tokens_seen": 3786744,
      "step": 5820
    },
    {
      "epoch": 3.4345518867924527,
      "grad_norm": 2.65614652633667,
      "learning_rate": 9.844223443214942e-06,
      "loss": 0.5713,
      "num_input_tokens_seen": 3789624,
      "step": 5825
    },
    {
      "epoch": 3.4375,
      "grad_norm": 1.4563069343566895,
      "learning_rate": 9.843585620262153e-06,
      "loss": 0.4781,
      "num_input_tokens_seen": 3793816,
      "step": 5830
    },
    {
      "epoch": 3.4404481132075473,
      "grad_norm": 10.31403636932373,
      "learning_rate": 9.842946514949197e-06,
      "loss": 0.5455,
      "num_input_tokens_seen": 3796248,
      "step": 5835
    },
    {
      "epoch": 3.4433962264150946,
      "grad_norm": 1.3758074045181274,
      "learning_rate": 9.842306127445279e-06,
      "loss": 0.5237,
      "num_input_tokens_seen": 3799000,
      "step": 5840
    },
    {
      "epoch": 3.4463443396226414,
      "grad_norm": 1.8348426818847656,
      "learning_rate": 9.841664457919944e-06,
      "loss": 0.4977,
      "num_input_tokens_seen": 3801976,
      "step": 5845
    },
    {
      "epoch": 3.4492924528301887,
      "grad_norm": 2.4773263931274414,
      "learning_rate": 9.841021506543079e-06,
      "loss": 0.544,
      "num_input_tokens_seen": 3805432,
      "step": 5850
    },
    {
      "epoch": 3.452240566037736,
      "grad_norm": 1.555792212486267,
      "learning_rate": 9.840377273484904e-06,
      "loss": 0.4171,
      "num_input_tokens_seen": 3809144,
      "step": 5855
    },
    {
      "epoch": 3.455188679245283,
      "grad_norm": 3.9127581119537354,
      "learning_rate": 9.839731758915986e-06,
      "loss": 0.5361,
      "num_input_tokens_seen": 3812472,
      "step": 5860
    },
    {
      "epoch": 3.45813679245283,
      "grad_norm": 2.701272964477539,
      "learning_rate": 9.839084963007226e-06,
      "loss": 0.4048,
      "num_input_tokens_seen": 3815512,
      "step": 5865
    },
    {
      "epoch": 3.4610849056603774,
      "grad_norm": 1.1864343881607056,
      "learning_rate": 9.838436885929868e-06,
      "loss": 0.7297,
      "num_input_tokens_seen": 3817816,
      "step": 5870
    },
    {
      "epoch": 3.4640330188679247,
      "grad_norm": 2.4107208251953125,
      "learning_rate": 9.837787527855492e-06,
      "loss": 0.363,
      "num_input_tokens_seen": 3820568,
      "step": 5875
    },
    {
      "epoch": 3.4669811320754715,
      "grad_norm": 2.6572368144989014,
      "learning_rate": 9.837136888956017e-06,
      "loss": 0.4708,
      "num_input_tokens_seen": 3824088,
      "step": 5880
    },
    {
      "epoch": 3.469929245283019,
      "grad_norm": 1.6846486330032349,
      "learning_rate": 9.836484969403705e-06,
      "loss": 0.4792,
      "num_input_tokens_seen": 3827192,
      "step": 5885
    },
    {
      "epoch": 3.472877358490566,
      "grad_norm": 3.556777238845825,
      "learning_rate": 9.835831769371152e-06,
      "loss": 0.4112,
      "num_input_tokens_seen": 3829400,
      "step": 5890
    },
    {
      "epoch": 3.4758254716981134,
      "grad_norm": 2.8453385829925537,
      "learning_rate": 9.835177289031298e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 3831480,
      "step": 5895
    },
    {
      "epoch": 3.4787735849056602,
      "grad_norm": 1.375619888305664,
      "learning_rate": 9.834521528557419e-06,
      "loss": 0.5183,
      "num_input_tokens_seen": 3835032,
      "step": 5900
    },
    {
      "epoch": 3.4817216981132075,
      "grad_norm": 1.86631178855896,
      "learning_rate": 9.833864488123128e-06,
      "loss": 0.3635,
      "num_input_tokens_seen": 3838008,
      "step": 5905
    },
    {
      "epoch": 3.484669811320755,
      "grad_norm": 1.8732240200042725,
      "learning_rate": 9.83320616790238e-06,
      "loss": 0.4173,
      "num_input_tokens_seen": 3841432,
      "step": 5910
    },
    {
      "epoch": 3.487617924528302,
      "grad_norm": 1.6845920085906982,
      "learning_rate": 9.832546568069472e-06,
      "loss": 0.4998,
      "num_input_tokens_seen": 3844344,
      "step": 5915
    },
    {
      "epoch": 3.490566037735849,
      "grad_norm": 2.3121297359466553,
      "learning_rate": 9.831885688799031e-06,
      "loss": 0.4914,
      "num_input_tokens_seen": 3847544,
      "step": 5920
    },
    {
      "epoch": 3.4935141509433962,
      "grad_norm": 1.8233729600906372,
      "learning_rate": 9.83122353026603e-06,
      "loss": 0.8029,
      "num_input_tokens_seen": 3850616,
      "step": 5925
    },
    {
      "epoch": 3.4964622641509435,
      "grad_norm": 1.4017064571380615,
      "learning_rate": 9.830560092645778e-06,
      "loss": 0.444,
      "num_input_tokens_seen": 3854104,
      "step": 5930
    },
    {
      "epoch": 3.4994103773584904,
      "grad_norm": 1.9475303888320923,
      "learning_rate": 9.829895376113923e-06,
      "loss": 0.4993,
      "num_input_tokens_seen": 3858296,
      "step": 5935
    },
    {
      "epoch": 3.5023584905660377,
      "grad_norm": 5.680035591125488,
      "learning_rate": 9.829229380846452e-06,
      "loss": 0.4349,
      "num_input_tokens_seen": 3860952,
      "step": 5940
    },
    {
      "epoch": 3.505306603773585,
      "grad_norm": 2.709632396697998,
      "learning_rate": 9.82856210701969e-06,
      "loss": 0.5228,
      "num_input_tokens_seen": 3863800,
      "step": 5945
    },
    {
      "epoch": 3.5082547169811322,
      "grad_norm": 2.132789134979248,
      "learning_rate": 9.827893554810298e-06,
      "loss": 0.5692,
      "num_input_tokens_seen": 3867800,
      "step": 5950
    },
    {
      "epoch": 3.5112028301886795,
      "grad_norm": 2.1033480167388916,
      "learning_rate": 9.827223724395281e-06,
      "loss": 0.5173,
      "num_input_tokens_seen": 3870360,
      "step": 5955
    },
    {
      "epoch": 3.5141509433962264,
      "grad_norm": 1.6832857131958008,
      "learning_rate": 9.82655261595198e-06,
      "loss": 0.4577,
      "num_input_tokens_seen": 3873432,
      "step": 5960
    },
    {
      "epoch": 3.5170990566037736,
      "grad_norm": 2.3264310359954834,
      "learning_rate": 9.825880229658073e-06,
      "loss": 0.5061,
      "num_input_tokens_seen": 3876376,
      "step": 5965
    },
    {
      "epoch": 3.5200471698113205,
      "grad_norm": 1.7049273252487183,
      "learning_rate": 9.825206565691576e-06,
      "loss": 0.5066,
      "num_input_tokens_seen": 3879352,
      "step": 5970
    },
    {
      "epoch": 3.5229952830188678,
      "grad_norm": 1.656238317489624,
      "learning_rate": 9.824531624230844e-06,
      "loss": 0.4673,
      "num_input_tokens_seen": 3882712,
      "step": 5975
    },
    {
      "epoch": 3.525943396226415,
      "grad_norm": 1.9921181201934814,
      "learning_rate": 9.823855405454573e-06,
      "loss": 0.5376,
      "num_input_tokens_seen": 3886072,
      "step": 5980
    },
    {
      "epoch": 3.5288915094339623,
      "grad_norm": 2.7016568183898926,
      "learning_rate": 9.823177909541795e-06,
      "loss": 0.4141,
      "num_input_tokens_seen": 3889304,
      "step": 5985
    },
    {
      "epoch": 3.5318396226415096,
      "grad_norm": 3.4773778915405273,
      "learning_rate": 9.822499136671877e-06,
      "loss": 0.3906,
      "num_input_tokens_seen": 3892600,
      "step": 5990
    },
    {
      "epoch": 3.5347877358490565,
      "grad_norm": 2.423518419265747,
      "learning_rate": 9.82181908702453e-06,
      "loss": 0.4666,
      "num_input_tokens_seen": 3895992,
      "step": 5995
    },
    {
      "epoch": 3.5377358490566038,
      "grad_norm": 2.0235812664031982,
      "learning_rate": 9.821137760779797e-06,
      "loss": 0.6857,
      "num_input_tokens_seen": 3899480,
      "step": 6000
    },
    {
      "epoch": 3.540683962264151,
      "grad_norm": 2.6161768436431885,
      "learning_rate": 9.820455158118065e-06,
      "loss": 0.4807,
      "num_input_tokens_seen": 3901880,
      "step": 6005
    },
    {
      "epoch": 3.543632075471698,
      "grad_norm": 2.06429123878479,
      "learning_rate": 9.819771279220053e-06,
      "loss": 0.4873,
      "num_input_tokens_seen": 3906808,
      "step": 6010
    },
    {
      "epoch": 3.546580188679245,
      "grad_norm": 1.429808497428894,
      "learning_rate": 9.819086124266825e-06,
      "loss": 0.4121,
      "num_input_tokens_seen": 3910296,
      "step": 6015
    },
    {
      "epoch": 3.5495283018867925,
      "grad_norm": 3.465043783187866,
      "learning_rate": 9.818399693439778e-06,
      "loss": 0.5161,
      "num_input_tokens_seen": 3913400,
      "step": 6020
    },
    {
      "epoch": 3.5524764150943398,
      "grad_norm": 1.3960872888565063,
      "learning_rate": 9.817711986920644e-06,
      "loss": 0.4391,
      "num_input_tokens_seen": 3915896,
      "step": 6025
    },
    {
      "epoch": 3.555424528301887,
      "grad_norm": 2.4132487773895264,
      "learning_rate": 9.817023004891497e-06,
      "loss": 0.4052,
      "num_input_tokens_seen": 3919352,
      "step": 6030
    },
    {
      "epoch": 3.558372641509434,
      "grad_norm": 7.3894171714782715,
      "learning_rate": 9.816332747534752e-06,
      "loss": 0.6553,
      "num_input_tokens_seen": 3921560,
      "step": 6035
    },
    {
      "epoch": 3.561320754716981,
      "grad_norm": 5.019872665405273,
      "learning_rate": 9.815641215033153e-06,
      "loss": 0.6066,
      "num_input_tokens_seen": 3924824,
      "step": 6040
    },
    {
      "epoch": 3.5642688679245285,
      "grad_norm": 0.8585574626922607,
      "learning_rate": 9.814948407569789e-06,
      "loss": 0.4676,
      "num_input_tokens_seen": 3928216,
      "step": 6045
    },
    {
      "epoch": 3.5672169811320753,
      "grad_norm": 2.468637228012085,
      "learning_rate": 9.814254325328082e-06,
      "loss": 0.8209,
      "num_input_tokens_seen": 3931192,
      "step": 6050
    },
    {
      "epoch": 3.5701650943396226,
      "grad_norm": 1.4459161758422852,
      "learning_rate": 9.813558968491794e-06,
      "loss": 0.3313,
      "num_input_tokens_seen": 3934072,
      "step": 6055
    },
    {
      "epoch": 3.57311320754717,
      "grad_norm": 2.385265350341797,
      "learning_rate": 9.812862337245024e-06,
      "loss": 0.4213,
      "num_input_tokens_seen": 3937048,
      "step": 6060
    },
    {
      "epoch": 3.576061320754717,
      "grad_norm": 1.4659218788146973,
      "learning_rate": 9.812164431772208e-06,
      "loss": 0.4467,
      "num_input_tokens_seen": 3940472,
      "step": 6065
    },
    {
      "epoch": 3.579009433962264,
      "grad_norm": 2.363335132598877,
      "learning_rate": 9.81146525225812e-06,
      "loss": 0.4232,
      "num_input_tokens_seen": 3943352,
      "step": 6070
    },
    {
      "epoch": 3.5819575471698113,
      "grad_norm": 2.0104377269744873,
      "learning_rate": 9.810764798887868e-06,
      "loss": 0.4158,
      "num_input_tokens_seen": 3948120,
      "step": 6075
    },
    {
      "epoch": 3.5849056603773586,
      "grad_norm": 2.774599313735962,
      "learning_rate": 9.810063071846905e-06,
      "loss": 0.4391,
      "num_input_tokens_seen": 3951640,
      "step": 6080
    },
    {
      "epoch": 3.5878537735849054,
      "grad_norm": 3.1206107139587402,
      "learning_rate": 9.809360071321013e-06,
      "loss": 0.5251,
      "num_input_tokens_seen": 3954232,
      "step": 6085
    },
    {
      "epoch": 3.5908018867924527,
      "grad_norm": 1.4326025247573853,
      "learning_rate": 9.808655797496314e-06,
      "loss": 0.3834,
      "num_input_tokens_seen": 3957592,
      "step": 6090
    },
    {
      "epoch": 3.59375,
      "grad_norm": 2.290938138961792,
      "learning_rate": 9.807950250559268e-06,
      "loss": 0.4787,
      "num_input_tokens_seen": 3960408,
      "step": 6095
    },
    {
      "epoch": 3.5966981132075473,
      "grad_norm": 2.7492825984954834,
      "learning_rate": 9.807243430696673e-06,
      "loss": 0.4039,
      "num_input_tokens_seen": 3963768,
      "step": 6100
    },
    {
      "epoch": 3.5996462264150946,
      "grad_norm": 2.6364223957061768,
      "learning_rate": 9.806535338095661e-06,
      "loss": 0.3762,
      "num_input_tokens_seen": 3967128,
      "step": 6105
    },
    {
      "epoch": 3.6025943396226414,
      "grad_norm": 6.124648571014404,
      "learning_rate": 9.805825972943706e-06,
      "loss": 0.4224,
      "num_input_tokens_seen": 3969304,
      "step": 6110
    },
    {
      "epoch": 3.6055424528301887,
      "grad_norm": 1.3641059398651123,
      "learning_rate": 9.80511533542861e-06,
      "loss": 0.3702,
      "num_input_tokens_seen": 3972024,
      "step": 6115
    },
    {
      "epoch": 3.608490566037736,
      "grad_norm": 2.9206619262695312,
      "learning_rate": 9.80440342573852e-06,
      "loss": 0.4523,
      "num_input_tokens_seen": 3975032,
      "step": 6120
    },
    {
      "epoch": 3.611438679245283,
      "grad_norm": 1.459501028060913,
      "learning_rate": 9.803690244061919e-06,
      "loss": 0.4127,
      "num_input_tokens_seen": 3977944,
      "step": 6125
    },
    {
      "epoch": 3.61438679245283,
      "grad_norm": 2.5439250469207764,
      "learning_rate": 9.802975790587621e-06,
      "loss": 0.4905,
      "num_input_tokens_seen": 3981304,
      "step": 6130
    },
    {
      "epoch": 3.6173349056603774,
      "grad_norm": 1.7963510751724243,
      "learning_rate": 9.802260065504783e-06,
      "loss": 0.5802,
      "num_input_tokens_seen": 3984792,
      "step": 6135
    },
    {
      "epoch": 3.6202830188679247,
      "grad_norm": 1.4936938285827637,
      "learning_rate": 9.801543069002897e-06,
      "loss": 0.3999,
      "num_input_tokens_seen": 3987832,
      "step": 6140
    },
    {
      "epoch": 3.6232311320754715,
      "grad_norm": 16.707948684692383,
      "learning_rate": 9.80082480127179e-06,
      "loss": 0.3942,
      "num_input_tokens_seen": 3991800,
      "step": 6145
    },
    {
      "epoch": 3.626179245283019,
      "grad_norm": 1.1759858131408691,
      "learning_rate": 9.800105262501628e-06,
      "loss": 0.4654,
      "num_input_tokens_seen": 3995096,
      "step": 6150
    },
    {
      "epoch": 3.629127358490566,
      "grad_norm": 1.2978332042694092,
      "learning_rate": 9.799384452882907e-06,
      "loss": 0.479,
      "num_input_tokens_seen": 3997880,
      "step": 6155
    },
    {
      "epoch": 3.632075471698113,
      "grad_norm": 2.4032607078552246,
      "learning_rate": 9.798662372606469e-06,
      "loss": 0.4028,
      "num_input_tokens_seen": 4001400,
      "step": 6160
    },
    {
      "epoch": 3.6350235849056602,
      "grad_norm": 2.098317861557007,
      "learning_rate": 9.797939021863487e-06,
      "loss": 0.4583,
      "num_input_tokens_seen": 4004344,
      "step": 6165
    },
    {
      "epoch": 3.6379716981132075,
      "grad_norm": 6.382221221923828,
      "learning_rate": 9.797214400845472e-06,
      "loss": 0.6341,
      "num_input_tokens_seen": 4007416,
      "step": 6170
    },
    {
      "epoch": 3.640919811320755,
      "grad_norm": 3.4683034420013428,
      "learning_rate": 9.796488509744269e-06,
      "loss": 0.5083,
      "num_input_tokens_seen": 4010488,
      "step": 6175
    },
    {
      "epoch": 3.643867924528302,
      "grad_norm": 2.5313379764556885,
      "learning_rate": 9.79576134875206e-06,
      "loss": 0.5237,
      "num_input_tokens_seen": 4013144,
      "step": 6180
    },
    {
      "epoch": 3.646816037735849,
      "grad_norm": 2.1807286739349365,
      "learning_rate": 9.795032918061367e-06,
      "loss": 0.512,
      "num_input_tokens_seen": 4016312,
      "step": 6185
    },
    {
      "epoch": 3.6497641509433962,
      "grad_norm": 1.6005451679229736,
      "learning_rate": 9.794303217865041e-06,
      "loss": 0.5066,
      "num_input_tokens_seen": 4019896,
      "step": 6190
    },
    {
      "epoch": 3.6527122641509435,
      "grad_norm": 1.542596697807312,
      "learning_rate": 9.79357224835628e-06,
      "loss": 0.5833,
      "num_input_tokens_seen": 4022872,
      "step": 6195
    },
    {
      "epoch": 3.6556603773584904,
      "grad_norm": 1.9818195104599,
      "learning_rate": 9.792840009728605e-06,
      "loss": 0.4617,
      "num_input_tokens_seen": 4025912,
      "step": 6200
    },
    {
      "epoch": 3.6586084905660377,
      "grad_norm": 1.441719651222229,
      "learning_rate": 9.79210650217588e-06,
      "loss": 0.4697,
      "num_input_tokens_seen": 4029208,
      "step": 6205
    },
    {
      "epoch": 3.661556603773585,
      "grad_norm": 1.5752439498901367,
      "learning_rate": 9.791371725892307e-06,
      "loss": 0.5245,
      "num_input_tokens_seen": 4032120,
      "step": 6210
    },
    {
      "epoch": 3.6645047169811322,
      "grad_norm": 1.0782500505447388,
      "learning_rate": 9.79063568107242e-06,
      "loss": 0.4678,
      "num_input_tokens_seen": 4035576,
      "step": 6215
    },
    {
      "epoch": 3.6674528301886795,
      "grad_norm": 1.9574594497680664,
      "learning_rate": 9.78989836791109e-06,
      "loss": 0.4948,
      "num_input_tokens_seen": 4039096,
      "step": 6220
    },
    {
      "epoch": 3.6704009433962264,
      "grad_norm": 1.5621784925460815,
      "learning_rate": 9.789159786603524e-06,
      "loss": 0.4846,
      "num_input_tokens_seen": 4045720,
      "step": 6225
    },
    {
      "epoch": 3.6733490566037736,
      "grad_norm": 1.8161766529083252,
      "learning_rate": 9.788419937345263e-06,
      "loss": 0.5236,
      "num_input_tokens_seen": 4048760,
      "step": 6230
    },
    {
      "epoch": 3.6762971698113205,
      "grad_norm": 1.7863855361938477,
      "learning_rate": 9.787678820332188e-06,
      "loss": 0.5792,
      "num_input_tokens_seen": 4052408,
      "step": 6235
    },
    {
      "epoch": 3.6792452830188678,
      "grad_norm": 1.747335433959961,
      "learning_rate": 9.78693643576051e-06,
      "loss": 0.3448,
      "num_input_tokens_seen": 4054872,
      "step": 6240
    },
    {
      "epoch": 3.682193396226415,
      "grad_norm": 3.1540074348449707,
      "learning_rate": 9.786192783826782e-06,
      "loss": 0.4117,
      "num_input_tokens_seen": 4057304,
      "step": 6245
    },
    {
      "epoch": 3.6851415094339623,
      "grad_norm": 1.4725329875946045,
      "learning_rate": 9.785447864727887e-06,
      "loss": 0.4217,
      "num_input_tokens_seen": 4060056,
      "step": 6250
    },
    {
      "epoch": 3.6880896226415096,
      "grad_norm": 1.585604190826416,
      "learning_rate": 9.784701678661045e-06,
      "loss": 0.45,
      "num_input_tokens_seen": 4062808,
      "step": 6255
    },
    {
      "epoch": 3.6910377358490565,
      "grad_norm": 1.9884909391403198,
      "learning_rate": 9.783954225823813e-06,
      "loss": 0.3605,
      "num_input_tokens_seen": 4066360,
      "step": 6260
    },
    {
      "epoch": 3.6939858490566038,
      "grad_norm": 1.9366538524627686,
      "learning_rate": 9.783205506414082e-06,
      "loss": 0.6737,
      "num_input_tokens_seen": 4071352,
      "step": 6265
    },
    {
      "epoch": 3.696933962264151,
      "grad_norm": 1.231418251991272,
      "learning_rate": 9.782455520630079e-06,
      "loss": 0.4795,
      "num_input_tokens_seen": 4076952,
      "step": 6270
    },
    {
      "epoch": 3.699882075471698,
      "grad_norm": 1.4096026420593262,
      "learning_rate": 9.781704268670364e-06,
      "loss": 0.5517,
      "num_input_tokens_seen": 4080952,
      "step": 6275
    },
    {
      "epoch": 3.702830188679245,
      "grad_norm": 1.0999194383621216,
      "learning_rate": 9.780951750733837e-06,
      "loss": 0.4417,
      "num_input_tokens_seen": 4084888,
      "step": 6280
    },
    {
      "epoch": 3.7057783018867925,
      "grad_norm": 1.1066210269927979,
      "learning_rate": 9.780197967019728e-06,
      "loss": 0.414,
      "num_input_tokens_seen": 4087640,
      "step": 6285
    },
    {
      "epoch": 3.7087264150943398,
      "grad_norm": 3.3654232025146484,
      "learning_rate": 9.779442917727608e-06,
      "loss": 0.5131,
      "num_input_tokens_seen": 4090584,
      "step": 6290
    },
    {
      "epoch": 3.711674528301887,
      "grad_norm": 2.059638738632202,
      "learning_rate": 9.778686603057377e-06,
      "loss": 0.428,
      "num_input_tokens_seen": 4096248,
      "step": 6295
    },
    {
      "epoch": 3.714622641509434,
      "grad_norm": 2.241614580154419,
      "learning_rate": 9.777929023209271e-06,
      "loss": 0.5105,
      "num_input_tokens_seen": 4099000,
      "step": 6300
    },
    {
      "epoch": 3.717570754716981,
      "grad_norm": 3.676776170730591,
      "learning_rate": 9.777170178383866e-06,
      "loss": 0.4514,
      "num_input_tokens_seen": 4102968,
      "step": 6305
    },
    {
      "epoch": 3.7205188679245285,
      "grad_norm": 2.588263511657715,
      "learning_rate": 9.776410068782068e-06,
      "loss": 0.4082,
      "num_input_tokens_seen": 4106136,
      "step": 6310
    },
    {
      "epoch": 3.7234669811320753,
      "grad_norm": 1.212329626083374,
      "learning_rate": 9.775648694605118e-06,
      "loss": 0.4888,
      "num_input_tokens_seen": 4109912,
      "step": 6315
    },
    {
      "epoch": 3.7264150943396226,
      "grad_norm": 2.049220323562622,
      "learning_rate": 9.774886056054593e-06,
      "loss": 0.5083,
      "num_input_tokens_seen": 4112920,
      "step": 6320
    },
    {
      "epoch": 3.72936320754717,
      "grad_norm": 1.1754775047302246,
      "learning_rate": 9.774122153332408e-06,
      "loss": 0.4457,
      "num_input_tokens_seen": 4116152,
      "step": 6325
    },
    {
      "epoch": 3.732311320754717,
      "grad_norm": 3.4246468544006348,
      "learning_rate": 9.773356986640807e-06,
      "loss": 0.4649,
      "num_input_tokens_seen": 4120280,
      "step": 6330
    },
    {
      "epoch": 3.735259433962264,
      "grad_norm": 2.438265085220337,
      "learning_rate": 9.772590556182373e-06,
      "loss": 0.4834,
      "num_input_tokens_seen": 4122840,
      "step": 6335
    },
    {
      "epoch": 3.7382075471698113,
      "grad_norm": 1.4794487953186035,
      "learning_rate": 9.77182286216002e-06,
      "loss": 0.3839,
      "num_input_tokens_seen": 4126232,
      "step": 6340
    },
    {
      "epoch": 3.7411556603773586,
      "grad_norm": 2.042238712310791,
      "learning_rate": 9.771053904776998e-06,
      "loss": 0.4966,
      "num_input_tokens_seen": 4129752,
      "step": 6345
    },
    {
      "epoch": 3.7441037735849054,
      "grad_norm": 2.4753189086914062,
      "learning_rate": 9.770283684236891e-06,
      "loss": 0.7019,
      "num_input_tokens_seen": 4132280,
      "step": 6350
    },
    {
      "epoch": 3.7470518867924527,
      "grad_norm": 1.3164738416671753,
      "learning_rate": 9.769512200743623e-06,
      "loss": 0.4648,
      "num_input_tokens_seen": 4135864,
      "step": 6355
    },
    {
      "epoch": 3.75,
      "grad_norm": 2.447172164916992,
      "learning_rate": 9.768739454501444e-06,
      "loss": 0.5098,
      "num_input_tokens_seen": 4138680,
      "step": 6360
    },
    {
      "epoch": 3.7529481132075473,
      "grad_norm": 5.2237725257873535,
      "learning_rate": 9.76796544571494e-06,
      "loss": 0.4696,
      "num_input_tokens_seen": 4141368,
      "step": 6365
    },
    {
      "epoch": 3.7558962264150946,
      "grad_norm": 2.027153253555298,
      "learning_rate": 9.767190174589036e-06,
      "loss": 0.448,
      "num_input_tokens_seen": 4144760,
      "step": 6370
    },
    {
      "epoch": 3.7588443396226414,
      "grad_norm": 3.7473416328430176,
      "learning_rate": 9.76641364132899e-06,
      "loss": 0.5654,
      "num_input_tokens_seen": 4147416,
      "step": 6375
    },
    {
      "epoch": 3.7617924528301887,
      "grad_norm": 1.8635953664779663,
      "learning_rate": 9.765635846140389e-06,
      "loss": 0.483,
      "num_input_tokens_seen": 4150328,
      "step": 6380
    },
    {
      "epoch": 3.764740566037736,
      "grad_norm": 1.0663868188858032,
      "learning_rate": 9.764856789229157e-06,
      "loss": 0.4791,
      "num_input_tokens_seen": 4153528,
      "step": 6385
    },
    {
      "epoch": 3.767688679245283,
      "grad_norm": 3.030426025390625,
      "learning_rate": 9.764076470801557e-06,
      "loss": 0.3479,
      "num_input_tokens_seen": 4159896,
      "step": 6390
    },
    {
      "epoch": 3.77063679245283,
      "grad_norm": 1.5392131805419922,
      "learning_rate": 9.763294891064182e-06,
      "loss": 0.4897,
      "num_input_tokens_seen": 4163160,
      "step": 6395
    },
    {
      "epoch": 3.7735849056603774,
      "grad_norm": 1.3763707876205444,
      "learning_rate": 9.762512050223951e-06,
      "loss": 0.5124,
      "num_input_tokens_seen": 4166168,
      "step": 6400
    },
    {
      "epoch": 3.7765330188679247,
      "grad_norm": 2.2697227001190186,
      "learning_rate": 9.761727948488132e-06,
      "loss": 0.4488,
      "num_input_tokens_seen": 4169016,
      "step": 6405
    },
    {
      "epoch": 3.7794811320754715,
      "grad_norm": 1.6069093942642212,
      "learning_rate": 9.760942586064315e-06,
      "loss": 0.6208,
      "num_input_tokens_seen": 4172248,
      "step": 6410
    },
    {
      "epoch": 3.782429245283019,
      "grad_norm": 2.738206624984741,
      "learning_rate": 9.760155963160431e-06,
      "loss": 0.4391,
      "num_input_tokens_seen": 4175000,
      "step": 6415
    },
    {
      "epoch": 3.785377358490566,
      "grad_norm": 2.1404919624328613,
      "learning_rate": 9.759368079984741e-06,
      "loss": 0.5175,
      "num_input_tokens_seen": 4178168,
      "step": 6420
    },
    {
      "epoch": 3.788325471698113,
      "grad_norm": 1.4808086156845093,
      "learning_rate": 9.758578936745839e-06,
      "loss": 0.5281,
      "num_input_tokens_seen": 4181688,
      "step": 6425
    },
    {
      "epoch": 3.7912735849056602,
      "grad_norm": 4.177740097045898,
      "learning_rate": 9.757788533652656e-06,
      "loss": 0.4679,
      "num_input_tokens_seen": 4184728,
      "step": 6430
    },
    {
      "epoch": 3.7942216981132075,
      "grad_norm": 1.8638392686843872,
      "learning_rate": 9.756996870914454e-06,
      "loss": 0.3743,
      "num_input_tokens_seen": 4188184,
      "step": 6435
    },
    {
      "epoch": 3.797169811320755,
      "grad_norm": 2.0354652404785156,
      "learning_rate": 9.756203948740828e-06,
      "loss": 0.4798,
      "num_input_tokens_seen": 4190808,
      "step": 6440
    },
    {
      "epoch": 3.800117924528302,
      "grad_norm": 1.6854146718978882,
      "learning_rate": 9.755409767341709e-06,
      "loss": 0.4782,
      "num_input_tokens_seen": 4194808,
      "step": 6445
    },
    {
      "epoch": 3.803066037735849,
      "grad_norm": 2.303384304046631,
      "learning_rate": 9.75461432692736e-06,
      "loss": 0.5435,
      "num_input_tokens_seen": 4197400,
      "step": 6450
    },
    {
      "epoch": 3.8060141509433962,
      "grad_norm": 1.7422261238098145,
      "learning_rate": 9.753817627708375e-06,
      "loss": 0.5429,
      "num_input_tokens_seen": 4199736,
      "step": 6455
    },
    {
      "epoch": 3.8089622641509435,
      "grad_norm": 1.3491671085357666,
      "learning_rate": 9.753019669895686e-06,
      "loss": 0.598,
      "num_input_tokens_seen": 4202680,
      "step": 6460
    },
    {
      "epoch": 3.8119103773584904,
      "grad_norm": 1.4366841316223145,
      "learning_rate": 9.752220453700556e-06,
      "loss": 0.5387,
      "num_input_tokens_seen": 4205464,
      "step": 6465
    },
    {
      "epoch": 3.8148584905660377,
      "grad_norm": 1.5678027868270874,
      "learning_rate": 9.75141997933458e-06,
      "loss": 0.5288,
      "num_input_tokens_seen": 4208856,
      "step": 6470
    },
    {
      "epoch": 3.817806603773585,
      "grad_norm": 1.3407434225082397,
      "learning_rate": 9.750618247009685e-06,
      "loss": 0.5233,
      "num_input_tokens_seen": 4212728,
      "step": 6475
    },
    {
      "epoch": 3.8207547169811322,
      "grad_norm": 1.1687155961990356,
      "learning_rate": 9.749815256938138e-06,
      "loss": 0.4468,
      "num_input_tokens_seen": 4215960,
      "step": 6480
    },
    {
      "epoch": 3.8237028301886795,
      "grad_norm": 1.9046894311904907,
      "learning_rate": 9.749011009332529e-06,
      "loss": 0.3658,
      "num_input_tokens_seen": 4219096,
      "step": 6485
    },
    {
      "epoch": 3.8266509433962264,
      "grad_norm": 2.041167736053467,
      "learning_rate": 9.748205504405787e-06,
      "loss": 0.4501,
      "num_input_tokens_seen": 4223032,
      "step": 6490
    },
    {
      "epoch": 3.8295990566037736,
      "grad_norm": 1.5436069965362549,
      "learning_rate": 9.747398742371177e-06,
      "loss": 0.4556,
      "num_input_tokens_seen": 4225656,
      "step": 6495
    },
    {
      "epoch": 3.8325471698113205,
      "grad_norm": 2.265415668487549,
      "learning_rate": 9.746590723442289e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 4229720,
      "step": 6500
    },
    {
      "epoch": 3.8354952830188678,
      "grad_norm": 2.1142921447753906,
      "learning_rate": 9.745781447833049e-06,
      "loss": 0.3584,
      "num_input_tokens_seen": 4233464,
      "step": 6505
    },
    {
      "epoch": 3.838443396226415,
      "grad_norm": 6.028593063354492,
      "learning_rate": 9.74497091575772e-06,
      "loss": 0.4143,
      "num_input_tokens_seen": 4236120,
      "step": 6510
    },
    {
      "epoch": 3.8413915094339623,
      "grad_norm": 2.431793451309204,
      "learning_rate": 9.744159127430888e-06,
      "loss": 0.6568,
      "num_input_tokens_seen": 4239448,
      "step": 6515
    },
    {
      "epoch": 3.8443396226415096,
      "grad_norm": 3.3506648540496826,
      "learning_rate": 9.743346083067482e-06,
      "loss": 0.3596,
      "num_input_tokens_seen": 4242616,
      "step": 6520
    },
    {
      "epoch": 3.8472877358490565,
      "grad_norm": 4.8003435134887695,
      "learning_rate": 9.742531782882758e-06,
      "loss": 0.5789,
      "num_input_tokens_seen": 4245528,
      "step": 6525
    },
    {
      "epoch": 3.8502358490566038,
      "grad_norm": 1.9072314500808716,
      "learning_rate": 9.741716227092305e-06,
      "loss": 0.4895,
      "num_input_tokens_seen": 4248120,
      "step": 6530
    },
    {
      "epoch": 3.853183962264151,
      "grad_norm": 1.9113293886184692,
      "learning_rate": 9.740899415912048e-06,
      "loss": 0.5454,
      "num_input_tokens_seen": 4250808,
      "step": 6535
    },
    {
      "epoch": 3.856132075471698,
      "grad_norm": 3.0875918865203857,
      "learning_rate": 9.740081349558236e-06,
      "loss": 0.4629,
      "num_input_tokens_seen": 4254328,
      "step": 6540
    },
    {
      "epoch": 3.859080188679245,
      "grad_norm": 1.70663321018219,
      "learning_rate": 9.739262028247459e-06,
      "loss": 0.6465,
      "num_input_tokens_seen": 4260024,
      "step": 6545
    },
    {
      "epoch": 3.8620283018867925,
      "grad_norm": 2.228905439376831,
      "learning_rate": 9.738441452196633e-06,
      "loss": 0.3888,
      "num_input_tokens_seen": 4262680,
      "step": 6550
    },
    {
      "epoch": 3.8649764150943398,
      "grad_norm": 3.6392974853515625,
      "learning_rate": 9.737619621623013e-06,
      "loss": 0.4758,
      "num_input_tokens_seen": 4265784,
      "step": 6555
    },
    {
      "epoch": 3.867924528301887,
      "grad_norm": 1.7249586582183838,
      "learning_rate": 9.73679653674418e-06,
      "loss": 0.5636,
      "num_input_tokens_seen": 4268792,
      "step": 6560
    },
    {
      "epoch": 3.870872641509434,
      "grad_norm": 5.577042102813721,
      "learning_rate": 9.735972197778047e-06,
      "loss": 0.5459,
      "num_input_tokens_seen": 4271576,
      "step": 6565
    },
    {
      "epoch": 3.873820754716981,
      "grad_norm": 2.455580711364746,
      "learning_rate": 9.735146604942867e-06,
      "loss": 0.3232,
      "num_input_tokens_seen": 4274360,
      "step": 6570
    },
    {
      "epoch": 3.8767688679245285,
      "grad_norm": 3.610812187194824,
      "learning_rate": 9.734319758457214e-06,
      "loss": 0.4813,
      "num_input_tokens_seen": 4277144,
      "step": 6575
    },
    {
      "epoch": 3.8797169811320753,
      "grad_norm": 1.7456868886947632,
      "learning_rate": 9.733491658540001e-06,
      "loss": 0.4674,
      "num_input_tokens_seen": 4280184,
      "step": 6580
    },
    {
      "epoch": 3.8826650943396226,
      "grad_norm": 1.9427430629730225,
      "learning_rate": 9.732662305410474e-06,
      "loss": 0.4806,
      "num_input_tokens_seen": 4282616,
      "step": 6585
    },
    {
      "epoch": 3.88561320754717,
      "grad_norm": 1.8242812156677246,
      "learning_rate": 9.731831699288203e-06,
      "loss": 0.7064,
      "num_input_tokens_seen": 4286072,
      "step": 6590
    },
    {
      "epoch": 3.888561320754717,
      "grad_norm": 1.366325855255127,
      "learning_rate": 9.730999840393096e-06,
      "loss": 0.4256,
      "num_input_tokens_seen": 4289496,
      "step": 6595
    },
    {
      "epoch": 3.891509433962264,
      "grad_norm": 2.40189266204834,
      "learning_rate": 9.730166728945391e-06,
      "loss": 0.4116,
      "num_input_tokens_seen": 4292280,
      "step": 6600
    },
    {
      "epoch": 3.8944575471698113,
      "grad_norm": 3.8786261081695557,
      "learning_rate": 9.72933236516566e-06,
      "loss": 0.5178,
      "num_input_tokens_seen": 4295096,
      "step": 6605
    },
    {
      "epoch": 3.8974056603773586,
      "grad_norm": 2.2663419246673584,
      "learning_rate": 9.728496749274806e-06,
      "loss": 0.434,
      "num_input_tokens_seen": 4299288,
      "step": 6610
    },
    {
      "epoch": 3.9003537735849054,
      "grad_norm": 1.5046064853668213,
      "learning_rate": 9.727659881494054e-06,
      "loss": 0.5354,
      "num_input_tokens_seen": 4304376,
      "step": 6615
    },
    {
      "epoch": 3.9033018867924527,
      "grad_norm": 2.110302686691284,
      "learning_rate": 9.726821762044975e-06,
      "loss": 0.5407,
      "num_input_tokens_seen": 4309432,
      "step": 6620
    },
    {
      "epoch": 3.90625,
      "grad_norm": 1.2773325443267822,
      "learning_rate": 9.725982391149465e-06,
      "loss": 0.4264,
      "num_input_tokens_seen": 4313432,
      "step": 6625
    },
    {
      "epoch": 3.9091981132075473,
      "grad_norm": 1.06777822971344,
      "learning_rate": 9.725141769029747e-06,
      "loss": 0.4589,
      "num_input_tokens_seen": 4317464,
      "step": 6630
    },
    {
      "epoch": 3.9121462264150946,
      "grad_norm": 1.6418951749801636,
      "learning_rate": 9.72429989590838e-06,
      "loss": 0.4014,
      "num_input_tokens_seen": 4320280,
      "step": 6635
    },
    {
      "epoch": 3.9150943396226414,
      "grad_norm": 1.787442684173584,
      "learning_rate": 9.723456772008257e-06,
      "loss": 0.5671,
      "num_input_tokens_seen": 4323576,
      "step": 6640
    },
    {
      "epoch": 3.9180424528301887,
      "grad_norm": 2.2333014011383057,
      "learning_rate": 9.722612397552598e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 4326616,
      "step": 6645
    },
    {
      "epoch": 3.920990566037736,
      "grad_norm": 1.1333509683609009,
      "learning_rate": 9.72176677276495e-06,
      "loss": 0.5446,
      "num_input_tokens_seen": 4330168,
      "step": 6650
    },
    {
      "epoch": 3.923938679245283,
      "grad_norm": 1.2473218441009521,
      "learning_rate": 9.7209198978692e-06,
      "loss": 0.4958,
      "num_input_tokens_seen": 4333464,
      "step": 6655
    },
    {
      "epoch": 3.92688679245283,
      "grad_norm": 2.5179078578948975,
      "learning_rate": 9.720071773089564e-06,
      "loss": 0.6165,
      "num_input_tokens_seen": 4336504,
      "step": 6660
    },
    {
      "epoch": 3.9298349056603774,
      "grad_norm": 1.5156223773956299,
      "learning_rate": 9.71922239865058e-06,
      "loss": 0.4006,
      "num_input_tokens_seen": 4340088,
      "step": 6665
    },
    {
      "epoch": 3.9327830188679247,
      "grad_norm": 1.633601188659668,
      "learning_rate": 9.718371774777131e-06,
      "loss": 0.3981,
      "num_input_tokens_seen": 4344120,
      "step": 6670
    },
    {
      "epoch": 3.9357311320754715,
      "grad_norm": 1.2970178127288818,
      "learning_rate": 9.717519901694416e-06,
      "loss": 0.6055,
      "num_input_tokens_seen": 4347928,
      "step": 6675
    },
    {
      "epoch": 3.938679245283019,
      "grad_norm": 2.250190019607544,
      "learning_rate": 9.716666779627978e-06,
      "loss": 0.5198,
      "num_input_tokens_seen": 4351192,
      "step": 6680
    },
    {
      "epoch": 3.941627358490566,
      "grad_norm": 4.729641914367676,
      "learning_rate": 9.715812408803681e-06,
      "loss": 0.4781,
      "num_input_tokens_seen": 4353592,
      "step": 6685
    },
    {
      "epoch": 3.944575471698113,
      "grad_norm": 1.5684490203857422,
      "learning_rate": 9.714956789447726e-06,
      "loss": 0.4144,
      "num_input_tokens_seen": 4356632,
      "step": 6690
    },
    {
      "epoch": 3.9475235849056602,
      "grad_norm": 1.032141089439392,
      "learning_rate": 9.71409992178664e-06,
      "loss": 0.3851,
      "num_input_tokens_seen": 4359864,
      "step": 6695
    },
    {
      "epoch": 3.9504716981132075,
      "grad_norm": 1.2516428232192993,
      "learning_rate": 9.713241806047282e-06,
      "loss": 0.4902,
      "num_input_tokens_seen": 4363480,
      "step": 6700
    },
    {
      "epoch": 3.953419811320755,
      "grad_norm": 1.492336392402649,
      "learning_rate": 9.712382442456845e-06,
      "loss": 0.5751,
      "num_input_tokens_seen": 4367544,
      "step": 6705
    },
    {
      "epoch": 3.956367924528302,
      "grad_norm": 1.9288151264190674,
      "learning_rate": 9.711521831242846e-06,
      "loss": 0.4691,
      "num_input_tokens_seen": 4370712,
      "step": 6710
    },
    {
      "epoch": 3.959316037735849,
      "grad_norm": 2.6786880493164062,
      "learning_rate": 9.710659972633137e-06,
      "loss": 0.3743,
      "num_input_tokens_seen": 4374136,
      "step": 6715
    },
    {
      "epoch": 3.9622641509433962,
      "grad_norm": 1.5475469827651978,
      "learning_rate": 9.709796866855899e-06,
      "loss": 0.3075,
      "num_input_tokens_seen": 4376984,
      "step": 6720
    },
    {
      "epoch": 3.9652122641509435,
      "grad_norm": 1.9753527641296387,
      "learning_rate": 9.70893251413964e-06,
      "loss": 0.5512,
      "num_input_tokens_seen": 4380664,
      "step": 6725
    },
    {
      "epoch": 3.9681603773584904,
      "grad_norm": 1.3781172037124634,
      "learning_rate": 9.708066914713205e-06,
      "loss": 0.5348,
      "num_input_tokens_seen": 4383800,
      "step": 6730
    },
    {
      "epoch": 3.9711084905660377,
      "grad_norm": 3.0246148109436035,
      "learning_rate": 9.707200068805764e-06,
      "loss": 0.6916,
      "num_input_tokens_seen": 4387576,
      "step": 6735
    },
    {
      "epoch": 3.974056603773585,
      "grad_norm": 1.4945316314697266,
      "learning_rate": 9.706331976646817e-06,
      "loss": 0.3833,
      "num_input_tokens_seen": 4391224,
      "step": 6740
    },
    {
      "epoch": 3.9770047169811322,
      "grad_norm": 2.9747040271759033,
      "learning_rate": 9.705462638466197e-06,
      "loss": 0.4771,
      "num_input_tokens_seen": 4394072,
      "step": 6745
    },
    {
      "epoch": 3.9799528301886795,
      "grad_norm": 1.7769852876663208,
      "learning_rate": 9.704592054494065e-06,
      "loss": 0.4634,
      "num_input_tokens_seen": 4397784,
      "step": 6750
    },
    {
      "epoch": 3.9829009433962264,
      "grad_norm": 1.8301317691802979,
      "learning_rate": 9.703720224960909e-06,
      "loss": 0.56,
      "num_input_tokens_seen": 4400984,
      "step": 6755
    },
    {
      "epoch": 3.9858490566037736,
      "grad_norm": 1.7535650730133057,
      "learning_rate": 9.702847150097552e-06,
      "loss": 0.4941,
      "num_input_tokens_seen": 4404760,
      "step": 6760
    },
    {
      "epoch": 3.9887971698113205,
      "grad_norm": 2.4453303813934326,
      "learning_rate": 9.701972830135143e-06,
      "loss": 0.5114,
      "num_input_tokens_seen": 4407928,
      "step": 6765
    },
    {
      "epoch": 3.9917452830188678,
      "grad_norm": 1.5601730346679688,
      "learning_rate": 9.701097265305164e-06,
      "loss": 0.4297,
      "num_input_tokens_seen": 4410552,
      "step": 6770
    },
    {
      "epoch": 3.994693396226415,
      "grad_norm": 1.6432669162750244,
      "learning_rate": 9.700220455839422e-06,
      "loss": 0.4422,
      "num_input_tokens_seen": 4414200,
      "step": 6775
    },
    {
      "epoch": 3.9976415094339623,
      "grad_norm": 2.0757100582122803,
      "learning_rate": 9.69934240197006e-06,
      "loss": 0.4471,
      "num_input_tokens_seen": 4417528,
      "step": 6780
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.5017365217208862,
      "eval_runtime": 19.0297,
      "eval_samples_per_second": 89.124,
      "eval_steps_per_second": 22.281,
      "num_input_tokens_seen": 4419816,
      "step": 6784
    },
    {
      "epoch": 4.00058962264151,
      "grad_norm": 1.3792768716812134,
      "learning_rate": 9.698463103929542e-06,
      "loss": 0.4823,
      "num_input_tokens_seen": 4420648,
      "step": 6785
    },
    {
      "epoch": 4.003537735849057,
      "grad_norm": 1.2368861436843872,
      "learning_rate": 9.697582561950669e-06,
      "loss": 0.5052,
      "num_input_tokens_seen": 4424936,
      "step": 6790
    },
    {
      "epoch": 4.006485849056604,
      "grad_norm": 1.7510918378829956,
      "learning_rate": 9.696700776266568e-06,
      "loss": 0.3518,
      "num_input_tokens_seen": 4428552,
      "step": 6795
    },
    {
      "epoch": 4.009433962264151,
      "grad_norm": 2.8072891235351562,
      "learning_rate": 9.695817747110694e-06,
      "loss": 0.5587,
      "num_input_tokens_seen": 4432008,
      "step": 6800
    },
    {
      "epoch": 4.012382075471698,
      "grad_norm": 1.63905668258667,
      "learning_rate": 9.694933474716831e-06,
      "loss": 0.5803,
      "num_input_tokens_seen": 4434664,
      "step": 6805
    },
    {
      "epoch": 4.015330188679245,
      "grad_norm": 1.6267446279525757,
      "learning_rate": 9.6940479593191e-06,
      "loss": 0.5447,
      "num_input_tokens_seen": 4437384,
      "step": 6810
    },
    {
      "epoch": 4.0182783018867925,
      "grad_norm": 1.9597493410110474,
      "learning_rate": 9.693161201151942e-06,
      "loss": 0.4246,
      "num_input_tokens_seen": 4439816,
      "step": 6815
    },
    {
      "epoch": 4.02122641509434,
      "grad_norm": 1.4212974309921265,
      "learning_rate": 9.692273200450128e-06,
      "loss": 0.5792,
      "num_input_tokens_seen": 4444392,
      "step": 6820
    },
    {
      "epoch": 4.024174528301887,
      "grad_norm": 1.8943935632705688,
      "learning_rate": 9.69138395744876e-06,
      "loss": 0.4606,
      "num_input_tokens_seen": 4447784,
      "step": 6825
    },
    {
      "epoch": 4.027122641509434,
      "grad_norm": 2.747095823287964,
      "learning_rate": 9.690493472383274e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 4450728,
      "step": 6830
    },
    {
      "epoch": 4.030070754716981,
      "grad_norm": 1.6090871095657349,
      "learning_rate": 9.689601745489423e-06,
      "loss": 0.4411,
      "num_input_tokens_seen": 4454696,
      "step": 6835
    },
    {
      "epoch": 4.033018867924528,
      "grad_norm": 1.3398933410644531,
      "learning_rate": 9.6887087770033e-06,
      "loss": 0.3798,
      "num_input_tokens_seen": 4458472,
      "step": 6840
    },
    {
      "epoch": 4.035966981132075,
      "grad_norm": 1.32926607131958,
      "learning_rate": 9.687814567161322e-06,
      "loss": 0.3346,
      "num_input_tokens_seen": 4461320,
      "step": 6845
    },
    {
      "epoch": 4.038915094339623,
      "grad_norm": 1.4502230882644653,
      "learning_rate": 9.686919116200232e-06,
      "loss": 0.429,
      "num_input_tokens_seen": 4464488,
      "step": 6850
    },
    {
      "epoch": 4.04186320754717,
      "grad_norm": 1.5732759237289429,
      "learning_rate": 9.686022424357108e-06,
      "loss": 0.3172,
      "num_input_tokens_seen": 4468296,
      "step": 6855
    },
    {
      "epoch": 4.044811320754717,
      "grad_norm": 1.155003547668457,
      "learning_rate": 9.685124491869353e-06,
      "loss": 0.4137,
      "num_input_tokens_seen": 4471208,
      "step": 6860
    },
    {
      "epoch": 4.0477594339622645,
      "grad_norm": 2.535128355026245,
      "learning_rate": 9.684225318974696e-06,
      "loss": 0.5474,
      "num_input_tokens_seen": 4473992,
      "step": 6865
    },
    {
      "epoch": 4.050707547169812,
      "grad_norm": 1.5267579555511475,
      "learning_rate": 9.683324905911197e-06,
      "loss": 0.4243,
      "num_input_tokens_seen": 4477288,
      "step": 6870
    },
    {
      "epoch": 4.053655660377358,
      "grad_norm": 1.2087308168411255,
      "learning_rate": 9.682423252917245e-06,
      "loss": 0.4147,
      "num_input_tokens_seen": 4480840,
      "step": 6875
    },
    {
      "epoch": 4.056603773584905,
      "grad_norm": 1.7894783020019531,
      "learning_rate": 9.681520360231557e-06,
      "loss": 0.3714,
      "num_input_tokens_seen": 4484360,
      "step": 6880
    },
    {
      "epoch": 4.059551886792453,
      "grad_norm": 1.3800687789916992,
      "learning_rate": 9.680616228093178e-06,
      "loss": 0.4868,
      "num_input_tokens_seen": 4487368,
      "step": 6885
    },
    {
      "epoch": 4.0625,
      "grad_norm": 1.6571564674377441,
      "learning_rate": 9.67971085674148e-06,
      "loss": 0.5315,
      "num_input_tokens_seen": 4491016,
      "step": 6890
    },
    {
      "epoch": 4.065448113207547,
      "grad_norm": 1.4735475778579712,
      "learning_rate": 9.678804246416164e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 4494824,
      "step": 6895
    },
    {
      "epoch": 4.068396226415095,
      "grad_norm": 2.686755418777466,
      "learning_rate": 9.677896397357259e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 4497704,
      "step": 6900
    },
    {
      "epoch": 4.071344339622642,
      "grad_norm": 1.6068991422653198,
      "learning_rate": 9.676987309805121e-06,
      "loss": 0.4551,
      "num_input_tokens_seen": 4500808,
      "step": 6905
    },
    {
      "epoch": 4.074292452830188,
      "grad_norm": 2.4640142917633057,
      "learning_rate": 9.67607698400044e-06,
      "loss": 0.5567,
      "num_input_tokens_seen": 4504520,
      "step": 6910
    },
    {
      "epoch": 4.0772405660377355,
      "grad_norm": 1.974229335784912,
      "learning_rate": 9.67516542018422e-06,
      "loss": 0.4292,
      "num_input_tokens_seen": 4507400,
      "step": 6915
    },
    {
      "epoch": 4.080188679245283,
      "grad_norm": 1.3728193044662476,
      "learning_rate": 9.67425261859781e-06,
      "loss": 0.4802,
      "num_input_tokens_seen": 4510536,
      "step": 6920
    },
    {
      "epoch": 4.08313679245283,
      "grad_norm": 2.1037697792053223,
      "learning_rate": 9.673338579482871e-06,
      "loss": 0.4879,
      "num_input_tokens_seen": 4513608,
      "step": 6925
    },
    {
      "epoch": 4.086084905660377,
      "grad_norm": 2.734231948852539,
      "learning_rate": 9.672423303081404e-06,
      "loss": 0.4857,
      "num_input_tokens_seen": 4516744,
      "step": 6930
    },
    {
      "epoch": 4.089033018867925,
      "grad_norm": 1.0829623937606812,
      "learning_rate": 9.67150678963573e-06,
      "loss": 0.4397,
      "num_input_tokens_seen": 4519976,
      "step": 6935
    },
    {
      "epoch": 4.091981132075472,
      "grad_norm": 1.3850733041763306,
      "learning_rate": 9.670589039388501e-06,
      "loss": 0.3999,
      "num_input_tokens_seen": 4523016,
      "step": 6940
    },
    {
      "epoch": 4.094929245283019,
      "grad_norm": 1.633005976676941,
      "learning_rate": 9.669670052582695e-06,
      "loss": 0.342,
      "num_input_tokens_seen": 4526728,
      "step": 6945
    },
    {
      "epoch": 4.097877358490566,
      "grad_norm": 1.788474678993225,
      "learning_rate": 9.668749829461617e-06,
      "loss": 0.492,
      "num_input_tokens_seen": 4531688,
      "step": 6950
    },
    {
      "epoch": 4.100825471698113,
      "grad_norm": 2.1207964420318604,
      "learning_rate": 9.667828370268898e-06,
      "loss": 0.41,
      "num_input_tokens_seen": 4534632,
      "step": 6955
    },
    {
      "epoch": 4.10377358490566,
      "grad_norm": 1.9003021717071533,
      "learning_rate": 9.666905675248505e-06,
      "loss": 0.3829,
      "num_input_tokens_seen": 4538376,
      "step": 6960
    },
    {
      "epoch": 4.1067216981132075,
      "grad_norm": 1.2470641136169434,
      "learning_rate": 9.66598174464472e-06,
      "loss": 0.4228,
      "num_input_tokens_seen": 4542152,
      "step": 6965
    },
    {
      "epoch": 4.109669811320755,
      "grad_norm": 1.5783416032791138,
      "learning_rate": 9.665056578702157e-06,
      "loss": 0.4163,
      "num_input_tokens_seen": 4546120,
      "step": 6970
    },
    {
      "epoch": 4.112617924528302,
      "grad_norm": 2.2033491134643555,
      "learning_rate": 9.66413017766576e-06,
      "loss": 0.6089,
      "num_input_tokens_seen": 4550504,
      "step": 6975
    },
    {
      "epoch": 4.115566037735849,
      "grad_norm": 2.0115277767181396,
      "learning_rate": 9.663202541780799e-06,
      "loss": 0.3702,
      "num_input_tokens_seen": 4553384,
      "step": 6980
    },
    {
      "epoch": 4.118514150943396,
      "grad_norm": 2.5282280445098877,
      "learning_rate": 9.662273671292866e-06,
      "loss": 0.3795,
      "num_input_tokens_seen": 4556456,
      "step": 6985
    },
    {
      "epoch": 4.121462264150943,
      "grad_norm": 1.7331839799880981,
      "learning_rate": 9.661343566447886e-06,
      "loss": 0.5846,
      "num_input_tokens_seen": 4559112,
      "step": 6990
    },
    {
      "epoch": 4.12441037735849,
      "grad_norm": 1.9574525356292725,
      "learning_rate": 9.660412227492107e-06,
      "loss": 0.6382,
      "num_input_tokens_seen": 4563112,
      "step": 6995
    },
    {
      "epoch": 4.127358490566038,
      "grad_norm": 2.7696988582611084,
      "learning_rate": 9.659479654672106e-06,
      "loss": 0.4016,
      "num_input_tokens_seen": 4566120,
      "step": 7000
    },
    {
      "epoch": 4.130306603773585,
      "grad_norm": 1.5588330030441284,
      "learning_rate": 9.658545848234784e-06,
      "loss": 0.4114,
      "num_input_tokens_seen": 4568936,
      "step": 7005
    },
    {
      "epoch": 4.133254716981132,
      "grad_norm": 2.5199122428894043,
      "learning_rate": 9.657610808427372e-06,
      "loss": 0.7058,
      "num_input_tokens_seen": 4571944,
      "step": 7010
    },
    {
      "epoch": 4.1362028301886795,
      "grad_norm": 2.1615517139434814,
      "learning_rate": 9.656674535497425e-06,
      "loss": 0.4962,
      "num_input_tokens_seen": 4575624,
      "step": 7015
    },
    {
      "epoch": 4.139150943396227,
      "grad_norm": 2.2504501342773438,
      "learning_rate": 9.655737029692827e-06,
      "loss": 0.5222,
      "num_input_tokens_seen": 4578280,
      "step": 7020
    },
    {
      "epoch": 4.142099056603773,
      "grad_norm": 1.3587526082992554,
      "learning_rate": 9.654798291261785e-06,
      "loss": 0.4279,
      "num_input_tokens_seen": 4582024,
      "step": 7025
    },
    {
      "epoch": 4.1450471698113205,
      "grad_norm": 2.7430689334869385,
      "learning_rate": 9.653858320452833e-06,
      "loss": 0.4952,
      "num_input_tokens_seen": 4586344,
      "step": 7030
    },
    {
      "epoch": 4.147995283018868,
      "grad_norm": 1.940010905265808,
      "learning_rate": 9.652917117514836e-06,
      "loss": 0.3986,
      "num_input_tokens_seen": 4589576,
      "step": 7035
    },
    {
      "epoch": 4.150943396226415,
      "grad_norm": 1.2919933795928955,
      "learning_rate": 9.651974682696975e-06,
      "loss": 0.3703,
      "num_input_tokens_seen": 4593736,
      "step": 7040
    },
    {
      "epoch": 4.153891509433962,
      "grad_norm": 4.005589008331299,
      "learning_rate": 9.651031016248773e-06,
      "loss": 0.5401,
      "num_input_tokens_seen": 4596072,
      "step": 7045
    },
    {
      "epoch": 4.15683962264151,
      "grad_norm": 1.7463839054107666,
      "learning_rate": 9.65008611842006e-06,
      "loss": 0.3957,
      "num_input_tokens_seen": 4599528,
      "step": 7050
    },
    {
      "epoch": 4.159787735849057,
      "grad_norm": 2.3930132389068604,
      "learning_rate": 9.64913998946101e-06,
      "loss": 0.5267,
      "num_input_tokens_seen": 4601736,
      "step": 7055
    },
    {
      "epoch": 4.162735849056604,
      "grad_norm": 2.5286080837249756,
      "learning_rate": 9.648192629622109e-06,
      "loss": 0.4864,
      "num_input_tokens_seen": 4606088,
      "step": 7060
    },
    {
      "epoch": 4.165683962264151,
      "grad_norm": 2.717069625854492,
      "learning_rate": 9.647244039154178e-06,
      "loss": 0.452,
      "num_input_tokens_seen": 4608712,
      "step": 7065
    },
    {
      "epoch": 4.168632075471698,
      "grad_norm": 1.8588204383850098,
      "learning_rate": 9.64629421830836e-06,
      "loss": 0.5189,
      "num_input_tokens_seen": 4611752,
      "step": 7070
    },
    {
      "epoch": 4.171580188679245,
      "grad_norm": 1.5143080949783325,
      "learning_rate": 9.64534316733612e-06,
      "loss": 0.3933,
      "num_input_tokens_seen": 4614184,
      "step": 7075
    },
    {
      "epoch": 4.1745283018867925,
      "grad_norm": 1.2861015796661377,
      "learning_rate": 9.644390886489258e-06,
      "loss": 0.5269,
      "num_input_tokens_seen": 4618856,
      "step": 7080
    },
    {
      "epoch": 4.17747641509434,
      "grad_norm": 3.0115578174591064,
      "learning_rate": 9.643437376019893e-06,
      "loss": 0.5745,
      "num_input_tokens_seen": 4622568,
      "step": 7085
    },
    {
      "epoch": 4.180424528301887,
      "grad_norm": 5.937365531921387,
      "learning_rate": 9.64248263618047e-06,
      "loss": 0.4525,
      "num_input_tokens_seen": 4626856,
      "step": 7090
    },
    {
      "epoch": 4.183372641509434,
      "grad_norm": 2.178421974182129,
      "learning_rate": 9.64152666722376e-06,
      "loss": 0.4558,
      "num_input_tokens_seen": 4631048,
      "step": 7095
    },
    {
      "epoch": 4.186320754716981,
      "grad_norm": 1.7571662664413452,
      "learning_rate": 9.640569469402863e-06,
      "loss": 0.4706,
      "num_input_tokens_seen": 4634088,
      "step": 7100
    },
    {
      "epoch": 4.189268867924528,
      "grad_norm": 1.0506714582443237,
      "learning_rate": 9.639611042971198e-06,
      "loss": 0.4729,
      "num_input_tokens_seen": 4638056,
      "step": 7105
    },
    {
      "epoch": 4.192216981132075,
      "grad_norm": 4.74109411239624,
      "learning_rate": 9.638651388182514e-06,
      "loss": 0.5465,
      "num_input_tokens_seen": 4640744,
      "step": 7110
    },
    {
      "epoch": 4.195165094339623,
      "grad_norm": 2.675098419189453,
      "learning_rate": 9.637690505290884e-06,
      "loss": 0.5278,
      "num_input_tokens_seen": 4643432,
      "step": 7115
    },
    {
      "epoch": 4.19811320754717,
      "grad_norm": 3.565408229827881,
      "learning_rate": 9.636728394550705e-06,
      "loss": 0.473,
      "num_input_tokens_seen": 4646184,
      "step": 7120
    },
    {
      "epoch": 4.201061320754717,
      "grad_norm": 3.1837704181671143,
      "learning_rate": 9.6357650562167e-06,
      "loss": 0.4971,
      "num_input_tokens_seen": 4649064,
      "step": 7125
    },
    {
      "epoch": 4.2040094339622645,
      "grad_norm": 3.152163028717041,
      "learning_rate": 9.634800490543918e-06,
      "loss": 0.4081,
      "num_input_tokens_seen": 4652360,
      "step": 7130
    },
    {
      "epoch": 4.206957547169812,
      "grad_norm": 2.869594097137451,
      "learning_rate": 9.633834697787731e-06,
      "loss": 0.4657,
      "num_input_tokens_seen": 4655080,
      "step": 7135
    },
    {
      "epoch": 4.209905660377358,
      "grad_norm": 2.0781044960021973,
      "learning_rate": 9.632867678203836e-06,
      "loss": 0.519,
      "num_input_tokens_seen": 4657384,
      "step": 7140
    },
    {
      "epoch": 4.212853773584905,
      "grad_norm": 1.5709443092346191,
      "learning_rate": 9.631899432048258e-06,
      "loss": 0.3374,
      "num_input_tokens_seen": 4660296,
      "step": 7145
    },
    {
      "epoch": 4.215801886792453,
      "grad_norm": 1.8899904489517212,
      "learning_rate": 9.630929959577343e-06,
      "loss": 0.5037,
      "num_input_tokens_seen": 4664360,
      "step": 7150
    },
    {
      "epoch": 4.21875,
      "grad_norm": 1.7032573223114014,
      "learning_rate": 9.629959261047764e-06,
      "loss": 0.448,
      "num_input_tokens_seen": 4668360,
      "step": 7155
    },
    {
      "epoch": 4.221698113207547,
      "grad_norm": 1.3508163690567017,
      "learning_rate": 9.628987336716513e-06,
      "loss": 0.5411,
      "num_input_tokens_seen": 4671528,
      "step": 7160
    },
    {
      "epoch": 4.224646226415095,
      "grad_norm": 4.669062614440918,
      "learning_rate": 9.628014186840918e-06,
      "loss": 0.4886,
      "num_input_tokens_seen": 4675016,
      "step": 7165
    },
    {
      "epoch": 4.227594339622642,
      "grad_norm": 1.9774128198623657,
      "learning_rate": 9.62703981167862e-06,
      "loss": 0.4901,
      "num_input_tokens_seen": 4677768,
      "step": 7170
    },
    {
      "epoch": 4.230542452830188,
      "grad_norm": 1.5904439687728882,
      "learning_rate": 9.626064211487592e-06,
      "loss": 0.4378,
      "num_input_tokens_seen": 4681064,
      "step": 7175
    },
    {
      "epoch": 4.2334905660377355,
      "grad_norm": 2.5728485584259033,
      "learning_rate": 9.625087386526125e-06,
      "loss": 0.5005,
      "num_input_tokens_seen": 4683784,
      "step": 7180
    },
    {
      "epoch": 4.236438679245283,
      "grad_norm": 1.2340127229690552,
      "learning_rate": 9.624109337052839e-06,
      "loss": 0.4803,
      "num_input_tokens_seen": 4687080,
      "step": 7185
    },
    {
      "epoch": 4.23938679245283,
      "grad_norm": 2.733752489089966,
      "learning_rate": 9.623130063326678e-06,
      "loss": 0.4852,
      "num_input_tokens_seen": 4689544,
      "step": 7190
    },
    {
      "epoch": 4.242334905660377,
      "grad_norm": 1.2143913507461548,
      "learning_rate": 9.622149565606909e-06,
      "loss": 0.4812,
      "num_input_tokens_seen": 4693640,
      "step": 7195
    },
    {
      "epoch": 4.245283018867925,
      "grad_norm": 1.6723196506500244,
      "learning_rate": 9.621167844153122e-06,
      "loss": 0.4102,
      "num_input_tokens_seen": 4696616,
      "step": 7200
    },
    {
      "epoch": 4.248231132075472,
      "grad_norm": 2.3170886039733887,
      "learning_rate": 9.620184899225231e-06,
      "loss": 0.557,
      "num_input_tokens_seen": 4700904,
      "step": 7205
    },
    {
      "epoch": 4.251179245283019,
      "grad_norm": 1.6726902723312378,
      "learning_rate": 9.619200731083477e-06,
      "loss": 0.507,
      "num_input_tokens_seen": 4703912,
      "step": 7210
    },
    {
      "epoch": 4.254127358490566,
      "grad_norm": 0.8947533369064331,
      "learning_rate": 9.618215339988422e-06,
      "loss": 0.6216,
      "num_input_tokens_seen": 4707048,
      "step": 7215
    },
    {
      "epoch": 4.257075471698113,
      "grad_norm": 2.406144857406616,
      "learning_rate": 9.617228726200951e-06,
      "loss": 0.4979,
      "num_input_tokens_seen": 4710312,
      "step": 7220
    },
    {
      "epoch": 4.26002358490566,
      "grad_norm": 2.2040109634399414,
      "learning_rate": 9.616240889982277e-06,
      "loss": 0.384,
      "num_input_tokens_seen": 4713384,
      "step": 7225
    },
    {
      "epoch": 4.2629716981132075,
      "grad_norm": 2.256894588470459,
      "learning_rate": 9.61525183159393e-06,
      "loss": 0.4548,
      "num_input_tokens_seen": 4716168,
      "step": 7230
    },
    {
      "epoch": 4.265919811320755,
      "grad_norm": 2.1924822330474854,
      "learning_rate": 9.614261551297774e-06,
      "loss": 0.3812,
      "num_input_tokens_seen": 4719368,
      "step": 7235
    },
    {
      "epoch": 4.268867924528302,
      "grad_norm": 2.008199453353882,
      "learning_rate": 9.613270049355983e-06,
      "loss": 0.4746,
      "num_input_tokens_seen": 4722536,
      "step": 7240
    },
    {
      "epoch": 4.271816037735849,
      "grad_norm": 2.0688059329986572,
      "learning_rate": 9.612277326031065e-06,
      "loss": 0.4638,
      "num_input_tokens_seen": 4726056,
      "step": 7245
    },
    {
      "epoch": 4.274764150943396,
      "grad_norm": 2.481250762939453,
      "learning_rate": 9.611283381585848e-06,
      "loss": 0.4333,
      "num_input_tokens_seen": 4728712,
      "step": 7250
    },
    {
      "epoch": 4.277712264150943,
      "grad_norm": 1.925898790359497,
      "learning_rate": 9.61028821628348e-06,
      "loss": 0.4202,
      "num_input_tokens_seen": 4732328,
      "step": 7255
    },
    {
      "epoch": 4.28066037735849,
      "grad_norm": 2.0609130859375,
      "learning_rate": 9.609291830387439e-06,
      "loss": 0.7485,
      "num_input_tokens_seen": 4735240,
      "step": 7260
    },
    {
      "epoch": 4.283608490566038,
      "grad_norm": 2.0005359649658203,
      "learning_rate": 9.608294224161523e-06,
      "loss": 0.3855,
      "num_input_tokens_seen": 4739112,
      "step": 7265
    },
    {
      "epoch": 4.286556603773585,
      "grad_norm": 1.4678510427474976,
      "learning_rate": 9.607295397869847e-06,
      "loss": 0.5807,
      "num_input_tokens_seen": 4741640,
      "step": 7270
    },
    {
      "epoch": 4.289504716981132,
      "grad_norm": 3.272409200668335,
      "learning_rate": 9.60629535177686e-06,
      "loss": 0.5031,
      "num_input_tokens_seen": 4744104,
      "step": 7275
    },
    {
      "epoch": 4.2924528301886795,
      "grad_norm": 1.7416832447052002,
      "learning_rate": 9.605294086147325e-06,
      "loss": 0.6573,
      "num_input_tokens_seen": 4747880,
      "step": 7280
    },
    {
      "epoch": 4.295400943396227,
      "grad_norm": 2.107044219970703,
      "learning_rate": 9.604291601246333e-06,
      "loss": 0.5157,
      "num_input_tokens_seen": 4750728,
      "step": 7285
    },
    {
      "epoch": 4.298349056603773,
      "grad_norm": 1.8313056230545044,
      "learning_rate": 9.603287897339299e-06,
      "loss": 0.3689,
      "num_input_tokens_seen": 4754088,
      "step": 7290
    },
    {
      "epoch": 4.3012971698113205,
      "grad_norm": 1.8223328590393066,
      "learning_rate": 9.602282974691953e-06,
      "loss": 0.3235,
      "num_input_tokens_seen": 4757256,
      "step": 7295
    },
    {
      "epoch": 4.304245283018868,
      "grad_norm": 2.2340760231018066,
      "learning_rate": 9.601276833570355e-06,
      "loss": 0.5108,
      "num_input_tokens_seen": 4760264,
      "step": 7300
    },
    {
      "epoch": 4.307193396226415,
      "grad_norm": 2.3296284675598145,
      "learning_rate": 9.600269474240885e-06,
      "loss": 0.3417,
      "num_input_tokens_seen": 4762440,
      "step": 7305
    },
    {
      "epoch": 4.310141509433962,
      "grad_norm": 2.4569525718688965,
      "learning_rate": 9.599260896970246e-06,
      "loss": 0.6009,
      "num_input_tokens_seen": 4766600,
      "step": 7310
    },
    {
      "epoch": 4.31308962264151,
      "grad_norm": 3.064113140106201,
      "learning_rate": 9.598251102025463e-06,
      "loss": 0.3808,
      "num_input_tokens_seen": 4769320,
      "step": 7315
    },
    {
      "epoch": 4.316037735849057,
      "grad_norm": 2.808861494064331,
      "learning_rate": 9.597240089673882e-06,
      "loss": 0.5002,
      "num_input_tokens_seen": 4772136,
      "step": 7320
    },
    {
      "epoch": 4.318985849056604,
      "grad_norm": 3.154987096786499,
      "learning_rate": 9.596227860183175e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 4774920,
      "step": 7325
    },
    {
      "epoch": 4.321933962264151,
      "grad_norm": 2.1308095455169678,
      "learning_rate": 9.595214413821334e-06,
      "loss": 0.3371,
      "num_input_tokens_seen": 4778632,
      "step": 7330
    },
    {
      "epoch": 4.324882075471698,
      "grad_norm": 2.2511134147644043,
      "learning_rate": 9.59419975085667e-06,
      "loss": 0.5201,
      "num_input_tokens_seen": 4782248,
      "step": 7335
    },
    {
      "epoch": 4.327830188679245,
      "grad_norm": 2.4492440223693848,
      "learning_rate": 9.593183871557826e-06,
      "loss": 0.4916,
      "num_input_tokens_seen": 4785256,
      "step": 7340
    },
    {
      "epoch": 4.3307783018867925,
      "grad_norm": 1.8863738775253296,
      "learning_rate": 9.592166776193754e-06,
      "loss": 0.3618,
      "num_input_tokens_seen": 4788360,
      "step": 7345
    },
    {
      "epoch": 4.33372641509434,
      "grad_norm": 2.3676235675811768,
      "learning_rate": 9.591148465033738e-06,
      "loss": 0.4221,
      "num_input_tokens_seen": 4791848,
      "step": 7350
    },
    {
      "epoch": 4.336674528301887,
      "grad_norm": 2.252833127975464,
      "learning_rate": 9.590128938347378e-06,
      "loss": 0.4216,
      "num_input_tokens_seen": 4794760,
      "step": 7355
    },
    {
      "epoch": 4.339622641509434,
      "grad_norm": 1.0753871202468872,
      "learning_rate": 9.589108196404599e-06,
      "loss": 0.4328,
      "num_input_tokens_seen": 4797608,
      "step": 7360
    },
    {
      "epoch": 4.342570754716981,
      "grad_norm": 2.2163984775543213,
      "learning_rate": 9.588086239475649e-06,
      "loss": 0.4152,
      "num_input_tokens_seen": 4801224,
      "step": 7365
    },
    {
      "epoch": 4.345518867924528,
      "grad_norm": 2.5468742847442627,
      "learning_rate": 9.587063067831092e-06,
      "loss": 0.4721,
      "num_input_tokens_seen": 4804488,
      "step": 7370
    },
    {
      "epoch": 4.348466981132075,
      "grad_norm": 1.7193576097488403,
      "learning_rate": 9.586038681741818e-06,
      "loss": 0.4086,
      "num_input_tokens_seen": 4806984,
      "step": 7375
    },
    {
      "epoch": 4.351415094339623,
      "grad_norm": 1.8865504264831543,
      "learning_rate": 9.58501308147904e-06,
      "loss": 0.6139,
      "num_input_tokens_seen": 4810280,
      "step": 7380
    },
    {
      "epoch": 4.35436320754717,
      "grad_norm": 1.2079781293869019,
      "learning_rate": 9.583986267314288e-06,
      "loss": 0.4688,
      "num_input_tokens_seen": 4813320,
      "step": 7385
    },
    {
      "epoch": 4.357311320754717,
      "grad_norm": 1.7007620334625244,
      "learning_rate": 9.582958239519416e-06,
      "loss": 0.5089,
      "num_input_tokens_seen": 4816296,
      "step": 7390
    },
    {
      "epoch": 4.3602594339622645,
      "grad_norm": 2.441124200820923,
      "learning_rate": 9.581928998366597e-06,
      "loss": 0.5664,
      "num_input_tokens_seen": 4819176,
      "step": 7395
    },
    {
      "epoch": 4.363207547169811,
      "grad_norm": 1.5746839046478271,
      "learning_rate": 9.58089854412833e-06,
      "loss": 0.3879,
      "num_input_tokens_seen": 4822632,
      "step": 7400
    },
    {
      "epoch": 4.366155660377358,
      "grad_norm": 2.8169047832489014,
      "learning_rate": 9.579866877077431e-06,
      "loss": 0.475,
      "num_input_tokens_seen": 4825480,
      "step": 7405
    },
    {
      "epoch": 4.369103773584905,
      "grad_norm": 1.3569514751434326,
      "learning_rate": 9.578833997487038e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 4828712,
      "step": 7410
    },
    {
      "epoch": 4.372051886792453,
      "grad_norm": 2.8722033500671387,
      "learning_rate": 9.57779990563061e-06,
      "loss": 0.418,
      "num_input_tokens_seen": 4831944,
      "step": 7415
    },
    {
      "epoch": 4.375,
      "grad_norm": 2.1287319660186768,
      "learning_rate": 9.576764601781928e-06,
      "loss": 0.413,
      "num_input_tokens_seen": 4835528,
      "step": 7420
    },
    {
      "epoch": 4.377948113207547,
      "grad_norm": 2.891575574874878,
      "learning_rate": 9.575728086215093e-06,
      "loss": 0.4642,
      "num_input_tokens_seen": 4838152,
      "step": 7425
    },
    {
      "epoch": 4.380896226415095,
      "grad_norm": 6.918198585510254,
      "learning_rate": 9.574690359204527e-06,
      "loss": 0.5152,
      "num_input_tokens_seen": 4840520,
      "step": 7430
    },
    {
      "epoch": 4.383844339622642,
      "grad_norm": 1.0360878705978394,
      "learning_rate": 9.573651421024972e-06,
      "loss": 0.3794,
      "num_input_tokens_seen": 4845064,
      "step": 7435
    },
    {
      "epoch": 4.386792452830189,
      "grad_norm": 3.0622286796569824,
      "learning_rate": 9.572611271951494e-06,
      "loss": 0.5777,
      "num_input_tokens_seen": 4847464,
      "step": 7440
    },
    {
      "epoch": 4.3897405660377355,
      "grad_norm": 2.212538957595825,
      "learning_rate": 9.571569912259473e-06,
      "loss": 0.5316,
      "num_input_tokens_seen": 4850728,
      "step": 7445
    },
    {
      "epoch": 4.392688679245283,
      "grad_norm": 1.8893916606903076,
      "learning_rate": 9.570527342224614e-06,
      "loss": 0.4438,
      "num_input_tokens_seen": 4854440,
      "step": 7450
    },
    {
      "epoch": 4.39563679245283,
      "grad_norm": 3.0040769577026367,
      "learning_rate": 9.569483562122945e-06,
      "loss": 0.499,
      "num_input_tokens_seen": 4857960,
      "step": 7455
    },
    {
      "epoch": 4.398584905660377,
      "grad_norm": 1.3081634044647217,
      "learning_rate": 9.568438572230811e-06,
      "loss": 0.3592,
      "num_input_tokens_seen": 4860840,
      "step": 7460
    },
    {
      "epoch": 4.401533018867925,
      "grad_norm": 1.9627350568771362,
      "learning_rate": 9.567392372824873e-06,
      "loss": 0.4627,
      "num_input_tokens_seen": 4863240,
      "step": 7465
    },
    {
      "epoch": 4.404481132075472,
      "grad_norm": 2.4167428016662598,
      "learning_rate": 9.566344964182123e-06,
      "loss": 0.4864,
      "num_input_tokens_seen": 4866248,
      "step": 7470
    },
    {
      "epoch": 4.407429245283019,
      "grad_norm": 3.1545193195343018,
      "learning_rate": 9.565296346579862e-06,
      "loss": 0.3894,
      "num_input_tokens_seen": 4869256,
      "step": 7475
    },
    {
      "epoch": 4.410377358490566,
      "grad_norm": 4.5815815925598145,
      "learning_rate": 9.564246520295719e-06,
      "loss": 0.5238,
      "num_input_tokens_seen": 4871912,
      "step": 7480
    },
    {
      "epoch": 4.413325471698113,
      "grad_norm": 1.6332738399505615,
      "learning_rate": 9.563195485607638e-06,
      "loss": 0.4309,
      "num_input_tokens_seen": 4875944,
      "step": 7485
    },
    {
      "epoch": 4.41627358490566,
      "grad_norm": 1.6208539009094238,
      "learning_rate": 9.562143242793885e-06,
      "loss": 0.4546,
      "num_input_tokens_seen": 4878280,
      "step": 7490
    },
    {
      "epoch": 4.4192216981132075,
      "grad_norm": 2.425055503845215,
      "learning_rate": 9.561089792133048e-06,
      "loss": 0.4125,
      "num_input_tokens_seen": 4881352,
      "step": 7495
    },
    {
      "epoch": 4.422169811320755,
      "grad_norm": 1.652378797531128,
      "learning_rate": 9.560035133904031e-06,
      "loss": 0.499,
      "num_input_tokens_seen": 4884808,
      "step": 7500
    },
    {
      "epoch": 4.425117924528302,
      "grad_norm": 1.2958364486694336,
      "learning_rate": 9.55897926838606e-06,
      "loss": 0.4404,
      "num_input_tokens_seen": 4888040,
      "step": 7505
    },
    {
      "epoch": 4.428066037735849,
      "grad_norm": 2.2129578590393066,
      "learning_rate": 9.55792219585868e-06,
      "loss": 0.4059,
      "num_input_tokens_seen": 4891144,
      "step": 7510
    },
    {
      "epoch": 4.431014150943396,
      "grad_norm": 2.1508052349090576,
      "learning_rate": 9.556863916601754e-06,
      "loss": 0.4684,
      "num_input_tokens_seen": 4894344,
      "step": 7515
    },
    {
      "epoch": 4.433962264150943,
      "grad_norm": 1.6770113706588745,
      "learning_rate": 9.555804430895467e-06,
      "loss": 0.5369,
      "num_input_tokens_seen": 4897352,
      "step": 7520
    },
    {
      "epoch": 4.43691037735849,
      "grad_norm": 1.7450464963912964,
      "learning_rate": 9.554743739020325e-06,
      "loss": 0.3056,
      "num_input_tokens_seen": 4900040,
      "step": 7525
    },
    {
      "epoch": 4.439858490566038,
      "grad_norm": 1.5154356956481934,
      "learning_rate": 9.553681841257146e-06,
      "loss": 0.5413,
      "num_input_tokens_seen": 4903560,
      "step": 7530
    },
    {
      "epoch": 4.442806603773585,
      "grad_norm": 2.1468966007232666,
      "learning_rate": 9.552618737887073e-06,
      "loss": 0.4726,
      "num_input_tokens_seen": 4907496,
      "step": 7535
    },
    {
      "epoch": 4.445754716981132,
      "grad_norm": 2.379988193511963,
      "learning_rate": 9.55155442919157e-06,
      "loss": 0.5243,
      "num_input_tokens_seen": 4910376,
      "step": 7540
    },
    {
      "epoch": 4.4487028301886795,
      "grad_norm": 2.343595266342163,
      "learning_rate": 9.550488915452416e-06,
      "loss": 0.4509,
      "num_input_tokens_seen": 4913512,
      "step": 7545
    },
    {
      "epoch": 4.451650943396227,
      "grad_norm": 2.2828774452209473,
      "learning_rate": 9.54942219695171e-06,
      "loss": 0.5485,
      "num_input_tokens_seen": 4916264,
      "step": 7550
    },
    {
      "epoch": 4.454599056603773,
      "grad_norm": 1.4305038452148438,
      "learning_rate": 9.54835427397187e-06,
      "loss": 0.4777,
      "num_input_tokens_seen": 4920200,
      "step": 7555
    },
    {
      "epoch": 4.4575471698113205,
      "grad_norm": 2.3910508155822754,
      "learning_rate": 9.547285146795634e-06,
      "loss": 0.5433,
      "num_input_tokens_seen": 4923304,
      "step": 7560
    },
    {
      "epoch": 4.460495283018868,
      "grad_norm": 2.9457247257232666,
      "learning_rate": 9.546214815706059e-06,
      "loss": 0.5329,
      "num_input_tokens_seen": 4926312,
      "step": 7565
    },
    {
      "epoch": 4.463443396226415,
      "grad_norm": 4.562378406524658,
      "learning_rate": 9.545143280986518e-06,
      "loss": 0.3715,
      "num_input_tokens_seen": 4930280,
      "step": 7570
    },
    {
      "epoch": 4.466391509433962,
      "grad_norm": 1.3466272354125977,
      "learning_rate": 9.544070542920703e-06,
      "loss": 0.5294,
      "num_input_tokens_seen": 4934024,
      "step": 7575
    },
    {
      "epoch": 4.46933962264151,
      "grad_norm": 1.6014982461929321,
      "learning_rate": 9.542996601792629e-06,
      "loss": 0.552,
      "num_input_tokens_seen": 4937096,
      "step": 7580
    },
    {
      "epoch": 4.472287735849057,
      "grad_norm": 4.151698589324951,
      "learning_rate": 9.541921457886624e-06,
      "loss": 0.5132,
      "num_input_tokens_seen": 4939848,
      "step": 7585
    },
    {
      "epoch": 4.475235849056604,
      "grad_norm": 2.1281020641326904,
      "learning_rate": 9.54084511148734e-06,
      "loss": 0.4083,
      "num_input_tokens_seen": 4942696,
      "step": 7590
    },
    {
      "epoch": 4.478183962264151,
      "grad_norm": 2.5933918952941895,
      "learning_rate": 9.539767562879742e-06,
      "loss": 0.4343,
      "num_input_tokens_seen": 4945704,
      "step": 7595
    },
    {
      "epoch": 4.481132075471698,
      "grad_norm": 2.99198579788208,
      "learning_rate": 9.538688812349118e-06,
      "loss": 0.5065,
      "num_input_tokens_seen": 4948680,
      "step": 7600
    },
    {
      "epoch": 4.484080188679245,
      "grad_norm": 1.916102647781372,
      "learning_rate": 9.537608860181069e-06,
      "loss": 0.437,
      "num_input_tokens_seen": 4951688,
      "step": 7605
    },
    {
      "epoch": 4.4870283018867925,
      "grad_norm": 1.8582398891448975,
      "learning_rate": 9.536527706661519e-06,
      "loss": 0.5244,
      "num_input_tokens_seen": 4955752,
      "step": 7610
    },
    {
      "epoch": 4.48997641509434,
      "grad_norm": 1.8352452516555786,
      "learning_rate": 9.535445352076707e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 4958888,
      "step": 7615
    },
    {
      "epoch": 4.492924528301887,
      "grad_norm": 1.9488977193832397,
      "learning_rate": 9.534361796713191e-06,
      "loss": 0.4895,
      "num_input_tokens_seen": 4961992,
      "step": 7620
    },
    {
      "epoch": 4.495872641509434,
      "grad_norm": 2.2810184955596924,
      "learning_rate": 9.533277040857847e-06,
      "loss": 0.4034,
      "num_input_tokens_seen": 4964616,
      "step": 7625
    },
    {
      "epoch": 4.498820754716981,
      "grad_norm": 2.05924129486084,
      "learning_rate": 9.53219108479787e-06,
      "loss": 0.3751,
      "num_input_tokens_seen": 4967912,
      "step": 7630
    },
    {
      "epoch": 4.501768867924528,
      "grad_norm": 2.593048334121704,
      "learning_rate": 9.53110392882077e-06,
      "loss": 0.2893,
      "num_input_tokens_seen": 4970280,
      "step": 7635
    },
    {
      "epoch": 4.504716981132075,
      "grad_norm": 1.8708491325378418,
      "learning_rate": 9.530015573214378e-06,
      "loss": 0.467,
      "num_input_tokens_seen": 4973224,
      "step": 7640
    },
    {
      "epoch": 4.507665094339623,
      "grad_norm": 1.2913099527359009,
      "learning_rate": 9.528926018266837e-06,
      "loss": 0.5468,
      "num_input_tokens_seen": 4976808,
      "step": 7645
    },
    {
      "epoch": 4.51061320754717,
      "grad_norm": 1.45063316822052,
      "learning_rate": 9.527835264266617e-06,
      "loss": 0.5315,
      "num_input_tokens_seen": 4982312,
      "step": 7650
    },
    {
      "epoch": 4.513561320754717,
      "grad_norm": 1.5617191791534424,
      "learning_rate": 9.526743311502496e-06,
      "loss": 0.4115,
      "num_input_tokens_seen": 4984744,
      "step": 7655
    },
    {
      "epoch": 4.5165094339622645,
      "grad_norm": 2.0885818004608154,
      "learning_rate": 9.525650160263573e-06,
      "loss": 0.4663,
      "num_input_tokens_seen": 4987816,
      "step": 7660
    },
    {
      "epoch": 4.519457547169811,
      "grad_norm": 1.497378945350647,
      "learning_rate": 9.524555810839267e-06,
      "loss": 0.5458,
      "num_input_tokens_seen": 4990856,
      "step": 7665
    },
    {
      "epoch": 4.522405660377358,
      "grad_norm": 2.121843099594116,
      "learning_rate": 9.523460263519309e-06,
      "loss": 0.3345,
      "num_input_tokens_seen": 4993384,
      "step": 7670
    },
    {
      "epoch": 4.525353773584905,
      "grad_norm": 1.7966277599334717,
      "learning_rate": 9.522363518593753e-06,
      "loss": 0.4313,
      "num_input_tokens_seen": 4996872,
      "step": 7675
    },
    {
      "epoch": 4.528301886792453,
      "grad_norm": 1.5318876504898071,
      "learning_rate": 9.521265576352963e-06,
      "loss": 0.4791,
      "num_input_tokens_seen": 4999304,
      "step": 7680
    },
    {
      "epoch": 4.53125,
      "grad_norm": 1.2461329698562622,
      "learning_rate": 9.520166437087628e-06,
      "loss": 0.5386,
      "num_input_tokens_seen": 5002760,
      "step": 7685
    },
    {
      "epoch": 4.534198113207547,
      "grad_norm": 1.2866296768188477,
      "learning_rate": 9.519066101088748e-06,
      "loss": 0.3854,
      "num_input_tokens_seen": 5005704,
      "step": 7690
    },
    {
      "epoch": 4.537146226415095,
      "grad_norm": 1.8826137781143188,
      "learning_rate": 9.51796456864764e-06,
      "loss": 0.3951,
      "num_input_tokens_seen": 5008616,
      "step": 7695
    },
    {
      "epoch": 4.540094339622642,
      "grad_norm": 1.1628481149673462,
      "learning_rate": 9.516861840055942e-06,
      "loss": 0.4036,
      "num_input_tokens_seen": 5011624,
      "step": 7700
    },
    {
      "epoch": 4.543042452830189,
      "grad_norm": 2.1325292587280273,
      "learning_rate": 9.515757915605604e-06,
      "loss": 0.5455,
      "num_input_tokens_seen": 5014760,
      "step": 7705
    },
    {
      "epoch": 4.5459905660377355,
      "grad_norm": 2.718200922012329,
      "learning_rate": 9.514652795588899e-06,
      "loss": 0.4415,
      "num_input_tokens_seen": 5018184,
      "step": 7710
    },
    {
      "epoch": 4.548938679245283,
      "grad_norm": 1.7247861623764038,
      "learning_rate": 9.513546480298405e-06,
      "loss": 0.5423,
      "num_input_tokens_seen": 5021448,
      "step": 7715
    },
    {
      "epoch": 4.55188679245283,
      "grad_norm": 1.761030673980713,
      "learning_rate": 9.512438970027032e-06,
      "loss": 0.4206,
      "num_input_tokens_seen": 5024456,
      "step": 7720
    },
    {
      "epoch": 4.554834905660377,
      "grad_norm": 1.4829206466674805,
      "learning_rate": 9.511330265067992e-06,
      "loss": 0.4736,
      "num_input_tokens_seen": 5026792,
      "step": 7725
    },
    {
      "epoch": 4.557783018867925,
      "grad_norm": 1.4282112121582031,
      "learning_rate": 9.510220365714822e-06,
      "loss": 0.3369,
      "num_input_tokens_seen": 5029128,
      "step": 7730
    },
    {
      "epoch": 4.560731132075472,
      "grad_norm": 1.5470753908157349,
      "learning_rate": 9.509109272261373e-06,
      "loss": 0.3452,
      "num_input_tokens_seen": 5033064,
      "step": 7735
    },
    {
      "epoch": 4.563679245283019,
      "grad_norm": 1.2875882387161255,
      "learning_rate": 9.50799698500181e-06,
      "loss": 0.3933,
      "num_input_tokens_seen": 5036392,
      "step": 7740
    },
    {
      "epoch": 4.566627358490566,
      "grad_norm": 2.0441648960113525,
      "learning_rate": 9.506883504230618e-06,
      "loss": 0.4035,
      "num_input_tokens_seen": 5040296,
      "step": 7745
    },
    {
      "epoch": 4.569575471698113,
      "grad_norm": 1.0374459028244019,
      "learning_rate": 9.505768830242593e-06,
      "loss": 0.4266,
      "num_input_tokens_seen": 5043688,
      "step": 7750
    },
    {
      "epoch": 4.57252358490566,
      "grad_norm": 2.1011946201324463,
      "learning_rate": 9.504652963332852e-06,
      "loss": 0.489,
      "num_input_tokens_seen": 5046152,
      "step": 7755
    },
    {
      "epoch": 4.5754716981132075,
      "grad_norm": 2.1412925720214844,
      "learning_rate": 9.503535903796825e-06,
      "loss": 0.496,
      "num_input_tokens_seen": 5048360,
      "step": 7760
    },
    {
      "epoch": 4.578419811320755,
      "grad_norm": 2.4934630393981934,
      "learning_rate": 9.50241765193026e-06,
      "loss": 0.3689,
      "num_input_tokens_seen": 5051112,
      "step": 7765
    },
    {
      "epoch": 4.581367924528302,
      "grad_norm": 5.191808223724365,
      "learning_rate": 9.501298208029214e-06,
      "loss": 0.4181,
      "num_input_tokens_seen": 5053480,
      "step": 7770
    },
    {
      "epoch": 4.584316037735849,
      "grad_norm": 1.1557424068450928,
      "learning_rate": 9.500177572390071e-06,
      "loss": 0.3971,
      "num_input_tokens_seen": 5057000,
      "step": 7775
    },
    {
      "epoch": 4.587264150943396,
      "grad_norm": 2.4938762187957764,
      "learning_rate": 9.49905574530952e-06,
      "loss": 0.4833,
      "num_input_tokens_seen": 5060072,
      "step": 7780
    },
    {
      "epoch": 4.590212264150943,
      "grad_norm": 4.864057540893555,
      "learning_rate": 9.497932727084571e-06,
      "loss": 0.626,
      "num_input_tokens_seen": 5062568,
      "step": 7785
    },
    {
      "epoch": 4.59316037735849,
      "grad_norm": 1.5662068128585815,
      "learning_rate": 9.496808518012545e-06,
      "loss": 0.4124,
      "num_input_tokens_seen": 5066600,
      "step": 7790
    },
    {
      "epoch": 4.596108490566038,
      "grad_norm": 1.232354760169983,
      "learning_rate": 9.495683118391087e-06,
      "loss": 0.4609,
      "num_input_tokens_seen": 5069384,
      "step": 7795
    },
    {
      "epoch": 4.599056603773585,
      "grad_norm": 1.6870357990264893,
      "learning_rate": 9.494556528518146e-06,
      "loss": 0.4144,
      "num_input_tokens_seen": 5072520,
      "step": 7800
    },
    {
      "epoch": 4.602004716981132,
      "grad_norm": 5.364602088928223,
      "learning_rate": 9.493428748691995e-06,
      "loss": 0.5683,
      "num_input_tokens_seen": 5075208,
      "step": 7805
    },
    {
      "epoch": 4.6049528301886795,
      "grad_norm": 1.4015411138534546,
      "learning_rate": 9.492299779211215e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 5077992,
      "step": 7810
    },
    {
      "epoch": 4.607900943396227,
      "grad_norm": 1.987358570098877,
      "learning_rate": 9.49116962037471e-06,
      "loss": 0.6076,
      "num_input_tokens_seen": 5080456,
      "step": 7815
    },
    {
      "epoch": 4.610849056603773,
      "grad_norm": 2.5312421321868896,
      "learning_rate": 9.490038272481691e-06,
      "loss": 0.4501,
      "num_input_tokens_seen": 5084168,
      "step": 7820
    },
    {
      "epoch": 4.6137971698113205,
      "grad_norm": 4.634452819824219,
      "learning_rate": 9.488905735831689e-06,
      "loss": 0.492,
      "num_input_tokens_seen": 5086824,
      "step": 7825
    },
    {
      "epoch": 4.616745283018868,
      "grad_norm": 3.1330456733703613,
      "learning_rate": 9.487772010724548e-06,
      "loss": 0.3705,
      "num_input_tokens_seen": 5089928,
      "step": 7830
    },
    {
      "epoch": 4.619693396226415,
      "grad_norm": 1.1904127597808838,
      "learning_rate": 9.486637097460425e-06,
      "loss": 0.3599,
      "num_input_tokens_seen": 5093928,
      "step": 7835
    },
    {
      "epoch": 4.622641509433962,
      "grad_norm": 2.6280577182769775,
      "learning_rate": 9.485500996339793e-06,
      "loss": 0.404,
      "num_input_tokens_seen": 5096680,
      "step": 7840
    },
    {
      "epoch": 4.62558962264151,
      "grad_norm": 1.6727854013442993,
      "learning_rate": 9.484363707663443e-06,
      "loss": 0.4298,
      "num_input_tokens_seen": 5099496,
      "step": 7845
    },
    {
      "epoch": 4.628537735849057,
      "grad_norm": 4.714191913604736,
      "learning_rate": 9.483225231732474e-06,
      "loss": 0.5356,
      "num_input_tokens_seen": 5102472,
      "step": 7850
    },
    {
      "epoch": 4.631485849056604,
      "grad_norm": 1.9147136211395264,
      "learning_rate": 9.482085568848302e-06,
      "loss": 0.4106,
      "num_input_tokens_seen": 5105768,
      "step": 7855
    },
    {
      "epoch": 4.634433962264151,
      "grad_norm": 1.8578300476074219,
      "learning_rate": 9.480944719312659e-06,
      "loss": 0.6216,
      "num_input_tokens_seen": 5109224,
      "step": 7860
    },
    {
      "epoch": 4.637382075471698,
      "grad_norm": 3.6490964889526367,
      "learning_rate": 9.47980268342759e-06,
      "loss": 0.5203,
      "num_input_tokens_seen": 5112392,
      "step": 7865
    },
    {
      "epoch": 4.640330188679245,
      "grad_norm": 2.039979934692383,
      "learning_rate": 9.478659461495456e-06,
      "loss": 0.4738,
      "num_input_tokens_seen": 5114376,
      "step": 7870
    },
    {
      "epoch": 4.6432783018867925,
      "grad_norm": 1.3141827583312988,
      "learning_rate": 9.477515053818926e-06,
      "loss": 0.4424,
      "num_input_tokens_seen": 5118536,
      "step": 7875
    },
    {
      "epoch": 4.64622641509434,
      "grad_norm": 3.704115867614746,
      "learning_rate": 9.476369460700988e-06,
      "loss": 0.4267,
      "num_input_tokens_seen": 5121736,
      "step": 7880
    },
    {
      "epoch": 4.649174528301887,
      "grad_norm": 2.4232606887817383,
      "learning_rate": 9.475222682444944e-06,
      "loss": 0.4979,
      "num_input_tokens_seen": 5124520,
      "step": 7885
    },
    {
      "epoch": 4.652122641509434,
      "grad_norm": 1.2228881120681763,
      "learning_rate": 9.474074719354406e-06,
      "loss": 0.4655,
      "num_input_tokens_seen": 5127912,
      "step": 7890
    },
    {
      "epoch": 4.655070754716981,
      "grad_norm": 1.836904764175415,
      "learning_rate": 9.472925571733306e-06,
      "loss": 0.4555,
      "num_input_tokens_seen": 5130824,
      "step": 7895
    },
    {
      "epoch": 4.658018867924528,
      "grad_norm": 1.9640544652938843,
      "learning_rate": 9.471775239885883e-06,
      "loss": 0.4919,
      "num_input_tokens_seen": 5134024,
      "step": 7900
    },
    {
      "epoch": 4.660966981132075,
      "grad_norm": 2.0581021308898926,
      "learning_rate": 9.470623724116693e-06,
      "loss": 0.4143,
      "num_input_tokens_seen": 5137672,
      "step": 7905
    },
    {
      "epoch": 4.663915094339623,
      "grad_norm": 1.1155809164047241,
      "learning_rate": 9.469471024730606e-06,
      "loss": 0.4364,
      "num_input_tokens_seen": 5141480,
      "step": 7910
    },
    {
      "epoch": 4.66686320754717,
      "grad_norm": 1.1886358261108398,
      "learning_rate": 9.4683171420328e-06,
      "loss": 0.5073,
      "num_input_tokens_seen": 5145800,
      "step": 7915
    },
    {
      "epoch": 4.669811320754717,
      "grad_norm": 2.5233659744262695,
      "learning_rate": 9.467162076328776e-06,
      "loss": 0.4013,
      "num_input_tokens_seen": 5149320,
      "step": 7920
    },
    {
      "epoch": 4.6727594339622645,
      "grad_norm": 3.3065855503082275,
      "learning_rate": 9.466005827924337e-06,
      "loss": 0.5339,
      "num_input_tokens_seen": 5152072,
      "step": 7925
    },
    {
      "epoch": 4.675707547169811,
      "grad_norm": 1.9489041566848755,
      "learning_rate": 9.46484839712561e-06,
      "loss": 0.6446,
      "num_input_tokens_seen": 5155304,
      "step": 7930
    },
    {
      "epoch": 4.678655660377358,
      "grad_norm": 4.527299880981445,
      "learning_rate": 9.463689784239026e-06,
      "loss": 0.4136,
      "num_input_tokens_seen": 5158216,
      "step": 7935
    },
    {
      "epoch": 4.681603773584905,
      "grad_norm": 2.0233635902404785,
      "learning_rate": 9.462529989571334e-06,
      "loss": 0.4616,
      "num_input_tokens_seen": 5161896,
      "step": 7940
    },
    {
      "epoch": 4.684551886792453,
      "grad_norm": 1.519830584526062,
      "learning_rate": 9.461369013429595e-06,
      "loss": 0.5737,
      "num_input_tokens_seen": 5167880,
      "step": 7945
    },
    {
      "epoch": 4.6875,
      "grad_norm": 2.645507335662842,
      "learning_rate": 9.460206856121183e-06,
      "loss": 0.5744,
      "num_input_tokens_seen": 5171112,
      "step": 7950
    },
    {
      "epoch": 4.690448113207547,
      "grad_norm": 1.7675485610961914,
      "learning_rate": 9.459043517953786e-06,
      "loss": 0.4463,
      "num_input_tokens_seen": 5174920,
      "step": 7955
    },
    {
      "epoch": 4.693396226415095,
      "grad_norm": 1.7541604042053223,
      "learning_rate": 9.457878999235396e-06,
      "loss": 0.4516,
      "num_input_tokens_seen": 5178088,
      "step": 7960
    },
    {
      "epoch": 4.696344339622642,
      "grad_norm": 1.7906488180160522,
      "learning_rate": 9.45671330027433e-06,
      "loss": 0.6278,
      "num_input_tokens_seen": 5181448,
      "step": 7965
    },
    {
      "epoch": 4.699292452830189,
      "grad_norm": 1.631229043006897,
      "learning_rate": 9.45554642137921e-06,
      "loss": 0.4615,
      "num_input_tokens_seen": 5183976,
      "step": 7970
    },
    {
      "epoch": 4.7022405660377355,
      "grad_norm": 1.7123756408691406,
      "learning_rate": 9.454378362858974e-06,
      "loss": 0.37,
      "num_input_tokens_seen": 5187688,
      "step": 7975
    },
    {
      "epoch": 4.705188679245283,
      "grad_norm": 1.5067150592803955,
      "learning_rate": 9.453209125022867e-06,
      "loss": 0.6165,
      "num_input_tokens_seen": 5190504,
      "step": 7980
    },
    {
      "epoch": 4.70813679245283,
      "grad_norm": 1.6526689529418945,
      "learning_rate": 9.452038708180453e-06,
      "loss": 0.4526,
      "num_input_tokens_seen": 5193032,
      "step": 7985
    },
    {
      "epoch": 4.711084905660377,
      "grad_norm": 2.604750633239746,
      "learning_rate": 9.450867112641603e-06,
      "loss": 0.4659,
      "num_input_tokens_seen": 5196968,
      "step": 7990
    },
    {
      "epoch": 4.714033018867925,
      "grad_norm": 2.1780645847320557,
      "learning_rate": 9.449694338716506e-06,
      "loss": 0.4777,
      "num_input_tokens_seen": 5199848,
      "step": 7995
    },
    {
      "epoch": 4.716981132075472,
      "grad_norm": 1.6224945783615112,
      "learning_rate": 9.448520386715653e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 5204040,
      "step": 8000
    },
    {
      "epoch": 4.719929245283019,
      "grad_norm": 1.8815759420394897,
      "learning_rate": 9.447345256949855e-06,
      "loss": 0.3828,
      "num_input_tokens_seen": 5207240,
      "step": 8005
    },
    {
      "epoch": 4.722877358490566,
      "grad_norm": 1.3700987100601196,
      "learning_rate": 9.446168949730234e-06,
      "loss": 0.5591,
      "num_input_tokens_seen": 5210312,
      "step": 8010
    },
    {
      "epoch": 4.725825471698113,
      "grad_norm": 1.7998849153518677,
      "learning_rate": 9.444991465368223e-06,
      "loss": 0.4712,
      "num_input_tokens_seen": 5214760,
      "step": 8015
    },
    {
      "epoch": 4.72877358490566,
      "grad_norm": 1.5192891359329224,
      "learning_rate": 9.443812804175562e-06,
      "loss": 0.5624,
      "num_input_tokens_seen": 5217960,
      "step": 8020
    },
    {
      "epoch": 4.7317216981132075,
      "grad_norm": 1.0249063968658447,
      "learning_rate": 9.44263296646431e-06,
      "loss": 0.4757,
      "num_input_tokens_seen": 5221576,
      "step": 8025
    },
    {
      "epoch": 4.734669811320755,
      "grad_norm": 2.260812520980835,
      "learning_rate": 9.441451952546835e-06,
      "loss": 0.5286,
      "num_input_tokens_seen": 5225160,
      "step": 8030
    },
    {
      "epoch": 4.737617924528302,
      "grad_norm": 2.1413991451263428,
      "learning_rate": 9.440269762735814e-06,
      "loss": 0.456,
      "num_input_tokens_seen": 5228072,
      "step": 8035
    },
    {
      "epoch": 4.740566037735849,
      "grad_norm": 2.1210973262786865,
      "learning_rate": 9.439086397344236e-06,
      "loss": 0.4229,
      "num_input_tokens_seen": 5231112,
      "step": 8040
    },
    {
      "epoch": 4.743514150943396,
      "grad_norm": 0.9826104640960693,
      "learning_rate": 9.437901856685404e-06,
      "loss": 0.3379,
      "num_input_tokens_seen": 5234440,
      "step": 8045
    },
    {
      "epoch": 4.746462264150943,
      "grad_norm": 1.7591389417648315,
      "learning_rate": 9.436716141072925e-06,
      "loss": 0.3924,
      "num_input_tokens_seen": 5237960,
      "step": 8050
    },
    {
      "epoch": 4.74941037735849,
      "grad_norm": 2.459364414215088,
      "learning_rate": 9.435529250820732e-06,
      "loss": 0.3906,
      "num_input_tokens_seen": 5240616,
      "step": 8055
    },
    {
      "epoch": 4.752358490566038,
      "grad_norm": 1.8316712379455566,
      "learning_rate": 9.43434118624305e-06,
      "loss": 0.3462,
      "num_input_tokens_seen": 5243784,
      "step": 8060
    },
    {
      "epoch": 4.755306603773585,
      "grad_norm": 1.7045954465866089,
      "learning_rate": 9.433151947654428e-06,
      "loss": 0.4659,
      "num_input_tokens_seen": 5247016,
      "step": 8065
    },
    {
      "epoch": 4.758254716981132,
      "grad_norm": 1.8809270858764648,
      "learning_rate": 9.431961535369724e-06,
      "loss": 0.6033,
      "num_input_tokens_seen": 5250504,
      "step": 8070
    },
    {
      "epoch": 4.7612028301886795,
      "grad_norm": 1.4781222343444824,
      "learning_rate": 9.430769949704103e-06,
      "loss": 0.5665,
      "num_input_tokens_seen": 5254056,
      "step": 8075
    },
    {
      "epoch": 4.764150943396227,
      "grad_norm": 2.366100311279297,
      "learning_rate": 9.42957719097304e-06,
      "loss": 0.328,
      "num_input_tokens_seen": 5257896,
      "step": 8080
    },
    {
      "epoch": 4.767099056603773,
      "grad_norm": 1.9611051082611084,
      "learning_rate": 9.42838325949233e-06,
      "loss": 0.4381,
      "num_input_tokens_seen": 5261544,
      "step": 8085
    },
    {
      "epoch": 4.7700471698113205,
      "grad_norm": 1.8819278478622437,
      "learning_rate": 9.427188155578062e-06,
      "loss": 0.5191,
      "num_input_tokens_seen": 5265352,
      "step": 8090
    },
    {
      "epoch": 4.772995283018868,
      "grad_norm": 3.05875563621521,
      "learning_rate": 9.42599187954665e-06,
      "loss": 0.5181,
      "num_input_tokens_seen": 5268648,
      "step": 8095
    },
    {
      "epoch": 4.775943396226415,
      "grad_norm": 2.150609254837036,
      "learning_rate": 9.424794431714814e-06,
      "loss": 0.5463,
      "num_input_tokens_seen": 5272200,
      "step": 8100
    },
    {
      "epoch": 4.778891509433962,
      "grad_norm": 2.5011281967163086,
      "learning_rate": 9.423595812399581e-06,
      "loss": 0.5231,
      "num_input_tokens_seen": 5275048,
      "step": 8105
    },
    {
      "epoch": 4.78183962264151,
      "grad_norm": 1.171621322631836,
      "learning_rate": 9.422396021918296e-06,
      "loss": 0.5194,
      "num_input_tokens_seen": 5278600,
      "step": 8110
    },
    {
      "epoch": 4.784787735849057,
      "grad_norm": 2.025660991668701,
      "learning_rate": 9.421195060588602e-06,
      "loss": 0.5531,
      "num_input_tokens_seen": 5282056,
      "step": 8115
    },
    {
      "epoch": 4.787735849056604,
      "grad_norm": 1.4467459917068481,
      "learning_rate": 9.419992928728461e-06,
      "loss": 0.3538,
      "num_input_tokens_seen": 5284808,
      "step": 8120
    },
    {
      "epoch": 4.790683962264151,
      "grad_norm": 2.263584852218628,
      "learning_rate": 9.418789626656144e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 5287880,
      "step": 8125
    },
    {
      "epoch": 4.793632075471698,
      "grad_norm": 1.2460007667541504,
      "learning_rate": 9.417585154690229e-06,
      "loss": 0.3562,
      "num_input_tokens_seen": 5290728,
      "step": 8130
    },
    {
      "epoch": 4.796580188679245,
      "grad_norm": 1.7463536262512207,
      "learning_rate": 9.416379513149605e-06,
      "loss": 0.3286,
      "num_input_tokens_seen": 5293640,
      "step": 8135
    },
    {
      "epoch": 4.7995283018867925,
      "grad_norm": 1.94441556930542,
      "learning_rate": 9.415172702353471e-06,
      "loss": 0.4564,
      "num_input_tokens_seen": 5297768,
      "step": 8140
    },
    {
      "epoch": 4.80247641509434,
      "grad_norm": 2.4404234886169434,
      "learning_rate": 9.413964722621339e-06,
      "loss": 0.4186,
      "num_input_tokens_seen": 5300680,
      "step": 8145
    },
    {
      "epoch": 4.805424528301887,
      "grad_norm": 1.5452872514724731,
      "learning_rate": 9.41275557427302e-06,
      "loss": 0.5172,
      "num_input_tokens_seen": 5303880,
      "step": 8150
    },
    {
      "epoch": 4.808372641509434,
      "grad_norm": 2.145986795425415,
      "learning_rate": 9.411545257628646e-06,
      "loss": 0.4725,
      "num_input_tokens_seen": 5307048,
      "step": 8155
    },
    {
      "epoch": 4.811320754716981,
      "grad_norm": 2.5573008060455322,
      "learning_rate": 9.410333773008653e-06,
      "loss": 0.3675,
      "num_input_tokens_seen": 5310088,
      "step": 8160
    },
    {
      "epoch": 4.814268867924528,
      "grad_norm": 2.309384822845459,
      "learning_rate": 9.409121120733784e-06,
      "loss": 0.4108,
      "num_input_tokens_seen": 5313320,
      "step": 8165
    },
    {
      "epoch": 4.817216981132075,
      "grad_norm": 1.8320255279541016,
      "learning_rate": 9.4079073011251e-06,
      "loss": 0.4226,
      "num_input_tokens_seen": 5317480,
      "step": 8170
    },
    {
      "epoch": 4.820165094339623,
      "grad_norm": 2.182582139968872,
      "learning_rate": 9.406692314503956e-06,
      "loss": 0.5109,
      "num_input_tokens_seen": 5320168,
      "step": 8175
    },
    {
      "epoch": 4.82311320754717,
      "grad_norm": 1.430658221244812,
      "learning_rate": 9.405476161192033e-06,
      "loss": 0.3242,
      "num_input_tokens_seen": 5326472,
      "step": 8180
    },
    {
      "epoch": 4.826061320754717,
      "grad_norm": 1.8482496738433838,
      "learning_rate": 9.40425884151131e-06,
      "loss": 0.4166,
      "num_input_tokens_seen": 5330728,
      "step": 8185
    },
    {
      "epoch": 4.8290094339622645,
      "grad_norm": 5.08083438873291,
      "learning_rate": 9.403040355784076e-06,
      "loss": 0.5636,
      "num_input_tokens_seen": 5333448,
      "step": 8190
    },
    {
      "epoch": 4.831957547169811,
      "grad_norm": 1.3805896043777466,
      "learning_rate": 9.401820704332932e-06,
      "loss": 0.4205,
      "num_input_tokens_seen": 5336360,
      "step": 8195
    },
    {
      "epoch": 4.834905660377358,
      "grad_norm": 1.5043219327926636,
      "learning_rate": 9.400599887480786e-06,
      "loss": 0.4176,
      "num_input_tokens_seen": 5339560,
      "step": 8200
    },
    {
      "epoch": 4.837853773584905,
      "grad_norm": 3.3727777004241943,
      "learning_rate": 9.399377905550854e-06,
      "loss": 0.6162,
      "num_input_tokens_seen": 5342344,
      "step": 8205
    },
    {
      "epoch": 4.840801886792453,
      "grad_norm": 1.4693987369537354,
      "learning_rate": 9.398154758866662e-06,
      "loss": 0.3588,
      "num_input_tokens_seen": 5346120,
      "step": 8210
    },
    {
      "epoch": 4.84375,
      "grad_norm": 2.595595121383667,
      "learning_rate": 9.396930447752041e-06,
      "loss": 0.4534,
      "num_input_tokens_seen": 5352616,
      "step": 8215
    },
    {
      "epoch": 4.846698113207547,
      "grad_norm": 1.3976528644561768,
      "learning_rate": 9.395704972531137e-06,
      "loss": 0.689,
      "num_input_tokens_seen": 5355464,
      "step": 8220
    },
    {
      "epoch": 4.849646226415095,
      "grad_norm": 1.3760422468185425,
      "learning_rate": 9.394478333528396e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 5358504,
      "step": 8225
    },
    {
      "epoch": 4.852594339622642,
      "grad_norm": 2.2249667644500732,
      "learning_rate": 9.393250531068576e-06,
      "loss": 0.364,
      "num_input_tokens_seen": 5360648,
      "step": 8230
    },
    {
      "epoch": 4.855542452830189,
      "grad_norm": 2.19929575920105,
      "learning_rate": 9.392021565476744e-06,
      "loss": 0.3551,
      "num_input_tokens_seen": 5366024,
      "step": 8235
    },
    {
      "epoch": 4.8584905660377355,
      "grad_norm": 1.030867099761963,
      "learning_rate": 9.390791437078274e-06,
      "loss": 0.3379,
      "num_input_tokens_seen": 5369352,
      "step": 8240
    },
    {
      "epoch": 4.861438679245283,
      "grad_norm": 1.9071762561798096,
      "learning_rate": 9.38956014619885e-06,
      "loss": 0.4944,
      "num_input_tokens_seen": 5372104,
      "step": 8245
    },
    {
      "epoch": 4.86438679245283,
      "grad_norm": 2.5750532150268555,
      "learning_rate": 9.388327693164456e-06,
      "loss": 0.6616,
      "num_input_tokens_seen": 5374728,
      "step": 8250
    },
    {
      "epoch": 4.867334905660377,
      "grad_norm": 4.7101826667785645,
      "learning_rate": 9.387094078301395e-06,
      "loss": 0.4208,
      "num_input_tokens_seen": 5380328,
      "step": 8255
    },
    {
      "epoch": 4.870283018867925,
      "grad_norm": 2.9712626934051514,
      "learning_rate": 9.385859301936269e-06,
      "loss": 0.3802,
      "num_input_tokens_seen": 5384488,
      "step": 8260
    },
    {
      "epoch": 4.873231132075472,
      "grad_norm": 3.576486349105835,
      "learning_rate": 9.38462336439599e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 5386888,
      "step": 8265
    },
    {
      "epoch": 4.876179245283019,
      "grad_norm": 4.8146443367004395,
      "learning_rate": 9.383386266007779e-06,
      "loss": 0.5905,
      "num_input_tokens_seen": 5389576,
      "step": 8270
    },
    {
      "epoch": 4.879127358490566,
      "grad_norm": 1.5878015756607056,
      "learning_rate": 9.382148007099164e-06,
      "loss": 0.4105,
      "num_input_tokens_seen": 5392744,
      "step": 8275
    },
    {
      "epoch": 4.882075471698113,
      "grad_norm": 1.122544288635254,
      "learning_rate": 9.380908587997977e-06,
      "loss": 0.3934,
      "num_input_tokens_seen": 5396328,
      "step": 8280
    },
    {
      "epoch": 4.88502358490566,
      "grad_norm": 1.813158392906189,
      "learning_rate": 9.37966800903236e-06,
      "loss": 0.3989,
      "num_input_tokens_seen": 5399496,
      "step": 8285
    },
    {
      "epoch": 4.8879716981132075,
      "grad_norm": 1.7239893674850464,
      "learning_rate": 9.378426270530762e-06,
      "loss": 0.4764,
      "num_input_tokens_seen": 5402312,
      "step": 8290
    },
    {
      "epoch": 4.890919811320755,
      "grad_norm": 3.5756137371063232,
      "learning_rate": 9.37718337282194e-06,
      "loss": 0.4632,
      "num_input_tokens_seen": 5405928,
      "step": 8295
    },
    {
      "epoch": 4.893867924528302,
      "grad_norm": 1.8245495557785034,
      "learning_rate": 9.375939316234956e-06,
      "loss": 0.4719,
      "num_input_tokens_seen": 5409608,
      "step": 8300
    },
    {
      "epoch": 4.896816037735849,
      "grad_norm": 1.3866806030273438,
      "learning_rate": 9.374694101099178e-06,
      "loss": 0.4291,
      "num_input_tokens_seen": 5412744,
      "step": 8305
    },
    {
      "epoch": 4.899764150943396,
      "grad_norm": 2.032897710800171,
      "learning_rate": 9.373447727744282e-06,
      "loss": 0.4423,
      "num_input_tokens_seen": 5417512,
      "step": 8310
    },
    {
      "epoch": 4.902712264150943,
      "grad_norm": 1.8219990730285645,
      "learning_rate": 9.372200196500253e-06,
      "loss": 0.4771,
      "num_input_tokens_seen": 5420648,
      "step": 8315
    },
    {
      "epoch": 4.90566037735849,
      "grad_norm": 1.8803461790084839,
      "learning_rate": 9.37095150769738e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 5423016,
      "step": 8320
    },
    {
      "epoch": 4.908608490566038,
      "grad_norm": 3.38843035697937,
      "learning_rate": 9.369701661666255e-06,
      "loss": 0.4001,
      "num_input_tokens_seen": 5425544,
      "step": 8325
    },
    {
      "epoch": 4.911556603773585,
      "grad_norm": 1.718065857887268,
      "learning_rate": 9.368450658737782e-06,
      "loss": 0.4381,
      "num_input_tokens_seen": 5429032,
      "step": 8330
    },
    {
      "epoch": 4.914504716981132,
      "grad_norm": 1.6822131872177124,
      "learning_rate": 9.367198499243173e-06,
      "loss": 0.5225,
      "num_input_tokens_seen": 5432040,
      "step": 8335
    },
    {
      "epoch": 4.9174528301886795,
      "grad_norm": 1.7496016025543213,
      "learning_rate": 9.365945183513938e-06,
      "loss": 0.4484,
      "num_input_tokens_seen": 5435240,
      "step": 8340
    },
    {
      "epoch": 4.920400943396227,
      "grad_norm": 1.6040019989013672,
      "learning_rate": 9.3646907118819e-06,
      "loss": 0.474,
      "num_input_tokens_seen": 5438952,
      "step": 8345
    },
    {
      "epoch": 4.923349056603773,
      "grad_norm": 2.2996151447296143,
      "learning_rate": 9.363435084679185e-06,
      "loss": 0.4824,
      "num_input_tokens_seen": 5441672,
      "step": 8350
    },
    {
      "epoch": 4.9262971698113205,
      "grad_norm": 2.250028610229492,
      "learning_rate": 9.362178302238227e-06,
      "loss": 0.4053,
      "num_input_tokens_seen": 5444488,
      "step": 8355
    },
    {
      "epoch": 4.929245283018868,
      "grad_norm": 3.3839643001556396,
      "learning_rate": 9.360920364891762e-06,
      "loss": 0.4813,
      "num_input_tokens_seen": 5447400,
      "step": 8360
    },
    {
      "epoch": 4.932193396226415,
      "grad_norm": 2.048722743988037,
      "learning_rate": 9.359661272972836e-06,
      "loss": 0.398,
      "num_input_tokens_seen": 5451112,
      "step": 8365
    },
    {
      "epoch": 4.935141509433962,
      "grad_norm": 1.8574259281158447,
      "learning_rate": 9.3584010268148e-06,
      "loss": 0.4676,
      "num_input_tokens_seen": 5454376,
      "step": 8370
    },
    {
      "epoch": 4.93808962264151,
      "grad_norm": 1.2148648500442505,
      "learning_rate": 9.357139626751308e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 5457896,
      "step": 8375
    },
    {
      "epoch": 4.941037735849057,
      "grad_norm": 1.3089103698730469,
      "learning_rate": 9.355877073116321e-06,
      "loss": 0.4814,
      "num_input_tokens_seen": 5460584,
      "step": 8380
    },
    {
      "epoch": 4.943985849056604,
      "grad_norm": 2.2912652492523193,
      "learning_rate": 9.354613366244108e-06,
      "loss": 0.4842,
      "num_input_tokens_seen": 5463368,
      "step": 8385
    },
    {
      "epoch": 4.946933962264151,
      "grad_norm": 2.569997549057007,
      "learning_rate": 9.353348506469236e-06,
      "loss": 0.6057,
      "num_input_tokens_seen": 5467720,
      "step": 8390
    },
    {
      "epoch": 4.949882075471698,
      "grad_norm": 2.156742811203003,
      "learning_rate": 9.352082494126586e-06,
      "loss": 0.4386,
      "num_input_tokens_seen": 5470792,
      "step": 8395
    },
    {
      "epoch": 4.952830188679245,
      "grad_norm": 3.5158495903015137,
      "learning_rate": 9.350815329551341e-06,
      "loss": 0.4343,
      "num_input_tokens_seen": 5474504,
      "step": 8400
    },
    {
      "epoch": 4.9557783018867925,
      "grad_norm": 2.7224245071411133,
      "learning_rate": 9.349547013078986e-06,
      "loss": 0.4271,
      "num_input_tokens_seen": 5478536,
      "step": 8405
    },
    {
      "epoch": 4.95872641509434,
      "grad_norm": 4.086310386657715,
      "learning_rate": 9.348277545045312e-06,
      "loss": 0.5504,
      "num_input_tokens_seen": 5480936,
      "step": 8410
    },
    {
      "epoch": 4.961674528301887,
      "grad_norm": 1.789497971534729,
      "learning_rate": 9.347006925786418e-06,
      "loss": 0.5185,
      "num_input_tokens_seen": 5484200,
      "step": 8415
    },
    {
      "epoch": 4.964622641509434,
      "grad_norm": 4.406622409820557,
      "learning_rate": 9.34573515563871e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 5487496,
      "step": 8420
    },
    {
      "epoch": 4.967570754716981,
      "grad_norm": 2.5554046630859375,
      "learning_rate": 9.344462234938885e-06,
      "loss": 0.3243,
      "num_input_tokens_seen": 5489864,
      "step": 8425
    },
    {
      "epoch": 4.970518867924528,
      "grad_norm": 1.692103624343872,
      "learning_rate": 9.343188164023962e-06,
      "loss": 0.5222,
      "num_input_tokens_seen": 5493704,
      "step": 8430
    },
    {
      "epoch": 4.973466981132075,
      "grad_norm": 2.1611404418945312,
      "learning_rate": 9.341912943231256e-06,
      "loss": 0.4547,
      "num_input_tokens_seen": 5499560,
      "step": 8435
    },
    {
      "epoch": 4.976415094339623,
      "grad_norm": 3.4097931385040283,
      "learning_rate": 9.340636572898383e-06,
      "loss": 0.648,
      "num_input_tokens_seen": 5501992,
      "step": 8440
    },
    {
      "epoch": 4.97936320754717,
      "grad_norm": 1.1731622219085693,
      "learning_rate": 9.339359053363272e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 5505352,
      "step": 8445
    },
    {
      "epoch": 4.982311320754717,
      "grad_norm": 2.038130044937134,
      "learning_rate": 9.338080384964148e-06,
      "loss": 0.5358,
      "num_input_tokens_seen": 5508232,
      "step": 8450
    },
    {
      "epoch": 4.9852594339622645,
      "grad_norm": 1.9956356287002563,
      "learning_rate": 9.336800568039548e-06,
      "loss": 0.4299,
      "num_input_tokens_seen": 5512872,
      "step": 8455
    },
    {
      "epoch": 4.988207547169811,
      "grad_norm": 1.1257824897766113,
      "learning_rate": 9.335519602928307e-06,
      "loss": 0.5236,
      "num_input_tokens_seen": 5516584,
      "step": 8460
    },
    {
      "epoch": 4.991155660377358,
      "grad_norm": 1.1254146099090576,
      "learning_rate": 9.334237489969565e-06,
      "loss": 0.4382,
      "num_input_tokens_seen": 5520392,
      "step": 8465
    },
    {
      "epoch": 4.994103773584905,
      "grad_norm": 1.9410003423690796,
      "learning_rate": 9.332954229502768e-06,
      "loss": 0.4476,
      "num_input_tokens_seen": 5523528,
      "step": 8470
    },
    {
      "epoch": 4.997051886792453,
      "grad_norm": 1.9509012699127197,
      "learning_rate": 9.331669821867665e-06,
      "loss": 0.3976,
      "num_input_tokens_seen": 5526216,
      "step": 8475
    },
    {
      "epoch": 5.0,
      "grad_norm": 4.298738956451416,
      "learning_rate": 9.33038426740431e-06,
      "loss": 0.4917,
      "num_input_tokens_seen": 5528680,
      "step": 8480
    },
    {
      "epoch": 5.002948113207547,
      "grad_norm": 2.41084885597229,
      "learning_rate": 9.329097566453055e-06,
      "loss": 0.4527,
      "num_input_tokens_seen": 5531464,
      "step": 8485
    },
    {
      "epoch": 5.005896226415095,
      "grad_norm": 0.9695172905921936,
      "learning_rate": 9.327809719354564e-06,
      "loss": 0.5777,
      "num_input_tokens_seen": 5534216,
      "step": 8490
    },
    {
      "epoch": 5.008844339622642,
      "grad_norm": 2.997445821762085,
      "learning_rate": 9.326520726449795e-06,
      "loss": 0.6475,
      "num_input_tokens_seen": 5537224,
      "step": 8495
    },
    {
      "epoch": 5.011792452830188,
      "grad_norm": 1.2498568296432495,
      "learning_rate": 9.32523058808002e-06,
      "loss": 0.4129,
      "num_input_tokens_seen": 5540904,
      "step": 8500
    },
    {
      "epoch": 5.0147405660377355,
      "grad_norm": 2.137699604034424,
      "learning_rate": 9.323939304586806e-06,
      "loss": 0.5609,
      "num_input_tokens_seen": 5544328,
      "step": 8505
    },
    {
      "epoch": 5.017688679245283,
      "grad_norm": 1.3748759031295776,
      "learning_rate": 9.322646876312025e-06,
      "loss": 0.3596,
      "num_input_tokens_seen": 5550984,
      "step": 8510
    },
    {
      "epoch": 5.02063679245283,
      "grad_norm": 2.105956792831421,
      "learning_rate": 9.321353303597854e-06,
      "loss": 0.375,
      "num_input_tokens_seen": 5555240,
      "step": 8515
    },
    {
      "epoch": 5.023584905660377,
      "grad_norm": 2.305466890335083,
      "learning_rate": 9.320058586786771e-06,
      "loss": 0.4131,
      "num_input_tokens_seen": 5559208,
      "step": 8520
    },
    {
      "epoch": 5.026533018867925,
      "grad_norm": 1.2828630208969116,
      "learning_rate": 9.31876272622156e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 5562824,
      "step": 8525
    },
    {
      "epoch": 5.029481132075472,
      "grad_norm": 1.771759271621704,
      "learning_rate": 9.317465722245305e-06,
      "loss": 0.4407,
      "num_input_tokens_seen": 5566216,
      "step": 8530
    },
    {
      "epoch": 5.032429245283019,
      "grad_norm": 0.84317547082901,
      "learning_rate": 9.316167575201391e-06,
      "loss": 0.3771,
      "num_input_tokens_seen": 5570696,
      "step": 8535
    },
    {
      "epoch": 5.035377358490566,
      "grad_norm": 1.3762197494506836,
      "learning_rate": 9.31486828543351e-06,
      "loss": 0.5153,
      "num_input_tokens_seen": 5574248,
      "step": 8540
    },
    {
      "epoch": 5.038325471698113,
      "grad_norm": 2.493542432785034,
      "learning_rate": 9.313567853285656e-06,
      "loss": 0.3103,
      "num_input_tokens_seen": 5577256,
      "step": 8545
    },
    {
      "epoch": 5.04127358490566,
      "grad_norm": 1.4198964834213257,
      "learning_rate": 9.31226627910212e-06,
      "loss": 0.4426,
      "num_input_tokens_seen": 5580872,
      "step": 8550
    },
    {
      "epoch": 5.0442216981132075,
      "grad_norm": 1.9185653924942017,
      "learning_rate": 9.310963563227504e-06,
      "loss": 0.4514,
      "num_input_tokens_seen": 5585256,
      "step": 8555
    },
    {
      "epoch": 5.047169811320755,
      "grad_norm": 1.3696937561035156,
      "learning_rate": 9.309659706006704e-06,
      "loss": 0.4937,
      "num_input_tokens_seen": 5587592,
      "step": 8560
    },
    {
      "epoch": 5.050117924528302,
      "grad_norm": 2.1512253284454346,
      "learning_rate": 9.308354707784925e-06,
      "loss": 0.3876,
      "num_input_tokens_seen": 5591464,
      "step": 8565
    },
    {
      "epoch": 5.053066037735849,
      "grad_norm": 4.915706634521484,
      "learning_rate": 9.307048568907669e-06,
      "loss": 0.4688,
      "num_input_tokens_seen": 5594344,
      "step": 8570
    },
    {
      "epoch": 5.056014150943396,
      "grad_norm": 2.2504372596740723,
      "learning_rate": 9.30574128972074e-06,
      "loss": 0.42,
      "num_input_tokens_seen": 5597192,
      "step": 8575
    },
    {
      "epoch": 5.058962264150943,
      "grad_norm": 1.4182058572769165,
      "learning_rate": 9.304432870570247e-06,
      "loss": 0.4025,
      "num_input_tokens_seen": 5599784,
      "step": 8580
    },
    {
      "epoch": 5.06191037735849,
      "grad_norm": 1.2327818870544434,
      "learning_rate": 9.303123311802605e-06,
      "loss": 0.4048,
      "num_input_tokens_seen": 5602792,
      "step": 8585
    },
    {
      "epoch": 5.064858490566038,
      "grad_norm": 3.1173484325408936,
      "learning_rate": 9.301812613764516e-06,
      "loss": 0.4445,
      "num_input_tokens_seen": 5605896,
      "step": 8590
    },
    {
      "epoch": 5.067806603773585,
      "grad_norm": 2.4708333015441895,
      "learning_rate": 9.300500776803001e-06,
      "loss": 0.4476,
      "num_input_tokens_seen": 5608712,
      "step": 8595
    },
    {
      "epoch": 5.070754716981132,
      "grad_norm": 1.5063966512680054,
      "learning_rate": 9.29918780126537e-06,
      "loss": 0.4131,
      "num_input_tokens_seen": 5611432,
      "step": 8600
    },
    {
      "epoch": 5.0737028301886795,
      "grad_norm": 1.5420846939086914,
      "learning_rate": 9.297873687499239e-06,
      "loss": 0.4427,
      "num_input_tokens_seen": 5614440,
      "step": 8605
    },
    {
      "epoch": 5.076650943396227,
      "grad_norm": 1.980177879333496,
      "learning_rate": 9.296558435852528e-06,
      "loss": 0.3489,
      "num_input_tokens_seen": 5617032,
      "step": 8610
    },
    {
      "epoch": 5.079599056603773,
      "grad_norm": 2.1566474437713623,
      "learning_rate": 9.295242046673454e-06,
      "loss": 0.4595,
      "num_input_tokens_seen": 5620072,
      "step": 8615
    },
    {
      "epoch": 5.0825471698113205,
      "grad_norm": 1.7096978425979614,
      "learning_rate": 9.293924520310535e-06,
      "loss": 0.501,
      "num_input_tokens_seen": 5623304,
      "step": 8620
    },
    {
      "epoch": 5.085495283018868,
      "grad_norm": 1.129740834236145,
      "learning_rate": 9.292605857112595e-06,
      "loss": 0.4741,
      "num_input_tokens_seen": 5627080,
      "step": 8625
    },
    {
      "epoch": 5.088443396226415,
      "grad_norm": 3.421769380569458,
      "learning_rate": 9.291286057428755e-06,
      "loss": 0.4497,
      "num_input_tokens_seen": 5630568,
      "step": 8630
    },
    {
      "epoch": 5.091391509433962,
      "grad_norm": 1.4722706079483032,
      "learning_rate": 9.289965121608436e-06,
      "loss": 0.2771,
      "num_input_tokens_seen": 5633864,
      "step": 8635
    },
    {
      "epoch": 5.09433962264151,
      "grad_norm": 1.4053847789764404,
      "learning_rate": 9.288643050001362e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 5636552,
      "step": 8640
    },
    {
      "epoch": 5.097287735849057,
      "grad_norm": 1.5366727113723755,
      "learning_rate": 9.287319842957557e-06,
      "loss": 0.5116,
      "num_input_tokens_seen": 5639304,
      "step": 8645
    },
    {
      "epoch": 5.100235849056604,
      "grad_norm": 2.2252590656280518,
      "learning_rate": 9.285995500827348e-06,
      "loss": 0.4196,
      "num_input_tokens_seen": 5641960,
      "step": 8650
    },
    {
      "epoch": 5.103183962264151,
      "grad_norm": 4.010032653808594,
      "learning_rate": 9.284670023961355e-06,
      "loss": 0.5389,
      "num_input_tokens_seen": 5645928,
      "step": 8655
    },
    {
      "epoch": 5.106132075471698,
      "grad_norm": 2.2784535884857178,
      "learning_rate": 9.28334341271051e-06,
      "loss": 0.5125,
      "num_input_tokens_seen": 5649160,
      "step": 8660
    },
    {
      "epoch": 5.109080188679245,
      "grad_norm": 3.3265559673309326,
      "learning_rate": 9.282015667426036e-06,
      "loss": 0.4747,
      "num_input_tokens_seen": 5651816,
      "step": 8665
    },
    {
      "epoch": 5.1120283018867925,
      "grad_norm": 1.9093157052993774,
      "learning_rate": 9.280686788459461e-06,
      "loss": 0.5281,
      "num_input_tokens_seen": 5656328,
      "step": 8670
    },
    {
      "epoch": 5.11497641509434,
      "grad_norm": 0.9107136726379395,
      "learning_rate": 9.279356776162606e-06,
      "loss": 0.4091,
      "num_input_tokens_seen": 5660008,
      "step": 8675
    },
    {
      "epoch": 5.117924528301887,
      "grad_norm": 1.8482780456542969,
      "learning_rate": 9.278025630887607e-06,
      "loss": 0.4998,
      "num_input_tokens_seen": 5662696,
      "step": 8680
    },
    {
      "epoch": 5.120872641509434,
      "grad_norm": 3.0636982917785645,
      "learning_rate": 9.27669335298688e-06,
      "loss": 0.3865,
      "num_input_tokens_seen": 5666216,
      "step": 8685
    },
    {
      "epoch": 5.123820754716981,
      "grad_norm": 2.0669620037078857,
      "learning_rate": 9.275359942813158e-06,
      "loss": 0.4275,
      "num_input_tokens_seen": 5669128,
      "step": 8690
    },
    {
      "epoch": 5.126768867924528,
      "grad_norm": 1.4080373048782349,
      "learning_rate": 9.274025400719466e-06,
      "loss": 0.4193,
      "num_input_tokens_seen": 5672168,
      "step": 8695
    },
    {
      "epoch": 5.129716981132075,
      "grad_norm": 4.376392364501953,
      "learning_rate": 9.27268972705913e-06,
      "loss": 0.4861,
      "num_input_tokens_seen": 5675528,
      "step": 8700
    },
    {
      "epoch": 5.132665094339623,
      "grad_norm": 1.2970664501190186,
      "learning_rate": 9.271352922185772e-06,
      "loss": 0.5187,
      "num_input_tokens_seen": 5680072,
      "step": 8705
    },
    {
      "epoch": 5.13561320754717,
      "grad_norm": 1.3828120231628418,
      "learning_rate": 9.270014986453321e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 5683688,
      "step": 8710
    },
    {
      "epoch": 5.138561320754717,
      "grad_norm": 2.581124782562256,
      "learning_rate": 9.268675920215999e-06,
      "loss": 0.5463,
      "num_input_tokens_seen": 5686600,
      "step": 8715
    },
    {
      "epoch": 5.1415094339622645,
      "grad_norm": 1.3863767385482788,
      "learning_rate": 9.26733572382833e-06,
      "loss": 0.4909,
      "num_input_tokens_seen": 5690376,
      "step": 8720
    },
    {
      "epoch": 5.144457547169812,
      "grad_norm": 2.3836750984191895,
      "learning_rate": 9.265994397645137e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 5692840,
      "step": 8725
    },
    {
      "epoch": 5.147405660377358,
      "grad_norm": 2.159472942352295,
      "learning_rate": 9.264651942021543e-06,
      "loss": 0.5319,
      "num_input_tokens_seen": 5697032,
      "step": 8730
    },
    {
      "epoch": 5.150353773584905,
      "grad_norm": 2.1057000160217285,
      "learning_rate": 9.263308357312966e-06,
      "loss": 0.4488,
      "num_input_tokens_seen": 5700552,
      "step": 8735
    },
    {
      "epoch": 5.153301886792453,
      "grad_norm": 1.8502203226089478,
      "learning_rate": 9.26196364387513e-06,
      "loss": 0.4837,
      "num_input_tokens_seen": 5705288,
      "step": 8740
    },
    {
      "epoch": 5.15625,
      "grad_norm": 1.8362536430358887,
      "learning_rate": 9.26061780206405e-06,
      "loss": 0.4218,
      "num_input_tokens_seen": 5708008,
      "step": 8745
    },
    {
      "epoch": 5.159198113207547,
      "grad_norm": 2.5595719814300537,
      "learning_rate": 9.259270832236043e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 5711240,
      "step": 8750
    },
    {
      "epoch": 5.162146226415095,
      "grad_norm": 2.4376637935638428,
      "learning_rate": 9.257922734747729e-06,
      "loss": 0.481,
      "num_input_tokens_seen": 5714312,
      "step": 8755
    },
    {
      "epoch": 5.165094339622642,
      "grad_norm": 1.6480398178100586,
      "learning_rate": 9.256573509956018e-06,
      "loss": 0.418,
      "num_input_tokens_seen": 5717320,
      "step": 8760
    },
    {
      "epoch": 5.168042452830188,
      "grad_norm": 1.7073054313659668,
      "learning_rate": 9.255223158218127e-06,
      "loss": 0.425,
      "num_input_tokens_seen": 5722088,
      "step": 8765
    },
    {
      "epoch": 5.1709905660377355,
      "grad_norm": 1.4120420217514038,
      "learning_rate": 9.253871679891566e-06,
      "loss": 0.3329,
      "num_input_tokens_seen": 5725256,
      "step": 8770
    },
    {
      "epoch": 5.173938679245283,
      "grad_norm": 2.0053324699401855,
      "learning_rate": 9.252519075334143e-06,
      "loss": 0.4566,
      "num_input_tokens_seen": 5728744,
      "step": 8775
    },
    {
      "epoch": 5.17688679245283,
      "grad_norm": 2.7283477783203125,
      "learning_rate": 9.251165344903969e-06,
      "loss": 0.4003,
      "num_input_tokens_seen": 5732168,
      "step": 8780
    },
    {
      "epoch": 5.179834905660377,
      "grad_norm": 1.87785804271698,
      "learning_rate": 9.249810488959448e-06,
      "loss": 0.5422,
      "num_input_tokens_seen": 5735752,
      "step": 8785
    },
    {
      "epoch": 5.182783018867925,
      "grad_norm": 1.3296831846237183,
      "learning_rate": 9.248454507859285e-06,
      "loss": 0.5083,
      "num_input_tokens_seen": 5738856,
      "step": 8790
    },
    {
      "epoch": 5.185731132075472,
      "grad_norm": 1.4376671314239502,
      "learning_rate": 9.247097401962482e-06,
      "loss": 0.4668,
      "num_input_tokens_seen": 5742696,
      "step": 8795
    },
    {
      "epoch": 5.188679245283019,
      "grad_norm": 2.4449105262756348,
      "learning_rate": 9.245739171628335e-06,
      "loss": 0.4625,
      "num_input_tokens_seen": 5745640,
      "step": 8800
    },
    {
      "epoch": 5.191627358490566,
      "grad_norm": 2.370718002319336,
      "learning_rate": 9.244379817216447e-06,
      "loss": 0.4479,
      "num_input_tokens_seen": 5749000,
      "step": 8805
    },
    {
      "epoch": 5.194575471698113,
      "grad_norm": 1.3816624879837036,
      "learning_rate": 9.243019339086708e-06,
      "loss": 0.3377,
      "num_input_tokens_seen": 5752808,
      "step": 8810
    },
    {
      "epoch": 5.19752358490566,
      "grad_norm": 2.0664796829223633,
      "learning_rate": 9.241657737599313e-06,
      "loss": 0.4219,
      "num_input_tokens_seen": 5756296,
      "step": 8815
    },
    {
      "epoch": 5.2004716981132075,
      "grad_norm": 2.264397144317627,
      "learning_rate": 9.240295013114752e-06,
      "loss": 0.4986,
      "num_input_tokens_seen": 5758984,
      "step": 8820
    },
    {
      "epoch": 5.203419811320755,
      "grad_norm": 2.9029133319854736,
      "learning_rate": 9.238931165993811e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 5762472,
      "step": 8825
    },
    {
      "epoch": 5.206367924528302,
      "grad_norm": 3.6382994651794434,
      "learning_rate": 9.237566196597577e-06,
      "loss": 0.5471,
      "num_input_tokens_seen": 5767592,
      "step": 8830
    },
    {
      "epoch": 5.209316037735849,
      "grad_norm": 2.148664951324463,
      "learning_rate": 9.236200105287427e-06,
      "loss": 0.5388,
      "num_input_tokens_seen": 5770792,
      "step": 8835
    },
    {
      "epoch": 5.212264150943396,
      "grad_norm": 2.5313024520874023,
      "learning_rate": 9.234832892425042e-06,
      "loss": 0.5583,
      "num_input_tokens_seen": 5773512,
      "step": 8840
    },
    {
      "epoch": 5.215212264150943,
      "grad_norm": 2.3997085094451904,
      "learning_rate": 9.2334645583724e-06,
      "loss": 0.5692,
      "num_input_tokens_seen": 5776456,
      "step": 8845
    },
    {
      "epoch": 5.21816037735849,
      "grad_norm": 1.5615992546081543,
      "learning_rate": 9.23209510349177e-06,
      "loss": 0.4607,
      "num_input_tokens_seen": 5779976,
      "step": 8850
    },
    {
      "epoch": 5.221108490566038,
      "grad_norm": 1.5125142335891724,
      "learning_rate": 9.230724528145722e-06,
      "loss": 0.3879,
      "num_input_tokens_seen": 5782376,
      "step": 8855
    },
    {
      "epoch": 5.224056603773585,
      "grad_norm": 1.617682695388794,
      "learning_rate": 9.229352832697122e-06,
      "loss": 0.3876,
      "num_input_tokens_seen": 5785704,
      "step": 8860
    },
    {
      "epoch": 5.227004716981132,
      "grad_norm": 1.7704075574874878,
      "learning_rate": 9.22798001750913e-06,
      "loss": 0.5026,
      "num_input_tokens_seen": 5789736,
      "step": 8865
    },
    {
      "epoch": 5.2299528301886795,
      "grad_norm": 2.8150506019592285,
      "learning_rate": 9.226606082945209e-06,
      "loss": 0.2945,
      "num_input_tokens_seen": 5792168,
      "step": 8870
    },
    {
      "epoch": 5.232900943396227,
      "grad_norm": 2.2224793434143066,
      "learning_rate": 9.225231029369112e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 5795112,
      "step": 8875
    },
    {
      "epoch": 5.235849056603773,
      "grad_norm": 1.9538326263427734,
      "learning_rate": 9.22385485714489e-06,
      "loss": 0.5022,
      "num_input_tokens_seen": 5798216,
      "step": 8880
    },
    {
      "epoch": 5.2387971698113205,
      "grad_norm": 3.0247950553894043,
      "learning_rate": 9.222477566636889e-06,
      "loss": 0.4593,
      "num_input_tokens_seen": 5801448,
      "step": 8885
    },
    {
      "epoch": 5.241745283018868,
      "grad_norm": 2.8815228939056396,
      "learning_rate": 9.221099158209757e-06,
      "loss": 0.5805,
      "num_input_tokens_seen": 5804072,
      "step": 8890
    },
    {
      "epoch": 5.244693396226415,
      "grad_norm": 2.2217464447021484,
      "learning_rate": 9.219719632228429e-06,
      "loss": 0.4423,
      "num_input_tokens_seen": 5808200,
      "step": 8895
    },
    {
      "epoch": 5.247641509433962,
      "grad_norm": 2.813058853149414,
      "learning_rate": 9.218338989058141e-06,
      "loss": 0.3634,
      "num_input_tokens_seen": 5810376,
      "step": 8900
    },
    {
      "epoch": 5.25058962264151,
      "grad_norm": 1.556547999382019,
      "learning_rate": 9.21695722906443e-06,
      "loss": 0.5097,
      "num_input_tokens_seen": 5813448,
      "step": 8905
    },
    {
      "epoch": 5.253537735849057,
      "grad_norm": 1.4735699892044067,
      "learning_rate": 9.215574352613115e-06,
      "loss": 0.5123,
      "num_input_tokens_seen": 5816584,
      "step": 8910
    },
    {
      "epoch": 5.256485849056604,
      "grad_norm": 7.339044570922852,
      "learning_rate": 9.214190360070323e-06,
      "loss": 0.4735,
      "num_input_tokens_seen": 5820424,
      "step": 8915
    },
    {
      "epoch": 5.259433962264151,
      "grad_norm": 1.8211698532104492,
      "learning_rate": 9.212805251802471e-06,
      "loss": 0.3986,
      "num_input_tokens_seen": 5823432,
      "step": 8920
    },
    {
      "epoch": 5.262382075471698,
      "grad_norm": 1.4225010871887207,
      "learning_rate": 9.211419028176273e-06,
      "loss": 0.4713,
      "num_input_tokens_seen": 5826568,
      "step": 8925
    },
    {
      "epoch": 5.265330188679245,
      "grad_norm": 2.2613701820373535,
      "learning_rate": 9.210031689558738e-06,
      "loss": 0.2991,
      "num_input_tokens_seen": 5829256,
      "step": 8930
    },
    {
      "epoch": 5.2682783018867925,
      "grad_norm": 1.9945213794708252,
      "learning_rate": 9.208643236317166e-06,
      "loss": 0.3871,
      "num_input_tokens_seen": 5833160,
      "step": 8935
    },
    {
      "epoch": 5.27122641509434,
      "grad_norm": 1.8355199098587036,
      "learning_rate": 9.20725366881916e-06,
      "loss": 0.4402,
      "num_input_tokens_seen": 5838632,
      "step": 8940
    },
    {
      "epoch": 5.274174528301887,
      "grad_norm": 1.651877999305725,
      "learning_rate": 9.205862987432614e-06,
      "loss": 0.3734,
      "num_input_tokens_seen": 5841416,
      "step": 8945
    },
    {
      "epoch": 5.277122641509434,
      "grad_norm": 2.063458204269409,
      "learning_rate": 9.204471192525715e-06,
      "loss": 0.4547,
      "num_input_tokens_seen": 5844296,
      "step": 8950
    },
    {
      "epoch": 5.280070754716981,
      "grad_norm": 2.0544967651367188,
      "learning_rate": 9.203078284466949e-06,
      "loss": 0.5152,
      "num_input_tokens_seen": 5847944,
      "step": 8955
    },
    {
      "epoch": 5.283018867924528,
      "grad_norm": 1.8793351650238037,
      "learning_rate": 9.201684263625091e-06,
      "loss": 0.4842,
      "num_input_tokens_seen": 5850312,
      "step": 8960
    },
    {
      "epoch": 5.285966981132075,
      "grad_norm": 1.6601569652557373,
      "learning_rate": 9.200289130369218e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 5853256,
      "step": 8965
    },
    {
      "epoch": 5.288915094339623,
      "grad_norm": 1.887472152709961,
      "learning_rate": 9.198892885068693e-06,
      "loss": 0.5796,
      "num_input_tokens_seen": 5856360,
      "step": 8970
    },
    {
      "epoch": 5.29186320754717,
      "grad_norm": 2.218308687210083,
      "learning_rate": 9.197495528093182e-06,
      "loss": 0.4164,
      "num_input_tokens_seen": 5859304,
      "step": 8975
    },
    {
      "epoch": 5.294811320754717,
      "grad_norm": 1.4595645666122437,
      "learning_rate": 9.196097059812639e-06,
      "loss": 0.456,
      "num_input_tokens_seen": 5862760,
      "step": 8980
    },
    {
      "epoch": 5.2977594339622645,
      "grad_norm": 1.2827430963516235,
      "learning_rate": 9.194697480597316e-06,
      "loss": 0.4149,
      "num_input_tokens_seen": 5865928,
      "step": 8985
    },
    {
      "epoch": 5.300707547169811,
      "grad_norm": 1.9678654670715332,
      "learning_rate": 9.193296790817755e-06,
      "loss": 0.4285,
      "num_input_tokens_seen": 5868296,
      "step": 8990
    },
    {
      "epoch": 5.303655660377358,
      "grad_norm": 1.339626669883728,
      "learning_rate": 9.1918949908448e-06,
      "loss": 0.3426,
      "num_input_tokens_seen": 5871976,
      "step": 8995
    },
    {
      "epoch": 5.306603773584905,
      "grad_norm": 1.176430106163025,
      "learning_rate": 9.190492081049578e-06,
      "loss": 0.3078,
      "num_input_tokens_seen": 5875880,
      "step": 9000
    },
    {
      "epoch": 5.309551886792453,
      "grad_norm": 1.0606156587600708,
      "learning_rate": 9.189088061803517e-06,
      "loss": 0.3029,
      "num_input_tokens_seen": 5878696,
      "step": 9005
    },
    {
      "epoch": 5.3125,
      "grad_norm": 2.1841821670532227,
      "learning_rate": 9.187682933478337e-06,
      "loss": 0.5204,
      "num_input_tokens_seen": 5881640,
      "step": 9010
    },
    {
      "epoch": 5.315448113207547,
      "grad_norm": 2.3424289226531982,
      "learning_rate": 9.186276696446054e-06,
      "loss": 0.4113,
      "num_input_tokens_seen": 5884264,
      "step": 9015
    },
    {
      "epoch": 5.318396226415095,
      "grad_norm": 1.3566397428512573,
      "learning_rate": 9.184869351078974e-06,
      "loss": 0.4451,
      "num_input_tokens_seen": 5888072,
      "step": 9020
    },
    {
      "epoch": 5.321344339622642,
      "grad_norm": 1.379604697227478,
      "learning_rate": 9.183460897749697e-06,
      "loss": 0.3669,
      "num_input_tokens_seen": 5891688,
      "step": 9025
    },
    {
      "epoch": 5.324292452830189,
      "grad_norm": 1.468673586845398,
      "learning_rate": 9.182051336831117e-06,
      "loss": 0.398,
      "num_input_tokens_seen": 5894280,
      "step": 9030
    },
    {
      "epoch": 5.3272405660377355,
      "grad_norm": 2.0117573738098145,
      "learning_rate": 9.180640668696424e-06,
      "loss": 0.3878,
      "num_input_tokens_seen": 5897672,
      "step": 9035
    },
    {
      "epoch": 5.330188679245283,
      "grad_norm": 2.4404358863830566,
      "learning_rate": 9.179228893719094e-06,
      "loss": 0.5118,
      "num_input_tokens_seen": 5900328,
      "step": 9040
    },
    {
      "epoch": 5.33313679245283,
      "grad_norm": 1.687803864479065,
      "learning_rate": 9.177816012272904e-06,
      "loss": 0.4041,
      "num_input_tokens_seen": 5903016,
      "step": 9045
    },
    {
      "epoch": 5.336084905660377,
      "grad_norm": 1.6854829788208008,
      "learning_rate": 9.17640202473192e-06,
      "loss": 0.589,
      "num_input_tokens_seen": 5905864,
      "step": 9050
    },
    {
      "epoch": 5.339033018867925,
      "grad_norm": 3.7648508548736572,
      "learning_rate": 9.1749869314705e-06,
      "loss": 0.4051,
      "num_input_tokens_seen": 5908904,
      "step": 9055
    },
    {
      "epoch": 5.341981132075472,
      "grad_norm": 3.0982625484466553,
      "learning_rate": 9.173570732863295e-06,
      "loss": 0.5738,
      "num_input_tokens_seen": 5912552,
      "step": 9060
    },
    {
      "epoch": 5.344929245283019,
      "grad_norm": 2.485663414001465,
      "learning_rate": 9.172153429285254e-06,
      "loss": 0.5138,
      "num_input_tokens_seen": 5914888,
      "step": 9065
    },
    {
      "epoch": 5.347877358490566,
      "grad_norm": 2.581028938293457,
      "learning_rate": 9.17073502111161e-06,
      "loss": 0.5569,
      "num_input_tokens_seen": 5917480,
      "step": 9070
    },
    {
      "epoch": 5.350825471698113,
      "grad_norm": 2.175549268722534,
      "learning_rate": 9.169315508717895e-06,
      "loss": 0.431,
      "num_input_tokens_seen": 5919944,
      "step": 9075
    },
    {
      "epoch": 5.35377358490566,
      "grad_norm": 2.520141124725342,
      "learning_rate": 9.167894892479932e-06,
      "loss": 0.4494,
      "num_input_tokens_seen": 5923240,
      "step": 9080
    },
    {
      "epoch": 5.3567216981132075,
      "grad_norm": 1.7953218221664429,
      "learning_rate": 9.16647317277383e-06,
      "loss": 0.4871,
      "num_input_tokens_seen": 5926344,
      "step": 9085
    },
    {
      "epoch": 5.359669811320755,
      "grad_norm": 3.2753944396972656,
      "learning_rate": 9.165050349976002e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 5929224,
      "step": 9090
    },
    {
      "epoch": 5.362617924528302,
      "grad_norm": 1.472706913948059,
      "learning_rate": 9.16362642446314e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 5933096,
      "step": 9095
    },
    {
      "epoch": 5.365566037735849,
      "grad_norm": 1.5615272521972656,
      "learning_rate": 9.162201396612242e-06,
      "loss": 0.449,
      "num_input_tokens_seen": 5936712,
      "step": 9100
    },
    {
      "epoch": 5.368514150943396,
      "grad_norm": 1.9115259647369385,
      "learning_rate": 9.160775266800583e-06,
      "loss": 0.6339,
      "num_input_tokens_seen": 5939368,
      "step": 9105
    },
    {
      "epoch": 5.371462264150943,
      "grad_norm": 1.7299535274505615,
      "learning_rate": 9.159348035405742e-06,
      "loss": 0.4358,
      "num_input_tokens_seen": 5942408,
      "step": 9110
    },
    {
      "epoch": 5.37441037735849,
      "grad_norm": 2.7527048587799072,
      "learning_rate": 9.157919702805582e-06,
      "loss": 0.4323,
      "num_input_tokens_seen": 5945480,
      "step": 9115
    },
    {
      "epoch": 5.377358490566038,
      "grad_norm": 1.146026611328125,
      "learning_rate": 9.156490269378262e-06,
      "loss": 0.3451,
      "num_input_tokens_seen": 5948872,
      "step": 9120
    },
    {
      "epoch": 5.380306603773585,
      "grad_norm": 2.4105098247528076,
      "learning_rate": 9.15505973550223e-06,
      "loss": 0.5244,
      "num_input_tokens_seen": 5952328,
      "step": 9125
    },
    {
      "epoch": 5.383254716981132,
      "grad_norm": 1.4499263763427734,
      "learning_rate": 9.153628101556223e-06,
      "loss": 0.4124,
      "num_input_tokens_seen": 5955720,
      "step": 9130
    },
    {
      "epoch": 5.3862028301886795,
      "grad_norm": 1.788260817527771,
      "learning_rate": 9.152195367919277e-06,
      "loss": 0.4264,
      "num_input_tokens_seen": 5958376,
      "step": 9135
    },
    {
      "epoch": 5.389150943396227,
      "grad_norm": 2.6777048110961914,
      "learning_rate": 9.150761534970713e-06,
      "loss": 0.3861,
      "num_input_tokens_seen": 5961096,
      "step": 9140
    },
    {
      "epoch": 5.392099056603773,
      "grad_norm": 2.493377208709717,
      "learning_rate": 9.149326603090144e-06,
      "loss": 0.4391,
      "num_input_tokens_seen": 5963944,
      "step": 9145
    },
    {
      "epoch": 5.3950471698113205,
      "grad_norm": 4.22123908996582,
      "learning_rate": 9.147890572657471e-06,
      "loss": 0.4189,
      "num_input_tokens_seen": 5966696,
      "step": 9150
    },
    {
      "epoch": 5.397995283018868,
      "grad_norm": 2.9517319202423096,
      "learning_rate": 9.146453444052895e-06,
      "loss": 0.6397,
      "num_input_tokens_seen": 5969864,
      "step": 9155
    },
    {
      "epoch": 5.400943396226415,
      "grad_norm": 1.664089322090149,
      "learning_rate": 9.145015217656899e-06,
      "loss": 0.5828,
      "num_input_tokens_seen": 5972840,
      "step": 9160
    },
    {
      "epoch": 5.403891509433962,
      "grad_norm": 1.492478370666504,
      "learning_rate": 9.14357589385026e-06,
      "loss": 0.5123,
      "num_input_tokens_seen": 5975464,
      "step": 9165
    },
    {
      "epoch": 5.40683962264151,
      "grad_norm": 1.6935193538665771,
      "learning_rate": 9.142135473014046e-06,
      "loss": 0.4051,
      "num_input_tokens_seen": 5978792,
      "step": 9170
    },
    {
      "epoch": 5.409787735849057,
      "grad_norm": 2.231369733810425,
      "learning_rate": 9.140693955529614e-06,
      "loss": 0.3788,
      "num_input_tokens_seen": 5981672,
      "step": 9175
    },
    {
      "epoch": 5.412735849056604,
      "grad_norm": 1.6467370986938477,
      "learning_rate": 9.13925134177861e-06,
      "loss": 0.2851,
      "num_input_tokens_seen": 5987752,
      "step": 9180
    },
    {
      "epoch": 5.415683962264151,
      "grad_norm": 1.6663157939910889,
      "learning_rate": 9.137807632142977e-06,
      "loss": 0.3544,
      "num_input_tokens_seen": 5990408,
      "step": 9185
    },
    {
      "epoch": 5.418632075471698,
      "grad_norm": 1.2874102592468262,
      "learning_rate": 9.136362827004937e-06,
      "loss": 0.4482,
      "num_input_tokens_seen": 5993704,
      "step": 9190
    },
    {
      "epoch": 5.421580188679245,
      "grad_norm": 2.3429033756256104,
      "learning_rate": 9.134916926747015e-06,
      "loss": 0.3738,
      "num_input_tokens_seen": 5997416,
      "step": 9195
    },
    {
      "epoch": 5.4245283018867925,
      "grad_norm": 2.629138708114624,
      "learning_rate": 9.133469931752016e-06,
      "loss": 0.5928,
      "num_input_tokens_seen": 6000584,
      "step": 9200
    },
    {
      "epoch": 5.42747641509434,
      "grad_norm": 1.18839430809021,
      "learning_rate": 9.132021842403035e-06,
      "loss": 0.3036,
      "num_input_tokens_seen": 6003080,
      "step": 9205
    },
    {
      "epoch": 5.430424528301887,
      "grad_norm": 3.4050393104553223,
      "learning_rate": 9.130572659083465e-06,
      "loss": 0.4712,
      "num_input_tokens_seen": 6006472,
      "step": 9210
    },
    {
      "epoch": 5.433372641509434,
      "grad_norm": 3.0528812408447266,
      "learning_rate": 9.129122382176982e-06,
      "loss": 0.4492,
      "num_input_tokens_seen": 6010440,
      "step": 9215
    },
    {
      "epoch": 5.436320754716981,
      "grad_norm": 1.8928977251052856,
      "learning_rate": 9.127671012067554e-06,
      "loss": 0.4778,
      "num_input_tokens_seen": 6013256,
      "step": 9220
    },
    {
      "epoch": 5.439268867924528,
      "grad_norm": 1.5489469766616821,
      "learning_rate": 9.126218549139434e-06,
      "loss": 0.4007,
      "num_input_tokens_seen": 6017192,
      "step": 9225
    },
    {
      "epoch": 5.442216981132075,
      "grad_norm": 1.6457371711730957,
      "learning_rate": 9.124764993777171e-06,
      "loss": 0.3953,
      "num_input_tokens_seen": 6020264,
      "step": 9230
    },
    {
      "epoch": 5.445165094339623,
      "grad_norm": 2.0792150497436523,
      "learning_rate": 9.1233103463656e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 6023752,
      "step": 9235
    },
    {
      "epoch": 5.44811320754717,
      "grad_norm": 3.946282148361206,
      "learning_rate": 9.121854607289842e-06,
      "loss": 0.3912,
      "num_input_tokens_seen": 6028328,
      "step": 9240
    },
    {
      "epoch": 5.451061320754717,
      "grad_norm": 2.694544792175293,
      "learning_rate": 9.120397776935314e-06,
      "loss": 0.5873,
      "num_input_tokens_seen": 6031272,
      "step": 9245
    },
    {
      "epoch": 5.4540094339622645,
      "grad_norm": 5.984225749969482,
      "learning_rate": 9.118939855687717e-06,
      "loss": 0.3539,
      "num_input_tokens_seen": 6034472,
      "step": 9250
    },
    {
      "epoch": 5.456957547169811,
      "grad_norm": 2.3550405502319336,
      "learning_rate": 9.117480843933043e-06,
      "loss": 0.4471,
      "num_input_tokens_seen": 6038760,
      "step": 9255
    },
    {
      "epoch": 5.459905660377358,
      "grad_norm": 2.8562934398651123,
      "learning_rate": 9.116020742057567e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 6042312,
      "step": 9260
    },
    {
      "epoch": 5.462853773584905,
      "grad_norm": 2.145179271697998,
      "learning_rate": 9.114559550447863e-06,
      "loss": 0.5925,
      "num_input_tokens_seen": 6044616,
      "step": 9265
    },
    {
      "epoch": 5.465801886792453,
      "grad_norm": 2.3533828258514404,
      "learning_rate": 9.113097269490784e-06,
      "loss": 0.5181,
      "num_input_tokens_seen": 6048424,
      "step": 9270
    },
    {
      "epoch": 5.46875,
      "grad_norm": 1.1036087274551392,
      "learning_rate": 9.111633899573476e-06,
      "loss": 0.4099,
      "num_input_tokens_seen": 6051016,
      "step": 9275
    },
    {
      "epoch": 5.471698113207547,
      "grad_norm": 2.8446006774902344,
      "learning_rate": 9.110169441083374e-06,
      "loss": 0.5384,
      "num_input_tokens_seen": 6054280,
      "step": 9280
    },
    {
      "epoch": 5.474646226415095,
      "grad_norm": 3.961167812347412,
      "learning_rate": 9.108703894408198e-06,
      "loss": 0.3662,
      "num_input_tokens_seen": 6057128,
      "step": 9285
    },
    {
      "epoch": 5.477594339622642,
      "grad_norm": 3.038400650024414,
      "learning_rate": 9.107237259935959e-06,
      "loss": 0.4008,
      "num_input_tokens_seen": 6060360,
      "step": 9290
    },
    {
      "epoch": 5.480542452830189,
      "grad_norm": 3.3277804851531982,
      "learning_rate": 9.105769538054954e-06,
      "loss": 0.4012,
      "num_input_tokens_seen": 6063976,
      "step": 9295
    },
    {
      "epoch": 5.4834905660377355,
      "grad_norm": 1.2873364686965942,
      "learning_rate": 9.104300729153768e-06,
      "loss": 0.4109,
      "num_input_tokens_seen": 6068488,
      "step": 9300
    },
    {
      "epoch": 5.486438679245283,
      "grad_norm": 1.5659147500991821,
      "learning_rate": 9.102830833621277e-06,
      "loss": 0.4521,
      "num_input_tokens_seen": 6071944,
      "step": 9305
    },
    {
      "epoch": 5.48938679245283,
      "grad_norm": 1.8612449169158936,
      "learning_rate": 9.101359851846639e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 6075144,
      "step": 9310
    },
    {
      "epoch": 5.492334905660377,
      "grad_norm": 3.327059745788574,
      "learning_rate": 9.099887784219305e-06,
      "loss": 0.5017,
      "num_input_tokens_seen": 6078952,
      "step": 9315
    },
    {
      "epoch": 5.495283018867925,
      "grad_norm": 1.9221395254135132,
      "learning_rate": 9.098414631129012e-06,
      "loss": 0.3857,
      "num_input_tokens_seen": 6081736,
      "step": 9320
    },
    {
      "epoch": 5.498231132075472,
      "grad_norm": 1.7748950719833374,
      "learning_rate": 9.09694039296578e-06,
      "loss": 0.3306,
      "num_input_tokens_seen": 6085128,
      "step": 9325
    },
    {
      "epoch": 5.501179245283019,
      "grad_norm": 2.848676919937134,
      "learning_rate": 9.095465070119924e-06,
      "loss": 0.5741,
      "num_input_tokens_seen": 6088680,
      "step": 9330
    },
    {
      "epoch": 5.504127358490566,
      "grad_norm": 2.822033166885376,
      "learning_rate": 9.09398866298204e-06,
      "loss": 0.6311,
      "num_input_tokens_seen": 6091880,
      "step": 9335
    },
    {
      "epoch": 5.507075471698113,
      "grad_norm": 1.184868574142456,
      "learning_rate": 9.09251117194301e-06,
      "loss": 0.424,
      "num_input_tokens_seen": 6095752,
      "step": 9340
    },
    {
      "epoch": 5.51002358490566,
      "grad_norm": 1.5121214389801025,
      "learning_rate": 9.091032597394012e-06,
      "loss": 0.4407,
      "num_input_tokens_seen": 6099176,
      "step": 9345
    },
    {
      "epoch": 5.5129716981132075,
      "grad_norm": 2.7798593044281006,
      "learning_rate": 9.089552939726503e-06,
      "loss": 0.4979,
      "num_input_tokens_seen": 6102152,
      "step": 9350
    },
    {
      "epoch": 5.515919811320755,
      "grad_norm": 1.714670181274414,
      "learning_rate": 9.088072199332227e-06,
      "loss": 0.5228,
      "num_input_tokens_seen": 6105256,
      "step": 9355
    },
    {
      "epoch": 5.518867924528302,
      "grad_norm": 2.2994697093963623,
      "learning_rate": 9.08659037660322e-06,
      "loss": 0.4303,
      "num_input_tokens_seen": 6108488,
      "step": 9360
    },
    {
      "epoch": 5.521816037735849,
      "grad_norm": 1.9496381282806396,
      "learning_rate": 9.085107471931797e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 6113256,
      "step": 9365
    },
    {
      "epoch": 5.524764150943396,
      "grad_norm": 2.571505546569824,
      "learning_rate": 9.083623485710564e-06,
      "loss": 0.4151,
      "num_input_tokens_seen": 6115848,
      "step": 9370
    },
    {
      "epoch": 5.527712264150943,
      "grad_norm": 1.8417683839797974,
      "learning_rate": 9.082138418332416e-06,
      "loss": 0.4596,
      "num_input_tokens_seen": 6118216,
      "step": 9375
    },
    {
      "epoch": 5.53066037735849,
      "grad_norm": 2.3361246585845947,
      "learning_rate": 9.080652270190527e-06,
      "loss": 0.3994,
      "num_input_tokens_seen": 6120520,
      "step": 9380
    },
    {
      "epoch": 5.533608490566038,
      "grad_norm": 2.540174961090088,
      "learning_rate": 9.079165041678363e-06,
      "loss": 0.3377,
      "num_input_tokens_seen": 6123880,
      "step": 9385
    },
    {
      "epoch": 5.536556603773585,
      "grad_norm": 1.8295546770095825,
      "learning_rate": 9.077676733189675e-06,
      "loss": 0.4339,
      "num_input_tokens_seen": 6127528,
      "step": 9390
    },
    {
      "epoch": 5.539504716981132,
      "grad_norm": 1.9093514680862427,
      "learning_rate": 9.076187345118496e-06,
      "loss": 0.3373,
      "num_input_tokens_seen": 6129864,
      "step": 9395
    },
    {
      "epoch": 5.5424528301886795,
      "grad_norm": 2.539487361907959,
      "learning_rate": 9.074696877859152e-06,
      "loss": 0.5797,
      "num_input_tokens_seen": 6133704,
      "step": 9400
    },
    {
      "epoch": 5.545400943396227,
      "grad_norm": 0.9788433909416199,
      "learning_rate": 9.073205331806248e-06,
      "loss": 0.3739,
      "num_input_tokens_seen": 6136968,
      "step": 9405
    },
    {
      "epoch": 5.548349056603773,
      "grad_norm": 3.93035626411438,
      "learning_rate": 9.071712707354676e-06,
      "loss": 0.3247,
      "num_input_tokens_seen": 6139880,
      "step": 9410
    },
    {
      "epoch": 5.5512971698113205,
      "grad_norm": 1.3705909252166748,
      "learning_rate": 9.070219004899618e-06,
      "loss": 0.5753,
      "num_input_tokens_seen": 6143112,
      "step": 9415
    },
    {
      "epoch": 5.554245283018868,
      "grad_norm": 3.2309653759002686,
      "learning_rate": 9.068724224836538e-06,
      "loss": 0.3399,
      "num_input_tokens_seen": 6146728,
      "step": 9420
    },
    {
      "epoch": 5.557193396226415,
      "grad_norm": 3.318100690841675,
      "learning_rate": 9.067228367561182e-06,
      "loss": 0.3981,
      "num_input_tokens_seen": 6149192,
      "step": 9425
    },
    {
      "epoch": 5.560141509433962,
      "grad_norm": 2.2322585582733154,
      "learning_rate": 9.06573143346959e-06,
      "loss": 0.5887,
      "num_input_tokens_seen": 6155560,
      "step": 9430
    },
    {
      "epoch": 5.56308962264151,
      "grad_norm": 2.0942013263702393,
      "learning_rate": 9.064233422958078e-06,
      "loss": 0.4602,
      "num_input_tokens_seen": 6159080,
      "step": 9435
    },
    {
      "epoch": 5.566037735849057,
      "grad_norm": 2.2779343128204346,
      "learning_rate": 9.062734336423248e-06,
      "loss": 0.4605,
      "num_input_tokens_seen": 6162632,
      "step": 9440
    },
    {
      "epoch": 5.568985849056604,
      "grad_norm": 3.158295154571533,
      "learning_rate": 9.061234174261998e-06,
      "loss": 0.7057,
      "num_input_tokens_seen": 6165928,
      "step": 9445
    },
    {
      "epoch": 5.571933962264151,
      "grad_norm": 2.1458961963653564,
      "learning_rate": 9.059732936871493e-06,
      "loss": 0.459,
      "num_input_tokens_seen": 6168680,
      "step": 9450
    },
    {
      "epoch": 5.574882075471698,
      "grad_norm": 1.3610783815383911,
      "learning_rate": 9.058230624649198e-06,
      "loss": 0.5412,
      "num_input_tokens_seen": 6171720,
      "step": 9455
    },
    {
      "epoch": 5.577830188679245,
      "grad_norm": 1.3831926584243774,
      "learning_rate": 9.056727237992856e-06,
      "loss": 0.357,
      "num_input_tokens_seen": 6175688,
      "step": 9460
    },
    {
      "epoch": 5.5807783018867925,
      "grad_norm": 2.4988064765930176,
      "learning_rate": 9.055222777300493e-06,
      "loss": 0.4673,
      "num_input_tokens_seen": 6178984,
      "step": 9465
    },
    {
      "epoch": 5.58372641509434,
      "grad_norm": 1.4757087230682373,
      "learning_rate": 9.053717242970423e-06,
      "loss": 0.4402,
      "num_input_tokens_seen": 6182184,
      "step": 9470
    },
    {
      "epoch": 5.586674528301887,
      "grad_norm": 1.7486714124679565,
      "learning_rate": 9.052210635401244e-06,
      "loss": 0.7418,
      "num_input_tokens_seen": 6184776,
      "step": 9475
    },
    {
      "epoch": 5.589622641509434,
      "grad_norm": 2.719128131866455,
      "learning_rate": 9.050702954991833e-06,
      "loss": 0.4825,
      "num_input_tokens_seen": 6187528,
      "step": 9480
    },
    {
      "epoch": 5.592570754716981,
      "grad_norm": 2.127504348754883,
      "learning_rate": 9.049194202141358e-06,
      "loss": 0.448,
      "num_input_tokens_seen": 6190472,
      "step": 9485
    },
    {
      "epoch": 5.595518867924528,
      "grad_norm": 2.7241947650909424,
      "learning_rate": 9.047684377249267e-06,
      "loss": 0.453,
      "num_input_tokens_seen": 6193192,
      "step": 9490
    },
    {
      "epoch": 5.598466981132075,
      "grad_norm": 1.6014137268066406,
      "learning_rate": 9.046173480715292e-06,
      "loss": 0.5519,
      "num_input_tokens_seen": 6196200,
      "step": 9495
    },
    {
      "epoch": 5.601415094339623,
      "grad_norm": 1.880471110343933,
      "learning_rate": 9.044661512939451e-06,
      "loss": 0.3322,
      "num_input_tokens_seen": 6198856,
      "step": 9500
    },
    {
      "epoch": 5.60436320754717,
      "grad_norm": 1.6292113065719604,
      "learning_rate": 9.043148474322043e-06,
      "loss": 0.6234,
      "num_input_tokens_seen": 6202024,
      "step": 9505
    },
    {
      "epoch": 5.607311320754717,
      "grad_norm": 1.6045300960540771,
      "learning_rate": 9.04163436526365e-06,
      "loss": 0.5384,
      "num_input_tokens_seen": 6205352,
      "step": 9510
    },
    {
      "epoch": 5.6102594339622645,
      "grad_norm": 1.8734464645385742,
      "learning_rate": 9.040119186165142e-06,
      "loss": 0.4535,
      "num_input_tokens_seen": 6207848,
      "step": 9515
    },
    {
      "epoch": 5.613207547169811,
      "grad_norm": 1.3760719299316406,
      "learning_rate": 9.038602937427665e-06,
      "loss": 0.3904,
      "num_input_tokens_seen": 6210920,
      "step": 9520
    },
    {
      "epoch": 5.616155660377358,
      "grad_norm": 1.415008783340454,
      "learning_rate": 9.037085619452658e-06,
      "loss": 0.6985,
      "num_input_tokens_seen": 6215432,
      "step": 9525
    },
    {
      "epoch": 5.619103773584905,
      "grad_norm": 3.0020735263824463,
      "learning_rate": 9.035567232641833e-06,
      "loss": 0.3224,
      "num_input_tokens_seen": 6218760,
      "step": 9530
    },
    {
      "epoch": 5.622051886792453,
      "grad_norm": 1.5325114727020264,
      "learning_rate": 9.03404777739719e-06,
      "loss": 0.4155,
      "num_input_tokens_seen": 6222952,
      "step": 9535
    },
    {
      "epoch": 5.625,
      "grad_norm": 0.9342206120491028,
      "learning_rate": 9.032527254121013e-06,
      "loss": 0.4727,
      "num_input_tokens_seen": 6226376,
      "step": 9540
    },
    {
      "epoch": 5.627948113207547,
      "grad_norm": 0.868590772151947,
      "learning_rate": 9.031005663215867e-06,
      "loss": 0.3267,
      "num_input_tokens_seen": 6229448,
      "step": 9545
    },
    {
      "epoch": 5.630896226415095,
      "grad_norm": 2.573518753051758,
      "learning_rate": 9.029483005084595e-06,
      "loss": 0.4122,
      "num_input_tokens_seen": 6233224,
      "step": 9550
    },
    {
      "epoch": 5.633844339622642,
      "grad_norm": 1.670257806777954,
      "learning_rate": 9.027959280130337e-06,
      "loss": 0.4086,
      "num_input_tokens_seen": 6236392,
      "step": 9555
    },
    {
      "epoch": 5.636792452830189,
      "grad_norm": 1.5887823104858398,
      "learning_rate": 9.026434488756496e-06,
      "loss": 0.469,
      "num_input_tokens_seen": 6239112,
      "step": 9560
    },
    {
      "epoch": 5.6397405660377355,
      "grad_norm": 1.4977518320083618,
      "learning_rate": 9.024908631366774e-06,
      "loss": 0.518,
      "num_input_tokens_seen": 6242568,
      "step": 9565
    },
    {
      "epoch": 5.642688679245283,
      "grad_norm": 2.1194980144500732,
      "learning_rate": 9.023381708365143e-06,
      "loss": 0.406,
      "num_input_tokens_seen": 6245640,
      "step": 9570
    },
    {
      "epoch": 5.64563679245283,
      "grad_norm": 2.9444210529327393,
      "learning_rate": 9.021853720155866e-06,
      "loss": 0.3532,
      "num_input_tokens_seen": 6248744,
      "step": 9575
    },
    {
      "epoch": 5.648584905660377,
      "grad_norm": 2.082366943359375,
      "learning_rate": 9.020324667143483e-06,
      "loss": 0.5186,
      "num_input_tokens_seen": 6251624,
      "step": 9580
    },
    {
      "epoch": 5.651533018867925,
      "grad_norm": 3.273745059967041,
      "learning_rate": 9.018794549732819e-06,
      "loss": 0.367,
      "num_input_tokens_seen": 6254664,
      "step": 9585
    },
    {
      "epoch": 5.654481132075472,
      "grad_norm": 1.9303058385849,
      "learning_rate": 9.017263368328977e-06,
      "loss": 0.4848,
      "num_input_tokens_seen": 6257928,
      "step": 9590
    },
    {
      "epoch": 5.657429245283019,
      "grad_norm": 1.6684821844100952,
      "learning_rate": 9.015731123337344e-06,
      "loss": 0.4894,
      "num_input_tokens_seen": 6261416,
      "step": 9595
    },
    {
      "epoch": 5.660377358490566,
      "grad_norm": 1.4447792768478394,
      "learning_rate": 9.01419781516359e-06,
      "loss": 0.3968,
      "num_input_tokens_seen": 6264520,
      "step": 9600
    },
    {
      "epoch": 5.663325471698113,
      "grad_norm": 2.670275926589966,
      "learning_rate": 9.012663444213664e-06,
      "loss": 0.4933,
      "num_input_tokens_seen": 6267944,
      "step": 9605
    },
    {
      "epoch": 5.66627358490566,
      "grad_norm": 2.5596470832824707,
      "learning_rate": 9.011128010893797e-06,
      "loss": 0.4628,
      "num_input_tokens_seen": 6271208,
      "step": 9610
    },
    {
      "epoch": 5.6692216981132075,
      "grad_norm": 1.652653455734253,
      "learning_rate": 9.009591515610503e-06,
      "loss": 0.4661,
      "num_input_tokens_seen": 6274312,
      "step": 9615
    },
    {
      "epoch": 5.672169811320755,
      "grad_norm": 3.2180609703063965,
      "learning_rate": 9.008053958770575e-06,
      "loss": 0.4437,
      "num_input_tokens_seen": 6277256,
      "step": 9620
    },
    {
      "epoch": 5.675117924528302,
      "grad_norm": 2.7684619426727295,
      "learning_rate": 9.006515340781087e-06,
      "loss": 0.4205,
      "num_input_tokens_seen": 6280712,
      "step": 9625
    },
    {
      "epoch": 5.678066037735849,
      "grad_norm": 1.142978310585022,
      "learning_rate": 9.004975662049396e-06,
      "loss": 0.3217,
      "num_input_tokens_seen": 6283816,
      "step": 9630
    },
    {
      "epoch": 5.681014150943396,
      "grad_norm": 3.529292106628418,
      "learning_rate": 9.003434922983138e-06,
      "loss": 0.4349,
      "num_input_tokens_seen": 6286248,
      "step": 9635
    },
    {
      "epoch": 5.683962264150943,
      "grad_norm": 1.9229927062988281,
      "learning_rate": 9.00189312399023e-06,
      "loss": 0.4269,
      "num_input_tokens_seen": 6289096,
      "step": 9640
    },
    {
      "epoch": 5.68691037735849,
      "grad_norm": 1.792043924331665,
      "learning_rate": 9.00035026547887e-06,
      "loss": 0.3732,
      "num_input_tokens_seen": 6292392,
      "step": 9645
    },
    {
      "epoch": 5.689858490566038,
      "grad_norm": 3.2065787315368652,
      "learning_rate": 8.998806347857537e-06,
      "loss": 0.469,
      "num_input_tokens_seen": 6295304,
      "step": 9650
    },
    {
      "epoch": 5.692806603773585,
      "grad_norm": 1.8897875547409058,
      "learning_rate": 8.99726137153499e-06,
      "loss": 0.4393,
      "num_input_tokens_seen": 6298088,
      "step": 9655
    },
    {
      "epoch": 5.695754716981132,
      "grad_norm": 2.4694504737854004,
      "learning_rate": 8.995715336920266e-06,
      "loss": 0.4891,
      "num_input_tokens_seen": 6301384,
      "step": 9660
    },
    {
      "epoch": 5.6987028301886795,
      "grad_norm": 2.761639356613159,
      "learning_rate": 8.994168244422687e-06,
      "loss": 0.5764,
      "num_input_tokens_seen": 6304392,
      "step": 9665
    },
    {
      "epoch": 5.701650943396227,
      "grad_norm": 1.1657084226608276,
      "learning_rate": 8.992620094451852e-06,
      "loss": 0.378,
      "num_input_tokens_seen": 6308552,
      "step": 9670
    },
    {
      "epoch": 5.704599056603773,
      "grad_norm": 3.3550541400909424,
      "learning_rate": 8.991070887417639e-06,
      "loss": 0.4368,
      "num_input_tokens_seen": 6311368,
      "step": 9675
    },
    {
      "epoch": 5.7075471698113205,
      "grad_norm": 2.946260452270508,
      "learning_rate": 8.989520623730208e-06,
      "loss": 0.4913,
      "num_input_tokens_seen": 6315112,
      "step": 9680
    },
    {
      "epoch": 5.710495283018868,
      "grad_norm": 1.4334410429000854,
      "learning_rate": 8.987969303799998e-06,
      "loss": 0.5032,
      "num_input_tokens_seen": 6319176,
      "step": 9685
    },
    {
      "epoch": 5.713443396226415,
      "grad_norm": 3.05387282371521,
      "learning_rate": 8.986416928037728e-06,
      "loss": 0.4111,
      "num_input_tokens_seen": 6321832,
      "step": 9690
    },
    {
      "epoch": 5.716391509433962,
      "grad_norm": 1.658649206161499,
      "learning_rate": 8.984863496854395e-06,
      "loss": 0.4271,
      "num_input_tokens_seen": 6324648,
      "step": 9695
    },
    {
      "epoch": 5.71933962264151,
      "grad_norm": 2.1617074012756348,
      "learning_rate": 8.983309010661279e-06,
      "loss": 0.4236,
      "num_input_tokens_seen": 6327304,
      "step": 9700
    },
    {
      "epoch": 5.722287735849057,
      "grad_norm": 1.6698461771011353,
      "learning_rate": 8.981753469869934e-06,
      "loss": 0.6688,
      "num_input_tokens_seen": 6330504,
      "step": 9705
    },
    {
      "epoch": 5.725235849056604,
      "grad_norm": 2.268144369125366,
      "learning_rate": 8.980196874892198e-06,
      "loss": 0.4468,
      "num_input_tokens_seen": 6333192,
      "step": 9710
    },
    {
      "epoch": 5.728183962264151,
      "grad_norm": 1.8669620752334595,
      "learning_rate": 8.978639226140184e-06,
      "loss": 0.3965,
      "num_input_tokens_seen": 6336488,
      "step": 9715
    },
    {
      "epoch": 5.731132075471698,
      "grad_norm": 2.112658739089966,
      "learning_rate": 8.977080524026289e-06,
      "loss": 0.3408,
      "num_input_tokens_seen": 6340296,
      "step": 9720
    },
    {
      "epoch": 5.734080188679245,
      "grad_norm": 1.7725245952606201,
      "learning_rate": 8.975520768963186e-06,
      "loss": 0.5933,
      "num_input_tokens_seen": 6342664,
      "step": 9725
    },
    {
      "epoch": 5.7370283018867925,
      "grad_norm": 2.735645055770874,
      "learning_rate": 8.973959961363825e-06,
      "loss": 0.4087,
      "num_input_tokens_seen": 6345384,
      "step": 9730
    },
    {
      "epoch": 5.73997641509434,
      "grad_norm": 2.775350332260132,
      "learning_rate": 8.972398101641438e-06,
      "loss": 0.4258,
      "num_input_tokens_seen": 6348104,
      "step": 9735
    },
    {
      "epoch": 5.742924528301887,
      "grad_norm": 2.370429039001465,
      "learning_rate": 8.970835190209532e-06,
      "loss": 0.4466,
      "num_input_tokens_seen": 6355592,
      "step": 9740
    },
    {
      "epoch": 5.745872641509434,
      "grad_norm": 4.120883464813232,
      "learning_rate": 8.969271227481899e-06,
      "loss": 0.5416,
      "num_input_tokens_seen": 6359144,
      "step": 9745
    },
    {
      "epoch": 5.748820754716981,
      "grad_norm": 2.135887384414673,
      "learning_rate": 8.967706213872599e-06,
      "loss": 0.4149,
      "num_input_tokens_seen": 6362696,
      "step": 9750
    },
    {
      "epoch": 5.751768867924528,
      "grad_norm": 1.5834332704544067,
      "learning_rate": 8.966140149795981e-06,
      "loss": 0.4449,
      "num_input_tokens_seen": 6366120,
      "step": 9755
    },
    {
      "epoch": 5.754716981132075,
      "grad_norm": 1.0197964906692505,
      "learning_rate": 8.964573035666663e-06,
      "loss": 0.3594,
      "num_input_tokens_seen": 6369224,
      "step": 9760
    },
    {
      "epoch": 5.757665094339623,
      "grad_norm": 3.677851676940918,
      "learning_rate": 8.96300487189955e-06,
      "loss": 0.4247,
      "num_input_tokens_seen": 6372328,
      "step": 9765
    },
    {
      "epoch": 5.76061320754717,
      "grad_norm": 2.2238149642944336,
      "learning_rate": 8.961435658909816e-06,
      "loss": 0.5206,
      "num_input_tokens_seen": 6375368,
      "step": 9770
    },
    {
      "epoch": 5.763561320754717,
      "grad_norm": 1.830399990081787,
      "learning_rate": 8.959865397112918e-06,
      "loss": 0.4524,
      "num_input_tokens_seen": 6378440,
      "step": 9775
    },
    {
      "epoch": 5.7665094339622645,
      "grad_norm": 2.8615005016326904,
      "learning_rate": 8.95829408692459e-06,
      "loss": 0.4644,
      "num_input_tokens_seen": 6381096,
      "step": 9780
    },
    {
      "epoch": 5.769457547169811,
      "grad_norm": 2.100790500640869,
      "learning_rate": 8.956721728760845e-06,
      "loss": 0.3863,
      "num_input_tokens_seen": 6384328,
      "step": 9785
    },
    {
      "epoch": 5.772405660377358,
      "grad_norm": 2.152247667312622,
      "learning_rate": 8.95514832303797e-06,
      "loss": 0.5591,
      "num_input_tokens_seen": 6388232,
      "step": 9790
    },
    {
      "epoch": 5.775353773584905,
      "grad_norm": 3.9218504428863525,
      "learning_rate": 8.953573870172528e-06,
      "loss": 0.3714,
      "num_input_tokens_seen": 6390952,
      "step": 9795
    },
    {
      "epoch": 5.778301886792453,
      "grad_norm": 3.4655866622924805,
      "learning_rate": 8.951998370581368e-06,
      "loss": 0.3962,
      "num_input_tokens_seen": 6394472,
      "step": 9800
    },
    {
      "epoch": 5.78125,
      "grad_norm": 1.9657557010650635,
      "learning_rate": 8.950421824681605e-06,
      "loss": 0.5625,
      "num_input_tokens_seen": 6398312,
      "step": 9805
    },
    {
      "epoch": 5.784198113207547,
      "grad_norm": 2.6667251586914062,
      "learning_rate": 8.948844232890638e-06,
      "loss": 0.4149,
      "num_input_tokens_seen": 6401768,
      "step": 9810
    },
    {
      "epoch": 5.787146226415095,
      "grad_norm": 1.686158537864685,
      "learning_rate": 8.947265595626144e-06,
      "loss": 0.5064,
      "num_input_tokens_seen": 6404872,
      "step": 9815
    },
    {
      "epoch": 5.790094339622642,
      "grad_norm": 2.5049242973327637,
      "learning_rate": 8.945685913306071e-06,
      "loss": 0.4599,
      "num_input_tokens_seen": 6408296,
      "step": 9820
    },
    {
      "epoch": 5.793042452830189,
      "grad_norm": 2.334519386291504,
      "learning_rate": 8.944105186348646e-06,
      "loss": 0.3621,
      "num_input_tokens_seen": 6411464,
      "step": 9825
    },
    {
      "epoch": 5.7959905660377355,
      "grad_norm": 1.850448489189148,
      "learning_rate": 8.942523415172377e-06,
      "loss": 0.4174,
      "num_input_tokens_seen": 6414408,
      "step": 9830
    },
    {
      "epoch": 5.798938679245283,
      "grad_norm": 2.590169668197632,
      "learning_rate": 8.94094060019604e-06,
      "loss": 0.3615,
      "num_input_tokens_seen": 6417448,
      "step": 9835
    },
    {
      "epoch": 5.80188679245283,
      "grad_norm": 2.3526647090911865,
      "learning_rate": 8.939356741838696e-06,
      "loss": 0.4851,
      "num_input_tokens_seen": 6420936,
      "step": 9840
    },
    {
      "epoch": 5.804834905660377,
      "grad_norm": 2.5362000465393066,
      "learning_rate": 8.937771840519677e-06,
      "loss": 0.3226,
      "num_input_tokens_seen": 6424264,
      "step": 9845
    },
    {
      "epoch": 5.807783018867925,
      "grad_norm": 1.8015542030334473,
      "learning_rate": 8.936185896658593e-06,
      "loss": 0.411,
      "num_input_tokens_seen": 6426920,
      "step": 9850
    },
    {
      "epoch": 5.810731132075472,
      "grad_norm": 2.584254741668701,
      "learning_rate": 8.934598910675329e-06,
      "loss": 0.3852,
      "num_input_tokens_seen": 6430216,
      "step": 9855
    },
    {
      "epoch": 5.813679245283019,
      "grad_norm": 1.8372210264205933,
      "learning_rate": 8.933010882990044e-06,
      "loss": 0.4039,
      "num_input_tokens_seen": 6434088,
      "step": 9860
    },
    {
      "epoch": 5.816627358490566,
      "grad_norm": 1.3932268619537354,
      "learning_rate": 8.93142181402318e-06,
      "loss": 0.367,
      "num_input_tokens_seen": 6437640,
      "step": 9865
    },
    {
      "epoch": 5.819575471698113,
      "grad_norm": 2.291452646255493,
      "learning_rate": 8.929831704195445e-06,
      "loss": 0.4027,
      "num_input_tokens_seen": 6440744,
      "step": 9870
    },
    {
      "epoch": 5.82252358490566,
      "grad_norm": 2.886775493621826,
      "learning_rate": 8.928240553927831e-06,
      "loss": 0.3891,
      "num_input_tokens_seen": 6443208,
      "step": 9875
    },
    {
      "epoch": 5.8254716981132075,
      "grad_norm": 2.408482551574707,
      "learning_rate": 8.926648363641602e-06,
      "loss": 0.3448,
      "num_input_tokens_seen": 6446312,
      "step": 9880
    },
    {
      "epoch": 5.828419811320755,
      "grad_norm": 2.5916571617126465,
      "learning_rate": 8.925055133758294e-06,
      "loss": 0.5208,
      "num_input_tokens_seen": 6449640,
      "step": 9885
    },
    {
      "epoch": 5.831367924528302,
      "grad_norm": 2.447073221206665,
      "learning_rate": 8.923460864699723e-06,
      "loss": 0.4145,
      "num_input_tokens_seen": 6453384,
      "step": 9890
    },
    {
      "epoch": 5.834316037735849,
      "grad_norm": 1.3931211233139038,
      "learning_rate": 8.921865556887979e-06,
      "loss": 0.5891,
      "num_input_tokens_seen": 6455880,
      "step": 9895
    },
    {
      "epoch": 5.837264150943396,
      "grad_norm": 1.6746543645858765,
      "learning_rate": 8.920269210745426e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 6458984,
      "step": 9900
    },
    {
      "epoch": 5.840212264150943,
      "grad_norm": 1.170090675354004,
      "learning_rate": 8.918671826694704e-06,
      "loss": 0.3977,
      "num_input_tokens_seen": 6462344,
      "step": 9905
    },
    {
      "epoch": 5.84316037735849,
      "grad_norm": 1.2894397974014282,
      "learning_rate": 8.91707340515873e-06,
      "loss": 0.4414,
      "num_input_tokens_seen": 6465288,
      "step": 9910
    },
    {
      "epoch": 5.846108490566038,
      "grad_norm": 1.420104742050171,
      "learning_rate": 8.915473946560688e-06,
      "loss": 0.4247,
      "num_input_tokens_seen": 6470728,
      "step": 9915
    },
    {
      "epoch": 5.849056603773585,
      "grad_norm": 4.945779323577881,
      "learning_rate": 8.913873451324044e-06,
      "loss": 0.5315,
      "num_input_tokens_seen": 6474376,
      "step": 9920
    },
    {
      "epoch": 5.852004716981132,
      "grad_norm": 1.4195010662078857,
      "learning_rate": 8.912271919872538e-06,
      "loss": 0.3625,
      "num_input_tokens_seen": 6477672,
      "step": 9925
    },
    {
      "epoch": 5.8549528301886795,
      "grad_norm": 2.7578988075256348,
      "learning_rate": 8.910669352630176e-06,
      "loss": 0.5733,
      "num_input_tokens_seen": 6480136,
      "step": 9930
    },
    {
      "epoch": 5.857900943396227,
      "grad_norm": 1.6564258337020874,
      "learning_rate": 8.909065750021253e-06,
      "loss": 0.3553,
      "num_input_tokens_seen": 6483240,
      "step": 9935
    },
    {
      "epoch": 5.860849056603773,
      "grad_norm": 3.189434289932251,
      "learning_rate": 8.907461112470323e-06,
      "loss": 0.5216,
      "num_input_tokens_seen": 6486600,
      "step": 9940
    },
    {
      "epoch": 5.8637971698113205,
      "grad_norm": 1.4162263870239258,
      "learning_rate": 8.905855440402225e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 6489128,
      "step": 9945
    },
    {
      "epoch": 5.866745283018868,
      "grad_norm": 2.5725746154785156,
      "learning_rate": 8.904248734242065e-06,
      "loss": 0.5143,
      "num_input_tokens_seen": 6491976,
      "step": 9950
    },
    {
      "epoch": 5.869693396226415,
      "grad_norm": 1.9021679162979126,
      "learning_rate": 8.902640994415226e-06,
      "loss": 0.4387,
      "num_input_tokens_seen": 6495464,
      "step": 9955
    },
    {
      "epoch": 5.872641509433962,
      "grad_norm": 4.451745986938477,
      "learning_rate": 8.901032221347364e-06,
      "loss": 0.4629,
      "num_input_tokens_seen": 6498344,
      "step": 9960
    },
    {
      "epoch": 5.87558962264151,
      "grad_norm": 2.2210371494293213,
      "learning_rate": 8.899422415464409e-06,
      "loss": 0.4315,
      "num_input_tokens_seen": 6501864,
      "step": 9965
    },
    {
      "epoch": 5.878537735849057,
      "grad_norm": 1.5052103996276855,
      "learning_rate": 8.897811577192565e-06,
      "loss": 0.5162,
      "num_input_tokens_seen": 6504776,
      "step": 9970
    },
    {
      "epoch": 5.881485849056604,
      "grad_norm": 1.7460362911224365,
      "learning_rate": 8.896199706958306e-06,
      "loss": 0.3986,
      "num_input_tokens_seen": 6507912,
      "step": 9975
    },
    {
      "epoch": 5.884433962264151,
      "grad_norm": 4.087330341339111,
      "learning_rate": 8.894586805188384e-06,
      "loss": 0.4826,
      "num_input_tokens_seen": 6510632,
      "step": 9980
    },
    {
      "epoch": 5.887382075471698,
      "grad_norm": 1.4754745960235596,
      "learning_rate": 8.892972872309821e-06,
      "loss": 0.3284,
      "num_input_tokens_seen": 6513640,
      "step": 9985
    },
    {
      "epoch": 5.890330188679245,
      "grad_norm": 2.0166385173797607,
      "learning_rate": 8.89135790874991e-06,
      "loss": 0.4495,
      "num_input_tokens_seen": 6517384,
      "step": 9990
    },
    {
      "epoch": 5.8932783018867925,
      "grad_norm": 2.142604351043701,
      "learning_rate": 8.889741914936224e-06,
      "loss": 0.4596,
      "num_input_tokens_seen": 6520680,
      "step": 9995
    },
    {
      "epoch": 5.89622641509434,
      "grad_norm": 1.4938353300094604,
      "learning_rate": 8.888124891296602e-06,
      "loss": 0.3534,
      "num_input_tokens_seen": 6524264,
      "step": 10000
    },
    {
      "epoch": 5.899174528301887,
      "grad_norm": 2.2481894493103027,
      "learning_rate": 8.886506838259156e-06,
      "loss": 0.4953,
      "num_input_tokens_seen": 6527624,
      "step": 10005
    },
    {
      "epoch": 5.902122641509434,
      "grad_norm": 2.555413246154785,
      "learning_rate": 8.884887756252279e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 6530184,
      "step": 10010
    },
    {
      "epoch": 5.905070754716981,
      "grad_norm": 2.8460235595703125,
      "learning_rate": 8.88326764570462e-06,
      "loss": 0.3933,
      "num_input_tokens_seen": 6534152,
      "step": 10015
    },
    {
      "epoch": 5.908018867924528,
      "grad_norm": 1.5925590991973877,
      "learning_rate": 8.88164650704512e-06,
      "loss": 0.3367,
      "num_input_tokens_seen": 6537544,
      "step": 10020
    },
    {
      "epoch": 5.910966981132075,
      "grad_norm": 1.4225480556488037,
      "learning_rate": 8.880024340702978e-06,
      "loss": 0.5126,
      "num_input_tokens_seen": 6540232,
      "step": 10025
    },
    {
      "epoch": 5.913915094339623,
      "grad_norm": 2.6845648288726807,
      "learning_rate": 8.878401147107667e-06,
      "loss": 0.3383,
      "num_input_tokens_seen": 6544328,
      "step": 10030
    },
    {
      "epoch": 5.91686320754717,
      "grad_norm": 1.5596427917480469,
      "learning_rate": 8.87677692668894e-06,
      "loss": 0.442,
      "num_input_tokens_seen": 6547624,
      "step": 10035
    },
    {
      "epoch": 5.919811320754717,
      "grad_norm": 3.904777765274048,
      "learning_rate": 8.875151679876813e-06,
      "loss": 0.6714,
      "num_input_tokens_seen": 6551080,
      "step": 10040
    },
    {
      "epoch": 5.9227594339622645,
      "grad_norm": 2.436663866043091,
      "learning_rate": 8.873525407101577e-06,
      "loss": 0.4746,
      "num_input_tokens_seen": 6554120,
      "step": 10045
    },
    {
      "epoch": 5.925707547169811,
      "grad_norm": 2.0067825317382812,
      "learning_rate": 8.871898108793796e-06,
      "loss": 0.4872,
      "num_input_tokens_seen": 6557000,
      "step": 10050
    },
    {
      "epoch": 5.928655660377358,
      "grad_norm": 1.824528694152832,
      "learning_rate": 8.870269785384304e-06,
      "loss": 0.4556,
      "num_input_tokens_seen": 6560328,
      "step": 10055
    },
    {
      "epoch": 5.931603773584905,
      "grad_norm": 2.4293510913848877,
      "learning_rate": 8.868640437304206e-06,
      "loss": 0.4022,
      "num_input_tokens_seen": 6563432,
      "step": 10060
    },
    {
      "epoch": 5.934551886792453,
      "grad_norm": 1.5488672256469727,
      "learning_rate": 8.86701006498488e-06,
      "loss": 0.5331,
      "num_input_tokens_seen": 6566472,
      "step": 10065
    },
    {
      "epoch": 5.9375,
      "grad_norm": 1.7801010608673096,
      "learning_rate": 8.865378668857972e-06,
      "loss": 0.4415,
      "num_input_tokens_seen": 6570376,
      "step": 10070
    },
    {
      "epoch": 5.940448113207547,
      "grad_norm": 1.7903923988342285,
      "learning_rate": 8.863746249355404e-06,
      "loss": 0.6459,
      "num_input_tokens_seen": 6572712,
      "step": 10075
    },
    {
      "epoch": 5.943396226415095,
      "grad_norm": 4.813990592956543,
      "learning_rate": 8.862112806909365e-06,
      "loss": 0.5066,
      "num_input_tokens_seen": 6575688,
      "step": 10080
    },
    {
      "epoch": 5.946344339622642,
      "grad_norm": 1.9035300016403198,
      "learning_rate": 8.860478341952314e-06,
      "loss": 0.5235,
      "num_input_tokens_seen": 6578696,
      "step": 10085
    },
    {
      "epoch": 5.949292452830189,
      "grad_norm": 2.4319918155670166,
      "learning_rate": 8.858842854916985e-06,
      "loss": 0.3276,
      "num_input_tokens_seen": 6581096,
      "step": 10090
    },
    {
      "epoch": 5.9522405660377355,
      "grad_norm": 2.5436289310455322,
      "learning_rate": 8.85720634623638e-06,
      "loss": 0.4287,
      "num_input_tokens_seen": 6584712,
      "step": 10095
    },
    {
      "epoch": 5.955188679245283,
      "grad_norm": 2.1763691902160645,
      "learning_rate": 8.855568816343769e-06,
      "loss": 0.4793,
      "num_input_tokens_seen": 6587624,
      "step": 10100
    },
    {
      "epoch": 5.95813679245283,
      "grad_norm": 1.643932819366455,
      "learning_rate": 8.8539302656727e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 6590568,
      "step": 10105
    },
    {
      "epoch": 5.961084905660377,
      "grad_norm": 1.4726020097732544,
      "learning_rate": 8.852290694656983e-06,
      "loss": 0.5579,
      "num_input_tokens_seen": 6593832,
      "step": 10110
    },
    {
      "epoch": 5.964033018867925,
      "grad_norm": 1.864038348197937,
      "learning_rate": 8.8506501037307e-06,
      "loss": 0.4579,
      "num_input_tokens_seen": 6596488,
      "step": 10115
    },
    {
      "epoch": 5.966981132075472,
      "grad_norm": 2.398531436920166,
      "learning_rate": 8.849008493328209e-06,
      "loss": 0.3538,
      "num_input_tokens_seen": 6599080,
      "step": 10120
    },
    {
      "epoch": 5.969929245283019,
      "grad_norm": 1.2692980766296387,
      "learning_rate": 8.847365863884131e-06,
      "loss": 0.4391,
      "num_input_tokens_seen": 6602088,
      "step": 10125
    },
    {
      "epoch": 5.972877358490566,
      "grad_norm": 1.5619392395019531,
      "learning_rate": 8.845722215833359e-06,
      "loss": 0.3039,
      "num_input_tokens_seen": 6604648,
      "step": 10130
    },
    {
      "epoch": 5.975825471698113,
      "grad_norm": 2.0308432579040527,
      "learning_rate": 8.844077549611056e-06,
      "loss": 0.3713,
      "num_input_tokens_seen": 6608552,
      "step": 10135
    },
    {
      "epoch": 5.97877358490566,
      "grad_norm": 1.2307082414627075,
      "learning_rate": 8.842431865652654e-06,
      "loss": 0.4916,
      "num_input_tokens_seen": 6612072,
      "step": 10140
    },
    {
      "epoch": 5.9817216981132075,
      "grad_norm": 1.4244396686553955,
      "learning_rate": 8.840785164393858e-06,
      "loss": 0.451,
      "num_input_tokens_seen": 6614440,
      "step": 10145
    },
    {
      "epoch": 5.984669811320755,
      "grad_norm": 2.163041830062866,
      "learning_rate": 8.839137446270634e-06,
      "loss": 0.4549,
      "num_input_tokens_seen": 6618024,
      "step": 10150
    },
    {
      "epoch": 5.987617924528302,
      "grad_norm": 1.280468463897705,
      "learning_rate": 8.837488711719226e-06,
      "loss": 0.4564,
      "num_input_tokens_seen": 6621192,
      "step": 10155
    },
    {
      "epoch": 5.990566037735849,
      "grad_norm": 1.685910940170288,
      "learning_rate": 8.835838961176143e-06,
      "loss": 0.4378,
      "num_input_tokens_seen": 6623592,
      "step": 10160
    },
    {
      "epoch": 5.993514150943396,
      "grad_norm": 1.6796917915344238,
      "learning_rate": 8.834188195078164e-06,
      "loss": 0.3693,
      "num_input_tokens_seen": 6626504,
      "step": 10165
    },
    {
      "epoch": 5.996462264150943,
      "grad_norm": 3.3206424713134766,
      "learning_rate": 8.832536413862337e-06,
      "loss": 0.3843,
      "num_input_tokens_seen": 6629800,
      "step": 10170
    },
    {
      "epoch": 5.99941037735849,
      "grad_norm": 4.527689456939697,
      "learning_rate": 8.830883617965976e-06,
      "loss": 0.5516,
      "num_input_tokens_seen": 6632680,
      "step": 10175
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.5020108222961426,
      "eval_runtime": 19.1833,
      "eval_samples_per_second": 88.41,
      "eval_steps_per_second": 22.103,
      "num_input_tokens_seen": 6632800,
      "step": 10176
    },
    {
      "epoch": 6.002358490566038,
      "grad_norm": 4.684268951416016,
      "learning_rate": 8.829229807826665e-06,
      "loss": 0.3397,
      "num_input_tokens_seen": 6634976,
      "step": 10180
    },
    {
      "epoch": 6.005306603773585,
      "grad_norm": 1.6458688974380493,
      "learning_rate": 8.82757498388226e-06,
      "loss": 0.3627,
      "num_input_tokens_seen": 6637536,
      "step": 10185
    },
    {
      "epoch": 6.008254716981132,
      "grad_norm": 1.398048996925354,
      "learning_rate": 8.825919146570884e-06,
      "loss": 0.3804,
      "num_input_tokens_seen": 6641152,
      "step": 10190
    },
    {
      "epoch": 6.0112028301886795,
      "grad_norm": 1.6877312660217285,
      "learning_rate": 8.824262296330925e-06,
      "loss": 0.4702,
      "num_input_tokens_seen": 6644064,
      "step": 10195
    },
    {
      "epoch": 6.014150943396227,
      "grad_norm": 2.7180442810058594,
      "learning_rate": 8.822604433601041e-06,
      "loss": 0.4001,
      "num_input_tokens_seen": 6647488,
      "step": 10200
    },
    {
      "epoch": 6.017099056603773,
      "grad_norm": 2.2489688396453857,
      "learning_rate": 8.820945558820158e-06,
      "loss": 0.3117,
      "num_input_tokens_seen": 6649792,
      "step": 10205
    },
    {
      "epoch": 6.0200471698113205,
      "grad_norm": 2.4209139347076416,
      "learning_rate": 8.81928567242747e-06,
      "loss": 0.4939,
      "num_input_tokens_seen": 6652672,
      "step": 10210
    },
    {
      "epoch": 6.022995283018868,
      "grad_norm": 2.0638842582702637,
      "learning_rate": 8.817624774862443e-06,
      "loss": 0.5602,
      "num_input_tokens_seen": 6656224,
      "step": 10215
    },
    {
      "epoch": 6.025943396226415,
      "grad_norm": 2.3370678424835205,
      "learning_rate": 8.815962866564803e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 6659840,
      "step": 10220
    },
    {
      "epoch": 6.028891509433962,
      "grad_norm": 1.5156012773513794,
      "learning_rate": 8.814299947974547e-06,
      "loss": 0.4868,
      "num_input_tokens_seen": 6663168,
      "step": 10225
    },
    {
      "epoch": 6.03183962264151,
      "grad_norm": 3.3772552013397217,
      "learning_rate": 8.812636019531942e-06,
      "loss": 0.4054,
      "num_input_tokens_seen": 6665792,
      "step": 10230
    },
    {
      "epoch": 6.034787735849057,
      "grad_norm": 2.0655391216278076,
      "learning_rate": 8.810971081677517e-06,
      "loss": 0.3985,
      "num_input_tokens_seen": 6668928,
      "step": 10235
    },
    {
      "epoch": 6.037735849056604,
      "grad_norm": 1.8166823387145996,
      "learning_rate": 8.809305134852076e-06,
      "loss": 0.5117,
      "num_input_tokens_seen": 6671616,
      "step": 10240
    },
    {
      "epoch": 6.040683962264151,
      "grad_norm": 1.442678451538086,
      "learning_rate": 8.807638179496684e-06,
      "loss": 0.5803,
      "num_input_tokens_seen": 6674912,
      "step": 10245
    },
    {
      "epoch": 6.043632075471698,
      "grad_norm": 1.806138515472412,
      "learning_rate": 8.805970216052673e-06,
      "loss": 0.586,
      "num_input_tokens_seen": 6679072,
      "step": 10250
    },
    {
      "epoch": 6.046580188679245,
      "grad_norm": 2.371598243713379,
      "learning_rate": 8.804301244961645e-06,
      "loss": 0.4251,
      "num_input_tokens_seen": 6681824,
      "step": 10255
    },
    {
      "epoch": 6.0495283018867925,
      "grad_norm": 2.1468987464904785,
      "learning_rate": 8.802631266665465e-06,
      "loss": 0.3436,
      "num_input_tokens_seen": 6685088,
      "step": 10260
    },
    {
      "epoch": 6.05247641509434,
      "grad_norm": 1.8991566896438599,
      "learning_rate": 8.80096028160627e-06,
      "loss": 0.485,
      "num_input_tokens_seen": 6687552,
      "step": 10265
    },
    {
      "epoch": 6.055424528301887,
      "grad_norm": 2.3988687992095947,
      "learning_rate": 8.799288290226457e-06,
      "loss": 0.3963,
      "num_input_tokens_seen": 6690560,
      "step": 10270
    },
    {
      "epoch": 6.058372641509434,
      "grad_norm": 2.086108922958374,
      "learning_rate": 8.797615292968698e-06,
      "loss": 0.4803,
      "num_input_tokens_seen": 6694432,
      "step": 10275
    },
    {
      "epoch": 6.061320754716981,
      "grad_norm": 1.2736926078796387,
      "learning_rate": 8.795941290275923e-06,
      "loss": 0.3813,
      "num_input_tokens_seen": 6697568,
      "step": 10280
    },
    {
      "epoch": 6.064268867924528,
      "grad_norm": 1.2940442562103271,
      "learning_rate": 8.79426628259133e-06,
      "loss": 0.3476,
      "num_input_tokens_seen": 6701184,
      "step": 10285
    },
    {
      "epoch": 6.067216981132075,
      "grad_norm": 3.109651803970337,
      "learning_rate": 8.792590270358389e-06,
      "loss": 0.4239,
      "num_input_tokens_seen": 6704160,
      "step": 10290
    },
    {
      "epoch": 6.070165094339623,
      "grad_norm": 1.497114896774292,
      "learning_rate": 8.790913254020827e-06,
      "loss": 0.4336,
      "num_input_tokens_seen": 6706432,
      "step": 10295
    },
    {
      "epoch": 6.07311320754717,
      "grad_norm": 3.4340968132019043,
      "learning_rate": 8.789235234022643e-06,
      "loss": 0.4422,
      "num_input_tokens_seen": 6709760,
      "step": 10300
    },
    {
      "epoch": 6.076061320754717,
      "grad_norm": 1.7172421216964722,
      "learning_rate": 8.787556210808101e-06,
      "loss": 0.2895,
      "num_input_tokens_seen": 6713248,
      "step": 10305
    },
    {
      "epoch": 6.0790094339622645,
      "grad_norm": 2.9005744457244873,
      "learning_rate": 8.78587618482173e-06,
      "loss": 0.5111,
      "num_input_tokens_seen": 6715904,
      "step": 10310
    },
    {
      "epoch": 6.081957547169812,
      "grad_norm": 0.6793528199195862,
      "learning_rate": 8.78419515650832e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 6721120,
      "step": 10315
    },
    {
      "epoch": 6.084905660377358,
      "grad_norm": 1.7809995412826538,
      "learning_rate": 8.782513126312934e-06,
      "loss": 0.4124,
      "num_input_tokens_seen": 6724160,
      "step": 10320
    },
    {
      "epoch": 6.087853773584905,
      "grad_norm": 4.138539791107178,
      "learning_rate": 8.780830094680897e-06,
      "loss": 0.5492,
      "num_input_tokens_seen": 6728096,
      "step": 10325
    },
    {
      "epoch": 6.090801886792453,
      "grad_norm": 3.349318027496338,
      "learning_rate": 8.779146062057797e-06,
      "loss": 0.4293,
      "num_input_tokens_seen": 6731392,
      "step": 10330
    },
    {
      "epoch": 6.09375,
      "grad_norm": 2.0829672813415527,
      "learning_rate": 8.777461028889492e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 6735776,
      "step": 10335
    },
    {
      "epoch": 6.096698113207547,
      "grad_norm": 1.5729119777679443,
      "learning_rate": 8.775774995622097e-06,
      "loss": 0.3952,
      "num_input_tokens_seen": 6738656,
      "step": 10340
    },
    {
      "epoch": 6.099646226415095,
      "grad_norm": 3.1893270015716553,
      "learning_rate": 8.774087962702e-06,
      "loss": 0.5506,
      "num_input_tokens_seen": 6741856,
      "step": 10345
    },
    {
      "epoch": 6.102594339622642,
      "grad_norm": 1.5361682176589966,
      "learning_rate": 8.772399930575849e-06,
      "loss": 0.4465,
      "num_input_tokens_seen": 6746240,
      "step": 10350
    },
    {
      "epoch": 6.105542452830188,
      "grad_norm": 1.7695817947387695,
      "learning_rate": 8.77071089969056e-06,
      "loss": 0.7837,
      "num_input_tokens_seen": 6749088,
      "step": 10355
    },
    {
      "epoch": 6.1084905660377355,
      "grad_norm": 4.132552146911621,
      "learning_rate": 8.769020870493309e-06,
      "loss": 0.3996,
      "num_input_tokens_seen": 6751392,
      "step": 10360
    },
    {
      "epoch": 6.111438679245283,
      "grad_norm": 1.7339767217636108,
      "learning_rate": 8.767329843431537e-06,
      "loss": 0.4884,
      "num_input_tokens_seen": 6755072,
      "step": 10365
    },
    {
      "epoch": 6.11438679245283,
      "grad_norm": 2.132688522338867,
      "learning_rate": 8.765637818952954e-06,
      "loss": 0.4144,
      "num_input_tokens_seen": 6757696,
      "step": 10370
    },
    {
      "epoch": 6.117334905660377,
      "grad_norm": 1.8360496759414673,
      "learning_rate": 8.76394479750553e-06,
      "loss": 0.344,
      "num_input_tokens_seen": 6760736,
      "step": 10375
    },
    {
      "epoch": 6.120283018867925,
      "grad_norm": 1.6938674449920654,
      "learning_rate": 8.762250779537499e-06,
      "loss": 0.5142,
      "num_input_tokens_seen": 6763680,
      "step": 10380
    },
    {
      "epoch": 6.123231132075472,
      "grad_norm": 1.5598729848861694,
      "learning_rate": 8.760555765497358e-06,
      "loss": 0.3207,
      "num_input_tokens_seen": 6766496,
      "step": 10385
    },
    {
      "epoch": 6.126179245283019,
      "grad_norm": 1.6034488677978516,
      "learning_rate": 8.758859755833873e-06,
      "loss": 0.3245,
      "num_input_tokens_seen": 6769280,
      "step": 10390
    },
    {
      "epoch": 6.129127358490566,
      "grad_norm": 2.908996343612671,
      "learning_rate": 8.757162750996066e-06,
      "loss": 0.4854,
      "num_input_tokens_seen": 6772064,
      "step": 10395
    },
    {
      "epoch": 6.132075471698113,
      "grad_norm": 1.359206199645996,
      "learning_rate": 8.755464751433229e-06,
      "loss": 0.4088,
      "num_input_tokens_seen": 6775200,
      "step": 10400
    },
    {
      "epoch": 6.13502358490566,
      "grad_norm": 1.4671485424041748,
      "learning_rate": 8.753765757594915e-06,
      "loss": 0.4999,
      "num_input_tokens_seen": 6778784,
      "step": 10405
    },
    {
      "epoch": 6.1379716981132075,
      "grad_norm": 1.521746039390564,
      "learning_rate": 8.752065769930938e-06,
      "loss": 0.4165,
      "num_input_tokens_seen": 6781536,
      "step": 10410
    },
    {
      "epoch": 6.140919811320755,
      "grad_norm": 1.831125259399414,
      "learning_rate": 8.75036478889138e-06,
      "loss": 0.3385,
      "num_input_tokens_seen": 6784064,
      "step": 10415
    },
    {
      "epoch": 6.143867924528302,
      "grad_norm": 1.5671486854553223,
      "learning_rate": 8.748662814926576e-06,
      "loss": 0.4212,
      "num_input_tokens_seen": 6788736,
      "step": 10420
    },
    {
      "epoch": 6.146816037735849,
      "grad_norm": 1.3478660583496094,
      "learning_rate": 8.746959848487139e-06,
      "loss": 0.3795,
      "num_input_tokens_seen": 6792256,
      "step": 10425
    },
    {
      "epoch": 6.149764150943396,
      "grad_norm": 2.5194122791290283,
      "learning_rate": 8.745255890023934e-06,
      "loss": 0.3895,
      "num_input_tokens_seen": 6795648,
      "step": 10430
    },
    {
      "epoch": 6.152712264150943,
      "grad_norm": 1.3107904195785522,
      "learning_rate": 8.74355093998809e-06,
      "loss": 0.3765,
      "num_input_tokens_seen": 6798784,
      "step": 10435
    },
    {
      "epoch": 6.15566037735849,
      "grad_norm": 1.340149164199829,
      "learning_rate": 8.741844998831001e-06,
      "loss": 0.3448,
      "num_input_tokens_seen": 6802432,
      "step": 10440
    },
    {
      "epoch": 6.158608490566038,
      "grad_norm": 1.7845377922058105,
      "learning_rate": 8.740138067004323e-06,
      "loss": 0.4029,
      "num_input_tokens_seen": 6805312,
      "step": 10445
    },
    {
      "epoch": 6.161556603773585,
      "grad_norm": 3.7810888290405273,
      "learning_rate": 8.738430144959973e-06,
      "loss": 0.4001,
      "num_input_tokens_seen": 6808640,
      "step": 10450
    },
    {
      "epoch": 6.164504716981132,
      "grad_norm": 1.5898064374923706,
      "learning_rate": 8.73672123315013e-06,
      "loss": 0.4271,
      "num_input_tokens_seen": 6811328,
      "step": 10455
    },
    {
      "epoch": 6.1674528301886795,
      "grad_norm": 1.4934734106063843,
      "learning_rate": 8.735011332027234e-06,
      "loss": 0.411,
      "num_input_tokens_seen": 6814400,
      "step": 10460
    },
    {
      "epoch": 6.170400943396227,
      "grad_norm": 3.3223869800567627,
      "learning_rate": 8.733300442043993e-06,
      "loss": 0.5188,
      "num_input_tokens_seen": 6817312,
      "step": 10465
    },
    {
      "epoch": 6.173349056603773,
      "grad_norm": 2.574805974960327,
      "learning_rate": 8.73158856365337e-06,
      "loss": 0.6431,
      "num_input_tokens_seen": 6820192,
      "step": 10470
    },
    {
      "epoch": 6.1762971698113205,
      "grad_norm": 4.194033622741699,
      "learning_rate": 8.729875697308592e-06,
      "loss": 0.3565,
      "num_input_tokens_seen": 6823264,
      "step": 10475
    },
    {
      "epoch": 6.179245283018868,
      "grad_norm": 4.226123332977295,
      "learning_rate": 8.728161843463148e-06,
      "loss": 0.4331,
      "num_input_tokens_seen": 6825888,
      "step": 10480
    },
    {
      "epoch": 6.182193396226415,
      "grad_norm": 3.0703721046447754,
      "learning_rate": 8.726447002570791e-06,
      "loss": 0.6685,
      "num_input_tokens_seen": 6829760,
      "step": 10485
    },
    {
      "epoch": 6.185141509433962,
      "grad_norm": 1.5446996688842773,
      "learning_rate": 8.724731175085526e-06,
      "loss": 0.4639,
      "num_input_tokens_seen": 6832416,
      "step": 10490
    },
    {
      "epoch": 6.18808962264151,
      "grad_norm": 1.5521007776260376,
      "learning_rate": 8.723014361461633e-06,
      "loss": 0.3098,
      "num_input_tokens_seen": 6835776,
      "step": 10495
    },
    {
      "epoch": 6.191037735849057,
      "grad_norm": 0.749037504196167,
      "learning_rate": 8.72129656215364e-06,
      "loss": 0.406,
      "num_input_tokens_seen": 6839328,
      "step": 10500
    },
    {
      "epoch": 6.193985849056604,
      "grad_norm": 1.5499939918518066,
      "learning_rate": 8.719577777616347e-06,
      "loss": 0.3585,
      "num_input_tokens_seen": 6841888,
      "step": 10505
    },
    {
      "epoch": 6.196933962264151,
      "grad_norm": 2.25122332572937,
      "learning_rate": 8.717858008304804e-06,
      "loss": 0.409,
      "num_input_tokens_seen": 6845056,
      "step": 10510
    },
    {
      "epoch": 6.199882075471698,
      "grad_norm": 1.3916805982589722,
      "learning_rate": 8.71613725467433e-06,
      "loss": 0.4572,
      "num_input_tokens_seen": 6847872,
      "step": 10515
    },
    {
      "epoch": 6.202830188679245,
      "grad_norm": 2.3830838203430176,
      "learning_rate": 8.714415517180506e-06,
      "loss": 0.3304,
      "num_input_tokens_seen": 6851776,
      "step": 10520
    },
    {
      "epoch": 6.2057783018867925,
      "grad_norm": 2.5637621879577637,
      "learning_rate": 8.712692796279164e-06,
      "loss": 0.38,
      "num_input_tokens_seen": 6856832,
      "step": 10525
    },
    {
      "epoch": 6.20872641509434,
      "grad_norm": 1.9720937013626099,
      "learning_rate": 8.710969092426401e-06,
      "loss": 0.3287,
      "num_input_tokens_seen": 6859296,
      "step": 10530
    },
    {
      "epoch": 6.211674528301887,
      "grad_norm": 2.9785029888153076,
      "learning_rate": 8.70924440607858e-06,
      "loss": 0.5329,
      "num_input_tokens_seen": 6862304,
      "step": 10535
    },
    {
      "epoch": 6.214622641509434,
      "grad_norm": 2.3837740421295166,
      "learning_rate": 8.707518737692315e-06,
      "loss": 0.4738,
      "num_input_tokens_seen": 6865248,
      "step": 10540
    },
    {
      "epoch": 6.217570754716981,
      "grad_norm": 1.1246404647827148,
      "learning_rate": 8.705792087724485e-06,
      "loss": 0.2997,
      "num_input_tokens_seen": 6868032,
      "step": 10545
    },
    {
      "epoch": 6.220518867924528,
      "grad_norm": 1.95211923122406,
      "learning_rate": 8.704064456632231e-06,
      "loss": 0.4927,
      "num_input_tokens_seen": 6871936,
      "step": 10550
    },
    {
      "epoch": 6.223466981132075,
      "grad_norm": 1.6259949207305908,
      "learning_rate": 8.702335844872946e-06,
      "loss": 0.3798,
      "num_input_tokens_seen": 6874720,
      "step": 10555
    },
    {
      "epoch": 6.226415094339623,
      "grad_norm": 1.6640855073928833,
      "learning_rate": 8.700606252904293e-06,
      "loss": 0.4181,
      "num_input_tokens_seen": 6877984,
      "step": 10560
    },
    {
      "epoch": 6.22936320754717,
      "grad_norm": 2.951138734817505,
      "learning_rate": 8.698875681184183e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 6881504,
      "step": 10565
    },
    {
      "epoch": 6.232311320754717,
      "grad_norm": 2.4996767044067383,
      "learning_rate": 8.697144130170797e-06,
      "loss": 0.4598,
      "num_input_tokens_seen": 6884832,
      "step": 10570
    },
    {
      "epoch": 6.2352594339622645,
      "grad_norm": 1.5430487394332886,
      "learning_rate": 8.695411600322568e-06,
      "loss": 0.4888,
      "num_input_tokens_seen": 6887648,
      "step": 10575
    },
    {
      "epoch": 6.238207547169812,
      "grad_norm": 1.2646154165267944,
      "learning_rate": 8.693678092098191e-06,
      "loss": 0.3685,
      "num_input_tokens_seen": 6890528,
      "step": 10580
    },
    {
      "epoch": 6.241155660377358,
      "grad_norm": 1.075602412223816,
      "learning_rate": 8.691943605956621e-06,
      "loss": 0.4171,
      "num_input_tokens_seen": 6894112,
      "step": 10585
    },
    {
      "epoch": 6.244103773584905,
      "grad_norm": 1.4442609548568726,
      "learning_rate": 8.690208142357069e-06,
      "loss": 0.4101,
      "num_input_tokens_seen": 6897440,
      "step": 10590
    },
    {
      "epoch": 6.247051886792453,
      "grad_norm": 1.982790470123291,
      "learning_rate": 8.68847170175901e-06,
      "loss": 0.3522,
      "num_input_tokens_seen": 6901088,
      "step": 10595
    },
    {
      "epoch": 6.25,
      "grad_norm": 1.2460302114486694,
      "learning_rate": 8.686734284622168e-06,
      "loss": 0.3493,
      "num_input_tokens_seen": 6903712,
      "step": 10600
    },
    {
      "epoch": 6.252948113207547,
      "grad_norm": 3.115110397338867,
      "learning_rate": 8.684995891406537e-06,
      "loss": 0.3964,
      "num_input_tokens_seen": 6906912,
      "step": 10605
    },
    {
      "epoch": 6.255896226415095,
      "grad_norm": 2.1191303730010986,
      "learning_rate": 8.683256522572362e-06,
      "loss": 0.5167,
      "num_input_tokens_seen": 6909536,
      "step": 10610
    },
    {
      "epoch": 6.258844339622642,
      "grad_norm": 3.4223861694335938,
      "learning_rate": 8.68151617858015e-06,
      "loss": 0.3299,
      "num_input_tokens_seen": 6912320,
      "step": 10615
    },
    {
      "epoch": 6.261792452830189,
      "grad_norm": 1.2667052745819092,
      "learning_rate": 8.67977485989066e-06,
      "loss": 0.4186,
      "num_input_tokens_seen": 6915456,
      "step": 10620
    },
    {
      "epoch": 6.2647405660377355,
      "grad_norm": 3.6430094242095947,
      "learning_rate": 8.67803256696492e-06,
      "loss": 0.467,
      "num_input_tokens_seen": 6919328,
      "step": 10625
    },
    {
      "epoch": 6.267688679245283,
      "grad_norm": 1.7098509073257446,
      "learning_rate": 8.676289300264205e-06,
      "loss": 0.4627,
      "num_input_tokens_seen": 6923104,
      "step": 10630
    },
    {
      "epoch": 6.27063679245283,
      "grad_norm": 2.6843483448028564,
      "learning_rate": 8.674545060250054e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 6926048,
      "step": 10635
    },
    {
      "epoch": 6.273584905660377,
      "grad_norm": 2.3614320755004883,
      "learning_rate": 8.672799847384263e-06,
      "loss": 0.3809,
      "num_input_tokens_seen": 6929792,
      "step": 10640
    },
    {
      "epoch": 6.276533018867925,
      "grad_norm": 1.6271038055419922,
      "learning_rate": 8.671053662128883e-06,
      "loss": 0.4686,
      "num_input_tokens_seen": 6932736,
      "step": 10645
    },
    {
      "epoch": 6.279481132075472,
      "grad_norm": 1.1814461946487427,
      "learning_rate": 8.669306504946223e-06,
      "loss": 0.4657,
      "num_input_tokens_seen": 6936032,
      "step": 10650
    },
    {
      "epoch": 6.282429245283019,
      "grad_norm": 4.465968608856201,
      "learning_rate": 8.667558376298854e-06,
      "loss": 0.4659,
      "num_input_tokens_seen": 6939136,
      "step": 10655
    },
    {
      "epoch": 6.285377358490566,
      "grad_norm": 2.177250862121582,
      "learning_rate": 8.665809276649597e-06,
      "loss": 0.3539,
      "num_input_tokens_seen": 6942656,
      "step": 10660
    },
    {
      "epoch": 6.288325471698113,
      "grad_norm": 2.0434207916259766,
      "learning_rate": 8.664059206461537e-06,
      "loss": 0.5863,
      "num_input_tokens_seen": 6945376,
      "step": 10665
    },
    {
      "epoch": 6.29127358490566,
      "grad_norm": 2.3921236991882324,
      "learning_rate": 8.662308166198009e-06,
      "loss": 0.4993,
      "num_input_tokens_seen": 6948448,
      "step": 10670
    },
    {
      "epoch": 6.2942216981132075,
      "grad_norm": 5.244947910308838,
      "learning_rate": 8.660556156322611e-06,
      "loss": 0.4374,
      "num_input_tokens_seen": 6951008,
      "step": 10675
    },
    {
      "epoch": 6.297169811320755,
      "grad_norm": 2.308397054672241,
      "learning_rate": 8.658803177299196e-06,
      "loss": 0.319,
      "num_input_tokens_seen": 6954400,
      "step": 10680
    },
    {
      "epoch": 6.300117924528302,
      "grad_norm": 2.06506609916687,
      "learning_rate": 8.65704922959187e-06,
      "loss": 0.4351,
      "num_input_tokens_seen": 6959168,
      "step": 10685
    },
    {
      "epoch": 6.303066037735849,
      "grad_norm": 1.74628746509552,
      "learning_rate": 8.655294313664998e-06,
      "loss": 0.4923,
      "num_input_tokens_seen": 6963456,
      "step": 10690
    },
    {
      "epoch": 6.306014150943396,
      "grad_norm": 1.2609224319458008,
      "learning_rate": 8.653538429983204e-06,
      "loss": 0.381,
      "num_input_tokens_seen": 6967232,
      "step": 10695
    },
    {
      "epoch": 6.308962264150943,
      "grad_norm": 2.5142407417297363,
      "learning_rate": 8.651781579011366e-06,
      "loss": 0.3356,
      "num_input_tokens_seen": 6970336,
      "step": 10700
    },
    {
      "epoch": 6.31191037735849,
      "grad_norm": 1.3669636249542236,
      "learning_rate": 8.650023761214615e-06,
      "loss": 0.474,
      "num_input_tokens_seen": 6973184,
      "step": 10705
    },
    {
      "epoch": 6.314858490566038,
      "grad_norm": 2.93194842338562,
      "learning_rate": 8.648264977058344e-06,
      "loss": 0.5343,
      "num_input_tokens_seen": 6976480,
      "step": 10710
    },
    {
      "epoch": 6.317806603773585,
      "grad_norm": 2.2444570064544678,
      "learning_rate": 8.646505227008197e-06,
      "loss": 0.5249,
      "num_input_tokens_seen": 6980032,
      "step": 10715
    },
    {
      "epoch": 6.320754716981132,
      "grad_norm": 2.386554002761841,
      "learning_rate": 8.644744511530074e-06,
      "loss": 0.4217,
      "num_input_tokens_seen": 6983296,
      "step": 10720
    },
    {
      "epoch": 6.3237028301886795,
      "grad_norm": 1.8733233213424683,
      "learning_rate": 8.642982831090135e-06,
      "loss": 0.5654,
      "num_input_tokens_seen": 6986784,
      "step": 10725
    },
    {
      "epoch": 6.326650943396227,
      "grad_norm": 1.7614340782165527,
      "learning_rate": 8.64122018615479e-06,
      "loss": 0.5456,
      "num_input_tokens_seen": 6991648,
      "step": 10730
    },
    {
      "epoch": 6.329599056603773,
      "grad_norm": 1.538223385810852,
      "learning_rate": 8.63945657719071e-06,
      "loss": 0.3512,
      "num_input_tokens_seen": 6995104,
      "step": 10735
    },
    {
      "epoch": 6.3325471698113205,
      "grad_norm": 1.4793990850448608,
      "learning_rate": 8.637692004664816e-06,
      "loss": 0.3562,
      "num_input_tokens_seen": 6997504,
      "step": 10740
    },
    {
      "epoch": 6.335495283018868,
      "grad_norm": 2.9595794677734375,
      "learning_rate": 8.635926469044284e-06,
      "loss": 0.5171,
      "num_input_tokens_seen": 7001312,
      "step": 10745
    },
    {
      "epoch": 6.338443396226415,
      "grad_norm": 2.223149299621582,
      "learning_rate": 8.63415997079655e-06,
      "loss": 0.352,
      "num_input_tokens_seen": 7004032,
      "step": 10750
    },
    {
      "epoch": 6.341391509433962,
      "grad_norm": 2.6209397315979004,
      "learning_rate": 8.6323925103893e-06,
      "loss": 0.436,
      "num_input_tokens_seen": 7006848,
      "step": 10755
    },
    {
      "epoch": 6.34433962264151,
      "grad_norm": 3.91560697555542,
      "learning_rate": 8.63062408829048e-06,
      "loss": 0.3455,
      "num_input_tokens_seen": 7009600,
      "step": 10760
    },
    {
      "epoch": 6.347287735849057,
      "grad_norm": 2.4054553508758545,
      "learning_rate": 8.628854704968285e-06,
      "loss": 0.3871,
      "num_input_tokens_seen": 7011904,
      "step": 10765
    },
    {
      "epoch": 6.350235849056604,
      "grad_norm": 3.0327184200286865,
      "learning_rate": 8.62708436089117e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 7015168,
      "step": 10770
    },
    {
      "epoch": 6.353183962264151,
      "grad_norm": 3.2022640705108643,
      "learning_rate": 8.625313056527836e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 7017632,
      "step": 10775
    },
    {
      "epoch": 6.356132075471698,
      "grad_norm": 1.177520751953125,
      "learning_rate": 8.623540792347244e-06,
      "loss": 0.3014,
      "num_input_tokens_seen": 7022304,
      "step": 10780
    },
    {
      "epoch": 6.359080188679245,
      "grad_norm": 2.0489003658294678,
      "learning_rate": 8.621767568818614e-06,
      "loss": 0.5369,
      "num_input_tokens_seen": 7025280,
      "step": 10785
    },
    {
      "epoch": 6.3620283018867925,
      "grad_norm": 1.0781079530715942,
      "learning_rate": 8.619993386411409e-06,
      "loss": 0.5328,
      "num_input_tokens_seen": 7029216,
      "step": 10790
    },
    {
      "epoch": 6.36497641509434,
      "grad_norm": 1.5288147926330566,
      "learning_rate": 8.618218245595356e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 7032192,
      "step": 10795
    },
    {
      "epoch": 6.367924528301887,
      "grad_norm": 1.3783318996429443,
      "learning_rate": 8.616442146840427e-06,
      "loss": 0.477,
      "num_input_tokens_seen": 7035072,
      "step": 10800
    },
    {
      "epoch": 6.370872641509434,
      "grad_norm": 2.512392520904541,
      "learning_rate": 8.614665090616854e-06,
      "loss": 0.4155,
      "num_input_tokens_seen": 7039072,
      "step": 10805
    },
    {
      "epoch": 6.373820754716981,
      "grad_norm": 2.566673994064331,
      "learning_rate": 8.61288707739512e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 7043584,
      "step": 10810
    },
    {
      "epoch": 6.376768867924528,
      "grad_norm": 1.7683953046798706,
      "learning_rate": 8.611108107645963e-06,
      "loss": 0.4746,
      "num_input_tokens_seen": 7047264,
      "step": 10815
    },
    {
      "epoch": 6.379716981132075,
      "grad_norm": 1.623326063156128,
      "learning_rate": 8.609328181840368e-06,
      "loss": 0.3146,
      "num_input_tokens_seen": 7050528,
      "step": 10820
    },
    {
      "epoch": 6.382665094339623,
      "grad_norm": 1.8695857524871826,
      "learning_rate": 8.607547300449585e-06,
      "loss": 0.4013,
      "num_input_tokens_seen": 7053920,
      "step": 10825
    },
    {
      "epoch": 6.38561320754717,
      "grad_norm": 2.623251438140869,
      "learning_rate": 8.605765463945105e-06,
      "loss": 0.444,
      "num_input_tokens_seen": 7056736,
      "step": 10830
    },
    {
      "epoch": 6.388561320754717,
      "grad_norm": 1.643649935722351,
      "learning_rate": 8.603982672798678e-06,
      "loss": 0.3847,
      "num_input_tokens_seen": 7059616,
      "step": 10835
    },
    {
      "epoch": 6.3915094339622645,
      "grad_norm": 1.6045095920562744,
      "learning_rate": 8.602198927482309e-06,
      "loss": 0.4994,
      "num_input_tokens_seen": 7063424,
      "step": 10840
    },
    {
      "epoch": 6.394457547169811,
      "grad_norm": 3.3581929206848145,
      "learning_rate": 8.600414228468245e-06,
      "loss": 0.3366,
      "num_input_tokens_seen": 7067520,
      "step": 10845
    },
    {
      "epoch": 6.397405660377358,
      "grad_norm": 2.165945291519165,
      "learning_rate": 8.598628576229e-06,
      "loss": 0.4935,
      "num_input_tokens_seen": 7070720,
      "step": 10850
    },
    {
      "epoch": 6.400353773584905,
      "grad_norm": 1.7261625528335571,
      "learning_rate": 8.596841971237328e-06,
      "loss": 0.3289,
      "num_input_tokens_seen": 7074656,
      "step": 10855
    },
    {
      "epoch": 6.403301886792453,
      "grad_norm": 1.53758704662323,
      "learning_rate": 8.595054413966246e-06,
      "loss": 0.4152,
      "num_input_tokens_seen": 7077088,
      "step": 10860
    },
    {
      "epoch": 6.40625,
      "grad_norm": 3.110621452331543,
      "learning_rate": 8.593265904889011e-06,
      "loss": 0.4701,
      "num_input_tokens_seen": 7080000,
      "step": 10865
    },
    {
      "epoch": 6.409198113207547,
      "grad_norm": 1.217392921447754,
      "learning_rate": 8.591476444479141e-06,
      "loss": 0.3213,
      "num_input_tokens_seen": 7083072,
      "step": 10870
    },
    {
      "epoch": 6.412146226415095,
      "grad_norm": 1.526443362236023,
      "learning_rate": 8.589686033210407e-06,
      "loss": 0.4516,
      "num_input_tokens_seen": 7085600,
      "step": 10875
    },
    {
      "epoch": 6.415094339622642,
      "grad_norm": 1.4059326648712158,
      "learning_rate": 8.587894671556823e-06,
      "loss": 0.4884,
      "num_input_tokens_seen": 7088960,
      "step": 10880
    },
    {
      "epoch": 6.418042452830189,
      "grad_norm": 3.5297603607177734,
      "learning_rate": 8.586102359992663e-06,
      "loss": 0.4484,
      "num_input_tokens_seen": 7092448,
      "step": 10885
    },
    {
      "epoch": 6.4209905660377355,
      "grad_norm": 2.4462599754333496,
      "learning_rate": 8.584309098992447e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 7094624,
      "step": 10890
    },
    {
      "epoch": 6.423938679245283,
      "grad_norm": 2.759803533554077,
      "learning_rate": 8.58251488903095e-06,
      "loss": 0.3845,
      "num_input_tokens_seen": 7097120,
      "step": 10895
    },
    {
      "epoch": 6.42688679245283,
      "grad_norm": 1.5519237518310547,
      "learning_rate": 8.580719730583196e-06,
      "loss": 0.3401,
      "num_input_tokens_seen": 7099616,
      "step": 10900
    },
    {
      "epoch": 6.429834905660377,
      "grad_norm": 1.7532367706298828,
      "learning_rate": 8.578923624124462e-06,
      "loss": 0.388,
      "num_input_tokens_seen": 7102592,
      "step": 10905
    },
    {
      "epoch": 6.432783018867925,
      "grad_norm": 3.7045364379882812,
      "learning_rate": 8.577126570130273e-06,
      "loss": 0.4829,
      "num_input_tokens_seen": 7105024,
      "step": 10910
    },
    {
      "epoch": 6.435731132075472,
      "grad_norm": 1.9576891660690308,
      "learning_rate": 8.575328569076408e-06,
      "loss": 0.3273,
      "num_input_tokens_seen": 7109056,
      "step": 10915
    },
    {
      "epoch": 6.438679245283019,
      "grad_norm": 3.592815399169922,
      "learning_rate": 8.573529621438896e-06,
      "loss": 0.4052,
      "num_input_tokens_seen": 7114080,
      "step": 10920
    },
    {
      "epoch": 6.441627358490566,
      "grad_norm": 2.876549005508423,
      "learning_rate": 8.571729727694015e-06,
      "loss": 0.437,
      "num_input_tokens_seen": 7118016,
      "step": 10925
    },
    {
      "epoch": 6.444575471698113,
      "grad_norm": 2.1755335330963135,
      "learning_rate": 8.569928888318298e-06,
      "loss": 0.4822,
      "num_input_tokens_seen": 7120288,
      "step": 10930
    },
    {
      "epoch": 6.44752358490566,
      "grad_norm": 3.2360498905181885,
      "learning_rate": 8.56812710378852e-06,
      "loss": 0.5622,
      "num_input_tokens_seen": 7123200,
      "step": 10935
    },
    {
      "epoch": 6.4504716981132075,
      "grad_norm": 3.8863954544067383,
      "learning_rate": 8.566324374581714e-06,
      "loss": 0.3988,
      "num_input_tokens_seen": 7126336,
      "step": 10940
    },
    {
      "epoch": 6.453419811320755,
      "grad_norm": 1.6692085266113281,
      "learning_rate": 8.564520701175158e-06,
      "loss": 0.4559,
      "num_input_tokens_seen": 7129312,
      "step": 10945
    },
    {
      "epoch": 6.456367924528302,
      "grad_norm": 3.6473913192749023,
      "learning_rate": 8.562716084046387e-06,
      "loss": 0.5083,
      "num_input_tokens_seen": 7133088,
      "step": 10950
    },
    {
      "epoch": 6.459316037735849,
      "grad_norm": 3.0011744499206543,
      "learning_rate": 8.560910523673177e-06,
      "loss": 0.4521,
      "num_input_tokens_seen": 7137248,
      "step": 10955
    },
    {
      "epoch": 6.462264150943396,
      "grad_norm": 1.306213140487671,
      "learning_rate": 8.55910402053356e-06,
      "loss": 0.2731,
      "num_input_tokens_seen": 7141312,
      "step": 10960
    },
    {
      "epoch": 6.465212264150943,
      "grad_norm": 1.998231291770935,
      "learning_rate": 8.557296575105814e-06,
      "loss": 0.5163,
      "num_input_tokens_seen": 7145312,
      "step": 10965
    },
    {
      "epoch": 6.46816037735849,
      "grad_norm": 2.068466901779175,
      "learning_rate": 8.555488187868469e-06,
      "loss": 0.3538,
      "num_input_tokens_seen": 7147840,
      "step": 10970
    },
    {
      "epoch": 6.471108490566038,
      "grad_norm": 2.4957549571990967,
      "learning_rate": 8.5536788593003e-06,
      "loss": 0.3487,
      "num_input_tokens_seen": 7152256,
      "step": 10975
    },
    {
      "epoch": 6.474056603773585,
      "grad_norm": 1.5608731508255005,
      "learning_rate": 8.55186858988034e-06,
      "loss": 0.4033,
      "num_input_tokens_seen": 7154976,
      "step": 10980
    },
    {
      "epoch": 6.477004716981132,
      "grad_norm": 1.6199347972869873,
      "learning_rate": 8.550057380087863e-06,
      "loss": 0.525,
      "num_input_tokens_seen": 7158112,
      "step": 10985
    },
    {
      "epoch": 6.4799528301886795,
      "grad_norm": 4.086792945861816,
      "learning_rate": 8.548245230402396e-06,
      "loss": 0.4632,
      "num_input_tokens_seen": 7162240,
      "step": 10990
    },
    {
      "epoch": 6.482900943396227,
      "grad_norm": 2.2521724700927734,
      "learning_rate": 8.546432141303711e-06,
      "loss": 0.4392,
      "num_input_tokens_seen": 7166880,
      "step": 10995
    },
    {
      "epoch": 6.485849056603773,
      "grad_norm": 1.2129896879196167,
      "learning_rate": 8.544618113271833e-06,
      "loss": 0.3647,
      "num_input_tokens_seen": 7170496,
      "step": 11000
    },
    {
      "epoch": 6.4887971698113205,
      "grad_norm": 1.4614900350570679,
      "learning_rate": 8.542803146787032e-06,
      "loss": 0.5893,
      "num_input_tokens_seen": 7174144,
      "step": 11005
    },
    {
      "epoch": 6.491745283018868,
      "grad_norm": 1.163750171661377,
      "learning_rate": 8.54098724232983e-06,
      "loss": 0.4448,
      "num_input_tokens_seen": 7176992,
      "step": 11010
    },
    {
      "epoch": 6.494693396226415,
      "grad_norm": 2.3187289237976074,
      "learning_rate": 8.539170400380994e-06,
      "loss": 0.531,
      "num_input_tokens_seen": 7179328,
      "step": 11015
    },
    {
      "epoch": 6.497641509433962,
      "grad_norm": 1.4366041421890259,
      "learning_rate": 8.537352621421542e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 7181952,
      "step": 11020
    },
    {
      "epoch": 6.50058962264151,
      "grad_norm": 1.4318405389785767,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.4972,
      "num_input_tokens_seen": 7185536,
      "step": 11025
    },
    {
      "epoch": 6.503537735849057,
      "grad_norm": 2.680626153945923,
      "learning_rate": 8.533714254396096e-06,
      "loss": 0.4412,
      "num_input_tokens_seen": 7188288,
      "step": 11030
    },
    {
      "epoch": 6.506485849056604,
      "grad_norm": 2.201526641845703,
      "learning_rate": 8.531893667293375e-06,
      "loss": 0.4875,
      "num_input_tokens_seen": 7192128,
      "step": 11035
    },
    {
      "epoch": 6.509433962264151,
      "grad_norm": 1.1782864332199097,
      "learning_rate": 8.530072145106585e-06,
      "loss": 0.3171,
      "num_input_tokens_seen": 7195104,
      "step": 11040
    },
    {
      "epoch": 6.512382075471698,
      "grad_norm": 2.1387217044830322,
      "learning_rate": 8.528249688317978e-06,
      "loss": 0.3848,
      "num_input_tokens_seen": 7197952,
      "step": 11045
    },
    {
      "epoch": 6.515330188679245,
      "grad_norm": 3.1230356693267822,
      "learning_rate": 8.526426297410062e-06,
      "loss": 0.3133,
      "num_input_tokens_seen": 7201216,
      "step": 11050
    },
    {
      "epoch": 6.5182783018867925,
      "grad_norm": 1.4122166633605957,
      "learning_rate": 8.524601972865586e-06,
      "loss": 0.402,
      "num_input_tokens_seen": 7207520,
      "step": 11055
    },
    {
      "epoch": 6.52122641509434,
      "grad_norm": 1.490643858909607,
      "learning_rate": 8.522776715167548e-06,
      "loss": 0.4625,
      "num_input_tokens_seen": 7211712,
      "step": 11060
    },
    {
      "epoch": 6.524174528301887,
      "grad_norm": 2.2613625526428223,
      "learning_rate": 8.520950524799192e-06,
      "loss": 0.2418,
      "num_input_tokens_seen": 7214112,
      "step": 11065
    },
    {
      "epoch": 6.527122641509434,
      "grad_norm": 2.9873292446136475,
      "learning_rate": 8.51912340224401e-06,
      "loss": 0.5313,
      "num_input_tokens_seen": 7219616,
      "step": 11070
    },
    {
      "epoch": 6.530070754716981,
      "grad_norm": 3.4120967388153076,
      "learning_rate": 8.51729534798574e-06,
      "loss": 0.3658,
      "num_input_tokens_seen": 7222240,
      "step": 11075
    },
    {
      "epoch": 6.533018867924528,
      "grad_norm": 2.680600881576538,
      "learning_rate": 8.515466362508369e-06,
      "loss": 0.4608,
      "num_input_tokens_seen": 7224832,
      "step": 11080
    },
    {
      "epoch": 6.535966981132075,
      "grad_norm": 1.951891303062439,
      "learning_rate": 8.513636446296125e-06,
      "loss": 0.3542,
      "num_input_tokens_seen": 7227648,
      "step": 11085
    },
    {
      "epoch": 6.538915094339623,
      "grad_norm": 1.6225849390029907,
      "learning_rate": 8.51180559983349e-06,
      "loss": 0.4697,
      "num_input_tokens_seen": 7231328,
      "step": 11090
    },
    {
      "epoch": 6.54186320754717,
      "grad_norm": 2.1897928714752197,
      "learning_rate": 8.50997382360519e-06,
      "loss": 0.4618,
      "num_input_tokens_seen": 7234592,
      "step": 11095
    },
    {
      "epoch": 6.544811320754717,
      "grad_norm": 2.547898054122925,
      "learning_rate": 8.508141118096191e-06,
      "loss": 0.5463,
      "num_input_tokens_seen": 7237600,
      "step": 11100
    },
    {
      "epoch": 6.5477594339622645,
      "grad_norm": 1.795732021331787,
      "learning_rate": 8.506307483791712e-06,
      "loss": 0.3954,
      "num_input_tokens_seen": 7240416,
      "step": 11105
    },
    {
      "epoch": 6.550707547169811,
      "grad_norm": 1.790718913078308,
      "learning_rate": 8.504472921177215e-06,
      "loss": 0.3508,
      "num_input_tokens_seen": 7243104,
      "step": 11110
    },
    {
      "epoch": 6.553655660377358,
      "grad_norm": 3.0685505867004395,
      "learning_rate": 8.502637430738409e-06,
      "loss": 0.5178,
      "num_input_tokens_seen": 7246336,
      "step": 11115
    },
    {
      "epoch": 6.556603773584905,
      "grad_norm": 2.019568681716919,
      "learning_rate": 8.500801012961248e-06,
      "loss": 0.4293,
      "num_input_tokens_seen": 7249152,
      "step": 11120
    },
    {
      "epoch": 6.559551886792453,
      "grad_norm": 1.8914834260940552,
      "learning_rate": 8.49896366833193e-06,
      "loss": 0.4648,
      "num_input_tokens_seen": 7252192,
      "step": 11125
    },
    {
      "epoch": 6.5625,
      "grad_norm": 1.3950716257095337,
      "learning_rate": 8.497125397336903e-06,
      "loss": 0.3835,
      "num_input_tokens_seen": 7255232,
      "step": 11130
    },
    {
      "epoch": 6.565448113207547,
      "grad_norm": 4.4465250968933105,
      "learning_rate": 8.495286200462854e-06,
      "loss": 0.4251,
      "num_input_tokens_seen": 7258016,
      "step": 11135
    },
    {
      "epoch": 6.568396226415095,
      "grad_norm": 1.8653334379196167,
      "learning_rate": 8.49344607819672e-06,
      "loss": 0.3453,
      "num_input_tokens_seen": 7260384,
      "step": 11140
    },
    {
      "epoch": 6.571344339622642,
      "grad_norm": 2.5832009315490723,
      "learning_rate": 8.49160503102568e-06,
      "loss": 0.5215,
      "num_input_tokens_seen": 7263264,
      "step": 11145
    },
    {
      "epoch": 6.574292452830189,
      "grad_norm": 2.406696319580078,
      "learning_rate": 8.489763059437161e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 7266464,
      "step": 11150
    },
    {
      "epoch": 6.5772405660377355,
      "grad_norm": 2.7078311443328857,
      "learning_rate": 8.487920163918833e-06,
      "loss": 0.3774,
      "num_input_tokens_seen": 7269472,
      "step": 11155
    },
    {
      "epoch": 6.580188679245283,
      "grad_norm": 1.8479723930358887,
      "learning_rate": 8.486076344958607e-06,
      "loss": 0.4179,
      "num_input_tokens_seen": 7273632,
      "step": 11160
    },
    {
      "epoch": 6.58313679245283,
      "grad_norm": 4.485570907592773,
      "learning_rate": 8.484231603044647e-06,
      "loss": 0.4313,
      "num_input_tokens_seen": 7277344,
      "step": 11165
    },
    {
      "epoch": 6.586084905660377,
      "grad_norm": 2.016378402709961,
      "learning_rate": 8.482385938665352e-06,
      "loss": 0.283,
      "num_input_tokens_seen": 7280128,
      "step": 11170
    },
    {
      "epoch": 6.589033018867925,
      "grad_norm": 1.260894536972046,
      "learning_rate": 8.480539352309373e-06,
      "loss": 0.4056,
      "num_input_tokens_seen": 7283648,
      "step": 11175
    },
    {
      "epoch": 6.591981132075472,
      "grad_norm": 3.133361339569092,
      "learning_rate": 8.478691844465598e-06,
      "loss": 0.4323,
      "num_input_tokens_seen": 7286272,
      "step": 11180
    },
    {
      "epoch": 6.594929245283019,
      "grad_norm": 4.142938137054443,
      "learning_rate": 8.476843415623168e-06,
      "loss": 0.3752,
      "num_input_tokens_seen": 7289568,
      "step": 11185
    },
    {
      "epoch": 6.597877358490566,
      "grad_norm": 2.6723999977111816,
      "learning_rate": 8.474994066271458e-06,
      "loss": 0.5469,
      "num_input_tokens_seen": 7293696,
      "step": 11190
    },
    {
      "epoch": 6.600825471698113,
      "grad_norm": 3.1884076595306396,
      "learning_rate": 8.473143796900089e-06,
      "loss": 0.5029,
      "num_input_tokens_seen": 7296160,
      "step": 11195
    },
    {
      "epoch": 6.60377358490566,
      "grad_norm": 3.000969171524048,
      "learning_rate": 8.471292607998936e-06,
      "loss": 0.5805,
      "num_input_tokens_seen": 7299936,
      "step": 11200
    },
    {
      "epoch": 6.6067216981132075,
      "grad_norm": 1.8377585411071777,
      "learning_rate": 8.469440500058104e-06,
      "loss": 0.4418,
      "num_input_tokens_seen": 7305664,
      "step": 11205
    },
    {
      "epoch": 6.609669811320755,
      "grad_norm": 2.339547872543335,
      "learning_rate": 8.467587473567945e-06,
      "loss": 0.6698,
      "num_input_tokens_seen": 7309472,
      "step": 11210
    },
    {
      "epoch": 6.612617924528302,
      "grad_norm": 1.1318994760513306,
      "learning_rate": 8.46573352901906e-06,
      "loss": 0.342,
      "num_input_tokens_seen": 7313152,
      "step": 11215
    },
    {
      "epoch": 6.615566037735849,
      "grad_norm": 4.024058818817139,
      "learning_rate": 8.463878666902286e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 7318112,
      "step": 11220
    },
    {
      "epoch": 6.618514150943396,
      "grad_norm": 2.114833116531372,
      "learning_rate": 8.462022887708706e-06,
      "loss": 0.3286,
      "num_input_tokens_seen": 7320800,
      "step": 11225
    },
    {
      "epoch": 6.621462264150943,
      "grad_norm": 2.0244078636169434,
      "learning_rate": 8.460166191929646e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 7324736,
      "step": 11230
    },
    {
      "epoch": 6.62441037735849,
      "grad_norm": 1.5146896839141846,
      "learning_rate": 8.458308580056675e-06,
      "loss": 0.4455,
      "num_input_tokens_seen": 7328352,
      "step": 11235
    },
    {
      "epoch": 6.627358490566038,
      "grad_norm": 1.4580230712890625,
      "learning_rate": 8.456450052581602e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 7332064,
      "step": 11240
    },
    {
      "epoch": 6.630306603773585,
      "grad_norm": 1.9581786394119263,
      "learning_rate": 8.45459060999648e-06,
      "loss": 0.5028,
      "num_input_tokens_seen": 7336224,
      "step": 11245
    },
    {
      "epoch": 6.633254716981132,
      "grad_norm": 2.674583673477173,
      "learning_rate": 8.452730252793608e-06,
      "loss": 0.5647,
      "num_input_tokens_seen": 7339424,
      "step": 11250
    },
    {
      "epoch": 6.6362028301886795,
      "grad_norm": 1.0306998491287231,
      "learning_rate": 8.450868981465519e-06,
      "loss": 0.3141,
      "num_input_tokens_seen": 7342912,
      "step": 11255
    },
    {
      "epoch": 6.639150943396227,
      "grad_norm": 1.9173943996429443,
      "learning_rate": 8.449006796504997e-06,
      "loss": 0.3394,
      "num_input_tokens_seen": 7345216,
      "step": 11260
    },
    {
      "epoch": 6.642099056603773,
      "grad_norm": 1.431261658668518,
      "learning_rate": 8.44714369840506e-06,
      "loss": 0.4621,
      "num_input_tokens_seen": 7348736,
      "step": 11265
    },
    {
      "epoch": 6.6450471698113205,
      "grad_norm": 1.7577106952667236,
      "learning_rate": 8.445279687658973e-06,
      "loss": 0.3726,
      "num_input_tokens_seen": 7351616,
      "step": 11270
    },
    {
      "epoch": 6.647995283018868,
      "grad_norm": 1.814050555229187,
      "learning_rate": 8.44341476476024e-06,
      "loss": 0.3986,
      "num_input_tokens_seen": 7359168,
      "step": 11275
    },
    {
      "epoch": 6.650943396226415,
      "grad_norm": 1.8595256805419922,
      "learning_rate": 8.441548930202608e-06,
      "loss": 0.403,
      "num_input_tokens_seen": 7361984,
      "step": 11280
    },
    {
      "epoch": 6.653891509433962,
      "grad_norm": 1.3324761390686035,
      "learning_rate": 8.439682184480065e-06,
      "loss": 0.4786,
      "num_input_tokens_seen": 7365216,
      "step": 11285
    },
    {
      "epoch": 6.65683962264151,
      "grad_norm": 4.095402240753174,
      "learning_rate": 8.437814528086837e-06,
      "loss": 0.5055,
      "num_input_tokens_seen": 7368512,
      "step": 11290
    },
    {
      "epoch": 6.659787735849057,
      "grad_norm": 1.4543209075927734,
      "learning_rate": 8.435945961517398e-06,
      "loss": 0.4474,
      "num_input_tokens_seen": 7371040,
      "step": 11295
    },
    {
      "epoch": 6.662735849056604,
      "grad_norm": 2.130070686340332,
      "learning_rate": 8.434076485266458e-06,
      "loss": 0.4074,
      "num_input_tokens_seen": 7374496,
      "step": 11300
    },
    {
      "epoch": 6.665683962264151,
      "grad_norm": 2.02514386177063,
      "learning_rate": 8.432206099828969e-06,
      "loss": 0.5354,
      "num_input_tokens_seen": 7378112,
      "step": 11305
    },
    {
      "epoch": 6.668632075471698,
      "grad_norm": 1.762580394744873,
      "learning_rate": 8.430334805700122e-06,
      "loss": 0.3799,
      "num_input_tokens_seen": 7381792,
      "step": 11310
    },
    {
      "epoch": 6.671580188679245,
      "grad_norm": 1.644616961479187,
      "learning_rate": 8.428462603375351e-06,
      "loss": 0.5497,
      "num_input_tokens_seen": 7384384,
      "step": 11315
    },
    {
      "epoch": 6.6745283018867925,
      "grad_norm": 1.644087314605713,
      "learning_rate": 8.426589493350332e-06,
      "loss": 0.5424,
      "num_input_tokens_seen": 7388032,
      "step": 11320
    },
    {
      "epoch": 6.67747641509434,
      "grad_norm": 3.4967098236083984,
      "learning_rate": 8.424715476120976e-06,
      "loss": 0.6295,
      "num_input_tokens_seen": 7391136,
      "step": 11325
    },
    {
      "epoch": 6.680424528301887,
      "grad_norm": 3.3440873622894287,
      "learning_rate": 8.422840552183437e-06,
      "loss": 0.5037,
      "num_input_tokens_seen": 7394560,
      "step": 11330
    },
    {
      "epoch": 6.683372641509434,
      "grad_norm": 1.3333353996276855,
      "learning_rate": 8.420964722034111e-06,
      "loss": 0.3815,
      "num_input_tokens_seen": 7398208,
      "step": 11335
    },
    {
      "epoch": 6.686320754716981,
      "grad_norm": 1.8587840795516968,
      "learning_rate": 8.41908798616963e-06,
      "loss": 0.4193,
      "num_input_tokens_seen": 7401600,
      "step": 11340
    },
    {
      "epoch": 6.689268867924528,
      "grad_norm": 2.1222939491271973,
      "learning_rate": 8.41721034508687e-06,
      "loss": 0.3391,
      "num_input_tokens_seen": 7405504,
      "step": 11345
    },
    {
      "epoch": 6.692216981132075,
      "grad_norm": 2.993468761444092,
      "learning_rate": 8.415331799282942e-06,
      "loss": 0.3748,
      "num_input_tokens_seen": 7408192,
      "step": 11350
    },
    {
      "epoch": 6.695165094339623,
      "grad_norm": 1.5288426876068115,
      "learning_rate": 8.413452349255205e-06,
      "loss": 0.4104,
      "num_input_tokens_seen": 7410752,
      "step": 11355
    },
    {
      "epoch": 6.69811320754717,
      "grad_norm": 2.077700138092041,
      "learning_rate": 8.411571995501245e-06,
      "loss": 0.4477,
      "num_input_tokens_seen": 7413376,
      "step": 11360
    },
    {
      "epoch": 6.701061320754717,
      "grad_norm": 1.739641785621643,
      "learning_rate": 8.409690738518895e-06,
      "loss": 0.5333,
      "num_input_tokens_seen": 7417344,
      "step": 11365
    },
    {
      "epoch": 6.7040094339622645,
      "grad_norm": 2.0545406341552734,
      "learning_rate": 8.407808578806229e-06,
      "loss": 0.3967,
      "num_input_tokens_seen": 7420608,
      "step": 11370
    },
    {
      "epoch": 6.706957547169811,
      "grad_norm": 1.7327436208724976,
      "learning_rate": 8.405925516861555e-06,
      "loss": 0.3422,
      "num_input_tokens_seen": 7423840,
      "step": 11375
    },
    {
      "epoch": 6.709905660377358,
      "grad_norm": 1.6333808898925781,
      "learning_rate": 8.40404155318342e-06,
      "loss": 0.4422,
      "num_input_tokens_seen": 7427840,
      "step": 11380
    },
    {
      "epoch": 6.712853773584905,
      "grad_norm": 2.378512144088745,
      "learning_rate": 8.402156688270613e-06,
      "loss": 0.5032,
      "num_input_tokens_seen": 7431328,
      "step": 11385
    },
    {
      "epoch": 6.715801886792453,
      "grad_norm": 1.7582159042358398,
      "learning_rate": 8.400270922622162e-06,
      "loss": 0.4468,
      "num_input_tokens_seen": 7435296,
      "step": 11390
    },
    {
      "epoch": 6.71875,
      "grad_norm": 2.4809610843658447,
      "learning_rate": 8.398384256737328e-06,
      "loss": 0.4916,
      "num_input_tokens_seen": 7438048,
      "step": 11395
    },
    {
      "epoch": 6.721698113207547,
      "grad_norm": 2.156104326248169,
      "learning_rate": 8.396496691115619e-06,
      "loss": 0.6071,
      "num_input_tokens_seen": 7441088,
      "step": 11400
    },
    {
      "epoch": 6.724646226415095,
      "grad_norm": 3.0070583820343018,
      "learning_rate": 8.39460822625677e-06,
      "loss": 0.4441,
      "num_input_tokens_seen": 7443360,
      "step": 11405
    },
    {
      "epoch": 6.727594339622642,
      "grad_norm": 0.8194558620452881,
      "learning_rate": 8.392718862660765e-06,
      "loss": 0.3033,
      "num_input_tokens_seen": 7447616,
      "step": 11410
    },
    {
      "epoch": 6.730542452830189,
      "grad_norm": 2.067647933959961,
      "learning_rate": 8.390828600827818e-06,
      "loss": 0.4327,
      "num_input_tokens_seen": 7450240,
      "step": 11415
    },
    {
      "epoch": 6.7334905660377355,
      "grad_norm": 1.7567627429962158,
      "learning_rate": 8.388937441258385e-06,
      "loss": 0.516,
      "num_input_tokens_seen": 7453664,
      "step": 11420
    },
    {
      "epoch": 6.736438679245283,
      "grad_norm": 2.1843059062957764,
      "learning_rate": 8.387045384453162e-06,
      "loss": 0.4303,
      "num_input_tokens_seen": 7456800,
      "step": 11425
    },
    {
      "epoch": 6.73938679245283,
      "grad_norm": 2.202421188354492,
      "learning_rate": 8.385152430913073e-06,
      "loss": 0.4387,
      "num_input_tokens_seen": 7459648,
      "step": 11430
    },
    {
      "epoch": 6.742334905660377,
      "grad_norm": 4.489969730377197,
      "learning_rate": 8.383258581139288e-06,
      "loss": 0.4903,
      "num_input_tokens_seen": 7463424,
      "step": 11435
    },
    {
      "epoch": 6.745283018867925,
      "grad_norm": 1.502013921737671,
      "learning_rate": 8.381363835633213e-06,
      "loss": 0.451,
      "num_input_tokens_seen": 7466144,
      "step": 11440
    },
    {
      "epoch": 6.748231132075472,
      "grad_norm": 1.7006384134292603,
      "learning_rate": 8.379468194896492e-06,
      "loss": 0.4346,
      "num_input_tokens_seen": 7468992,
      "step": 11445
    },
    {
      "epoch": 6.751179245283019,
      "grad_norm": 6.214755058288574,
      "learning_rate": 8.377571659431e-06,
      "loss": 0.5312,
      "num_input_tokens_seen": 7472288,
      "step": 11450
    },
    {
      "epoch": 6.754127358490566,
      "grad_norm": 3.390658140182495,
      "learning_rate": 8.375674229738855e-06,
      "loss": 0.4775,
      "num_input_tokens_seen": 7475648,
      "step": 11455
    },
    {
      "epoch": 6.757075471698113,
      "grad_norm": 2.702378988265991,
      "learning_rate": 8.37377590632241e-06,
      "loss": 0.5368,
      "num_input_tokens_seen": 7479712,
      "step": 11460
    },
    {
      "epoch": 6.76002358490566,
      "grad_norm": 3.6077725887298584,
      "learning_rate": 8.371876689684253e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 7482688,
      "step": 11465
    },
    {
      "epoch": 6.7629716981132075,
      "grad_norm": 1.4233475923538208,
      "learning_rate": 8.369976580327211e-06,
      "loss": 0.3797,
      "num_input_tokens_seen": 7486176,
      "step": 11470
    },
    {
      "epoch": 6.765919811320755,
      "grad_norm": 2.197141408920288,
      "learning_rate": 8.368075578754345e-06,
      "loss": 0.4165,
      "num_input_tokens_seen": 7488384,
      "step": 11475
    },
    {
      "epoch": 6.768867924528302,
      "grad_norm": 1.0275706052780151,
      "learning_rate": 8.366173685468952e-06,
      "loss": 0.4835,
      "num_input_tokens_seen": 7491680,
      "step": 11480
    },
    {
      "epoch": 6.771816037735849,
      "grad_norm": 1.7352893352508545,
      "learning_rate": 8.364270900974572e-06,
      "loss": 0.3695,
      "num_input_tokens_seen": 7494592,
      "step": 11485
    },
    {
      "epoch": 6.774764150943396,
      "grad_norm": 3.045090913772583,
      "learning_rate": 8.362367225774968e-06,
      "loss": 0.4804,
      "num_input_tokens_seen": 7497632,
      "step": 11490
    },
    {
      "epoch": 6.777712264150943,
      "grad_norm": 1.319770097732544,
      "learning_rate": 8.360462660374153e-06,
      "loss": 0.4748,
      "num_input_tokens_seen": 7500096,
      "step": 11495
    },
    {
      "epoch": 6.78066037735849,
      "grad_norm": 1.8948845863342285,
      "learning_rate": 8.358557205276365e-06,
      "loss": 0.4511,
      "num_input_tokens_seen": 7503392,
      "step": 11500
    },
    {
      "epoch": 6.783608490566038,
      "grad_norm": 1.3515570163726807,
      "learning_rate": 8.356650860986083e-06,
      "loss": 0.3506,
      "num_input_tokens_seen": 7506304,
      "step": 11505
    },
    {
      "epoch": 6.786556603773585,
      "grad_norm": 1.884287714958191,
      "learning_rate": 8.354743628008017e-06,
      "loss": 0.4215,
      "num_input_tokens_seen": 7509376,
      "step": 11510
    },
    {
      "epoch": 6.789504716981132,
      "grad_norm": 2.448120594024658,
      "learning_rate": 8.35283550684712e-06,
      "loss": 0.547,
      "num_input_tokens_seen": 7512480,
      "step": 11515
    },
    {
      "epoch": 6.7924528301886795,
      "grad_norm": 3.4300365447998047,
      "learning_rate": 8.350926498008572e-06,
      "loss": 0.586,
      "num_input_tokens_seen": 7515552,
      "step": 11520
    },
    {
      "epoch": 6.795400943396227,
      "grad_norm": 2.6889214515686035,
      "learning_rate": 8.34901660199779e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 7518656,
      "step": 11525
    },
    {
      "epoch": 6.798349056603773,
      "grad_norm": 1.9622069597244263,
      "learning_rate": 8.347105819320432e-06,
      "loss": 0.4835,
      "num_input_tokens_seen": 7521952,
      "step": 11530
    },
    {
      "epoch": 6.8012971698113205,
      "grad_norm": 2.5863547325134277,
      "learning_rate": 8.34519415048238e-06,
      "loss": 0.395,
      "num_input_tokens_seen": 7525376,
      "step": 11535
    },
    {
      "epoch": 6.804245283018868,
      "grad_norm": 2.9789795875549316,
      "learning_rate": 8.343281595989761e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 7528128,
      "step": 11540
    },
    {
      "epoch": 6.807193396226415,
      "grad_norm": 1.7870593070983887,
      "learning_rate": 8.341368156348933e-06,
      "loss": 0.4204,
      "num_input_tokens_seen": 7531328,
      "step": 11545
    },
    {
      "epoch": 6.810141509433962,
      "grad_norm": 8.119572639465332,
      "learning_rate": 8.339453832066482e-06,
      "loss": 0.4761,
      "num_input_tokens_seen": 7533792,
      "step": 11550
    },
    {
      "epoch": 6.81308962264151,
      "grad_norm": 2.8290703296661377,
      "learning_rate": 8.337538623649237e-06,
      "loss": 0.3677,
      "num_input_tokens_seen": 7537216,
      "step": 11555
    },
    {
      "epoch": 6.816037735849057,
      "grad_norm": 1.4916973114013672,
      "learning_rate": 8.33562253160426e-06,
      "loss": 0.5086,
      "num_input_tokens_seen": 7539712,
      "step": 11560
    },
    {
      "epoch": 6.818985849056604,
      "grad_norm": 3.5682928562164307,
      "learning_rate": 8.33370555643884e-06,
      "loss": 0.5623,
      "num_input_tokens_seen": 7542848,
      "step": 11565
    },
    {
      "epoch": 6.821933962264151,
      "grad_norm": 2.3088340759277344,
      "learning_rate": 8.331787698660507e-06,
      "loss": 0.3821,
      "num_input_tokens_seen": 7545888,
      "step": 11570
    },
    {
      "epoch": 6.824882075471698,
      "grad_norm": 1.5998623371124268,
      "learning_rate": 8.32986895877702e-06,
      "loss": 0.5281,
      "num_input_tokens_seen": 7548864,
      "step": 11575
    },
    {
      "epoch": 6.827830188679245,
      "grad_norm": 2.7926735877990723,
      "learning_rate": 8.327949337296378e-06,
      "loss": 0.3915,
      "num_input_tokens_seen": 7551232,
      "step": 11580
    },
    {
      "epoch": 6.8307783018867925,
      "grad_norm": 2.8393094539642334,
      "learning_rate": 8.326028834726803e-06,
      "loss": 0.4711,
      "num_input_tokens_seen": 7555040,
      "step": 11585
    },
    {
      "epoch": 6.83372641509434,
      "grad_norm": 2.076556921005249,
      "learning_rate": 8.324107451576762e-06,
      "loss": 0.4952,
      "num_input_tokens_seen": 7557664,
      "step": 11590
    },
    {
      "epoch": 6.836674528301887,
      "grad_norm": 1.7905797958374023,
      "learning_rate": 8.322185188354947e-06,
      "loss": 0.4391,
      "num_input_tokens_seen": 7561856,
      "step": 11595
    },
    {
      "epoch": 6.839622641509434,
      "grad_norm": 1.6176079511642456,
      "learning_rate": 8.320262045570284e-06,
      "loss": 0.3429,
      "num_input_tokens_seen": 7565248,
      "step": 11600
    },
    {
      "epoch": 6.842570754716981,
      "grad_norm": 2.686922311782837,
      "learning_rate": 8.318338023731937e-06,
      "loss": 0.3578,
      "num_input_tokens_seen": 7569536,
      "step": 11605
    },
    {
      "epoch": 6.845518867924528,
      "grad_norm": 2.6717967987060547,
      "learning_rate": 8.316413123349296e-06,
      "loss": 0.415,
      "num_input_tokens_seen": 7572288,
      "step": 11610
    },
    {
      "epoch": 6.848466981132075,
      "grad_norm": 1.9239338636398315,
      "learning_rate": 8.314487344931987e-06,
      "loss": 0.4335,
      "num_input_tokens_seen": 7575520,
      "step": 11615
    },
    {
      "epoch": 6.851415094339623,
      "grad_norm": 2.557251453399658,
      "learning_rate": 8.31256068898987e-06,
      "loss": 0.4561,
      "num_input_tokens_seen": 7579264,
      "step": 11620
    },
    {
      "epoch": 6.85436320754717,
      "grad_norm": 2.58467435836792,
      "learning_rate": 8.310633156033032e-06,
      "loss": 0.3629,
      "num_input_tokens_seen": 7582496,
      "step": 11625
    },
    {
      "epoch": 6.857311320754717,
      "grad_norm": 1.9937858581542969,
      "learning_rate": 8.3087047465718e-06,
      "loss": 0.4917,
      "num_input_tokens_seen": 7586144,
      "step": 11630
    },
    {
      "epoch": 6.8602594339622645,
      "grad_norm": 1.9986299276351929,
      "learning_rate": 8.306775461116727e-06,
      "loss": 0.4278,
      "num_input_tokens_seen": 7589184,
      "step": 11635
    },
    {
      "epoch": 6.863207547169811,
      "grad_norm": 1.6567435264587402,
      "learning_rate": 8.304845300178597e-06,
      "loss": 0.3716,
      "num_input_tokens_seen": 7592480,
      "step": 11640
    },
    {
      "epoch": 6.866155660377358,
      "grad_norm": 1.3341597318649292,
      "learning_rate": 8.302914264268433e-06,
      "loss": 0.4902,
      "num_input_tokens_seen": 7595584,
      "step": 11645
    },
    {
      "epoch": 6.869103773584905,
      "grad_norm": 1.602872371673584,
      "learning_rate": 8.300982353897482e-06,
      "loss": 0.3334,
      "num_input_tokens_seen": 7599008,
      "step": 11650
    },
    {
      "epoch": 6.872051886792453,
      "grad_norm": 2.1271252632141113,
      "learning_rate": 8.299049569577226e-06,
      "loss": 0.4292,
      "num_input_tokens_seen": 7601728,
      "step": 11655
    },
    {
      "epoch": 6.875,
      "grad_norm": 3.870276927947998,
      "learning_rate": 8.297115911819379e-06,
      "loss": 0.4201,
      "num_input_tokens_seen": 7604288,
      "step": 11660
    },
    {
      "epoch": 6.877948113207547,
      "grad_norm": 2.3912394046783447,
      "learning_rate": 8.295181381135884e-06,
      "loss": 0.4505,
      "num_input_tokens_seen": 7607648,
      "step": 11665
    },
    {
      "epoch": 6.880896226415095,
      "grad_norm": 1.5460768938064575,
      "learning_rate": 8.293245978038917e-06,
      "loss": 0.376,
      "num_input_tokens_seen": 7611360,
      "step": 11670
    },
    {
      "epoch": 6.883844339622642,
      "grad_norm": 2.4147145748138428,
      "learning_rate": 8.291309703040884e-06,
      "loss": 0.3271,
      "num_input_tokens_seen": 7617664,
      "step": 11675
    },
    {
      "epoch": 6.886792452830189,
      "grad_norm": 1.2025635242462158,
      "learning_rate": 8.289372556654422e-06,
      "loss": 0.4598,
      "num_input_tokens_seen": 7620480,
      "step": 11680
    },
    {
      "epoch": 6.8897405660377355,
      "grad_norm": 1.700533390045166,
      "learning_rate": 8.287434539392401e-06,
      "loss": 0.4873,
      "num_input_tokens_seen": 7624352,
      "step": 11685
    },
    {
      "epoch": 6.892688679245283,
      "grad_norm": 4.3826680183410645,
      "learning_rate": 8.285495651767916e-06,
      "loss": 0.4431,
      "num_input_tokens_seen": 7627296,
      "step": 11690
    },
    {
      "epoch": 6.89563679245283,
      "grad_norm": 2.450436592102051,
      "learning_rate": 8.283555894294297e-06,
      "loss": 0.4,
      "num_input_tokens_seen": 7630816,
      "step": 11695
    },
    {
      "epoch": 6.898584905660377,
      "grad_norm": 1.6770349740982056,
      "learning_rate": 8.281615267485105e-06,
      "loss": 0.5964,
      "num_input_tokens_seen": 7633920,
      "step": 11700
    },
    {
      "epoch": 6.901533018867925,
      "grad_norm": 1.8706599473953247,
      "learning_rate": 8.279673771854127e-06,
      "loss": 0.4768,
      "num_input_tokens_seen": 7637920,
      "step": 11705
    },
    {
      "epoch": 6.904481132075472,
      "grad_norm": 3.2704851627349854,
      "learning_rate": 8.277731407915386e-06,
      "loss": 0.541,
      "num_input_tokens_seen": 7640800,
      "step": 11710
    },
    {
      "epoch": 6.907429245283019,
      "grad_norm": 3.3839197158813477,
      "learning_rate": 8.275788176183126e-06,
      "loss": 0.3586,
      "num_input_tokens_seen": 7644480,
      "step": 11715
    },
    {
      "epoch": 6.910377358490566,
      "grad_norm": 1.4205790758132935,
      "learning_rate": 8.273844077171827e-06,
      "loss": 0.4119,
      "num_input_tokens_seen": 7647936,
      "step": 11720
    },
    {
      "epoch": 6.913325471698113,
      "grad_norm": 1.471185326576233,
      "learning_rate": 8.271899111396202e-06,
      "loss": 0.4406,
      "num_input_tokens_seen": 7651264,
      "step": 11725
    },
    {
      "epoch": 6.91627358490566,
      "grad_norm": 1.936965823173523,
      "learning_rate": 8.269953279371185e-06,
      "loss": 0.492,
      "num_input_tokens_seen": 7654496,
      "step": 11730
    },
    {
      "epoch": 6.9192216981132075,
      "grad_norm": 1.8686084747314453,
      "learning_rate": 8.268006581611945e-06,
      "loss": 0.4332,
      "num_input_tokens_seen": 7657856,
      "step": 11735
    },
    {
      "epoch": 6.922169811320755,
      "grad_norm": 2.4149816036224365,
      "learning_rate": 8.266059018633878e-06,
      "loss": 0.4429,
      "num_input_tokens_seen": 7660192,
      "step": 11740
    },
    {
      "epoch": 6.925117924528302,
      "grad_norm": 1.4742753505706787,
      "learning_rate": 8.264110590952609e-06,
      "loss": 0.3879,
      "num_input_tokens_seen": 7663904,
      "step": 11745
    },
    {
      "epoch": 6.928066037735849,
      "grad_norm": 1.7058687210083008,
      "learning_rate": 8.262161299083993e-06,
      "loss": 0.3115,
      "num_input_tokens_seen": 7666688,
      "step": 11750
    },
    {
      "epoch": 6.931014150943396,
      "grad_norm": 1.8989156484603882,
      "learning_rate": 8.260211143544117e-06,
      "loss": 0.3835,
      "num_input_tokens_seen": 7670080,
      "step": 11755
    },
    {
      "epoch": 6.933962264150943,
      "grad_norm": 2.674440860748291,
      "learning_rate": 8.258260124849288e-06,
      "loss": 0.4932,
      "num_input_tokens_seen": 7672768,
      "step": 11760
    },
    {
      "epoch": 6.93691037735849,
      "grad_norm": 1.8931361436843872,
      "learning_rate": 8.256308243516048e-06,
      "loss": 0.3607,
      "num_input_tokens_seen": 7675744,
      "step": 11765
    },
    {
      "epoch": 6.939858490566038,
      "grad_norm": 1.975437879562378,
      "learning_rate": 8.254355500061168e-06,
      "loss": 0.3709,
      "num_input_tokens_seen": 7678400,
      "step": 11770
    },
    {
      "epoch": 6.942806603773585,
      "grad_norm": 2.361424446105957,
      "learning_rate": 8.252401895001643e-06,
      "loss": 0.6518,
      "num_input_tokens_seen": 7681760,
      "step": 11775
    },
    {
      "epoch": 6.945754716981132,
      "grad_norm": 1.490308165550232,
      "learning_rate": 8.2504474288547e-06,
      "loss": 0.5036,
      "num_input_tokens_seen": 7684480,
      "step": 11780
    },
    {
      "epoch": 6.9487028301886795,
      "grad_norm": 1.8954814672470093,
      "learning_rate": 8.248492102137791e-06,
      "loss": 0.4585,
      "num_input_tokens_seen": 7686816,
      "step": 11785
    },
    {
      "epoch": 6.951650943396227,
      "grad_norm": 1.5862764120101929,
      "learning_rate": 8.246535915368596e-06,
      "loss": 0.4664,
      "num_input_tokens_seen": 7691072,
      "step": 11790
    },
    {
      "epoch": 6.954599056603773,
      "grad_norm": 1.515276312828064,
      "learning_rate": 8.24457886906503e-06,
      "loss": 0.3612,
      "num_input_tokens_seen": 7695168,
      "step": 11795
    },
    {
      "epoch": 6.9575471698113205,
      "grad_norm": 2.1550660133361816,
      "learning_rate": 8.242620963745222e-06,
      "loss": 0.3018,
      "num_input_tokens_seen": 7698560,
      "step": 11800
    },
    {
      "epoch": 6.960495283018868,
      "grad_norm": 2.737398386001587,
      "learning_rate": 8.240662199927538e-06,
      "loss": 0.4631,
      "num_input_tokens_seen": 7702016,
      "step": 11805
    },
    {
      "epoch": 6.963443396226415,
      "grad_norm": 3.3583672046661377,
      "learning_rate": 8.238702578130573e-06,
      "loss": 0.4655,
      "num_input_tokens_seen": 7705376,
      "step": 11810
    },
    {
      "epoch": 6.966391509433962,
      "grad_norm": 2.2687034606933594,
      "learning_rate": 8.23674209887314e-06,
      "loss": 0.6098,
      "num_input_tokens_seen": 7708128,
      "step": 11815
    },
    {
      "epoch": 6.96933962264151,
      "grad_norm": 1.68133544921875,
      "learning_rate": 8.234780762674288e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 7711776,
      "step": 11820
    },
    {
      "epoch": 6.972287735849057,
      "grad_norm": 1.64499032497406,
      "learning_rate": 8.232818570053286e-06,
      "loss": 0.4944,
      "num_input_tokens_seen": 7714976,
      "step": 11825
    },
    {
      "epoch": 6.975235849056604,
      "grad_norm": 2.100989818572998,
      "learning_rate": 8.230855521529637e-06,
      "loss": 0.483,
      "num_input_tokens_seen": 7718432,
      "step": 11830
    },
    {
      "epoch": 6.978183962264151,
      "grad_norm": 1.3968454599380493,
      "learning_rate": 8.228891617623064e-06,
      "loss": 0.6142,
      "num_input_tokens_seen": 7720640,
      "step": 11835
    },
    {
      "epoch": 6.981132075471698,
      "grad_norm": 2.4207065105438232,
      "learning_rate": 8.22692685885352e-06,
      "loss": 0.4694,
      "num_input_tokens_seen": 7723232,
      "step": 11840
    },
    {
      "epoch": 6.984080188679245,
      "grad_norm": 2.4904093742370605,
      "learning_rate": 8.224961245741183e-06,
      "loss": 0.3445,
      "num_input_tokens_seen": 7727712,
      "step": 11845
    },
    {
      "epoch": 6.9870283018867925,
      "grad_norm": 1.526089072227478,
      "learning_rate": 8.222994778806457e-06,
      "loss": 0.4319,
      "num_input_tokens_seen": 7730816,
      "step": 11850
    },
    {
      "epoch": 6.98997641509434,
      "grad_norm": 2.62384295463562,
      "learning_rate": 8.221027458569972e-06,
      "loss": 0.511,
      "num_input_tokens_seen": 7734336,
      "step": 11855
    },
    {
      "epoch": 6.992924528301887,
      "grad_norm": 2.4000675678253174,
      "learning_rate": 8.219059285552586e-06,
      "loss": 0.4371,
      "num_input_tokens_seen": 7737504,
      "step": 11860
    },
    {
      "epoch": 6.995872641509434,
      "grad_norm": 2.3739373683929443,
      "learning_rate": 8.21709026027538e-06,
      "loss": 0.5436,
      "num_input_tokens_seen": 7741952,
      "step": 11865
    },
    {
      "epoch": 6.998820754716981,
      "grad_norm": 2.0545756816864014,
      "learning_rate": 8.215120383259664e-06,
      "loss": 0.4869,
      "num_input_tokens_seen": 7745280,
      "step": 11870
    },
    {
      "epoch": 7.001768867924528,
      "grad_norm": 1.3189345598220825,
      "learning_rate": 8.21314965502697e-06,
      "loss": 0.4837,
      "num_input_tokens_seen": 7747712,
      "step": 11875
    },
    {
      "epoch": 7.004716981132075,
      "grad_norm": 1.4737904071807861,
      "learning_rate": 8.211178076099056e-06,
      "loss": 0.3015,
      "num_input_tokens_seen": 7750848,
      "step": 11880
    },
    {
      "epoch": 7.007665094339623,
      "grad_norm": 2.4677116870880127,
      "learning_rate": 8.209205646997909e-06,
      "loss": 0.3273,
      "num_input_tokens_seen": 7754688,
      "step": 11885
    },
    {
      "epoch": 7.01061320754717,
      "grad_norm": 1.9036543369293213,
      "learning_rate": 8.207232368245735e-06,
      "loss": 0.3342,
      "num_input_tokens_seen": 7758688,
      "step": 11890
    },
    {
      "epoch": 7.013561320754717,
      "grad_norm": 1.569035291671753,
      "learning_rate": 8.205258240364968e-06,
      "loss": 0.3225,
      "num_input_tokens_seen": 7762048,
      "step": 11895
    },
    {
      "epoch": 7.0165094339622645,
      "grad_norm": 2.288790225982666,
      "learning_rate": 8.203283263878268e-06,
      "loss": 0.3935,
      "num_input_tokens_seen": 7764448,
      "step": 11900
    },
    {
      "epoch": 7.019457547169812,
      "grad_norm": 2.6840505599975586,
      "learning_rate": 8.201307439308518e-06,
      "loss": 0.3824,
      "num_input_tokens_seen": 7766816,
      "step": 11905
    },
    {
      "epoch": 7.022405660377358,
      "grad_norm": 2.3313894271850586,
      "learning_rate": 8.199330767178828e-06,
      "loss": 0.4043,
      "num_input_tokens_seen": 7770688,
      "step": 11910
    },
    {
      "epoch": 7.025353773584905,
      "grad_norm": 2.881484031677246,
      "learning_rate": 8.19735324801253e-06,
      "loss": 0.5366,
      "num_input_tokens_seen": 7773632,
      "step": 11915
    },
    {
      "epoch": 7.028301886792453,
      "grad_norm": 2.8905062675476074,
      "learning_rate": 8.195374882333178e-06,
      "loss": 0.5753,
      "num_input_tokens_seen": 7776224,
      "step": 11920
    },
    {
      "epoch": 7.03125,
      "grad_norm": 2.180483341217041,
      "learning_rate": 8.193395670664555e-06,
      "loss": 0.4906,
      "num_input_tokens_seen": 7778816,
      "step": 11925
    },
    {
      "epoch": 7.034198113207547,
      "grad_norm": 1.6871193647384644,
      "learning_rate": 8.191415613530667e-06,
      "loss": 0.4363,
      "num_input_tokens_seen": 7781056,
      "step": 11930
    },
    {
      "epoch": 7.037146226415095,
      "grad_norm": 2.294992446899414,
      "learning_rate": 8.189434711455739e-06,
      "loss": 0.4459,
      "num_input_tokens_seen": 7784064,
      "step": 11935
    },
    {
      "epoch": 7.040094339622642,
      "grad_norm": 1.1303610801696777,
      "learning_rate": 8.187452964964226e-06,
      "loss": 0.416,
      "num_input_tokens_seen": 7787616,
      "step": 11940
    },
    {
      "epoch": 7.043042452830188,
      "grad_norm": 1.7973695993423462,
      "learning_rate": 8.185470374580805e-06,
      "loss": 0.5803,
      "num_input_tokens_seen": 7791168,
      "step": 11945
    },
    {
      "epoch": 7.0459905660377355,
      "grad_norm": 2.594310998916626,
      "learning_rate": 8.183486940830371e-06,
      "loss": 0.4739,
      "num_input_tokens_seen": 7794784,
      "step": 11950
    },
    {
      "epoch": 7.048938679245283,
      "grad_norm": 1.482409119606018,
      "learning_rate": 8.18150266423805e-06,
      "loss": 0.4084,
      "num_input_tokens_seen": 7797984,
      "step": 11955
    },
    {
      "epoch": 7.05188679245283,
      "grad_norm": 1.6465197801589966,
      "learning_rate": 8.179517545329188e-06,
      "loss": 0.4017,
      "num_input_tokens_seen": 7801472,
      "step": 11960
    },
    {
      "epoch": 7.054834905660377,
      "grad_norm": 2.653254508972168,
      "learning_rate": 8.177531584629353e-06,
      "loss": 0.2901,
      "num_input_tokens_seen": 7803808,
      "step": 11965
    },
    {
      "epoch": 7.057783018867925,
      "grad_norm": 2.5850114822387695,
      "learning_rate": 8.175544782664335e-06,
      "loss": 0.4454,
      "num_input_tokens_seen": 7806528,
      "step": 11970
    },
    {
      "epoch": 7.060731132075472,
      "grad_norm": 1.5508613586425781,
      "learning_rate": 8.173557139960151e-06,
      "loss": 0.4679,
      "num_input_tokens_seen": 7809856,
      "step": 11975
    },
    {
      "epoch": 7.063679245283019,
      "grad_norm": 2.1336863040924072,
      "learning_rate": 8.17156865704304e-06,
      "loss": 0.3541,
      "num_input_tokens_seen": 7815200,
      "step": 11980
    },
    {
      "epoch": 7.066627358490566,
      "grad_norm": 2.1790597438812256,
      "learning_rate": 8.169579334439453e-06,
      "loss": 0.5896,
      "num_input_tokens_seen": 7817824,
      "step": 11985
    },
    {
      "epoch": 7.069575471698113,
      "grad_norm": 2.0576560497283936,
      "learning_rate": 8.16758917267608e-06,
      "loss": 0.3398,
      "num_input_tokens_seen": 7821856,
      "step": 11990
    },
    {
      "epoch": 7.07252358490566,
      "grad_norm": 2.7320148944854736,
      "learning_rate": 8.165598172279822e-06,
      "loss": 0.4701,
      "num_input_tokens_seen": 7824544,
      "step": 11995
    },
    {
      "epoch": 7.0754716981132075,
      "grad_norm": 3.282135009765625,
      "learning_rate": 8.163606333777804e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 7827328,
      "step": 12000
    },
    {
      "epoch": 7.078419811320755,
      "grad_norm": 1.6186277866363525,
      "learning_rate": 8.161613657697374e-06,
      "loss": 0.421,
      "num_input_tokens_seen": 7830880,
      "step": 12005
    },
    {
      "epoch": 7.081367924528302,
      "grad_norm": 2.187924861907959,
      "learning_rate": 8.159620144566103e-06,
      "loss": 0.3233,
      "num_input_tokens_seen": 7833920,
      "step": 12010
    },
    {
      "epoch": 7.084316037735849,
      "grad_norm": 2.0952911376953125,
      "learning_rate": 8.157625794911782e-06,
      "loss": 0.3889,
      "num_input_tokens_seen": 7836768,
      "step": 12015
    },
    {
      "epoch": 7.087264150943396,
      "grad_norm": 5.083712577819824,
      "learning_rate": 8.155630609262424e-06,
      "loss": 0.3213,
      "num_input_tokens_seen": 7840576,
      "step": 12020
    },
    {
      "epoch": 7.090212264150943,
      "grad_norm": 1.8995298147201538,
      "learning_rate": 8.153634588146262e-06,
      "loss": 0.371,
      "num_input_tokens_seen": 7844000,
      "step": 12025
    },
    {
      "epoch": 7.09316037735849,
      "grad_norm": 2.383364677429199,
      "learning_rate": 8.15163773209175e-06,
      "loss": 0.4403,
      "num_input_tokens_seen": 7847072,
      "step": 12030
    },
    {
      "epoch": 7.096108490566038,
      "grad_norm": 2.094858169555664,
      "learning_rate": 8.149640041627566e-06,
      "loss": 0.3444,
      "num_input_tokens_seen": 7851264,
      "step": 12035
    },
    {
      "epoch": 7.099056603773585,
      "grad_norm": 1.9967538118362427,
      "learning_rate": 8.147641517282608e-06,
      "loss": 0.5278,
      "num_input_tokens_seen": 7854656,
      "step": 12040
    },
    {
      "epoch": 7.102004716981132,
      "grad_norm": 2.886871099472046,
      "learning_rate": 8.145642159585992e-06,
      "loss": 0.5214,
      "num_input_tokens_seen": 7859520,
      "step": 12045
    },
    {
      "epoch": 7.1049528301886795,
      "grad_norm": 1.9680029153823853,
      "learning_rate": 8.143641969067057e-06,
      "loss": 0.3462,
      "num_input_tokens_seen": 7862400,
      "step": 12050
    },
    {
      "epoch": 7.107900943396227,
      "grad_norm": 2.7757222652435303,
      "learning_rate": 8.141640946255362e-06,
      "loss": 0.3447,
      "num_input_tokens_seen": 7865088,
      "step": 12055
    },
    {
      "epoch": 7.110849056603773,
      "grad_norm": 2.1096463203430176,
      "learning_rate": 8.139639091680687e-06,
      "loss": 0.3318,
      "num_input_tokens_seen": 7868736,
      "step": 12060
    },
    {
      "epoch": 7.1137971698113205,
      "grad_norm": 2.77583646774292,
      "learning_rate": 8.137636405873031e-06,
      "loss": 0.5793,
      "num_input_tokens_seen": 7871552,
      "step": 12065
    },
    {
      "epoch": 7.116745283018868,
      "grad_norm": 2.2539403438568115,
      "learning_rate": 8.135632889362614e-06,
      "loss": 0.4085,
      "num_input_tokens_seen": 7874240,
      "step": 12070
    },
    {
      "epoch": 7.119693396226415,
      "grad_norm": 3.021364212036133,
      "learning_rate": 8.133628542679879e-06,
      "loss": 0.4844,
      "num_input_tokens_seen": 7877632,
      "step": 12075
    },
    {
      "epoch": 7.122641509433962,
      "grad_norm": 4.49337100982666,
      "learning_rate": 8.131623366355478e-06,
      "loss": 0.4264,
      "num_input_tokens_seen": 7880160,
      "step": 12080
    },
    {
      "epoch": 7.12558962264151,
      "grad_norm": 1.7100971937179565,
      "learning_rate": 8.129617360920297e-06,
      "loss": 0.4005,
      "num_input_tokens_seen": 7883136,
      "step": 12085
    },
    {
      "epoch": 7.128537735849057,
      "grad_norm": 2.2806668281555176,
      "learning_rate": 8.12761052690543e-06,
      "loss": 0.2636,
      "num_input_tokens_seen": 7885408,
      "step": 12090
    },
    {
      "epoch": 7.131485849056604,
      "grad_norm": 2.138929605484009,
      "learning_rate": 8.125602864842197e-06,
      "loss": 0.4684,
      "num_input_tokens_seen": 7888224,
      "step": 12095
    },
    {
      "epoch": 7.134433962264151,
      "grad_norm": 2.133676528930664,
      "learning_rate": 8.123594375262135e-06,
      "loss": 0.4637,
      "num_input_tokens_seen": 7891808,
      "step": 12100
    },
    {
      "epoch": 7.137382075471698,
      "grad_norm": 2.332988739013672,
      "learning_rate": 8.121585058697e-06,
      "loss": 0.4083,
      "num_input_tokens_seen": 7894912,
      "step": 12105
    },
    {
      "epoch": 7.140330188679245,
      "grad_norm": 1.7388025522232056,
      "learning_rate": 8.119574915678767e-06,
      "loss": 0.5051,
      "num_input_tokens_seen": 7898560,
      "step": 12110
    },
    {
      "epoch": 7.1432783018867925,
      "grad_norm": 2.162344455718994,
      "learning_rate": 8.117563946739632e-06,
      "loss": 0.345,
      "num_input_tokens_seen": 7901248,
      "step": 12115
    },
    {
      "epoch": 7.14622641509434,
      "grad_norm": 2.1779322624206543,
      "learning_rate": 8.115552152412006e-06,
      "loss": 0.3347,
      "num_input_tokens_seen": 7905088,
      "step": 12120
    },
    {
      "epoch": 7.149174528301887,
      "grad_norm": 2.9986374378204346,
      "learning_rate": 8.11353953322852e-06,
      "loss": 0.4253,
      "num_input_tokens_seen": 7907936,
      "step": 12125
    },
    {
      "epoch": 7.152122641509434,
      "grad_norm": 5.630772590637207,
      "learning_rate": 8.111526089722024e-06,
      "loss": 0.424,
      "num_input_tokens_seen": 7910496,
      "step": 12130
    },
    {
      "epoch": 7.155070754716981,
      "grad_norm": 1.6223100423812866,
      "learning_rate": 8.109511822425586e-06,
      "loss": 0.4137,
      "num_input_tokens_seen": 7913856,
      "step": 12135
    },
    {
      "epoch": 7.158018867924528,
      "grad_norm": 2.9513845443725586,
      "learning_rate": 8.107496731872491e-06,
      "loss": 0.3229,
      "num_input_tokens_seen": 7919200,
      "step": 12140
    },
    {
      "epoch": 7.160966981132075,
      "grad_norm": 3.971353769302368,
      "learning_rate": 8.105480818596243e-06,
      "loss": 0.3888,
      "num_input_tokens_seen": 7922048,
      "step": 12145
    },
    {
      "epoch": 7.163915094339623,
      "grad_norm": 2.063004493713379,
      "learning_rate": 8.103464083130566e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 7924672,
      "step": 12150
    },
    {
      "epoch": 7.16686320754717,
      "grad_norm": 2.603506326675415,
      "learning_rate": 8.101446526009397e-06,
      "loss": 0.4345,
      "num_input_tokens_seen": 7927744,
      "step": 12155
    },
    {
      "epoch": 7.169811320754717,
      "grad_norm": 2.417618989944458,
      "learning_rate": 8.099428147766894e-06,
      "loss": 0.347,
      "num_input_tokens_seen": 7930816,
      "step": 12160
    },
    {
      "epoch": 7.1727594339622645,
      "grad_norm": 3.124481439590454,
      "learning_rate": 8.097408948937431e-06,
      "loss": 0.4496,
      "num_input_tokens_seen": 7933408,
      "step": 12165
    },
    {
      "epoch": 7.175707547169812,
      "grad_norm": 1.0909606218338013,
      "learning_rate": 8.095388930055599e-06,
      "loss": 0.3722,
      "num_input_tokens_seen": 7937216,
      "step": 12170
    },
    {
      "epoch": 7.178655660377358,
      "grad_norm": 3.3752760887145996,
      "learning_rate": 8.093368091656209e-06,
      "loss": 0.3512,
      "num_input_tokens_seen": 7940800,
      "step": 12175
    },
    {
      "epoch": 7.181603773584905,
      "grad_norm": 2.5349836349487305,
      "learning_rate": 8.091346434274284e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 7944000,
      "step": 12180
    },
    {
      "epoch": 7.184551886792453,
      "grad_norm": 2.0393357276916504,
      "learning_rate": 8.089323958445068e-06,
      "loss": 0.4007,
      "num_input_tokens_seen": 7946144,
      "step": 12185
    },
    {
      "epoch": 7.1875,
      "grad_norm": 3.878053903579712,
      "learning_rate": 8.08730066470402e-06,
      "loss": 0.4061,
      "num_input_tokens_seen": 7948416,
      "step": 12190
    },
    {
      "epoch": 7.190448113207547,
      "grad_norm": 2.6362884044647217,
      "learning_rate": 8.085276553586814e-06,
      "loss": 0.3687,
      "num_input_tokens_seen": 7950976,
      "step": 12195
    },
    {
      "epoch": 7.193396226415095,
      "grad_norm": 1.5253713130950928,
      "learning_rate": 8.083251625629345e-06,
      "loss": 0.4074,
      "num_input_tokens_seen": 7953824,
      "step": 12200
    },
    {
      "epoch": 7.196344339622642,
      "grad_norm": 3.064913272857666,
      "learning_rate": 8.08122588136772e-06,
      "loss": 0.4512,
      "num_input_tokens_seen": 7956576,
      "step": 12205
    },
    {
      "epoch": 7.199292452830188,
      "grad_norm": 2.297760248184204,
      "learning_rate": 8.079199321338262e-06,
      "loss": 0.4588,
      "num_input_tokens_seen": 7959584,
      "step": 12210
    },
    {
      "epoch": 7.2022405660377355,
      "grad_norm": 1.8918788433074951,
      "learning_rate": 8.077171946077516e-06,
      "loss": 0.411,
      "num_input_tokens_seen": 7963712,
      "step": 12215
    },
    {
      "epoch": 7.205188679245283,
      "grad_norm": 3.343000888824463,
      "learning_rate": 8.075143756122232e-06,
      "loss": 0.415,
      "num_input_tokens_seen": 7967680,
      "step": 12220
    },
    {
      "epoch": 7.20813679245283,
      "grad_norm": 2.3498241901397705,
      "learning_rate": 8.073114752009388e-06,
      "loss": 0.379,
      "num_input_tokens_seen": 7970464,
      "step": 12225
    },
    {
      "epoch": 7.211084905660377,
      "grad_norm": 2.0956473350524902,
      "learning_rate": 8.071084934276168e-06,
      "loss": 0.5107,
      "num_input_tokens_seen": 7973696,
      "step": 12230
    },
    {
      "epoch": 7.214033018867925,
      "grad_norm": 1.7145568132400513,
      "learning_rate": 8.069054303459976e-06,
      "loss": 0.3574,
      "num_input_tokens_seen": 7976352,
      "step": 12235
    },
    {
      "epoch": 7.216981132075472,
      "grad_norm": 2.117650270462036,
      "learning_rate": 8.06702286009843e-06,
      "loss": 0.5213,
      "num_input_tokens_seen": 7979712,
      "step": 12240
    },
    {
      "epoch": 7.219929245283019,
      "grad_norm": 2.0798380374908447,
      "learning_rate": 8.064990604729363e-06,
      "loss": 0.4471,
      "num_input_tokens_seen": 7982656,
      "step": 12245
    },
    {
      "epoch": 7.222877358490566,
      "grad_norm": 2.9221270084381104,
      "learning_rate": 8.062957537890827e-06,
      "loss": 0.3829,
      "num_input_tokens_seen": 7985408,
      "step": 12250
    },
    {
      "epoch": 7.225825471698113,
      "grad_norm": 2.000239372253418,
      "learning_rate": 8.060923660121081e-06,
      "loss": 0.4369,
      "num_input_tokens_seen": 7988128,
      "step": 12255
    },
    {
      "epoch": 7.22877358490566,
      "grad_norm": 2.1756668090820312,
      "learning_rate": 8.058888971958603e-06,
      "loss": 0.4046,
      "num_input_tokens_seen": 7991136,
      "step": 12260
    },
    {
      "epoch": 7.2317216981132075,
      "grad_norm": 1.5905725955963135,
      "learning_rate": 8.056853473942085e-06,
      "loss": 0.3102,
      "num_input_tokens_seen": 7994176,
      "step": 12265
    },
    {
      "epoch": 7.234669811320755,
      "grad_norm": 6.826207160949707,
      "learning_rate": 8.054817166610438e-06,
      "loss": 0.4642,
      "num_input_tokens_seen": 7997184,
      "step": 12270
    },
    {
      "epoch": 7.237617924528302,
      "grad_norm": 1.7690850496292114,
      "learning_rate": 8.052780050502781e-06,
      "loss": 0.3713,
      "num_input_tokens_seen": 8001056,
      "step": 12275
    },
    {
      "epoch": 7.240566037735849,
      "grad_norm": 2.6282012462615967,
      "learning_rate": 8.050742126158448e-06,
      "loss": 0.4258,
      "num_input_tokens_seen": 8004416,
      "step": 12280
    },
    {
      "epoch": 7.243514150943396,
      "grad_norm": 1.553707242012024,
      "learning_rate": 8.04870339411699e-06,
      "loss": 0.5149,
      "num_input_tokens_seen": 8008768,
      "step": 12285
    },
    {
      "epoch": 7.246462264150943,
      "grad_norm": 2.539104461669922,
      "learning_rate": 8.046663854918166e-06,
      "loss": 0.4366,
      "num_input_tokens_seen": 8012192,
      "step": 12290
    },
    {
      "epoch": 7.24941037735849,
      "grad_norm": 1.9790639877319336,
      "learning_rate": 8.044623509101959e-06,
      "loss": 0.3766,
      "num_input_tokens_seen": 8015616,
      "step": 12295
    },
    {
      "epoch": 7.252358490566038,
      "grad_norm": 2.757936954498291,
      "learning_rate": 8.042582357208557e-06,
      "loss": 0.3876,
      "num_input_tokens_seen": 8018752,
      "step": 12300
    },
    {
      "epoch": 7.255306603773585,
      "grad_norm": 1.4044712781906128,
      "learning_rate": 8.04054039977836e-06,
      "loss": 0.4784,
      "num_input_tokens_seen": 8022656,
      "step": 12305
    },
    {
      "epoch": 7.258254716981132,
      "grad_norm": 0.9646516442298889,
      "learning_rate": 8.038497637351992e-06,
      "loss": 0.3863,
      "num_input_tokens_seen": 8025984,
      "step": 12310
    },
    {
      "epoch": 7.2612028301886795,
      "grad_norm": 4.003399848937988,
      "learning_rate": 8.036454070470276e-06,
      "loss": 0.523,
      "num_input_tokens_seen": 8028352,
      "step": 12315
    },
    {
      "epoch": 7.264150943396227,
      "grad_norm": 2.3622994422912598,
      "learning_rate": 8.03440969967426e-06,
      "loss": 0.4538,
      "num_input_tokens_seen": 8030912,
      "step": 12320
    },
    {
      "epoch": 7.267099056603773,
      "grad_norm": 2.335052967071533,
      "learning_rate": 8.032364525505198e-06,
      "loss": 0.3874,
      "num_input_tokens_seen": 8034432,
      "step": 12325
    },
    {
      "epoch": 7.2700471698113205,
      "grad_norm": 2.0673749446868896,
      "learning_rate": 8.030318548504561e-06,
      "loss": 0.6288,
      "num_input_tokens_seen": 8037440,
      "step": 12330
    },
    {
      "epoch": 7.272995283018868,
      "grad_norm": 1.7539925575256348,
      "learning_rate": 8.028271769214026e-06,
      "loss": 0.3659,
      "num_input_tokens_seen": 8040064,
      "step": 12335
    },
    {
      "epoch": 7.275943396226415,
      "grad_norm": 2.6530513763427734,
      "learning_rate": 8.02622418817549e-06,
      "loss": 0.4535,
      "num_input_tokens_seen": 8043840,
      "step": 12340
    },
    {
      "epoch": 7.278891509433962,
      "grad_norm": 3.1688613891601562,
      "learning_rate": 8.024175805931056e-06,
      "loss": 0.5111,
      "num_input_tokens_seen": 8047392,
      "step": 12345
    },
    {
      "epoch": 7.28183962264151,
      "grad_norm": 2.2103347778320312,
      "learning_rate": 8.022126623023045e-06,
      "loss": 0.398,
      "num_input_tokens_seen": 8050080,
      "step": 12350
    },
    {
      "epoch": 7.284787735849057,
      "grad_norm": 1.5307804346084595,
      "learning_rate": 8.020076639993987e-06,
      "loss": 0.4353,
      "num_input_tokens_seen": 8053504,
      "step": 12355
    },
    {
      "epoch": 7.287735849056604,
      "grad_norm": 1.8239808082580566,
      "learning_rate": 8.018025857386622e-06,
      "loss": 0.3766,
      "num_input_tokens_seen": 8056256,
      "step": 12360
    },
    {
      "epoch": 7.290683962264151,
      "grad_norm": 1.2403656244277954,
      "learning_rate": 8.015974275743905e-06,
      "loss": 0.3381,
      "num_input_tokens_seen": 8059616,
      "step": 12365
    },
    {
      "epoch": 7.293632075471698,
      "grad_norm": 2.1340689659118652,
      "learning_rate": 8.013921895609e-06,
      "loss": 0.2365,
      "num_input_tokens_seen": 8062944,
      "step": 12370
    },
    {
      "epoch": 7.296580188679245,
      "grad_norm": 1.7445706129074097,
      "learning_rate": 8.011868717525283e-06,
      "loss": 0.3709,
      "num_input_tokens_seen": 8067008,
      "step": 12375
    },
    {
      "epoch": 7.2995283018867925,
      "grad_norm": 3.3448872566223145,
      "learning_rate": 8.009814742036343e-06,
      "loss": 0.4789,
      "num_input_tokens_seen": 8069632,
      "step": 12380
    },
    {
      "epoch": 7.30247641509434,
      "grad_norm": 2.8965771198272705,
      "learning_rate": 8.007759969685979e-06,
      "loss": 0.3749,
      "num_input_tokens_seen": 8072160,
      "step": 12385
    },
    {
      "epoch": 7.305424528301887,
      "grad_norm": 2.9302456378936768,
      "learning_rate": 8.005704401018199e-06,
      "loss": 0.4917,
      "num_input_tokens_seen": 8075488,
      "step": 12390
    },
    {
      "epoch": 7.308372641509434,
      "grad_norm": 4.608858585357666,
      "learning_rate": 8.003648036577226e-06,
      "loss": 0.4914,
      "num_input_tokens_seen": 8079040,
      "step": 12395
    },
    {
      "epoch": 7.311320754716981,
      "grad_norm": 2.5446343421936035,
      "learning_rate": 8.00159087690749e-06,
      "loss": 0.3792,
      "num_input_tokens_seen": 8081728,
      "step": 12400
    },
    {
      "epoch": 7.314268867924528,
      "grad_norm": 1.6422923803329468,
      "learning_rate": 7.999532922553635e-06,
      "loss": 0.3497,
      "num_input_tokens_seen": 8084832,
      "step": 12405
    },
    {
      "epoch": 7.317216981132075,
      "grad_norm": 1.8990668058395386,
      "learning_rate": 7.997474174060508e-06,
      "loss": 0.4747,
      "num_input_tokens_seen": 8088416,
      "step": 12410
    },
    {
      "epoch": 7.320165094339623,
      "grad_norm": 2.0357236862182617,
      "learning_rate": 7.995414631973179e-06,
      "loss": 0.2893,
      "num_input_tokens_seen": 8091584,
      "step": 12415
    },
    {
      "epoch": 7.32311320754717,
      "grad_norm": 2.379852056503296,
      "learning_rate": 7.993354296836914e-06,
      "loss": 0.2884,
      "num_input_tokens_seen": 8094208,
      "step": 12420
    },
    {
      "epoch": 7.326061320754717,
      "grad_norm": 2.0941812992095947,
      "learning_rate": 7.991293169197198e-06,
      "loss": 0.3217,
      "num_input_tokens_seen": 8096864,
      "step": 12425
    },
    {
      "epoch": 7.3290094339622645,
      "grad_norm": 3.003298759460449,
      "learning_rate": 7.989231249599725e-06,
      "loss": 0.4849,
      "num_input_tokens_seen": 8099712,
      "step": 12430
    },
    {
      "epoch": 7.331957547169811,
      "grad_norm": 2.5988895893096924,
      "learning_rate": 7.987168538590395e-06,
      "loss": 0.3672,
      "num_input_tokens_seen": 8102400,
      "step": 12435
    },
    {
      "epoch": 7.334905660377358,
      "grad_norm": 0.8338944911956787,
      "learning_rate": 7.985105036715322e-06,
      "loss": 0.2029,
      "num_input_tokens_seen": 8108288,
      "step": 12440
    },
    {
      "epoch": 7.337853773584905,
      "grad_norm": 1.8371944427490234,
      "learning_rate": 7.983040744520823e-06,
      "loss": 0.2693,
      "num_input_tokens_seen": 8111008,
      "step": 12445
    },
    {
      "epoch": 7.340801886792453,
      "grad_norm": 1.773409128189087,
      "learning_rate": 7.980975662553432e-06,
      "loss": 0.4571,
      "num_input_tokens_seen": 8115008,
      "step": 12450
    },
    {
      "epoch": 7.34375,
      "grad_norm": 1.9357980489730835,
      "learning_rate": 7.978909791359888e-06,
      "loss": 0.3693,
      "num_input_tokens_seen": 8118752,
      "step": 12455
    },
    {
      "epoch": 7.346698113207547,
      "grad_norm": 1.732206106185913,
      "learning_rate": 7.976843131487136e-06,
      "loss": 0.5031,
      "num_input_tokens_seen": 8122048,
      "step": 12460
    },
    {
      "epoch": 7.349646226415095,
      "grad_norm": 2.9061222076416016,
      "learning_rate": 7.974775683482337e-06,
      "loss": 0.4317,
      "num_input_tokens_seen": 8124448,
      "step": 12465
    },
    {
      "epoch": 7.352594339622642,
      "grad_norm": 1.3417195081710815,
      "learning_rate": 7.972707447892855e-06,
      "loss": 0.5957,
      "num_input_tokens_seen": 8128096,
      "step": 12470
    },
    {
      "epoch": 7.355542452830189,
      "grad_norm": 2.0616581439971924,
      "learning_rate": 7.970638425266264e-06,
      "loss": 0.5007,
      "num_input_tokens_seen": 8131488,
      "step": 12475
    },
    {
      "epoch": 7.3584905660377355,
      "grad_norm": 2.0164268016815186,
      "learning_rate": 7.968568616150349e-06,
      "loss": 0.3962,
      "num_input_tokens_seen": 8134688,
      "step": 12480
    },
    {
      "epoch": 7.361438679245283,
      "grad_norm": 1.1332072019577026,
      "learning_rate": 7.966498021093096e-06,
      "loss": 0.3502,
      "num_input_tokens_seen": 8138368,
      "step": 12485
    },
    {
      "epoch": 7.36438679245283,
      "grad_norm": 19.94340705871582,
      "learning_rate": 7.96442664064271e-06,
      "loss": 0.5062,
      "num_input_tokens_seen": 8141408,
      "step": 12490
    },
    {
      "epoch": 7.367334905660377,
      "grad_norm": 2.6861815452575684,
      "learning_rate": 7.962354475347593e-06,
      "loss": 0.4325,
      "num_input_tokens_seen": 8143872,
      "step": 12495
    },
    {
      "epoch": 7.370283018867925,
      "grad_norm": 1.2617459297180176,
      "learning_rate": 7.960281525756364e-06,
      "loss": 0.2763,
      "num_input_tokens_seen": 8148096,
      "step": 12500
    },
    {
      "epoch": 7.373231132075472,
      "grad_norm": 3.2648725509643555,
      "learning_rate": 7.95820779241784e-06,
      "loss": 0.5184,
      "num_input_tokens_seen": 8150848,
      "step": 12505
    },
    {
      "epoch": 7.376179245283019,
      "grad_norm": 2.480302095413208,
      "learning_rate": 7.956133275881055e-06,
      "loss": 0.4826,
      "num_input_tokens_seen": 8155232,
      "step": 12510
    },
    {
      "epoch": 7.379127358490566,
      "grad_norm": 1.5568878650665283,
      "learning_rate": 7.954057976695244e-06,
      "loss": 0.4783,
      "num_input_tokens_seen": 8158912,
      "step": 12515
    },
    {
      "epoch": 7.382075471698113,
      "grad_norm": 1.5117324590682983,
      "learning_rate": 7.951981895409854e-06,
      "loss": 0.3174,
      "num_input_tokens_seen": 8161696,
      "step": 12520
    },
    {
      "epoch": 7.38502358490566,
      "grad_norm": 3.119663953781128,
      "learning_rate": 7.949905032574534e-06,
      "loss": 0.4056,
      "num_input_tokens_seen": 8166272,
      "step": 12525
    },
    {
      "epoch": 7.3879716981132075,
      "grad_norm": 4.333902835845947,
      "learning_rate": 7.947827388739145e-06,
      "loss": 0.5553,
      "num_input_tokens_seen": 8169504,
      "step": 12530
    },
    {
      "epoch": 7.390919811320755,
      "grad_norm": 3.6118650436401367,
      "learning_rate": 7.945748964453747e-06,
      "loss": 0.468,
      "num_input_tokens_seen": 8172800,
      "step": 12535
    },
    {
      "epoch": 7.393867924528302,
      "grad_norm": 1.6763180494308472,
      "learning_rate": 7.943669760268618e-06,
      "loss": 0.511,
      "num_input_tokens_seen": 8176480,
      "step": 12540
    },
    {
      "epoch": 7.396816037735849,
      "grad_norm": 2.1456193923950195,
      "learning_rate": 7.941589776734232e-06,
      "loss": 0.4246,
      "num_input_tokens_seen": 8179616,
      "step": 12545
    },
    {
      "epoch": 7.399764150943396,
      "grad_norm": 2.367192029953003,
      "learning_rate": 7.939509014401277e-06,
      "loss": 0.3964,
      "num_input_tokens_seen": 8183488,
      "step": 12550
    },
    {
      "epoch": 7.402712264150943,
      "grad_norm": 3.201099395751953,
      "learning_rate": 7.93742747382064e-06,
      "loss": 0.4537,
      "num_input_tokens_seen": 8186592,
      "step": 12555
    },
    {
      "epoch": 7.40566037735849,
      "grad_norm": 1.5436292886734009,
      "learning_rate": 7.93534515554342e-06,
      "loss": 0.3649,
      "num_input_tokens_seen": 8190048,
      "step": 12560
    },
    {
      "epoch": 7.408608490566038,
      "grad_norm": 2.39947509765625,
      "learning_rate": 7.933262060120918e-06,
      "loss": 0.4374,
      "num_input_tokens_seen": 8193440,
      "step": 12565
    },
    {
      "epoch": 7.411556603773585,
      "grad_norm": 1.8190984725952148,
      "learning_rate": 7.931178188104646e-06,
      "loss": 0.4411,
      "num_input_tokens_seen": 8196736,
      "step": 12570
    },
    {
      "epoch": 7.414504716981132,
      "grad_norm": 1.9123783111572266,
      "learning_rate": 7.929093540046317e-06,
      "loss": 0.5102,
      "num_input_tokens_seen": 8199456,
      "step": 12575
    },
    {
      "epoch": 7.4174528301886795,
      "grad_norm": 2.235212802886963,
      "learning_rate": 7.927008116497848e-06,
      "loss": 0.6973,
      "num_input_tokens_seen": 8203360,
      "step": 12580
    },
    {
      "epoch": 7.420400943396227,
      "grad_norm": 1.9296789169311523,
      "learning_rate": 7.924921918011366e-06,
      "loss": 0.4744,
      "num_input_tokens_seen": 8207264,
      "step": 12585
    },
    {
      "epoch": 7.423349056603773,
      "grad_norm": 2.060218095779419,
      "learning_rate": 7.9228349451392e-06,
      "loss": 0.3945,
      "num_input_tokens_seen": 8210528,
      "step": 12590
    },
    {
      "epoch": 7.4262971698113205,
      "grad_norm": 1.3163892030715942,
      "learning_rate": 7.920747198433884e-06,
      "loss": 0.3177,
      "num_input_tokens_seen": 8213728,
      "step": 12595
    },
    {
      "epoch": 7.429245283018868,
      "grad_norm": 1.5851162672042847,
      "learning_rate": 7.91865867844816e-06,
      "loss": 0.4901,
      "num_input_tokens_seen": 8217856,
      "step": 12600
    },
    {
      "epoch": 7.432193396226415,
      "grad_norm": 1.8003064393997192,
      "learning_rate": 7.916569385734976e-06,
      "loss": 0.3317,
      "num_input_tokens_seen": 8220000,
      "step": 12605
    },
    {
      "epoch": 7.435141509433962,
      "grad_norm": 1.8168344497680664,
      "learning_rate": 7.914479320847474e-06,
      "loss": 0.3992,
      "num_input_tokens_seen": 8223040,
      "step": 12610
    },
    {
      "epoch": 7.43808962264151,
      "grad_norm": 2.325937509536743,
      "learning_rate": 7.912388484339012e-06,
      "loss": 0.3922,
      "num_input_tokens_seen": 8225344,
      "step": 12615
    },
    {
      "epoch": 7.441037735849057,
      "grad_norm": 3.7619271278381348,
      "learning_rate": 7.910296876763147e-06,
      "loss": 0.4717,
      "num_input_tokens_seen": 8227840,
      "step": 12620
    },
    {
      "epoch": 7.443985849056604,
      "grad_norm": 1.8331016302108765,
      "learning_rate": 7.90820449867364e-06,
      "loss": 0.4376,
      "num_input_tokens_seen": 8231232,
      "step": 12625
    },
    {
      "epoch": 7.446933962264151,
      "grad_norm": 2.1001436710357666,
      "learning_rate": 7.90611135062446e-06,
      "loss": 0.414,
      "num_input_tokens_seen": 8234400,
      "step": 12630
    },
    {
      "epoch": 7.449882075471698,
      "grad_norm": 2.465538740158081,
      "learning_rate": 7.904017433169775e-06,
      "loss": 0.4386,
      "num_input_tokens_seen": 8237312,
      "step": 12635
    },
    {
      "epoch": 7.452830188679245,
      "grad_norm": 5.104916095733643,
      "learning_rate": 7.901922746863957e-06,
      "loss": 0.4274,
      "num_input_tokens_seen": 8240864,
      "step": 12640
    },
    {
      "epoch": 7.4557783018867925,
      "grad_norm": 1.9842617511749268,
      "learning_rate": 7.899827292261589e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 8244256,
      "step": 12645
    },
    {
      "epoch": 7.45872641509434,
      "grad_norm": 2.355825424194336,
      "learning_rate": 7.897731069917444e-06,
      "loss": 0.3408,
      "num_input_tokens_seen": 8248064,
      "step": 12650
    },
    {
      "epoch": 7.461674528301887,
      "grad_norm": 3.1829774379730225,
      "learning_rate": 7.895634080386512e-06,
      "loss": 0.4101,
      "num_input_tokens_seen": 8251648,
      "step": 12655
    },
    {
      "epoch": 7.464622641509434,
      "grad_norm": 2.1236677169799805,
      "learning_rate": 7.893536324223977e-06,
      "loss": 0.4067,
      "num_input_tokens_seen": 8255456,
      "step": 12660
    },
    {
      "epoch": 7.467570754716981,
      "grad_norm": 2.7250266075134277,
      "learning_rate": 7.89143780198523e-06,
      "loss": 0.4408,
      "num_input_tokens_seen": 8258208,
      "step": 12665
    },
    {
      "epoch": 7.470518867924528,
      "grad_norm": 3.2099993228912354,
      "learning_rate": 7.889338514225862e-06,
      "loss": 0.3645,
      "num_input_tokens_seen": 8260960,
      "step": 12670
    },
    {
      "epoch": 7.473466981132075,
      "grad_norm": 4.372264862060547,
      "learning_rate": 7.887238461501671e-06,
      "loss": 0.4477,
      "num_input_tokens_seen": 8263360,
      "step": 12675
    },
    {
      "epoch": 7.476415094339623,
      "grad_norm": 1.862304449081421,
      "learning_rate": 7.885137644368654e-06,
      "loss": 0.294,
      "num_input_tokens_seen": 8266016,
      "step": 12680
    },
    {
      "epoch": 7.47936320754717,
      "grad_norm": 2.1847269535064697,
      "learning_rate": 7.883036063383012e-06,
      "loss": 0.4599,
      "num_input_tokens_seen": 8268640,
      "step": 12685
    },
    {
      "epoch": 7.482311320754717,
      "grad_norm": 3.9076075553894043,
      "learning_rate": 7.880933719101148e-06,
      "loss": 0.3729,
      "num_input_tokens_seen": 8271456,
      "step": 12690
    },
    {
      "epoch": 7.4852594339622645,
      "grad_norm": 2.3612866401672363,
      "learning_rate": 7.878830612079664e-06,
      "loss": 0.381,
      "num_input_tokens_seen": 8274656,
      "step": 12695
    },
    {
      "epoch": 7.488207547169811,
      "grad_norm": 5.0508131980896,
      "learning_rate": 7.876726742875369e-06,
      "loss": 0.4714,
      "num_input_tokens_seen": 8277472,
      "step": 12700
    },
    {
      "epoch": 7.491155660377358,
      "grad_norm": 3.3084397315979004,
      "learning_rate": 7.874622112045269e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 8280800,
      "step": 12705
    },
    {
      "epoch": 7.494103773584905,
      "grad_norm": 2.756094455718994,
      "learning_rate": 7.872516720146578e-06,
      "loss": 0.3916,
      "num_input_tokens_seen": 8283712,
      "step": 12710
    },
    {
      "epoch": 7.497051886792453,
      "grad_norm": 1.5952520370483398,
      "learning_rate": 7.870410567736705e-06,
      "loss": 0.37,
      "num_input_tokens_seen": 8287872,
      "step": 12715
    },
    {
      "epoch": 7.5,
      "grad_norm": 1.5211774110794067,
      "learning_rate": 7.868303655373264e-06,
      "loss": 0.3059,
      "num_input_tokens_seen": 8291328,
      "step": 12720
    },
    {
      "epoch": 7.502948113207547,
      "grad_norm": 2.261403799057007,
      "learning_rate": 7.866195983614066e-06,
      "loss": 0.5107,
      "num_input_tokens_seen": 8293920,
      "step": 12725
    },
    {
      "epoch": 7.505896226415095,
      "grad_norm": 4.688958168029785,
      "learning_rate": 7.864087553017133e-06,
      "loss": 0.5651,
      "num_input_tokens_seen": 8296704,
      "step": 12730
    },
    {
      "epoch": 7.508844339622642,
      "grad_norm": 3.901465892791748,
      "learning_rate": 7.861978364140674e-06,
      "loss": 0.4373,
      "num_input_tokens_seen": 8301792,
      "step": 12735
    },
    {
      "epoch": 7.511792452830189,
      "grad_norm": 2.922650098800659,
      "learning_rate": 7.859868417543109e-06,
      "loss": 0.4092,
      "num_input_tokens_seen": 8305312,
      "step": 12740
    },
    {
      "epoch": 7.5147405660377355,
      "grad_norm": 1.942541241645813,
      "learning_rate": 7.857757713783055e-06,
      "loss": 0.3922,
      "num_input_tokens_seen": 8308832,
      "step": 12745
    },
    {
      "epoch": 7.517688679245283,
      "grad_norm": 2.7667229175567627,
      "learning_rate": 7.855646253419331e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 8311200,
      "step": 12750
    },
    {
      "epoch": 7.52063679245283,
      "grad_norm": 1.7393347024917603,
      "learning_rate": 7.853534037010952e-06,
      "loss": 0.4667,
      "num_input_tokens_seen": 8313952,
      "step": 12755
    },
    {
      "epoch": 7.523584905660377,
      "grad_norm": 2.5969932079315186,
      "learning_rate": 7.851421065117142e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 8317024,
      "step": 12760
    },
    {
      "epoch": 7.526533018867925,
      "grad_norm": 3.351426601409912,
      "learning_rate": 7.849307338297314e-06,
      "loss": 0.382,
      "num_input_tokens_seen": 8319648,
      "step": 12765
    },
    {
      "epoch": 7.529481132075472,
      "grad_norm": 1.6915512084960938,
      "learning_rate": 7.847192857111087e-06,
      "loss": 0.3355,
      "num_input_tokens_seen": 8322848,
      "step": 12770
    },
    {
      "epoch": 7.532429245283019,
      "grad_norm": 2.4099693298339844,
      "learning_rate": 7.845077622118282e-06,
      "loss": 0.5652,
      "num_input_tokens_seen": 8325664,
      "step": 12775
    },
    {
      "epoch": 7.535377358490566,
      "grad_norm": 1.341578483581543,
      "learning_rate": 7.842961633878916e-06,
      "loss": 0.544,
      "num_input_tokens_seen": 8330144,
      "step": 12780
    },
    {
      "epoch": 7.538325471698113,
      "grad_norm": 2.356422185897827,
      "learning_rate": 7.840844892953204e-06,
      "loss": 0.4593,
      "num_input_tokens_seen": 8333344,
      "step": 12785
    },
    {
      "epoch": 7.54127358490566,
      "grad_norm": 3.681206703186035,
      "learning_rate": 7.838727399901562e-06,
      "loss": 0.4677,
      "num_input_tokens_seen": 8337920,
      "step": 12790
    },
    {
      "epoch": 7.5442216981132075,
      "grad_norm": 1.7842581272125244,
      "learning_rate": 7.836609155284607e-06,
      "loss": 0.399,
      "num_input_tokens_seen": 8340704,
      "step": 12795
    },
    {
      "epoch": 7.547169811320755,
      "grad_norm": 3.0184223651885986,
      "learning_rate": 7.834490159663154e-06,
      "loss": 0.3755,
      "num_input_tokens_seen": 8343872,
      "step": 12800
    },
    {
      "epoch": 7.550117924528302,
      "grad_norm": 3.3758602142333984,
      "learning_rate": 7.832370413598215e-06,
      "loss": 0.3324,
      "num_input_tokens_seen": 8347456,
      "step": 12805
    },
    {
      "epoch": 7.553066037735849,
      "grad_norm": 1.961235761642456,
      "learning_rate": 7.830249917651003e-06,
      "loss": 0.5419,
      "num_input_tokens_seen": 8350688,
      "step": 12810
    },
    {
      "epoch": 7.556014150943396,
      "grad_norm": 2.4225265979766846,
      "learning_rate": 7.828128672382926e-06,
      "loss": 0.4354,
      "num_input_tokens_seen": 8353920,
      "step": 12815
    },
    {
      "epoch": 7.558962264150943,
      "grad_norm": 1.5328577756881714,
      "learning_rate": 7.826006678355596e-06,
      "loss": 0.3481,
      "num_input_tokens_seen": 8357024,
      "step": 12820
    },
    {
      "epoch": 7.56191037735849,
      "grad_norm": 2.3475513458251953,
      "learning_rate": 7.823883936130817e-06,
      "loss": 0.3697,
      "num_input_tokens_seen": 8360160,
      "step": 12825
    },
    {
      "epoch": 7.564858490566038,
      "grad_norm": 2.133432626724243,
      "learning_rate": 7.821760446270597e-06,
      "loss": 0.3563,
      "num_input_tokens_seen": 8362976,
      "step": 12830
    },
    {
      "epoch": 7.567806603773585,
      "grad_norm": 1.855751395225525,
      "learning_rate": 7.819636209337136e-06,
      "loss": 0.3888,
      "num_input_tokens_seen": 8365888,
      "step": 12835
    },
    {
      "epoch": 7.570754716981132,
      "grad_norm": 1.731449842453003,
      "learning_rate": 7.817511225892838e-06,
      "loss": 0.4236,
      "num_input_tokens_seen": 8369120,
      "step": 12840
    },
    {
      "epoch": 7.5737028301886795,
      "grad_norm": 2.062988519668579,
      "learning_rate": 7.8153854965003e-06,
      "loss": 0.4603,
      "num_input_tokens_seen": 8373376,
      "step": 12845
    },
    {
      "epoch": 7.576650943396227,
      "grad_norm": 2.6432371139526367,
      "learning_rate": 7.813259021722319e-06,
      "loss": 0.3501,
      "num_input_tokens_seen": 8375808,
      "step": 12850
    },
    {
      "epoch": 7.579599056603773,
      "grad_norm": 2.4975013732910156,
      "learning_rate": 7.811131802121885e-06,
      "loss": 0.5657,
      "num_input_tokens_seen": 8378720,
      "step": 12855
    },
    {
      "epoch": 7.5825471698113205,
      "grad_norm": 2.2664835453033447,
      "learning_rate": 7.809003838262193e-06,
      "loss": 0.3197,
      "num_input_tokens_seen": 8381728,
      "step": 12860
    },
    {
      "epoch": 7.585495283018868,
      "grad_norm": 2.14251446723938,
      "learning_rate": 7.806875130706628e-06,
      "loss": 0.4422,
      "num_input_tokens_seen": 8384480,
      "step": 12865
    },
    {
      "epoch": 7.588443396226415,
      "grad_norm": 3.790902614593506,
      "learning_rate": 7.804745680018775e-06,
      "loss": 0.3221,
      "num_input_tokens_seen": 8387104,
      "step": 12870
    },
    {
      "epoch": 7.591391509433962,
      "grad_norm": 1.4395138025283813,
      "learning_rate": 7.802615486762418e-06,
      "loss": 0.3857,
      "num_input_tokens_seen": 8390560,
      "step": 12875
    },
    {
      "epoch": 7.59433962264151,
      "grad_norm": 2.2490885257720947,
      "learning_rate": 7.800484551501528e-06,
      "loss": 0.4363,
      "num_input_tokens_seen": 8393472,
      "step": 12880
    },
    {
      "epoch": 7.597287735849057,
      "grad_norm": 2.8760266304016113,
      "learning_rate": 7.798352874800285e-06,
      "loss": 0.4374,
      "num_input_tokens_seen": 8396416,
      "step": 12885
    },
    {
      "epoch": 7.600235849056604,
      "grad_norm": 1.5365777015686035,
      "learning_rate": 7.79622045722306e-06,
      "loss": 0.3626,
      "num_input_tokens_seen": 8400096,
      "step": 12890
    },
    {
      "epoch": 7.603183962264151,
      "grad_norm": 4.174517631530762,
      "learning_rate": 7.794087299334416e-06,
      "loss": 0.3912,
      "num_input_tokens_seen": 8403552,
      "step": 12895
    },
    {
      "epoch": 7.606132075471698,
      "grad_norm": 1.8114674091339111,
      "learning_rate": 7.79195340169912e-06,
      "loss": 0.4015,
      "num_input_tokens_seen": 8407648,
      "step": 12900
    },
    {
      "epoch": 7.609080188679245,
      "grad_norm": 1.6811645030975342,
      "learning_rate": 7.789818764882127e-06,
      "loss": 0.471,
      "num_input_tokens_seen": 8410720,
      "step": 12905
    },
    {
      "epoch": 7.6120283018867925,
      "grad_norm": 2.1953039169311523,
      "learning_rate": 7.78768338944859e-06,
      "loss": 0.31,
      "num_input_tokens_seen": 8413408,
      "step": 12910
    },
    {
      "epoch": 7.61497641509434,
      "grad_norm": 1.9849573373794556,
      "learning_rate": 7.785547275963865e-06,
      "loss": 0.3788,
      "num_input_tokens_seen": 8416768,
      "step": 12915
    },
    {
      "epoch": 7.617924528301887,
      "grad_norm": 2.6337649822235107,
      "learning_rate": 7.783410424993492e-06,
      "loss": 0.3913,
      "num_input_tokens_seen": 8419072,
      "step": 12920
    },
    {
      "epoch": 7.620872641509434,
      "grad_norm": 2.12776255607605,
      "learning_rate": 7.781272837103213e-06,
      "loss": 0.4123,
      "num_input_tokens_seen": 8421728,
      "step": 12925
    },
    {
      "epoch": 7.623820754716981,
      "grad_norm": 2.0554819107055664,
      "learning_rate": 7.779134512858964e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 8425184,
      "step": 12930
    },
    {
      "epoch": 7.626768867924528,
      "grad_norm": 8.579044342041016,
      "learning_rate": 7.776995452826876e-06,
      "loss": 0.4383,
      "num_input_tokens_seen": 8428256,
      "step": 12935
    },
    {
      "epoch": 7.629716981132075,
      "grad_norm": 2.6147043704986572,
      "learning_rate": 7.774855657573274e-06,
      "loss": 0.4513,
      "num_input_tokens_seen": 8431872,
      "step": 12940
    },
    {
      "epoch": 7.632665094339623,
      "grad_norm": 1.7213010787963867,
      "learning_rate": 7.772715127664676e-06,
      "loss": 0.3603,
      "num_input_tokens_seen": 8434656,
      "step": 12945
    },
    {
      "epoch": 7.63561320754717,
      "grad_norm": 3.4233412742614746,
      "learning_rate": 7.7705738636678e-06,
      "loss": 0.3734,
      "num_input_tokens_seen": 8437504,
      "step": 12950
    },
    {
      "epoch": 7.638561320754717,
      "grad_norm": 1.9577927589416504,
      "learning_rate": 7.768431866149552e-06,
      "loss": 0.3585,
      "num_input_tokens_seen": 8440960,
      "step": 12955
    },
    {
      "epoch": 7.6415094339622645,
      "grad_norm": 2.636037588119507,
      "learning_rate": 7.766289135677035e-06,
      "loss": 0.4011,
      "num_input_tokens_seen": 8444128,
      "step": 12960
    },
    {
      "epoch": 7.644457547169811,
      "grad_norm": 1.8809701204299927,
      "learning_rate": 7.764145672817549e-06,
      "loss": 0.498,
      "num_input_tokens_seen": 8448192,
      "step": 12965
    },
    {
      "epoch": 7.647405660377358,
      "grad_norm": 2.8841614723205566,
      "learning_rate": 7.762001478138583e-06,
      "loss": 0.4189,
      "num_input_tokens_seen": 8451744,
      "step": 12970
    },
    {
      "epoch": 7.650353773584905,
      "grad_norm": 3.031399965286255,
      "learning_rate": 7.759856552207822e-06,
      "loss": 0.4304,
      "num_input_tokens_seen": 8455040,
      "step": 12975
    },
    {
      "epoch": 7.653301886792453,
      "grad_norm": 4.450901508331299,
      "learning_rate": 7.757710895593144e-06,
      "loss": 0.3564,
      "num_input_tokens_seen": 8457792,
      "step": 12980
    },
    {
      "epoch": 7.65625,
      "grad_norm": 2.7953264713287354,
      "learning_rate": 7.755564508862623e-06,
      "loss": 0.4801,
      "num_input_tokens_seen": 8461440,
      "step": 12985
    },
    {
      "epoch": 7.659198113207547,
      "grad_norm": 2.5081403255462646,
      "learning_rate": 7.753417392584522e-06,
      "loss": 0.527,
      "num_input_tokens_seen": 8464576,
      "step": 12990
    },
    {
      "epoch": 7.662146226415095,
      "grad_norm": 2.059462070465088,
      "learning_rate": 7.751269547327298e-06,
      "loss": 0.301,
      "num_input_tokens_seen": 8468160,
      "step": 12995
    },
    {
      "epoch": 7.665094339622642,
      "grad_norm": 2.0491855144500732,
      "learning_rate": 7.749120973659606e-06,
      "loss": 0.3607,
      "num_input_tokens_seen": 8471296,
      "step": 13000
    },
    {
      "epoch": 7.668042452830189,
      "grad_norm": 2.485041379928589,
      "learning_rate": 7.746971672150286e-06,
      "loss": 0.3674,
      "num_input_tokens_seen": 8473728,
      "step": 13005
    },
    {
      "epoch": 7.6709905660377355,
      "grad_norm": 3.1727583408355713,
      "learning_rate": 7.74482164336838e-06,
      "loss": 0.5391,
      "num_input_tokens_seen": 8477600,
      "step": 13010
    },
    {
      "epoch": 7.673938679245283,
      "grad_norm": 1.7440383434295654,
      "learning_rate": 7.742670887883111e-06,
      "loss": 0.5365,
      "num_input_tokens_seen": 8480704,
      "step": 13015
    },
    {
      "epoch": 7.67688679245283,
      "grad_norm": 2.0993595123291016,
      "learning_rate": 7.740519406263905e-06,
      "loss": 0.3469,
      "num_input_tokens_seen": 8483872,
      "step": 13020
    },
    {
      "epoch": 7.679834905660377,
      "grad_norm": 3.6898202896118164,
      "learning_rate": 7.738367199080376e-06,
      "loss": 0.2866,
      "num_input_tokens_seen": 8486848,
      "step": 13025
    },
    {
      "epoch": 7.682783018867925,
      "grad_norm": 2.1749861240386963,
      "learning_rate": 7.73621426690233e-06,
      "loss": 0.57,
      "num_input_tokens_seen": 8489536,
      "step": 13030
    },
    {
      "epoch": 7.685731132075472,
      "grad_norm": 8.308359146118164,
      "learning_rate": 7.734060610299764e-06,
      "loss": 0.4558,
      "num_input_tokens_seen": 8492320,
      "step": 13035
    },
    {
      "epoch": 7.688679245283019,
      "grad_norm": 2.591728687286377,
      "learning_rate": 7.731906229842869e-06,
      "loss": 0.5196,
      "num_input_tokens_seen": 8495328,
      "step": 13040
    },
    {
      "epoch": 7.691627358490566,
      "grad_norm": 2.6472785472869873,
      "learning_rate": 7.729751126102023e-06,
      "loss": 0.4421,
      "num_input_tokens_seen": 8498368,
      "step": 13045
    },
    {
      "epoch": 7.694575471698113,
      "grad_norm": 1.0644233226776123,
      "learning_rate": 7.727595299647805e-06,
      "loss": 0.3836,
      "num_input_tokens_seen": 8501632,
      "step": 13050
    },
    {
      "epoch": 7.69752358490566,
      "grad_norm": 1.721089243888855,
      "learning_rate": 7.725438751050973e-06,
      "loss": 0.3607,
      "num_input_tokens_seen": 8504800,
      "step": 13055
    },
    {
      "epoch": 7.7004716981132075,
      "grad_norm": 2.9734160900115967,
      "learning_rate": 7.723281480882489e-06,
      "loss": 0.5088,
      "num_input_tokens_seen": 8507264,
      "step": 13060
    },
    {
      "epoch": 7.703419811320755,
      "grad_norm": 4.0242600440979,
      "learning_rate": 7.721123489713494e-06,
      "loss": 0.3502,
      "num_input_tokens_seen": 8511040,
      "step": 13065
    },
    {
      "epoch": 7.706367924528302,
      "grad_norm": 3.9850783348083496,
      "learning_rate": 7.718964778115328e-06,
      "loss": 0.394,
      "num_input_tokens_seen": 8513952,
      "step": 13070
    },
    {
      "epoch": 7.709316037735849,
      "grad_norm": 1.89198899269104,
      "learning_rate": 7.716805346659519e-06,
      "loss": 0.4228,
      "num_input_tokens_seen": 8517344,
      "step": 13075
    },
    {
      "epoch": 7.712264150943396,
      "grad_norm": 3.7127890586853027,
      "learning_rate": 7.714645195917788e-06,
      "loss": 0.4467,
      "num_input_tokens_seen": 8520448,
      "step": 13080
    },
    {
      "epoch": 7.715212264150943,
      "grad_norm": 2.149409532546997,
      "learning_rate": 7.712484326462038e-06,
      "loss": 0.4963,
      "num_input_tokens_seen": 8523328,
      "step": 13085
    },
    {
      "epoch": 7.71816037735849,
      "grad_norm": 1.1930314302444458,
      "learning_rate": 7.710322738864375e-06,
      "loss": 0.3976,
      "num_input_tokens_seen": 8526816,
      "step": 13090
    },
    {
      "epoch": 7.721108490566038,
      "grad_norm": 3.9999423027038574,
      "learning_rate": 7.708160433697085e-06,
      "loss": 0.4499,
      "num_input_tokens_seen": 8529504,
      "step": 13095
    },
    {
      "epoch": 7.724056603773585,
      "grad_norm": 1.6784738302230835,
      "learning_rate": 7.705997411532649e-06,
      "loss": 0.3348,
      "num_input_tokens_seen": 8533056,
      "step": 13100
    },
    {
      "epoch": 7.727004716981132,
      "grad_norm": 2.267273426055908,
      "learning_rate": 7.703833672943735e-06,
      "loss": 0.4559,
      "num_input_tokens_seen": 8536288,
      "step": 13105
    },
    {
      "epoch": 7.7299528301886795,
      "grad_norm": 2.2425405979156494,
      "learning_rate": 7.701669218503206e-06,
      "loss": 0.4687,
      "num_input_tokens_seen": 8538880,
      "step": 13110
    },
    {
      "epoch": 7.732900943396227,
      "grad_norm": 3.9774842262268066,
      "learning_rate": 7.699504048784106e-06,
      "loss": 0.402,
      "num_input_tokens_seen": 8544736,
      "step": 13115
    },
    {
      "epoch": 7.735849056603773,
      "grad_norm": 2.8074254989624023,
      "learning_rate": 7.697338164359675e-06,
      "loss": 0.4157,
      "num_input_tokens_seen": 8547136,
      "step": 13120
    },
    {
      "epoch": 7.7387971698113205,
      "grad_norm": 2.6301400661468506,
      "learning_rate": 7.69517156580334e-06,
      "loss": 0.3464,
      "num_input_tokens_seen": 8549984,
      "step": 13125
    },
    {
      "epoch": 7.741745283018868,
      "grad_norm": 2.555373191833496,
      "learning_rate": 7.693004253688716e-06,
      "loss": 0.4093,
      "num_input_tokens_seen": 8553792,
      "step": 13130
    },
    {
      "epoch": 7.744693396226415,
      "grad_norm": 1.6086993217468262,
      "learning_rate": 7.690836228589613e-06,
      "loss": 0.3818,
      "num_input_tokens_seen": 8556480,
      "step": 13135
    },
    {
      "epoch": 7.747641509433962,
      "grad_norm": 3.822707176208496,
      "learning_rate": 7.688667491080019e-06,
      "loss": 0.4145,
      "num_input_tokens_seen": 8559616,
      "step": 13140
    },
    {
      "epoch": 7.75058962264151,
      "grad_norm": 2.938126802444458,
      "learning_rate": 7.686498041734121e-06,
      "loss": 0.4131,
      "num_input_tokens_seen": 8562400,
      "step": 13145
    },
    {
      "epoch": 7.753537735849057,
      "grad_norm": 2.0569612979888916,
      "learning_rate": 7.684327881126285e-06,
      "loss": 0.4227,
      "num_input_tokens_seen": 8566336,
      "step": 13150
    },
    {
      "epoch": 7.756485849056604,
      "grad_norm": 1.5294172763824463,
      "learning_rate": 7.682157009831078e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 8570976,
      "step": 13155
    },
    {
      "epoch": 7.759433962264151,
      "grad_norm": 2.7983970642089844,
      "learning_rate": 7.67998542842324e-06,
      "loss": 0.4101,
      "num_input_tokens_seen": 8574176,
      "step": 13160
    },
    {
      "epoch": 7.762382075471698,
      "grad_norm": 1.6997811794281006,
      "learning_rate": 7.677813137477711e-06,
      "loss": 0.4454,
      "num_input_tokens_seen": 8577120,
      "step": 13165
    },
    {
      "epoch": 7.765330188679245,
      "grad_norm": 2.4485061168670654,
      "learning_rate": 7.675640137569614e-06,
      "loss": 0.441,
      "num_input_tokens_seen": 8579936,
      "step": 13170
    },
    {
      "epoch": 7.7682783018867925,
      "grad_norm": 2.12693190574646,
      "learning_rate": 7.673466429274257e-06,
      "loss": 0.5256,
      "num_input_tokens_seen": 8582304,
      "step": 13175
    },
    {
      "epoch": 7.77122641509434,
      "grad_norm": 1.5442661046981812,
      "learning_rate": 7.671292013167143e-06,
      "loss": 0.3135,
      "num_input_tokens_seen": 8585184,
      "step": 13180
    },
    {
      "epoch": 7.774174528301887,
      "grad_norm": 1.6561390161514282,
      "learning_rate": 7.669116889823955e-06,
      "loss": 0.4471,
      "num_input_tokens_seen": 8588128,
      "step": 13185
    },
    {
      "epoch": 7.777122641509434,
      "grad_norm": 1.3347402811050415,
      "learning_rate": 7.666941059820567e-06,
      "loss": 0.3611,
      "num_input_tokens_seen": 8591808,
      "step": 13190
    },
    {
      "epoch": 7.780070754716981,
      "grad_norm": 3.6079633235931396,
      "learning_rate": 7.66476452373304e-06,
      "loss": 0.4664,
      "num_input_tokens_seen": 8594752,
      "step": 13195
    },
    {
      "epoch": 7.783018867924528,
      "grad_norm": 1.9773160219192505,
      "learning_rate": 7.66258728213762e-06,
      "loss": 0.3859,
      "num_input_tokens_seen": 8598304,
      "step": 13200
    },
    {
      "epoch": 7.785966981132075,
      "grad_norm": 3.327244520187378,
      "learning_rate": 7.66040933561074e-06,
      "loss": 0.3913,
      "num_input_tokens_seen": 8600992,
      "step": 13205
    },
    {
      "epoch": 7.788915094339623,
      "grad_norm": 2.380021810531616,
      "learning_rate": 7.658230684729027e-06,
      "loss": 0.4422,
      "num_input_tokens_seen": 8608192,
      "step": 13210
    },
    {
      "epoch": 7.79186320754717,
      "grad_norm": 1.989427924156189,
      "learning_rate": 7.656051330069282e-06,
      "loss": 0.3475,
      "num_input_tokens_seen": 8612576,
      "step": 13215
    },
    {
      "epoch": 7.794811320754717,
      "grad_norm": 3.6624433994293213,
      "learning_rate": 7.6538712722085e-06,
      "loss": 0.4225,
      "num_input_tokens_seen": 8615680,
      "step": 13220
    },
    {
      "epoch": 7.7977594339622645,
      "grad_norm": 2.0149941444396973,
      "learning_rate": 7.651690511723862e-06,
      "loss": 0.3988,
      "num_input_tokens_seen": 8618080,
      "step": 13225
    },
    {
      "epoch": 7.800707547169811,
      "grad_norm": 1.566019892692566,
      "learning_rate": 7.64950904919273e-06,
      "loss": 0.2902,
      "num_input_tokens_seen": 8620608,
      "step": 13230
    },
    {
      "epoch": 7.803655660377358,
      "grad_norm": 2.344214677810669,
      "learning_rate": 7.647326885192662e-06,
      "loss": 0.5519,
      "num_input_tokens_seen": 8623424,
      "step": 13235
    },
    {
      "epoch": 7.806603773584905,
      "grad_norm": 2.9266605377197266,
      "learning_rate": 7.645144020301392e-06,
      "loss": 0.5043,
      "num_input_tokens_seen": 8627360,
      "step": 13240
    },
    {
      "epoch": 7.809551886792453,
      "grad_norm": 2.5658066272735596,
      "learning_rate": 7.64296045509684e-06,
      "loss": 0.4525,
      "num_input_tokens_seen": 8629888,
      "step": 13245
    },
    {
      "epoch": 7.8125,
      "grad_norm": 2.396303176879883,
      "learning_rate": 7.64077619015712e-06,
      "loss": 0.5715,
      "num_input_tokens_seen": 8633408,
      "step": 13250
    },
    {
      "epoch": 7.815448113207547,
      "grad_norm": 2.0013973712921143,
      "learning_rate": 7.638591226060519e-06,
      "loss": 0.4835,
      "num_input_tokens_seen": 8636512,
      "step": 13255
    },
    {
      "epoch": 7.818396226415095,
      "grad_norm": 3.196708917617798,
      "learning_rate": 7.636405563385522e-06,
      "loss": 0.4835,
      "num_input_tokens_seen": 8640320,
      "step": 13260
    },
    {
      "epoch": 7.821344339622642,
      "grad_norm": 1.9235656261444092,
      "learning_rate": 7.634219202710789e-06,
      "loss": 0.5227,
      "num_input_tokens_seen": 8643872,
      "step": 13265
    },
    {
      "epoch": 7.824292452830189,
      "grad_norm": 1.7198208570480347,
      "learning_rate": 7.632032144615168e-06,
      "loss": 0.2569,
      "num_input_tokens_seen": 8646880,
      "step": 13270
    },
    {
      "epoch": 7.8272405660377355,
      "grad_norm": 2.324612617492676,
      "learning_rate": 7.629844389677695e-06,
      "loss": 0.4049,
      "num_input_tokens_seen": 8649568,
      "step": 13275
    },
    {
      "epoch": 7.830188679245283,
      "grad_norm": 2.9373276233673096,
      "learning_rate": 7.627655938477586e-06,
      "loss": 0.3216,
      "num_input_tokens_seen": 8652672,
      "step": 13280
    },
    {
      "epoch": 7.83313679245283,
      "grad_norm": 1.9209688901901245,
      "learning_rate": 7.6254667915942415e-06,
      "loss": 0.3919,
      "num_input_tokens_seen": 8656192,
      "step": 13285
    },
    {
      "epoch": 7.836084905660377,
      "grad_norm": 1.7439285516738892,
      "learning_rate": 7.62327694960725e-06,
      "loss": 0.455,
      "num_input_tokens_seen": 8659776,
      "step": 13290
    },
    {
      "epoch": 7.839033018867925,
      "grad_norm": 3.3810596466064453,
      "learning_rate": 7.621086413096379e-06,
      "loss": 0.4131,
      "num_input_tokens_seen": 8662688,
      "step": 13295
    },
    {
      "epoch": 7.841981132075472,
      "grad_norm": 4.054974555969238,
      "learning_rate": 7.618895182641584e-06,
      "loss": 0.3886,
      "num_input_tokens_seen": 8665760,
      "step": 13300
    },
    {
      "epoch": 7.844929245283019,
      "grad_norm": 2.0524401664733887,
      "learning_rate": 7.6167032588230035e-06,
      "loss": 0.5656,
      "num_input_tokens_seen": 8669280,
      "step": 13305
    },
    {
      "epoch": 7.847877358490566,
      "grad_norm": 2.329983711242676,
      "learning_rate": 7.614510642220958e-06,
      "loss": 0.3872,
      "num_input_tokens_seen": 8673184,
      "step": 13310
    },
    {
      "epoch": 7.850825471698113,
      "grad_norm": 1.6023428440093994,
      "learning_rate": 7.612317333415951e-06,
      "loss": 0.3348,
      "num_input_tokens_seen": 8676640,
      "step": 13315
    },
    {
      "epoch": 7.85377358490566,
      "grad_norm": 3.4624242782592773,
      "learning_rate": 7.610123332988673e-06,
      "loss": 0.5106,
      "num_input_tokens_seen": 8679744,
      "step": 13320
    },
    {
      "epoch": 7.8567216981132075,
      "grad_norm": 1.1404262781143188,
      "learning_rate": 7.607928641519992e-06,
      "loss": 0.3815,
      "num_input_tokens_seen": 8682752,
      "step": 13325
    },
    {
      "epoch": 7.859669811320755,
      "grad_norm": 2.5558249950408936,
      "learning_rate": 7.605733259590964e-06,
      "loss": 0.4225,
      "num_input_tokens_seen": 8685472,
      "step": 13330
    },
    {
      "epoch": 7.862617924528302,
      "grad_norm": 1.731750249862671,
      "learning_rate": 7.603537187782826e-06,
      "loss": 0.3393,
      "num_input_tokens_seen": 8689632,
      "step": 13335
    },
    {
      "epoch": 7.865566037735849,
      "grad_norm": 2.082778215408325,
      "learning_rate": 7.601340426676996e-06,
      "loss": 0.4447,
      "num_input_tokens_seen": 8693440,
      "step": 13340
    },
    {
      "epoch": 7.868514150943396,
      "grad_norm": 2.76826810836792,
      "learning_rate": 7.599142976855077e-06,
      "loss": 0.3355,
      "num_input_tokens_seen": 8696960,
      "step": 13345
    },
    {
      "epoch": 7.871462264150943,
      "grad_norm": 7.586110591888428,
      "learning_rate": 7.596944838898854e-06,
      "loss": 0.3982,
      "num_input_tokens_seen": 8700864,
      "step": 13350
    },
    {
      "epoch": 7.87441037735849,
      "grad_norm": 3.6489715576171875,
      "learning_rate": 7.594746013390293e-06,
      "loss": 0.432,
      "num_input_tokens_seen": 8703904,
      "step": 13355
    },
    {
      "epoch": 7.877358490566038,
      "grad_norm": 1.4688109159469604,
      "learning_rate": 7.59254650091154e-06,
      "loss": 0.4019,
      "num_input_tokens_seen": 8707552,
      "step": 13360
    },
    {
      "epoch": 7.880306603773585,
      "grad_norm": 2.336695909500122,
      "learning_rate": 7.59034630204493e-06,
      "loss": 0.3984,
      "num_input_tokens_seen": 8710464,
      "step": 13365
    },
    {
      "epoch": 7.883254716981132,
      "grad_norm": 2.1053175926208496,
      "learning_rate": 7.588145417372972e-06,
      "loss": 0.3975,
      "num_input_tokens_seen": 8713216,
      "step": 13370
    },
    {
      "epoch": 7.8862028301886795,
      "grad_norm": 4.455665111541748,
      "learning_rate": 7.585943847478361e-06,
      "loss": 0.4141,
      "num_input_tokens_seen": 8716512,
      "step": 13375
    },
    {
      "epoch": 7.889150943396227,
      "grad_norm": 2.159214735031128,
      "learning_rate": 7.583741592943971e-06,
      "loss": 0.3418,
      "num_input_tokens_seen": 8719840,
      "step": 13380
    },
    {
      "epoch": 7.892099056603773,
      "grad_norm": 1.2077945470809937,
      "learning_rate": 7.581538654352859e-06,
      "loss": 0.3369,
      "num_input_tokens_seen": 8723296,
      "step": 13385
    },
    {
      "epoch": 7.8950471698113205,
      "grad_norm": 1.9655312299728394,
      "learning_rate": 7.579335032288262e-06,
      "loss": 0.4314,
      "num_input_tokens_seen": 8726272,
      "step": 13390
    },
    {
      "epoch": 7.897995283018868,
      "grad_norm": 2.34226655960083,
      "learning_rate": 7.577130727333598e-06,
      "loss": 0.3827,
      "num_input_tokens_seen": 8729216,
      "step": 13395
    },
    {
      "epoch": 7.900943396226415,
      "grad_norm": 4.47019624710083,
      "learning_rate": 7.5749257400724695e-06,
      "loss": 0.4882,
      "num_input_tokens_seen": 8731648,
      "step": 13400
    },
    {
      "epoch": 7.903891509433962,
      "grad_norm": 1.927394986152649,
      "learning_rate": 7.572720071088653e-06,
      "loss": 0.4285,
      "num_input_tokens_seen": 8735104,
      "step": 13405
    },
    {
      "epoch": 7.90683962264151,
      "grad_norm": 2.8650500774383545,
      "learning_rate": 7.570513720966108e-06,
      "loss": 0.4292,
      "num_input_tokens_seen": 8738432,
      "step": 13410
    },
    {
      "epoch": 7.909787735849057,
      "grad_norm": 2.021859645843506,
      "learning_rate": 7.56830669028898e-06,
      "loss": 0.3657,
      "num_input_tokens_seen": 8742080,
      "step": 13415
    },
    {
      "epoch": 7.912735849056604,
      "grad_norm": 2.4289166927337646,
      "learning_rate": 7.566098979641588e-06,
      "loss": 0.4759,
      "num_input_tokens_seen": 8745120,
      "step": 13420
    },
    {
      "epoch": 7.915683962264151,
      "grad_norm": 4.298847198486328,
      "learning_rate": 7.563890589608427e-06,
      "loss": 0.4181,
      "num_input_tokens_seen": 8750592,
      "step": 13425
    },
    {
      "epoch": 7.918632075471698,
      "grad_norm": 1.7805455923080444,
      "learning_rate": 7.561681520774187e-06,
      "loss": 0.4343,
      "num_input_tokens_seen": 8754432,
      "step": 13430
    },
    {
      "epoch": 7.921580188679245,
      "grad_norm": 1.4597715139389038,
      "learning_rate": 7.559471773723721e-06,
      "loss": 0.5774,
      "num_input_tokens_seen": 8758016,
      "step": 13435
    },
    {
      "epoch": 7.9245283018867925,
      "grad_norm": 1.4705971479415894,
      "learning_rate": 7.557261349042073e-06,
      "loss": 0.3994,
      "num_input_tokens_seen": 8760544,
      "step": 13440
    },
    {
      "epoch": 7.92747641509434,
      "grad_norm": 2.225756883621216,
      "learning_rate": 7.555050247314464e-06,
      "loss": 0.4744,
      "num_input_tokens_seen": 8765792,
      "step": 13445
    },
    {
      "epoch": 7.930424528301887,
      "grad_norm": 2.0553815364837646,
      "learning_rate": 7.552838469126289e-06,
      "loss": 0.3427,
      "num_input_tokens_seen": 8768608,
      "step": 13450
    },
    {
      "epoch": 7.933372641509434,
      "grad_norm": 1.825537085533142,
      "learning_rate": 7.550626015063125e-06,
      "loss": 0.3828,
      "num_input_tokens_seen": 8771680,
      "step": 13455
    },
    {
      "epoch": 7.936320754716981,
      "grad_norm": 2.6737914085388184,
      "learning_rate": 7.548412885710734e-06,
      "loss": 0.3649,
      "num_input_tokens_seen": 8774720,
      "step": 13460
    },
    {
      "epoch": 7.939268867924528,
      "grad_norm": 1.4219367504119873,
      "learning_rate": 7.546199081655048e-06,
      "loss": 0.3616,
      "num_input_tokens_seen": 8778080,
      "step": 13465
    },
    {
      "epoch": 7.942216981132075,
      "grad_norm": 2.9162654876708984,
      "learning_rate": 7.54398460348218e-06,
      "loss": 0.5237,
      "num_input_tokens_seen": 8781568,
      "step": 13470
    },
    {
      "epoch": 7.945165094339623,
      "grad_norm": 3.0545060634613037,
      "learning_rate": 7.541769451778425e-06,
      "loss": 0.4307,
      "num_input_tokens_seen": 8784768,
      "step": 13475
    },
    {
      "epoch": 7.94811320754717,
      "grad_norm": 3.1044540405273438,
      "learning_rate": 7.5395536271302536e-06,
      "loss": 0.3514,
      "num_input_tokens_seen": 8788032,
      "step": 13480
    },
    {
      "epoch": 7.951061320754717,
      "grad_norm": 1.6437549591064453,
      "learning_rate": 7.5373371301243136e-06,
      "loss": 0.3308,
      "num_input_tokens_seen": 8793408,
      "step": 13485
    },
    {
      "epoch": 7.9540094339622645,
      "grad_norm": 1.6758924722671509,
      "learning_rate": 7.535119961347433e-06,
      "loss": 0.3728,
      "num_input_tokens_seen": 8796704,
      "step": 13490
    },
    {
      "epoch": 7.956957547169811,
      "grad_norm": 2.367798089981079,
      "learning_rate": 7.532902121386618e-06,
      "loss": 0.3715,
      "num_input_tokens_seen": 8800448,
      "step": 13495
    },
    {
      "epoch": 7.959905660377358,
      "grad_norm": 3.0903031826019287,
      "learning_rate": 7.530683610829051e-06,
      "loss": 0.5002,
      "num_input_tokens_seen": 8803296,
      "step": 13500
    },
    {
      "epoch": 7.962853773584905,
      "grad_norm": 2.4797983169555664,
      "learning_rate": 7.5284644302620906e-06,
      "loss": 0.4646,
      "num_input_tokens_seen": 8805984,
      "step": 13505
    },
    {
      "epoch": 7.965801886792453,
      "grad_norm": 2.400684118270874,
      "learning_rate": 7.526244580273274e-06,
      "loss": 0.5778,
      "num_input_tokens_seen": 8811072,
      "step": 13510
    },
    {
      "epoch": 7.96875,
      "grad_norm": 3.4902589321136475,
      "learning_rate": 7.524024061450318e-06,
      "loss": 0.3653,
      "num_input_tokens_seen": 8814144,
      "step": 13515
    },
    {
      "epoch": 7.971698113207547,
      "grad_norm": 2.744936466217041,
      "learning_rate": 7.521802874381115e-06,
      "loss": 0.4454,
      "num_input_tokens_seen": 8816928,
      "step": 13520
    },
    {
      "epoch": 7.974646226415095,
      "grad_norm": 2.6063666343688965,
      "learning_rate": 7.519581019653731e-06,
      "loss": 0.4588,
      "num_input_tokens_seen": 8819168,
      "step": 13525
    },
    {
      "epoch": 7.977594339622642,
      "grad_norm": 1.2602781057357788,
      "learning_rate": 7.517358497856413e-06,
      "loss": 0.4142,
      "num_input_tokens_seen": 8822688,
      "step": 13530
    },
    {
      "epoch": 7.980542452830189,
      "grad_norm": 2.7582015991210938,
      "learning_rate": 7.515135309577584e-06,
      "loss": 0.3912,
      "num_input_tokens_seen": 8825952,
      "step": 13535
    },
    {
      "epoch": 7.9834905660377355,
      "grad_norm": 2.1712729930877686,
      "learning_rate": 7.5129114554058425e-06,
      "loss": 0.5284,
      "num_input_tokens_seen": 8829184,
      "step": 13540
    },
    {
      "epoch": 7.986438679245283,
      "grad_norm": 2.7428975105285645,
      "learning_rate": 7.510686935929963e-06,
      "loss": 0.4749,
      "num_input_tokens_seen": 8832768,
      "step": 13545
    },
    {
      "epoch": 7.98938679245283,
      "grad_norm": 4.409809589385986,
      "learning_rate": 7.5084617517388965e-06,
      "loss": 0.3945,
      "num_input_tokens_seen": 8835744,
      "step": 13550
    },
    {
      "epoch": 7.992334905660377,
      "grad_norm": 3.4153990745544434,
      "learning_rate": 7.506235903421771e-06,
      "loss": 0.4586,
      "num_input_tokens_seen": 8839232,
      "step": 13555
    },
    {
      "epoch": 7.995283018867925,
      "grad_norm": 3.569145679473877,
      "learning_rate": 7.504009391567889e-06,
      "loss": 0.3999,
      "num_input_tokens_seen": 8842720,
      "step": 13560
    },
    {
      "epoch": 7.998231132075472,
      "grad_norm": 1.6880728006362915,
      "learning_rate": 7.501782216766729e-06,
      "loss": 0.3557,
      "num_input_tokens_seen": 8845792,
      "step": 13565
    },
    {
      "epoch": 8.0,
      "eval_loss": 0.5099420547485352,
      "eval_runtime": 19.0498,
      "eval_samples_per_second": 89.03,
      "eval_steps_per_second": 22.257,
      "num_input_tokens_seen": 8848168,
      "step": 13568
    },
    {
      "epoch": 8.00117924528302,
      "grad_norm": 2.666106939315796,
      "learning_rate": 7.499554379607944e-06,
      "loss": 0.429,
      "num_input_tokens_seen": 8849256,
      "step": 13570
    },
    {
      "epoch": 8.004127358490566,
      "grad_norm": 1.7458851337432861,
      "learning_rate": 7.497325880681365e-06,
      "loss": 0.3561,
      "num_input_tokens_seen": 8852584,
      "step": 13575
    },
    {
      "epoch": 8.007075471698114,
      "grad_norm": 2.776974678039551,
      "learning_rate": 7.495096720576994e-06,
      "loss": 0.4698,
      "num_input_tokens_seen": 8855496,
      "step": 13580
    },
    {
      "epoch": 8.01002358490566,
      "grad_norm": 2.4449968338012695,
      "learning_rate": 7.492866899885017e-06,
      "loss": 0.3735,
      "num_input_tokens_seen": 8858056,
      "step": 13585
    },
    {
      "epoch": 8.012971698113208,
      "grad_norm": 2.096892833709717,
      "learning_rate": 7.490636419195782e-06,
      "loss": 0.5267,
      "num_input_tokens_seen": 8863240,
      "step": 13590
    },
    {
      "epoch": 8.015919811320755,
      "grad_norm": 2.1281867027282715,
      "learning_rate": 7.488405279099821e-06,
      "loss": 0.4313,
      "num_input_tokens_seen": 8866536,
      "step": 13595
    },
    {
      "epoch": 8.018867924528301,
      "grad_norm": 2.3399698734283447,
      "learning_rate": 7.48617348018784e-06,
      "loss": 0.4359,
      "num_input_tokens_seen": 8869000,
      "step": 13600
    },
    {
      "epoch": 8.02181603773585,
      "grad_norm": 1.100244402885437,
      "learning_rate": 7.4839410230507134e-06,
      "loss": 0.3305,
      "num_input_tokens_seen": 8871656,
      "step": 13605
    },
    {
      "epoch": 8.024764150943396,
      "grad_norm": 1.7875359058380127,
      "learning_rate": 7.481707908279496e-06,
      "loss": 0.3883,
      "num_input_tokens_seen": 8874792,
      "step": 13610
    },
    {
      "epoch": 8.027712264150944,
      "grad_norm": 1.9129972457885742,
      "learning_rate": 7.4794741364654144e-06,
      "loss": 0.2673,
      "num_input_tokens_seen": 8877896,
      "step": 13615
    },
    {
      "epoch": 8.03066037735849,
      "grad_norm": 3.9335618019104004,
      "learning_rate": 7.477239708199871e-06,
      "loss": 0.4106,
      "num_input_tokens_seen": 8881096,
      "step": 13620
    },
    {
      "epoch": 8.033608490566039,
      "grad_norm": 2.187635660171509,
      "learning_rate": 7.475004624074434e-06,
      "loss": 0.362,
      "num_input_tokens_seen": 8883496,
      "step": 13625
    },
    {
      "epoch": 8.036556603773585,
      "grad_norm": 1.756363034248352,
      "learning_rate": 7.4727688846808595e-06,
      "loss": 0.5068,
      "num_input_tokens_seen": 8886856,
      "step": 13630
    },
    {
      "epoch": 8.039504716981131,
      "grad_norm": 2.8471572399139404,
      "learning_rate": 7.4705324906110654e-06,
      "loss": 0.4106,
      "num_input_tokens_seen": 8890280,
      "step": 13635
    },
    {
      "epoch": 8.04245283018868,
      "grad_norm": 1.8761793375015259,
      "learning_rate": 7.4682954424571466e-06,
      "loss": 0.4369,
      "num_input_tokens_seen": 8893384,
      "step": 13640
    },
    {
      "epoch": 8.045400943396226,
      "grad_norm": 2.593594551086426,
      "learning_rate": 7.466057740811372e-06,
      "loss": 0.5718,
      "num_input_tokens_seen": 8896136,
      "step": 13645
    },
    {
      "epoch": 8.048349056603774,
      "grad_norm": 1.4992259740829468,
      "learning_rate": 7.463819386266182e-06,
      "loss": 0.4577,
      "num_input_tokens_seen": 8899944,
      "step": 13650
    },
    {
      "epoch": 8.05129716981132,
      "grad_norm": 2.8070898056030273,
      "learning_rate": 7.461580379414191e-06,
      "loss": 0.3305,
      "num_input_tokens_seen": 8903400,
      "step": 13655
    },
    {
      "epoch": 8.054245283018869,
      "grad_norm": 4.676871299743652,
      "learning_rate": 7.459340720848187e-06,
      "loss": 0.3973,
      "num_input_tokens_seen": 8906856,
      "step": 13660
    },
    {
      "epoch": 8.057193396226415,
      "grad_norm": 1.3727142810821533,
      "learning_rate": 7.457100411161128e-06,
      "loss": 0.3616,
      "num_input_tokens_seen": 8909832,
      "step": 13665
    },
    {
      "epoch": 8.060141509433961,
      "grad_norm": 5.175848007202148,
      "learning_rate": 7.454859450946144e-06,
      "loss": 0.5317,
      "num_input_tokens_seen": 8912904,
      "step": 13670
    },
    {
      "epoch": 8.06308962264151,
      "grad_norm": 3.9332563877105713,
      "learning_rate": 7.4526178407965396e-06,
      "loss": 0.4443,
      "num_input_tokens_seen": 8916200,
      "step": 13675
    },
    {
      "epoch": 8.066037735849056,
      "grad_norm": 2.150294065475464,
      "learning_rate": 7.450375581305794e-06,
      "loss": 0.4242,
      "num_input_tokens_seen": 8919272,
      "step": 13680
    },
    {
      "epoch": 8.068985849056604,
      "grad_norm": 1.47675621509552,
      "learning_rate": 7.448132673067552e-06,
      "loss": 0.3985,
      "num_input_tokens_seen": 8923272,
      "step": 13685
    },
    {
      "epoch": 8.07193396226415,
      "grad_norm": 2.911299467086792,
      "learning_rate": 7.445889116675634e-06,
      "loss": 0.6907,
      "num_input_tokens_seen": 8927528,
      "step": 13690
    },
    {
      "epoch": 8.074882075471699,
      "grad_norm": 2.4006049633026123,
      "learning_rate": 7.443644912724031e-06,
      "loss": 0.3503,
      "num_input_tokens_seen": 8929992,
      "step": 13695
    },
    {
      "epoch": 8.077830188679245,
      "grad_norm": 2.804715633392334,
      "learning_rate": 7.441400061806907e-06,
      "loss": 0.4373,
      "num_input_tokens_seen": 8933864,
      "step": 13700
    },
    {
      "epoch": 8.080778301886792,
      "grad_norm": 1.6213233470916748,
      "learning_rate": 7.439154564518592e-06,
      "loss": 0.5058,
      "num_input_tokens_seen": 8937832,
      "step": 13705
    },
    {
      "epoch": 8.08372641509434,
      "grad_norm": 1.528058648109436,
      "learning_rate": 7.436908421453597e-06,
      "loss": 0.3998,
      "num_input_tokens_seen": 8941224,
      "step": 13710
    },
    {
      "epoch": 8.086674528301886,
      "grad_norm": 1.7929173707962036,
      "learning_rate": 7.434661633206593e-06,
      "loss": 0.3248,
      "num_input_tokens_seen": 8944168,
      "step": 13715
    },
    {
      "epoch": 8.089622641509434,
      "grad_norm": 3.8835673332214355,
      "learning_rate": 7.4324142003724286e-06,
      "loss": 0.3541,
      "num_input_tokens_seen": 8946792,
      "step": 13720
    },
    {
      "epoch": 8.09257075471698,
      "grad_norm": 1.6249452829360962,
      "learning_rate": 7.430166123546122e-06,
      "loss": 0.4109,
      "num_input_tokens_seen": 8950216,
      "step": 13725
    },
    {
      "epoch": 8.095518867924529,
      "grad_norm": 1.541869878768921,
      "learning_rate": 7.427917403322862e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 8954312,
      "step": 13730
    },
    {
      "epoch": 8.098466981132075,
      "grad_norm": 2.542351245880127,
      "learning_rate": 7.425668040298003e-06,
      "loss": 0.4719,
      "num_input_tokens_seen": 8958376,
      "step": 13735
    },
    {
      "epoch": 8.101415094339623,
      "grad_norm": 2.4646389484405518,
      "learning_rate": 7.4234180350670785e-06,
      "loss": 0.336,
      "num_input_tokens_seen": 8961384,
      "step": 13740
    },
    {
      "epoch": 8.10436320754717,
      "grad_norm": 2.8588569164276123,
      "learning_rate": 7.421167388225785e-06,
      "loss": 0.3946,
      "num_input_tokens_seen": 8964264,
      "step": 13745
    },
    {
      "epoch": 8.107311320754716,
      "grad_norm": 1.9972561597824097,
      "learning_rate": 7.41891610036999e-06,
      "loss": 0.4015,
      "num_input_tokens_seen": 8966728,
      "step": 13750
    },
    {
      "epoch": 8.110259433962264,
      "grad_norm": 2.407116174697876,
      "learning_rate": 7.416664172095732e-06,
      "loss": 0.2923,
      "num_input_tokens_seen": 8969544,
      "step": 13755
    },
    {
      "epoch": 8.11320754716981,
      "grad_norm": 2.901962995529175,
      "learning_rate": 7.414411603999221e-06,
      "loss": 0.3625,
      "num_input_tokens_seen": 8972776,
      "step": 13760
    },
    {
      "epoch": 8.116155660377359,
      "grad_norm": 3.874617099761963,
      "learning_rate": 7.4121583966768295e-06,
      "loss": 0.3316,
      "num_input_tokens_seen": 8975624,
      "step": 13765
    },
    {
      "epoch": 8.119103773584905,
      "grad_norm": 1.8574072122573853,
      "learning_rate": 7.409904550725109e-06,
      "loss": 0.4319,
      "num_input_tokens_seen": 8980392,
      "step": 13770
    },
    {
      "epoch": 8.122051886792454,
      "grad_norm": 1.4014862775802612,
      "learning_rate": 7.407650066740771e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 8984008,
      "step": 13775
    },
    {
      "epoch": 8.125,
      "grad_norm": 4.182077407836914,
      "learning_rate": 7.405394945320702e-06,
      "loss": 0.3382,
      "num_input_tokens_seen": 8987304,
      "step": 13780
    },
    {
      "epoch": 8.127948113207546,
      "grad_norm": 2.386167526245117,
      "learning_rate": 7.403139187061955e-06,
      "loss": 0.4656,
      "num_input_tokens_seen": 8990216,
      "step": 13785
    },
    {
      "epoch": 8.130896226415095,
      "grad_norm": 3.9744343757629395,
      "learning_rate": 7.400882792561752e-06,
      "loss": 0.4925,
      "num_input_tokens_seen": 8993480,
      "step": 13790
    },
    {
      "epoch": 8.133844339622641,
      "grad_norm": 3.4291646480560303,
      "learning_rate": 7.39862576241748e-06,
      "loss": 0.3748,
      "num_input_tokens_seen": 8999112,
      "step": 13795
    },
    {
      "epoch": 8.13679245283019,
      "grad_norm": 2.106797695159912,
      "learning_rate": 7.396368097226703e-06,
      "loss": 0.4129,
      "num_input_tokens_seen": 9002824,
      "step": 13800
    },
    {
      "epoch": 8.139740566037736,
      "grad_norm": 3.0510683059692383,
      "learning_rate": 7.394109797587144e-06,
      "loss": 0.4775,
      "num_input_tokens_seen": 9005736,
      "step": 13805
    },
    {
      "epoch": 8.142688679245284,
      "grad_norm": 2.3914542198181152,
      "learning_rate": 7.3918508640966956e-06,
      "loss": 0.4175,
      "num_input_tokens_seen": 9009416,
      "step": 13810
    },
    {
      "epoch": 8.14563679245283,
      "grad_norm": 1.9719914197921753,
      "learning_rate": 7.389591297353424e-06,
      "loss": 0.4633,
      "num_input_tokens_seen": 9012456,
      "step": 13815
    },
    {
      "epoch": 8.148584905660377,
      "grad_norm": 2.599912643432617,
      "learning_rate": 7.3873310979555565e-06,
      "loss": 0.376,
      "num_input_tokens_seen": 9015528,
      "step": 13820
    },
    {
      "epoch": 8.151533018867925,
      "grad_norm": 2.6653919219970703,
      "learning_rate": 7.385070266501495e-06,
      "loss": 0.3802,
      "num_input_tokens_seen": 9019112,
      "step": 13825
    },
    {
      "epoch": 8.154481132075471,
      "grad_norm": 2.2161753177642822,
      "learning_rate": 7.382808803589798e-06,
      "loss": 0.3897,
      "num_input_tokens_seen": 9021736,
      "step": 13830
    },
    {
      "epoch": 8.15742924528302,
      "grad_norm": 1.1827857494354248,
      "learning_rate": 7.380546709819204e-06,
      "loss": 0.4062,
      "num_input_tokens_seen": 9024936,
      "step": 13835
    },
    {
      "epoch": 8.160377358490566,
      "grad_norm": 2.122030258178711,
      "learning_rate": 7.378283985788608e-06,
      "loss": 0.372,
      "num_input_tokens_seen": 9027560,
      "step": 13840
    },
    {
      "epoch": 8.163325471698114,
      "grad_norm": 2.7912421226501465,
      "learning_rate": 7.376020632097076e-06,
      "loss": 0.4653,
      "num_input_tokens_seen": 9030504,
      "step": 13845
    },
    {
      "epoch": 8.16627358490566,
      "grad_norm": 1.2989771366119385,
      "learning_rate": 7.373756649343841e-06,
      "loss": 0.2937,
      "num_input_tokens_seen": 9033704,
      "step": 13850
    },
    {
      "epoch": 8.169221698113208,
      "grad_norm": 1.4684436321258545,
      "learning_rate": 7.371492038128305e-06,
      "loss": 0.3975,
      "num_input_tokens_seen": 9037800,
      "step": 13855
    },
    {
      "epoch": 8.172169811320755,
      "grad_norm": 2.1536645889282227,
      "learning_rate": 7.36922679905003e-06,
      "loss": 0.3701,
      "num_input_tokens_seen": 9041160,
      "step": 13860
    },
    {
      "epoch": 8.175117924528301,
      "grad_norm": 2.0741193294525146,
      "learning_rate": 7.366960932708749e-06,
      "loss": 0.3279,
      "num_input_tokens_seen": 9044200,
      "step": 13865
    },
    {
      "epoch": 8.17806603773585,
      "grad_norm": 2.3600454330444336,
      "learning_rate": 7.364694439704361e-06,
      "loss": 0.348,
      "num_input_tokens_seen": 9046856,
      "step": 13870
    },
    {
      "epoch": 8.181014150943396,
      "grad_norm": 4.38322639465332,
      "learning_rate": 7.3624273206369264e-06,
      "loss": 0.2514,
      "num_input_tokens_seen": 9049288,
      "step": 13875
    },
    {
      "epoch": 8.183962264150944,
      "grad_norm": 3.125810384750366,
      "learning_rate": 7.360159576106681e-06,
      "loss": 0.3868,
      "num_input_tokens_seen": 9051976,
      "step": 13880
    },
    {
      "epoch": 8.18691037735849,
      "grad_norm": 1.6389169692993164,
      "learning_rate": 7.357891206714014e-06,
      "loss": 0.2955,
      "num_input_tokens_seen": 9055336,
      "step": 13885
    },
    {
      "epoch": 8.189858490566039,
      "grad_norm": 3.010446071624756,
      "learning_rate": 7.355622213059487e-06,
      "loss": 0.3901,
      "num_input_tokens_seen": 9058120,
      "step": 13890
    },
    {
      "epoch": 8.192806603773585,
      "grad_norm": 2.7184367179870605,
      "learning_rate": 7.353352595743829e-06,
      "loss": 0.2998,
      "num_input_tokens_seen": 9061160,
      "step": 13895
    },
    {
      "epoch": 8.195754716981131,
      "grad_norm": 1.5839126110076904,
      "learning_rate": 7.351082355367928e-06,
      "loss": 0.4522,
      "num_input_tokens_seen": 9064552,
      "step": 13900
    },
    {
      "epoch": 8.19870283018868,
      "grad_norm": 2.3468291759490967,
      "learning_rate": 7.34881149253284e-06,
      "loss": 0.3219,
      "num_input_tokens_seen": 9068488,
      "step": 13905
    },
    {
      "epoch": 8.201650943396226,
      "grad_norm": 2.5472218990325928,
      "learning_rate": 7.346540007839787e-06,
      "loss": 0.4691,
      "num_input_tokens_seen": 9071752,
      "step": 13910
    },
    {
      "epoch": 8.204599056603774,
      "grad_norm": 5.368915557861328,
      "learning_rate": 7.344267901890154e-06,
      "loss": 0.4471,
      "num_input_tokens_seen": 9074120,
      "step": 13915
    },
    {
      "epoch": 8.20754716981132,
      "grad_norm": 3.0006468296051025,
      "learning_rate": 7.341995175285491e-06,
      "loss": 0.4004,
      "num_input_tokens_seen": 9076968,
      "step": 13920
    },
    {
      "epoch": 8.210495283018869,
      "grad_norm": 2.642289638519287,
      "learning_rate": 7.339721828627512e-06,
      "loss": 0.4406,
      "num_input_tokens_seen": 9080264,
      "step": 13925
    },
    {
      "epoch": 8.213443396226415,
      "grad_norm": 3.0328216552734375,
      "learning_rate": 7.337447862518096e-06,
      "loss": 0.2811,
      "num_input_tokens_seen": 9082984,
      "step": 13930
    },
    {
      "epoch": 8.216391509433961,
      "grad_norm": 1.7147077322006226,
      "learning_rate": 7.335173277559282e-06,
      "loss": 0.3452,
      "num_input_tokens_seen": 9086344,
      "step": 13935
    },
    {
      "epoch": 8.21933962264151,
      "grad_norm": 2.7895584106445312,
      "learning_rate": 7.332898074353281e-06,
      "loss": 0.3691,
      "num_input_tokens_seen": 9089448,
      "step": 13940
    },
    {
      "epoch": 8.222287735849056,
      "grad_norm": 1.8917620182037354,
      "learning_rate": 7.330622253502461e-06,
      "loss": 0.2899,
      "num_input_tokens_seen": 9092680,
      "step": 13945
    },
    {
      "epoch": 8.225235849056604,
      "grad_norm": 1.8764305114746094,
      "learning_rate": 7.3283458156093534e-06,
      "loss": 0.2974,
      "num_input_tokens_seen": 9095304,
      "step": 13950
    },
    {
      "epoch": 8.22818396226415,
      "grad_norm": 1.8710672855377197,
      "learning_rate": 7.326068761276657e-06,
      "loss": 0.3705,
      "num_input_tokens_seen": 9098504,
      "step": 13955
    },
    {
      "epoch": 8.231132075471699,
      "grad_norm": 2.5115270614624023,
      "learning_rate": 7.323791091107231e-06,
      "loss": 0.36,
      "num_input_tokens_seen": 9101512,
      "step": 13960
    },
    {
      "epoch": 8.234080188679245,
      "grad_norm": 5.925174236297607,
      "learning_rate": 7.3215128057040986e-06,
      "loss": 0.4934,
      "num_input_tokens_seen": 9103880,
      "step": 13965
    },
    {
      "epoch": 8.237028301886792,
      "grad_norm": 6.538854598999023,
      "learning_rate": 7.319233905670447e-06,
      "loss": 0.4085,
      "num_input_tokens_seen": 9106440,
      "step": 13970
    },
    {
      "epoch": 8.23997641509434,
      "grad_norm": 2.83817720413208,
      "learning_rate": 7.316954391609622e-06,
      "loss": 0.3234,
      "num_input_tokens_seen": 9109576,
      "step": 13975
    },
    {
      "epoch": 8.242924528301886,
      "grad_norm": 2.789663791656494,
      "learning_rate": 7.314674264125137e-06,
      "loss": 0.4577,
      "num_input_tokens_seen": 9113128,
      "step": 13980
    },
    {
      "epoch": 8.245872641509434,
      "grad_norm": 2.0470354557037354,
      "learning_rate": 7.312393523820665e-06,
      "loss": 0.46,
      "num_input_tokens_seen": 9116552,
      "step": 13985
    },
    {
      "epoch": 8.24882075471698,
      "grad_norm": 2.7704360485076904,
      "learning_rate": 7.310112171300041e-06,
      "loss": 0.4312,
      "num_input_tokens_seen": 9119656,
      "step": 13990
    },
    {
      "epoch": 8.251768867924529,
      "grad_norm": 3.5966460704803467,
      "learning_rate": 7.307830207167263e-06,
      "loss": 0.4338,
      "num_input_tokens_seen": 9122568,
      "step": 13995
    },
    {
      "epoch": 8.254716981132075,
      "grad_norm": 4.281689167022705,
      "learning_rate": 7.305547632026493e-06,
      "loss": 0.4016,
      "num_input_tokens_seen": 9125160,
      "step": 14000
    },
    {
      "epoch": 8.257665094339623,
      "grad_norm": 3.0443994998931885,
      "learning_rate": 7.3032644464820515e-06,
      "loss": 0.4734,
      "num_input_tokens_seen": 9128520,
      "step": 14005
    },
    {
      "epoch": 8.26061320754717,
      "grad_norm": 2.897758960723877,
      "learning_rate": 7.30098065113842e-06,
      "loss": 0.3919,
      "num_input_tokens_seen": 9132104,
      "step": 14010
    },
    {
      "epoch": 8.263561320754716,
      "grad_norm": 1.5462449789047241,
      "learning_rate": 7.298696246600244e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 9135432,
      "step": 14015
    },
    {
      "epoch": 8.266509433962264,
      "grad_norm": 2.461265802383423,
      "learning_rate": 7.2964112334723315e-06,
      "loss": 0.3581,
      "num_input_tokens_seen": 9139944,
      "step": 14020
    },
    {
      "epoch": 8.26945754716981,
      "grad_norm": 2.317335605621338,
      "learning_rate": 7.294125612359647e-06,
      "loss": 0.389,
      "num_input_tokens_seen": 9143208,
      "step": 14025
    },
    {
      "epoch": 8.272405660377359,
      "grad_norm": 2.1711039543151855,
      "learning_rate": 7.291839383867318e-06,
      "loss": 0.2363,
      "num_input_tokens_seen": 9146216,
      "step": 14030
    },
    {
      "epoch": 8.275353773584905,
      "grad_norm": 1.928035855293274,
      "learning_rate": 7.289552548600638e-06,
      "loss": 0.3073,
      "num_input_tokens_seen": 9149288,
      "step": 14035
    },
    {
      "epoch": 8.278301886792454,
      "grad_norm": 3.0396032333374023,
      "learning_rate": 7.287265107165052e-06,
      "loss": 0.3135,
      "num_input_tokens_seen": 9153448,
      "step": 14040
    },
    {
      "epoch": 8.28125,
      "grad_norm": 2.2305355072021484,
      "learning_rate": 7.284977060166171e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 9157832,
      "step": 14045
    },
    {
      "epoch": 8.284198113207546,
      "grad_norm": 2.923834800720215,
      "learning_rate": 7.282688408209766e-06,
      "loss": 0.4026,
      "num_input_tokens_seen": 9161096,
      "step": 14050
    },
    {
      "epoch": 8.287146226415095,
      "grad_norm": 3.395155429840088,
      "learning_rate": 7.2803991519017655e-06,
      "loss": 0.4475,
      "num_input_tokens_seen": 9164104,
      "step": 14055
    },
    {
      "epoch": 8.290094339622641,
      "grad_norm": 2.713153600692749,
      "learning_rate": 7.2781092918482634e-06,
      "loss": 0.4329,
      "num_input_tokens_seen": 9166728,
      "step": 14060
    },
    {
      "epoch": 8.29304245283019,
      "grad_norm": 2.379946708679199,
      "learning_rate": 7.275818828655508e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 9170056,
      "step": 14065
    },
    {
      "epoch": 8.295990566037736,
      "grad_norm": 3.942443609237671,
      "learning_rate": 7.27352776292991e-06,
      "loss": 0.3621,
      "num_input_tokens_seen": 9173192,
      "step": 14070
    },
    {
      "epoch": 8.298938679245284,
      "grad_norm": 2.1148619651794434,
      "learning_rate": 7.271236095278036e-06,
      "loss": 0.4185,
      "num_input_tokens_seen": 9176616,
      "step": 14075
    },
    {
      "epoch": 8.30188679245283,
      "grad_norm": 3.0360612869262695,
      "learning_rate": 7.2689438263066195e-06,
      "loss": 0.4694,
      "num_input_tokens_seen": 9179976,
      "step": 14080
    },
    {
      "epoch": 8.304834905660377,
      "grad_norm": 3.7810540199279785,
      "learning_rate": 7.266650956622546e-06,
      "loss": 0.4752,
      "num_input_tokens_seen": 9183080,
      "step": 14085
    },
    {
      "epoch": 8.307783018867925,
      "grad_norm": 1.861411690711975,
      "learning_rate": 7.2643574868328625e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 9185416,
      "step": 14090
    },
    {
      "epoch": 8.310731132075471,
      "grad_norm": 2.0562405586242676,
      "learning_rate": 7.262063417544776e-06,
      "loss": 0.3161,
      "num_input_tokens_seen": 9188392,
      "step": 14095
    },
    {
      "epoch": 8.31367924528302,
      "grad_norm": 4.4509596824646,
      "learning_rate": 7.25976874936565e-06,
      "loss": 0.4071,
      "num_input_tokens_seen": 9192360,
      "step": 14100
    },
    {
      "epoch": 8.316627358490566,
      "grad_norm": 4.977168083190918,
      "learning_rate": 7.257473482903009e-06,
      "loss": 0.4578,
      "num_input_tokens_seen": 9194728,
      "step": 14105
    },
    {
      "epoch": 8.319575471698114,
      "grad_norm": 2.217033863067627,
      "learning_rate": 7.255177618764534e-06,
      "loss": 0.4118,
      "num_input_tokens_seen": 9198216,
      "step": 14110
    },
    {
      "epoch": 8.32252358490566,
      "grad_norm": 2.2413735389709473,
      "learning_rate": 7.252881157558065e-06,
      "loss": 0.4386,
      "num_input_tokens_seen": 9201768,
      "step": 14115
    },
    {
      "epoch": 8.325471698113208,
      "grad_norm": 3.2492992877960205,
      "learning_rate": 7.250584099891602e-06,
      "loss": 0.3745,
      "num_input_tokens_seen": 9204712,
      "step": 14120
    },
    {
      "epoch": 8.328419811320755,
      "grad_norm": 3.902268648147583,
      "learning_rate": 7.248286446373296e-06,
      "loss": 0.3689,
      "num_input_tokens_seen": 9207112,
      "step": 14125
    },
    {
      "epoch": 8.331367924528301,
      "grad_norm": 1.6623175144195557,
      "learning_rate": 7.245988197611466e-06,
      "loss": 0.3721,
      "num_input_tokens_seen": 9210568,
      "step": 14130
    },
    {
      "epoch": 8.33431603773585,
      "grad_norm": 2.9371471405029297,
      "learning_rate": 7.2436893542145805e-06,
      "loss": 0.3289,
      "num_input_tokens_seen": 9214248,
      "step": 14135
    },
    {
      "epoch": 8.337264150943396,
      "grad_norm": 2.0764291286468506,
      "learning_rate": 7.241389916791269e-06,
      "loss": 0.4597,
      "num_input_tokens_seen": 9218376,
      "step": 14140
    },
    {
      "epoch": 8.340212264150944,
      "grad_norm": 1.462090015411377,
      "learning_rate": 7.239089885950317e-06,
      "loss": 0.4129,
      "num_input_tokens_seen": 9222152,
      "step": 14145
    },
    {
      "epoch": 8.34316037735849,
      "grad_norm": 1.7906250953674316,
      "learning_rate": 7.236789262300667e-06,
      "loss": 0.3677,
      "num_input_tokens_seen": 9225448,
      "step": 14150
    },
    {
      "epoch": 8.346108490566039,
      "grad_norm": 1.7789106369018555,
      "learning_rate": 7.23448804645142e-06,
      "loss": 0.4122,
      "num_input_tokens_seen": 9228904,
      "step": 14155
    },
    {
      "epoch": 8.349056603773585,
      "grad_norm": 2.3959455490112305,
      "learning_rate": 7.232186239011834e-06,
      "loss": 0.3969,
      "num_input_tokens_seen": 9232456,
      "step": 14160
    },
    {
      "epoch": 8.352004716981131,
      "grad_norm": 3.3744118213653564,
      "learning_rate": 7.2298838405913195e-06,
      "loss": 0.3695,
      "num_input_tokens_seen": 9235496,
      "step": 14165
    },
    {
      "epoch": 8.35495283018868,
      "grad_norm": 2.167912721633911,
      "learning_rate": 7.227580851799448e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 9238184,
      "step": 14170
    },
    {
      "epoch": 8.357900943396226,
      "grad_norm": 2.4132373332977295,
      "learning_rate": 7.2252772732459455e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 9240680,
      "step": 14175
    },
    {
      "epoch": 8.360849056603774,
      "grad_norm": 6.015293121337891,
      "learning_rate": 7.222973105540696e-06,
      "loss": 0.3944,
      "num_input_tokens_seen": 9243944,
      "step": 14180
    },
    {
      "epoch": 8.36379716981132,
      "grad_norm": 1.475831151008606,
      "learning_rate": 7.2206683492937345e-06,
      "loss": 0.3471,
      "num_input_tokens_seen": 9246792,
      "step": 14185
    },
    {
      "epoch": 8.366745283018869,
      "grad_norm": 1.8131763935089111,
      "learning_rate": 7.218363005115259e-06,
      "loss": 0.3932,
      "num_input_tokens_seen": 9251048,
      "step": 14190
    },
    {
      "epoch": 8.369693396226415,
      "grad_norm": 2.7515087127685547,
      "learning_rate": 7.216057073615617e-06,
      "loss": 0.4088,
      "num_input_tokens_seen": 9254856,
      "step": 14195
    },
    {
      "epoch": 8.372641509433961,
      "grad_norm": 4.056880950927734,
      "learning_rate": 7.21375055540531e-06,
      "loss": 0.4401,
      "num_input_tokens_seen": 9257832,
      "step": 14200
    },
    {
      "epoch": 8.37558962264151,
      "grad_norm": 2.2605767250061035,
      "learning_rate": 7.211443451095007e-06,
      "loss": 0.3207,
      "num_input_tokens_seen": 9261352,
      "step": 14205
    },
    {
      "epoch": 8.378537735849056,
      "grad_norm": 1.5928245782852173,
      "learning_rate": 7.2091357612955185e-06,
      "loss": 0.3891,
      "num_input_tokens_seen": 9265096,
      "step": 14210
    },
    {
      "epoch": 8.381485849056604,
      "grad_norm": 2.0999646186828613,
      "learning_rate": 7.206827486617816e-06,
      "loss": 0.3951,
      "num_input_tokens_seen": 9268904,
      "step": 14215
    },
    {
      "epoch": 8.38443396226415,
      "grad_norm": 3.4141688346862793,
      "learning_rate": 7.204518627673026e-06,
      "loss": 0.5359,
      "num_input_tokens_seen": 9271464,
      "step": 14220
    },
    {
      "epoch": 8.387382075471699,
      "grad_norm": 2.6523115634918213,
      "learning_rate": 7.202209185072428e-06,
      "loss": 0.549,
      "num_input_tokens_seen": 9274280,
      "step": 14225
    },
    {
      "epoch": 8.390330188679245,
      "grad_norm": 2.240976572036743,
      "learning_rate": 7.199899159427457e-06,
      "loss": 0.3769,
      "num_input_tokens_seen": 9277224,
      "step": 14230
    },
    {
      "epoch": 8.393278301886792,
      "grad_norm": 1.6264857053756714,
      "learning_rate": 7.1975885513497035e-06,
      "loss": 0.3814,
      "num_input_tokens_seen": 9281224,
      "step": 14235
    },
    {
      "epoch": 8.39622641509434,
      "grad_norm": 4.339727401733398,
      "learning_rate": 7.195277361450909e-06,
      "loss": 0.4956,
      "num_input_tokens_seen": 9283912,
      "step": 14240
    },
    {
      "epoch": 8.399174528301886,
      "grad_norm": 2.5912351608276367,
      "learning_rate": 7.192965590342973e-06,
      "loss": 0.628,
      "num_input_tokens_seen": 9288616,
      "step": 14245
    },
    {
      "epoch": 8.402122641509434,
      "grad_norm": 2.866222381591797,
      "learning_rate": 7.190653238637945e-06,
      "loss": 0.493,
      "num_input_tokens_seen": 9291400,
      "step": 14250
    },
    {
      "epoch": 8.40507075471698,
      "grad_norm": 3.4372546672821045,
      "learning_rate": 7.18834030694803e-06,
      "loss": 0.4921,
      "num_input_tokens_seen": 9294632,
      "step": 14255
    },
    {
      "epoch": 8.408018867924529,
      "grad_norm": 3.0348970890045166,
      "learning_rate": 7.186026795885589e-06,
      "loss": 0.4218,
      "num_input_tokens_seen": 9298664,
      "step": 14260
    },
    {
      "epoch": 8.410966981132075,
      "grad_norm": 2.5941033363342285,
      "learning_rate": 7.183712706063133e-06,
      "loss": 0.3715,
      "num_input_tokens_seen": 9301544,
      "step": 14265
    },
    {
      "epoch": 8.413915094339623,
      "grad_norm": 2.5344462394714355,
      "learning_rate": 7.1813980380933255e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 9304648,
      "step": 14270
    },
    {
      "epoch": 8.41686320754717,
      "grad_norm": 0.8082200288772583,
      "learning_rate": 7.179082792588986e-06,
      "loss": 0.3954,
      "num_input_tokens_seen": 9307752,
      "step": 14275
    },
    {
      "epoch": 8.419811320754716,
      "grad_norm": 2.5596089363098145,
      "learning_rate": 7.176766970163087e-06,
      "loss": 0.3884,
      "num_input_tokens_seen": 9310472,
      "step": 14280
    },
    {
      "epoch": 8.422759433962264,
      "grad_norm": 2.9833126068115234,
      "learning_rate": 7.17445057142875e-06,
      "loss": 0.3593,
      "num_input_tokens_seen": 9314312,
      "step": 14285
    },
    {
      "epoch": 8.42570754716981,
      "grad_norm": 3.1281986236572266,
      "learning_rate": 7.172133596999253e-06,
      "loss": 0.3455,
      "num_input_tokens_seen": 9317704,
      "step": 14290
    },
    {
      "epoch": 8.428655660377359,
      "grad_norm": 2.5291125774383545,
      "learning_rate": 7.1698160474880255e-06,
      "loss": 0.5131,
      "num_input_tokens_seen": 9321000,
      "step": 14295
    },
    {
      "epoch": 8.431603773584905,
      "grad_norm": 1.9944555759429932,
      "learning_rate": 7.167497923508648e-06,
      "loss": 0.3584,
      "num_input_tokens_seen": 9323688,
      "step": 14300
    },
    {
      "epoch": 8.434551886792454,
      "grad_norm": 2.2621262073516846,
      "learning_rate": 7.165179225674854e-06,
      "loss": 0.4258,
      "num_input_tokens_seen": 9326568,
      "step": 14305
    },
    {
      "epoch": 8.4375,
      "grad_norm": 2.481161594390869,
      "learning_rate": 7.1628599546005276e-06,
      "loss": 0.338,
      "num_input_tokens_seen": 9329224,
      "step": 14310
    },
    {
      "epoch": 8.440448113207546,
      "grad_norm": 1.8259048461914062,
      "learning_rate": 7.160540110899708e-06,
      "loss": 0.3929,
      "num_input_tokens_seen": 9332360,
      "step": 14315
    },
    {
      "epoch": 8.443396226415095,
      "grad_norm": 4.227758407592773,
      "learning_rate": 7.158219695186582e-06,
      "loss": 0.5698,
      "num_input_tokens_seen": 9335176,
      "step": 14320
    },
    {
      "epoch": 8.446344339622641,
      "grad_norm": 2.5277490615844727,
      "learning_rate": 7.1558987080754905e-06,
      "loss": 0.456,
      "num_input_tokens_seen": 9339624,
      "step": 14325
    },
    {
      "epoch": 8.44929245283019,
      "grad_norm": 1.8370674848556519,
      "learning_rate": 7.1535771501809245e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 9342696,
      "step": 14330
    },
    {
      "epoch": 8.452240566037736,
      "grad_norm": 3.5850014686584473,
      "learning_rate": 7.151255022117527e-06,
      "loss": 0.3994,
      "num_input_tokens_seen": 9345000,
      "step": 14335
    },
    {
      "epoch": 8.455188679245284,
      "grad_norm": 1.6765283346176147,
      "learning_rate": 7.148932324500091e-06,
      "loss": 0.2395,
      "num_input_tokens_seen": 9348072,
      "step": 14340
    },
    {
      "epoch": 8.45813679245283,
      "grad_norm": 2.8801708221435547,
      "learning_rate": 7.146609057943559e-06,
      "loss": 0.3975,
      "num_input_tokens_seen": 9351592,
      "step": 14345
    },
    {
      "epoch": 8.461084905660377,
      "grad_norm": 3.0807976722717285,
      "learning_rate": 7.14428522306303e-06,
      "loss": 0.4588,
      "num_input_tokens_seen": 9354312,
      "step": 14350
    },
    {
      "epoch": 8.464033018867925,
      "grad_norm": 1.7765644788742065,
      "learning_rate": 7.141960820473745e-06,
      "loss": 0.302,
      "num_input_tokens_seen": 9357288,
      "step": 14355
    },
    {
      "epoch": 8.466981132075471,
      "grad_norm": 2.011965751647949,
      "learning_rate": 7.139635850791102e-06,
      "loss": 0.4624,
      "num_input_tokens_seen": 9360040,
      "step": 14360
    },
    {
      "epoch": 8.46992924528302,
      "grad_norm": 2.324897527694702,
      "learning_rate": 7.137310314630647e-06,
      "loss": 0.4772,
      "num_input_tokens_seen": 9363112,
      "step": 14365
    },
    {
      "epoch": 8.472877358490566,
      "grad_norm": 3.6103196144104004,
      "learning_rate": 7.134984212608074e-06,
      "loss": 0.3807,
      "num_input_tokens_seen": 9369000,
      "step": 14370
    },
    {
      "epoch": 8.475825471698114,
      "grad_norm": 2.622830629348755,
      "learning_rate": 7.13265754533923e-06,
      "loss": 0.3546,
      "num_input_tokens_seen": 9372264,
      "step": 14375
    },
    {
      "epoch": 8.47877358490566,
      "grad_norm": 13.428444862365723,
      "learning_rate": 7.130330313440109e-06,
      "loss": 0.4717,
      "num_input_tokens_seen": 9375176,
      "step": 14380
    },
    {
      "epoch": 8.481721698113208,
      "grad_norm": 2.7904152870178223,
      "learning_rate": 7.128002517526856e-06,
      "loss": 0.3451,
      "num_input_tokens_seen": 9378088,
      "step": 14385
    },
    {
      "epoch": 8.484669811320755,
      "grad_norm": 2.313046455383301,
      "learning_rate": 7.1256741582157654e-06,
      "loss": 0.3807,
      "num_input_tokens_seen": 9381448,
      "step": 14390
    },
    {
      "epoch": 8.487617924528301,
      "grad_norm": 1.846826195716858,
      "learning_rate": 7.123345236123282e-06,
      "loss": 0.3888,
      "num_input_tokens_seen": 9385576,
      "step": 14395
    },
    {
      "epoch": 8.49056603773585,
      "grad_norm": 2.162078857421875,
      "learning_rate": 7.121015751865994e-06,
      "loss": 0.3748,
      "num_input_tokens_seen": 9388296,
      "step": 14400
    },
    {
      "epoch": 8.493514150943396,
      "grad_norm": 1.653348684310913,
      "learning_rate": 7.118685706060645e-06,
      "loss": 0.4586,
      "num_input_tokens_seen": 9392552,
      "step": 14405
    },
    {
      "epoch": 8.496462264150944,
      "grad_norm": 1.6599537134170532,
      "learning_rate": 7.116355099324126e-06,
      "loss": 0.3036,
      "num_input_tokens_seen": 9396296,
      "step": 14410
    },
    {
      "epoch": 8.49941037735849,
      "grad_norm": 3.6936824321746826,
      "learning_rate": 7.114023932273471e-06,
      "loss": 0.4239,
      "num_input_tokens_seen": 9398408,
      "step": 14415
    },
    {
      "epoch": 8.502358490566039,
      "grad_norm": 1.5160828828811646,
      "learning_rate": 7.111692205525871e-06,
      "loss": 0.3825,
      "num_input_tokens_seen": 9401352,
      "step": 14420
    },
    {
      "epoch": 8.505306603773585,
      "grad_norm": 2.0115528106689453,
      "learning_rate": 7.109359919698658e-06,
      "loss": 0.4372,
      "num_input_tokens_seen": 9404360,
      "step": 14425
    },
    {
      "epoch": 8.508254716981131,
      "grad_norm": 4.964980125427246,
      "learning_rate": 7.107027075409316e-06,
      "loss": 0.3583,
      "num_input_tokens_seen": 9407080,
      "step": 14430
    },
    {
      "epoch": 8.51120283018868,
      "grad_norm": 3.491225242614746,
      "learning_rate": 7.104693673275475e-06,
      "loss": 0.5223,
      "num_input_tokens_seen": 9410216,
      "step": 14435
    },
    {
      "epoch": 8.514150943396226,
      "grad_norm": 2.5000360012054443,
      "learning_rate": 7.1023597139149115e-06,
      "loss": 0.53,
      "num_input_tokens_seen": 9413352,
      "step": 14440
    },
    {
      "epoch": 8.517099056603774,
      "grad_norm": 2.085036277770996,
      "learning_rate": 7.100025197945555e-06,
      "loss": 0.2691,
      "num_input_tokens_seen": 9417128,
      "step": 14445
    },
    {
      "epoch": 8.52004716981132,
      "grad_norm": 2.489471197128296,
      "learning_rate": 7.097690125985476e-06,
      "loss": 0.3329,
      "num_input_tokens_seen": 9419656,
      "step": 14450
    },
    {
      "epoch": 8.522995283018869,
      "grad_norm": 3.95540452003479,
      "learning_rate": 7.095354498652895e-06,
      "loss": 0.4832,
      "num_input_tokens_seen": 9422568,
      "step": 14455
    },
    {
      "epoch": 8.525943396226415,
      "grad_norm": 4.1191229820251465,
      "learning_rate": 7.093018316566182e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 9425032,
      "step": 14460
    },
    {
      "epoch": 8.528891509433961,
      "grad_norm": 2.906787633895874,
      "learning_rate": 7.0906815803438465e-06,
      "loss": 0.5568,
      "num_input_tokens_seen": 9428904,
      "step": 14465
    },
    {
      "epoch": 8.53183962264151,
      "grad_norm": 4.482585906982422,
      "learning_rate": 7.088344290604554e-06,
      "loss": 0.4224,
      "num_input_tokens_seen": 9431912,
      "step": 14470
    },
    {
      "epoch": 8.534787735849056,
      "grad_norm": 1.3960684537887573,
      "learning_rate": 7.086006447967111e-06,
      "loss": 0.3612,
      "num_input_tokens_seen": 9435464,
      "step": 14475
    },
    {
      "epoch": 8.537735849056604,
      "grad_norm": 3.8985610008239746,
      "learning_rate": 7.08366805305047e-06,
      "loss": 0.5258,
      "num_input_tokens_seen": 9437832,
      "step": 14480
    },
    {
      "epoch": 8.54068396226415,
      "grad_norm": 4.417331218719482,
      "learning_rate": 7.08132910647373e-06,
      "loss": 0.4489,
      "num_input_tokens_seen": 9440328,
      "step": 14485
    },
    {
      "epoch": 8.543632075471699,
      "grad_norm": 3.271285057067871,
      "learning_rate": 7.078989608856142e-06,
      "loss": 0.3863,
      "num_input_tokens_seen": 9443304,
      "step": 14490
    },
    {
      "epoch": 8.546580188679245,
      "grad_norm": 0.3511248826980591,
      "learning_rate": 7.076649560817092e-06,
      "loss": 0.4744,
      "num_input_tokens_seen": 9449608,
      "step": 14495
    },
    {
      "epoch": 8.549528301886792,
      "grad_norm": 1.8501567840576172,
      "learning_rate": 7.0743089629761245e-06,
      "loss": 0.342,
      "num_input_tokens_seen": 9452520,
      "step": 14500
    },
    {
      "epoch": 8.55247641509434,
      "grad_norm": 2.4843695163726807,
      "learning_rate": 7.071967815952917e-06,
      "loss": 0.288,
      "num_input_tokens_seen": 9454728,
      "step": 14505
    },
    {
      "epoch": 8.555424528301886,
      "grad_norm": 2.3058955669403076,
      "learning_rate": 7.0696261203673e-06,
      "loss": 0.3987,
      "num_input_tokens_seen": 9457832,
      "step": 14510
    },
    {
      "epoch": 8.558372641509434,
      "grad_norm": 3.246218681335449,
      "learning_rate": 7.067283876839249e-06,
      "loss": 0.3172,
      "num_input_tokens_seen": 9460424,
      "step": 14515
    },
    {
      "epoch": 8.56132075471698,
      "grad_norm": 1.9781414270401,
      "learning_rate": 7.064941085988884e-06,
      "loss": 0.4047,
      "num_input_tokens_seen": 9463848,
      "step": 14520
    },
    {
      "epoch": 8.564268867924529,
      "grad_norm": 1.635388731956482,
      "learning_rate": 7.062597748436464e-06,
      "loss": 0.3855,
      "num_input_tokens_seen": 9467368,
      "step": 14525
    },
    {
      "epoch": 8.567216981132075,
      "grad_norm": 5.490896701812744,
      "learning_rate": 7.060253864802402e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 9470280,
      "step": 14530
    },
    {
      "epoch": 8.570165094339622,
      "grad_norm": 3.234849452972412,
      "learning_rate": 7.05790943570725e-06,
      "loss": 0.2998,
      "num_input_tokens_seen": 9473128,
      "step": 14535
    },
    {
      "epoch": 8.57311320754717,
      "grad_norm": 2.327120065689087,
      "learning_rate": 7.055564461771704e-06,
      "loss": 0.3409,
      "num_input_tokens_seen": 9477224,
      "step": 14540
    },
    {
      "epoch": 8.576061320754716,
      "grad_norm": 1.818238377571106,
      "learning_rate": 7.053218943616611e-06,
      "loss": 0.3431,
      "num_input_tokens_seen": 9480488,
      "step": 14545
    },
    {
      "epoch": 8.579009433962264,
      "grad_norm": 2.931459903717041,
      "learning_rate": 7.050872881862952e-06,
      "loss": 0.3548,
      "num_input_tokens_seen": 9483752,
      "step": 14550
    },
    {
      "epoch": 8.58195754716981,
      "grad_norm": 2.881089925765991,
      "learning_rate": 7.04852627713186e-06,
      "loss": 0.3602,
      "num_input_tokens_seen": 9487144,
      "step": 14555
    },
    {
      "epoch": 8.584905660377359,
      "grad_norm": 2.698629856109619,
      "learning_rate": 7.0461791300446055e-06,
      "loss": 0.3691,
      "num_input_tokens_seen": 9489960,
      "step": 14560
    },
    {
      "epoch": 8.587853773584905,
      "grad_norm": 4.952970027923584,
      "learning_rate": 7.043831441222611e-06,
      "loss": 0.5597,
      "num_input_tokens_seen": 9494568,
      "step": 14565
    },
    {
      "epoch": 8.590801886792454,
      "grad_norm": 3.6802735328674316,
      "learning_rate": 7.04148321128743e-06,
      "loss": 0.447,
      "num_input_tokens_seen": 9498216,
      "step": 14570
    },
    {
      "epoch": 8.59375,
      "grad_norm": 3.7266273498535156,
      "learning_rate": 7.039134440860773e-06,
      "loss": 0.4238,
      "num_input_tokens_seen": 9502408,
      "step": 14575
    },
    {
      "epoch": 8.596698113207546,
      "grad_norm": 1.7840392589569092,
      "learning_rate": 7.036785130564484e-06,
      "loss": 0.3865,
      "num_input_tokens_seen": 9505320,
      "step": 14580
    },
    {
      "epoch": 8.599646226415095,
      "grad_norm": 1.9658892154693604,
      "learning_rate": 7.0344352810205544e-06,
      "loss": 0.432,
      "num_input_tokens_seen": 9508232,
      "step": 14585
    },
    {
      "epoch": 8.602594339622641,
      "grad_norm": 2.8039045333862305,
      "learning_rate": 7.032084892851115e-06,
      "loss": 0.3003,
      "num_input_tokens_seen": 9511336,
      "step": 14590
    },
    {
      "epoch": 8.60554245283019,
      "grad_norm": 2.4071102142333984,
      "learning_rate": 7.0297339666784425e-06,
      "loss": 0.4037,
      "num_input_tokens_seen": 9514600,
      "step": 14595
    },
    {
      "epoch": 8.608490566037736,
      "grad_norm": 1.9058986902236938,
      "learning_rate": 7.0273825031249556e-06,
      "loss": 0.3441,
      "num_input_tokens_seen": 9517608,
      "step": 14600
    },
    {
      "epoch": 8.611438679245284,
      "grad_norm": 2.071061134338379,
      "learning_rate": 7.025030502813213e-06,
      "loss": 0.296,
      "num_input_tokens_seen": 9521000,
      "step": 14605
    },
    {
      "epoch": 8.61438679245283,
      "grad_norm": 4.542070388793945,
      "learning_rate": 7.022677966365917e-06,
      "loss": 0.4603,
      "num_input_tokens_seen": 9523720,
      "step": 14610
    },
    {
      "epoch": 8.617334905660378,
      "grad_norm": 2.9570059776306152,
      "learning_rate": 7.020324894405913e-06,
      "loss": 0.3647,
      "num_input_tokens_seen": 9527080,
      "step": 14615
    },
    {
      "epoch": 8.620283018867925,
      "grad_norm": 3.77510142326355,
      "learning_rate": 7.017971287556185e-06,
      "loss": 0.4807,
      "num_input_tokens_seen": 9531368,
      "step": 14620
    },
    {
      "epoch": 8.623231132075471,
      "grad_norm": 3.354464530944824,
      "learning_rate": 7.015617146439863e-06,
      "loss": 0.3272,
      "num_input_tokens_seen": 9534376,
      "step": 14625
    },
    {
      "epoch": 8.62617924528302,
      "grad_norm": 1.8330391645431519,
      "learning_rate": 7.0132624716802125e-06,
      "loss": 0.4818,
      "num_input_tokens_seen": 9538504,
      "step": 14630
    },
    {
      "epoch": 8.629127358490566,
      "grad_norm": 1.7349047660827637,
      "learning_rate": 7.0109072639006474e-06,
      "loss": 0.3811,
      "num_input_tokens_seen": 9541864,
      "step": 14635
    },
    {
      "epoch": 8.632075471698114,
      "grad_norm": 5.677450180053711,
      "learning_rate": 7.008551523724717e-06,
      "loss": 0.4787,
      "num_input_tokens_seen": 9544712,
      "step": 14640
    },
    {
      "epoch": 8.63502358490566,
      "grad_norm": 1.7557170391082764,
      "learning_rate": 7.006195251776116e-06,
      "loss": 0.4003,
      "num_input_tokens_seen": 9547016,
      "step": 14645
    },
    {
      "epoch": 8.637971698113208,
      "grad_norm": 1.952061653137207,
      "learning_rate": 7.003838448678674e-06,
      "loss": 0.3155,
      "num_input_tokens_seen": 9549736,
      "step": 14650
    },
    {
      "epoch": 8.640919811320755,
      "grad_norm": 2.693075656890869,
      "learning_rate": 7.0014811150563675e-06,
      "loss": 0.4608,
      "num_input_tokens_seen": 9552936,
      "step": 14655
    },
    {
      "epoch": 8.643867924528301,
      "grad_norm": 3.187349319458008,
      "learning_rate": 6.999123251533311e-06,
      "loss": 0.4517,
      "num_input_tokens_seen": 9557064,
      "step": 14660
    },
    {
      "epoch": 8.64681603773585,
      "grad_norm": 3.3136374950408936,
      "learning_rate": 6.996764858733756e-06,
      "loss": 0.4811,
      "num_input_tokens_seen": 9560968,
      "step": 14665
    },
    {
      "epoch": 8.649764150943396,
      "grad_norm": 2.0213358402252197,
      "learning_rate": 6.994405937282099e-06,
      "loss": 0.4053,
      "num_input_tokens_seen": 9564680,
      "step": 14670
    },
    {
      "epoch": 8.652712264150944,
      "grad_norm": 5.869167327880859,
      "learning_rate": 6.9920464878028745e-06,
      "loss": 0.4502,
      "num_input_tokens_seen": 9567944,
      "step": 14675
    },
    {
      "epoch": 8.65566037735849,
      "grad_norm": 2.5663812160491943,
      "learning_rate": 6.989686510920758e-06,
      "loss": 0.4961,
      "num_input_tokens_seen": 9571272,
      "step": 14680
    },
    {
      "epoch": 8.658608490566039,
      "grad_norm": 3.2090396881103516,
      "learning_rate": 6.9873260072605634e-06,
      "loss": 0.3187,
      "num_input_tokens_seen": 9576104,
      "step": 14685
    },
    {
      "epoch": 8.661556603773585,
      "grad_norm": 5.881671905517578,
      "learning_rate": 6.984964977447243e-06,
      "loss": 0.4937,
      "num_input_tokens_seen": 9578920,
      "step": 14690
    },
    {
      "epoch": 8.664504716981131,
      "grad_norm": 2.5772156715393066,
      "learning_rate": 6.982603422105889e-06,
      "loss": 0.3695,
      "num_input_tokens_seen": 9584136,
      "step": 14695
    },
    {
      "epoch": 8.66745283018868,
      "grad_norm": 4.4444355964660645,
      "learning_rate": 6.980241341861736e-06,
      "loss": 0.4547,
      "num_input_tokens_seen": 9589768,
      "step": 14700
    },
    {
      "epoch": 8.670400943396226,
      "grad_norm": 2.446840286254883,
      "learning_rate": 6.977878737340153e-06,
      "loss": 0.4855,
      "num_input_tokens_seen": 9596744,
      "step": 14705
    },
    {
      "epoch": 8.673349056603774,
      "grad_norm": 2.2228715419769287,
      "learning_rate": 6.97551560916665e-06,
      "loss": 0.3848,
      "num_input_tokens_seen": 9600520,
      "step": 14710
    },
    {
      "epoch": 8.67629716981132,
      "grad_norm": 2.2739105224609375,
      "learning_rate": 6.973151957966875e-06,
      "loss": 0.5385,
      "num_input_tokens_seen": 9603432,
      "step": 14715
    },
    {
      "epoch": 8.679245283018869,
      "grad_norm": 3.178424596786499,
      "learning_rate": 6.970787784366616e-06,
      "loss": 0.4285,
      "num_input_tokens_seen": 9607208,
      "step": 14720
    },
    {
      "epoch": 8.682193396226415,
      "grad_norm": 2.009650468826294,
      "learning_rate": 6.968423088991797e-06,
      "loss": 0.3787,
      "num_input_tokens_seen": 9610088,
      "step": 14725
    },
    {
      "epoch": 8.685141509433961,
      "grad_norm": 2.3175225257873535,
      "learning_rate": 6.966057872468481e-06,
      "loss": 0.399,
      "num_input_tokens_seen": 9613288,
      "step": 14730
    },
    {
      "epoch": 8.68808962264151,
      "grad_norm": 1.885446548461914,
      "learning_rate": 6.963692135422872e-06,
      "loss": 0.4141,
      "num_input_tokens_seen": 9617928,
      "step": 14735
    },
    {
      "epoch": 8.691037735849056,
      "grad_norm": 1.8774847984313965,
      "learning_rate": 6.961325878481305e-06,
      "loss": 0.3259,
      "num_input_tokens_seen": 9620712,
      "step": 14740
    },
    {
      "epoch": 8.693985849056604,
      "grad_norm": 2.97818922996521,
      "learning_rate": 6.958959102270259e-06,
      "loss": 0.5477,
      "num_input_tokens_seen": 9623784,
      "step": 14745
    },
    {
      "epoch": 8.69693396226415,
      "grad_norm": 1.4760496616363525,
      "learning_rate": 6.95659180741635e-06,
      "loss": 0.3008,
      "num_input_tokens_seen": 9626856,
      "step": 14750
    },
    {
      "epoch": 8.699882075471699,
      "grad_norm": 2.454509973526001,
      "learning_rate": 6.954223994546326e-06,
      "loss": 0.4244,
      "num_input_tokens_seen": 9629800,
      "step": 14755
    },
    {
      "epoch": 8.702830188679245,
      "grad_norm": 3.469494581222534,
      "learning_rate": 6.951855664287077e-06,
      "loss": 0.3438,
      "num_input_tokens_seen": 9633384,
      "step": 14760
    },
    {
      "epoch": 8.705778301886792,
      "grad_norm": 1.54270601272583,
      "learning_rate": 6.9494868172656304e-06,
      "loss": 0.3257,
      "num_input_tokens_seen": 9636872,
      "step": 14765
    },
    {
      "epoch": 8.70872641509434,
      "grad_norm": 2.034594774246216,
      "learning_rate": 6.947117454109146e-06,
      "loss": 0.3777,
      "num_input_tokens_seen": 9639720,
      "step": 14770
    },
    {
      "epoch": 8.711674528301886,
      "grad_norm": 2.6863253116607666,
      "learning_rate": 6.944747575444924e-06,
      "loss": 0.567,
      "num_input_tokens_seen": 9642600,
      "step": 14775
    },
    {
      "epoch": 8.714622641509434,
      "grad_norm": 3.1776413917541504,
      "learning_rate": 6.942377181900399e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 9646568,
      "step": 14780
    },
    {
      "epoch": 8.71757075471698,
      "grad_norm": 3.6220107078552246,
      "learning_rate": 6.940006274103146e-06,
      "loss": 0.3925,
      "num_input_tokens_seen": 9649448,
      "step": 14785
    },
    {
      "epoch": 8.720518867924529,
      "grad_norm": 2.513629913330078,
      "learning_rate": 6.93763485268087e-06,
      "loss": 0.4324,
      "num_input_tokens_seen": 9652008,
      "step": 14790
    },
    {
      "epoch": 8.723466981132075,
      "grad_norm": 2.0231757164001465,
      "learning_rate": 6.935262918261416e-06,
      "loss": 0.3284,
      "num_input_tokens_seen": 9655048,
      "step": 14795
    },
    {
      "epoch": 8.726415094339622,
      "grad_norm": 2.069519281387329,
      "learning_rate": 6.932890471472764e-06,
      "loss": 0.5942,
      "num_input_tokens_seen": 9657960,
      "step": 14800
    },
    {
      "epoch": 8.72936320754717,
      "grad_norm": 2.7226979732513428,
      "learning_rate": 6.930517512943029e-06,
      "loss": 0.4625,
      "num_input_tokens_seen": 9661224,
      "step": 14805
    },
    {
      "epoch": 8.732311320754716,
      "grad_norm": 2.512676239013672,
      "learning_rate": 6.928144043300463e-06,
      "loss": 0.2971,
      "num_input_tokens_seen": 9664392,
      "step": 14810
    },
    {
      "epoch": 8.735259433962264,
      "grad_norm": 2.156874418258667,
      "learning_rate": 6.925770063173451e-06,
      "loss": 0.5073,
      "num_input_tokens_seen": 9668904,
      "step": 14815
    },
    {
      "epoch": 8.73820754716981,
      "grad_norm": 2.5475564002990723,
      "learning_rate": 6.923395573190514e-06,
      "loss": 0.3598,
      "num_input_tokens_seen": 9672424,
      "step": 14820
    },
    {
      "epoch": 8.741155660377359,
      "grad_norm": 2.8670053482055664,
      "learning_rate": 6.921020573980313e-06,
      "loss": 0.3798,
      "num_input_tokens_seen": 9675624,
      "step": 14825
    },
    {
      "epoch": 8.744103773584905,
      "grad_norm": 2.135509729385376,
      "learning_rate": 6.918645066171634e-06,
      "loss": 0.3263,
      "num_input_tokens_seen": 9678920,
      "step": 14830
    },
    {
      "epoch": 8.747051886792454,
      "grad_norm": 4.993575572967529,
      "learning_rate": 6.916269050393404e-06,
      "loss": 0.3873,
      "num_input_tokens_seen": 9681448,
      "step": 14835
    },
    {
      "epoch": 8.75,
      "grad_norm": 2.099391222000122,
      "learning_rate": 6.913892527274686e-06,
      "loss": 0.5164,
      "num_input_tokens_seen": 9684392,
      "step": 14840
    },
    {
      "epoch": 8.752948113207546,
      "grad_norm": 6.596562385559082,
      "learning_rate": 6.9115154974446716e-06,
      "loss": 0.4245,
      "num_input_tokens_seen": 9687720,
      "step": 14845
    },
    {
      "epoch": 8.755896226415095,
      "grad_norm": 2.492990016937256,
      "learning_rate": 6.909137961532692e-06,
      "loss": 0.2978,
      "num_input_tokens_seen": 9690632,
      "step": 14850
    },
    {
      "epoch": 8.758844339622641,
      "grad_norm": 4.106266498565674,
      "learning_rate": 6.906759920168209e-06,
      "loss": 0.4541,
      "num_input_tokens_seen": 9693544,
      "step": 14855
    },
    {
      "epoch": 8.76179245283019,
      "grad_norm": 1.2404447793960571,
      "learning_rate": 6.90438137398082e-06,
      "loss": 0.382,
      "num_input_tokens_seen": 9697992,
      "step": 14860
    },
    {
      "epoch": 8.764740566037736,
      "grad_norm": 1.9230403900146484,
      "learning_rate": 6.902002323600252e-06,
      "loss": 0.3407,
      "num_input_tokens_seen": 9700744,
      "step": 14865
    },
    {
      "epoch": 8.767688679245284,
      "grad_norm": 5.1434526443481445,
      "learning_rate": 6.899622769656373e-06,
      "loss": 0.3362,
      "num_input_tokens_seen": 9703752,
      "step": 14870
    },
    {
      "epoch": 8.77063679245283,
      "grad_norm": 2.4574313163757324,
      "learning_rate": 6.897242712779179e-06,
      "loss": 0.3287,
      "num_input_tokens_seen": 9707272,
      "step": 14875
    },
    {
      "epoch": 8.773584905660378,
      "grad_norm": 1.8372465372085571,
      "learning_rate": 6.894862153598802e-06,
      "loss": 0.3483,
      "num_input_tokens_seen": 9710280,
      "step": 14880
    },
    {
      "epoch": 8.776533018867925,
      "grad_norm": 2.3414535522460938,
      "learning_rate": 6.892481092745502e-06,
      "loss": 0.4507,
      "num_input_tokens_seen": 9714216,
      "step": 14885
    },
    {
      "epoch": 8.779481132075471,
      "grad_norm": 2.5674240589141846,
      "learning_rate": 6.890099530849677e-06,
      "loss": 0.6275,
      "num_input_tokens_seen": 9717224,
      "step": 14890
    },
    {
      "epoch": 8.78242924528302,
      "grad_norm": 1.6274032592773438,
      "learning_rate": 6.887717468541855e-06,
      "loss": 0.4574,
      "num_input_tokens_seen": 9720872,
      "step": 14895
    },
    {
      "epoch": 8.785377358490566,
      "grad_norm": 3.696709394454956,
      "learning_rate": 6.885334906452696e-06,
      "loss": 0.4132,
      "num_input_tokens_seen": 9723624,
      "step": 14900
    },
    {
      "epoch": 8.788325471698114,
      "grad_norm": 4.480981826782227,
      "learning_rate": 6.882951845212997e-06,
      "loss": 0.3908,
      "num_input_tokens_seen": 9726248,
      "step": 14905
    },
    {
      "epoch": 8.79127358490566,
      "grad_norm": 2.964874744415283,
      "learning_rate": 6.880568285453682e-06,
      "loss": 0.4575,
      "num_input_tokens_seen": 9728456,
      "step": 14910
    },
    {
      "epoch": 8.794221698113208,
      "grad_norm": 2.3584275245666504,
      "learning_rate": 6.878184227805807e-06,
      "loss": 0.3126,
      "num_input_tokens_seen": 9732360,
      "step": 14915
    },
    {
      "epoch": 8.797169811320755,
      "grad_norm": 1.721100926399231,
      "learning_rate": 6.8757996729005645e-06,
      "loss": 0.3775,
      "num_input_tokens_seen": 9736072,
      "step": 14920
    },
    {
      "epoch": 8.800117924528301,
      "grad_norm": 2.1002752780914307,
      "learning_rate": 6.8734146213692756e-06,
      "loss": 0.356,
      "num_input_tokens_seen": 9738856,
      "step": 14925
    },
    {
      "epoch": 8.80306603773585,
      "grad_norm": 2.216747283935547,
      "learning_rate": 6.87102907384339e-06,
      "loss": 0.3749,
      "num_input_tokens_seen": 9741992,
      "step": 14930
    },
    {
      "epoch": 8.806014150943396,
      "grad_norm": 2.1929080486297607,
      "learning_rate": 6.868643030954494e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 9745896,
      "step": 14935
    },
    {
      "epoch": 8.808962264150944,
      "grad_norm": 1.8580706119537354,
      "learning_rate": 6.866256493334302e-06,
      "loss": 0.4265,
      "num_input_tokens_seen": 9749576,
      "step": 14940
    },
    {
      "epoch": 8.81191037735849,
      "grad_norm": 2.153745174407959,
      "learning_rate": 6.863869461614659e-06,
      "loss": 0.3347,
      "num_input_tokens_seen": 9753672,
      "step": 14945
    },
    {
      "epoch": 8.814858490566039,
      "grad_norm": 2.6148741245269775,
      "learning_rate": 6.861481936427545e-06,
      "loss": 0.3544,
      "num_input_tokens_seen": 9757320,
      "step": 14950
    },
    {
      "epoch": 8.817806603773585,
      "grad_norm": 4.132111549377441,
      "learning_rate": 6.859093918405067e-06,
      "loss": 0.3839,
      "num_input_tokens_seen": 9760328,
      "step": 14955
    },
    {
      "epoch": 8.820754716981131,
      "grad_norm": 4.385087490081787,
      "learning_rate": 6.856705408179458e-06,
      "loss": 0.5049,
      "num_input_tokens_seen": 9763496,
      "step": 14960
    },
    {
      "epoch": 8.82370283018868,
      "grad_norm": 1.8727986812591553,
      "learning_rate": 6.854316406383093e-06,
      "loss": 0.419,
      "num_input_tokens_seen": 9767592,
      "step": 14965
    },
    {
      "epoch": 8.826650943396226,
      "grad_norm": 2.9489119052886963,
      "learning_rate": 6.8519269136484665e-06,
      "loss": 0.4227,
      "num_input_tokens_seen": 9770440,
      "step": 14970
    },
    {
      "epoch": 8.829599056603774,
      "grad_norm": 3.0117952823638916,
      "learning_rate": 6.849536930608208e-06,
      "loss": 0.389,
      "num_input_tokens_seen": 9772552,
      "step": 14975
    },
    {
      "epoch": 8.83254716981132,
      "grad_norm": 1.9244282245635986,
      "learning_rate": 6.847146457895078e-06,
      "loss": 0.3432,
      "num_input_tokens_seen": 9775752,
      "step": 14980
    },
    {
      "epoch": 8.835495283018869,
      "grad_norm": 1.5959678888320923,
      "learning_rate": 6.8447554961419615e-06,
      "loss": 0.3584,
      "num_input_tokens_seen": 9778760,
      "step": 14985
    },
    {
      "epoch": 8.838443396226415,
      "grad_norm": 1.6212234497070312,
      "learning_rate": 6.842364045981876e-06,
      "loss": 0.313,
      "num_input_tokens_seen": 9782280,
      "step": 14990
    },
    {
      "epoch": 8.841391509433961,
      "grad_norm": 2.307650566101074,
      "learning_rate": 6.83997210804797e-06,
      "loss": 0.3883,
      "num_input_tokens_seen": 9786440,
      "step": 14995
    },
    {
      "epoch": 8.84433962264151,
      "grad_norm": 1.8725664615631104,
      "learning_rate": 6.837579682973519e-06,
      "loss": 0.4152,
      "num_input_tokens_seen": 9789832,
      "step": 15000
    },
    {
      "epoch": 8.847287735849056,
      "grad_norm": 2.428133249282837,
      "learning_rate": 6.835186771391926e-06,
      "loss": 0.2716,
      "num_input_tokens_seen": 9792360,
      "step": 15005
    },
    {
      "epoch": 8.850235849056604,
      "grad_norm": 3.086512327194214,
      "learning_rate": 6.8327933739367266e-06,
      "loss": 0.4118,
      "num_input_tokens_seen": 9794824,
      "step": 15010
    },
    {
      "epoch": 8.85318396226415,
      "grad_norm": 3.4672436714172363,
      "learning_rate": 6.830399491241584e-06,
      "loss": 0.2922,
      "num_input_tokens_seen": 9798280,
      "step": 15015
    },
    {
      "epoch": 8.856132075471699,
      "grad_norm": 1.7747465372085571,
      "learning_rate": 6.828005123940287e-06,
      "loss": 0.3109,
      "num_input_tokens_seen": 9803720,
      "step": 15020
    },
    {
      "epoch": 8.859080188679245,
      "grad_norm": 3.624042510986328,
      "learning_rate": 6.825610272666754e-06,
      "loss": 0.4407,
      "num_input_tokens_seen": 9806504,
      "step": 15025
    },
    {
      "epoch": 8.862028301886792,
      "grad_norm": 3.3068394660949707,
      "learning_rate": 6.823214938055034e-06,
      "loss": 0.4007,
      "num_input_tokens_seen": 9809192,
      "step": 15030
    },
    {
      "epoch": 8.86497641509434,
      "grad_norm": 2.7866663932800293,
      "learning_rate": 6.8208191207393e-06,
      "loss": 0.48,
      "num_input_tokens_seen": 9812008,
      "step": 15035
    },
    {
      "epoch": 8.867924528301886,
      "grad_norm": 1.5656015872955322,
      "learning_rate": 6.818422821353859e-06,
      "loss": 0.3435,
      "num_input_tokens_seen": 9815464,
      "step": 15040
    },
    {
      "epoch": 8.870872641509434,
      "grad_norm": 4.182625770568848,
      "learning_rate": 6.816026040533139e-06,
      "loss": 0.4625,
      "num_input_tokens_seen": 9818888,
      "step": 15045
    },
    {
      "epoch": 8.87382075471698,
      "grad_norm": 2.716383695602417,
      "learning_rate": 6.8136287789116966e-06,
      "loss": 0.3678,
      "num_input_tokens_seen": 9822952,
      "step": 15050
    },
    {
      "epoch": 8.876768867924529,
      "grad_norm": 1.5944288969039917,
      "learning_rate": 6.81123103712422e-06,
      "loss": 0.4056,
      "num_input_tokens_seen": 9827144,
      "step": 15055
    },
    {
      "epoch": 8.879716981132075,
      "grad_norm": 2.8928966522216797,
      "learning_rate": 6.808832815805518e-06,
      "loss": 0.3675,
      "num_input_tokens_seen": 9830440,
      "step": 15060
    },
    {
      "epoch": 8.882665094339622,
      "grad_norm": 2.9204976558685303,
      "learning_rate": 6.806434115590534e-06,
      "loss": 0.443,
      "num_input_tokens_seen": 9833192,
      "step": 15065
    },
    {
      "epoch": 8.88561320754717,
      "grad_norm": 2.4361791610717773,
      "learning_rate": 6.804034937114332e-06,
      "loss": 0.3941,
      "num_input_tokens_seen": 9835880,
      "step": 15070
    },
    {
      "epoch": 8.888561320754716,
      "grad_norm": 2.6314823627471924,
      "learning_rate": 6.8016352810121064e-06,
      "loss": 0.3446,
      "num_input_tokens_seen": 9838248,
      "step": 15075
    },
    {
      "epoch": 8.891509433962264,
      "grad_norm": 2.179109811782837,
      "learning_rate": 6.799235147919176e-06,
      "loss": 0.4545,
      "num_input_tokens_seen": 9841480,
      "step": 15080
    },
    {
      "epoch": 8.89445754716981,
      "grad_norm": 2.0048701763153076,
      "learning_rate": 6.796834538470985e-06,
      "loss": 0.4262,
      "num_input_tokens_seen": 9844392,
      "step": 15085
    },
    {
      "epoch": 8.897405660377359,
      "grad_norm": 1.8004496097564697,
      "learning_rate": 6.794433453303106e-06,
      "loss": 0.4196,
      "num_input_tokens_seen": 9847848,
      "step": 15090
    },
    {
      "epoch": 8.900353773584905,
      "grad_norm": 2.1938366889953613,
      "learning_rate": 6.792031893051238e-06,
      "loss": 0.4999,
      "num_input_tokens_seen": 9851432,
      "step": 15095
    },
    {
      "epoch": 8.903301886792454,
      "grad_norm": 3.094050168991089,
      "learning_rate": 6.789629858351201e-06,
      "loss": 0.3777,
      "num_input_tokens_seen": 9854984,
      "step": 15100
    },
    {
      "epoch": 8.90625,
      "grad_norm": 2.8745994567871094,
      "learning_rate": 6.787227349838946e-06,
      "loss": 0.4076,
      "num_input_tokens_seen": 9857800,
      "step": 15105
    },
    {
      "epoch": 8.909198113207546,
      "grad_norm": 2.2255163192749023,
      "learning_rate": 6.784824368150548e-06,
      "loss": 0.3255,
      "num_input_tokens_seen": 9860296,
      "step": 15110
    },
    {
      "epoch": 8.912146226415095,
      "grad_norm": 2.0890579223632812,
      "learning_rate": 6.7824209139222076e-06,
      "loss": 0.3604,
      "num_input_tokens_seen": 9862984,
      "step": 15115
    },
    {
      "epoch": 8.915094339622641,
      "grad_norm": 2.5852580070495605,
      "learning_rate": 6.780016987790248e-06,
      "loss": 0.3428,
      "num_input_tokens_seen": 9866120,
      "step": 15120
    },
    {
      "epoch": 8.91804245283019,
      "grad_norm": 2.4474847316741943,
      "learning_rate": 6.7776125903911194e-06,
      "loss": 0.4906,
      "num_input_tokens_seen": 9870120,
      "step": 15125
    },
    {
      "epoch": 8.920990566037736,
      "grad_norm": 3.444129705429077,
      "learning_rate": 6.775207722361396e-06,
      "loss": 0.5517,
      "num_input_tokens_seen": 9872808,
      "step": 15130
    },
    {
      "epoch": 8.923938679245284,
      "grad_norm": 2.122300624847412,
      "learning_rate": 6.772802384337778e-06,
      "loss": 0.4352,
      "num_input_tokens_seen": 9876456,
      "step": 15135
    },
    {
      "epoch": 8.92688679245283,
      "grad_norm": 3.071918249130249,
      "learning_rate": 6.770396576957088e-06,
      "loss": 0.2946,
      "num_input_tokens_seen": 9878728,
      "step": 15140
    },
    {
      "epoch": 8.929834905660378,
      "grad_norm": 1.8663431406021118,
      "learning_rate": 6.767990300856274e-06,
      "loss": 0.5341,
      "num_input_tokens_seen": 9881704,
      "step": 15145
    },
    {
      "epoch": 8.932783018867925,
      "grad_norm": 2.977050304412842,
      "learning_rate": 6.765583556672408e-06,
      "loss": 0.3521,
      "num_input_tokens_seen": 9884392,
      "step": 15150
    },
    {
      "epoch": 8.935731132075471,
      "grad_norm": 1.5496002435684204,
      "learning_rate": 6.763176345042687e-06,
      "loss": 0.2488,
      "num_input_tokens_seen": 9887528,
      "step": 15155
    },
    {
      "epoch": 8.93867924528302,
      "grad_norm": 5.135709762573242,
      "learning_rate": 6.760768666604429e-06,
      "loss": 0.6035,
      "num_input_tokens_seen": 9890376,
      "step": 15160
    },
    {
      "epoch": 8.941627358490566,
      "grad_norm": 4.098505973815918,
      "learning_rate": 6.758360521995079e-06,
      "loss": 0.4492,
      "num_input_tokens_seen": 9893320,
      "step": 15165
    },
    {
      "epoch": 8.944575471698114,
      "grad_norm": 2.8324337005615234,
      "learning_rate": 6.755951911852202e-06,
      "loss": 0.3756,
      "num_input_tokens_seen": 9896232,
      "step": 15170
    },
    {
      "epoch": 8.94752358490566,
      "grad_norm": 4.494495391845703,
      "learning_rate": 6.7535428368134885e-06,
      "loss": 0.4015,
      "num_input_tokens_seen": 9899464,
      "step": 15175
    },
    {
      "epoch": 8.950471698113208,
      "grad_norm": 2.36110520362854,
      "learning_rate": 6.751133297516752e-06,
      "loss": 0.5217,
      "num_input_tokens_seen": 9902728,
      "step": 15180
    },
    {
      "epoch": 8.953419811320755,
      "grad_norm": 2.2041213512420654,
      "learning_rate": 6.748723294599928e-06,
      "loss": 0.35,
      "num_input_tokens_seen": 9906280,
      "step": 15185
    },
    {
      "epoch": 8.956367924528301,
      "grad_norm": 1.1619752645492554,
      "learning_rate": 6.746312828701075e-06,
      "loss": 0.4208,
      "num_input_tokens_seen": 9909960,
      "step": 15190
    },
    {
      "epoch": 8.95931603773585,
      "grad_norm": 4.294003009796143,
      "learning_rate": 6.743901900458374e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 9912744,
      "step": 15195
    },
    {
      "epoch": 8.962264150943396,
      "grad_norm": 2.167071580886841,
      "learning_rate": 6.741490510510129e-06,
      "loss": 0.3991,
      "num_input_tokens_seen": 9915976,
      "step": 15200
    },
    {
      "epoch": 8.965212264150944,
      "grad_norm": 2.4379770755767822,
      "learning_rate": 6.7390786594947665e-06,
      "loss": 0.2997,
      "num_input_tokens_seen": 9918536,
      "step": 15205
    },
    {
      "epoch": 8.96816037735849,
      "grad_norm": 2.330822229385376,
      "learning_rate": 6.7366663480508335e-06,
      "loss": 0.453,
      "num_input_tokens_seen": 9921416,
      "step": 15210
    },
    {
      "epoch": 8.971108490566039,
      "grad_norm": 1.5783472061157227,
      "learning_rate": 6.734253576817002e-06,
      "loss": 0.3343,
      "num_input_tokens_seen": 9925544,
      "step": 15215
    },
    {
      "epoch": 8.974056603773585,
      "grad_norm": 3.272941827774048,
      "learning_rate": 6.731840346432061e-06,
      "loss": 0.3874,
      "num_input_tokens_seen": 9928936,
      "step": 15220
    },
    {
      "epoch": 8.977004716981131,
      "grad_norm": 2.0508549213409424,
      "learning_rate": 6.729426657534922e-06,
      "loss": 0.4687,
      "num_input_tokens_seen": 9931912,
      "step": 15225
    },
    {
      "epoch": 8.97995283018868,
      "grad_norm": 3.4232406616210938,
      "learning_rate": 6.727012510764624e-06,
      "loss": 0.4709,
      "num_input_tokens_seen": 9934984,
      "step": 15230
    },
    {
      "epoch": 8.982900943396226,
      "grad_norm": 1.9945486783981323,
      "learning_rate": 6.724597906760322e-06,
      "loss": 0.4471,
      "num_input_tokens_seen": 9937800,
      "step": 15235
    },
    {
      "epoch": 8.985849056603774,
      "grad_norm": 2.603271007537842,
      "learning_rate": 6.722182846161289e-06,
      "loss": 0.5142,
      "num_input_tokens_seen": 9940712,
      "step": 15240
    },
    {
      "epoch": 8.98879716981132,
      "grad_norm": 1.8966410160064697,
      "learning_rate": 6.719767329606926e-06,
      "loss": 0.4572,
      "num_input_tokens_seen": 9944584,
      "step": 15245
    },
    {
      "epoch": 8.991745283018869,
      "grad_norm": 3.0709445476531982,
      "learning_rate": 6.717351357736751e-06,
      "loss": 0.5779,
      "num_input_tokens_seen": 9947368,
      "step": 15250
    },
    {
      "epoch": 8.994693396226415,
      "grad_norm": 1.5476329326629639,
      "learning_rate": 6.7149349311904025e-06,
      "loss": 0.3254,
      "num_input_tokens_seen": 9950088,
      "step": 15255
    },
    {
      "epoch": 8.997641509433961,
      "grad_norm": 3.120551347732544,
      "learning_rate": 6.712518050607642e-06,
      "loss": 0.3661,
      "num_input_tokens_seen": 9953096,
      "step": 15260
    },
    {
      "epoch": 9.00058962264151,
      "grad_norm": 4.0768208503723145,
      "learning_rate": 6.710100716628345e-06,
      "loss": 0.3108,
      "num_input_tokens_seen": 9954992,
      "step": 15265
    },
    {
      "epoch": 9.003537735849056,
      "grad_norm": 2.141944169998169,
      "learning_rate": 6.707682929892513e-06,
      "loss": 0.4337,
      "num_input_tokens_seen": 9957808,
      "step": 15270
    },
    {
      "epoch": 9.006485849056604,
      "grad_norm": 3.5433034896850586,
      "learning_rate": 6.705264691040266e-06,
      "loss": 0.3454,
      "num_input_tokens_seen": 9961520,
      "step": 15275
    },
    {
      "epoch": 9.00943396226415,
      "grad_norm": 3.2581379413604736,
      "learning_rate": 6.7028460007118435e-06,
      "loss": 0.3543,
      "num_input_tokens_seen": 9964144,
      "step": 15280
    },
    {
      "epoch": 9.012382075471699,
      "grad_norm": 2.69348406791687,
      "learning_rate": 6.700426859547602e-06,
      "loss": 0.2855,
      "num_input_tokens_seen": 9967056,
      "step": 15285
    },
    {
      "epoch": 9.015330188679245,
      "grad_norm": 1.4692714214324951,
      "learning_rate": 6.6980072681880224e-06,
      "loss": 0.3572,
      "num_input_tokens_seen": 9970800,
      "step": 15290
    },
    {
      "epoch": 9.018278301886792,
      "grad_norm": 3.3459842205047607,
      "learning_rate": 6.695587227273699e-06,
      "loss": 0.2975,
      "num_input_tokens_seen": 9973392,
      "step": 15295
    },
    {
      "epoch": 9.02122641509434,
      "grad_norm": 2.84409236907959,
      "learning_rate": 6.69316673744535e-06,
      "loss": 0.3454,
      "num_input_tokens_seen": 9976176,
      "step": 15300
    },
    {
      "epoch": 9.024174528301886,
      "grad_norm": 3.748802900314331,
      "learning_rate": 6.6907457993438115e-06,
      "loss": 0.4048,
      "num_input_tokens_seen": 9979440,
      "step": 15305
    },
    {
      "epoch": 9.027122641509434,
      "grad_norm": 3.250750780105591,
      "learning_rate": 6.688324413610036e-06,
      "loss": 0.4487,
      "num_input_tokens_seen": 9982192,
      "step": 15310
    },
    {
      "epoch": 9.03007075471698,
      "grad_norm": 2.582509756088257,
      "learning_rate": 6.685902580885094e-06,
      "loss": 0.2786,
      "num_input_tokens_seen": 9984848,
      "step": 15315
    },
    {
      "epoch": 9.033018867924529,
      "grad_norm": 4.707276821136475,
      "learning_rate": 6.6834803018101794e-06,
      "loss": 0.3064,
      "num_input_tokens_seen": 9987568,
      "step": 15320
    },
    {
      "epoch": 9.035966981132075,
      "grad_norm": 2.529352903366089,
      "learning_rate": 6.681057577026599e-06,
      "loss": 0.3202,
      "num_input_tokens_seen": 9990736,
      "step": 15325
    },
    {
      "epoch": 9.038915094339623,
      "grad_norm": 2.9558496475219727,
      "learning_rate": 6.6786344071757795e-06,
      "loss": 0.4524,
      "num_input_tokens_seen": 9994000,
      "step": 15330
    },
    {
      "epoch": 9.04186320754717,
      "grad_norm": 3.077704429626465,
      "learning_rate": 6.676210792899267e-06,
      "loss": 0.3168,
      "num_input_tokens_seen": 9996272,
      "step": 15335
    },
    {
      "epoch": 9.044811320754716,
      "grad_norm": 3.257589101791382,
      "learning_rate": 6.6737867348387235e-06,
      "loss": 0.3955,
      "num_input_tokens_seen": 9999632,
      "step": 15340
    },
    {
      "epoch": 9.047759433962264,
      "grad_norm": 2.5129952430725098,
      "learning_rate": 6.671362233635926e-06,
      "loss": 0.3593,
      "num_input_tokens_seen": 10002512,
      "step": 15345
    },
    {
      "epoch": 9.05070754716981,
      "grad_norm": 2.338437795639038,
      "learning_rate": 6.668937289932775e-06,
      "loss": 0.3448,
      "num_input_tokens_seen": 10005136,
      "step": 15350
    },
    {
      "epoch": 9.053655660377359,
      "grad_norm": 3.7598979473114014,
      "learning_rate": 6.666511904371285e-06,
      "loss": 0.2999,
      "num_input_tokens_seen": 10008336,
      "step": 15355
    },
    {
      "epoch": 9.056603773584905,
      "grad_norm": 3.086458206176758,
      "learning_rate": 6.664086077593586e-06,
      "loss": 0.2692,
      "num_input_tokens_seen": 10011248,
      "step": 15360
    },
    {
      "epoch": 9.059551886792454,
      "grad_norm": 4.133688926696777,
      "learning_rate": 6.661659810241924e-06,
      "loss": 0.5523,
      "num_input_tokens_seen": 10014448,
      "step": 15365
    },
    {
      "epoch": 9.0625,
      "grad_norm": 2.739734411239624,
      "learning_rate": 6.659233102958667e-06,
      "loss": 0.3496,
      "num_input_tokens_seen": 10017616,
      "step": 15370
    },
    {
      "epoch": 9.065448113207546,
      "grad_norm": 2.8032898902893066,
      "learning_rate": 6.6568059563862965e-06,
      "loss": 0.3469,
      "num_input_tokens_seen": 10020432,
      "step": 15375
    },
    {
      "epoch": 9.068396226415095,
      "grad_norm": 2.7908413410186768,
      "learning_rate": 6.654378371167407e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 10024624,
      "step": 15380
    },
    {
      "epoch": 9.071344339622641,
      "grad_norm": 1.3122307062149048,
      "learning_rate": 6.651950347944715e-06,
      "loss": 0.3473,
      "num_input_tokens_seen": 10029040,
      "step": 15385
    },
    {
      "epoch": 9.07429245283019,
      "grad_norm": 3.0658631324768066,
      "learning_rate": 6.649521887361048e-06,
      "loss": 0.3784,
      "num_input_tokens_seen": 10034064,
      "step": 15390
    },
    {
      "epoch": 9.077240566037736,
      "grad_norm": 2.026421070098877,
      "learning_rate": 6.647092990059352e-06,
      "loss": 0.4726,
      "num_input_tokens_seen": 10037200,
      "step": 15395
    },
    {
      "epoch": 9.080188679245284,
      "grad_norm": 1.9497674703598022,
      "learning_rate": 6.644663656682689e-06,
      "loss": 0.4033,
      "num_input_tokens_seen": 10040368,
      "step": 15400
    },
    {
      "epoch": 9.08313679245283,
      "grad_norm": 2.7108278274536133,
      "learning_rate": 6.642233887874234e-06,
      "loss": 0.3821,
      "num_input_tokens_seen": 10044208,
      "step": 15405
    },
    {
      "epoch": 9.086084905660377,
      "grad_norm": 5.38900899887085,
      "learning_rate": 6.639803684277279e-06,
      "loss": 0.4293,
      "num_input_tokens_seen": 10047120,
      "step": 15410
    },
    {
      "epoch": 9.089033018867925,
      "grad_norm": 4.527899265289307,
      "learning_rate": 6.637373046535233e-06,
      "loss": 0.3601,
      "num_input_tokens_seen": 10050288,
      "step": 15415
    },
    {
      "epoch": 9.091981132075471,
      "grad_norm": 4.853857517242432,
      "learning_rate": 6.634941975291617e-06,
      "loss": 0.4904,
      "num_input_tokens_seen": 10054480,
      "step": 15420
    },
    {
      "epoch": 9.09492924528302,
      "grad_norm": 2.3710391521453857,
      "learning_rate": 6.632510471190065e-06,
      "loss": 0.3969,
      "num_input_tokens_seen": 10057616,
      "step": 15425
    },
    {
      "epoch": 9.097877358490566,
      "grad_norm": 1.9033665657043457,
      "learning_rate": 6.630078534874332e-06,
      "loss": 0.3054,
      "num_input_tokens_seen": 10060144,
      "step": 15430
    },
    {
      "epoch": 9.100825471698114,
      "grad_norm": 2.2946364879608154,
      "learning_rate": 6.62764616698828e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 10062928,
      "step": 15435
    },
    {
      "epoch": 9.10377358490566,
      "grad_norm": 2.050767660140991,
      "learning_rate": 6.625213368175889e-06,
      "loss": 0.3988,
      "num_input_tokens_seen": 10065840,
      "step": 15440
    },
    {
      "epoch": 9.106721698113208,
      "grad_norm": 2.644279956817627,
      "learning_rate": 6.622780139081256e-06,
      "loss": 0.3401,
      "num_input_tokens_seen": 10068912,
      "step": 15445
    },
    {
      "epoch": 9.109669811320755,
      "grad_norm": 2.5647025108337402,
      "learning_rate": 6.620346480348589e-06,
      "loss": 0.4874,
      "num_input_tokens_seen": 10071728,
      "step": 15450
    },
    {
      "epoch": 9.112617924528301,
      "grad_norm": 3.909867525100708,
      "learning_rate": 6.617912392622206e-06,
      "loss": 0.3906,
      "num_input_tokens_seen": 10074672,
      "step": 15455
    },
    {
      "epoch": 9.11556603773585,
      "grad_norm": 1.249127984046936,
      "learning_rate": 6.615477876546544e-06,
      "loss": 0.2949,
      "num_input_tokens_seen": 10078160,
      "step": 15460
    },
    {
      "epoch": 9.118514150943396,
      "grad_norm": 4.260204315185547,
      "learning_rate": 6.6130429327661535e-06,
      "loss": 0.3605,
      "num_input_tokens_seen": 10080912,
      "step": 15465
    },
    {
      "epoch": 9.121462264150944,
      "grad_norm": 1.778310775756836,
      "learning_rate": 6.610607561925694e-06,
      "loss": 0.3987,
      "num_input_tokens_seen": 10084048,
      "step": 15470
    },
    {
      "epoch": 9.12441037735849,
      "grad_norm": 1.6034519672393799,
      "learning_rate": 6.608171764669941e-06,
      "loss": 0.36,
      "num_input_tokens_seen": 10087312,
      "step": 15475
    },
    {
      "epoch": 9.127358490566039,
      "grad_norm": 3.438253402709961,
      "learning_rate": 6.605735541643783e-06,
      "loss": 0.4232,
      "num_input_tokens_seen": 10091184,
      "step": 15480
    },
    {
      "epoch": 9.130306603773585,
      "grad_norm": 3.0842971801757812,
      "learning_rate": 6.603298893492219e-06,
      "loss": 0.3992,
      "num_input_tokens_seen": 10096400,
      "step": 15485
    },
    {
      "epoch": 9.133254716981131,
      "grad_norm": 3.1788229942321777,
      "learning_rate": 6.6008618208603625e-06,
      "loss": 0.38,
      "num_input_tokens_seen": 10099216,
      "step": 15490
    },
    {
      "epoch": 9.13620283018868,
      "grad_norm": 2.855682373046875,
      "learning_rate": 6.598424324393442e-06,
      "loss": 0.3689,
      "num_input_tokens_seen": 10102224,
      "step": 15495
    },
    {
      "epoch": 9.139150943396226,
      "grad_norm": 2.0665063858032227,
      "learning_rate": 6.595986404736792e-06,
      "loss": 0.4944,
      "num_input_tokens_seen": 10105904,
      "step": 15500
    },
    {
      "epoch": 9.142099056603774,
      "grad_norm": 3.4416561126708984,
      "learning_rate": 6.5935480625358615e-06,
      "loss": 0.3883,
      "num_input_tokens_seen": 10108976,
      "step": 15505
    },
    {
      "epoch": 9.14504716981132,
      "grad_norm": 3.0720560550689697,
      "learning_rate": 6.591109298436216e-06,
      "loss": 0.3664,
      "num_input_tokens_seen": 10112048,
      "step": 15510
    },
    {
      "epoch": 9.147995283018869,
      "grad_norm": 2.3089096546173096,
      "learning_rate": 6.5886701130835255e-06,
      "loss": 0.336,
      "num_input_tokens_seen": 10115344,
      "step": 15515
    },
    {
      "epoch": 9.150943396226415,
      "grad_norm": 2.373603105545044,
      "learning_rate": 6.586230507123574e-06,
      "loss": 0.2463,
      "num_input_tokens_seen": 10118288,
      "step": 15520
    },
    {
      "epoch": 9.153891509433961,
      "grad_norm": 4.4341301918029785,
      "learning_rate": 6.583790481202261e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 10122000,
      "step": 15525
    },
    {
      "epoch": 9.15683962264151,
      "grad_norm": 2.1818697452545166,
      "learning_rate": 6.5813500359655925e-06,
      "loss": 0.4549,
      "num_input_tokens_seen": 10124976,
      "step": 15530
    },
    {
      "epoch": 9.159787735849056,
      "grad_norm": 2.5069644451141357,
      "learning_rate": 6.578909172059687e-06,
      "loss": 0.3342,
      "num_input_tokens_seen": 10129424,
      "step": 15535
    },
    {
      "epoch": 9.162735849056604,
      "grad_norm": 2.1268208026885986,
      "learning_rate": 6.576467890130772e-06,
      "loss": 0.5142,
      "num_input_tokens_seen": 10132336,
      "step": 15540
    },
    {
      "epoch": 9.16568396226415,
      "grad_norm": 1.7799327373504639,
      "learning_rate": 6.574026190825191e-06,
      "loss": 0.3993,
      "num_input_tokens_seen": 10135824,
      "step": 15545
    },
    {
      "epoch": 9.168632075471699,
      "grad_norm": 2.2952194213867188,
      "learning_rate": 6.57158407478939e-06,
      "loss": 0.584,
      "num_input_tokens_seen": 10139120,
      "step": 15550
    },
    {
      "epoch": 9.171580188679245,
      "grad_norm": 2.8885700702667236,
      "learning_rate": 6.569141542669935e-06,
      "loss": 0.3488,
      "num_input_tokens_seen": 10141584,
      "step": 15555
    },
    {
      "epoch": 9.174528301886792,
      "grad_norm": 3.753981351852417,
      "learning_rate": 6.566698595113492e-06,
      "loss": 0.4841,
      "num_input_tokens_seen": 10144048,
      "step": 15560
    },
    {
      "epoch": 9.17747641509434,
      "grad_norm": 1.7688887119293213,
      "learning_rate": 6.564255232766843e-06,
      "loss": 0.4002,
      "num_input_tokens_seen": 10147568,
      "step": 15565
    },
    {
      "epoch": 9.180424528301886,
      "grad_norm": 1.7112520933151245,
      "learning_rate": 6.561811456276881e-06,
      "loss": 0.3668,
      "num_input_tokens_seen": 10151088,
      "step": 15570
    },
    {
      "epoch": 9.183372641509434,
      "grad_norm": 2.597309112548828,
      "learning_rate": 6.559367266290605e-06,
      "loss": 0.4924,
      "num_input_tokens_seen": 10154352,
      "step": 15575
    },
    {
      "epoch": 9.18632075471698,
      "grad_norm": 1.7510815858840942,
      "learning_rate": 6.556922663455123e-06,
      "loss": 0.4176,
      "num_input_tokens_seen": 10157776,
      "step": 15580
    },
    {
      "epoch": 9.189268867924529,
      "grad_norm": 3.1420514583587646,
      "learning_rate": 6.554477648417657e-06,
      "loss": 0.2943,
      "num_input_tokens_seen": 10160528,
      "step": 15585
    },
    {
      "epoch": 9.192216981132075,
      "grad_norm": 1.717568039894104,
      "learning_rate": 6.552032221825535e-06,
      "loss": 0.439,
      "num_input_tokens_seen": 10164080,
      "step": 15590
    },
    {
      "epoch": 9.195165094339623,
      "grad_norm": 1.507290005683899,
      "learning_rate": 6.549586384326192e-06,
      "loss": 0.3956,
      "num_input_tokens_seen": 10167600,
      "step": 15595
    },
    {
      "epoch": 9.19811320754717,
      "grad_norm": 4.158390045166016,
      "learning_rate": 6.547140136567176e-06,
      "loss": 0.3982,
      "num_input_tokens_seen": 10170512,
      "step": 15600
    },
    {
      "epoch": 9.201061320754716,
      "grad_norm": 1.8670555353164673,
      "learning_rate": 6.544693479196142e-06,
      "loss": 0.3391,
      "num_input_tokens_seen": 10174928,
      "step": 15605
    },
    {
      "epoch": 9.204009433962264,
      "grad_norm": 2.9960825443267822,
      "learning_rate": 6.542246412860851e-06,
      "loss": 0.44,
      "num_input_tokens_seen": 10177552,
      "step": 15610
    },
    {
      "epoch": 9.20695754716981,
      "grad_norm": 2.3815324306488037,
      "learning_rate": 6.5397989382091754e-06,
      "loss": 0.5209,
      "num_input_tokens_seen": 10181840,
      "step": 15615
    },
    {
      "epoch": 9.209905660377359,
      "grad_norm": 2.127275228500366,
      "learning_rate": 6.537351055889096e-06,
      "loss": 0.2597,
      "num_input_tokens_seen": 10184880,
      "step": 15620
    },
    {
      "epoch": 9.212853773584905,
      "grad_norm": 3.5977749824523926,
      "learning_rate": 6.534902766548698e-06,
      "loss": 0.3355,
      "num_input_tokens_seen": 10188272,
      "step": 15625
    },
    {
      "epoch": 9.215801886792454,
      "grad_norm": 2.0789361000061035,
      "learning_rate": 6.532454070836176e-06,
      "loss": 0.3663,
      "num_input_tokens_seen": 10190704,
      "step": 15630
    },
    {
      "epoch": 9.21875,
      "grad_norm": 2.914146900177002,
      "learning_rate": 6.530004969399836e-06,
      "loss": 0.3209,
      "num_input_tokens_seen": 10194096,
      "step": 15635
    },
    {
      "epoch": 9.221698113207546,
      "grad_norm": 2.6274631023406982,
      "learning_rate": 6.5275554628880865e-06,
      "loss": 0.2777,
      "num_input_tokens_seen": 10197616,
      "step": 15640
    },
    {
      "epoch": 9.224646226415095,
      "grad_norm": 2.249664068222046,
      "learning_rate": 6.525105551949444e-06,
      "loss": 0.4667,
      "num_input_tokens_seen": 10201392,
      "step": 15645
    },
    {
      "epoch": 9.227594339622641,
      "grad_norm": 4.6555094718933105,
      "learning_rate": 6.522655237232535e-06,
      "loss": 0.4882,
      "num_input_tokens_seen": 10204304,
      "step": 15650
    },
    {
      "epoch": 9.23054245283019,
      "grad_norm": 3.014333724975586,
      "learning_rate": 6.5202045193860885e-06,
      "loss": 0.3778,
      "num_input_tokens_seen": 10207280,
      "step": 15655
    },
    {
      "epoch": 9.233490566037736,
      "grad_norm": 2.7347233295440674,
      "learning_rate": 6.517753399058944e-06,
      "loss": 0.3208,
      "num_input_tokens_seen": 10209904,
      "step": 15660
    },
    {
      "epoch": 9.236438679245284,
      "grad_norm": 2.4949793815612793,
      "learning_rate": 6.515301876900047e-06,
      "loss": 0.6337,
      "num_input_tokens_seen": 10213360,
      "step": 15665
    },
    {
      "epoch": 9.23938679245283,
      "grad_norm": 3.1196861267089844,
      "learning_rate": 6.512849953558445e-06,
      "loss": 0.4203,
      "num_input_tokens_seen": 10216496,
      "step": 15670
    },
    {
      "epoch": 9.242334905660377,
      "grad_norm": 1.582646131515503,
      "learning_rate": 6.510397629683301e-06,
      "loss": 0.3568,
      "num_input_tokens_seen": 10219984,
      "step": 15675
    },
    {
      "epoch": 9.245283018867925,
      "grad_norm": 2.570546865463257,
      "learning_rate": 6.507944905923872e-06,
      "loss": 0.3507,
      "num_input_tokens_seen": 10223280,
      "step": 15680
    },
    {
      "epoch": 9.248231132075471,
      "grad_norm": 2.599398612976074,
      "learning_rate": 6.505491782929531e-06,
      "loss": 0.3952,
      "num_input_tokens_seen": 10226640,
      "step": 15685
    },
    {
      "epoch": 9.25117924528302,
      "grad_norm": 1.7579734325408936,
      "learning_rate": 6.5030382613497535e-06,
      "loss": 0.3722,
      "num_input_tokens_seen": 10230288,
      "step": 15690
    },
    {
      "epoch": 9.254127358490566,
      "grad_norm": 5.006028175354004,
      "learning_rate": 6.500584341834119e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 10233136,
      "step": 15695
    },
    {
      "epoch": 9.257075471698114,
      "grad_norm": 3.505951404571533,
      "learning_rate": 6.498130025032312e-06,
      "loss": 0.3248,
      "num_input_tokens_seen": 10235792,
      "step": 15700
    },
    {
      "epoch": 9.26002358490566,
      "grad_norm": 2.3119189739227295,
      "learning_rate": 6.495675311594123e-06,
      "loss": 0.3616,
      "num_input_tokens_seen": 10239088,
      "step": 15705
    },
    {
      "epoch": 9.262971698113208,
      "grad_norm": 2.6068005561828613,
      "learning_rate": 6.493220202169452e-06,
      "loss": 0.3967,
      "num_input_tokens_seen": 10242320,
      "step": 15710
    },
    {
      "epoch": 9.265919811320755,
      "grad_norm": 2.2857918739318848,
      "learning_rate": 6.490764697408295e-06,
      "loss": 0.3508,
      "num_input_tokens_seen": 10246160,
      "step": 15715
    },
    {
      "epoch": 9.268867924528301,
      "grad_norm": 0.8318555951118469,
      "learning_rate": 6.48830879796076e-06,
      "loss": 0.2591,
      "num_input_tokens_seen": 10251632,
      "step": 15720
    },
    {
      "epoch": 9.27181603773585,
      "grad_norm": 2.582273483276367,
      "learning_rate": 6.4858525044770546e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 10254224,
      "step": 15725
    },
    {
      "epoch": 9.274764150943396,
      "grad_norm": 3.2986502647399902,
      "learning_rate": 6.483395817607497e-06,
      "loss": 0.3786,
      "num_input_tokens_seen": 10256976,
      "step": 15730
    },
    {
      "epoch": 9.277712264150944,
      "grad_norm": 3.8442792892456055,
      "learning_rate": 6.480938738002504e-06,
      "loss": 0.3651,
      "num_input_tokens_seen": 10260368,
      "step": 15735
    },
    {
      "epoch": 9.28066037735849,
      "grad_norm": 2.140795946121216,
      "learning_rate": 6.478481266312597e-06,
      "loss": 0.37,
      "num_input_tokens_seen": 10263920,
      "step": 15740
    },
    {
      "epoch": 9.283608490566039,
      "grad_norm": 1.9786944389343262,
      "learning_rate": 6.476023403188403e-06,
      "loss": 0.357,
      "num_input_tokens_seen": 10268048,
      "step": 15745
    },
    {
      "epoch": 9.286556603773585,
      "grad_norm": 1.7013970613479614,
      "learning_rate": 6.473565149280651e-06,
      "loss": 0.4381,
      "num_input_tokens_seen": 10271280,
      "step": 15750
    },
    {
      "epoch": 9.289504716981131,
      "grad_norm": 1.8829222917556763,
      "learning_rate": 6.471106505240175e-06,
      "loss": 0.4675,
      "num_input_tokens_seen": 10274256,
      "step": 15755
    },
    {
      "epoch": 9.29245283018868,
      "grad_norm": 2.178178071975708,
      "learning_rate": 6.468647471717914e-06,
      "loss": 0.384,
      "num_input_tokens_seen": 10276784,
      "step": 15760
    },
    {
      "epoch": 9.295400943396226,
      "grad_norm": 4.0366997718811035,
      "learning_rate": 6.466188049364902e-06,
      "loss": 0.4048,
      "num_input_tokens_seen": 10279024,
      "step": 15765
    },
    {
      "epoch": 9.298349056603774,
      "grad_norm": 2.3506648540496826,
      "learning_rate": 6.463728238832288e-06,
      "loss": 0.4125,
      "num_input_tokens_seen": 10282352,
      "step": 15770
    },
    {
      "epoch": 9.30129716981132,
      "grad_norm": 1.6107803583145142,
      "learning_rate": 6.461268040771311e-06,
      "loss": 0.31,
      "num_input_tokens_seen": 10285104,
      "step": 15775
    },
    {
      "epoch": 9.304245283018869,
      "grad_norm": 2.6596596240997314,
      "learning_rate": 6.458807455833326e-06,
      "loss": 0.4125,
      "num_input_tokens_seen": 10287824,
      "step": 15780
    },
    {
      "epoch": 9.307193396226415,
      "grad_norm": 1.9756290912628174,
      "learning_rate": 6.456346484669778e-06,
      "loss": 0.4139,
      "num_input_tokens_seen": 10292016,
      "step": 15785
    },
    {
      "epoch": 9.310141509433961,
      "grad_norm": 2.5869123935699463,
      "learning_rate": 6.4538851279322225e-06,
      "loss": 0.3732,
      "num_input_tokens_seen": 10295568,
      "step": 15790
    },
    {
      "epoch": 9.31308962264151,
      "grad_norm": 2.1338016986846924,
      "learning_rate": 6.451423386272312e-06,
      "loss": 0.3142,
      "num_input_tokens_seen": 10302928,
      "step": 15795
    },
    {
      "epoch": 9.316037735849056,
      "grad_norm": 1.8033570051193237,
      "learning_rate": 6.448961260341806e-06,
      "loss": 0.4111,
      "num_input_tokens_seen": 10306544,
      "step": 15800
    },
    {
      "epoch": 9.318985849056604,
      "grad_norm": 1.202943205833435,
      "learning_rate": 6.446498750792563e-06,
      "loss": 0.3325,
      "num_input_tokens_seen": 10309424,
      "step": 15805
    },
    {
      "epoch": 9.32193396226415,
      "grad_norm": 4.276802062988281,
      "learning_rate": 6.444035858276538e-06,
      "loss": 0.3642,
      "num_input_tokens_seen": 10312496,
      "step": 15810
    },
    {
      "epoch": 9.324882075471699,
      "grad_norm": 4.620153427124023,
      "learning_rate": 6.441572583445799e-06,
      "loss": 0.3718,
      "num_input_tokens_seen": 10315376,
      "step": 15815
    },
    {
      "epoch": 9.327830188679245,
      "grad_norm": 2.3456342220306396,
      "learning_rate": 6.439108926952504e-06,
      "loss": 0.3244,
      "num_input_tokens_seen": 10318544,
      "step": 15820
    },
    {
      "epoch": 9.330778301886792,
      "grad_norm": 2.2263245582580566,
      "learning_rate": 6.43664488944892e-06,
      "loss": 0.4835,
      "num_input_tokens_seen": 10321904,
      "step": 15825
    },
    {
      "epoch": 9.33372641509434,
      "grad_norm": 2.76908278465271,
      "learning_rate": 6.434180471587409e-06,
      "loss": 0.3416,
      "num_input_tokens_seen": 10325392,
      "step": 15830
    },
    {
      "epoch": 9.336674528301886,
      "grad_norm": 2.044100284576416,
      "learning_rate": 6.431715674020438e-06,
      "loss": 0.3338,
      "num_input_tokens_seen": 10328304,
      "step": 15835
    },
    {
      "epoch": 9.339622641509434,
      "grad_norm": 2.7489852905273438,
      "learning_rate": 6.429250497400571e-06,
      "loss": 0.4561,
      "num_input_tokens_seen": 10334224,
      "step": 15840
    },
    {
      "epoch": 9.34257075471698,
      "grad_norm": 7.595127582550049,
      "learning_rate": 6.426784942380475e-06,
      "loss": 0.501,
      "num_input_tokens_seen": 10337200,
      "step": 15845
    },
    {
      "epoch": 9.345518867924529,
      "grad_norm": 12.796364784240723,
      "learning_rate": 6.424319009612917e-06,
      "loss": 0.3969,
      "num_input_tokens_seen": 10340560,
      "step": 15850
    },
    {
      "epoch": 9.348466981132075,
      "grad_norm": 2.534548759460449,
      "learning_rate": 6.421852699750763e-06,
      "loss": 0.3936,
      "num_input_tokens_seen": 10343856,
      "step": 15855
    },
    {
      "epoch": 9.351415094339623,
      "grad_norm": 2.4600634574890137,
      "learning_rate": 6.4193860134469775e-06,
      "loss": 0.3257,
      "num_input_tokens_seen": 10346224,
      "step": 15860
    },
    {
      "epoch": 9.35436320754717,
      "grad_norm": 2.678802013397217,
      "learning_rate": 6.416918951354629e-06,
      "loss": 0.3133,
      "num_input_tokens_seen": 10349296,
      "step": 15865
    },
    {
      "epoch": 9.357311320754716,
      "grad_norm": 2.17466402053833,
      "learning_rate": 6.4144515141268816e-06,
      "loss": 0.2675,
      "num_input_tokens_seen": 10352176,
      "step": 15870
    },
    {
      "epoch": 9.360259433962264,
      "grad_norm": 2.9604885578155518,
      "learning_rate": 6.411983702416999e-06,
      "loss": 0.3042,
      "num_input_tokens_seen": 10355472,
      "step": 15875
    },
    {
      "epoch": 9.36320754716981,
      "grad_norm": 4.104915618896484,
      "learning_rate": 6.409515516878346e-06,
      "loss": 0.3803,
      "num_input_tokens_seen": 10358608,
      "step": 15880
    },
    {
      "epoch": 9.366155660377359,
      "grad_norm": 4.079616546630859,
      "learning_rate": 6.407046958164387e-06,
      "loss": 0.4691,
      "num_input_tokens_seen": 10362000,
      "step": 15885
    },
    {
      "epoch": 9.369103773584905,
      "grad_norm": 2.215550184249878,
      "learning_rate": 6.404578026928679e-06,
      "loss": 0.4357,
      "num_input_tokens_seen": 10364784,
      "step": 15890
    },
    {
      "epoch": 9.372051886792454,
      "grad_norm": 3.5875542163848877,
      "learning_rate": 6.402108723824887e-06,
      "loss": 0.4585,
      "num_input_tokens_seen": 10367760,
      "step": 15895
    },
    {
      "epoch": 9.375,
      "grad_norm": 1.7194088697433472,
      "learning_rate": 6.399639049506767e-06,
      "loss": 0.3507,
      "num_input_tokens_seen": 10374096,
      "step": 15900
    },
    {
      "epoch": 9.377948113207546,
      "grad_norm": 5.651604652404785,
      "learning_rate": 6.397169004628177e-06,
      "loss": 0.4052,
      "num_input_tokens_seen": 10377168,
      "step": 15905
    },
    {
      "epoch": 9.380896226415095,
      "grad_norm": 2.679089069366455,
      "learning_rate": 6.39469858984307e-06,
      "loss": 0.5029,
      "num_input_tokens_seen": 10380400,
      "step": 15910
    },
    {
      "epoch": 9.383844339622641,
      "grad_norm": 2.7768306732177734,
      "learning_rate": 6.3922278058055024e-06,
      "loss": 0.3237,
      "num_input_tokens_seen": 10383248,
      "step": 15915
    },
    {
      "epoch": 9.38679245283019,
      "grad_norm": 1.9655019044876099,
      "learning_rate": 6.389756653169622e-06,
      "loss": 0.3945,
      "num_input_tokens_seen": 10386320,
      "step": 15920
    },
    {
      "epoch": 9.389740566037736,
      "grad_norm": 2.027432680130005,
      "learning_rate": 6.387285132589678e-06,
      "loss": 0.3749,
      "num_input_tokens_seen": 10389424,
      "step": 15925
    },
    {
      "epoch": 9.392688679245284,
      "grad_norm": 2.1318700313568115,
      "learning_rate": 6.3848132447200166e-06,
      "loss": 0.3635,
      "num_input_tokens_seen": 10393072,
      "step": 15930
    },
    {
      "epoch": 9.39563679245283,
      "grad_norm": 2.002270221710205,
      "learning_rate": 6.38234099021508e-06,
      "loss": 0.3938,
      "num_input_tokens_seen": 10396304,
      "step": 15935
    },
    {
      "epoch": 9.398584905660377,
      "grad_norm": 2.3423099517822266,
      "learning_rate": 6.379868369729409e-06,
      "loss": 0.333,
      "num_input_tokens_seen": 10399664,
      "step": 15940
    },
    {
      "epoch": 9.401533018867925,
      "grad_norm": 2.790731191635132,
      "learning_rate": 6.377395383917642e-06,
      "loss": 0.4399,
      "num_input_tokens_seen": 10402896,
      "step": 15945
    },
    {
      "epoch": 9.404481132075471,
      "grad_norm": 2.076799154281616,
      "learning_rate": 6.374922033434507e-06,
      "loss": 0.3293,
      "num_input_tokens_seen": 10405296,
      "step": 15950
    },
    {
      "epoch": 9.40742924528302,
      "grad_norm": 2.4376301765441895,
      "learning_rate": 6.372448318934842e-06,
      "loss": 0.4693,
      "num_input_tokens_seen": 10408496,
      "step": 15955
    },
    {
      "epoch": 9.410377358490566,
      "grad_norm": 2.3520314693450928,
      "learning_rate": 6.369974241073569e-06,
      "loss": 0.3165,
      "num_input_tokens_seen": 10411888,
      "step": 15960
    },
    {
      "epoch": 9.413325471698114,
      "grad_norm": 2.0389559268951416,
      "learning_rate": 6.367499800505709e-06,
      "loss": 0.36,
      "num_input_tokens_seen": 10415216,
      "step": 15965
    },
    {
      "epoch": 9.41627358490566,
      "grad_norm": 3.260810613632202,
      "learning_rate": 6.365024997886384e-06,
      "loss": 0.3442,
      "num_input_tokens_seen": 10418288,
      "step": 15970
    },
    {
      "epoch": 9.419221698113208,
      "grad_norm": 2.6975209712982178,
      "learning_rate": 6.362549833870808e-06,
      "loss": 0.3624,
      "num_input_tokens_seen": 10420912,
      "step": 15975
    },
    {
      "epoch": 9.422169811320755,
      "grad_norm": 1.567344069480896,
      "learning_rate": 6.360074309114293e-06,
      "loss": 0.3843,
      "num_input_tokens_seen": 10423408,
      "step": 15980
    },
    {
      "epoch": 9.425117924528301,
      "grad_norm": 2.631965398788452,
      "learning_rate": 6.357598424272241e-06,
      "loss": 0.3235,
      "num_input_tokens_seen": 10427792,
      "step": 15985
    },
    {
      "epoch": 9.42806603773585,
      "grad_norm": 3.1568689346313477,
      "learning_rate": 6.355122180000156e-06,
      "loss": 0.4691,
      "num_input_tokens_seen": 10430928,
      "step": 15990
    },
    {
      "epoch": 9.431014150943396,
      "grad_norm": 3.0079305171966553,
      "learning_rate": 6.352645576953635e-06,
      "loss": 0.4132,
      "num_input_tokens_seen": 10434704,
      "step": 15995
    },
    {
      "epoch": 9.433962264150944,
      "grad_norm": 1.860533595085144,
      "learning_rate": 6.350168615788366e-06,
      "loss": 0.3727,
      "num_input_tokens_seen": 10438096,
      "step": 16000
    },
    {
      "epoch": 9.43691037735849,
      "grad_norm": 2.781571865081787,
      "learning_rate": 6.347691297160137e-06,
      "loss": 0.3399,
      "num_input_tokens_seen": 10441936,
      "step": 16005
    },
    {
      "epoch": 9.439858490566039,
      "grad_norm": 4.923755168914795,
      "learning_rate": 6.34521362172483e-06,
      "loss": 0.3689,
      "num_input_tokens_seen": 10445552,
      "step": 16010
    },
    {
      "epoch": 9.442806603773585,
      "grad_norm": 2.797358512878418,
      "learning_rate": 6.342735590138417e-06,
      "loss": 0.3888,
      "num_input_tokens_seen": 10448848,
      "step": 16015
    },
    {
      "epoch": 9.445754716981131,
      "grad_norm": 2.8626868724823,
      "learning_rate": 6.340257203056972e-06,
      "loss": 0.4097,
      "num_input_tokens_seen": 10452592,
      "step": 16020
    },
    {
      "epoch": 9.44870283018868,
      "grad_norm": 2.9776132106781006,
      "learning_rate": 6.3377784611366554e-06,
      "loss": 0.439,
      "num_input_tokens_seen": 10456272,
      "step": 16025
    },
    {
      "epoch": 9.451650943396226,
      "grad_norm": 3.118195056915283,
      "learning_rate": 6.335299365033726e-06,
      "loss": 0.4599,
      "num_input_tokens_seen": 10460080,
      "step": 16030
    },
    {
      "epoch": 9.454599056603774,
      "grad_norm": 2.9336352348327637,
      "learning_rate": 6.3328199154045346e-06,
      "loss": 0.3853,
      "num_input_tokens_seen": 10463248,
      "step": 16035
    },
    {
      "epoch": 9.45754716981132,
      "grad_norm": 1.3094220161437988,
      "learning_rate": 6.3303401129055265e-06,
      "loss": 0.4637,
      "num_input_tokens_seen": 10466896,
      "step": 16040
    },
    {
      "epoch": 9.460495283018869,
      "grad_norm": 1.9588351249694824,
      "learning_rate": 6.32785995819324e-06,
      "loss": 0.3629,
      "num_input_tokens_seen": 10470128,
      "step": 16045
    },
    {
      "epoch": 9.463443396226415,
      "grad_norm": 2.59356951713562,
      "learning_rate": 6.3253794519243075e-06,
      "loss": 0.4269,
      "num_input_tokens_seen": 10473840,
      "step": 16050
    },
    {
      "epoch": 9.466391509433961,
      "grad_norm": 2.6218159198760986,
      "learning_rate": 6.322898594755452e-06,
      "loss": 0.3994,
      "num_input_tokens_seen": 10477424,
      "step": 16055
    },
    {
      "epoch": 9.46933962264151,
      "grad_norm": 1.5984671115875244,
      "learning_rate": 6.320417387343492e-06,
      "loss": 0.3488,
      "num_input_tokens_seen": 10480688,
      "step": 16060
    },
    {
      "epoch": 9.472287735849056,
      "grad_norm": 2.4238035678863525,
      "learning_rate": 6.3179358303453386e-06,
      "loss": 0.357,
      "num_input_tokens_seen": 10484720,
      "step": 16065
    },
    {
      "epoch": 9.475235849056604,
      "grad_norm": 2.3635451793670654,
      "learning_rate": 6.315453924417995e-06,
      "loss": 0.3579,
      "num_input_tokens_seen": 10487952,
      "step": 16070
    },
    {
      "epoch": 9.47818396226415,
      "grad_norm": 2.6382174491882324,
      "learning_rate": 6.312971670218554e-06,
      "loss": 0.3065,
      "num_input_tokens_seen": 10491664,
      "step": 16075
    },
    {
      "epoch": 9.481132075471699,
      "grad_norm": 1.791094183921814,
      "learning_rate": 6.3104890684042055e-06,
      "loss": 0.4577,
      "num_input_tokens_seen": 10494416,
      "step": 16080
    },
    {
      "epoch": 9.484080188679245,
      "grad_norm": 3.0495352745056152,
      "learning_rate": 6.308006119632228e-06,
      "loss": 0.3334,
      "num_input_tokens_seen": 10496816,
      "step": 16085
    },
    {
      "epoch": 9.487028301886792,
      "grad_norm": 2.028726816177368,
      "learning_rate": 6.305522824559993e-06,
      "loss": 0.3879,
      "num_input_tokens_seen": 10500304,
      "step": 16090
    },
    {
      "epoch": 9.48997641509434,
      "grad_norm": 2.6754910945892334,
      "learning_rate": 6.303039183844965e-06,
      "loss": 0.4932,
      "num_input_tokens_seen": 10502992,
      "step": 16095
    },
    {
      "epoch": 9.492924528301886,
      "grad_norm": 3.8133046627044678,
      "learning_rate": 6.300555198144697e-06,
      "loss": 0.4101,
      "num_input_tokens_seen": 10505200,
      "step": 16100
    },
    {
      "epoch": 9.495872641509434,
      "grad_norm": 2.085482120513916,
      "learning_rate": 6.2980708681168335e-06,
      "loss": 0.4393,
      "num_input_tokens_seen": 10508176,
      "step": 16105
    },
    {
      "epoch": 9.49882075471698,
      "grad_norm": 2.685556411743164,
      "learning_rate": 6.2955861944191145e-06,
      "loss": 0.3644,
      "num_input_tokens_seen": 10510928,
      "step": 16110
    },
    {
      "epoch": 9.501768867924529,
      "grad_norm": 2.5360171794891357,
      "learning_rate": 6.293101177709367e-06,
      "loss": 0.4075,
      "num_input_tokens_seen": 10514000,
      "step": 16115
    },
    {
      "epoch": 9.504716981132075,
      "grad_norm": 4.248354434967041,
      "learning_rate": 6.290615818645509e-06,
      "loss": 0.3591,
      "num_input_tokens_seen": 10517392,
      "step": 16120
    },
    {
      "epoch": 9.507665094339622,
      "grad_norm": 3.3375558853149414,
      "learning_rate": 6.288130117885552e-06,
      "loss": 0.4013,
      "num_input_tokens_seen": 10520688,
      "step": 16125
    },
    {
      "epoch": 9.51061320754717,
      "grad_norm": 2.1879544258117676,
      "learning_rate": 6.285644076087594e-06,
      "loss": 0.3585,
      "num_input_tokens_seen": 10523280,
      "step": 16130
    },
    {
      "epoch": 9.513561320754716,
      "grad_norm": 2.019651412963867,
      "learning_rate": 6.283157693909826e-06,
      "loss": 0.3922,
      "num_input_tokens_seen": 10526928,
      "step": 16135
    },
    {
      "epoch": 9.516509433962264,
      "grad_norm": 4.027740478515625,
      "learning_rate": 6.280670972010528e-06,
      "loss": 0.2761,
      "num_input_tokens_seen": 10529968,
      "step": 16140
    },
    {
      "epoch": 9.51945754716981,
      "grad_norm": 3.4406824111938477,
      "learning_rate": 6.278183911048072e-06,
      "loss": 0.3571,
      "num_input_tokens_seen": 10533200,
      "step": 16145
    },
    {
      "epoch": 9.522405660377359,
      "grad_norm": 3.4908158779144287,
      "learning_rate": 6.275696511680915e-06,
      "loss": 0.3913,
      "num_input_tokens_seen": 10536048,
      "step": 16150
    },
    {
      "epoch": 9.525353773584905,
      "grad_norm": 3.276486396789551,
      "learning_rate": 6.27320877456761e-06,
      "loss": 0.506,
      "num_input_tokens_seen": 10539760,
      "step": 16155
    },
    {
      "epoch": 9.528301886792454,
      "grad_norm": 2.7534854412078857,
      "learning_rate": 6.270720700366793e-06,
      "loss": 0.6174,
      "num_input_tokens_seen": 10544016,
      "step": 16160
    },
    {
      "epoch": 9.53125,
      "grad_norm": 1.7970536947250366,
      "learning_rate": 6.2682322897371974e-06,
      "loss": 0.4251,
      "num_input_tokens_seen": 10547984,
      "step": 16165
    },
    {
      "epoch": 9.534198113207546,
      "grad_norm": 2.959211587905884,
      "learning_rate": 6.265743543337634e-06,
      "loss": 0.3509,
      "num_input_tokens_seen": 10551568,
      "step": 16170
    },
    {
      "epoch": 9.537146226415095,
      "grad_norm": 1.4761197566986084,
      "learning_rate": 6.2632544618270156e-06,
      "loss": 0.2753,
      "num_input_tokens_seen": 10554608,
      "step": 16175
    },
    {
      "epoch": 9.540094339622641,
      "grad_norm": 2.2524514198303223,
      "learning_rate": 6.260765045864332e-06,
      "loss": 0.4476,
      "num_input_tokens_seen": 10558160,
      "step": 16180
    },
    {
      "epoch": 9.54304245283019,
      "grad_norm": 2.325085401535034,
      "learning_rate": 6.258275296108669e-06,
      "loss": 0.4223,
      "num_input_tokens_seen": 10560720,
      "step": 16185
    },
    {
      "epoch": 9.545990566037736,
      "grad_norm": 1.6599845886230469,
      "learning_rate": 6.2557852132191985e-06,
      "loss": 0.4747,
      "num_input_tokens_seen": 10564752,
      "step": 16190
    },
    {
      "epoch": 9.548938679245284,
      "grad_norm": 4.046700954437256,
      "learning_rate": 6.253294797855182e-06,
      "loss": 0.3604,
      "num_input_tokens_seen": 10567600,
      "step": 16195
    },
    {
      "epoch": 9.55188679245283,
      "grad_norm": 2.2707836627960205,
      "learning_rate": 6.250804050675964e-06,
      "loss": 0.4247,
      "num_input_tokens_seen": 10571280,
      "step": 16200
    },
    {
      "epoch": 9.554834905660378,
      "grad_norm": 3.4458892345428467,
      "learning_rate": 6.248312972340984e-06,
      "loss": 0.4556,
      "num_input_tokens_seen": 10574928,
      "step": 16205
    },
    {
      "epoch": 9.557783018867925,
      "grad_norm": 1.3727914094924927,
      "learning_rate": 6.2458215635097656e-06,
      "loss": 0.5248,
      "num_input_tokens_seen": 10578864,
      "step": 16210
    },
    {
      "epoch": 9.560731132075471,
      "grad_norm": 2.6479766368865967,
      "learning_rate": 6.2433298248419175e-06,
      "loss": 0.3577,
      "num_input_tokens_seen": 10581648,
      "step": 16215
    },
    {
      "epoch": 9.56367924528302,
      "grad_norm": 2.6002891063690186,
      "learning_rate": 6.2408377569971405e-06,
      "loss": 0.3685,
      "num_input_tokens_seen": 10585424,
      "step": 16220
    },
    {
      "epoch": 9.566627358490566,
      "grad_norm": 6.889472961425781,
      "learning_rate": 6.238345360635221e-06,
      "loss": 0.4587,
      "num_input_tokens_seen": 10587856,
      "step": 16225
    },
    {
      "epoch": 9.569575471698114,
      "grad_norm": 2.6355085372924805,
      "learning_rate": 6.2358526364160274e-06,
      "loss": 0.4316,
      "num_input_tokens_seen": 10591536,
      "step": 16230
    },
    {
      "epoch": 9.57252358490566,
      "grad_norm": 1.6638799905776978,
      "learning_rate": 6.2333595849995245e-06,
      "loss": 0.338,
      "num_input_tokens_seen": 10594768,
      "step": 16235
    },
    {
      "epoch": 9.575471698113208,
      "grad_norm": 2.500528335571289,
      "learning_rate": 6.230866207045756e-06,
      "loss": 0.3746,
      "num_input_tokens_seen": 10598384,
      "step": 16240
    },
    {
      "epoch": 9.578419811320755,
      "grad_norm": 2.298816204071045,
      "learning_rate": 6.228372503214853e-06,
      "loss": 0.2456,
      "num_input_tokens_seen": 10601776,
      "step": 16245
    },
    {
      "epoch": 9.581367924528301,
      "grad_norm": 3.5345942974090576,
      "learning_rate": 6.225878474167035e-06,
      "loss": 0.6167,
      "num_input_tokens_seen": 10607760,
      "step": 16250
    },
    {
      "epoch": 9.58431603773585,
      "grad_norm": 3.6032612323760986,
      "learning_rate": 6.22338412056261e-06,
      "loss": 0.6155,
      "num_input_tokens_seen": 10610832,
      "step": 16255
    },
    {
      "epoch": 9.587264150943396,
      "grad_norm": 1.9593465328216553,
      "learning_rate": 6.220889443061966e-06,
      "loss": 0.3425,
      "num_input_tokens_seen": 10614192,
      "step": 16260
    },
    {
      "epoch": 9.590212264150944,
      "grad_norm": 5.425792694091797,
      "learning_rate": 6.2183944423255796e-06,
      "loss": 0.4821,
      "num_input_tokens_seen": 10616848,
      "step": 16265
    },
    {
      "epoch": 9.59316037735849,
      "grad_norm": 3.7418839931488037,
      "learning_rate": 6.215899119014015e-06,
      "loss": 0.3529,
      "num_input_tokens_seen": 10620144,
      "step": 16270
    },
    {
      "epoch": 9.596108490566039,
      "grad_norm": 1.9273862838745117,
      "learning_rate": 6.2134034737879175e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 10622832,
      "step": 16275
    },
    {
      "epoch": 9.599056603773585,
      "grad_norm": 2.1943199634552,
      "learning_rate": 6.2109075073080205e-06,
      "loss": 0.4255,
      "num_input_tokens_seen": 10625552,
      "step": 16280
    },
    {
      "epoch": 9.602004716981131,
      "grad_norm": 2.018352508544922,
      "learning_rate": 6.2084112202351425e-06,
      "loss": 0.3471,
      "num_input_tokens_seen": 10628368,
      "step": 16285
    },
    {
      "epoch": 9.60495283018868,
      "grad_norm": 3.232924222946167,
      "learning_rate": 6.205914613230186e-06,
      "loss": 0.3244,
      "num_input_tokens_seen": 10631920,
      "step": 16290
    },
    {
      "epoch": 9.607900943396226,
      "grad_norm": 1.989395260810852,
      "learning_rate": 6.203417686954138e-06,
      "loss": 0.3593,
      "num_input_tokens_seen": 10635152,
      "step": 16295
    },
    {
      "epoch": 9.610849056603774,
      "grad_norm": 2.1981143951416016,
      "learning_rate": 6.2009204420680706e-06,
      "loss": 0.3664,
      "num_input_tokens_seen": 10638096,
      "step": 16300
    },
    {
      "epoch": 9.61379716981132,
      "grad_norm": 2.201779365539551,
      "learning_rate": 6.198422879233141e-06,
      "loss": 0.4008,
      "num_input_tokens_seen": 10641136,
      "step": 16305
    },
    {
      "epoch": 9.616745283018869,
      "grad_norm": 2.6620354652404785,
      "learning_rate": 6.1959249991105895e-06,
      "loss": 0.4521,
      "num_input_tokens_seen": 10644176,
      "step": 16310
    },
    {
      "epoch": 9.619693396226415,
      "grad_norm": 1.9121938943862915,
      "learning_rate": 6.19342680236174e-06,
      "loss": 0.3356,
      "num_input_tokens_seen": 10646608,
      "step": 16315
    },
    {
      "epoch": 9.622641509433961,
      "grad_norm": 3.327310562133789,
      "learning_rate": 6.190928289648003e-06,
      "loss": 0.4796,
      "num_input_tokens_seen": 10650672,
      "step": 16320
    },
    {
      "epoch": 9.62558962264151,
      "grad_norm": 2.184406042098999,
      "learning_rate": 6.188429461630866e-06,
      "loss": 0.3043,
      "num_input_tokens_seen": 10653680,
      "step": 16325
    },
    {
      "epoch": 9.628537735849056,
      "grad_norm": 4.023909091949463,
      "learning_rate": 6.185930318971909e-06,
      "loss": 0.3331,
      "num_input_tokens_seen": 10656976,
      "step": 16330
    },
    {
      "epoch": 9.631485849056604,
      "grad_norm": 2.4213967323303223,
      "learning_rate": 6.1834308623327885e-06,
      "loss": 0.441,
      "num_input_tokens_seen": 10659664,
      "step": 16335
    },
    {
      "epoch": 9.63443396226415,
      "grad_norm": 11.27309513092041,
      "learning_rate": 6.180931092375247e-06,
      "loss": 0.3928,
      "num_input_tokens_seen": 10663792,
      "step": 16340
    },
    {
      "epoch": 9.637382075471699,
      "grad_norm": 3.23309326171875,
      "learning_rate": 6.1784310097611075e-06,
      "loss": 0.3496,
      "num_input_tokens_seen": 10666160,
      "step": 16345
    },
    {
      "epoch": 9.640330188679245,
      "grad_norm": 2.8784143924713135,
      "learning_rate": 6.1759306151522815e-06,
      "loss": 0.345,
      "num_input_tokens_seen": 10668976,
      "step": 16350
    },
    {
      "epoch": 9.643278301886792,
      "grad_norm": 3.293720245361328,
      "learning_rate": 6.173429909210755e-06,
      "loss": 0.2711,
      "num_input_tokens_seen": 10672784,
      "step": 16355
    },
    {
      "epoch": 9.64622641509434,
      "grad_norm": 2.3329575061798096,
      "learning_rate": 6.170928892598606e-06,
      "loss": 0.4054,
      "num_input_tokens_seen": 10676336,
      "step": 16360
    },
    {
      "epoch": 9.649174528301886,
      "grad_norm": 1.9740569591522217,
      "learning_rate": 6.168427565977984e-06,
      "loss": 0.3172,
      "num_input_tokens_seen": 10679344,
      "step": 16365
    },
    {
      "epoch": 9.652122641509434,
      "grad_norm": 2.0854477882385254,
      "learning_rate": 6.165925930011129e-06,
      "loss": 0.3845,
      "num_input_tokens_seen": 10682416,
      "step": 16370
    },
    {
      "epoch": 9.65507075471698,
      "grad_norm": 2.4912619590759277,
      "learning_rate": 6.163423985360359e-06,
      "loss": 0.4313,
      "num_input_tokens_seen": 10685616,
      "step": 16375
    },
    {
      "epoch": 9.658018867924529,
      "grad_norm": 1.7878752946853638,
      "learning_rate": 6.160921732688076e-06,
      "loss": 0.3738,
      "num_input_tokens_seen": 10688848,
      "step": 16380
    },
    {
      "epoch": 9.660966981132075,
      "grad_norm": 2.225677013397217,
      "learning_rate": 6.158419172656759e-06,
      "loss": 0.5055,
      "num_input_tokens_seen": 10692048,
      "step": 16385
    },
    {
      "epoch": 9.663915094339622,
      "grad_norm": 3.603638172149658,
      "learning_rate": 6.155916305928974e-06,
      "loss": 0.4606,
      "num_input_tokens_seen": 10695472,
      "step": 16390
    },
    {
      "epoch": 9.66686320754717,
      "grad_norm": 2.2072694301605225,
      "learning_rate": 6.153413133167366e-06,
      "loss": 0.3599,
      "num_input_tokens_seen": 10698768,
      "step": 16395
    },
    {
      "epoch": 9.669811320754716,
      "grad_norm": 2.3150978088378906,
      "learning_rate": 6.1509096550346596e-06,
      "loss": 0.3355,
      "num_input_tokens_seen": 10702288,
      "step": 16400
    },
    {
      "epoch": 9.672759433962264,
      "grad_norm": 1.9942560195922852,
      "learning_rate": 6.148405872193661e-06,
      "loss": 0.4104,
      "num_input_tokens_seen": 10706096,
      "step": 16405
    },
    {
      "epoch": 9.67570754716981,
      "grad_norm": 1.9642401933670044,
      "learning_rate": 6.14590178530726e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 10709392,
      "step": 16410
    },
    {
      "epoch": 9.678655660377359,
      "grad_norm": 2.665734052658081,
      "learning_rate": 6.143397395038422e-06,
      "loss": 0.3584,
      "num_input_tokens_seen": 10713360,
      "step": 16415
    },
    {
      "epoch": 9.681603773584905,
      "grad_norm": 2.785623550415039,
      "learning_rate": 6.140892702050196e-06,
      "loss": 0.4353,
      "num_input_tokens_seen": 10716464,
      "step": 16420
    },
    {
      "epoch": 9.684551886792454,
      "grad_norm": 3.070402145385742,
      "learning_rate": 6.138387707005711e-06,
      "loss": 0.4035,
      "num_input_tokens_seen": 10719824,
      "step": 16425
    },
    {
      "epoch": 9.6875,
      "grad_norm": 2.493583917617798,
      "learning_rate": 6.135882410568172e-06,
      "loss": 0.4111,
      "num_input_tokens_seen": 10722768,
      "step": 16430
    },
    {
      "epoch": 9.690448113207546,
      "grad_norm": 2.19046688079834,
      "learning_rate": 6.133376813400872e-06,
      "loss": 0.4062,
      "num_input_tokens_seen": 10724944,
      "step": 16435
    },
    {
      "epoch": 9.693396226415095,
      "grad_norm": 2.1911120414733887,
      "learning_rate": 6.130870916167175e-06,
      "loss": 0.4833,
      "num_input_tokens_seen": 10728048,
      "step": 16440
    },
    {
      "epoch": 9.696344339622641,
      "grad_norm": 1.8458647727966309,
      "learning_rate": 6.128364719530528e-06,
      "loss": 0.5316,
      "num_input_tokens_seen": 10730768,
      "step": 16445
    },
    {
      "epoch": 9.69929245283019,
      "grad_norm": 2.6938154697418213,
      "learning_rate": 6.125858224154459e-06,
      "loss": 0.4095,
      "num_input_tokens_seen": 10734000,
      "step": 16450
    },
    {
      "epoch": 9.702240566037736,
      "grad_norm": 3.6339621543884277,
      "learning_rate": 6.123351430702576e-06,
      "loss": 0.3504,
      "num_input_tokens_seen": 10736592,
      "step": 16455
    },
    {
      "epoch": 9.705188679245284,
      "grad_norm": 1.8029247522354126,
      "learning_rate": 6.1208443398385575e-06,
      "loss": 0.294,
      "num_input_tokens_seen": 10739888,
      "step": 16460
    },
    {
      "epoch": 9.70813679245283,
      "grad_norm": 1.8889799118041992,
      "learning_rate": 6.118336952226169e-06,
      "loss": 0.5239,
      "num_input_tokens_seen": 10743984,
      "step": 16465
    },
    {
      "epoch": 9.711084905660378,
      "grad_norm": 2.1132171154022217,
      "learning_rate": 6.115829268529254e-06,
      "loss": 0.3339,
      "num_input_tokens_seen": 10747792,
      "step": 16470
    },
    {
      "epoch": 9.714033018867925,
      "grad_norm": 3.6159753799438477,
      "learning_rate": 6.1133212894117326e-06,
      "loss": 0.3415,
      "num_input_tokens_seen": 10751152,
      "step": 16475
    },
    {
      "epoch": 9.716981132075471,
      "grad_norm": 3.731701374053955,
      "learning_rate": 6.1108130155375986e-06,
      "loss": 0.4803,
      "num_input_tokens_seen": 10754192,
      "step": 16480
    },
    {
      "epoch": 9.71992924528302,
      "grad_norm": 1.4406051635742188,
      "learning_rate": 6.108304447570933e-06,
      "loss": 0.3669,
      "num_input_tokens_seen": 10757584,
      "step": 16485
    },
    {
      "epoch": 9.722877358490566,
      "grad_norm": 2.607942581176758,
      "learning_rate": 6.105795586175888e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 10760816,
      "step": 16490
    },
    {
      "epoch": 9.725825471698114,
      "grad_norm": 3.76261830329895,
      "learning_rate": 6.1032864320166954e-06,
      "loss": 0.4062,
      "num_input_tokens_seen": 10764880,
      "step": 16495
    },
    {
      "epoch": 9.72877358490566,
      "grad_norm": 3.5371530055999756,
      "learning_rate": 6.100776985757666e-06,
      "loss": 0.358,
      "num_input_tokens_seen": 10767632,
      "step": 16500
    },
    {
      "epoch": 9.731721698113208,
      "grad_norm": 2.4419028759002686,
      "learning_rate": 6.098267248063186e-06,
      "loss": 0.4597,
      "num_input_tokens_seen": 10770288,
      "step": 16505
    },
    {
      "epoch": 9.734669811320755,
      "grad_norm": 5.362765789031982,
      "learning_rate": 6.0957572195977165e-06,
      "loss": 0.4452,
      "num_input_tokens_seen": 10773296,
      "step": 16510
    },
    {
      "epoch": 9.737617924528301,
      "grad_norm": 3.4774794578552246,
      "learning_rate": 6.0932469010258025e-06,
      "loss": 0.4105,
      "num_input_tokens_seen": 10776048,
      "step": 16515
    },
    {
      "epoch": 9.74056603773585,
      "grad_norm": 3.1349384784698486,
      "learning_rate": 6.0907362930120594e-06,
      "loss": 0.3266,
      "num_input_tokens_seen": 10780080,
      "step": 16520
    },
    {
      "epoch": 9.743514150943396,
      "grad_norm": 3.8733880519866943,
      "learning_rate": 6.088225396221181e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 10783344,
      "step": 16525
    },
    {
      "epoch": 9.746462264150944,
      "grad_norm": 2.1874942779541016,
      "learning_rate": 6.0857142113179415e-06,
      "loss": 0.3171,
      "num_input_tokens_seen": 10786000,
      "step": 16530
    },
    {
      "epoch": 9.74941037735849,
      "grad_norm": 2.977844715118408,
      "learning_rate": 6.083202738967182e-06,
      "loss": 0.4202,
      "num_input_tokens_seen": 10789744,
      "step": 16535
    },
    {
      "epoch": 9.752358490566039,
      "grad_norm": 2.73738956451416,
      "learning_rate": 6.0806909798338324e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 10793168,
      "step": 16540
    },
    {
      "epoch": 9.755306603773585,
      "grad_norm": 1.7961540222167969,
      "learning_rate": 6.0781789345828854e-06,
      "loss": 0.3651,
      "num_input_tokens_seen": 10796016,
      "step": 16545
    },
    {
      "epoch": 9.758254716981131,
      "grad_norm": 3.155907392501831,
      "learning_rate": 6.0756666038794195e-06,
      "loss": 0.4898,
      "num_input_tokens_seen": 10798928,
      "step": 16550
    },
    {
      "epoch": 9.76120283018868,
      "grad_norm": 1.6378077268600464,
      "learning_rate": 6.073153988388586e-06,
      "loss": 0.3439,
      "num_input_tokens_seen": 10802416,
      "step": 16555
    },
    {
      "epoch": 9.764150943396226,
      "grad_norm": 2.97456693649292,
      "learning_rate": 6.070641088775608e-06,
      "loss": 0.3928,
      "num_input_tokens_seen": 10805104,
      "step": 16560
    },
    {
      "epoch": 9.767099056603774,
      "grad_norm": 2.215664863586426,
      "learning_rate": 6.068127905705787e-06,
      "loss": 0.4216,
      "num_input_tokens_seen": 10808496,
      "step": 16565
    },
    {
      "epoch": 9.77004716981132,
      "grad_norm": 2.191291570663452,
      "learning_rate": 6.065614439844501e-06,
      "loss": 0.4061,
      "num_input_tokens_seen": 10811632,
      "step": 16570
    },
    {
      "epoch": 9.772995283018869,
      "grad_norm": 2.213578224182129,
      "learning_rate": 6.063100691857198e-06,
      "loss": 0.4434,
      "num_input_tokens_seen": 10814640,
      "step": 16575
    },
    {
      "epoch": 9.775943396226415,
      "grad_norm": 4.050931930541992,
      "learning_rate": 6.060586662409407e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 10817840,
      "step": 16580
    },
    {
      "epoch": 9.778891509433961,
      "grad_norm": 1.9550971984863281,
      "learning_rate": 6.058072352166724e-06,
      "loss": 0.4667,
      "num_input_tokens_seen": 10820944,
      "step": 16585
    },
    {
      "epoch": 9.78183962264151,
      "grad_norm": 2.4437146186828613,
      "learning_rate": 6.055557761794826e-06,
      "loss": 0.3563,
      "num_input_tokens_seen": 10823760,
      "step": 16590
    },
    {
      "epoch": 9.784787735849056,
      "grad_norm": 2.827178955078125,
      "learning_rate": 6.053042891959462e-06,
      "loss": 0.3519,
      "num_input_tokens_seen": 10826224,
      "step": 16595
    },
    {
      "epoch": 9.787735849056604,
      "grad_norm": 4.610492706298828,
      "learning_rate": 6.050527743326455e-06,
      "loss": 0.3644,
      "num_input_tokens_seen": 10830288,
      "step": 16600
    },
    {
      "epoch": 9.79068396226415,
      "grad_norm": 3.2514307498931885,
      "learning_rate": 6.048012316561699e-06,
      "loss": 0.3833,
      "num_input_tokens_seen": 10833392,
      "step": 16605
    },
    {
      "epoch": 9.793632075471699,
      "grad_norm": 2.8535454273223877,
      "learning_rate": 6.045496612331166e-06,
      "loss": 0.377,
      "num_input_tokens_seen": 10836688,
      "step": 16610
    },
    {
      "epoch": 9.796580188679245,
      "grad_norm": 2.023005485534668,
      "learning_rate": 6.0429806313009e-06,
      "loss": 0.5257,
      "num_input_tokens_seen": 10839248,
      "step": 16615
    },
    {
      "epoch": 9.799528301886792,
      "grad_norm": 2.504904270172119,
      "learning_rate": 6.040464374137015e-06,
      "loss": 0.2781,
      "num_input_tokens_seen": 10842416,
      "step": 16620
    },
    {
      "epoch": 9.80247641509434,
      "grad_norm": 2.58198618888855,
      "learning_rate": 6.0379478415057045e-06,
      "loss": 0.3882,
      "num_input_tokens_seen": 10845136,
      "step": 16625
    },
    {
      "epoch": 9.805424528301886,
      "grad_norm": 1.8210455179214478,
      "learning_rate": 6.035431034073228e-06,
      "loss": 0.2995,
      "num_input_tokens_seen": 10848272,
      "step": 16630
    },
    {
      "epoch": 9.808372641509434,
      "grad_norm": 2.355684757232666,
      "learning_rate": 6.0329139525059235e-06,
      "loss": 0.3818,
      "num_input_tokens_seen": 10850992,
      "step": 16635
    },
    {
      "epoch": 9.81132075471698,
      "grad_norm": 1.702268362045288,
      "learning_rate": 6.030396597470198e-06,
      "loss": 0.4443,
      "num_input_tokens_seen": 10854096,
      "step": 16640
    },
    {
      "epoch": 9.814268867924529,
      "grad_norm": 2.5137674808502197,
      "learning_rate": 6.027878969632534e-06,
      "loss": 0.3796,
      "num_input_tokens_seen": 10857936,
      "step": 16645
    },
    {
      "epoch": 9.817216981132075,
      "grad_norm": 2.0676541328430176,
      "learning_rate": 6.025361069659482e-06,
      "loss": 0.4803,
      "num_input_tokens_seen": 10860784,
      "step": 16650
    },
    {
      "epoch": 9.820165094339622,
      "grad_norm": 2.474544048309326,
      "learning_rate": 6.022842898217668e-06,
      "loss": 0.3006,
      "num_input_tokens_seen": 10863888,
      "step": 16655
    },
    {
      "epoch": 9.82311320754717,
      "grad_norm": 1.9515637159347534,
      "learning_rate": 6.020324455973788e-06,
      "loss": 0.4809,
      "num_input_tokens_seen": 10867632,
      "step": 16660
    },
    {
      "epoch": 9.826061320754716,
      "grad_norm": 1.8432711362838745,
      "learning_rate": 6.017805743594612e-06,
      "loss": 0.3865,
      "num_input_tokens_seen": 10870192,
      "step": 16665
    },
    {
      "epoch": 9.829009433962264,
      "grad_norm": 2.3254454135894775,
      "learning_rate": 6.0152867617469776e-06,
      "loss": 0.3491,
      "num_input_tokens_seen": 10873264,
      "step": 16670
    },
    {
      "epoch": 9.83195754716981,
      "grad_norm": 1.9653053283691406,
      "learning_rate": 6.012767511097799e-06,
      "loss": 0.3091,
      "num_input_tokens_seen": 10876656,
      "step": 16675
    },
    {
      "epoch": 9.834905660377359,
      "grad_norm": 5.938626766204834,
      "learning_rate": 6.010247992314055e-06,
      "loss": 0.4253,
      "num_input_tokens_seen": 10879824,
      "step": 16680
    },
    {
      "epoch": 9.837853773584905,
      "grad_norm": 6.53312873840332,
      "learning_rate": 6.007728206062802e-06,
      "loss": 0.3466,
      "num_input_tokens_seen": 10882480,
      "step": 16685
    },
    {
      "epoch": 9.840801886792454,
      "grad_norm": 3.500185251235962,
      "learning_rate": 6.005208153011163e-06,
      "loss": 0.4112,
      "num_input_tokens_seen": 10885648,
      "step": 16690
    },
    {
      "epoch": 9.84375,
      "grad_norm": 2.181725025177002,
      "learning_rate": 6.0026878338263335e-06,
      "loss": 0.3063,
      "num_input_tokens_seen": 10888880,
      "step": 16695
    },
    {
      "epoch": 9.846698113207546,
      "grad_norm": 6.721580505371094,
      "learning_rate": 6.000167249175579e-06,
      "loss": 0.4255,
      "num_input_tokens_seen": 10892304,
      "step": 16700
    },
    {
      "epoch": 9.849646226415095,
      "grad_norm": 4.082174301147461,
      "learning_rate": 5.997646399726236e-06,
      "loss": 0.5917,
      "num_input_tokens_seen": 10894960,
      "step": 16705
    },
    {
      "epoch": 9.852594339622641,
      "grad_norm": 2.4891750812530518,
      "learning_rate": 5.995125286145707e-06,
      "loss": 0.3054,
      "num_input_tokens_seen": 10898352,
      "step": 16710
    },
    {
      "epoch": 9.85554245283019,
      "grad_norm": 2.6735830307006836,
      "learning_rate": 5.99260390910147e-06,
      "loss": 0.4061,
      "num_input_tokens_seen": 10901456,
      "step": 16715
    },
    {
      "epoch": 9.858490566037736,
      "grad_norm": 1.327249526977539,
      "learning_rate": 5.990082269261071e-06,
      "loss": 0.3434,
      "num_input_tokens_seen": 10904560,
      "step": 16720
    },
    {
      "epoch": 9.861438679245284,
      "grad_norm": 2.5125489234924316,
      "learning_rate": 5.987560367292123e-06,
      "loss": 0.4097,
      "num_input_tokens_seen": 10908080,
      "step": 16725
    },
    {
      "epoch": 9.86438679245283,
      "grad_norm": 1.7410204410552979,
      "learning_rate": 5.985038203862313e-06,
      "loss": 0.3271,
      "num_input_tokens_seen": 10911056,
      "step": 16730
    },
    {
      "epoch": 9.867334905660378,
      "grad_norm": 1.4541611671447754,
      "learning_rate": 5.982515779639393e-06,
      "loss": 0.2912,
      "num_input_tokens_seen": 10914384,
      "step": 16735
    },
    {
      "epoch": 9.870283018867925,
      "grad_norm": 2.2535011768341064,
      "learning_rate": 5.979993095291186e-06,
      "loss": 0.4713,
      "num_input_tokens_seen": 10918000,
      "step": 16740
    },
    {
      "epoch": 9.873231132075471,
      "grad_norm": 1.45979905128479,
      "learning_rate": 5.977470151485582e-06,
      "loss": 0.2117,
      "num_input_tokens_seen": 10921104,
      "step": 16745
    },
    {
      "epoch": 9.87617924528302,
      "grad_norm": 2.174992322921753,
      "learning_rate": 5.974946948890544e-06,
      "loss": 0.328,
      "num_input_tokens_seen": 10924848,
      "step": 16750
    },
    {
      "epoch": 9.879127358490566,
      "grad_norm": 1.6254414319992065,
      "learning_rate": 5.9724234881740994e-06,
      "loss": 0.411,
      "num_input_tokens_seen": 10928208,
      "step": 16755
    },
    {
      "epoch": 9.882075471698114,
      "grad_norm": 1.5500922203063965,
      "learning_rate": 5.9698997700043445e-06,
      "loss": 0.334,
      "num_input_tokens_seen": 10931920,
      "step": 16760
    },
    {
      "epoch": 9.88502358490566,
      "grad_norm": 2.435666799545288,
      "learning_rate": 5.9673757950494475e-06,
      "loss": 0.2803,
      "num_input_tokens_seen": 10936016,
      "step": 16765
    },
    {
      "epoch": 9.887971698113208,
      "grad_norm": 3.388235330581665,
      "learning_rate": 5.964851563977639e-06,
      "loss": 0.4116,
      "num_input_tokens_seen": 10938224,
      "step": 16770
    },
    {
      "epoch": 9.890919811320755,
      "grad_norm": 3.7855443954467773,
      "learning_rate": 5.962327077457219e-06,
      "loss": 0.2549,
      "num_input_tokens_seen": 10940816,
      "step": 16775
    },
    {
      "epoch": 9.893867924528301,
      "grad_norm": 3.4862096309661865,
      "learning_rate": 5.959802336156558e-06,
      "loss": 0.4353,
      "num_input_tokens_seen": 10943536,
      "step": 16780
    },
    {
      "epoch": 9.89681603773585,
      "grad_norm": 1.2363417148590088,
      "learning_rate": 5.957277340744094e-06,
      "loss": 0.367,
      "num_input_tokens_seen": 10946480,
      "step": 16785
    },
    {
      "epoch": 9.899764150943396,
      "grad_norm": 3.063689708709717,
      "learning_rate": 5.954752091888326e-06,
      "loss": 0.5593,
      "num_input_tokens_seen": 10949968,
      "step": 16790
    },
    {
      "epoch": 9.902712264150944,
      "grad_norm": 2.1563096046447754,
      "learning_rate": 5.952226590257829e-06,
      "loss": 0.5114,
      "num_input_tokens_seen": 10952656,
      "step": 16795
    },
    {
      "epoch": 9.90566037735849,
      "grad_norm": 3.8924951553344727,
      "learning_rate": 5.949700836521239e-06,
      "loss": 0.3066,
      "num_input_tokens_seen": 10954960,
      "step": 16800
    },
    {
      "epoch": 9.908608490566039,
      "grad_norm": 2.554572820663452,
      "learning_rate": 5.947174831347257e-06,
      "loss": 0.3967,
      "num_input_tokens_seen": 10958256,
      "step": 16805
    },
    {
      "epoch": 9.911556603773585,
      "grad_norm": 1.4245203733444214,
      "learning_rate": 5.944648575404657e-06,
      "loss": 0.4072,
      "num_input_tokens_seen": 10963056,
      "step": 16810
    },
    {
      "epoch": 9.914504716981131,
      "grad_norm": 3.73224139213562,
      "learning_rate": 5.942122069362276e-06,
      "loss": 0.4447,
      "num_input_tokens_seen": 10966384,
      "step": 16815
    },
    {
      "epoch": 9.91745283018868,
      "grad_norm": 3.2709403038024902,
      "learning_rate": 5.939595313889016e-06,
      "loss": 0.4194,
      "num_input_tokens_seen": 10969680,
      "step": 16820
    },
    {
      "epoch": 9.920400943396226,
      "grad_norm": 2.644888162612915,
      "learning_rate": 5.937068309653848e-06,
      "loss": 0.3625,
      "num_input_tokens_seen": 10973360,
      "step": 16825
    },
    {
      "epoch": 9.923349056603774,
      "grad_norm": 4.134654998779297,
      "learning_rate": 5.934541057325807e-06,
      "loss": 0.3913,
      "num_input_tokens_seen": 10976976,
      "step": 16830
    },
    {
      "epoch": 9.92629716981132,
      "grad_norm": 1.81215238571167,
      "learning_rate": 5.932013557573992e-06,
      "loss": 0.4516,
      "num_input_tokens_seen": 10980560,
      "step": 16835
    },
    {
      "epoch": 9.929245283018869,
      "grad_norm": 4.7528300285339355,
      "learning_rate": 5.929485811067572e-06,
      "loss": 0.3088,
      "num_input_tokens_seen": 10984080,
      "step": 16840
    },
    {
      "epoch": 9.932193396226415,
      "grad_norm": 2.3289639949798584,
      "learning_rate": 5.926957818475778e-06,
      "loss": 0.346,
      "num_input_tokens_seen": 10987056,
      "step": 16845
    },
    {
      "epoch": 9.935141509433961,
      "grad_norm": 3.151592969894409,
      "learning_rate": 5.924429580467905e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 10990480,
      "step": 16850
    },
    {
      "epoch": 9.93808962264151,
      "grad_norm": 2.5649797916412354,
      "learning_rate": 5.921901097713317e-06,
      "loss": 0.3044,
      "num_input_tokens_seen": 10993520,
      "step": 16855
    },
    {
      "epoch": 9.941037735849056,
      "grad_norm": 3.349287271499634,
      "learning_rate": 5.919372370881442e-06,
      "loss": 0.2404,
      "num_input_tokens_seen": 10996208,
      "step": 16860
    },
    {
      "epoch": 9.943985849056604,
      "grad_norm": 2.22196364402771,
      "learning_rate": 5.916843400641768e-06,
      "loss": 0.413,
      "num_input_tokens_seen": 11000016,
      "step": 16865
    },
    {
      "epoch": 9.94693396226415,
      "grad_norm": 2.7622289657592773,
      "learning_rate": 5.914314187663851e-06,
      "loss": 0.3892,
      "num_input_tokens_seen": 11003024,
      "step": 16870
    },
    {
      "epoch": 9.949882075471699,
      "grad_norm": 3.572880506515503,
      "learning_rate": 5.911784732617314e-06,
      "loss": 0.3314,
      "num_input_tokens_seen": 11005168,
      "step": 16875
    },
    {
      "epoch": 9.952830188679245,
      "grad_norm": 2.1173453330993652,
      "learning_rate": 5.90925503617184e-06,
      "loss": 0.3725,
      "num_input_tokens_seen": 11008336,
      "step": 16880
    },
    {
      "epoch": 9.955778301886792,
      "grad_norm": 4.31459903717041,
      "learning_rate": 5.9067250989971745e-06,
      "loss": 0.4573,
      "num_input_tokens_seen": 11011408,
      "step": 16885
    },
    {
      "epoch": 9.95872641509434,
      "grad_norm": 2.175584554672241,
      "learning_rate": 5.904194921763133e-06,
      "loss": 0.2939,
      "num_input_tokens_seen": 11014928,
      "step": 16890
    },
    {
      "epoch": 9.961674528301886,
      "grad_norm": 2.476794958114624,
      "learning_rate": 5.901664505139589e-06,
      "loss": 0.3057,
      "num_input_tokens_seen": 11018096,
      "step": 16895
    },
    {
      "epoch": 9.964622641509434,
      "grad_norm": 6.294628143310547,
      "learning_rate": 5.8991338497964814e-06,
      "loss": 0.5172,
      "num_input_tokens_seen": 11021200,
      "step": 16900
    },
    {
      "epoch": 9.96757075471698,
      "grad_norm": 2.1673519611358643,
      "learning_rate": 5.896602956403812e-06,
      "loss": 0.3481,
      "num_input_tokens_seen": 11027280,
      "step": 16905
    },
    {
      "epoch": 9.970518867924529,
      "grad_norm": 4.833362102508545,
      "learning_rate": 5.894071825631645e-06,
      "loss": 0.3502,
      "num_input_tokens_seen": 11030608,
      "step": 16910
    },
    {
      "epoch": 9.973466981132075,
      "grad_norm": 2.756204128265381,
      "learning_rate": 5.891540458150109e-06,
      "loss": 0.4036,
      "num_input_tokens_seen": 11035056,
      "step": 16915
    },
    {
      "epoch": 9.976415094339622,
      "grad_norm": 2.4853250980377197,
      "learning_rate": 5.889008854629395e-06,
      "loss": 0.5076,
      "num_input_tokens_seen": 11038352,
      "step": 16920
    },
    {
      "epoch": 9.97936320754717,
      "grad_norm": 3.54835844039917,
      "learning_rate": 5.886477015739754e-06,
      "loss": 0.5762,
      "num_input_tokens_seen": 11040912,
      "step": 16925
    },
    {
      "epoch": 9.982311320754716,
      "grad_norm": 2.323976516723633,
      "learning_rate": 5.883944942151502e-06,
      "loss": 0.3954,
      "num_input_tokens_seen": 11044528,
      "step": 16930
    },
    {
      "epoch": 9.985259433962264,
      "grad_norm": 3.151970386505127,
      "learning_rate": 5.88141263453502e-06,
      "loss": 0.3634,
      "num_input_tokens_seen": 11047152,
      "step": 16935
    },
    {
      "epoch": 9.98820754716981,
      "grad_norm": 5.0896124839782715,
      "learning_rate": 5.878880093560744e-06,
      "loss": 0.2998,
      "num_input_tokens_seen": 11050544,
      "step": 16940
    },
    {
      "epoch": 9.991155660377359,
      "grad_norm": 3.026899576187134,
      "learning_rate": 5.876347319899173e-06,
      "loss": 0.4117,
      "num_input_tokens_seen": 11053040,
      "step": 16945
    },
    {
      "epoch": 9.994103773584905,
      "grad_norm": 2.2079579830169678,
      "learning_rate": 5.873814314220874e-06,
      "loss": 0.5502,
      "num_input_tokens_seen": 11056656,
      "step": 16950
    },
    {
      "epoch": 9.997051886792454,
      "grad_norm": 1.888657808303833,
      "learning_rate": 5.871281077196469e-06,
      "loss": 0.3848,
      "num_input_tokens_seen": 11059440,
      "step": 16955
    },
    {
      "epoch": 10.0,
      "grad_norm": 7.1668782234191895,
      "learning_rate": 5.868747609496643e-06,
      "loss": 0.4427,
      "num_input_tokens_seen": 11063280,
      "step": 16960
    },
    {
      "epoch": 10.0,
      "eval_loss": 0.52943354845047,
      "eval_runtime": 19.2396,
      "eval_samples_per_second": 88.152,
      "eval_steps_per_second": 22.038,
      "num_input_tokens_seen": 11063280,
      "step": 16960
    },
    {
      "epoch": 10.002948113207546,
      "grad_norm": 2.236830234527588,
      "learning_rate": 5.8662139117921435e-06,
      "loss": 0.3382,
      "num_input_tokens_seen": 11066480,
      "step": 16965
    },
    {
      "epoch": 10.005896226415095,
      "grad_norm": 2.191300630569458,
      "learning_rate": 5.8636799847537785e-06,
      "loss": 0.3696,
      "num_input_tokens_seen": 11070448,
      "step": 16970
    },
    {
      "epoch": 10.008844339622641,
      "grad_norm": 2.762096643447876,
      "learning_rate": 5.861145829052415e-06,
      "loss": 0.3161,
      "num_input_tokens_seen": 11073232,
      "step": 16975
    },
    {
      "epoch": 10.01179245283019,
      "grad_norm": 2.7102532386779785,
      "learning_rate": 5.858611445358982e-06,
      "loss": 0.201,
      "num_input_tokens_seen": 11076016,
      "step": 16980
    },
    {
      "epoch": 10.014740566037736,
      "grad_norm": 2.283935546875,
      "learning_rate": 5.856076834344468e-06,
      "loss": 0.3956,
      "num_input_tokens_seen": 11079760,
      "step": 16985
    },
    {
      "epoch": 10.017688679245284,
      "grad_norm": 3.5570545196533203,
      "learning_rate": 5.853541996679924e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 11082608,
      "step": 16990
    },
    {
      "epoch": 10.02063679245283,
      "grad_norm": 1.843102216720581,
      "learning_rate": 5.851006933036456e-06,
      "loss": 0.3841,
      "num_input_tokens_seen": 11085616,
      "step": 16995
    },
    {
      "epoch": 10.023584905660377,
      "grad_norm": 1.6436667442321777,
      "learning_rate": 5.848471644085236e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 11090096,
      "step": 17000
    },
    {
      "epoch": 10.026533018867925,
      "grad_norm": 5.827091217041016,
      "learning_rate": 5.84593613049749e-06,
      "loss": 0.3709,
      "num_input_tokens_seen": 11092848,
      "step": 17005
    },
    {
      "epoch": 10.029481132075471,
      "grad_norm": 4.524348735809326,
      "learning_rate": 5.843400392944509e-06,
      "loss": 0.4307,
      "num_input_tokens_seen": 11095792,
      "step": 17010
    },
    {
      "epoch": 10.03242924528302,
      "grad_norm": 1.5599381923675537,
      "learning_rate": 5.840864432097639e-06,
      "loss": 0.4455,
      "num_input_tokens_seen": 11100528,
      "step": 17015
    },
    {
      "epoch": 10.035377358490566,
      "grad_norm": 3.4543261528015137,
      "learning_rate": 5.838328248628286e-06,
      "loss": 0.3511,
      "num_input_tokens_seen": 11103824,
      "step": 17020
    },
    {
      "epoch": 10.038325471698114,
      "grad_norm": 2.0398054122924805,
      "learning_rate": 5.835791843207916e-06,
      "loss": 0.4317,
      "num_input_tokens_seen": 11107184,
      "step": 17025
    },
    {
      "epoch": 10.04127358490566,
      "grad_norm": 3.838715076446533,
      "learning_rate": 5.833255216508056e-06,
      "loss": 0.4389,
      "num_input_tokens_seen": 11114512,
      "step": 17030
    },
    {
      "epoch": 10.044221698113208,
      "grad_norm": 2.4425644874572754,
      "learning_rate": 5.830718369200284e-06,
      "loss": 0.4194,
      "num_input_tokens_seen": 11116912,
      "step": 17035
    },
    {
      "epoch": 10.047169811320755,
      "grad_norm": 2.2100231647491455,
      "learning_rate": 5.828181301956244e-06,
      "loss": 0.3677,
      "num_input_tokens_seen": 11120176,
      "step": 17040
    },
    {
      "epoch": 10.050117924528301,
      "grad_norm": 3.5590505599975586,
      "learning_rate": 5.825644015447636e-06,
      "loss": 0.3028,
      "num_input_tokens_seen": 11122896,
      "step": 17045
    },
    {
      "epoch": 10.05306603773585,
      "grad_norm": 3.900620222091675,
      "learning_rate": 5.823106510346216e-06,
      "loss": 0.436,
      "num_input_tokens_seen": 11126256,
      "step": 17050
    },
    {
      "epoch": 10.056014150943396,
      "grad_norm": 2.0913264751434326,
      "learning_rate": 5.820568787323798e-06,
      "loss": 0.3686,
      "num_input_tokens_seen": 11129040,
      "step": 17055
    },
    {
      "epoch": 10.058962264150944,
      "grad_norm": 5.313127040863037,
      "learning_rate": 5.818030847052258e-06,
      "loss": 0.3068,
      "num_input_tokens_seen": 11132016,
      "step": 17060
    },
    {
      "epoch": 10.06191037735849,
      "grad_norm": 2.636955499649048,
      "learning_rate": 5.815492690203528e-06,
      "loss": 0.3379,
      "num_input_tokens_seen": 11134864,
      "step": 17065
    },
    {
      "epoch": 10.064858490566039,
      "grad_norm": 2.718418836593628,
      "learning_rate": 5.812954317449591e-06,
      "loss": 0.609,
      "num_input_tokens_seen": 11138032,
      "step": 17070
    },
    {
      "epoch": 10.067806603773585,
      "grad_norm": 2.9297893047332764,
      "learning_rate": 5.810415729462496e-06,
      "loss": 0.3786,
      "num_input_tokens_seen": 11140944,
      "step": 17075
    },
    {
      "epoch": 10.070754716981131,
      "grad_norm": 3.6357223987579346,
      "learning_rate": 5.807876926914344e-06,
      "loss": 0.3007,
      "num_input_tokens_seen": 11144496,
      "step": 17080
    },
    {
      "epoch": 10.07370283018868,
      "grad_norm": 3.955145835876465,
      "learning_rate": 5.8053379104772935e-06,
      "loss": 0.3227,
      "num_input_tokens_seen": 11147440,
      "step": 17085
    },
    {
      "epoch": 10.076650943396226,
      "grad_norm": 4.106929302215576,
      "learning_rate": 5.802798680823562e-06,
      "loss": 0.416,
      "num_input_tokens_seen": 11150032,
      "step": 17090
    },
    {
      "epoch": 10.079599056603774,
      "grad_norm": 2.101675033569336,
      "learning_rate": 5.8002592386254186e-06,
      "loss": 0.3803,
      "num_input_tokens_seen": 11152912,
      "step": 17095
    },
    {
      "epoch": 10.08254716981132,
      "grad_norm": 2.315084457397461,
      "learning_rate": 5.797719584555192e-06,
      "loss": 0.3678,
      "num_input_tokens_seen": 11155920,
      "step": 17100
    },
    {
      "epoch": 10.085495283018869,
      "grad_norm": 5.475011348724365,
      "learning_rate": 5.795179719285269e-06,
      "loss": 0.4083,
      "num_input_tokens_seen": 11158896,
      "step": 17105
    },
    {
      "epoch": 10.088443396226415,
      "grad_norm": 2.77616548538208,
      "learning_rate": 5.792639643488086e-06,
      "loss": 0.3617,
      "num_input_tokens_seen": 11161424,
      "step": 17110
    },
    {
      "epoch": 10.091391509433961,
      "grad_norm": 4.017038822174072,
      "learning_rate": 5.7900993578361434e-06,
      "loss": 0.3653,
      "num_input_tokens_seen": 11164304,
      "step": 17115
    },
    {
      "epoch": 10.09433962264151,
      "grad_norm": 1.4543569087982178,
      "learning_rate": 5.7875588630019895e-06,
      "loss": 0.3009,
      "num_input_tokens_seen": 11168048,
      "step": 17120
    },
    {
      "epoch": 10.097287735849056,
      "grad_norm": 3.10024094581604,
      "learning_rate": 5.7850181596582335e-06,
      "loss": 0.2757,
      "num_input_tokens_seen": 11170288,
      "step": 17125
    },
    {
      "epoch": 10.100235849056604,
      "grad_norm": 2.2524056434631348,
      "learning_rate": 5.782477248477535e-06,
      "loss": 0.441,
      "num_input_tokens_seen": 11173296,
      "step": 17130
    },
    {
      "epoch": 10.10318396226415,
      "grad_norm": 1.926977515220642,
      "learning_rate": 5.779936130132614e-06,
      "loss": 0.3755,
      "num_input_tokens_seen": 11177072,
      "step": 17135
    },
    {
      "epoch": 10.106132075471699,
      "grad_norm": 3.904543161392212,
      "learning_rate": 5.777394805296242e-06,
      "loss": 0.3886,
      "num_input_tokens_seen": 11180720,
      "step": 17140
    },
    {
      "epoch": 10.109080188679245,
      "grad_norm": 4.499124050140381,
      "learning_rate": 5.774853274641243e-06,
      "loss": 0.3377,
      "num_input_tokens_seen": 11184144,
      "step": 17145
    },
    {
      "epoch": 10.112028301886792,
      "grad_norm": 2.6898717880249023,
      "learning_rate": 5.772311538840501e-06,
      "loss": 0.3566,
      "num_input_tokens_seen": 11186928,
      "step": 17150
    },
    {
      "epoch": 10.11497641509434,
      "grad_norm": 2.7684848308563232,
      "learning_rate": 5.76976959856695e-06,
      "loss": 0.375,
      "num_input_tokens_seen": 11189808,
      "step": 17155
    },
    {
      "epoch": 10.117924528301886,
      "grad_norm": 4.081057071685791,
      "learning_rate": 5.767227454493582e-06,
      "loss": 0.3196,
      "num_input_tokens_seen": 11192816,
      "step": 17160
    },
    {
      "epoch": 10.120872641509434,
      "grad_norm": 2.6376779079437256,
      "learning_rate": 5.764685107293436e-06,
      "loss": 0.371,
      "num_input_tokens_seen": 11196240,
      "step": 17165
    },
    {
      "epoch": 10.12382075471698,
      "grad_norm": 1.6304056644439697,
      "learning_rate": 5.762142557639614e-06,
      "loss": 0.2972,
      "num_input_tokens_seen": 11199888,
      "step": 17170
    },
    {
      "epoch": 10.126768867924529,
      "grad_norm": 2.4335174560546875,
      "learning_rate": 5.759599806205266e-06,
      "loss": 0.3689,
      "num_input_tokens_seen": 11202864,
      "step": 17175
    },
    {
      "epoch": 10.129716981132075,
      "grad_norm": 1.5857563018798828,
      "learning_rate": 5.757056853663594e-06,
      "loss": 0.3085,
      "num_input_tokens_seen": 11205840,
      "step": 17180
    },
    {
      "epoch": 10.132665094339623,
      "grad_norm": 3.1204757690429688,
      "learning_rate": 5.754513700687858e-06,
      "loss": 0.3276,
      "num_input_tokens_seen": 11208112,
      "step": 17185
    },
    {
      "epoch": 10.13561320754717,
      "grad_norm": 3.415346145629883,
      "learning_rate": 5.751970347951369e-06,
      "loss": 0.2732,
      "num_input_tokens_seen": 11210608,
      "step": 17190
    },
    {
      "epoch": 10.138561320754716,
      "grad_norm": 2.639396905899048,
      "learning_rate": 5.7494267961274875e-06,
      "loss": 0.4057,
      "num_input_tokens_seen": 11213872,
      "step": 17195
    },
    {
      "epoch": 10.141509433962264,
      "grad_norm": 4.661757946014404,
      "learning_rate": 5.746883045889633e-06,
      "loss": 0.2573,
      "num_input_tokens_seen": 11216848,
      "step": 17200
    },
    {
      "epoch": 10.14445754716981,
      "grad_norm": 2.5506951808929443,
      "learning_rate": 5.744339097911272e-06,
      "loss": 0.3192,
      "num_input_tokens_seen": 11219792,
      "step": 17205
    },
    {
      "epoch": 10.147405660377359,
      "grad_norm": 3.494114637374878,
      "learning_rate": 5.741794952865928e-06,
      "loss": 0.3808,
      "num_input_tokens_seen": 11222544,
      "step": 17210
    },
    {
      "epoch": 10.150353773584905,
      "grad_norm": 2.6537415981292725,
      "learning_rate": 5.739250611427173e-06,
      "loss": 0.4416,
      "num_input_tokens_seen": 11225808,
      "step": 17215
    },
    {
      "epoch": 10.153301886792454,
      "grad_norm": 4.799945831298828,
      "learning_rate": 5.736706074268633e-06,
      "loss": 0.5447,
      "num_input_tokens_seen": 11229776,
      "step": 17220
    },
    {
      "epoch": 10.15625,
      "grad_norm": 3.400574207305908,
      "learning_rate": 5.734161342063984e-06,
      "loss": 0.3847,
      "num_input_tokens_seen": 11232976,
      "step": 17225
    },
    {
      "epoch": 10.159198113207546,
      "grad_norm": 1.9110426902770996,
      "learning_rate": 5.731616415486957e-06,
      "loss": 0.3187,
      "num_input_tokens_seen": 11235984,
      "step": 17230
    },
    {
      "epoch": 10.162146226415095,
      "grad_norm": 2.2381649017333984,
      "learning_rate": 5.72907129521133e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 11239472,
      "step": 17235
    },
    {
      "epoch": 10.165094339622641,
      "grad_norm": 3.986616373062134,
      "learning_rate": 5.726525981910935e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 11243312,
      "step": 17240
    },
    {
      "epoch": 10.16804245283019,
      "grad_norm": 1.8503303527832031,
      "learning_rate": 5.723980476259658e-06,
      "loss": 0.3964,
      "num_input_tokens_seen": 11246736,
      "step": 17245
    },
    {
      "epoch": 10.170990566037736,
      "grad_norm": 2.1283187866210938,
      "learning_rate": 5.721434778931426e-06,
      "loss": 0.4034,
      "num_input_tokens_seen": 11249488,
      "step": 17250
    },
    {
      "epoch": 10.173938679245284,
      "grad_norm": 2.6528496742248535,
      "learning_rate": 5.7188888906002284e-06,
      "loss": 0.3016,
      "num_input_tokens_seen": 11253328,
      "step": 17255
    },
    {
      "epoch": 10.17688679245283,
      "grad_norm": 2.4719715118408203,
      "learning_rate": 5.716342811940098e-06,
      "loss": 0.3584,
      "num_input_tokens_seen": 11256944,
      "step": 17260
    },
    {
      "epoch": 10.179834905660377,
      "grad_norm": 1.5065561532974243,
      "learning_rate": 5.713796543625123e-06,
      "loss": 0.3687,
      "num_input_tokens_seen": 11260240,
      "step": 17265
    },
    {
      "epoch": 10.182783018867925,
      "grad_norm": 2.675222635269165,
      "learning_rate": 5.711250086329435e-06,
      "loss": 0.2955,
      "num_input_tokens_seen": 11262864,
      "step": 17270
    },
    {
      "epoch": 10.185731132075471,
      "grad_norm": 3.1739578247070312,
      "learning_rate": 5.7087034407272225e-06,
      "loss": 0.3456,
      "num_input_tokens_seen": 11265744,
      "step": 17275
    },
    {
      "epoch": 10.18867924528302,
      "grad_norm": 4.033857822418213,
      "learning_rate": 5.70615660749272e-06,
      "loss": 0.3833,
      "num_input_tokens_seen": 11268528,
      "step": 17280
    },
    {
      "epoch": 10.191627358490566,
      "grad_norm": 6.158642768859863,
      "learning_rate": 5.7036095873002106e-06,
      "loss": 0.5305,
      "num_input_tokens_seen": 11271824,
      "step": 17285
    },
    {
      "epoch": 10.194575471698114,
      "grad_norm": 2.364011526107788,
      "learning_rate": 5.701062380824032e-06,
      "loss": 0.2883,
      "num_input_tokens_seen": 11275216,
      "step": 17290
    },
    {
      "epoch": 10.19752358490566,
      "grad_norm": 2.7802627086639404,
      "learning_rate": 5.698514988738566e-06,
      "loss": 0.3354,
      "num_input_tokens_seen": 11278064,
      "step": 17295
    },
    {
      "epoch": 10.200471698113208,
      "grad_norm": 2.436126947402954,
      "learning_rate": 5.6959674117182465e-06,
      "loss": 0.421,
      "num_input_tokens_seen": 11282736,
      "step": 17300
    },
    {
      "epoch": 10.203419811320755,
      "grad_norm": 4.761617660522461,
      "learning_rate": 5.693419650437554e-06,
      "loss": 0.402,
      "num_input_tokens_seen": 11285200,
      "step": 17305
    },
    {
      "epoch": 10.206367924528301,
      "grad_norm": 3.340930700302124,
      "learning_rate": 5.690871705571022e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 11287376,
      "step": 17310
    },
    {
      "epoch": 10.20931603773585,
      "grad_norm": 3.365468978881836,
      "learning_rate": 5.688323577793229e-06,
      "loss": 0.4477,
      "num_input_tokens_seen": 11290288,
      "step": 17315
    },
    {
      "epoch": 10.212264150943396,
      "grad_norm": 2.726369619369507,
      "learning_rate": 5.685775267778801e-06,
      "loss": 0.4168,
      "num_input_tokens_seen": 11292976,
      "step": 17320
    },
    {
      "epoch": 10.215212264150944,
      "grad_norm": 5.126517295837402,
      "learning_rate": 5.683226776202416e-06,
      "loss": 0.4327,
      "num_input_tokens_seen": 11296656,
      "step": 17325
    },
    {
      "epoch": 10.21816037735849,
      "grad_norm": 3.199646234512329,
      "learning_rate": 5.680678103738798e-06,
      "loss": 0.48,
      "num_input_tokens_seen": 11299760,
      "step": 17330
    },
    {
      "epoch": 10.221108490566039,
      "grad_norm": 4.901650905609131,
      "learning_rate": 5.678129251062717e-06,
      "loss": 0.3417,
      "num_input_tokens_seen": 11302736,
      "step": 17335
    },
    {
      "epoch": 10.224056603773585,
      "grad_norm": 1.9152823686599731,
      "learning_rate": 5.675580218848995e-06,
      "loss": 0.3446,
      "num_input_tokens_seen": 11305968,
      "step": 17340
    },
    {
      "epoch": 10.227004716981131,
      "grad_norm": 4.549933910369873,
      "learning_rate": 5.673031007772498e-06,
      "loss": 0.3756,
      "num_input_tokens_seen": 11309040,
      "step": 17345
    },
    {
      "epoch": 10.22995283018868,
      "grad_norm": 3.1808485984802246,
      "learning_rate": 5.670481618508141e-06,
      "loss": 0.3604,
      "num_input_tokens_seen": 11311760,
      "step": 17350
    },
    {
      "epoch": 10.232900943396226,
      "grad_norm": 5.211045265197754,
      "learning_rate": 5.667932051730887e-06,
      "loss": 0.341,
      "num_input_tokens_seen": 11314704,
      "step": 17355
    },
    {
      "epoch": 10.235849056603774,
      "grad_norm": 1.9566785097122192,
      "learning_rate": 5.6653823081157434e-06,
      "loss": 0.3543,
      "num_input_tokens_seen": 11317808,
      "step": 17360
    },
    {
      "epoch": 10.23879716981132,
      "grad_norm": 3.3826732635498047,
      "learning_rate": 5.662832388337766e-06,
      "loss": 0.397,
      "num_input_tokens_seen": 11320816,
      "step": 17365
    },
    {
      "epoch": 10.241745283018869,
      "grad_norm": 2.209038257598877,
      "learning_rate": 5.660282293072057e-06,
      "loss": 0.3498,
      "num_input_tokens_seen": 11325232,
      "step": 17370
    },
    {
      "epoch": 10.244693396226415,
      "grad_norm": 2.366002082824707,
      "learning_rate": 5.657732022993765e-06,
      "loss": 0.3768,
      "num_input_tokens_seen": 11328368,
      "step": 17375
    },
    {
      "epoch": 10.247641509433961,
      "grad_norm": 5.412224292755127,
      "learning_rate": 5.655181578778085e-06,
      "loss": 0.4547,
      "num_input_tokens_seen": 11331312,
      "step": 17380
    },
    {
      "epoch": 10.25058962264151,
      "grad_norm": 2.7893948554992676,
      "learning_rate": 5.65263096110026e-06,
      "loss": 0.2458,
      "num_input_tokens_seen": 11334032,
      "step": 17385
    },
    {
      "epoch": 10.253537735849056,
      "grad_norm": 2.4446887969970703,
      "learning_rate": 5.650080170635573e-06,
      "loss": 0.3472,
      "num_input_tokens_seen": 11337328,
      "step": 17390
    },
    {
      "epoch": 10.256485849056604,
      "grad_norm": 1.897607445716858,
      "learning_rate": 5.647529208059359e-06,
      "loss": 0.3258,
      "num_input_tokens_seen": 11340240,
      "step": 17395
    },
    {
      "epoch": 10.25943396226415,
      "grad_norm": 1.8987586498260498,
      "learning_rate": 5.6449780740469985e-06,
      "loss": 0.4787,
      "num_input_tokens_seen": 11343536,
      "step": 17400
    },
    {
      "epoch": 10.262382075471699,
      "grad_norm": 1.4581321477890015,
      "learning_rate": 5.642426769273912e-06,
      "loss": 0.3464,
      "num_input_tokens_seen": 11346544,
      "step": 17405
    },
    {
      "epoch": 10.265330188679245,
      "grad_norm": 2.3183844089508057,
      "learning_rate": 5.63987529441557e-06,
      "loss": 0.2924,
      "num_input_tokens_seen": 11350096,
      "step": 17410
    },
    {
      "epoch": 10.268278301886792,
      "grad_norm": 1.7459776401519775,
      "learning_rate": 5.637323650147487e-06,
      "loss": 0.4884,
      "num_input_tokens_seen": 11352688,
      "step": 17415
    },
    {
      "epoch": 10.27122641509434,
      "grad_norm": 2.4574997425079346,
      "learning_rate": 5.63477183714522e-06,
      "loss": 0.281,
      "num_input_tokens_seen": 11355824,
      "step": 17420
    },
    {
      "epoch": 10.274174528301886,
      "grad_norm": 3.4797680377960205,
      "learning_rate": 5.632219856084373e-06,
      "loss": 0.4866,
      "num_input_tokens_seen": 11359152,
      "step": 17425
    },
    {
      "epoch": 10.277122641509434,
      "grad_norm": 1.973578929901123,
      "learning_rate": 5.6296677076405944e-06,
      "loss": 0.4205,
      "num_input_tokens_seen": 11362320,
      "step": 17430
    },
    {
      "epoch": 10.28007075471698,
      "grad_norm": 2.9002130031585693,
      "learning_rate": 5.627115392489578e-06,
      "loss": 0.3303,
      "num_input_tokens_seen": 11366608,
      "step": 17435
    },
    {
      "epoch": 10.283018867924529,
      "grad_norm": 4.200732231140137,
      "learning_rate": 5.624562911307058e-06,
      "loss": 0.3174,
      "num_input_tokens_seen": 11369616,
      "step": 17440
    },
    {
      "epoch": 10.285966981132075,
      "grad_norm": 2.2556450366973877,
      "learning_rate": 5.622010264768813e-06,
      "loss": 0.4381,
      "num_input_tokens_seen": 11372624,
      "step": 17445
    },
    {
      "epoch": 10.288915094339623,
      "grad_norm": 2.140825033187866,
      "learning_rate": 5.619457453550673e-06,
      "loss": 0.2626,
      "num_input_tokens_seen": 11376144,
      "step": 17450
    },
    {
      "epoch": 10.29186320754717,
      "grad_norm": 2.6157894134521484,
      "learning_rate": 5.616904478328502e-06,
      "loss": 0.4153,
      "num_input_tokens_seen": 11379216,
      "step": 17455
    },
    {
      "epoch": 10.294811320754716,
      "grad_norm": 2.5347180366516113,
      "learning_rate": 5.6143513397782104e-06,
      "loss": 0.4119,
      "num_input_tokens_seen": 11382032,
      "step": 17460
    },
    {
      "epoch": 10.297759433962264,
      "grad_norm": 5.237630844116211,
      "learning_rate": 5.611798038575755e-06,
      "loss": 0.4468,
      "num_input_tokens_seen": 11384752,
      "step": 17465
    },
    {
      "epoch": 10.30070754716981,
      "grad_norm": 2.3017730712890625,
      "learning_rate": 5.609244575397131e-06,
      "loss": 0.2419,
      "num_input_tokens_seen": 11388048,
      "step": 17470
    },
    {
      "epoch": 10.303655660377359,
      "grad_norm": 2.804633617401123,
      "learning_rate": 5.606690950918381e-06,
      "loss": 0.3855,
      "num_input_tokens_seen": 11391152,
      "step": 17475
    },
    {
      "epoch": 10.306603773584905,
      "grad_norm": 2.553508996963501,
      "learning_rate": 5.604137165815586e-06,
      "loss": 0.3195,
      "num_input_tokens_seen": 11394864,
      "step": 17480
    },
    {
      "epoch": 10.309551886792454,
      "grad_norm": 4.214234828948975,
      "learning_rate": 5.601583220764874e-06,
      "loss": 0.3805,
      "num_input_tokens_seen": 11399984,
      "step": 17485
    },
    {
      "epoch": 10.3125,
      "grad_norm": 2.570779323577881,
      "learning_rate": 5.599029116442409e-06,
      "loss": 0.3897,
      "num_input_tokens_seen": 11402864,
      "step": 17490
    },
    {
      "epoch": 10.315448113207546,
      "grad_norm": 3.488511323928833,
      "learning_rate": 5.596474853524406e-06,
      "loss": 0.3776,
      "num_input_tokens_seen": 11405232,
      "step": 17495
    },
    {
      "epoch": 10.318396226415095,
      "grad_norm": 4.43650484085083,
      "learning_rate": 5.593920432687115e-06,
      "loss": 0.5069,
      "num_input_tokens_seen": 11408048,
      "step": 17500
    },
    {
      "epoch": 10.321344339622641,
      "grad_norm": 3.2700893878936768,
      "learning_rate": 5.591365854606829e-06,
      "loss": 0.3743,
      "num_input_tokens_seen": 11412016,
      "step": 17505
    },
    {
      "epoch": 10.32429245283019,
      "grad_norm": 2.827975034713745,
      "learning_rate": 5.588811119959885e-06,
      "loss": 0.3291,
      "num_input_tokens_seen": 11414960,
      "step": 17510
    },
    {
      "epoch": 10.327240566037736,
      "grad_norm": 2.1097817420959473,
      "learning_rate": 5.58625622942266e-06,
      "loss": 0.684,
      "num_input_tokens_seen": 11417840,
      "step": 17515
    },
    {
      "epoch": 10.330188679245284,
      "grad_norm": 2.6729228496551514,
      "learning_rate": 5.58370118367157e-06,
      "loss": 0.2969,
      "num_input_tokens_seen": 11421360,
      "step": 17520
    },
    {
      "epoch": 10.33313679245283,
      "grad_norm": 2.0294246673583984,
      "learning_rate": 5.581145983383077e-06,
      "loss": 0.4068,
      "num_input_tokens_seen": 11424368,
      "step": 17525
    },
    {
      "epoch": 10.336084905660377,
      "grad_norm": 2.1000876426696777,
      "learning_rate": 5.57859062923368e-06,
      "loss": 0.2624,
      "num_input_tokens_seen": 11427056,
      "step": 17530
    },
    {
      "epoch": 10.339033018867925,
      "grad_norm": 1.9153977632522583,
      "learning_rate": 5.5760351218999194e-06,
      "loss": 0.3341,
      "num_input_tokens_seen": 11431152,
      "step": 17535
    },
    {
      "epoch": 10.341981132075471,
      "grad_norm": 3.605034828186035,
      "learning_rate": 5.573479462058379e-06,
      "loss": 0.4114,
      "num_input_tokens_seen": 11433680,
      "step": 17540
    },
    {
      "epoch": 10.34492924528302,
      "grad_norm": 2.6693365573883057,
      "learning_rate": 5.570923650385679e-06,
      "loss": 0.3669,
      "num_input_tokens_seen": 11436208,
      "step": 17545
    },
    {
      "epoch": 10.347877358490566,
      "grad_norm": 3.5500359535217285,
      "learning_rate": 5.568367687558481e-06,
      "loss": 0.4004,
      "num_input_tokens_seen": 11438384,
      "step": 17550
    },
    {
      "epoch": 10.350825471698114,
      "grad_norm": 5.69774055480957,
      "learning_rate": 5.56581157425349e-06,
      "loss": 0.4195,
      "num_input_tokens_seen": 11440976,
      "step": 17555
    },
    {
      "epoch": 10.35377358490566,
      "grad_norm": 2.398529291152954,
      "learning_rate": 5.563255311147446e-06,
      "loss": 0.3413,
      "num_input_tokens_seen": 11444816,
      "step": 17560
    },
    {
      "epoch": 10.356721698113208,
      "grad_norm": 4.133174419403076,
      "learning_rate": 5.560698898917129e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 11447376,
      "step": 17565
    },
    {
      "epoch": 10.359669811320755,
      "grad_norm": 2.2507314682006836,
      "learning_rate": 5.558142338239365e-06,
      "loss": 0.3708,
      "num_input_tokens_seen": 11450896,
      "step": 17570
    },
    {
      "epoch": 10.362617924528301,
      "grad_norm": 1.8855146169662476,
      "learning_rate": 5.55558562979101e-06,
      "loss": 0.4119,
      "num_input_tokens_seen": 11453584,
      "step": 17575
    },
    {
      "epoch": 10.36556603773585,
      "grad_norm": 3.634397506713867,
      "learning_rate": 5.553028774248964e-06,
      "loss": 0.4158,
      "num_input_tokens_seen": 11456272,
      "step": 17580
    },
    {
      "epoch": 10.368514150943396,
      "grad_norm": 3.053724765777588,
      "learning_rate": 5.5504717722901665e-06,
      "loss": 0.3593,
      "num_input_tokens_seen": 11459216,
      "step": 17585
    },
    {
      "epoch": 10.371462264150944,
      "grad_norm": 2.730050802230835,
      "learning_rate": 5.547914624591597e-06,
      "loss": 0.5275,
      "num_input_tokens_seen": 11462160,
      "step": 17590
    },
    {
      "epoch": 10.37441037735849,
      "grad_norm": 3.473198890686035,
      "learning_rate": 5.545357331830269e-06,
      "loss": 0.4625,
      "num_input_tokens_seen": 11465488,
      "step": 17595
    },
    {
      "epoch": 10.377358490566039,
      "grad_norm": 2.4932329654693604,
      "learning_rate": 5.542799894683235e-06,
      "loss": 0.3093,
      "num_input_tokens_seen": 11467792,
      "step": 17600
    },
    {
      "epoch": 10.380306603773585,
      "grad_norm": 3.3443124294281006,
      "learning_rate": 5.540242313827591e-06,
      "loss": 0.3896,
      "num_input_tokens_seen": 11470896,
      "step": 17605
    },
    {
      "epoch": 10.383254716981131,
      "grad_norm": 8.577765464782715,
      "learning_rate": 5.537684589940466e-06,
      "loss": 0.3657,
      "num_input_tokens_seen": 11473648,
      "step": 17610
    },
    {
      "epoch": 10.38620283018868,
      "grad_norm": 5.007880687713623,
      "learning_rate": 5.535126723699029e-06,
      "loss": 0.4337,
      "num_input_tokens_seen": 11477136,
      "step": 17615
    },
    {
      "epoch": 10.389150943396226,
      "grad_norm": 3.1569900512695312,
      "learning_rate": 5.532568715780485e-06,
      "loss": 0.3108,
      "num_input_tokens_seen": 11480080,
      "step": 17620
    },
    {
      "epoch": 10.392099056603774,
      "grad_norm": 2.7042129039764404,
      "learning_rate": 5.530010566862077e-06,
      "loss": 0.4411,
      "num_input_tokens_seen": 11482992,
      "step": 17625
    },
    {
      "epoch": 10.39504716981132,
      "grad_norm": 2.844331741333008,
      "learning_rate": 5.527452277621089e-06,
      "loss": 0.3397,
      "num_input_tokens_seen": 11486512,
      "step": 17630
    },
    {
      "epoch": 10.397995283018869,
      "grad_norm": 2.6111152172088623,
      "learning_rate": 5.524893848734837e-06,
      "loss": 0.4316,
      "num_input_tokens_seen": 11490448,
      "step": 17635
    },
    {
      "epoch": 10.400943396226415,
      "grad_norm": 2.7033448219299316,
      "learning_rate": 5.522335280880676e-06,
      "loss": 0.4368,
      "num_input_tokens_seen": 11493520,
      "step": 17640
    },
    {
      "epoch": 10.403891509433961,
      "grad_norm": 3.7705066204071045,
      "learning_rate": 5.519776574735999e-06,
      "loss": 0.3717,
      "num_input_tokens_seen": 11497232,
      "step": 17645
    },
    {
      "epoch": 10.40683962264151,
      "grad_norm": 2.8128511905670166,
      "learning_rate": 5.5172177309782325e-06,
      "loss": 0.3409,
      "num_input_tokens_seen": 11501072,
      "step": 17650
    },
    {
      "epoch": 10.409787735849056,
      "grad_norm": 3.1056880950927734,
      "learning_rate": 5.514658750284844e-06,
      "loss": 0.4006,
      "num_input_tokens_seen": 11504880,
      "step": 17655
    },
    {
      "epoch": 10.412735849056604,
      "grad_norm": 2.853879451751709,
      "learning_rate": 5.512099633333332e-06,
      "loss": 0.317,
      "num_input_tokens_seen": 11508336,
      "step": 17660
    },
    {
      "epoch": 10.41568396226415,
      "grad_norm": 4.7368974685668945,
      "learning_rate": 5.509540380801236e-06,
      "loss": 0.4241,
      "num_input_tokens_seen": 11511344,
      "step": 17665
    },
    {
      "epoch": 10.418632075471699,
      "grad_norm": 4.0707902908325195,
      "learning_rate": 5.506980993366129e-06,
      "loss": 0.3733,
      "num_input_tokens_seen": 11515984,
      "step": 17670
    },
    {
      "epoch": 10.421580188679245,
      "grad_norm": 2.349018096923828,
      "learning_rate": 5.504421471705616e-06,
      "loss": 0.4111,
      "num_input_tokens_seen": 11518992,
      "step": 17675
    },
    {
      "epoch": 10.424528301886792,
      "grad_norm": 6.048584461212158,
      "learning_rate": 5.501861816497346e-06,
      "loss": 0.4107,
      "num_input_tokens_seen": 11523312,
      "step": 17680
    },
    {
      "epoch": 10.42747641509434,
      "grad_norm": 2.32818865776062,
      "learning_rate": 5.499302028418998e-06,
      "loss": 0.3727,
      "num_input_tokens_seen": 11526960,
      "step": 17685
    },
    {
      "epoch": 10.430424528301886,
      "grad_norm": 1.6205062866210938,
      "learning_rate": 5.496742108148285e-06,
      "loss": 0.4616,
      "num_input_tokens_seen": 11530096,
      "step": 17690
    },
    {
      "epoch": 10.433372641509434,
      "grad_norm": 3.2586190700531006,
      "learning_rate": 5.494182056362959e-06,
      "loss": 0.308,
      "num_input_tokens_seen": 11533328,
      "step": 17695
    },
    {
      "epoch": 10.43632075471698,
      "grad_norm": 3.2934730052948,
      "learning_rate": 5.491621873740804e-06,
      "loss": 0.4698,
      "num_input_tokens_seen": 11536336,
      "step": 17700
    },
    {
      "epoch": 10.439268867924529,
      "grad_norm": 1.8386735916137695,
      "learning_rate": 5.4890615609596375e-06,
      "loss": 0.5364,
      "num_input_tokens_seen": 11538928,
      "step": 17705
    },
    {
      "epoch": 10.442216981132075,
      "grad_norm": 2.0543055534362793,
      "learning_rate": 5.486501118697317e-06,
      "loss": 0.3848,
      "num_input_tokens_seen": 11541520,
      "step": 17710
    },
    {
      "epoch": 10.445165094339623,
      "grad_norm": 1.2610985040664673,
      "learning_rate": 5.483940547631727e-06,
      "loss": 0.3601,
      "num_input_tokens_seen": 11544976,
      "step": 17715
    },
    {
      "epoch": 10.44811320754717,
      "grad_norm": 1.874112844467163,
      "learning_rate": 5.481379848440792e-06,
      "loss": 0.5124,
      "num_input_tokens_seen": 11549072,
      "step": 17720
    },
    {
      "epoch": 10.451061320754716,
      "grad_norm": 1.7469552755355835,
      "learning_rate": 5.478819021802468e-06,
      "loss": 0.3794,
      "num_input_tokens_seen": 11552080,
      "step": 17725
    },
    {
      "epoch": 10.454009433962264,
      "grad_norm": 3.6463067531585693,
      "learning_rate": 5.476258068394743e-06,
      "loss": 0.2933,
      "num_input_tokens_seen": 11555920,
      "step": 17730
    },
    {
      "epoch": 10.45695754716981,
      "grad_norm": 2.2082536220550537,
      "learning_rate": 5.473696988895644e-06,
      "loss": 0.2379,
      "num_input_tokens_seen": 11560112,
      "step": 17735
    },
    {
      "epoch": 10.459905660377359,
      "grad_norm": 3.2026994228363037,
      "learning_rate": 5.471135783983224e-06,
      "loss": 0.3868,
      "num_input_tokens_seen": 11563184,
      "step": 17740
    },
    {
      "epoch": 10.462853773584905,
      "grad_norm": 2.426938056945801,
      "learning_rate": 5.4685744543355745e-06,
      "loss": 0.341,
      "num_input_tokens_seen": 11566384,
      "step": 17745
    },
    {
      "epoch": 10.465801886792454,
      "grad_norm": 2.6454083919525146,
      "learning_rate": 5.466013000630819e-06,
      "loss": 0.3955,
      "num_input_tokens_seen": 11569392,
      "step": 17750
    },
    {
      "epoch": 10.46875,
      "grad_norm": 1.6985721588134766,
      "learning_rate": 5.463451423547114e-06,
      "loss": 0.3669,
      "num_input_tokens_seen": 11573264,
      "step": 17755
    },
    {
      "epoch": 10.471698113207546,
      "grad_norm": 1.9474225044250488,
      "learning_rate": 5.460889723762647e-06,
      "loss": 0.4188,
      "num_input_tokens_seen": 11576816,
      "step": 17760
    },
    {
      "epoch": 10.474646226415095,
      "grad_norm": 3.021440267562866,
      "learning_rate": 5.458327901955639e-06,
      "loss": 0.3481,
      "num_input_tokens_seen": 11580656,
      "step": 17765
    },
    {
      "epoch": 10.477594339622641,
      "grad_norm": 2.984830379486084,
      "learning_rate": 5.455765958804344e-06,
      "loss": 0.3316,
      "num_input_tokens_seen": 11583504,
      "step": 17770
    },
    {
      "epoch": 10.48054245283019,
      "grad_norm": 1.9020577669143677,
      "learning_rate": 5.4532038949870455e-06,
      "loss": 0.3063,
      "num_input_tokens_seen": 11587056,
      "step": 17775
    },
    {
      "epoch": 10.483490566037736,
      "grad_norm": 1.5156751871109009,
      "learning_rate": 5.450641711182066e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 11590448,
      "step": 17780
    },
    {
      "epoch": 10.486438679245284,
      "grad_norm": 3.9102132320404053,
      "learning_rate": 5.448079408067748e-06,
      "loss": 0.3105,
      "num_input_tokens_seen": 11593200,
      "step": 17785
    },
    {
      "epoch": 10.48938679245283,
      "grad_norm": 2.7179441452026367,
      "learning_rate": 5.4455169863224775e-06,
      "loss": 0.3819,
      "num_input_tokens_seen": 11596880,
      "step": 17790
    },
    {
      "epoch": 10.492334905660377,
      "grad_norm": 2.3332467079162598,
      "learning_rate": 5.442954446624664e-06,
      "loss": 0.4312,
      "num_input_tokens_seen": 11599632,
      "step": 17795
    },
    {
      "epoch": 10.495283018867925,
      "grad_norm": 4.443470478057861,
      "learning_rate": 5.440391789652752e-06,
      "loss": 0.5003,
      "num_input_tokens_seen": 11602416,
      "step": 17800
    },
    {
      "epoch": 10.498231132075471,
      "grad_norm": 6.064513683319092,
      "learning_rate": 5.437829016085216e-06,
      "loss": 0.271,
      "num_input_tokens_seen": 11605008,
      "step": 17805
    },
    {
      "epoch": 10.50117924528302,
      "grad_norm": 3.5286128520965576,
      "learning_rate": 5.435266126600561e-06,
      "loss": 0.3225,
      "num_input_tokens_seen": 11607856,
      "step": 17810
    },
    {
      "epoch": 10.504127358490566,
      "grad_norm": 1.9397250413894653,
      "learning_rate": 5.4327031218773215e-06,
      "loss": 0.4079,
      "num_input_tokens_seen": 11611184,
      "step": 17815
    },
    {
      "epoch": 10.507075471698114,
      "grad_norm": 2.8401334285736084,
      "learning_rate": 5.430140002594067e-06,
      "loss": 0.3475,
      "num_input_tokens_seen": 11614704,
      "step": 17820
    },
    {
      "epoch": 10.51002358490566,
      "grad_norm": 6.736853122711182,
      "learning_rate": 5.4275767694293934e-06,
      "loss": 0.378,
      "num_input_tokens_seen": 11617584,
      "step": 17825
    },
    {
      "epoch": 10.512971698113208,
      "grad_norm": 2.2772741317749023,
      "learning_rate": 5.425013423061926e-06,
      "loss": 0.4474,
      "num_input_tokens_seen": 11620464,
      "step": 17830
    },
    {
      "epoch": 10.515919811320755,
      "grad_norm": 4.8379902839660645,
      "learning_rate": 5.422449964170324e-06,
      "loss": 0.2538,
      "num_input_tokens_seen": 11623696,
      "step": 17835
    },
    {
      "epoch": 10.518867924528301,
      "grad_norm": 3.0294594764709473,
      "learning_rate": 5.419886393433275e-06,
      "loss": 0.397,
      "num_input_tokens_seen": 11626928,
      "step": 17840
    },
    {
      "epoch": 10.52181603773585,
      "grad_norm": 1.425459861755371,
      "learning_rate": 5.417322711529491e-06,
      "loss": 0.3661,
      "num_input_tokens_seen": 11629808,
      "step": 17845
    },
    {
      "epoch": 10.524764150943396,
      "grad_norm": 2.107189655303955,
      "learning_rate": 5.4147589191377224e-06,
      "loss": 0.3338,
      "num_input_tokens_seen": 11633584,
      "step": 17850
    },
    {
      "epoch": 10.527712264150944,
      "grad_norm": 2.7019550800323486,
      "learning_rate": 5.412195016936742e-06,
      "loss": 0.5934,
      "num_input_tokens_seen": 11636816,
      "step": 17855
    },
    {
      "epoch": 10.53066037735849,
      "grad_norm": 2.7533226013183594,
      "learning_rate": 5.409631005605354e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 11640048,
      "step": 17860
    },
    {
      "epoch": 10.533608490566039,
      "grad_norm": 2.909158229827881,
      "learning_rate": 5.407066885822391e-06,
      "loss": 0.2739,
      "num_input_tokens_seen": 11643184,
      "step": 17865
    },
    {
      "epoch": 10.536556603773585,
      "grad_norm": 2.6675121784210205,
      "learning_rate": 5.404502658266717e-06,
      "loss": 0.3957,
      "num_input_tokens_seen": 11646608,
      "step": 17870
    },
    {
      "epoch": 10.539504716981131,
      "grad_norm": 3.662813663482666,
      "learning_rate": 5.4019383236172195e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 11652528,
      "step": 17875
    },
    {
      "epoch": 10.54245283018868,
      "grad_norm": 1.8791899681091309,
      "learning_rate": 5.39937388255282e-06,
      "loss": 0.4047,
      "num_input_tokens_seen": 11656016,
      "step": 17880
    },
    {
      "epoch": 10.545400943396226,
      "grad_norm": 1.9273725748062134,
      "learning_rate": 5.3968093357524645e-06,
      "loss": 0.3189,
      "num_input_tokens_seen": 11659152,
      "step": 17885
    },
    {
      "epoch": 10.548349056603774,
      "grad_norm": 2.5352768898010254,
      "learning_rate": 5.3942446838951245e-06,
      "loss": 0.3136,
      "num_input_tokens_seen": 11662416,
      "step": 17890
    },
    {
      "epoch": 10.55129716981132,
      "grad_norm": 4.709670543670654,
      "learning_rate": 5.3916799276598074e-06,
      "loss": 0.4149,
      "num_input_tokens_seen": 11666288,
      "step": 17895
    },
    {
      "epoch": 10.554245283018869,
      "grad_norm": 2.5742149353027344,
      "learning_rate": 5.3891150677255425e-06,
      "loss": 0.3559,
      "num_input_tokens_seen": 11669648,
      "step": 17900
    },
    {
      "epoch": 10.557193396226415,
      "grad_norm": 2.2333507537841797,
      "learning_rate": 5.386550104771384e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 11673808,
      "step": 17905
    },
    {
      "epoch": 10.560141509433961,
      "grad_norm": 3.6738831996917725,
      "learning_rate": 5.3839850394764205e-06,
      "loss": 0.3883,
      "num_input_tokens_seen": 11677136,
      "step": 17910
    },
    {
      "epoch": 10.56308962264151,
      "grad_norm": 1.8079829216003418,
      "learning_rate": 5.381419872519763e-06,
      "loss": 0.307,
      "num_input_tokens_seen": 11679984,
      "step": 17915
    },
    {
      "epoch": 10.566037735849056,
      "grad_norm": 2.831439733505249,
      "learning_rate": 5.378854604580549e-06,
      "loss": 0.273,
      "num_input_tokens_seen": 11682992,
      "step": 17920
    },
    {
      "epoch": 10.568985849056604,
      "grad_norm": 2.5175082683563232,
      "learning_rate": 5.376289236337946e-06,
      "loss": 0.3645,
      "num_input_tokens_seen": 11687024,
      "step": 17925
    },
    {
      "epoch": 10.57193396226415,
      "grad_norm": 1.7944520711898804,
      "learning_rate": 5.373723768471147e-06,
      "loss": 0.3354,
      "num_input_tokens_seen": 11690160,
      "step": 17930
    },
    {
      "epoch": 10.574882075471699,
      "grad_norm": 2.9446234703063965,
      "learning_rate": 5.37115820165937e-06,
      "loss": 0.3308,
      "num_input_tokens_seen": 11694256,
      "step": 17935
    },
    {
      "epoch": 10.577830188679245,
      "grad_norm": 3.007157802581787,
      "learning_rate": 5.368592536581858e-06,
      "loss": 0.2975,
      "num_input_tokens_seen": 11697808,
      "step": 17940
    },
    {
      "epoch": 10.580778301886792,
      "grad_norm": 3.278438091278076,
      "learning_rate": 5.366026773917885e-06,
      "loss": 0.337,
      "num_input_tokens_seen": 11701904,
      "step": 17945
    },
    {
      "epoch": 10.58372641509434,
      "grad_norm": 2.635357618331909,
      "learning_rate": 5.363460914346746e-06,
      "loss": 0.325,
      "num_input_tokens_seen": 11704464,
      "step": 17950
    },
    {
      "epoch": 10.586674528301886,
      "grad_norm": 3.6010220050811768,
      "learning_rate": 5.360894958547762e-06,
      "loss": 0.3533,
      "num_input_tokens_seen": 11707408,
      "step": 17955
    },
    {
      "epoch": 10.589622641509434,
      "grad_norm": 2.788458824157715,
      "learning_rate": 5.358328907200284e-06,
      "loss": 0.3972,
      "num_input_tokens_seen": 11710544,
      "step": 17960
    },
    {
      "epoch": 10.59257075471698,
      "grad_norm": 1.6266142129898071,
      "learning_rate": 5.355762760983682e-06,
      "loss": 0.2769,
      "num_input_tokens_seen": 11714768,
      "step": 17965
    },
    {
      "epoch": 10.595518867924529,
      "grad_norm": 2.9958503246307373,
      "learning_rate": 5.353196520577356e-06,
      "loss": 0.4167,
      "num_input_tokens_seen": 11718256,
      "step": 17970
    },
    {
      "epoch": 10.598466981132075,
      "grad_norm": 2.616300106048584,
      "learning_rate": 5.35063018666073e-06,
      "loss": 0.4275,
      "num_input_tokens_seen": 11721040,
      "step": 17975
    },
    {
      "epoch": 10.601415094339622,
      "grad_norm": 2.9427073001861572,
      "learning_rate": 5.3480637599132515e-06,
      "loss": 0.4001,
      "num_input_tokens_seen": 11724592,
      "step": 17980
    },
    {
      "epoch": 10.60436320754717,
      "grad_norm": 2.1944713592529297,
      "learning_rate": 5.34549724101439e-06,
      "loss": 0.3692,
      "num_input_tokens_seen": 11729200,
      "step": 17985
    },
    {
      "epoch": 10.607311320754716,
      "grad_norm": 3.8543853759765625,
      "learning_rate": 5.342930630643646e-06,
      "loss": 0.4728,
      "num_input_tokens_seen": 11732368,
      "step": 17990
    },
    {
      "epoch": 10.610259433962264,
      "grad_norm": 1.748189091682434,
      "learning_rate": 5.340363929480541e-06,
      "loss": 0.3403,
      "num_input_tokens_seen": 11735856,
      "step": 17995
    },
    {
      "epoch": 10.61320754716981,
      "grad_norm": 2.9466164112091064,
      "learning_rate": 5.3377971382046164e-06,
      "loss": 0.4121,
      "num_input_tokens_seen": 11739184,
      "step": 18000
    },
    {
      "epoch": 10.616155660377359,
      "grad_norm": 4.3717427253723145,
      "learning_rate": 5.335230257495446e-06,
      "loss": 0.4749,
      "num_input_tokens_seen": 11741680,
      "step": 18005
    },
    {
      "epoch": 10.619103773584905,
      "grad_norm": 1.578360676765442,
      "learning_rate": 5.3326632880326205e-06,
      "loss": 0.3491,
      "num_input_tokens_seen": 11745616,
      "step": 18010
    },
    {
      "epoch": 10.622051886792454,
      "grad_norm": 3.349447727203369,
      "learning_rate": 5.3300962304957515e-06,
      "loss": 0.3061,
      "num_input_tokens_seen": 11748528,
      "step": 18015
    },
    {
      "epoch": 10.625,
      "grad_norm": 3.4133248329162598,
      "learning_rate": 5.327529085564487e-06,
      "loss": 0.4662,
      "num_input_tokens_seen": 11751056,
      "step": 18020
    },
    {
      "epoch": 10.627948113207546,
      "grad_norm": 2.131540060043335,
      "learning_rate": 5.324961853918485e-06,
      "loss": 0.3393,
      "num_input_tokens_seen": 11754864,
      "step": 18025
    },
    {
      "epoch": 10.630896226415095,
      "grad_norm": 2.8473474979400635,
      "learning_rate": 5.32239453623743e-06,
      "loss": 0.371,
      "num_input_tokens_seen": 11758256,
      "step": 18030
    },
    {
      "epoch": 10.633844339622641,
      "grad_norm": 4.949124813079834,
      "learning_rate": 5.3198271332010335e-06,
      "loss": 0.3751,
      "num_input_tokens_seen": 11762096,
      "step": 18035
    },
    {
      "epoch": 10.63679245283019,
      "grad_norm": 3.5244696140289307,
      "learning_rate": 5.317259645489024e-06,
      "loss": 0.3324,
      "num_input_tokens_seen": 11765872,
      "step": 18040
    },
    {
      "epoch": 10.639740566037736,
      "grad_norm": 2.2902731895446777,
      "learning_rate": 5.314692073781157e-06,
      "loss": 0.2893,
      "num_input_tokens_seen": 11768528,
      "step": 18045
    },
    {
      "epoch": 10.642688679245284,
      "grad_norm": 1.5570361614227295,
      "learning_rate": 5.312124418757207e-06,
      "loss": 0.3557,
      "num_input_tokens_seen": 11772208,
      "step": 18050
    },
    {
      "epoch": 10.64563679245283,
      "grad_norm": 2.6260485649108887,
      "learning_rate": 5.309556681096972e-06,
      "loss": 0.5397,
      "num_input_tokens_seen": 11776528,
      "step": 18055
    },
    {
      "epoch": 10.648584905660378,
      "grad_norm": 2.0930910110473633,
      "learning_rate": 5.306988861480271e-06,
      "loss": 0.3975,
      "num_input_tokens_seen": 11780176,
      "step": 18060
    },
    {
      "epoch": 10.651533018867925,
      "grad_norm": 2.8467299938201904,
      "learning_rate": 5.304420960586946e-06,
      "loss": 0.3357,
      "num_input_tokens_seen": 11783216,
      "step": 18065
    },
    {
      "epoch": 10.654481132075471,
      "grad_norm": 1.5697345733642578,
      "learning_rate": 5.3018529790968606e-06,
      "loss": 0.3338,
      "num_input_tokens_seen": 11789904,
      "step": 18070
    },
    {
      "epoch": 10.65742924528302,
      "grad_norm": 2.2542638778686523,
      "learning_rate": 5.299284917689898e-06,
      "loss": 0.4727,
      "num_input_tokens_seen": 11793040,
      "step": 18075
    },
    {
      "epoch": 10.660377358490566,
      "grad_norm": 5.108577728271484,
      "learning_rate": 5.296716777045962e-06,
      "loss": 0.3884,
      "num_input_tokens_seen": 11796336,
      "step": 18080
    },
    {
      "epoch": 10.663325471698114,
      "grad_norm": 2.5337820053100586,
      "learning_rate": 5.294148557844983e-06,
      "loss": 0.3869,
      "num_input_tokens_seen": 11799568,
      "step": 18085
    },
    {
      "epoch": 10.66627358490566,
      "grad_norm": 2.8313708305358887,
      "learning_rate": 5.291580260766904e-06,
      "loss": 0.3317,
      "num_input_tokens_seen": 11802960,
      "step": 18090
    },
    {
      "epoch": 10.669221698113208,
      "grad_norm": 2.5068671703338623,
      "learning_rate": 5.289011886491694e-06,
      "loss": 0.3257,
      "num_input_tokens_seen": 11807088,
      "step": 18095
    },
    {
      "epoch": 10.672169811320755,
      "grad_norm": 2.98797607421875,
      "learning_rate": 5.286443435699342e-06,
      "loss": 0.3056,
      "num_input_tokens_seen": 11811920,
      "step": 18100
    },
    {
      "epoch": 10.675117924528301,
      "grad_norm": 1.6492878198623657,
      "learning_rate": 5.283874909069855e-06,
      "loss": 0.407,
      "num_input_tokens_seen": 11814800,
      "step": 18105
    },
    {
      "epoch": 10.67806603773585,
      "grad_norm": 2.4847452640533447,
      "learning_rate": 5.281306307283263e-06,
      "loss": 0.445,
      "num_input_tokens_seen": 11818224,
      "step": 18110
    },
    {
      "epoch": 10.681014150943396,
      "grad_norm": 3.850813865661621,
      "learning_rate": 5.2787376310196145e-06,
      "loss": 0.3669,
      "num_input_tokens_seen": 11821424,
      "step": 18115
    },
    {
      "epoch": 10.683962264150944,
      "grad_norm": 4.274418830871582,
      "learning_rate": 5.276168880958977e-06,
      "loss": 0.355,
      "num_input_tokens_seen": 11823888,
      "step": 18120
    },
    {
      "epoch": 10.68691037735849,
      "grad_norm": 6.18873929977417,
      "learning_rate": 5.273600057781437e-06,
      "loss": 0.3494,
      "num_input_tokens_seen": 11826896,
      "step": 18125
    },
    {
      "epoch": 10.689858490566039,
      "grad_norm": 4.320852756500244,
      "learning_rate": 5.271031162167103e-06,
      "loss": 0.362,
      "num_input_tokens_seen": 11830224,
      "step": 18130
    },
    {
      "epoch": 10.692806603773585,
      "grad_norm": 2.789550304412842,
      "learning_rate": 5.268462194796101e-06,
      "loss": 0.3199,
      "num_input_tokens_seen": 11833456,
      "step": 18135
    },
    {
      "epoch": 10.695754716981131,
      "grad_norm": 2.754539966583252,
      "learning_rate": 5.265893156348576e-06,
      "loss": 0.4827,
      "num_input_tokens_seen": 11836848,
      "step": 18140
    },
    {
      "epoch": 10.69870283018868,
      "grad_norm": 3.3421356678009033,
      "learning_rate": 5.2633240475046925e-06,
      "loss": 0.4115,
      "num_input_tokens_seen": 11839504,
      "step": 18145
    },
    {
      "epoch": 10.701650943396226,
      "grad_norm": 2.5830318927764893,
      "learning_rate": 5.2607548689446305e-06,
      "loss": 0.4453,
      "num_input_tokens_seen": 11842640,
      "step": 18150
    },
    {
      "epoch": 10.704599056603774,
      "grad_norm": 5.124602794647217,
      "learning_rate": 5.258185621348595e-06,
      "loss": 0.3442,
      "num_input_tokens_seen": 11844592,
      "step": 18155
    },
    {
      "epoch": 10.70754716981132,
      "grad_norm": 2.321617603302002,
      "learning_rate": 5.255616305396801e-06,
      "loss": 0.5397,
      "num_input_tokens_seen": 11847696,
      "step": 18160
    },
    {
      "epoch": 10.710495283018869,
      "grad_norm": 4.331507205963135,
      "learning_rate": 5.253046921769491e-06,
      "loss": 0.3407,
      "num_input_tokens_seen": 11850672,
      "step": 18165
    },
    {
      "epoch": 10.713443396226415,
      "grad_norm": 2.513171911239624,
      "learning_rate": 5.250477471146916e-06,
      "loss": 0.3258,
      "num_input_tokens_seen": 11854480,
      "step": 18170
    },
    {
      "epoch": 10.716391509433961,
      "grad_norm": 3.0567197799682617,
      "learning_rate": 5.2479079542093535e-06,
      "loss": 0.2342,
      "num_input_tokens_seen": 11858032,
      "step": 18175
    },
    {
      "epoch": 10.71933962264151,
      "grad_norm": 3.2821991443634033,
      "learning_rate": 5.245338371637091e-06,
      "loss": 0.3329,
      "num_input_tokens_seen": 11861296,
      "step": 18180
    },
    {
      "epoch": 10.722287735849056,
      "grad_norm": 3.178295612335205,
      "learning_rate": 5.242768724110437e-06,
      "loss": 0.4341,
      "num_input_tokens_seen": 11863344,
      "step": 18185
    },
    {
      "epoch": 10.725235849056604,
      "grad_norm": 3.619694232940674,
      "learning_rate": 5.240199012309717e-06,
      "loss": 0.3417,
      "num_input_tokens_seen": 11866704,
      "step": 18190
    },
    {
      "epoch": 10.72818396226415,
      "grad_norm": 2.2598514556884766,
      "learning_rate": 5.237629236915273e-06,
      "loss": 0.3639,
      "num_input_tokens_seen": 11869648,
      "step": 18195
    },
    {
      "epoch": 10.731132075471699,
      "grad_norm": 2.342189073562622,
      "learning_rate": 5.235059398607464e-06,
      "loss": 0.4036,
      "num_input_tokens_seen": 11873040,
      "step": 18200
    },
    {
      "epoch": 10.734080188679245,
      "grad_norm": 2.396108388900757,
      "learning_rate": 5.232489498066665e-06,
      "loss": 0.328,
      "num_input_tokens_seen": 11876176,
      "step": 18205
    },
    {
      "epoch": 10.737028301886792,
      "grad_norm": 3.746102809906006,
      "learning_rate": 5.229919535973272e-06,
      "loss": 0.4602,
      "num_input_tokens_seen": 11878864,
      "step": 18210
    },
    {
      "epoch": 10.73997641509434,
      "grad_norm": 4.024622917175293,
      "learning_rate": 5.2273495130076905e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 11882192,
      "step": 18215
    },
    {
      "epoch": 10.742924528301886,
      "grad_norm": 3.0105676651000977,
      "learning_rate": 5.224779429850344e-06,
      "loss": 0.3297,
      "num_input_tokens_seen": 11885040,
      "step": 18220
    },
    {
      "epoch": 10.745872641509434,
      "grad_norm": 3.0587942600250244,
      "learning_rate": 5.222209287181677e-06,
      "loss": 0.3419,
      "num_input_tokens_seen": 11888112,
      "step": 18225
    },
    {
      "epoch": 10.74882075471698,
      "grad_norm": 1.7900869846343994,
      "learning_rate": 5.219639085682142e-06,
      "loss": 0.2961,
      "num_input_tokens_seen": 11891440,
      "step": 18230
    },
    {
      "epoch": 10.751768867924529,
      "grad_norm": 1.9678398370742798,
      "learning_rate": 5.2170688260322124e-06,
      "loss": 0.3792,
      "num_input_tokens_seen": 11894224,
      "step": 18235
    },
    {
      "epoch": 10.754716981132075,
      "grad_norm": 3.475299119949341,
      "learning_rate": 5.214498508912376e-06,
      "loss": 0.4167,
      "num_input_tokens_seen": 11898384,
      "step": 18240
    },
    {
      "epoch": 10.757665094339622,
      "grad_norm": 5.283947467803955,
      "learning_rate": 5.211928135003135e-06,
      "loss": 0.293,
      "num_input_tokens_seen": 11901296,
      "step": 18245
    },
    {
      "epoch": 10.76061320754717,
      "grad_norm": 2.512531042098999,
      "learning_rate": 5.209357704985007e-06,
      "loss": 0.4455,
      "num_input_tokens_seen": 11904336,
      "step": 18250
    },
    {
      "epoch": 10.763561320754716,
      "grad_norm": 3.9295547008514404,
      "learning_rate": 5.206787219538524e-06,
      "loss": 0.3935,
      "num_input_tokens_seen": 11907760,
      "step": 18255
    },
    {
      "epoch": 10.766509433962264,
      "grad_norm": 3.534097671508789,
      "learning_rate": 5.204216679344234e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 11910544,
      "step": 18260
    },
    {
      "epoch": 10.76945754716981,
      "grad_norm": 3.6369435787200928,
      "learning_rate": 5.201646085082696e-06,
      "loss": 0.3224,
      "num_input_tokens_seen": 11913200,
      "step": 18265
    },
    {
      "epoch": 10.772405660377359,
      "grad_norm": 1.9674863815307617,
      "learning_rate": 5.199075437434491e-06,
      "loss": 0.3407,
      "num_input_tokens_seen": 11915856,
      "step": 18270
    },
    {
      "epoch": 10.775353773584905,
      "grad_norm": 3.5680670738220215,
      "learning_rate": 5.1965047370802046e-06,
      "loss": 0.4124,
      "num_input_tokens_seen": 11919280,
      "step": 18275
    },
    {
      "epoch": 10.778301886792454,
      "grad_norm": 3.434519052505493,
      "learning_rate": 5.193933984700441e-06,
      "loss": 0.4124,
      "num_input_tokens_seen": 11922192,
      "step": 18280
    },
    {
      "epoch": 10.78125,
      "grad_norm": 4.892963886260986,
      "learning_rate": 5.1913631809758216e-06,
      "loss": 0.3011,
      "num_input_tokens_seen": 11925584,
      "step": 18285
    },
    {
      "epoch": 10.784198113207546,
      "grad_norm": 2.604424476623535,
      "learning_rate": 5.188792326586973e-06,
      "loss": 0.4597,
      "num_input_tokens_seen": 11928720,
      "step": 18290
    },
    {
      "epoch": 10.787146226415095,
      "grad_norm": 3.5810248851776123,
      "learning_rate": 5.186221422214544e-06,
      "loss": 0.3354,
      "num_input_tokens_seen": 11933232,
      "step": 18295
    },
    {
      "epoch": 10.790094339622641,
      "grad_norm": 3.4061319828033447,
      "learning_rate": 5.1836504685391885e-06,
      "loss": 0.3631,
      "num_input_tokens_seen": 11936208,
      "step": 18300
    },
    {
      "epoch": 10.79304245283019,
      "grad_norm": 3.354257106781006,
      "learning_rate": 5.181079466241582e-06,
      "loss": 0.3286,
      "num_input_tokens_seen": 11939504,
      "step": 18305
    },
    {
      "epoch": 10.795990566037736,
      "grad_norm": 2.7241334915161133,
      "learning_rate": 5.178508416002406e-06,
      "loss": 0.4243,
      "num_input_tokens_seen": 11941872,
      "step": 18310
    },
    {
      "epoch": 10.798938679245284,
      "grad_norm": 0.5349394083023071,
      "learning_rate": 5.175937318502357e-06,
      "loss": 0.2514,
      "num_input_tokens_seen": 11949072,
      "step": 18315
    },
    {
      "epoch": 10.80188679245283,
      "grad_norm": 1.9773993492126465,
      "learning_rate": 5.173366174422147e-06,
      "loss": 0.3245,
      "num_input_tokens_seen": 11952144,
      "step": 18320
    },
    {
      "epoch": 10.804834905660378,
      "grad_norm": 2.267798900604248,
      "learning_rate": 5.170794984442492e-06,
      "loss": 0.6438,
      "num_input_tokens_seen": 11954640,
      "step": 18325
    },
    {
      "epoch": 10.807783018867925,
      "grad_norm": 1.7042869329452515,
      "learning_rate": 5.16822374924413e-06,
      "loss": 0.2986,
      "num_input_tokens_seen": 11957968,
      "step": 18330
    },
    {
      "epoch": 10.810731132075471,
      "grad_norm": 3.274555206298828,
      "learning_rate": 5.165652469507806e-06,
      "loss": 0.4318,
      "num_input_tokens_seen": 11960944,
      "step": 18335
    },
    {
      "epoch": 10.81367924528302,
      "grad_norm": 2.5985512733459473,
      "learning_rate": 5.163081145914276e-06,
      "loss": 0.3977,
      "num_input_tokens_seen": 11964496,
      "step": 18340
    },
    {
      "epoch": 10.816627358490566,
      "grad_norm": 2.8874995708465576,
      "learning_rate": 5.160509779144311e-06,
      "loss": 0.3632,
      "num_input_tokens_seen": 11967152,
      "step": 18345
    },
    {
      "epoch": 10.819575471698114,
      "grad_norm": 4.250257968902588,
      "learning_rate": 5.157938369878688e-06,
      "loss": 0.394,
      "num_input_tokens_seen": 11969744,
      "step": 18350
    },
    {
      "epoch": 10.82252358490566,
      "grad_norm": 2.335707426071167,
      "learning_rate": 5.155366918798203e-06,
      "loss": 0.2809,
      "num_input_tokens_seen": 11972400,
      "step": 18355
    },
    {
      "epoch": 10.825471698113208,
      "grad_norm": 1.4992411136627197,
      "learning_rate": 5.152795426583654e-06,
      "loss": 0.3309,
      "num_input_tokens_seen": 11974992,
      "step": 18360
    },
    {
      "epoch": 10.828419811320755,
      "grad_norm": 2.5281145572662354,
      "learning_rate": 5.15022389391586e-06,
      "loss": 0.3745,
      "num_input_tokens_seen": 11978096,
      "step": 18365
    },
    {
      "epoch": 10.831367924528301,
      "grad_norm": 3.674957275390625,
      "learning_rate": 5.147652321475642e-06,
      "loss": 0.5072,
      "num_input_tokens_seen": 11981008,
      "step": 18370
    },
    {
      "epoch": 10.83431603773585,
      "grad_norm": 1.9721482992172241,
      "learning_rate": 5.145080709943835e-06,
      "loss": 0.3936,
      "num_input_tokens_seen": 11984336,
      "step": 18375
    },
    {
      "epoch": 10.837264150943396,
      "grad_norm": 4.812885284423828,
      "learning_rate": 5.142509060001285e-06,
      "loss": 0.3093,
      "num_input_tokens_seen": 11986736,
      "step": 18380
    },
    {
      "epoch": 10.840212264150944,
      "grad_norm": 3.333235263824463,
      "learning_rate": 5.139937372328847e-06,
      "loss": 0.4739,
      "num_input_tokens_seen": 11990608,
      "step": 18385
    },
    {
      "epoch": 10.84316037735849,
      "grad_norm": 5.352551460266113,
      "learning_rate": 5.1373656476073876e-06,
      "loss": 0.3725,
      "num_input_tokens_seen": 11993552,
      "step": 18390
    },
    {
      "epoch": 10.846108490566039,
      "grad_norm": 4.58211088180542,
      "learning_rate": 5.134793886517779e-06,
      "loss": 0.3775,
      "num_input_tokens_seen": 11995952,
      "step": 18395
    },
    {
      "epoch": 10.849056603773585,
      "grad_norm": 2.18815016746521,
      "learning_rate": 5.1322220897409105e-06,
      "loss": 0.3981,
      "num_input_tokens_seen": 11999376,
      "step": 18400
    },
    {
      "epoch": 10.852004716981131,
      "grad_norm": 2.474853038787842,
      "learning_rate": 5.129650257957671e-06,
      "loss": 0.4274,
      "num_input_tokens_seen": 12003312,
      "step": 18405
    },
    {
      "epoch": 10.85495283018868,
      "grad_norm": 4.461069583892822,
      "learning_rate": 5.12707839184897e-06,
      "loss": 0.4039,
      "num_input_tokens_seen": 12006416,
      "step": 18410
    },
    {
      "epoch": 10.857900943396226,
      "grad_norm": 2.0852251052856445,
      "learning_rate": 5.124506492095716e-06,
      "loss": 0.3129,
      "num_input_tokens_seen": 12010160,
      "step": 18415
    },
    {
      "epoch": 10.860849056603774,
      "grad_norm": 1.6385436058044434,
      "learning_rate": 5.121934559378831e-06,
      "loss": 0.3494,
      "num_input_tokens_seen": 12014256,
      "step": 18420
    },
    {
      "epoch": 10.86379716981132,
      "grad_norm": 2.8635153770446777,
      "learning_rate": 5.1193625943792456e-06,
      "loss": 0.426,
      "num_input_tokens_seen": 12017456,
      "step": 18425
    },
    {
      "epoch": 10.866745283018869,
      "grad_norm": 2.319417953491211,
      "learning_rate": 5.116790597777901e-06,
      "loss": 0.3499,
      "num_input_tokens_seen": 12020720,
      "step": 18430
    },
    {
      "epoch": 10.869693396226415,
      "grad_norm": 1.647322177886963,
      "learning_rate": 5.11421857025574e-06,
      "loss": 0.2392,
      "num_input_tokens_seen": 12024080,
      "step": 18435
    },
    {
      "epoch": 10.872641509433961,
      "grad_norm": 5.10393762588501,
      "learning_rate": 5.111646512493721e-06,
      "loss": 0.3892,
      "num_input_tokens_seen": 12026736,
      "step": 18440
    },
    {
      "epoch": 10.87558962264151,
      "grad_norm": 1.9618160724639893,
      "learning_rate": 5.109074425172806e-06,
      "loss": 0.3227,
      "num_input_tokens_seen": 12029296,
      "step": 18445
    },
    {
      "epoch": 10.878537735849056,
      "grad_norm": 1.7240840196609497,
      "learning_rate": 5.106502308973967e-06,
      "loss": 0.3229,
      "num_input_tokens_seen": 12032880,
      "step": 18450
    },
    {
      "epoch": 10.881485849056604,
      "grad_norm": 2.490644693374634,
      "learning_rate": 5.103930164578184e-06,
      "loss": 0.4329,
      "num_input_tokens_seen": 12036912,
      "step": 18455
    },
    {
      "epoch": 10.88443396226415,
      "grad_norm": 2.762875556945801,
      "learning_rate": 5.101357992666441e-06,
      "loss": 0.3234,
      "num_input_tokens_seen": 12039632,
      "step": 18460
    },
    {
      "epoch": 10.887382075471699,
      "grad_norm": 2.833361864089966,
      "learning_rate": 5.098785793919733e-06,
      "loss": 0.422,
      "num_input_tokens_seen": 12043632,
      "step": 18465
    },
    {
      "epoch": 10.890330188679245,
      "grad_norm": 2.667381763458252,
      "learning_rate": 5.096213569019061e-06,
      "loss": 0.4758,
      "num_input_tokens_seen": 12047824,
      "step": 18470
    },
    {
      "epoch": 10.893278301886792,
      "grad_norm": 2.444791793823242,
      "learning_rate": 5.0936413186454315e-06,
      "loss": 0.2931,
      "num_input_tokens_seen": 12050480,
      "step": 18475
    },
    {
      "epoch": 10.89622641509434,
      "grad_norm": 2.135464906692505,
      "learning_rate": 5.0910690434798584e-06,
      "loss": 0.4547,
      "num_input_tokens_seen": 12053680,
      "step": 18480
    },
    {
      "epoch": 10.899174528301886,
      "grad_norm": 3.5650410652160645,
      "learning_rate": 5.088496744203364e-06,
      "loss": 0.3956,
      "num_input_tokens_seen": 12056688,
      "step": 18485
    },
    {
      "epoch": 10.902122641509434,
      "grad_norm": 2.479853391647339,
      "learning_rate": 5.085924421496976e-06,
      "loss": 0.3798,
      "num_input_tokens_seen": 12060720,
      "step": 18490
    },
    {
      "epoch": 10.90507075471698,
      "grad_norm": 3.08607816696167,
      "learning_rate": 5.083352076041725e-06,
      "loss": 0.2465,
      "num_input_tokens_seen": 12063216,
      "step": 18495
    },
    {
      "epoch": 10.908018867924529,
      "grad_norm": 3.8438076972961426,
      "learning_rate": 5.080779708518654e-06,
      "loss": 0.3978,
      "num_input_tokens_seen": 12066800,
      "step": 18500
    },
    {
      "epoch": 10.910966981132075,
      "grad_norm": 2.1845197677612305,
      "learning_rate": 5.078207319608807e-06,
      "loss": 0.3492,
      "num_input_tokens_seen": 12070320,
      "step": 18505
    },
    {
      "epoch": 10.913915094339622,
      "grad_norm": 2.693627119064331,
      "learning_rate": 5.075634909993235e-06,
      "loss": 0.3543,
      "num_input_tokens_seen": 12073296,
      "step": 18510
    },
    {
      "epoch": 10.91686320754717,
      "grad_norm": 3.956373453140259,
      "learning_rate": 5.073062480352995e-06,
      "loss": 0.336,
      "num_input_tokens_seen": 12075472,
      "step": 18515
    },
    {
      "epoch": 10.919811320754716,
      "grad_norm": 8.80846118927002,
      "learning_rate": 5.070490031369149e-06,
      "loss": 0.395,
      "num_input_tokens_seen": 12078384,
      "step": 18520
    },
    {
      "epoch": 10.922759433962264,
      "grad_norm": 2.1584599018096924,
      "learning_rate": 5.067917563722762e-06,
      "loss": 0.3274,
      "num_input_tokens_seen": 12081808,
      "step": 18525
    },
    {
      "epoch": 10.92570754716981,
      "grad_norm": 2.879453659057617,
      "learning_rate": 5.065345078094907e-06,
      "loss": 0.4302,
      "num_input_tokens_seen": 12085200,
      "step": 18530
    },
    {
      "epoch": 10.928655660377359,
      "grad_norm": 1.6405314207077026,
      "learning_rate": 5.062772575166663e-06,
      "loss": 0.3497,
      "num_input_tokens_seen": 12087888,
      "step": 18535
    },
    {
      "epoch": 10.931603773584905,
      "grad_norm": 3.7632956504821777,
      "learning_rate": 5.0602000556191075e-06,
      "loss": 0.2891,
      "num_input_tokens_seen": 12090608,
      "step": 18540
    },
    {
      "epoch": 10.934551886792454,
      "grad_norm": 2.590761661529541,
      "learning_rate": 5.0576275201333284e-06,
      "loss": 0.2786,
      "num_input_tokens_seen": 12094960,
      "step": 18545
    },
    {
      "epoch": 10.9375,
      "grad_norm": 3.6195850372314453,
      "learning_rate": 5.055054969390415e-06,
      "loss": 0.3768,
      "num_input_tokens_seen": 12098352,
      "step": 18550
    },
    {
      "epoch": 10.940448113207546,
      "grad_norm": 3.2892050743103027,
      "learning_rate": 5.052482404071461e-06,
      "loss": 0.4303,
      "num_input_tokens_seen": 12101840,
      "step": 18555
    },
    {
      "epoch": 10.943396226415095,
      "grad_norm": 3.4126620292663574,
      "learning_rate": 5.049909824857564e-06,
      "loss": 0.4604,
      "num_input_tokens_seen": 12106256,
      "step": 18560
    },
    {
      "epoch": 10.946344339622641,
      "grad_norm": 2.6609344482421875,
      "learning_rate": 5.047337232429827e-06,
      "loss": 0.4776,
      "num_input_tokens_seen": 12108528,
      "step": 18565
    },
    {
      "epoch": 10.94929245283019,
      "grad_norm": 2.55163836479187,
      "learning_rate": 5.044764627469354e-06,
      "loss": 0.3411,
      "num_input_tokens_seen": 12111920,
      "step": 18570
    },
    {
      "epoch": 10.952240566037736,
      "grad_norm": 2.01534366607666,
      "learning_rate": 5.042192010657251e-06,
      "loss": 0.3128,
      "num_input_tokens_seen": 12115024,
      "step": 18575
    },
    {
      "epoch": 10.955188679245284,
      "grad_norm": 2.3214597702026367,
      "learning_rate": 5.039619382674632e-06,
      "loss": 0.3344,
      "num_input_tokens_seen": 12118096,
      "step": 18580
    },
    {
      "epoch": 10.95813679245283,
      "grad_norm": 3.8189620971679688,
      "learning_rate": 5.0370467442026115e-06,
      "loss": 0.3442,
      "num_input_tokens_seen": 12122224,
      "step": 18585
    },
    {
      "epoch": 10.961084905660378,
      "grad_norm": 2.1392595767974854,
      "learning_rate": 5.034474095922304e-06,
      "loss": 0.3679,
      "num_input_tokens_seen": 12125552,
      "step": 18590
    },
    {
      "epoch": 10.964033018867925,
      "grad_norm": 3.241791009902954,
      "learning_rate": 5.031901438514832e-06,
      "loss": 0.2092,
      "num_input_tokens_seen": 12129232,
      "step": 18595
    },
    {
      "epoch": 10.966981132075471,
      "grad_norm": 2.5336267948150635,
      "learning_rate": 5.0293287726613185e-06,
      "loss": 0.4918,
      "num_input_tokens_seen": 12132176,
      "step": 18600
    },
    {
      "epoch": 10.96992924528302,
      "grad_norm": 2.9993631839752197,
      "learning_rate": 5.0267560990428836e-06,
      "loss": 0.4941,
      "num_input_tokens_seen": 12135600,
      "step": 18605
    },
    {
      "epoch": 10.972877358490566,
      "grad_norm": 4.378379821777344,
      "learning_rate": 5.024183418340657e-06,
      "loss": 0.3432,
      "num_input_tokens_seen": 12139088,
      "step": 18610
    },
    {
      "epoch": 10.975825471698114,
      "grad_norm": 2.6746349334716797,
      "learning_rate": 5.021610731235766e-06,
      "loss": 0.2516,
      "num_input_tokens_seen": 12143024,
      "step": 18615
    },
    {
      "epoch": 10.97877358490566,
      "grad_norm": 3.42079758644104,
      "learning_rate": 5.01903803840934e-06,
      "loss": 0.3362,
      "num_input_tokens_seen": 12146640,
      "step": 18620
    },
    {
      "epoch": 10.981721698113208,
      "grad_norm": 2.9208247661590576,
      "learning_rate": 5.016465340542514e-06,
      "loss": 0.3411,
      "num_input_tokens_seen": 12150704,
      "step": 18625
    },
    {
      "epoch": 10.984669811320755,
      "grad_norm": 3.3420186042785645,
      "learning_rate": 5.013892638316417e-06,
      "loss": 0.3694,
      "num_input_tokens_seen": 12155920,
      "step": 18630
    },
    {
      "epoch": 10.987617924528301,
      "grad_norm": 2.7794246673583984,
      "learning_rate": 5.011319932412182e-06,
      "loss": 0.377,
      "num_input_tokens_seen": 12158352,
      "step": 18635
    },
    {
      "epoch": 10.99056603773585,
      "grad_norm": 1.4129530191421509,
      "learning_rate": 5.008747223510947e-06,
      "loss": 0.2736,
      "num_input_tokens_seen": 12162512,
      "step": 18640
    },
    {
      "epoch": 10.993514150943396,
      "grad_norm": 7.344161510467529,
      "learning_rate": 5.006174512293849e-06,
      "loss": 0.3488,
      "num_input_tokens_seen": 12165296,
      "step": 18645
    },
    {
      "epoch": 10.996462264150944,
      "grad_norm": 5.310723781585693,
      "learning_rate": 5.003601799442019e-06,
      "loss": 0.3677,
      "num_input_tokens_seen": 12168720,
      "step": 18650
    },
    {
      "epoch": 10.99941037735849,
      "grad_norm": 2.111420154571533,
      "learning_rate": 5.0010290856366e-06,
      "loss": 0.4537,
      "num_input_tokens_seen": 12171856,
      "step": 18655
    },
    {
      "epoch": 11.002358490566039,
      "grad_norm": 2.004443645477295,
      "learning_rate": 4.998456371558726e-06,
      "loss": 0.2664,
      "num_input_tokens_seen": 12174800,
      "step": 18660
    },
    {
      "epoch": 11.005306603773585,
      "grad_norm": 2.8950324058532715,
      "learning_rate": 4.995883657889531e-06,
      "loss": 0.4325,
      "num_input_tokens_seen": 12178544,
      "step": 18665
    },
    {
      "epoch": 11.008254716981131,
      "grad_norm": 2.3139302730560303,
      "learning_rate": 4.993310945310158e-06,
      "loss": 0.3309,
      "num_input_tokens_seen": 12182192,
      "step": 18670
    },
    {
      "epoch": 11.01120283018868,
      "grad_norm": 1.9467076063156128,
      "learning_rate": 4.9907382345017416e-06,
      "loss": 0.3315,
      "num_input_tokens_seen": 12185552,
      "step": 18675
    },
    {
      "epoch": 11.014150943396226,
      "grad_norm": 2.683283805847168,
      "learning_rate": 4.988165526145416e-06,
      "loss": 0.4825,
      "num_input_tokens_seen": 12188176,
      "step": 18680
    },
    {
      "epoch": 11.017099056603774,
      "grad_norm": 2.354342460632324,
      "learning_rate": 4.985592820922319e-06,
      "loss": 0.29,
      "num_input_tokens_seen": 12191152,
      "step": 18685
    },
    {
      "epoch": 11.02004716981132,
      "grad_norm": 5.716548442840576,
      "learning_rate": 4.983020119513586e-06,
      "loss": 0.3524,
      "num_input_tokens_seen": 12194448,
      "step": 18690
    },
    {
      "epoch": 11.022995283018869,
      "grad_norm": 2.2911159992218018,
      "learning_rate": 4.9804474226003465e-06,
      "loss": 0.3198,
      "num_input_tokens_seen": 12197776,
      "step": 18695
    },
    {
      "epoch": 11.025943396226415,
      "grad_norm": 3.390439748764038,
      "learning_rate": 4.977874730863739e-06,
      "loss": 0.342,
      "num_input_tokens_seen": 12201424,
      "step": 18700
    },
    {
      "epoch": 11.028891509433961,
      "grad_norm": 2.525070905685425,
      "learning_rate": 4.975302044984889e-06,
      "loss": 0.3029,
      "num_input_tokens_seen": 12204016,
      "step": 18705
    },
    {
      "epoch": 11.03183962264151,
      "grad_norm": 2.57283091545105,
      "learning_rate": 4.972729365644931e-06,
      "loss": 0.3925,
      "num_input_tokens_seen": 12207120,
      "step": 18710
    },
    {
      "epoch": 11.034787735849056,
      "grad_norm": 2.4361491203308105,
      "learning_rate": 4.97015669352499e-06,
      "loss": 0.5008,
      "num_input_tokens_seen": 12210992,
      "step": 18715
    },
    {
      "epoch": 11.037735849056604,
      "grad_norm": 2.788032293319702,
      "learning_rate": 4.967584029306194e-06,
      "loss": 0.3703,
      "num_input_tokens_seen": 12215184,
      "step": 18720
    },
    {
      "epoch": 11.04068396226415,
      "grad_norm": 2.1505331993103027,
      "learning_rate": 4.965011373669666e-06,
      "loss": 0.3677,
      "num_input_tokens_seen": 12218928,
      "step": 18725
    },
    {
      "epoch": 11.043632075471699,
      "grad_norm": 2.590747356414795,
      "learning_rate": 4.962438727296527e-06,
      "loss": 0.3451,
      "num_input_tokens_seen": 12222032,
      "step": 18730
    },
    {
      "epoch": 11.046580188679245,
      "grad_norm": 3.482480764389038,
      "learning_rate": 4.959866090867897e-06,
      "loss": 0.4006,
      "num_input_tokens_seen": 12225104,
      "step": 18735
    },
    {
      "epoch": 11.049528301886792,
      "grad_norm": 2.307365894317627,
      "learning_rate": 4.957293465064893e-06,
      "loss": 0.3898,
      "num_input_tokens_seen": 12227824,
      "step": 18740
    },
    {
      "epoch": 11.05247641509434,
      "grad_norm": 5.037140369415283,
      "learning_rate": 4.954720850568627e-06,
      "loss": 0.3607,
      "num_input_tokens_seen": 12230448,
      "step": 18745
    },
    {
      "epoch": 11.055424528301886,
      "grad_norm": 4.02696418762207,
      "learning_rate": 4.952148248060212e-06,
      "loss": 0.3053,
      "num_input_tokens_seen": 12233776,
      "step": 18750
    },
    {
      "epoch": 11.058372641509434,
      "grad_norm": 3.7996461391448975,
      "learning_rate": 4.949575658220755e-06,
      "loss": 0.2564,
      "num_input_tokens_seen": 12236240,
      "step": 18755
    },
    {
      "epoch": 11.06132075471698,
      "grad_norm": 4.318322658538818,
      "learning_rate": 4.947003081731359e-06,
      "loss": 0.4753,
      "num_input_tokens_seen": 12239504,
      "step": 18760
    },
    {
      "epoch": 11.064268867924529,
      "grad_norm": 2.4440433979034424,
      "learning_rate": 4.944430519273126e-06,
      "loss": 0.3391,
      "num_input_tokens_seen": 12242416,
      "step": 18765
    },
    {
      "epoch": 11.067216981132075,
      "grad_norm": 3.1278347969055176,
      "learning_rate": 4.941857971527152e-06,
      "loss": 0.4231,
      "num_input_tokens_seen": 12246192,
      "step": 18770
    },
    {
      "epoch": 11.070165094339623,
      "grad_norm": 2.9099042415618896,
      "learning_rate": 4.93928543917453e-06,
      "loss": 0.3938,
      "num_input_tokens_seen": 12250096,
      "step": 18775
    },
    {
      "epoch": 11.07311320754717,
      "grad_norm": 3.319836378097534,
      "learning_rate": 4.93671292289635e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 12254032,
      "step": 18780
    },
    {
      "epoch": 11.076061320754716,
      "grad_norm": 2.8216004371643066,
      "learning_rate": 4.934140423373698e-06,
      "loss": 0.3173,
      "num_input_tokens_seen": 12257424,
      "step": 18785
    },
    {
      "epoch": 11.079009433962264,
      "grad_norm": 2.5357778072357178,
      "learning_rate": 4.931567941287651e-06,
      "loss": 0.263,
      "num_input_tokens_seen": 12260944,
      "step": 18790
    },
    {
      "epoch": 11.08195754716981,
      "grad_norm": 4.195082664489746,
      "learning_rate": 4.9289954773192875e-06,
      "loss": 0.3273,
      "num_input_tokens_seen": 12263056,
      "step": 18795
    },
    {
      "epoch": 11.084905660377359,
      "grad_norm": 4.919189453125,
      "learning_rate": 4.926423032149677e-06,
      "loss": 0.2669,
      "num_input_tokens_seen": 12266480,
      "step": 18800
    },
    {
      "epoch": 11.087853773584905,
      "grad_norm": 2.735436201095581,
      "learning_rate": 4.923850606459883e-06,
      "loss": 0.3957,
      "num_input_tokens_seen": 12269328,
      "step": 18805
    },
    {
      "epoch": 11.090801886792454,
      "grad_norm": 2.6453256607055664,
      "learning_rate": 4.921278200930972e-06,
      "loss": 0.2954,
      "num_input_tokens_seen": 12271408,
      "step": 18810
    },
    {
      "epoch": 11.09375,
      "grad_norm": 3.1545779705047607,
      "learning_rate": 4.918705816243996e-06,
      "loss": 0.4298,
      "num_input_tokens_seen": 12274384,
      "step": 18815
    },
    {
      "epoch": 11.096698113207546,
      "grad_norm": 5.975141525268555,
      "learning_rate": 4.916133453080007e-06,
      "loss": 0.3106,
      "num_input_tokens_seen": 12277648,
      "step": 18820
    },
    {
      "epoch": 11.099646226415095,
      "grad_norm": 2.9268081188201904,
      "learning_rate": 4.913561112120046e-06,
      "loss": 0.4145,
      "num_input_tokens_seen": 12282096,
      "step": 18825
    },
    {
      "epoch": 11.102594339622641,
      "grad_norm": 3.140249729156494,
      "learning_rate": 4.910988794045154e-06,
      "loss": 0.4101,
      "num_input_tokens_seen": 12284912,
      "step": 18830
    },
    {
      "epoch": 11.10554245283019,
      "grad_norm": 1.6862032413482666,
      "learning_rate": 4.9084164995363626e-06,
      "loss": 0.3242,
      "num_input_tokens_seen": 12287824,
      "step": 18835
    },
    {
      "epoch": 11.108490566037736,
      "grad_norm": 3.221494197845459,
      "learning_rate": 4.905844229274697e-06,
      "loss": 0.4541,
      "num_input_tokens_seen": 12291152,
      "step": 18840
    },
    {
      "epoch": 11.111438679245284,
      "grad_norm": 2.7325901985168457,
      "learning_rate": 4.903271983941177e-06,
      "loss": 0.3456,
      "num_input_tokens_seen": 12293904,
      "step": 18845
    },
    {
      "epoch": 11.11438679245283,
      "grad_norm": 3.1134402751922607,
      "learning_rate": 4.900699764216818e-06,
      "loss": 0.3313,
      "num_input_tokens_seen": 12296432,
      "step": 18850
    },
    {
      "epoch": 11.117334905660377,
      "grad_norm": 2.9292571544647217,
      "learning_rate": 4.898127570782622e-06,
      "loss": 0.3298,
      "num_input_tokens_seen": 12299376,
      "step": 18855
    },
    {
      "epoch": 11.120283018867925,
      "grad_norm": 2.5798046588897705,
      "learning_rate": 4.895555404319592e-06,
      "loss": 0.3568,
      "num_input_tokens_seen": 12301712,
      "step": 18860
    },
    {
      "epoch": 11.123231132075471,
      "grad_norm": 7.139198303222656,
      "learning_rate": 4.89298326550872e-06,
      "loss": 0.4002,
      "num_input_tokens_seen": 12304496,
      "step": 18865
    },
    {
      "epoch": 11.12617924528302,
      "grad_norm": 5.531651496887207,
      "learning_rate": 4.8904111550309876e-06,
      "loss": 0.4819,
      "num_input_tokens_seen": 12307344,
      "step": 18870
    },
    {
      "epoch": 11.129127358490566,
      "grad_norm": 3.0349009037017822,
      "learning_rate": 4.8878390735673755e-06,
      "loss": 0.3284,
      "num_input_tokens_seen": 12310576,
      "step": 18875
    },
    {
      "epoch": 11.132075471698114,
      "grad_norm": 6.8454742431640625,
      "learning_rate": 4.8852670217988505e-06,
      "loss": 0.324,
      "num_input_tokens_seen": 12315120,
      "step": 18880
    },
    {
      "epoch": 11.13502358490566,
      "grad_norm": 1.9079841375350952,
      "learning_rate": 4.882695000406377e-06,
      "loss": 0.3037,
      "num_input_tokens_seen": 12318320,
      "step": 18885
    },
    {
      "epoch": 11.137971698113208,
      "grad_norm": 2.1093997955322266,
      "learning_rate": 4.880123010070909e-06,
      "loss": 0.344,
      "num_input_tokens_seen": 12321328,
      "step": 18890
    },
    {
      "epoch": 11.140919811320755,
      "grad_norm": 5.495284557342529,
      "learning_rate": 4.877551051473388e-06,
      "loss": 0.2915,
      "num_input_tokens_seen": 12324432,
      "step": 18895
    },
    {
      "epoch": 11.143867924528301,
      "grad_norm": 2.76992130279541,
      "learning_rate": 4.874979125294755e-06,
      "loss": 0.4132,
      "num_input_tokens_seen": 12327632,
      "step": 18900
    },
    {
      "epoch": 11.14681603773585,
      "grad_norm": 3.3313262462615967,
      "learning_rate": 4.872407232215937e-06,
      "loss": 0.4083,
      "num_input_tokens_seen": 12330544,
      "step": 18905
    },
    {
      "epoch": 11.149764150943396,
      "grad_norm": 2.287444591522217,
      "learning_rate": 4.8698353729178546e-06,
      "loss": 0.3859,
      "num_input_tokens_seen": 12333968,
      "step": 18910
    },
    {
      "epoch": 11.152712264150944,
      "grad_norm": 3.676532030105591,
      "learning_rate": 4.867263548081418e-06,
      "loss": 0.3358,
      "num_input_tokens_seen": 12337104,
      "step": 18915
    },
    {
      "epoch": 11.15566037735849,
      "grad_norm": 2.8962314128875732,
      "learning_rate": 4.8646917583875304e-06,
      "loss": 0.236,
      "num_input_tokens_seen": 12340656,
      "step": 18920
    },
    {
      "epoch": 11.158608490566039,
      "grad_norm": 2.3854353427886963,
      "learning_rate": 4.862120004517082e-06,
      "loss": 0.4384,
      "num_input_tokens_seen": 12344304,
      "step": 18925
    },
    {
      "epoch": 11.161556603773585,
      "grad_norm": 2.607245683670044,
      "learning_rate": 4.859548287150956e-06,
      "loss": 0.294,
      "num_input_tokens_seen": 12350800,
      "step": 18930
    },
    {
      "epoch": 11.164504716981131,
      "grad_norm": 5.347283840179443,
      "learning_rate": 4.8569766069700275e-06,
      "loss": 0.3419,
      "num_input_tokens_seen": 12354096,
      "step": 18935
    },
    {
      "epoch": 11.16745283018868,
      "grad_norm": 1.8244376182556152,
      "learning_rate": 4.854404964655158e-06,
      "loss": 0.3323,
      "num_input_tokens_seen": 12357520,
      "step": 18940
    },
    {
      "epoch": 11.170400943396226,
      "grad_norm": 2.336599826812744,
      "learning_rate": 4.8518333608872015e-06,
      "loss": 0.2989,
      "num_input_tokens_seen": 12360592,
      "step": 18945
    },
    {
      "epoch": 11.173349056603774,
      "grad_norm": 2.4436724185943604,
      "learning_rate": 4.849261796347002e-06,
      "loss": 0.4278,
      "num_input_tokens_seen": 12363120,
      "step": 18950
    },
    {
      "epoch": 11.17629716981132,
      "grad_norm": 2.5173592567443848,
      "learning_rate": 4.846690271715391e-06,
      "loss": 0.3995,
      "num_input_tokens_seen": 12366896,
      "step": 18955
    },
    {
      "epoch": 11.179245283018869,
      "grad_norm": 1.7357096672058105,
      "learning_rate": 4.844118787673191e-06,
      "loss": 0.4895,
      "num_input_tokens_seen": 12370096,
      "step": 18960
    },
    {
      "epoch": 11.182193396226415,
      "grad_norm": 1.5885047912597656,
      "learning_rate": 4.841547344901214e-06,
      "loss": 0.2992,
      "num_input_tokens_seen": 12373808,
      "step": 18965
    },
    {
      "epoch": 11.185141509433961,
      "grad_norm": 1.6408543586730957,
      "learning_rate": 4.838975944080261e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 12378064,
      "step": 18970
    },
    {
      "epoch": 11.18808962264151,
      "grad_norm": 2.300551414489746,
      "learning_rate": 4.83640458589112e-06,
      "loss": 0.3382,
      "num_input_tokens_seen": 12381744,
      "step": 18975
    },
    {
      "epoch": 11.191037735849056,
      "grad_norm": 2.3883817195892334,
      "learning_rate": 4.833833271014571e-06,
      "loss": 0.3703,
      "num_input_tokens_seen": 12384656,
      "step": 18980
    },
    {
      "epoch": 11.193985849056604,
      "grad_norm": 2.3296713829040527,
      "learning_rate": 4.831262000131379e-06,
      "loss": 0.3804,
      "num_input_tokens_seen": 12388336,
      "step": 18985
    },
    {
      "epoch": 11.19693396226415,
      "grad_norm": 2.4810731410980225,
      "learning_rate": 4.828690773922299e-06,
      "loss": 0.2396,
      "num_input_tokens_seen": 12390704,
      "step": 18990
    },
    {
      "epoch": 11.199882075471699,
      "grad_norm": 2.139838457107544,
      "learning_rate": 4.826119593068074e-06,
      "loss": 0.238,
      "num_input_tokens_seen": 12395024,
      "step": 18995
    },
    {
      "epoch": 11.202830188679245,
      "grad_norm": 5.266247749328613,
      "learning_rate": 4.8235484582494375e-06,
      "loss": 0.3546,
      "num_input_tokens_seen": 12398192,
      "step": 19000
    },
    {
      "epoch": 11.205778301886792,
      "grad_norm": 3.186474561691284,
      "learning_rate": 4.8209773701471076e-06,
      "loss": 0.6617,
      "num_input_tokens_seen": 12401392,
      "step": 19005
    },
    {
      "epoch": 11.20872641509434,
      "grad_norm": 3.941420555114746,
      "learning_rate": 4.818406329441789e-06,
      "loss": 0.5045,
      "num_input_tokens_seen": 12404912,
      "step": 19010
    },
    {
      "epoch": 11.211674528301886,
      "grad_norm": 3.098647356033325,
      "learning_rate": 4.815835336814179e-06,
      "loss": 0.4558,
      "num_input_tokens_seen": 12411408,
      "step": 19015
    },
    {
      "epoch": 11.214622641509434,
      "grad_norm": 4.1012349128723145,
      "learning_rate": 4.813264392944957e-06,
      "loss": 0.3711,
      "num_input_tokens_seen": 12414480,
      "step": 19020
    },
    {
      "epoch": 11.21757075471698,
      "grad_norm": 2.6398797035217285,
      "learning_rate": 4.8106934985147905e-06,
      "loss": 0.285,
      "num_input_tokens_seen": 12418160,
      "step": 19025
    },
    {
      "epoch": 11.220518867924529,
      "grad_norm": 2.8535752296447754,
      "learning_rate": 4.808122654204338e-06,
      "loss": 0.3588,
      "num_input_tokens_seen": 12421360,
      "step": 19030
    },
    {
      "epoch": 11.223466981132075,
      "grad_norm": 6.178938865661621,
      "learning_rate": 4.805551860694239e-06,
      "loss": 0.4009,
      "num_input_tokens_seen": 12423888,
      "step": 19035
    },
    {
      "epoch": 11.226415094339623,
      "grad_norm": 1.7846553325653076,
      "learning_rate": 4.802981118665122e-06,
      "loss": 0.3855,
      "num_input_tokens_seen": 12428208,
      "step": 19040
    },
    {
      "epoch": 11.22936320754717,
      "grad_norm": 3.182849884033203,
      "learning_rate": 4.800410428797604e-06,
      "loss": 0.4328,
      "num_input_tokens_seen": 12431568,
      "step": 19045
    },
    {
      "epoch": 11.232311320754716,
      "grad_norm": 2.6362509727478027,
      "learning_rate": 4.797839791772286e-06,
      "loss": 0.3515,
      "num_input_tokens_seen": 12434384,
      "step": 19050
    },
    {
      "epoch": 11.235259433962264,
      "grad_norm": 5.0005693435668945,
      "learning_rate": 4.795269208269752e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 12436784,
      "step": 19055
    },
    {
      "epoch": 11.23820754716981,
      "grad_norm": 2.8635447025299072,
      "learning_rate": 4.792698678970579e-06,
      "loss": 0.3189,
      "num_input_tokens_seen": 12440592,
      "step": 19060
    },
    {
      "epoch": 11.241155660377359,
      "grad_norm": 3.8279361724853516,
      "learning_rate": 4.7901282045553245e-06,
      "loss": 0.3805,
      "num_input_tokens_seen": 12443344,
      "step": 19065
    },
    {
      "epoch": 11.244103773584905,
      "grad_norm": 2.2663140296936035,
      "learning_rate": 4.787557785704531e-06,
      "loss": 0.4064,
      "num_input_tokens_seen": 12446736,
      "step": 19070
    },
    {
      "epoch": 11.247051886792454,
      "grad_norm": 2.024879217147827,
      "learning_rate": 4.784987423098731e-06,
      "loss": 0.3028,
      "num_input_tokens_seen": 12450128,
      "step": 19075
    },
    {
      "epoch": 11.25,
      "grad_norm": 3.1784489154815674,
      "learning_rate": 4.7824171174184354e-06,
      "loss": 0.3516,
      "num_input_tokens_seen": 12452976,
      "step": 19080
    },
    {
      "epoch": 11.252948113207546,
      "grad_norm": 4.035409927368164,
      "learning_rate": 4.779846869344146e-06,
      "loss": 0.3235,
      "num_input_tokens_seen": 12455984,
      "step": 19085
    },
    {
      "epoch": 11.255896226415095,
      "grad_norm": 2.4025628566741943,
      "learning_rate": 4.777276679556346e-06,
      "loss": 0.2961,
      "num_input_tokens_seen": 12459088,
      "step": 19090
    },
    {
      "epoch": 11.258844339622641,
      "grad_norm": 1.878843903541565,
      "learning_rate": 4.774706548735507e-06,
      "loss": 0.3704,
      "num_input_tokens_seen": 12461712,
      "step": 19095
    },
    {
      "epoch": 11.26179245283019,
      "grad_norm": 2.100754737854004,
      "learning_rate": 4.77213647756208e-06,
      "loss": 0.3257,
      "num_input_tokens_seen": 12465072,
      "step": 19100
    },
    {
      "epoch": 11.264740566037736,
      "grad_norm": 3.777226448059082,
      "learning_rate": 4.769566466716501e-06,
      "loss": 0.3267,
      "num_input_tokens_seen": 12467984,
      "step": 19105
    },
    {
      "epoch": 11.267688679245284,
      "grad_norm": 2.3994710445404053,
      "learning_rate": 4.766996516879195e-06,
      "loss": 0.5089,
      "num_input_tokens_seen": 12471280,
      "step": 19110
    },
    {
      "epoch": 11.27063679245283,
      "grad_norm": 2.4616613388061523,
      "learning_rate": 4.764426628730564e-06,
      "loss": 0.4872,
      "num_input_tokens_seen": 12474736,
      "step": 19115
    },
    {
      "epoch": 11.273584905660377,
      "grad_norm": 2.025953769683838,
      "learning_rate": 4.761856802950999e-06,
      "loss": 0.2715,
      "num_input_tokens_seen": 12478064,
      "step": 19120
    },
    {
      "epoch": 11.276533018867925,
      "grad_norm": 3.4830243587493896,
      "learning_rate": 4.759287040220872e-06,
      "loss": 0.5084,
      "num_input_tokens_seen": 12480944,
      "step": 19125
    },
    {
      "epoch": 11.279481132075471,
      "grad_norm": 4.849188804626465,
      "learning_rate": 4.756717341220538e-06,
      "loss": 0.2879,
      "num_input_tokens_seen": 12483728,
      "step": 19130
    },
    {
      "epoch": 11.28242924528302,
      "grad_norm": 2.362220048904419,
      "learning_rate": 4.7541477066303365e-06,
      "loss": 0.4629,
      "num_input_tokens_seen": 12486160,
      "step": 19135
    },
    {
      "epoch": 11.285377358490566,
      "grad_norm": 4.428398132324219,
      "learning_rate": 4.75157813713059e-06,
      "loss": 0.3607,
      "num_input_tokens_seen": 12489744,
      "step": 19140
    },
    {
      "epoch": 11.288325471698114,
      "grad_norm": 6.311544895172119,
      "learning_rate": 4.7490086334016e-06,
      "loss": 0.3728,
      "num_input_tokens_seen": 12492304,
      "step": 19145
    },
    {
      "epoch": 11.29127358490566,
      "grad_norm": 5.3923115730285645,
      "learning_rate": 4.746439196123659e-06,
      "loss": 0.319,
      "num_input_tokens_seen": 12496144,
      "step": 19150
    },
    {
      "epoch": 11.294221698113208,
      "grad_norm": 6.6790947914123535,
      "learning_rate": 4.743869825977032e-06,
      "loss": 0.3354,
      "num_input_tokens_seen": 12499152,
      "step": 19155
    },
    {
      "epoch": 11.297169811320755,
      "grad_norm": 4.919562816619873,
      "learning_rate": 4.741300523641972e-06,
      "loss": 0.3769,
      "num_input_tokens_seen": 12501168,
      "step": 19160
    },
    {
      "epoch": 11.300117924528301,
      "grad_norm": 3.749565362930298,
      "learning_rate": 4.738731289798715e-06,
      "loss": 0.3989,
      "num_input_tokens_seen": 12504240,
      "step": 19165
    },
    {
      "epoch": 11.30306603773585,
      "grad_norm": 3.7977521419525146,
      "learning_rate": 4.7361621251274744e-06,
      "loss": 0.4335,
      "num_input_tokens_seen": 12507952,
      "step": 19170
    },
    {
      "epoch": 11.306014150943396,
      "grad_norm": 3.0528500080108643,
      "learning_rate": 4.733593030308446e-06,
      "loss": 0.4033,
      "num_input_tokens_seen": 12511536,
      "step": 19175
    },
    {
      "epoch": 11.308962264150944,
      "grad_norm": 2.1469242572784424,
      "learning_rate": 4.731024006021814e-06,
      "loss": 0.3032,
      "num_input_tokens_seen": 12514256,
      "step": 19180
    },
    {
      "epoch": 11.31191037735849,
      "grad_norm": 2.782423496246338,
      "learning_rate": 4.728455052947732e-06,
      "loss": 0.332,
      "num_input_tokens_seen": 12517936,
      "step": 19185
    },
    {
      "epoch": 11.314858490566039,
      "grad_norm": 5.1573076248168945,
      "learning_rate": 4.725886171766349e-06,
      "loss": 0.3474,
      "num_input_tokens_seen": 12520784,
      "step": 19190
    },
    {
      "epoch": 11.317806603773585,
      "grad_norm": 3.7818949222564697,
      "learning_rate": 4.723317363157781e-06,
      "loss": 0.2432,
      "num_input_tokens_seen": 12524016,
      "step": 19195
    },
    {
      "epoch": 11.320754716981131,
      "grad_norm": 2.8232312202453613,
      "learning_rate": 4.720748627802135e-06,
      "loss": 0.3114,
      "num_input_tokens_seen": 12526960,
      "step": 19200
    },
    {
      "epoch": 11.32370283018868,
      "grad_norm": 2.4666051864624023,
      "learning_rate": 4.718179966379492e-06,
      "loss": 0.3086,
      "num_input_tokens_seen": 12530736,
      "step": 19205
    },
    {
      "epoch": 11.326650943396226,
      "grad_norm": 4.753872871398926,
      "learning_rate": 4.715611379569919e-06,
      "loss": 0.299,
      "num_input_tokens_seen": 12533520,
      "step": 19210
    },
    {
      "epoch": 11.329599056603774,
      "grad_norm": 4.082564353942871,
      "learning_rate": 4.713042868053458e-06,
      "loss": 0.3886,
      "num_input_tokens_seen": 12536656,
      "step": 19215
    },
    {
      "epoch": 11.33254716981132,
      "grad_norm": 2.0688021183013916,
      "learning_rate": 4.7104744325101345e-06,
      "loss": 0.4137,
      "num_input_tokens_seen": 12540112,
      "step": 19220
    },
    {
      "epoch": 11.335495283018869,
      "grad_norm": 4.766937732696533,
      "learning_rate": 4.7079060736199525e-06,
      "loss": 0.2594,
      "num_input_tokens_seen": 12542640,
      "step": 19225
    },
    {
      "epoch": 11.338443396226415,
      "grad_norm": 2.9978954792022705,
      "learning_rate": 4.705337792062897e-06,
      "loss": 0.3725,
      "num_input_tokens_seen": 12545808,
      "step": 19230
    },
    {
      "epoch": 11.341391509433961,
      "grad_norm": 3.5566420555114746,
      "learning_rate": 4.702769588518931e-06,
      "loss": 0.4459,
      "num_input_tokens_seen": 12548528,
      "step": 19235
    },
    {
      "epoch": 11.34433962264151,
      "grad_norm": 2.732017755508423,
      "learning_rate": 4.700201463667996e-06,
      "loss": 0.427,
      "num_input_tokens_seen": 12551792,
      "step": 19240
    },
    {
      "epoch": 11.347287735849056,
      "grad_norm": 4.120290279388428,
      "learning_rate": 4.697633418190017e-06,
      "loss": 0.4721,
      "num_input_tokens_seen": 12555184,
      "step": 19245
    },
    {
      "epoch": 11.350235849056604,
      "grad_norm": 3.1967711448669434,
      "learning_rate": 4.695065452764893e-06,
      "loss": 0.4315,
      "num_input_tokens_seen": 12558160,
      "step": 19250
    },
    {
      "epoch": 11.35318396226415,
      "grad_norm": 3.2372758388519287,
      "learning_rate": 4.692497568072505e-06,
      "loss": 0.333,
      "num_input_tokens_seen": 12562000,
      "step": 19255
    },
    {
      "epoch": 11.356132075471699,
      "grad_norm": 3.3249053955078125,
      "learning_rate": 4.689929764792711e-06,
      "loss": 0.3627,
      "num_input_tokens_seen": 12564720,
      "step": 19260
    },
    {
      "epoch": 11.359080188679245,
      "grad_norm": 2.384399890899658,
      "learning_rate": 4.687362043605349e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 12568208,
      "step": 19265
    },
    {
      "epoch": 11.362028301886792,
      "grad_norm": 2.355854034423828,
      "learning_rate": 4.684794405190231e-06,
      "loss": 0.3423,
      "num_input_tokens_seen": 12571152,
      "step": 19270
    },
    {
      "epoch": 11.36497641509434,
      "grad_norm": 3.4837660789489746,
      "learning_rate": 4.682226850227155e-06,
      "loss": 0.3624,
      "num_input_tokens_seen": 12574896,
      "step": 19275
    },
    {
      "epoch": 11.367924528301886,
      "grad_norm": 2.0178909301757812,
      "learning_rate": 4.6796593793958875e-06,
      "loss": 0.3234,
      "num_input_tokens_seen": 12578384,
      "step": 19280
    },
    {
      "epoch": 11.370872641509434,
      "grad_norm": 3.545686721801758,
      "learning_rate": 4.6770919933761815e-06,
      "loss": 0.337,
      "num_input_tokens_seen": 12581168,
      "step": 19285
    },
    {
      "epoch": 11.37382075471698,
      "grad_norm": 2.779470920562744,
      "learning_rate": 4.6745246928477615e-06,
      "loss": 0.4259,
      "num_input_tokens_seen": 12584272,
      "step": 19290
    },
    {
      "epoch": 11.376768867924529,
      "grad_norm": 3.018712282180786,
      "learning_rate": 4.671957478490332e-06,
      "loss": 0.5189,
      "num_input_tokens_seen": 12587440,
      "step": 19295
    },
    {
      "epoch": 11.379716981132075,
      "grad_norm": 2.346397876739502,
      "learning_rate": 4.669390350983574e-06,
      "loss": 0.6201,
      "num_input_tokens_seen": 12592208,
      "step": 19300
    },
    {
      "epoch": 11.382665094339623,
      "grad_norm": 2.698775053024292,
      "learning_rate": 4.666823311007145e-06,
      "loss": 0.2941,
      "num_input_tokens_seen": 12595504,
      "step": 19305
    },
    {
      "epoch": 11.38561320754717,
      "grad_norm": 4.345825672149658,
      "learning_rate": 4.664256359240681e-06,
      "loss": 0.3889,
      "num_input_tokens_seen": 12597776,
      "step": 19310
    },
    {
      "epoch": 11.388561320754716,
      "grad_norm": 2.9866530895233154,
      "learning_rate": 4.661689496363793e-06,
      "loss": 0.2552,
      "num_input_tokens_seen": 12600368,
      "step": 19315
    },
    {
      "epoch": 11.391509433962264,
      "grad_norm": 1.9587651491165161,
      "learning_rate": 4.659122723056068e-06,
      "loss": 0.32,
      "num_input_tokens_seen": 12603664,
      "step": 19320
    },
    {
      "epoch": 11.39445754716981,
      "grad_norm": 2.97615647315979,
      "learning_rate": 4.656556039997072e-06,
      "loss": 0.3748,
      "num_input_tokens_seen": 12606544,
      "step": 19325
    },
    {
      "epoch": 11.397405660377359,
      "grad_norm": 2.449540138244629,
      "learning_rate": 4.653989447866345e-06,
      "loss": 0.3998,
      "num_input_tokens_seen": 12610160,
      "step": 19330
    },
    {
      "epoch": 11.400353773584905,
      "grad_norm": 3.9174702167510986,
      "learning_rate": 4.651422947343401e-06,
      "loss": 0.29,
      "num_input_tokens_seen": 12613584,
      "step": 19335
    },
    {
      "epoch": 11.403301886792454,
      "grad_norm": 2.467662811279297,
      "learning_rate": 4.6488565391077355e-06,
      "loss": 0.2793,
      "num_input_tokens_seen": 12616560,
      "step": 19340
    },
    {
      "epoch": 11.40625,
      "grad_norm": 3.8434083461761475,
      "learning_rate": 4.646290223838815e-06,
      "loss": 0.3551,
      "num_input_tokens_seen": 12619184,
      "step": 19345
    },
    {
      "epoch": 11.409198113207546,
      "grad_norm": 2.0932297706604004,
      "learning_rate": 4.64372400221608e-06,
      "loss": 0.4112,
      "num_input_tokens_seen": 12622384,
      "step": 19350
    },
    {
      "epoch": 11.412146226415095,
      "grad_norm": 2.6615967750549316,
      "learning_rate": 4.641157874918953e-06,
      "loss": 0.4107,
      "num_input_tokens_seen": 12625296,
      "step": 19355
    },
    {
      "epoch": 11.415094339622641,
      "grad_norm": 2.929427146911621,
      "learning_rate": 4.6385918426268245e-06,
      "loss": 0.3373,
      "num_input_tokens_seen": 12628496,
      "step": 19360
    },
    {
      "epoch": 11.41804245283019,
      "grad_norm": 2.00095272064209,
      "learning_rate": 4.636025906019062e-06,
      "loss": 0.3276,
      "num_input_tokens_seen": 12631824,
      "step": 19365
    },
    {
      "epoch": 11.420990566037736,
      "grad_norm": 5.305765151977539,
      "learning_rate": 4.6334600657750115e-06,
      "loss": 0.4927,
      "num_input_tokens_seen": 12634672,
      "step": 19370
    },
    {
      "epoch": 11.423938679245284,
      "grad_norm": 4.474178791046143,
      "learning_rate": 4.6308943225739855e-06,
      "loss": 0.4817,
      "num_input_tokens_seen": 12637648,
      "step": 19375
    },
    {
      "epoch": 11.42688679245283,
      "grad_norm": 3.0600359439849854,
      "learning_rate": 4.628328677095281e-06,
      "loss": 0.4268,
      "num_input_tokens_seen": 12640752,
      "step": 19380
    },
    {
      "epoch": 11.429834905660377,
      "grad_norm": 5.619620323181152,
      "learning_rate": 4.625763130018159e-06,
      "loss": 0.3703,
      "num_input_tokens_seen": 12643952,
      "step": 19385
    },
    {
      "epoch": 11.432783018867925,
      "grad_norm": 2.64030122756958,
      "learning_rate": 4.6231976820218635e-06,
      "loss": 0.3529,
      "num_input_tokens_seen": 12647216,
      "step": 19390
    },
    {
      "epoch": 11.435731132075471,
      "grad_norm": 3.766113519668579,
      "learning_rate": 4.620632333785605e-06,
      "loss": 0.3568,
      "num_input_tokens_seen": 12650224,
      "step": 19395
    },
    {
      "epoch": 11.43867924528302,
      "grad_norm": 2.5583231449127197,
      "learning_rate": 4.618067085988569e-06,
      "loss": 0.3297,
      "num_input_tokens_seen": 12653488,
      "step": 19400
    },
    {
      "epoch": 11.441627358490566,
      "grad_norm": 3.055851697921753,
      "learning_rate": 4.615501939309921e-06,
      "loss": 0.2986,
      "num_input_tokens_seen": 12657232,
      "step": 19405
    },
    {
      "epoch": 11.444575471698114,
      "grad_norm": 3.5484273433685303,
      "learning_rate": 4.612936894428791e-06,
      "loss": 0.3339,
      "num_input_tokens_seen": 12660400,
      "step": 19410
    },
    {
      "epoch": 11.44752358490566,
      "grad_norm": 3.0802512168884277,
      "learning_rate": 4.610371952024285e-06,
      "loss": 0.5552,
      "num_input_tokens_seen": 12664944,
      "step": 19415
    },
    {
      "epoch": 11.450471698113208,
      "grad_norm": 3.457401990890503,
      "learning_rate": 4.607807112775485e-06,
      "loss": 0.2934,
      "num_input_tokens_seen": 12668272,
      "step": 19420
    },
    {
      "epoch": 11.453419811320755,
      "grad_norm": 2.4113500118255615,
      "learning_rate": 4.605242377361441e-06,
      "loss": 0.3197,
      "num_input_tokens_seen": 12672272,
      "step": 19425
    },
    {
      "epoch": 11.456367924528301,
      "grad_norm": 1.6858993768692017,
      "learning_rate": 4.60267774646118e-06,
      "loss": 0.3842,
      "num_input_tokens_seen": 12675760,
      "step": 19430
    },
    {
      "epoch": 11.45931603773585,
      "grad_norm": 3.6357429027557373,
      "learning_rate": 4.600113220753698e-06,
      "loss": 0.4433,
      "num_input_tokens_seen": 12678896,
      "step": 19435
    },
    {
      "epoch": 11.462264150943396,
      "grad_norm": 3.8535728454589844,
      "learning_rate": 4.597548800917964e-06,
      "loss": 0.387,
      "num_input_tokens_seen": 12682256,
      "step": 19440
    },
    {
      "epoch": 11.465212264150944,
      "grad_norm": 2.2702760696411133,
      "learning_rate": 4.594984487632919e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 12686544,
      "step": 19445
    },
    {
      "epoch": 11.46816037735849,
      "grad_norm": 1.961531162261963,
      "learning_rate": 4.592420281577478e-06,
      "loss": 0.3709,
      "num_input_tokens_seen": 12689520,
      "step": 19450
    },
    {
      "epoch": 11.471108490566039,
      "grad_norm": 2.4612793922424316,
      "learning_rate": 4.589856183430521e-06,
      "loss": 0.3176,
      "num_input_tokens_seen": 12692336,
      "step": 19455
    },
    {
      "epoch": 11.474056603773585,
      "grad_norm": 3.8923072814941406,
      "learning_rate": 4.58729219387091e-06,
      "loss": 0.3825,
      "num_input_tokens_seen": 12696784,
      "step": 19460
    },
    {
      "epoch": 11.477004716981131,
      "grad_norm": 3.540701389312744,
      "learning_rate": 4.584728313577468e-06,
      "loss": 0.2807,
      "num_input_tokens_seen": 12701584,
      "step": 19465
    },
    {
      "epoch": 11.47995283018868,
      "grad_norm": 5.023279190063477,
      "learning_rate": 4.582164543228993e-06,
      "loss": 0.3338,
      "num_input_tokens_seen": 12704624,
      "step": 19470
    },
    {
      "epoch": 11.482900943396226,
      "grad_norm": 2.656906843185425,
      "learning_rate": 4.5796008835042574e-06,
      "loss": 0.3215,
      "num_input_tokens_seen": 12708368,
      "step": 19475
    },
    {
      "epoch": 11.485849056603774,
      "grad_norm": 2.5755786895751953,
      "learning_rate": 4.577037335082e-06,
      "loss": 0.4306,
      "num_input_tokens_seen": 12712304,
      "step": 19480
    },
    {
      "epoch": 11.48879716981132,
      "grad_norm": 4.281582832336426,
      "learning_rate": 4.574473898640933e-06,
      "loss": 0.437,
      "num_input_tokens_seen": 12715600,
      "step": 19485
    },
    {
      "epoch": 11.491745283018869,
      "grad_norm": 2.88093638420105,
      "learning_rate": 4.571910574859732e-06,
      "loss": 0.358,
      "num_input_tokens_seen": 12718192,
      "step": 19490
    },
    {
      "epoch": 11.494693396226415,
      "grad_norm": 2.4662725925445557,
      "learning_rate": 4.5693473644170535e-06,
      "loss": 0.3712,
      "num_input_tokens_seen": 12723344,
      "step": 19495
    },
    {
      "epoch": 11.497641509433961,
      "grad_norm": 4.7870564460754395,
      "learning_rate": 4.566784267991516e-06,
      "loss": 0.395,
      "num_input_tokens_seen": 12726480,
      "step": 19500
    },
    {
      "epoch": 11.50058962264151,
      "grad_norm": 4.4008283615112305,
      "learning_rate": 4.564221286261709e-06,
      "loss": 0.4189,
      "num_input_tokens_seen": 12729936,
      "step": 19505
    },
    {
      "epoch": 11.503537735849056,
      "grad_norm": 4.64227819442749,
      "learning_rate": 4.5616584199061964e-06,
      "loss": 0.4949,
      "num_input_tokens_seen": 12732560,
      "step": 19510
    },
    {
      "epoch": 11.506485849056604,
      "grad_norm": 2.7179620265960693,
      "learning_rate": 4.559095669603506e-06,
      "loss": 0.3673,
      "num_input_tokens_seen": 12736592,
      "step": 19515
    },
    {
      "epoch": 11.50943396226415,
      "grad_norm": 2.6628541946411133,
      "learning_rate": 4.556533036032136e-06,
      "loss": 0.2726,
      "num_input_tokens_seen": 12739824,
      "step": 19520
    },
    {
      "epoch": 11.512382075471699,
      "grad_norm": 3.0179929733276367,
      "learning_rate": 4.553970519870557e-06,
      "loss": 0.3249,
      "num_input_tokens_seen": 12742864,
      "step": 19525
    },
    {
      "epoch": 11.515330188679245,
      "grad_norm": 4.008169174194336,
      "learning_rate": 4.551408121797205e-06,
      "loss": 0.3526,
      "num_input_tokens_seen": 12746480,
      "step": 19530
    },
    {
      "epoch": 11.518278301886792,
      "grad_norm": 2.7741737365722656,
      "learning_rate": 4.548845842490486e-06,
      "loss": 0.3497,
      "num_input_tokens_seen": 12749104,
      "step": 19535
    },
    {
      "epoch": 11.52122641509434,
      "grad_norm": 3.064547061920166,
      "learning_rate": 4.5462836826287745e-06,
      "loss": 0.2583,
      "num_input_tokens_seen": 12753360,
      "step": 19540
    },
    {
      "epoch": 11.524174528301886,
      "grad_norm": 3.454883337020874,
      "learning_rate": 4.543721642890414e-06,
      "loss": 0.3361,
      "num_input_tokens_seen": 12756208,
      "step": 19545
    },
    {
      "epoch": 11.527122641509434,
      "grad_norm": 8.238747596740723,
      "learning_rate": 4.541159723953714e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 12759120,
      "step": 19550
    },
    {
      "epoch": 11.53007075471698,
      "grad_norm": 4.033265113830566,
      "learning_rate": 4.538597926496955e-06,
      "loss": 0.3877,
      "num_input_tokens_seen": 12761968,
      "step": 19555
    },
    {
      "epoch": 11.533018867924529,
      "grad_norm": 3.488154411315918,
      "learning_rate": 4.536036251198384e-06,
      "loss": 0.3654,
      "num_input_tokens_seen": 12765232,
      "step": 19560
    },
    {
      "epoch": 11.535966981132075,
      "grad_norm": 1.8889278173446655,
      "learning_rate": 4.5334746987362124e-06,
      "loss": 0.3726,
      "num_input_tokens_seen": 12768976,
      "step": 19565
    },
    {
      "epoch": 11.538915094339622,
      "grad_norm": 2.3891000747680664,
      "learning_rate": 4.530913269788627e-06,
      "loss": 0.3561,
      "num_input_tokens_seen": 12772816,
      "step": 19570
    },
    {
      "epoch": 11.54186320754717,
      "grad_norm": 2.2001259326934814,
      "learning_rate": 4.528351965033775e-06,
      "loss": 0.4169,
      "num_input_tokens_seen": 12777488,
      "step": 19575
    },
    {
      "epoch": 11.544811320754716,
      "grad_norm": 2.8217122554779053,
      "learning_rate": 4.525790785149774e-06,
      "loss": 0.3557,
      "num_input_tokens_seen": 12781040,
      "step": 19580
    },
    {
      "epoch": 11.547759433962264,
      "grad_norm": 6.920278072357178,
      "learning_rate": 4.523229730814705e-06,
      "loss": 0.4655,
      "num_input_tokens_seen": 12784080,
      "step": 19585
    },
    {
      "epoch": 11.55070754716981,
      "grad_norm": 2.715130090713501,
      "learning_rate": 4.52066880270662e-06,
      "loss": 0.2927,
      "num_input_tokens_seen": 12786736,
      "step": 19590
    },
    {
      "epoch": 11.553655660377359,
      "grad_norm": 2.232081890106201,
      "learning_rate": 4.518108001503536e-06,
      "loss": 0.3374,
      "num_input_tokens_seen": 12790000,
      "step": 19595
    },
    {
      "epoch": 11.556603773584905,
      "grad_norm": 2.4387500286102295,
      "learning_rate": 4.515547327883434e-06,
      "loss": 0.3287,
      "num_input_tokens_seen": 12793456,
      "step": 19600
    },
    {
      "epoch": 11.559551886792454,
      "grad_norm": 2.5698351860046387,
      "learning_rate": 4.512986782524266e-06,
      "loss": 0.3315,
      "num_input_tokens_seen": 12796784,
      "step": 19605
    },
    {
      "epoch": 11.5625,
      "grad_norm": 1.8755861520767212,
      "learning_rate": 4.510426366103946e-06,
      "loss": 0.253,
      "num_input_tokens_seen": 12799632,
      "step": 19610
    },
    {
      "epoch": 11.565448113207546,
      "grad_norm": 3.3739631175994873,
      "learning_rate": 4.5078660793003544e-06,
      "loss": 0.4082,
      "num_input_tokens_seen": 12802736,
      "step": 19615
    },
    {
      "epoch": 11.568396226415095,
      "grad_norm": 3.7542672157287598,
      "learning_rate": 4.505305922791341e-06,
      "loss": 0.3562,
      "num_input_tokens_seen": 12806384,
      "step": 19620
    },
    {
      "epoch": 11.571344339622641,
      "grad_norm": 2.9826900959014893,
      "learning_rate": 4.502745897254716e-06,
      "loss": 0.3332,
      "num_input_tokens_seen": 12809456,
      "step": 19625
    },
    {
      "epoch": 11.57429245283019,
      "grad_norm": 2.2935402393341064,
      "learning_rate": 4.500186003368257e-06,
      "loss": 0.337,
      "num_input_tokens_seen": 12812528,
      "step": 19630
    },
    {
      "epoch": 11.577240566037736,
      "grad_norm": 3.0368728637695312,
      "learning_rate": 4.497626241809709e-06,
      "loss": 0.3819,
      "num_input_tokens_seen": 12816048,
      "step": 19635
    },
    {
      "epoch": 11.580188679245284,
      "grad_norm": 2.782705307006836,
      "learning_rate": 4.4950666132567775e-06,
      "loss": 0.2827,
      "num_input_tokens_seen": 12818992,
      "step": 19640
    },
    {
      "epoch": 11.58313679245283,
      "grad_norm": 4.114466667175293,
      "learning_rate": 4.492507118387136e-06,
      "loss": 0.4781,
      "num_input_tokens_seen": 12822064,
      "step": 19645
    },
    {
      "epoch": 11.586084905660378,
      "grad_norm": 1.5947636365890503,
      "learning_rate": 4.4899477578784235e-06,
      "loss": 0.335,
      "num_input_tokens_seen": 12825904,
      "step": 19650
    },
    {
      "epoch": 11.589033018867925,
      "grad_norm": 2.4638819694519043,
      "learning_rate": 4.487388532408239e-06,
      "loss": 0.2272,
      "num_input_tokens_seen": 12830160,
      "step": 19655
    },
    {
      "epoch": 11.591981132075471,
      "grad_norm": 4.413480281829834,
      "learning_rate": 4.48482944265415e-06,
      "loss": 0.338,
      "num_input_tokens_seen": 12833360,
      "step": 19660
    },
    {
      "epoch": 11.59492924528302,
      "grad_norm": 3.8675954341888428,
      "learning_rate": 4.482270489293685e-06,
      "loss": 0.2952,
      "num_input_tokens_seen": 12836208,
      "step": 19665
    },
    {
      "epoch": 11.597877358490566,
      "grad_norm": 2.7623445987701416,
      "learning_rate": 4.479711673004341e-06,
      "loss": 0.3761,
      "num_input_tokens_seen": 12839984,
      "step": 19670
    },
    {
      "epoch": 11.600825471698114,
      "grad_norm": 2.715082883834839,
      "learning_rate": 4.477152994463575e-06,
      "loss": 0.2245,
      "num_input_tokens_seen": 12843088,
      "step": 19675
    },
    {
      "epoch": 11.60377358490566,
      "grad_norm": 2.952995538711548,
      "learning_rate": 4.474594454348805e-06,
      "loss": 0.3974,
      "num_input_tokens_seen": 12846736,
      "step": 19680
    },
    {
      "epoch": 11.606721698113208,
      "grad_norm": 3.521667242050171,
      "learning_rate": 4.472036053337419e-06,
      "loss": 0.3732,
      "num_input_tokens_seen": 12849200,
      "step": 19685
    },
    {
      "epoch": 11.609669811320755,
      "grad_norm": 2.310652732849121,
      "learning_rate": 4.469477792106761e-06,
      "loss": 0.358,
      "num_input_tokens_seen": 12852080,
      "step": 19690
    },
    {
      "epoch": 11.612617924528301,
      "grad_norm": 2.3622961044311523,
      "learning_rate": 4.466919671334146e-06,
      "loss": 0.3108,
      "num_input_tokens_seen": 12854864,
      "step": 19695
    },
    {
      "epoch": 11.61556603773585,
      "grad_norm": 3.1215553283691406,
      "learning_rate": 4.4643616916968456e-06,
      "loss": 0.3447,
      "num_input_tokens_seen": 12857776,
      "step": 19700
    },
    {
      "epoch": 11.618514150943396,
      "grad_norm": 3.8869869709014893,
      "learning_rate": 4.461803853872095e-06,
      "loss": 0.3862,
      "num_input_tokens_seen": 12861360,
      "step": 19705
    },
    {
      "epoch": 11.621462264150944,
      "grad_norm": 2.6695749759674072,
      "learning_rate": 4.459246158537094e-06,
      "loss": 0.2916,
      "num_input_tokens_seen": 12864976,
      "step": 19710
    },
    {
      "epoch": 11.62441037735849,
      "grad_norm": 2.401380777359009,
      "learning_rate": 4.456688606369003e-06,
      "loss": 0.3191,
      "num_input_tokens_seen": 12867856,
      "step": 19715
    },
    {
      "epoch": 11.627358490566039,
      "grad_norm": 1.7639538049697876,
      "learning_rate": 4.4541311980449446e-06,
      "loss": 0.251,
      "num_input_tokens_seen": 12876464,
      "step": 19720
    },
    {
      "epoch": 11.630306603773585,
      "grad_norm": 4.165428161621094,
      "learning_rate": 4.4515739342420045e-06,
      "loss": 0.3129,
      "num_input_tokens_seen": 12879952,
      "step": 19725
    },
    {
      "epoch": 11.633254716981131,
      "grad_norm": 2.7343051433563232,
      "learning_rate": 4.4490168156372294e-06,
      "loss": 0.4182,
      "num_input_tokens_seen": 12883088,
      "step": 19730
    },
    {
      "epoch": 11.63620283018868,
      "grad_norm": 2.850191354751587,
      "learning_rate": 4.446459842907626e-06,
      "loss": 0.3258,
      "num_input_tokens_seen": 12885840,
      "step": 19735
    },
    {
      "epoch": 11.639150943396226,
      "grad_norm": 2.14508318901062,
      "learning_rate": 4.443903016730165e-06,
      "loss": 0.3486,
      "num_input_tokens_seen": 12888656,
      "step": 19740
    },
    {
      "epoch": 11.642099056603774,
      "grad_norm": 3.1416988372802734,
      "learning_rate": 4.4413463377817775e-06,
      "loss": 0.3017,
      "num_input_tokens_seen": 12892880,
      "step": 19745
    },
    {
      "epoch": 11.64504716981132,
      "grad_norm": 2.0887463092803955,
      "learning_rate": 4.438789806739353e-06,
      "loss": 0.3611,
      "num_input_tokens_seen": 12895792,
      "step": 19750
    },
    {
      "epoch": 11.647995283018869,
      "grad_norm": 4.08836030960083,
      "learning_rate": 4.436233424279746e-06,
      "loss": 0.3311,
      "num_input_tokens_seen": 12898672,
      "step": 19755
    },
    {
      "epoch": 11.650943396226415,
      "grad_norm": 2.666257619857788,
      "learning_rate": 4.433677191079771e-06,
      "loss": 0.4604,
      "num_input_tokens_seen": 12902576,
      "step": 19760
    },
    {
      "epoch": 11.653891509433961,
      "grad_norm": 2.1791186332702637,
      "learning_rate": 4.431121107816201e-06,
      "loss": 0.3461,
      "num_input_tokens_seen": 12905616,
      "step": 19765
    },
    {
      "epoch": 11.65683962264151,
      "grad_norm": 2.8505682945251465,
      "learning_rate": 4.4285651751657676e-06,
      "loss": 0.4085,
      "num_input_tokens_seen": 12908752,
      "step": 19770
    },
    {
      "epoch": 11.659787735849056,
      "grad_norm": 2.407656669616699,
      "learning_rate": 4.4260093938051685e-06,
      "loss": 0.4832,
      "num_input_tokens_seen": 12911824,
      "step": 19775
    },
    {
      "epoch": 11.662735849056604,
      "grad_norm": 3.032538652420044,
      "learning_rate": 4.423453764411056e-06,
      "loss": 0.2572,
      "num_input_tokens_seen": 12914992,
      "step": 19780
    },
    {
      "epoch": 11.66568396226415,
      "grad_norm": 5.86004638671875,
      "learning_rate": 4.4208982876600425e-06,
      "loss": 0.3632,
      "num_input_tokens_seen": 12917744,
      "step": 19785
    },
    {
      "epoch": 11.668632075471699,
      "grad_norm": 2.6655428409576416,
      "learning_rate": 4.418342964228705e-06,
      "loss": 0.4312,
      "num_input_tokens_seen": 12921424,
      "step": 19790
    },
    {
      "epoch": 11.671580188679245,
      "grad_norm": 2.5237255096435547,
      "learning_rate": 4.415787794793574e-06,
      "loss": 0.349,
      "num_input_tokens_seen": 12924336,
      "step": 19795
    },
    {
      "epoch": 11.674528301886792,
      "grad_norm": 2.736051559448242,
      "learning_rate": 4.4132327800311414e-06,
      "loss": 0.3619,
      "num_input_tokens_seen": 12927728,
      "step": 19800
    },
    {
      "epoch": 11.67747641509434,
      "grad_norm": 3.1326725482940674,
      "learning_rate": 4.41067792061786e-06,
      "loss": 0.4579,
      "num_input_tokens_seen": 12932336,
      "step": 19805
    },
    {
      "epoch": 11.680424528301886,
      "grad_norm": 3.5012478828430176,
      "learning_rate": 4.408123217230139e-06,
      "loss": 0.324,
      "num_input_tokens_seen": 12936688,
      "step": 19810
    },
    {
      "epoch": 11.683372641509434,
      "grad_norm": 2.7109692096710205,
      "learning_rate": 4.4055686705443464e-06,
      "loss": 0.3054,
      "num_input_tokens_seen": 12939120,
      "step": 19815
    },
    {
      "epoch": 11.68632075471698,
      "grad_norm": 4.324853420257568,
      "learning_rate": 4.4030142812368106e-06,
      "loss": 0.5332,
      "num_input_tokens_seen": 12941712,
      "step": 19820
    },
    {
      "epoch": 11.689268867924529,
      "grad_norm": 2.4789862632751465,
      "learning_rate": 4.400460049983817e-06,
      "loss": 0.3176,
      "num_input_tokens_seen": 12944560,
      "step": 19825
    },
    {
      "epoch": 11.692216981132075,
      "grad_norm": 4.248828887939453,
      "learning_rate": 4.397905977461608e-06,
      "loss": 0.3162,
      "num_input_tokens_seen": 12947792,
      "step": 19830
    },
    {
      "epoch": 11.695165094339622,
      "grad_norm": 3.450928211212158,
      "learning_rate": 4.395352064346387e-06,
      "loss": 0.381,
      "num_input_tokens_seen": 12951184,
      "step": 19835
    },
    {
      "epoch": 11.69811320754717,
      "grad_norm": 3.4724326133728027,
      "learning_rate": 4.392798311314314e-06,
      "loss": 0.3052,
      "num_input_tokens_seen": 12953872,
      "step": 19840
    },
    {
      "epoch": 11.701061320754716,
      "grad_norm": 5.270394325256348,
      "learning_rate": 4.390244719041502e-06,
      "loss": 0.3202,
      "num_input_tokens_seen": 12956848,
      "step": 19845
    },
    {
      "epoch": 11.704009433962264,
      "grad_norm": 2.672513008117676,
      "learning_rate": 4.387691288204029e-06,
      "loss": 0.3853,
      "num_input_tokens_seen": 12959792,
      "step": 19850
    },
    {
      "epoch": 11.70695754716981,
      "grad_norm": 3.9002599716186523,
      "learning_rate": 4.385138019477928e-06,
      "loss": 0.4453,
      "num_input_tokens_seen": 12963728,
      "step": 19855
    },
    {
      "epoch": 11.709905660377359,
      "grad_norm": 2.542320728302002,
      "learning_rate": 4.3825849135391864e-06,
      "loss": 0.3877,
      "num_input_tokens_seen": 12966928,
      "step": 19860
    },
    {
      "epoch": 11.712853773584905,
      "grad_norm": 2.5021073818206787,
      "learning_rate": 4.380031971063748e-06,
      "loss": 0.3766,
      "num_input_tokens_seen": 12970544,
      "step": 19865
    },
    {
      "epoch": 11.715801886792454,
      "grad_norm": 3.959275245666504,
      "learning_rate": 4.37747919272752e-06,
      "loss": 0.2545,
      "num_input_tokens_seen": 12973904,
      "step": 19870
    },
    {
      "epoch": 11.71875,
      "grad_norm": 3.515876293182373,
      "learning_rate": 4.374926579206357e-06,
      "loss": 0.3655,
      "num_input_tokens_seen": 12977232,
      "step": 19875
    },
    {
      "epoch": 11.721698113207546,
      "grad_norm": 2.9236373901367188,
      "learning_rate": 4.372374131176075e-06,
      "loss": 0.3731,
      "num_input_tokens_seen": 12980016,
      "step": 19880
    },
    {
      "epoch": 11.724646226415095,
      "grad_norm": 3.3351829051971436,
      "learning_rate": 4.369821849312449e-06,
      "loss": 0.3131,
      "num_input_tokens_seen": 12982736,
      "step": 19885
    },
    {
      "epoch": 11.727594339622641,
      "grad_norm": 3.3169569969177246,
      "learning_rate": 4.367269734291203e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 12986480,
      "step": 19890
    },
    {
      "epoch": 11.73054245283019,
      "grad_norm": 2.5217974185943604,
      "learning_rate": 4.364717786788022e-06,
      "loss": 0.3393,
      "num_input_tokens_seen": 12989296,
      "step": 19895
    },
    {
      "epoch": 11.733490566037736,
      "grad_norm": 2.3677427768707275,
      "learning_rate": 4.362166007478545e-06,
      "loss": 0.304,
      "num_input_tokens_seen": 12993008,
      "step": 19900
    },
    {
      "epoch": 11.736438679245284,
      "grad_norm": 3.122501850128174,
      "learning_rate": 4.3596143970383665e-06,
      "loss": 0.3506,
      "num_input_tokens_seen": 12996528,
      "step": 19905
    },
    {
      "epoch": 11.73938679245283,
      "grad_norm": 1.9580241441726685,
      "learning_rate": 4.357062956143035e-06,
      "loss": 0.2346,
      "num_input_tokens_seen": 12999888,
      "step": 19910
    },
    {
      "epoch": 11.742334905660378,
      "grad_norm": 5.113658905029297,
      "learning_rate": 4.354511685468059e-06,
      "loss": 0.3916,
      "num_input_tokens_seen": 13003216,
      "step": 19915
    },
    {
      "epoch": 11.745283018867925,
      "grad_norm": 3.3515193462371826,
      "learning_rate": 4.351960585688894e-06,
      "loss": 0.4338,
      "num_input_tokens_seen": 13006736,
      "step": 19920
    },
    {
      "epoch": 11.748231132075471,
      "grad_norm": 4.886715888977051,
      "learning_rate": 4.349409657480959e-06,
      "loss": 0.3118,
      "num_input_tokens_seen": 13009680,
      "step": 19925
    },
    {
      "epoch": 11.75117924528302,
      "grad_norm": 2.104062795639038,
      "learning_rate": 4.34685890151962e-06,
      "loss": 0.2942,
      "num_input_tokens_seen": 13013040,
      "step": 19930
    },
    {
      "epoch": 11.754127358490566,
      "grad_norm": 2.9694595336914062,
      "learning_rate": 4.344308318480201e-06,
      "loss": 0.3928,
      "num_input_tokens_seen": 13015504,
      "step": 19935
    },
    {
      "epoch": 11.757075471698114,
      "grad_norm": 2.2153384685516357,
      "learning_rate": 4.341757909037981e-06,
      "loss": 0.434,
      "num_input_tokens_seen": 13019152,
      "step": 19940
    },
    {
      "epoch": 11.76002358490566,
      "grad_norm": 5.004147052764893,
      "learning_rate": 4.33920767386819e-06,
      "loss": 0.3215,
      "num_input_tokens_seen": 13022128,
      "step": 19945
    },
    {
      "epoch": 11.762971698113208,
      "grad_norm": 2.9987664222717285,
      "learning_rate": 4.336657613646017e-06,
      "loss": 0.3521,
      "num_input_tokens_seen": 13024784,
      "step": 19950
    },
    {
      "epoch": 11.765919811320755,
      "grad_norm": 1.5151793956756592,
      "learning_rate": 4.3341077290466e-06,
      "loss": 0.2877,
      "num_input_tokens_seen": 13028176,
      "step": 19955
    },
    {
      "epoch": 11.768867924528301,
      "grad_norm": 3.2525863647460938,
      "learning_rate": 4.331558020745031e-06,
      "loss": 0.3249,
      "num_input_tokens_seen": 13032272,
      "step": 19960
    },
    {
      "epoch": 11.77181603773585,
      "grad_norm": 4.664340496063232,
      "learning_rate": 4.329008489416357e-06,
      "loss": 0.3644,
      "num_input_tokens_seen": 13034832,
      "step": 19965
    },
    {
      "epoch": 11.774764150943396,
      "grad_norm": 3.4732789993286133,
      "learning_rate": 4.326459135735576e-06,
      "loss": 0.4821,
      "num_input_tokens_seen": 13037200,
      "step": 19970
    },
    {
      "epoch": 11.777712264150944,
      "grad_norm": 2.2168478965759277,
      "learning_rate": 4.323909960377644e-06,
      "loss": 0.2726,
      "num_input_tokens_seen": 13041680,
      "step": 19975
    },
    {
      "epoch": 11.78066037735849,
      "grad_norm": 2.3234710693359375,
      "learning_rate": 4.3213609640174625e-06,
      "loss": 0.3055,
      "num_input_tokens_seen": 13044528,
      "step": 19980
    },
    {
      "epoch": 11.783608490566039,
      "grad_norm": 2.16900372505188,
      "learning_rate": 4.318812147329889e-06,
      "loss": 0.4243,
      "num_input_tokens_seen": 13047248,
      "step": 19985
    },
    {
      "epoch": 11.786556603773585,
      "grad_norm": 2.0913307666778564,
      "learning_rate": 4.316263510989737e-06,
      "loss": 0.3191,
      "num_input_tokens_seen": 13050928,
      "step": 19990
    },
    {
      "epoch": 11.789504716981131,
      "grad_norm": 5.8368096351623535,
      "learning_rate": 4.313715055671768e-06,
      "loss": 0.4503,
      "num_input_tokens_seen": 13053616,
      "step": 19995
    },
    {
      "epoch": 11.79245283018868,
      "grad_norm": 6.281067371368408,
      "learning_rate": 4.311166782050694e-06,
      "loss": 0.3559,
      "num_input_tokens_seen": 13056400,
      "step": 20000
    },
    {
      "epoch": 11.795400943396226,
      "grad_norm": 2.9095821380615234,
      "learning_rate": 4.308618690801184e-06,
      "loss": 0.3114,
      "num_input_tokens_seen": 13059184,
      "step": 20005
    },
    {
      "epoch": 11.798349056603774,
      "grad_norm": 1.770971655845642,
      "learning_rate": 4.3060707825978564e-06,
      "loss": 0.2989,
      "num_input_tokens_seen": 13062576,
      "step": 20010
    },
    {
      "epoch": 11.80129716981132,
      "grad_norm": 2.262758731842041,
      "learning_rate": 4.303523058115278e-06,
      "loss": 0.2936,
      "num_input_tokens_seen": 13065712,
      "step": 20015
    },
    {
      "epoch": 11.804245283018869,
      "grad_norm": 4.326164245605469,
      "learning_rate": 4.300975518027972e-06,
      "loss": 0.4175,
      "num_input_tokens_seen": 13068848,
      "step": 20020
    },
    {
      "epoch": 11.807193396226415,
      "grad_norm": 5.070953369140625,
      "learning_rate": 4.298428163010411e-06,
      "loss": 0.4302,
      "num_input_tokens_seen": 13071408,
      "step": 20025
    },
    {
      "epoch": 11.810141509433961,
      "grad_norm": 4.121811866760254,
      "learning_rate": 4.295880993737016e-06,
      "loss": 0.3091,
      "num_input_tokens_seen": 13074736,
      "step": 20030
    },
    {
      "epoch": 11.81308962264151,
      "grad_norm": 3.1230826377868652,
      "learning_rate": 4.293334010882164e-06,
      "loss": 0.2936,
      "num_input_tokens_seen": 13077168,
      "step": 20035
    },
    {
      "epoch": 11.816037735849056,
      "grad_norm": 2.900219202041626,
      "learning_rate": 4.290787215120178e-06,
      "loss": 0.3821,
      "num_input_tokens_seen": 13080656,
      "step": 20040
    },
    {
      "epoch": 11.818985849056604,
      "grad_norm": 3.019371747970581,
      "learning_rate": 4.2882406071253315e-06,
      "loss": 0.3946,
      "num_input_tokens_seen": 13084496,
      "step": 20045
    },
    {
      "epoch": 11.82193396226415,
      "grad_norm": 3.298809051513672,
      "learning_rate": 4.285694187571852e-06,
      "loss": 0.326,
      "num_input_tokens_seen": 13086864,
      "step": 20050
    },
    {
      "epoch": 11.824882075471699,
      "grad_norm": 3.491697072982788,
      "learning_rate": 4.283147957133917e-06,
      "loss": 0.3551,
      "num_input_tokens_seen": 13090256,
      "step": 20055
    },
    {
      "epoch": 11.827830188679245,
      "grad_norm": 3.8841865062713623,
      "learning_rate": 4.280601916485648e-06,
      "loss": 0.5957,
      "num_input_tokens_seen": 13094544,
      "step": 20060
    },
    {
      "epoch": 11.830778301886792,
      "grad_norm": 3.2441141605377197,
      "learning_rate": 4.278056066301123e-06,
      "loss": 0.373,
      "num_input_tokens_seen": 13098096,
      "step": 20065
    },
    {
      "epoch": 11.83372641509434,
      "grad_norm": 5.585747241973877,
      "learning_rate": 4.275510407254366e-06,
      "loss": 0.3699,
      "num_input_tokens_seen": 13101648,
      "step": 20070
    },
    {
      "epoch": 11.836674528301886,
      "grad_norm": 5.809953212738037,
      "learning_rate": 4.27296494001935e-06,
      "loss": 0.5472,
      "num_input_tokens_seen": 13103760,
      "step": 20075
    },
    {
      "epoch": 11.839622641509434,
      "grad_norm": 2.4274473190307617,
      "learning_rate": 4.2704196652700005e-06,
      "loss": 0.3275,
      "num_input_tokens_seen": 13106256,
      "step": 20080
    },
    {
      "epoch": 11.84257075471698,
      "grad_norm": 7.530351161956787,
      "learning_rate": 4.2678745836801895e-06,
      "loss": 0.3523,
      "num_input_tokens_seen": 13109072,
      "step": 20085
    },
    {
      "epoch": 11.845518867924529,
      "grad_norm": 2.7343080043792725,
      "learning_rate": 4.265329695923736e-06,
      "loss": 0.3538,
      "num_input_tokens_seen": 13111312,
      "step": 20090
    },
    {
      "epoch": 11.848466981132075,
      "grad_norm": 2.927044153213501,
      "learning_rate": 4.262785002674412e-06,
      "loss": 0.3267,
      "num_input_tokens_seen": 13115408,
      "step": 20095
    },
    {
      "epoch": 11.851415094339622,
      "grad_norm": 3.643876552581787,
      "learning_rate": 4.260240504605938e-06,
      "loss": 0.347,
      "num_input_tokens_seen": 13117968,
      "step": 20100
    },
    {
      "epoch": 11.85436320754717,
      "grad_norm": 2.1152777671813965,
      "learning_rate": 4.2576962023919775e-06,
      "loss": 0.3731,
      "num_input_tokens_seen": 13121680,
      "step": 20105
    },
    {
      "epoch": 11.857311320754716,
      "grad_norm": 2.127545118331909,
      "learning_rate": 4.255152096706145e-06,
      "loss": 0.3094,
      "num_input_tokens_seen": 13124336,
      "step": 20110
    },
    {
      "epoch": 11.860259433962264,
      "grad_norm": 2.6178183555603027,
      "learning_rate": 4.252608188222007e-06,
      "loss": 0.2977,
      "num_input_tokens_seen": 13128240,
      "step": 20115
    },
    {
      "epoch": 11.86320754716981,
      "grad_norm": 4.012906074523926,
      "learning_rate": 4.250064477613071e-06,
      "loss": 0.3837,
      "num_input_tokens_seen": 13130928,
      "step": 20120
    },
    {
      "epoch": 11.866155660377359,
      "grad_norm": 8.452526092529297,
      "learning_rate": 4.247520965552797e-06,
      "loss": 0.3878,
      "num_input_tokens_seen": 13133360,
      "step": 20125
    },
    {
      "epoch": 11.869103773584905,
      "grad_norm": 3.6721739768981934,
      "learning_rate": 4.24497765271459e-06,
      "loss": 0.3664,
      "num_input_tokens_seen": 13135760,
      "step": 20130
    },
    {
      "epoch": 11.872051886792454,
      "grad_norm": 2.9720897674560547,
      "learning_rate": 4.242434539771804e-06,
      "loss": 0.23,
      "num_input_tokens_seen": 13138736,
      "step": 20135
    },
    {
      "epoch": 11.875,
      "grad_norm": 4.060837745666504,
      "learning_rate": 4.239891627397737e-06,
      "loss": 0.3659,
      "num_input_tokens_seen": 13142320,
      "step": 20140
    },
    {
      "epoch": 11.877948113207546,
      "grad_norm": 3.6309335231781006,
      "learning_rate": 4.237348916265638e-06,
      "loss": 0.314,
      "num_input_tokens_seen": 13145424,
      "step": 20145
    },
    {
      "epoch": 11.880896226415095,
      "grad_norm": 5.251545429229736,
      "learning_rate": 4.234806407048702e-06,
      "loss": 0.3222,
      "num_input_tokens_seen": 13148272,
      "step": 20150
    },
    {
      "epoch": 11.883844339622641,
      "grad_norm": 3.9953396320343018,
      "learning_rate": 4.232264100420066e-06,
      "loss": 0.3203,
      "num_input_tokens_seen": 13152944,
      "step": 20155
    },
    {
      "epoch": 11.88679245283019,
      "grad_norm": 2.1548473834991455,
      "learning_rate": 4.229721997052819e-06,
      "loss": 0.3395,
      "num_input_tokens_seen": 13157200,
      "step": 20160
    },
    {
      "epoch": 11.889740566037736,
      "grad_norm": 3.2899868488311768,
      "learning_rate": 4.227180097619993e-06,
      "loss": 0.2976,
      "num_input_tokens_seen": 13160080,
      "step": 20165
    },
    {
      "epoch": 11.892688679245284,
      "grad_norm": 3.432182788848877,
      "learning_rate": 4.224638402794566e-06,
      "loss": 0.3071,
      "num_input_tokens_seen": 13163088,
      "step": 20170
    },
    {
      "epoch": 11.89563679245283,
      "grad_norm": 2.252676010131836,
      "learning_rate": 4.222096913249464e-06,
      "loss": 0.3437,
      "num_input_tokens_seen": 13167440,
      "step": 20175
    },
    {
      "epoch": 11.898584905660378,
      "grad_norm": 6.0276594161987305,
      "learning_rate": 4.219555629657558e-06,
      "loss": 0.3684,
      "num_input_tokens_seen": 13170256,
      "step": 20180
    },
    {
      "epoch": 11.901533018867925,
      "grad_norm": 2.1762845516204834,
      "learning_rate": 4.21701455269166e-06,
      "loss": 0.1981,
      "num_input_tokens_seen": 13175504,
      "step": 20185
    },
    {
      "epoch": 11.904481132075471,
      "grad_norm": 3.138164520263672,
      "learning_rate": 4.2144736830245356e-06,
      "loss": 0.3874,
      "num_input_tokens_seen": 13178768,
      "step": 20190
    },
    {
      "epoch": 11.90742924528302,
      "grad_norm": 2.4305903911590576,
      "learning_rate": 4.211933021328889e-06,
      "loss": 0.2877,
      "num_input_tokens_seen": 13181456,
      "step": 20195
    },
    {
      "epoch": 11.910377358490566,
      "grad_norm": 2.7549068927764893,
      "learning_rate": 4.20939256827737e-06,
      "loss": 0.432,
      "num_input_tokens_seen": 13185008,
      "step": 20200
    },
    {
      "epoch": 11.913325471698114,
      "grad_norm": 3.528595447540283,
      "learning_rate": 4.206852324542578e-06,
      "loss": 0.3265,
      "num_input_tokens_seen": 13188976,
      "step": 20205
    },
    {
      "epoch": 11.91627358490566,
      "grad_norm": 3.036463499069214,
      "learning_rate": 4.2043122907970496e-06,
      "loss": 0.3963,
      "num_input_tokens_seen": 13191856,
      "step": 20210
    },
    {
      "epoch": 11.919221698113208,
      "grad_norm": 2.2963151931762695,
      "learning_rate": 4.2017724677132715e-06,
      "loss": 0.3803,
      "num_input_tokens_seen": 13194800,
      "step": 20215
    },
    {
      "epoch": 11.922169811320755,
      "grad_norm": 2.1888458728790283,
      "learning_rate": 4.1992328559636734e-06,
      "loss": 0.4519,
      "num_input_tokens_seen": 13197328,
      "step": 20220
    },
    {
      "epoch": 11.925117924528301,
      "grad_norm": 2.5733821392059326,
      "learning_rate": 4.196693456220628e-06,
      "loss": 0.3107,
      "num_input_tokens_seen": 13201104,
      "step": 20225
    },
    {
      "epoch": 11.92806603773585,
      "grad_norm": 2.525388479232788,
      "learning_rate": 4.194154269156452e-06,
      "loss": 0.3754,
      "num_input_tokens_seen": 13204624,
      "step": 20230
    },
    {
      "epoch": 11.931014150943396,
      "grad_norm": 3.1193606853485107,
      "learning_rate": 4.191615295443404e-06,
      "loss": 0.4644,
      "num_input_tokens_seen": 13207312,
      "step": 20235
    },
    {
      "epoch": 11.933962264150944,
      "grad_norm": 2.6548690795898438,
      "learning_rate": 4.189076535753692e-06,
      "loss": 0.2378,
      "num_input_tokens_seen": 13211184,
      "step": 20240
    },
    {
      "epoch": 11.93691037735849,
      "grad_norm": 4.048726558685303,
      "learning_rate": 4.186537990759464e-06,
      "loss": 0.3686,
      "num_input_tokens_seen": 13214064,
      "step": 20245
    },
    {
      "epoch": 11.939858490566039,
      "grad_norm": 16.716733932495117,
      "learning_rate": 4.183999661132806e-06,
      "loss": 0.4844,
      "num_input_tokens_seen": 13217136,
      "step": 20250
    },
    {
      "epoch": 11.942806603773585,
      "grad_norm": 4.2716240882873535,
      "learning_rate": 4.181461547545756e-06,
      "loss": 0.2784,
      "num_input_tokens_seen": 13220176,
      "step": 20255
    },
    {
      "epoch": 11.945754716981131,
      "grad_norm": 2.855415105819702,
      "learning_rate": 4.178923650670289e-06,
      "loss": 0.369,
      "num_input_tokens_seen": 13222608,
      "step": 20260
    },
    {
      "epoch": 11.94870283018868,
      "grad_norm": 3.642786741256714,
      "learning_rate": 4.176385971178324e-06,
      "loss": 0.4154,
      "num_input_tokens_seen": 13225904,
      "step": 20265
    },
    {
      "epoch": 11.951650943396226,
      "grad_norm": 2.053516149520874,
      "learning_rate": 4.1738485097417225e-06,
      "loss": 0.376,
      "num_input_tokens_seen": 13229136,
      "step": 20270
    },
    {
      "epoch": 11.954599056603774,
      "grad_norm": 2.9656100273132324,
      "learning_rate": 4.1713112670322886e-06,
      "loss": 0.3338,
      "num_input_tokens_seen": 13231792,
      "step": 20275
    },
    {
      "epoch": 11.95754716981132,
      "grad_norm": 1.9471689462661743,
      "learning_rate": 4.168774243721768e-06,
      "loss": 0.3771,
      "num_input_tokens_seen": 13238160,
      "step": 20280
    },
    {
      "epoch": 11.960495283018869,
      "grad_norm": 6.785888195037842,
      "learning_rate": 4.166237440481849e-06,
      "loss": 0.3378,
      "num_input_tokens_seen": 13242448,
      "step": 20285
    },
    {
      "epoch": 11.963443396226415,
      "grad_norm": 2.9863133430480957,
      "learning_rate": 4.163700857984162e-06,
      "loss": 0.3214,
      "num_input_tokens_seen": 13245648,
      "step": 20290
    },
    {
      "epoch": 11.966391509433961,
      "grad_norm": 2.418766498565674,
      "learning_rate": 4.161164496900275e-06,
      "loss": 0.5059,
      "num_input_tokens_seen": 13248752,
      "step": 20295
    },
    {
      "epoch": 11.96933962264151,
      "grad_norm": 3.253074884414673,
      "learning_rate": 4.1586283579017036e-06,
      "loss": 0.3935,
      "num_input_tokens_seen": 13251536,
      "step": 20300
    },
    {
      "epoch": 11.972287735849056,
      "grad_norm": 2.796755075454712,
      "learning_rate": 4.156092441659901e-06,
      "loss": 0.2989,
      "num_input_tokens_seen": 13254672,
      "step": 20305
    },
    {
      "epoch": 11.975235849056604,
      "grad_norm": 2.7266576290130615,
      "learning_rate": 4.153556748846261e-06,
      "loss": 0.3406,
      "num_input_tokens_seen": 13257680,
      "step": 20310
    },
    {
      "epoch": 11.97818396226415,
      "grad_norm": 5.865846157073975,
      "learning_rate": 4.15102128013212e-06,
      "loss": 0.2483,
      "num_input_tokens_seen": 13260656,
      "step": 20315
    },
    {
      "epoch": 11.981132075471699,
      "grad_norm": 3.521742820739746,
      "learning_rate": 4.1484860361887544e-06,
      "loss": 0.3695,
      "num_input_tokens_seen": 13263888,
      "step": 20320
    },
    {
      "epoch": 11.984080188679245,
      "grad_norm": 4.036994457244873,
      "learning_rate": 4.145951017687379e-06,
      "loss": 0.4148,
      "num_input_tokens_seen": 13266640,
      "step": 20325
    },
    {
      "epoch": 11.987028301886792,
      "grad_norm": 4.236042022705078,
      "learning_rate": 4.1434162252991524e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 13270096,
      "step": 20330
    },
    {
      "epoch": 11.98997641509434,
      "grad_norm": 3.2309505939483643,
      "learning_rate": 4.140881659695173e-06,
      "loss": 0.3419,
      "num_input_tokens_seen": 13273200,
      "step": 20335
    },
    {
      "epoch": 11.992924528301886,
      "grad_norm": 2.9114160537719727,
      "learning_rate": 4.138347321546477e-06,
      "loss": 0.4011,
      "num_input_tokens_seen": 13276400,
      "step": 20340
    },
    {
      "epoch": 11.995872641509434,
      "grad_norm": 2.664261817932129,
      "learning_rate": 4.13581321152404e-06,
      "loss": 0.3915,
      "num_input_tokens_seen": 13278704,
      "step": 20345
    },
    {
      "epoch": 11.99882075471698,
      "grad_norm": 4.473373889923096,
      "learning_rate": 4.133279330298781e-06,
      "loss": 0.4077,
      "num_input_tokens_seen": 13281168,
      "step": 20350
    },
    {
      "epoch": 12.0,
      "eval_loss": 0.5487748980522156,
      "eval_runtime": 19.3819,
      "eval_samples_per_second": 87.504,
      "eval_steps_per_second": 21.876,
      "num_input_tokens_seen": 13281912,
      "step": 20352
    },
    {
      "epoch": 12.001768867924529,
      "grad_norm": 2.6249730587005615,
      "learning_rate": 4.130745678541555e-06,
      "loss": 0.353,
      "num_input_tokens_seen": 13283864,
      "step": 20355
    },
    {
      "epoch": 12.004716981132075,
      "grad_norm": 1.9240474700927734,
      "learning_rate": 4.128212256923155e-06,
      "loss": 0.3585,
      "num_input_tokens_seen": 13288760,
      "step": 20360
    },
    {
      "epoch": 12.007665094339623,
      "grad_norm": 2.769350051879883,
      "learning_rate": 4.125679066114318e-06,
      "loss": 0.3218,
      "num_input_tokens_seen": 13291704,
      "step": 20365
    },
    {
      "epoch": 12.01061320754717,
      "grad_norm": 2.614417552947998,
      "learning_rate": 4.123146106785717e-06,
      "loss": 0.4386,
      "num_input_tokens_seen": 13296344,
      "step": 20370
    },
    {
      "epoch": 12.013561320754716,
      "grad_norm": 5.73816442489624,
      "learning_rate": 4.1206133796079625e-06,
      "loss": 0.3174,
      "num_input_tokens_seen": 13300216,
      "step": 20375
    },
    {
      "epoch": 12.016509433962264,
      "grad_norm": 2.831481456756592,
      "learning_rate": 4.1180808852516065e-06,
      "loss": 0.3052,
      "num_input_tokens_seen": 13303832,
      "step": 20380
    },
    {
      "epoch": 12.01945754716981,
      "grad_norm": 2.5774428844451904,
      "learning_rate": 4.115548624387136e-06,
      "loss": 0.3133,
      "num_input_tokens_seen": 13306424,
      "step": 20385
    },
    {
      "epoch": 12.022405660377359,
      "grad_norm": 3.81384539604187,
      "learning_rate": 4.113016597684979e-06,
      "loss": 0.3835,
      "num_input_tokens_seen": 13309752,
      "step": 20390
    },
    {
      "epoch": 12.025353773584905,
      "grad_norm": 3.5844943523406982,
      "learning_rate": 4.110484805815502e-06,
      "loss": 0.2298,
      "num_input_tokens_seen": 13313528,
      "step": 20395
    },
    {
      "epoch": 12.028301886792454,
      "grad_norm": 4.244783878326416,
      "learning_rate": 4.107953249449005e-06,
      "loss": 0.3466,
      "num_input_tokens_seen": 13317304,
      "step": 20400
    },
    {
      "epoch": 12.03125,
      "grad_norm": 2.51422119140625,
      "learning_rate": 4.105421929255729e-06,
      "loss": 0.3609,
      "num_input_tokens_seen": 13320536,
      "step": 20405
    },
    {
      "epoch": 12.034198113207546,
      "grad_norm": 3.5618269443511963,
      "learning_rate": 4.102890845905854e-06,
      "loss": 0.382,
      "num_input_tokens_seen": 13323896,
      "step": 20410
    },
    {
      "epoch": 12.037146226415095,
      "grad_norm": 3.4842803478240967,
      "learning_rate": 4.1003600000694935e-06,
      "loss": 0.2472,
      "num_input_tokens_seen": 13328152,
      "step": 20415
    },
    {
      "epoch": 12.040094339622641,
      "grad_norm": 2.1029341220855713,
      "learning_rate": 4.0978293924167e-06,
      "loss": 0.2954,
      "num_input_tokens_seen": 13331448,
      "step": 20420
    },
    {
      "epoch": 12.04304245283019,
      "grad_norm": 3.15633225440979,
      "learning_rate": 4.095299023617461e-06,
      "loss": 0.2799,
      "num_input_tokens_seen": 13335736,
      "step": 20425
    },
    {
      "epoch": 12.045990566037736,
      "grad_norm": 1.4970883131027222,
      "learning_rate": 4.092768894341707e-06,
      "loss": 0.3648,
      "num_input_tokens_seen": 13338776,
      "step": 20430
    },
    {
      "epoch": 12.048938679245284,
      "grad_norm": 4.482364654541016,
      "learning_rate": 4.090239005259298e-06,
      "loss": 0.3692,
      "num_input_tokens_seen": 13341656,
      "step": 20435
    },
    {
      "epoch": 12.05188679245283,
      "grad_norm": 3.632020950317383,
      "learning_rate": 4.087709357040033e-06,
      "loss": 0.4255,
      "num_input_tokens_seen": 13344376,
      "step": 20440
    },
    {
      "epoch": 12.054834905660377,
      "grad_norm": 8.917466163635254,
      "learning_rate": 4.085179950353648e-06,
      "loss": 0.415,
      "num_input_tokens_seen": 13349592,
      "step": 20445
    },
    {
      "epoch": 12.057783018867925,
      "grad_norm": 3.203906536102295,
      "learning_rate": 4.0826507858698135e-06,
      "loss": 0.348,
      "num_input_tokens_seen": 13352632,
      "step": 20450
    },
    {
      "epoch": 12.060731132075471,
      "grad_norm": 2.2375380992889404,
      "learning_rate": 4.080121864258136e-06,
      "loss": 0.4618,
      "num_input_tokens_seen": 13356184,
      "step": 20455
    },
    {
      "epoch": 12.06367924528302,
      "grad_norm": 3.8850653171539307,
      "learning_rate": 4.077593186188161e-06,
      "loss": 0.2893,
      "num_input_tokens_seen": 13359448,
      "step": 20460
    },
    {
      "epoch": 12.066627358490566,
      "grad_norm": 3.051764726638794,
      "learning_rate": 4.075064752329364e-06,
      "loss": 0.4575,
      "num_input_tokens_seen": 13362136,
      "step": 20465
    },
    {
      "epoch": 12.069575471698114,
      "grad_norm": 4.132770538330078,
      "learning_rate": 4.0725365633511605e-06,
      "loss": 0.3701,
      "num_input_tokens_seen": 13364952,
      "step": 20470
    },
    {
      "epoch": 12.07252358490566,
      "grad_norm": 3.0670416355133057,
      "learning_rate": 4.070008619922899e-06,
      "loss": 0.4149,
      "num_input_tokens_seen": 13368504,
      "step": 20475
    },
    {
      "epoch": 12.075471698113208,
      "grad_norm": 3.4873197078704834,
      "learning_rate": 4.067480922713864e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 13371704,
      "step": 20480
    },
    {
      "epoch": 12.078419811320755,
      "grad_norm": 5.753536224365234,
      "learning_rate": 4.064953472393273e-06,
      "loss": 0.359,
      "num_input_tokens_seen": 13374904,
      "step": 20485
    },
    {
      "epoch": 12.081367924528301,
      "grad_norm": 1.3289717435836792,
      "learning_rate": 4.06242626963028e-06,
      "loss": 0.3922,
      "num_input_tokens_seen": 13378296,
      "step": 20490
    },
    {
      "epoch": 12.08431603773585,
      "grad_norm": 2.2111828327178955,
      "learning_rate": 4.059899315093972e-06,
      "loss": 0.2894,
      "num_input_tokens_seen": 13380696,
      "step": 20495
    },
    {
      "epoch": 12.087264150943396,
      "grad_norm": 4.29170036315918,
      "learning_rate": 4.057372609453374e-06,
      "loss": 0.2079,
      "num_input_tokens_seen": 13383512,
      "step": 20500
    },
    {
      "epoch": 12.090212264150944,
      "grad_norm": 4.303058624267578,
      "learning_rate": 4.054846153377439e-06,
      "loss": 0.2893,
      "num_input_tokens_seen": 13386072,
      "step": 20505
    },
    {
      "epoch": 12.09316037735849,
      "grad_norm": 2.777421712875366,
      "learning_rate": 4.052319947535058e-06,
      "loss": 0.328,
      "num_input_tokens_seen": 13389496,
      "step": 20510
    },
    {
      "epoch": 12.096108490566039,
      "grad_norm": 2.4721410274505615,
      "learning_rate": 4.049793992595056e-06,
      "loss": 0.2193,
      "num_input_tokens_seen": 13392824,
      "step": 20515
    },
    {
      "epoch": 12.099056603773585,
      "grad_norm": 6.346024036407471,
      "learning_rate": 4.047268289226187e-06,
      "loss": 0.434,
      "num_input_tokens_seen": 13396248,
      "step": 20520
    },
    {
      "epoch": 12.102004716981131,
      "grad_norm": 3.03840970993042,
      "learning_rate": 4.044742838097147e-06,
      "loss": 0.3271,
      "num_input_tokens_seen": 13400696,
      "step": 20525
    },
    {
      "epoch": 12.10495283018868,
      "grad_norm": 3.6480863094329834,
      "learning_rate": 4.0422176398765564e-06,
      "loss": 0.3603,
      "num_input_tokens_seen": 13403416,
      "step": 20530
    },
    {
      "epoch": 12.107900943396226,
      "grad_norm": 2.697359561920166,
      "learning_rate": 4.039692695232975e-06,
      "loss": 0.3258,
      "num_input_tokens_seen": 13406232,
      "step": 20535
    },
    {
      "epoch": 12.110849056603774,
      "grad_norm": 4.425070285797119,
      "learning_rate": 4.03716800483489e-06,
      "loss": 0.4166,
      "num_input_tokens_seen": 13409592,
      "step": 20540
    },
    {
      "epoch": 12.11379716981132,
      "grad_norm": 5.141991138458252,
      "learning_rate": 4.034643569350726e-06,
      "loss": 0.4068,
      "num_input_tokens_seen": 13412600,
      "step": 20545
    },
    {
      "epoch": 12.116745283018869,
      "grad_norm": 1.7947977781295776,
      "learning_rate": 4.032119389448837e-06,
      "loss": 0.3674,
      "num_input_tokens_seen": 13415704,
      "step": 20550
    },
    {
      "epoch": 12.119693396226415,
      "grad_norm": 3.791771173477173,
      "learning_rate": 4.0295954657975115e-06,
      "loss": 0.3078,
      "num_input_tokens_seen": 13418488,
      "step": 20555
    },
    {
      "epoch": 12.122641509433961,
      "grad_norm": 3.4496617317199707,
      "learning_rate": 4.027071799064968e-06,
      "loss": 0.4106,
      "num_input_tokens_seen": 13422328,
      "step": 20560
    },
    {
      "epoch": 12.12558962264151,
      "grad_norm": 3.7593331336975098,
      "learning_rate": 4.02454838991936e-06,
      "loss": 0.4722,
      "num_input_tokens_seen": 13425624,
      "step": 20565
    },
    {
      "epoch": 12.128537735849056,
      "grad_norm": 4.181211471557617,
      "learning_rate": 4.022025239028768e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 13428760,
      "step": 20570
    },
    {
      "epoch": 12.131485849056604,
      "grad_norm": 6.438934326171875,
      "learning_rate": 4.0195023470612095e-06,
      "loss": 0.3318,
      "num_input_tokens_seen": 13431224,
      "step": 20575
    },
    {
      "epoch": 12.13443396226415,
      "grad_norm": 5.056939601898193,
      "learning_rate": 4.016979714684631e-06,
      "loss": 0.3496,
      "num_input_tokens_seen": 13433464,
      "step": 20580
    },
    {
      "epoch": 12.137382075471699,
      "grad_norm": 4.1657819747924805,
      "learning_rate": 4.014457342566909e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 13436280,
      "step": 20585
    },
    {
      "epoch": 12.140330188679245,
      "grad_norm": 3.3484060764312744,
      "learning_rate": 4.011935231375853e-06,
      "loss": 0.2917,
      "num_input_tokens_seen": 13438808,
      "step": 20590
    },
    {
      "epoch": 12.143278301886792,
      "grad_norm": 3.923607349395752,
      "learning_rate": 4.009413381779203e-06,
      "loss": 0.4555,
      "num_input_tokens_seen": 13441592,
      "step": 20595
    },
    {
      "epoch": 12.14622641509434,
      "grad_norm": 2.997964859008789,
      "learning_rate": 4.00689179444463e-06,
      "loss": 0.3184,
      "num_input_tokens_seen": 13444728,
      "step": 20600
    },
    {
      "epoch": 12.149174528301886,
      "grad_norm": 2.206421136856079,
      "learning_rate": 4.004370470039733e-06,
      "loss": 0.335,
      "num_input_tokens_seen": 13448280,
      "step": 20605
    },
    {
      "epoch": 12.152122641509434,
      "grad_norm": 2.3633475303649902,
      "learning_rate": 4.001849409232046e-06,
      "loss": 0.3919,
      "num_input_tokens_seen": 13451256,
      "step": 20610
    },
    {
      "epoch": 12.15507075471698,
      "grad_norm": 2.888762950897217,
      "learning_rate": 3.9993286126890274e-06,
      "loss": 0.2797,
      "num_input_tokens_seen": 13454232,
      "step": 20615
    },
    {
      "epoch": 12.158018867924529,
      "grad_norm": 2.391286849975586,
      "learning_rate": 3.996808081078074e-06,
      "loss": 0.3011,
      "num_input_tokens_seen": 13457464,
      "step": 20620
    },
    {
      "epoch": 12.160966981132075,
      "grad_norm": 3.058185338973999,
      "learning_rate": 3.9942878150665035e-06,
      "loss": 0.2561,
      "num_input_tokens_seen": 13460728,
      "step": 20625
    },
    {
      "epoch": 12.163915094339623,
      "grad_norm": 2.7282187938690186,
      "learning_rate": 3.991767815321569e-06,
      "loss": 0.3655,
      "num_input_tokens_seen": 13464280,
      "step": 20630
    },
    {
      "epoch": 12.16686320754717,
      "grad_norm": 2.987643003463745,
      "learning_rate": 3.9892480825104504e-06,
      "loss": 0.4409,
      "num_input_tokens_seen": 13467640,
      "step": 20635
    },
    {
      "epoch": 12.169811320754716,
      "grad_norm": 5.0638604164123535,
      "learning_rate": 3.986728617300257e-06,
      "loss": 0.3553,
      "num_input_tokens_seen": 13470424,
      "step": 20640
    },
    {
      "epoch": 12.172759433962264,
      "grad_norm": 2.9158055782318115,
      "learning_rate": 3.984209420358031e-06,
      "loss": 0.3627,
      "num_input_tokens_seen": 13473240,
      "step": 20645
    },
    {
      "epoch": 12.17570754716981,
      "grad_norm": 3.2295644283294678,
      "learning_rate": 3.981690492350738e-06,
      "loss": 0.4265,
      "num_input_tokens_seen": 13476984,
      "step": 20650
    },
    {
      "epoch": 12.178655660377359,
      "grad_norm": 3.7602219581604004,
      "learning_rate": 3.979171833945276e-06,
      "loss": 0.3657,
      "num_input_tokens_seen": 13480632,
      "step": 20655
    },
    {
      "epoch": 12.181603773584905,
      "grad_norm": 3.4003725051879883,
      "learning_rate": 3.97665344580847e-06,
      "loss": 0.4206,
      "num_input_tokens_seen": 13484696,
      "step": 20660
    },
    {
      "epoch": 12.184551886792454,
      "grad_norm": 2.5987823009490967,
      "learning_rate": 3.974135328607075e-06,
      "loss": 0.2996,
      "num_input_tokens_seen": 13488056,
      "step": 20665
    },
    {
      "epoch": 12.1875,
      "grad_norm": 3.8917205333709717,
      "learning_rate": 3.971617483007773e-06,
      "loss": 0.3679,
      "num_input_tokens_seen": 13491576,
      "step": 20670
    },
    {
      "epoch": 12.190448113207546,
      "grad_norm": 4.310510158538818,
      "learning_rate": 3.969099909677174e-06,
      "loss": 0.3933,
      "num_input_tokens_seen": 13493816,
      "step": 20675
    },
    {
      "epoch": 12.193396226415095,
      "grad_norm": 4.217745780944824,
      "learning_rate": 3.9665826092818165e-06,
      "loss": 0.2716,
      "num_input_tokens_seen": 13496600,
      "step": 20680
    },
    {
      "epoch": 12.196344339622641,
      "grad_norm": 2.611581802368164,
      "learning_rate": 3.964065582488168e-06,
      "loss": 0.3145,
      "num_input_tokens_seen": 13499608,
      "step": 20685
    },
    {
      "epoch": 12.19929245283019,
      "grad_norm": 4.735732078552246,
      "learning_rate": 3.961548829962622e-06,
      "loss": 0.3863,
      "num_input_tokens_seen": 13502712,
      "step": 20690
    },
    {
      "epoch": 12.202240566037736,
      "grad_norm": 3.4904375076293945,
      "learning_rate": 3.959032352371496e-06,
      "loss": 0.3898,
      "num_input_tokens_seen": 13505848,
      "step": 20695
    },
    {
      "epoch": 12.205188679245284,
      "grad_norm": 2.2928225994110107,
      "learning_rate": 3.956516150381043e-06,
      "loss": 0.3046,
      "num_input_tokens_seen": 13508344,
      "step": 20700
    },
    {
      "epoch": 12.20813679245283,
      "grad_norm": 4.314245223999023,
      "learning_rate": 3.954000224657436e-06,
      "loss": 0.3733,
      "num_input_tokens_seen": 13510840,
      "step": 20705
    },
    {
      "epoch": 12.211084905660377,
      "grad_norm": 2.632591962814331,
      "learning_rate": 3.951484575866776e-06,
      "loss": 0.2855,
      "num_input_tokens_seen": 13514072,
      "step": 20710
    },
    {
      "epoch": 12.214033018867925,
      "grad_norm": 2.3361353874206543,
      "learning_rate": 3.948969204675096e-06,
      "loss": 0.3604,
      "num_input_tokens_seen": 13516856,
      "step": 20715
    },
    {
      "epoch": 12.216981132075471,
      "grad_norm": 2.4873592853546143,
      "learning_rate": 3.946454111748346e-06,
      "loss": 0.3176,
      "num_input_tokens_seen": 13521144,
      "step": 20720
    },
    {
      "epoch": 12.21992924528302,
      "grad_norm": 3.214561700820923,
      "learning_rate": 3.943939297752413e-06,
      "loss": 0.3252,
      "num_input_tokens_seen": 13525336,
      "step": 20725
    },
    {
      "epoch": 12.222877358490566,
      "grad_norm": 6.3854169845581055,
      "learning_rate": 3.9414247633531e-06,
      "loss": 0.4091,
      "num_input_tokens_seen": 13528312,
      "step": 20730
    },
    {
      "epoch": 12.225825471698114,
      "grad_norm": 4.805942535400391,
      "learning_rate": 3.9389105092161454e-06,
      "loss": 0.3953,
      "num_input_tokens_seen": 13530616,
      "step": 20735
    },
    {
      "epoch": 12.22877358490566,
      "grad_norm": 2.2574925422668457,
      "learning_rate": 3.936396536007205e-06,
      "loss": 0.3276,
      "num_input_tokens_seen": 13534328,
      "step": 20740
    },
    {
      "epoch": 12.231721698113208,
      "grad_norm": 2.8402884006500244,
      "learning_rate": 3.933882844391866e-06,
      "loss": 0.3409,
      "num_input_tokens_seen": 13537336,
      "step": 20745
    },
    {
      "epoch": 12.234669811320755,
      "grad_norm": 2.713965892791748,
      "learning_rate": 3.931369435035639e-06,
      "loss": 0.294,
      "num_input_tokens_seen": 13539800,
      "step": 20750
    },
    {
      "epoch": 12.237617924528301,
      "grad_norm": 4.26702356338501,
      "learning_rate": 3.92885630860396e-06,
      "loss": 0.3727,
      "num_input_tokens_seen": 13542680,
      "step": 20755
    },
    {
      "epoch": 12.24056603773585,
      "grad_norm": 4.106425762176514,
      "learning_rate": 3.926343465762189e-06,
      "loss": 0.392,
      "num_input_tokens_seen": 13545080,
      "step": 20760
    },
    {
      "epoch": 12.243514150943396,
      "grad_norm": 1.8896358013153076,
      "learning_rate": 3.923830907175613e-06,
      "loss": 0.3523,
      "num_input_tokens_seen": 13548984,
      "step": 20765
    },
    {
      "epoch": 12.246462264150944,
      "grad_norm": 2.64504337310791,
      "learning_rate": 3.921318633509442e-06,
      "loss": 0.2164,
      "num_input_tokens_seen": 13552312,
      "step": 20770
    },
    {
      "epoch": 12.24941037735849,
      "grad_norm": 2.726611614227295,
      "learning_rate": 3.918806645428811e-06,
      "loss": 0.3877,
      "num_input_tokens_seen": 13556600,
      "step": 20775
    },
    {
      "epoch": 12.252358490566039,
      "grad_norm": 3.683722972869873,
      "learning_rate": 3.916294943598781e-06,
      "loss": 0.4143,
      "num_input_tokens_seen": 13559448,
      "step": 20780
    },
    {
      "epoch": 12.255306603773585,
      "grad_norm": 2.6430466175079346,
      "learning_rate": 3.913783528684336e-06,
      "loss": 0.4822,
      "num_input_tokens_seen": 13562936,
      "step": 20785
    },
    {
      "epoch": 12.258254716981131,
      "grad_norm": 3.876765489578247,
      "learning_rate": 3.911272401350381e-06,
      "loss": 0.3081,
      "num_input_tokens_seen": 13566328,
      "step": 20790
    },
    {
      "epoch": 12.26120283018868,
      "grad_norm": 4.667914867401123,
      "learning_rate": 3.9087615622617525e-06,
      "loss": 0.3079,
      "num_input_tokens_seen": 13569592,
      "step": 20795
    },
    {
      "epoch": 12.264150943396226,
      "grad_norm": 3.3990318775177,
      "learning_rate": 3.906251012083202e-06,
      "loss": 0.3767,
      "num_input_tokens_seen": 13572760,
      "step": 20800
    },
    {
      "epoch": 12.267099056603774,
      "grad_norm": 4.712822437286377,
      "learning_rate": 3.9037407514794085e-06,
      "loss": 0.353,
      "num_input_tokens_seen": 13576184,
      "step": 20805
    },
    {
      "epoch": 12.27004716981132,
      "grad_norm": 3.084927797317505,
      "learning_rate": 3.901230781114976e-06,
      "loss": 0.4948,
      "num_input_tokens_seen": 13579512,
      "step": 20810
    },
    {
      "epoch": 12.272995283018869,
      "grad_norm": 4.119962215423584,
      "learning_rate": 3.898721101654431e-06,
      "loss": 0.3364,
      "num_input_tokens_seen": 13582680,
      "step": 20815
    },
    {
      "epoch": 12.275943396226415,
      "grad_norm": 3.716001272201538,
      "learning_rate": 3.896211713762221e-06,
      "loss": 0.3751,
      "num_input_tokens_seen": 13585976,
      "step": 20820
    },
    {
      "epoch": 12.278891509433961,
      "grad_norm": 4.973601818084717,
      "learning_rate": 3.893702618102715e-06,
      "loss": 0.4929,
      "num_input_tokens_seen": 13588728,
      "step": 20825
    },
    {
      "epoch": 12.28183962264151,
      "grad_norm": 2.5675604343414307,
      "learning_rate": 3.891193815340211e-06,
      "loss": 0.3092,
      "num_input_tokens_seen": 13591512,
      "step": 20830
    },
    {
      "epoch": 12.284787735849056,
      "grad_norm": 2.8471450805664062,
      "learning_rate": 3.888685306138922e-06,
      "loss": 0.4348,
      "num_input_tokens_seen": 13595352,
      "step": 20835
    },
    {
      "epoch": 12.287735849056604,
      "grad_norm": 4.959554195404053,
      "learning_rate": 3.886177091162987e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 13599064,
      "step": 20840
    },
    {
      "epoch": 12.29068396226415,
      "grad_norm": 1.6358941793441772,
      "learning_rate": 3.883669171076468e-06,
      "loss": 0.2732,
      "num_input_tokens_seen": 13602296,
      "step": 20845
    },
    {
      "epoch": 12.293632075471699,
      "grad_norm": 2.0686709880828857,
      "learning_rate": 3.881161546543348e-06,
      "loss": 0.3853,
      "num_input_tokens_seen": 13605720,
      "step": 20850
    },
    {
      "epoch": 12.296580188679245,
      "grad_norm": 9.706639289855957,
      "learning_rate": 3.8786542182275295e-06,
      "loss": 0.4167,
      "num_input_tokens_seen": 13608664,
      "step": 20855
    },
    {
      "epoch": 12.299528301886792,
      "grad_norm": 2.397244930267334,
      "learning_rate": 3.87614718679284e-06,
      "loss": 0.409,
      "num_input_tokens_seen": 13611992,
      "step": 20860
    },
    {
      "epoch": 12.30247641509434,
      "grad_norm": 2.125706672668457,
      "learning_rate": 3.8736404529030255e-06,
      "loss": 0.4036,
      "num_input_tokens_seen": 13616184,
      "step": 20865
    },
    {
      "epoch": 12.305424528301886,
      "grad_norm": 3.891131639480591,
      "learning_rate": 3.871134017221756e-06,
      "loss": 0.3698,
      "num_input_tokens_seen": 13619096,
      "step": 20870
    },
    {
      "epoch": 12.308372641509434,
      "grad_norm": 2.491774559020996,
      "learning_rate": 3.868627880412621e-06,
      "loss": 0.481,
      "num_input_tokens_seen": 13622008,
      "step": 20875
    },
    {
      "epoch": 12.31132075471698,
      "grad_norm": 2.9624111652374268,
      "learning_rate": 3.86612204313913e-06,
      "loss": 0.277,
      "num_input_tokens_seen": 13625976,
      "step": 20880
    },
    {
      "epoch": 12.314268867924529,
      "grad_norm": 2.9760162830352783,
      "learning_rate": 3.863616506064714e-06,
      "loss": 0.3385,
      "num_input_tokens_seen": 13629272,
      "step": 20885
    },
    {
      "epoch": 12.317216981132075,
      "grad_norm": 3.2171952724456787,
      "learning_rate": 3.861111269852727e-06,
      "loss": 0.2594,
      "num_input_tokens_seen": 13632376,
      "step": 20890
    },
    {
      "epoch": 12.320165094339623,
      "grad_norm": 2.1058924198150635,
      "learning_rate": 3.858606335166439e-06,
      "loss": 0.337,
      "num_input_tokens_seen": 13635352,
      "step": 20895
    },
    {
      "epoch": 12.32311320754717,
      "grad_norm": 2.2584102153778076,
      "learning_rate": 3.8561017026690415e-06,
      "loss": 0.3032,
      "num_input_tokens_seen": 13638456,
      "step": 20900
    },
    {
      "epoch": 12.326061320754716,
      "grad_norm": 3.1981778144836426,
      "learning_rate": 3.8535973730236495e-06,
      "loss": 0.3073,
      "num_input_tokens_seen": 13640856,
      "step": 20905
    },
    {
      "epoch": 12.329009433962264,
      "grad_norm": 1.7064399719238281,
      "learning_rate": 3.851093346893294e-06,
      "loss": 0.3552,
      "num_input_tokens_seen": 13644152,
      "step": 20910
    },
    {
      "epoch": 12.33195754716981,
      "grad_norm": 3.0838658809661865,
      "learning_rate": 3.848589624940927e-06,
      "loss": 0.3963,
      "num_input_tokens_seen": 13647320,
      "step": 20915
    },
    {
      "epoch": 12.334905660377359,
      "grad_norm": 4.527851581573486,
      "learning_rate": 3.846086207829417e-06,
      "loss": 0.3738,
      "num_input_tokens_seen": 13650328,
      "step": 20920
    },
    {
      "epoch": 12.337853773584905,
      "grad_norm": 3.5663979053497314,
      "learning_rate": 3.843583096221559e-06,
      "loss": 0.4254,
      "num_input_tokens_seen": 13653688,
      "step": 20925
    },
    {
      "epoch": 12.340801886792454,
      "grad_norm": 3.951063632965088,
      "learning_rate": 3.8410802907800596e-06,
      "loss": 0.4677,
      "num_input_tokens_seen": 13657048,
      "step": 20930
    },
    {
      "epoch": 12.34375,
      "grad_norm": 3.8893046379089355,
      "learning_rate": 3.838577792167546e-06,
      "loss": 0.4233,
      "num_input_tokens_seen": 13660920,
      "step": 20935
    },
    {
      "epoch": 12.346698113207546,
      "grad_norm": 6.41038179397583,
      "learning_rate": 3.836075601046569e-06,
      "loss": 0.395,
      "num_input_tokens_seen": 13663896,
      "step": 20940
    },
    {
      "epoch": 12.349646226415095,
      "grad_norm": 2.8852298259735107,
      "learning_rate": 3.833573718079594e-06,
      "loss": 0.4834,
      "num_input_tokens_seen": 13667192,
      "step": 20945
    },
    {
      "epoch": 12.352594339622641,
      "grad_norm": 3.619802951812744,
      "learning_rate": 3.831072143929002e-06,
      "loss": 0.3511,
      "num_input_tokens_seen": 13669976,
      "step": 20950
    },
    {
      "epoch": 12.35554245283019,
      "grad_norm": 2.214292049407959,
      "learning_rate": 3.828570879257098e-06,
      "loss": 0.4502,
      "num_input_tokens_seen": 13672728,
      "step": 20955
    },
    {
      "epoch": 12.358490566037736,
      "grad_norm": 3.0870726108551025,
      "learning_rate": 3.826069924726102e-06,
      "loss": 0.3107,
      "num_input_tokens_seen": 13676056,
      "step": 20960
    },
    {
      "epoch": 12.361438679245284,
      "grad_norm": 4.07956600189209,
      "learning_rate": 3.823569280998154e-06,
      "loss": 0.4432,
      "num_input_tokens_seen": 13679032,
      "step": 20965
    },
    {
      "epoch": 12.36438679245283,
      "grad_norm": 4.918158531188965,
      "learning_rate": 3.8210689487353095e-06,
      "loss": 0.3114,
      "num_input_tokens_seen": 13682040,
      "step": 20970
    },
    {
      "epoch": 12.367334905660377,
      "grad_norm": 2.5436489582061768,
      "learning_rate": 3.818568928599539e-06,
      "loss": 0.371,
      "num_input_tokens_seen": 13684984,
      "step": 20975
    },
    {
      "epoch": 12.370283018867925,
      "grad_norm": 3.9167094230651855,
      "learning_rate": 3.816069221252739e-06,
      "loss": 0.2964,
      "num_input_tokens_seen": 13687832,
      "step": 20980
    },
    {
      "epoch": 12.373231132075471,
      "grad_norm": 3.0221219062805176,
      "learning_rate": 3.813569827356715e-06,
      "loss": 0.2916,
      "num_input_tokens_seen": 13692472,
      "step": 20985
    },
    {
      "epoch": 12.37617924528302,
      "grad_norm": 3.067626953125,
      "learning_rate": 3.811070747573191e-06,
      "loss": 0.2198,
      "num_input_tokens_seen": 13698392,
      "step": 20990
    },
    {
      "epoch": 12.379127358490566,
      "grad_norm": 2.4305059909820557,
      "learning_rate": 3.8085719825638098e-06,
      "loss": 0.302,
      "num_input_tokens_seen": 13702424,
      "step": 20995
    },
    {
      "epoch": 12.382075471698114,
      "grad_norm": 2.4149930477142334,
      "learning_rate": 3.806073532990132e-06,
      "loss": 0.3146,
      "num_input_tokens_seen": 13705240,
      "step": 21000
    },
    {
      "epoch": 12.38502358490566,
      "grad_norm": 2.159437894821167,
      "learning_rate": 3.8035753995136333e-06,
      "loss": 0.3507,
      "num_input_tokens_seen": 13708664,
      "step": 21005
    },
    {
      "epoch": 12.387971698113208,
      "grad_norm": 3.321687936782837,
      "learning_rate": 3.801077582795702e-06,
      "loss": 0.2927,
      "num_input_tokens_seen": 13711992,
      "step": 21010
    },
    {
      "epoch": 12.390919811320755,
      "grad_norm": 2.6007816791534424,
      "learning_rate": 3.7985800834976494e-06,
      "loss": 0.2707,
      "num_input_tokens_seen": 13714648,
      "step": 21015
    },
    {
      "epoch": 12.393867924528301,
      "grad_norm": 2.081460952758789,
      "learning_rate": 3.7960829022806965e-06,
      "loss": 0.2912,
      "num_input_tokens_seen": 13717880,
      "step": 21020
    },
    {
      "epoch": 12.39681603773585,
      "grad_norm": 5.388912200927734,
      "learning_rate": 3.793586039805984e-06,
      "loss": 0.3051,
      "num_input_tokens_seen": 13720600,
      "step": 21025
    },
    {
      "epoch": 12.399764150943396,
      "grad_norm": 2.776482105255127,
      "learning_rate": 3.791089496734567e-06,
      "loss": 0.3643,
      "num_input_tokens_seen": 13723800,
      "step": 21030
    },
    {
      "epoch": 12.402712264150944,
      "grad_norm": 4.252889156341553,
      "learning_rate": 3.7885932737274163e-06,
      "loss": 0.3376,
      "num_input_tokens_seen": 13727032,
      "step": 21035
    },
    {
      "epoch": 12.40566037735849,
      "grad_norm": 5.593424320220947,
      "learning_rate": 3.7860973714454156e-06,
      "loss": 0.3379,
      "num_input_tokens_seen": 13730040,
      "step": 21040
    },
    {
      "epoch": 12.408608490566039,
      "grad_norm": 2.809781074523926,
      "learning_rate": 3.7836017905493695e-06,
      "loss": 0.3424,
      "num_input_tokens_seen": 13733560,
      "step": 21045
    },
    {
      "epoch": 12.411556603773585,
      "grad_norm": 2.951747179031372,
      "learning_rate": 3.7811065316999908e-06,
      "loss": 0.3554,
      "num_input_tokens_seen": 13737080,
      "step": 21050
    },
    {
      "epoch": 12.414504716981131,
      "grad_norm": 1.792353630065918,
      "learning_rate": 3.7786115955579105e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 13741432,
      "step": 21055
    },
    {
      "epoch": 12.41745283018868,
      "grad_norm": 1.9442509412765503,
      "learning_rate": 3.7761169827836746e-06,
      "loss": 0.3953,
      "num_input_tokens_seen": 13749272,
      "step": 21060
    },
    {
      "epoch": 12.420400943396226,
      "grad_norm": 3.4836626052856445,
      "learning_rate": 3.773622694037743e-06,
      "loss": 0.3446,
      "num_input_tokens_seen": 13752312,
      "step": 21065
    },
    {
      "epoch": 12.423349056603774,
      "grad_norm": 2.4343535900115967,
      "learning_rate": 3.7711287299804865e-06,
      "loss": 0.4225,
      "num_input_tokens_seen": 13756088,
      "step": 21070
    },
    {
      "epoch": 12.42629716981132,
      "grad_norm": 3.199510335922241,
      "learning_rate": 3.768635091272197e-06,
      "loss": 0.3629,
      "num_input_tokens_seen": 13759192,
      "step": 21075
    },
    {
      "epoch": 12.429245283018869,
      "grad_norm": 2.261979579925537,
      "learning_rate": 3.7661417785730732e-06,
      "loss": 0.3697,
      "num_input_tokens_seen": 13762392,
      "step": 21080
    },
    {
      "epoch": 12.432193396226415,
      "grad_norm": 4.748734474182129,
      "learning_rate": 3.7636487925432304e-06,
      "loss": 0.4001,
      "num_input_tokens_seen": 13765560,
      "step": 21085
    },
    {
      "epoch": 12.435141509433961,
      "grad_norm": 2.526278018951416,
      "learning_rate": 3.761156133842697e-06,
      "loss": 0.3054,
      "num_input_tokens_seen": 13768792,
      "step": 21090
    },
    {
      "epoch": 12.43808962264151,
      "grad_norm": 2.734201669692993,
      "learning_rate": 3.7586638031314182e-06,
      "loss": 0.2721,
      "num_input_tokens_seen": 13771544,
      "step": 21095
    },
    {
      "epoch": 12.441037735849056,
      "grad_norm": 2.3232436180114746,
      "learning_rate": 3.7561718010692477e-06,
      "loss": 0.2734,
      "num_input_tokens_seen": 13774264,
      "step": 21100
    },
    {
      "epoch": 12.443985849056604,
      "grad_norm": 4.892324447631836,
      "learning_rate": 3.7536801283159523e-06,
      "loss": 0.45,
      "num_input_tokens_seen": 13777176,
      "step": 21105
    },
    {
      "epoch": 12.44693396226415,
      "grad_norm": 4.02307653427124,
      "learning_rate": 3.7511887855312155e-06,
      "loss": 0.3517,
      "num_input_tokens_seen": 13779640,
      "step": 21110
    },
    {
      "epoch": 12.449882075471699,
      "grad_norm": 3.155975341796875,
      "learning_rate": 3.74869777337463e-06,
      "loss": 0.2688,
      "num_input_tokens_seen": 13782872,
      "step": 21115
    },
    {
      "epoch": 12.452830188679245,
      "grad_norm": 3.7623212337493896,
      "learning_rate": 3.7462070925057004e-06,
      "loss": 0.3517,
      "num_input_tokens_seen": 13787512,
      "step": 21120
    },
    {
      "epoch": 12.455778301886792,
      "grad_norm": 2.7868552207946777,
      "learning_rate": 3.7437167435838472e-06,
      "loss": 0.2963,
      "num_input_tokens_seen": 13790584,
      "step": 21125
    },
    {
      "epoch": 12.45872641509434,
      "grad_norm": 3.0571517944335938,
      "learning_rate": 3.741226727268401e-06,
      "loss": 0.4306,
      "num_input_tokens_seen": 13793976,
      "step": 21130
    },
    {
      "epoch": 12.461674528301886,
      "grad_norm": 5.736512660980225,
      "learning_rate": 3.7387370442186027e-06,
      "loss": 0.3287,
      "num_input_tokens_seen": 13797528,
      "step": 21135
    },
    {
      "epoch": 12.464622641509434,
      "grad_norm": 3.231822967529297,
      "learning_rate": 3.736247695093609e-06,
      "loss": 0.3129,
      "num_input_tokens_seen": 13800696,
      "step": 21140
    },
    {
      "epoch": 12.46757075471698,
      "grad_norm": 2.9698550701141357,
      "learning_rate": 3.7337586805524838e-06,
      "loss": 0.3624,
      "num_input_tokens_seen": 13804312,
      "step": 21145
    },
    {
      "epoch": 12.470518867924529,
      "grad_norm": 4.9800801277160645,
      "learning_rate": 3.731270001254205e-06,
      "loss": 0.3027,
      "num_input_tokens_seen": 13807096,
      "step": 21150
    },
    {
      "epoch": 12.473466981132075,
      "grad_norm": 2.5995781421661377,
      "learning_rate": 3.728781657857661e-06,
      "loss": 0.3137,
      "num_input_tokens_seen": 13810744,
      "step": 21155
    },
    {
      "epoch": 12.476415094339623,
      "grad_norm": 4.284379482269287,
      "learning_rate": 3.726293651021653e-06,
      "loss": 0.3676,
      "num_input_tokens_seen": 13814296,
      "step": 21160
    },
    {
      "epoch": 12.47936320754717,
      "grad_norm": 5.0136895179748535,
      "learning_rate": 3.7238059814048888e-06,
      "loss": 0.2998,
      "num_input_tokens_seen": 13817112,
      "step": 21165
    },
    {
      "epoch": 12.482311320754716,
      "grad_norm": 6.4987006187438965,
      "learning_rate": 3.7213186496659916e-06,
      "loss": 0.4281,
      "num_input_tokens_seen": 13819832,
      "step": 21170
    },
    {
      "epoch": 12.485259433962264,
      "grad_norm": 13.371336936950684,
      "learning_rate": 3.718831656463493e-06,
      "loss": 0.4268,
      "num_input_tokens_seen": 13822712,
      "step": 21175
    },
    {
      "epoch": 12.48820754716981,
      "grad_norm": 3.0891411304473877,
      "learning_rate": 3.7163450024558345e-06,
      "loss": 0.3813,
      "num_input_tokens_seen": 13825624,
      "step": 21180
    },
    {
      "epoch": 12.491155660377359,
      "grad_norm": 2.3158745765686035,
      "learning_rate": 3.7138586883013694e-06,
      "loss": 0.2945,
      "num_input_tokens_seen": 13828440,
      "step": 21185
    },
    {
      "epoch": 12.494103773584905,
      "grad_norm": 3.035343885421753,
      "learning_rate": 3.7113727146583573e-06,
      "loss": 0.325,
      "num_input_tokens_seen": 13831544,
      "step": 21190
    },
    {
      "epoch": 12.497051886792454,
      "grad_norm": 2.359293222427368,
      "learning_rate": 3.7088870821849748e-06,
      "loss": 0.4517,
      "num_input_tokens_seen": 13834520,
      "step": 21195
    },
    {
      "epoch": 12.5,
      "grad_norm": 3.258863925933838,
      "learning_rate": 3.706401791539301e-06,
      "loss": 0.4052,
      "num_input_tokens_seen": 13838040,
      "step": 21200
    },
    {
      "epoch": 12.502948113207546,
      "grad_norm": 4.702422142028809,
      "learning_rate": 3.703916843379328e-06,
      "loss": 0.3706,
      "num_input_tokens_seen": 13841688,
      "step": 21205
    },
    {
      "epoch": 12.505896226415095,
      "grad_norm": 4.060920238494873,
      "learning_rate": 3.7014322383629575e-06,
      "loss": 0.265,
      "num_input_tokens_seen": 13845144,
      "step": 21210
    },
    {
      "epoch": 12.508844339622641,
      "grad_norm": 3.204749822616577,
      "learning_rate": 3.6989479771479976e-06,
      "loss": 0.4326,
      "num_input_tokens_seen": 13847928,
      "step": 21215
    },
    {
      "epoch": 12.51179245283019,
      "grad_norm": 3.049386739730835,
      "learning_rate": 3.696464060392169e-06,
      "loss": 0.3763,
      "num_input_tokens_seen": 13851320,
      "step": 21220
    },
    {
      "epoch": 12.514740566037736,
      "grad_norm": 2.6858530044555664,
      "learning_rate": 3.6939804887530962e-06,
      "loss": 0.2907,
      "num_input_tokens_seen": 13854520,
      "step": 21225
    },
    {
      "epoch": 12.517688679245284,
      "grad_norm": 2.4298501014709473,
      "learning_rate": 3.6914972628883196e-06,
      "loss": 0.4019,
      "num_input_tokens_seen": 13858648,
      "step": 21230
    },
    {
      "epoch": 12.52063679245283,
      "grad_norm": 6.548820495605469,
      "learning_rate": 3.6890143834552814e-06,
      "loss": 0.347,
      "num_input_tokens_seen": 13861880,
      "step": 21235
    },
    {
      "epoch": 12.523584905660378,
      "grad_norm": 4.843287467956543,
      "learning_rate": 3.6865318511113347e-06,
      "loss": 0.3717,
      "num_input_tokens_seen": 13864376,
      "step": 21240
    },
    {
      "epoch": 12.526533018867925,
      "grad_norm": 2.7702043056488037,
      "learning_rate": 3.684049666513742e-06,
      "loss": 0.2905,
      "num_input_tokens_seen": 13867288,
      "step": 21245
    },
    {
      "epoch": 12.529481132075471,
      "grad_norm": 3.4120352268218994,
      "learning_rate": 3.6815678303196715e-06,
      "loss": 0.3394,
      "num_input_tokens_seen": 13870264,
      "step": 21250
    },
    {
      "epoch": 12.53242924528302,
      "grad_norm": 1.9582855701446533,
      "learning_rate": 3.6790863431861988e-06,
      "loss": 0.278,
      "num_input_tokens_seen": 13874360,
      "step": 21255
    },
    {
      "epoch": 12.535377358490566,
      "grad_norm": 3.2264082431793213,
      "learning_rate": 3.676605205770311e-06,
      "loss": 0.4347,
      "num_input_tokens_seen": 13876664,
      "step": 21260
    },
    {
      "epoch": 12.538325471698114,
      "grad_norm": 2.8540754318237305,
      "learning_rate": 3.674124418728898e-06,
      "loss": 0.3338,
      "num_input_tokens_seen": 13880088,
      "step": 21265
    },
    {
      "epoch": 12.54127358490566,
      "grad_norm": 5.460129737854004,
      "learning_rate": 3.671643982718759e-06,
      "loss": 0.3735,
      "num_input_tokens_seen": 13883160,
      "step": 21270
    },
    {
      "epoch": 12.544221698113208,
      "grad_norm": 2.6225204467773438,
      "learning_rate": 3.6691638983966017e-06,
      "loss": 0.3897,
      "num_input_tokens_seen": 13886648,
      "step": 21275
    },
    {
      "epoch": 12.547169811320755,
      "grad_norm": 3.329176664352417,
      "learning_rate": 3.666684166419037e-06,
      "loss": 0.4713,
      "num_input_tokens_seen": 13889656,
      "step": 21280
    },
    {
      "epoch": 12.550117924528301,
      "grad_norm": 2.0760645866394043,
      "learning_rate": 3.6642047874425833e-06,
      "loss": 0.2621,
      "num_input_tokens_seen": 13893144,
      "step": 21285
    },
    {
      "epoch": 12.55306603773585,
      "grad_norm": 3.7049195766448975,
      "learning_rate": 3.661725762123671e-06,
      "loss": 0.3365,
      "num_input_tokens_seen": 13896056,
      "step": 21290
    },
    {
      "epoch": 12.556014150943396,
      "grad_norm": 4.06227445602417,
      "learning_rate": 3.659247091118631e-06,
      "loss": 0.3294,
      "num_input_tokens_seen": 13898360,
      "step": 21295
    },
    {
      "epoch": 12.558962264150944,
      "grad_norm": 2.698856830596924,
      "learning_rate": 3.6567687750837027e-06,
      "loss": 0.2824,
      "num_input_tokens_seen": 13901464,
      "step": 21300
    },
    {
      "epoch": 12.56191037735849,
      "grad_norm": 5.062452793121338,
      "learning_rate": 3.6542908146750287e-06,
      "loss": 0.3537,
      "num_input_tokens_seen": 13904504,
      "step": 21305
    },
    {
      "epoch": 12.564858490566039,
      "grad_norm": 3.246971368789673,
      "learning_rate": 3.6518132105486624e-06,
      "loss": 0.2647,
      "num_input_tokens_seen": 13907896,
      "step": 21310
    },
    {
      "epoch": 12.567806603773585,
      "grad_norm": 3.4587795734405518,
      "learning_rate": 3.649335963360559e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 13911320,
      "step": 21315
    },
    {
      "epoch": 12.570754716981131,
      "grad_norm": 3.8048665523529053,
      "learning_rate": 3.6468590737665795e-06,
      "loss": 0.3035,
      "num_input_tokens_seen": 13915000,
      "step": 21320
    },
    {
      "epoch": 12.57370283018868,
      "grad_norm": 3.344170570373535,
      "learning_rate": 3.6443825424224926e-06,
      "loss": 0.4027,
      "num_input_tokens_seen": 13917560,
      "step": 21325
    },
    {
      "epoch": 12.576650943396226,
      "grad_norm": 3.8680036067962646,
      "learning_rate": 3.64190636998397e-06,
      "loss": 0.2905,
      "num_input_tokens_seen": 13920120,
      "step": 21330
    },
    {
      "epoch": 12.579599056603774,
      "grad_norm": 2.8840997219085693,
      "learning_rate": 3.639430557106588e-06,
      "loss": 0.3219,
      "num_input_tokens_seen": 13923032,
      "step": 21335
    },
    {
      "epoch": 12.58254716981132,
      "grad_norm": 5.633397102355957,
      "learning_rate": 3.6369551044458314e-06,
      "loss": 0.3395,
      "num_input_tokens_seen": 13926840,
      "step": 21340
    },
    {
      "epoch": 12.585495283018869,
      "grad_norm": 6.972689151763916,
      "learning_rate": 3.6344800126570846e-06,
      "loss": 0.4165,
      "num_input_tokens_seen": 13929400,
      "step": 21345
    },
    {
      "epoch": 12.588443396226415,
      "grad_norm": 6.015566825866699,
      "learning_rate": 3.6320052823956385e-06,
      "loss": 0.2435,
      "num_input_tokens_seen": 13932120,
      "step": 21350
    },
    {
      "epoch": 12.591391509433961,
      "grad_norm": 5.883251190185547,
      "learning_rate": 3.6295309143166906e-06,
      "loss": 0.234,
      "num_input_tokens_seen": 13935064,
      "step": 21355
    },
    {
      "epoch": 12.59433962264151,
      "grad_norm": 3.9262642860412598,
      "learning_rate": 3.62705690907534e-06,
      "loss": 0.3314,
      "num_input_tokens_seen": 13938584,
      "step": 21360
    },
    {
      "epoch": 12.597287735849056,
      "grad_norm": 2.9461798667907715,
      "learning_rate": 3.624583267326588e-06,
      "loss": 0.3161,
      "num_input_tokens_seen": 13941656,
      "step": 21365
    },
    {
      "epoch": 12.600235849056604,
      "grad_norm": 4.13564920425415,
      "learning_rate": 3.6221099897253454e-06,
      "loss": 0.3433,
      "num_input_tokens_seen": 13944472,
      "step": 21370
    },
    {
      "epoch": 12.60318396226415,
      "grad_norm": 3.43648099899292,
      "learning_rate": 3.619637076926421e-06,
      "loss": 0.3209,
      "num_input_tokens_seen": 13947512,
      "step": 21375
    },
    {
      "epoch": 12.606132075471699,
      "grad_norm": 5.738428115844727,
      "learning_rate": 3.617164529584528e-06,
      "loss": 0.3816,
      "num_input_tokens_seen": 13950808,
      "step": 21380
    },
    {
      "epoch": 12.609080188679245,
      "grad_norm": 2.4414560794830322,
      "learning_rate": 3.614692348354286e-06,
      "loss": 0.2666,
      "num_input_tokens_seen": 13954072,
      "step": 21385
    },
    {
      "epoch": 12.612028301886792,
      "grad_norm": 2.1865594387054443,
      "learning_rate": 3.612220533890216e-06,
      "loss": 0.3719,
      "num_input_tokens_seen": 13957016,
      "step": 21390
    },
    {
      "epoch": 12.61497641509434,
      "grad_norm": 1.8305792808532715,
      "learning_rate": 3.609749086846741e-06,
      "loss": 0.2846,
      "num_input_tokens_seen": 13960024,
      "step": 21395
    },
    {
      "epoch": 12.617924528301886,
      "grad_norm": 2.4060370922088623,
      "learning_rate": 3.607278007878186e-06,
      "loss": 0.3851,
      "num_input_tokens_seen": 13963320,
      "step": 21400
    },
    {
      "epoch": 12.620872641509434,
      "grad_norm": 3.3054118156433105,
      "learning_rate": 3.6048072976387817e-06,
      "loss": 0.3583,
      "num_input_tokens_seen": 13966424,
      "step": 21405
    },
    {
      "epoch": 12.62382075471698,
      "grad_norm": 3.4472761154174805,
      "learning_rate": 3.6023369567826585e-06,
      "loss": 0.3542,
      "num_input_tokens_seen": 13969688,
      "step": 21410
    },
    {
      "epoch": 12.626768867924529,
      "grad_norm": 4.033164024353027,
      "learning_rate": 3.599866985963848e-06,
      "loss": 0.3137,
      "num_input_tokens_seen": 13975928,
      "step": 21415
    },
    {
      "epoch": 12.629716981132075,
      "grad_norm": 3.188581943511963,
      "learning_rate": 3.5973973858362885e-06,
      "loss": 0.3308,
      "num_input_tokens_seen": 13979128,
      "step": 21420
    },
    {
      "epoch": 12.632665094339622,
      "grad_norm": 3.5570056438446045,
      "learning_rate": 3.594928157053816e-06,
      "loss": 0.3491,
      "num_input_tokens_seen": 13982488,
      "step": 21425
    },
    {
      "epoch": 12.63561320754717,
      "grad_norm": 2.1358227729797363,
      "learning_rate": 3.592459300270168e-06,
      "loss": 0.283,
      "num_input_tokens_seen": 13986232,
      "step": 21430
    },
    {
      "epoch": 12.638561320754716,
      "grad_norm": 3.0476815700531006,
      "learning_rate": 3.589990816138988e-06,
      "loss": 0.3874,
      "num_input_tokens_seen": 13989016,
      "step": 21435
    },
    {
      "epoch": 12.641509433962264,
      "grad_norm": 2.768232583999634,
      "learning_rate": 3.587522705313816e-06,
      "loss": 0.3531,
      "num_input_tokens_seen": 13998008,
      "step": 21440
    },
    {
      "epoch": 12.64445754716981,
      "grad_norm": 3.8550169467926025,
      "learning_rate": 3.585054968448094e-06,
      "loss": 0.3027,
      "num_input_tokens_seen": 14001240,
      "step": 21445
    },
    {
      "epoch": 12.647405660377359,
      "grad_norm": 3.9696741104125977,
      "learning_rate": 3.5825876061951686e-06,
      "loss": 0.4439,
      "num_input_tokens_seen": 14004024,
      "step": 21450
    },
    {
      "epoch": 12.650353773584905,
      "grad_norm": 2.8694238662719727,
      "learning_rate": 3.5801206192082818e-06,
      "loss": 0.2395,
      "num_input_tokens_seen": 14006712,
      "step": 21455
    },
    {
      "epoch": 12.653301886792454,
      "grad_norm": 4.577816486358643,
      "learning_rate": 3.577654008140582e-06,
      "loss": 0.3093,
      "num_input_tokens_seen": 14010648,
      "step": 21460
    },
    {
      "epoch": 12.65625,
      "grad_norm": 2.8944475650787354,
      "learning_rate": 3.5751877736451123e-06,
      "loss": 0.2982,
      "num_input_tokens_seen": 14013080,
      "step": 21465
    },
    {
      "epoch": 12.659198113207546,
      "grad_norm": 3.420577049255371,
      "learning_rate": 3.5727219163748205e-06,
      "loss": 0.3245,
      "num_input_tokens_seen": 14016088,
      "step": 21470
    },
    {
      "epoch": 12.662146226415095,
      "grad_norm": 5.421682357788086,
      "learning_rate": 3.570256436982552e-06,
      "loss": 0.3602,
      "num_input_tokens_seen": 14018392,
      "step": 21475
    },
    {
      "epoch": 12.665094339622641,
      "grad_norm": 2.6409735679626465,
      "learning_rate": 3.5677913361210536e-06,
      "loss": 0.3106,
      "num_input_tokens_seen": 14021560,
      "step": 21480
    },
    {
      "epoch": 12.66804245283019,
      "grad_norm": 4.527918815612793,
      "learning_rate": 3.565326614442972e-06,
      "loss": 0.3472,
      "num_input_tokens_seen": 14025176,
      "step": 21485
    },
    {
      "epoch": 12.670990566037736,
      "grad_norm": 3.998150110244751,
      "learning_rate": 3.5628622726008523e-06,
      "loss": 0.3692,
      "num_input_tokens_seen": 14028408,
      "step": 21490
    },
    {
      "epoch": 12.673938679245284,
      "grad_norm": 4.9621429443359375,
      "learning_rate": 3.56039831124714e-06,
      "loss": 0.4373,
      "num_input_tokens_seen": 14032024,
      "step": 21495
    },
    {
      "epoch": 12.67688679245283,
      "grad_norm": 4.055542469024658,
      "learning_rate": 3.557934731034179e-06,
      "loss": 0.27,
      "num_input_tokens_seen": 14034456,
      "step": 21500
    },
    {
      "epoch": 12.679834905660378,
      "grad_norm": 2.7810192108154297,
      "learning_rate": 3.5554715326142126e-06,
      "loss": 0.352,
      "num_input_tokens_seen": 14038584,
      "step": 21505
    },
    {
      "epoch": 12.682783018867925,
      "grad_norm": 4.445689678192139,
      "learning_rate": 3.553008716639384e-06,
      "loss": 0.2779,
      "num_input_tokens_seen": 14041272,
      "step": 21510
    },
    {
      "epoch": 12.685731132075471,
      "grad_norm": 3.112377405166626,
      "learning_rate": 3.5505462837617338e-06,
      "loss": 0.3029,
      "num_input_tokens_seen": 14045272,
      "step": 21515
    },
    {
      "epoch": 12.68867924528302,
      "grad_norm": 4.456690788269043,
      "learning_rate": 3.5480842346332013e-06,
      "loss": 0.3017,
      "num_input_tokens_seen": 14048280,
      "step": 21520
    },
    {
      "epoch": 12.691627358490566,
      "grad_norm": 3.0293211936950684,
      "learning_rate": 3.5456225699056256e-06,
      "loss": 0.3993,
      "num_input_tokens_seen": 14052184,
      "step": 21525
    },
    {
      "epoch": 12.694575471698114,
      "grad_norm": 3.8884997367858887,
      "learning_rate": 3.5431612902307426e-06,
      "loss": 0.3501,
      "num_input_tokens_seen": 14055768,
      "step": 21530
    },
    {
      "epoch": 12.69752358490566,
      "grad_norm": 3.3717899322509766,
      "learning_rate": 3.540700396260186e-06,
      "loss": 0.3326,
      "num_input_tokens_seen": 14058072,
      "step": 21535
    },
    {
      "epoch": 12.700471698113208,
      "grad_norm": 6.659128665924072,
      "learning_rate": 3.538239888645489e-06,
      "loss": 0.4521,
      "num_input_tokens_seen": 14064184,
      "step": 21540
    },
    {
      "epoch": 12.703419811320755,
      "grad_norm": 4.410452842712402,
      "learning_rate": 3.535779768038082e-06,
      "loss": 0.3504,
      "num_input_tokens_seen": 14067736,
      "step": 21545
    },
    {
      "epoch": 12.706367924528301,
      "grad_norm": 2.695115566253662,
      "learning_rate": 3.5333200350892905e-06,
      "loss": 0.2695,
      "num_input_tokens_seen": 14071000,
      "step": 21550
    },
    {
      "epoch": 12.70931603773585,
      "grad_norm": 2.7699124813079834,
      "learning_rate": 3.530860690450342e-06,
      "loss": 0.4768,
      "num_input_tokens_seen": 14073880,
      "step": 21555
    },
    {
      "epoch": 12.712264150943396,
      "grad_norm": 2.480238437652588,
      "learning_rate": 3.528401734772357e-06,
      "loss": 0.3675,
      "num_input_tokens_seen": 14076440,
      "step": 21560
    },
    {
      "epoch": 12.715212264150944,
      "grad_norm": 4.693593502044678,
      "learning_rate": 3.5259431687063538e-06,
      "loss": 0.3076,
      "num_input_tokens_seen": 14080056,
      "step": 21565
    },
    {
      "epoch": 12.71816037735849,
      "grad_norm": 3.145026683807373,
      "learning_rate": 3.523484992903249e-06,
      "loss": 0.3149,
      "num_input_tokens_seen": 14082648,
      "step": 21570
    },
    {
      "epoch": 12.721108490566039,
      "grad_norm": 3.3825626373291016,
      "learning_rate": 3.5210272080138573e-06,
      "loss": 0.3692,
      "num_input_tokens_seen": 14084824,
      "step": 21575
    },
    {
      "epoch": 12.724056603773585,
      "grad_norm": 6.032142162322998,
      "learning_rate": 3.518569814688887e-06,
      "loss": 0.3751,
      "num_input_tokens_seen": 14087736,
      "step": 21580
    },
    {
      "epoch": 12.727004716981131,
      "grad_norm": 2.63630747795105,
      "learning_rate": 3.5161128135789414e-06,
      "loss": 0.3009,
      "num_input_tokens_seen": 14090840,
      "step": 21585
    },
    {
      "epoch": 12.72995283018868,
      "grad_norm": 5.657066822052002,
      "learning_rate": 3.513656205334525e-06,
      "loss": 0.285,
      "num_input_tokens_seen": 14094008,
      "step": 21590
    },
    {
      "epoch": 12.732900943396226,
      "grad_norm": 2.8135833740234375,
      "learning_rate": 3.5111999906060336e-06,
      "loss": 0.3837,
      "num_input_tokens_seen": 14097176,
      "step": 21595
    },
    {
      "epoch": 12.735849056603774,
      "grad_norm": 3.0824592113494873,
      "learning_rate": 3.50874417004376e-06,
      "loss": 0.2791,
      "num_input_tokens_seen": 14100536,
      "step": 21600
    },
    {
      "epoch": 12.73879716981132,
      "grad_norm": 4.154845237731934,
      "learning_rate": 3.5062887442978956e-06,
      "loss": 0.344,
      "num_input_tokens_seen": 14103576,
      "step": 21605
    },
    {
      "epoch": 12.741745283018869,
      "grad_norm": 1.583423137664795,
      "learning_rate": 3.503833714018524e-06,
      "loss": 0.3228,
      "num_input_tokens_seen": 14106872,
      "step": 21610
    },
    {
      "epoch": 12.744693396226415,
      "grad_norm": 3.2842671871185303,
      "learning_rate": 3.5013790798556228e-06,
      "loss": 0.3708,
      "num_input_tokens_seen": 14110712,
      "step": 21615
    },
    {
      "epoch": 12.747641509433961,
      "grad_norm": 3.0496938228607178,
      "learning_rate": 3.4989248424590705e-06,
      "loss": 0.3733,
      "num_input_tokens_seen": 14114008,
      "step": 21620
    },
    {
      "epoch": 12.75058962264151,
      "grad_norm": 3.9015731811523438,
      "learning_rate": 3.4964710024786354e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 14117112,
      "step": 21625
    },
    {
      "epoch": 12.753537735849056,
      "grad_norm": 4.6696319580078125,
      "learning_rate": 3.4940175605639813e-06,
      "loss": 0.3433,
      "num_input_tokens_seen": 14119896,
      "step": 21630
    },
    {
      "epoch": 12.756485849056604,
      "grad_norm": 4.118602275848389,
      "learning_rate": 3.4915645173646694e-06,
      "loss": 0.5534,
      "num_input_tokens_seen": 14122840,
      "step": 21635
    },
    {
      "epoch": 12.75943396226415,
      "grad_norm": 7.686024188995361,
      "learning_rate": 3.489111873530153e-06,
      "loss": 0.4638,
      "num_input_tokens_seen": 14125304,
      "step": 21640
    },
    {
      "epoch": 12.762382075471699,
      "grad_norm": 4.889338493347168,
      "learning_rate": 3.4866596297097776e-06,
      "loss": 0.4115,
      "num_input_tokens_seen": 14128856,
      "step": 21645
    },
    {
      "epoch": 12.765330188679245,
      "grad_norm": 3.0874056816101074,
      "learning_rate": 3.484207786552789e-06,
      "loss": 0.3267,
      "num_input_tokens_seen": 14132024,
      "step": 21650
    },
    {
      "epoch": 12.768278301886792,
      "grad_norm": 2.975493907928467,
      "learning_rate": 3.4817563447083214e-06,
      "loss": 0.3325,
      "num_input_tokens_seen": 14135192,
      "step": 21655
    },
    {
      "epoch": 12.77122641509434,
      "grad_norm": 4.667953968048096,
      "learning_rate": 3.4793053048254044e-06,
      "loss": 0.3522,
      "num_input_tokens_seen": 14138776,
      "step": 21660
    },
    {
      "epoch": 12.774174528301886,
      "grad_norm": 11.657426834106445,
      "learning_rate": 3.47685466755296e-06,
      "loss": 0.3625,
      "num_input_tokens_seen": 14142136,
      "step": 21665
    },
    {
      "epoch": 12.777122641509434,
      "grad_norm": 4.863961219787598,
      "learning_rate": 3.474404433539809e-06,
      "loss": 0.3591,
      "num_input_tokens_seen": 14145400,
      "step": 21670
    },
    {
      "epoch": 12.78007075471698,
      "grad_norm": 4.171825408935547,
      "learning_rate": 3.4719546034346598e-06,
      "loss": 0.3448,
      "num_input_tokens_seen": 14149016,
      "step": 21675
    },
    {
      "epoch": 12.783018867924529,
      "grad_norm": 2.2246859073638916,
      "learning_rate": 3.4695051778861125e-06,
      "loss": 0.3569,
      "num_input_tokens_seen": 14152088,
      "step": 21680
    },
    {
      "epoch": 12.785966981132075,
      "grad_norm": 2.7536182403564453,
      "learning_rate": 3.4670561575426677e-06,
      "loss": 0.3916,
      "num_input_tokens_seen": 14154840,
      "step": 21685
    },
    {
      "epoch": 12.788915094339622,
      "grad_norm": 5.0762104988098145,
      "learning_rate": 3.4646075430527115e-06,
      "loss": 0.3309,
      "num_input_tokens_seen": 14158392,
      "step": 21690
    },
    {
      "epoch": 12.79186320754717,
      "grad_norm": 4.05437707901001,
      "learning_rate": 3.4621593350645236e-06,
      "loss": 0.3953,
      "num_input_tokens_seen": 14161560,
      "step": 21695
    },
    {
      "epoch": 12.794811320754716,
      "grad_norm": 3.5340235233306885,
      "learning_rate": 3.4597115342262817e-06,
      "loss": 0.5333,
      "num_input_tokens_seen": 14165272,
      "step": 21700
    },
    {
      "epoch": 12.797759433962264,
      "grad_norm": 2.641016721725464,
      "learning_rate": 3.4572641411860484e-06,
      "loss": 0.2882,
      "num_input_tokens_seen": 14168312,
      "step": 21705
    },
    {
      "epoch": 12.80070754716981,
      "grad_norm": 4.4181647300720215,
      "learning_rate": 3.454817156591782e-06,
      "loss": 0.335,
      "num_input_tokens_seen": 14171544,
      "step": 21710
    },
    {
      "epoch": 12.803655660377359,
      "grad_norm": 3.4987332820892334,
      "learning_rate": 3.4523705810913344e-06,
      "loss": 0.3581,
      "num_input_tokens_seen": 14174008,
      "step": 21715
    },
    {
      "epoch": 12.806603773584905,
      "grad_norm": 8.13829517364502,
      "learning_rate": 3.449924415332443e-06,
      "loss": 0.298,
      "num_input_tokens_seen": 14177560,
      "step": 21720
    },
    {
      "epoch": 12.809551886792454,
      "grad_norm": 3.8514766693115234,
      "learning_rate": 3.447478659962745e-06,
      "loss": 0.3413,
      "num_input_tokens_seen": 14180632,
      "step": 21725
    },
    {
      "epoch": 12.8125,
      "grad_norm": 4.029857158660889,
      "learning_rate": 3.4450333156297625e-06,
      "loss": 0.4951,
      "num_input_tokens_seen": 14183960,
      "step": 21730
    },
    {
      "epoch": 12.815448113207546,
      "grad_norm": 4.088692665100098,
      "learning_rate": 3.44258838298091e-06,
      "loss": 0.3659,
      "num_input_tokens_seen": 14186616,
      "step": 21735
    },
    {
      "epoch": 12.818396226415095,
      "grad_norm": 3.341370105743408,
      "learning_rate": 3.440143862663497e-06,
      "loss": 0.3283,
      "num_input_tokens_seen": 14189688,
      "step": 21740
    },
    {
      "epoch": 12.821344339622641,
      "grad_norm": 4.7779412269592285,
      "learning_rate": 3.4376997553247183e-06,
      "loss": 0.2844,
      "num_input_tokens_seen": 14192984,
      "step": 21745
    },
    {
      "epoch": 12.82429245283019,
      "grad_norm": 3.5784049034118652,
      "learning_rate": 3.4352560616116617e-06,
      "loss": 0.3077,
      "num_input_tokens_seen": 14196856,
      "step": 21750
    },
    {
      "epoch": 12.827240566037736,
      "grad_norm": 3.40222430229187,
      "learning_rate": 3.4328127821713077e-06,
      "loss": 0.5122,
      "num_input_tokens_seen": 14199736,
      "step": 21755
    },
    {
      "epoch": 12.830188679245284,
      "grad_norm": 3.3480069637298584,
      "learning_rate": 3.430369917650521e-06,
      "loss": 0.301,
      "num_input_tokens_seen": 14202872,
      "step": 21760
    },
    {
      "epoch": 12.83313679245283,
      "grad_norm": 2.4822418689727783,
      "learning_rate": 3.427927468696066e-06,
      "loss": 0.3869,
      "num_input_tokens_seen": 14206232,
      "step": 21765
    },
    {
      "epoch": 12.836084905660378,
      "grad_norm": 2.68818736076355,
      "learning_rate": 3.425485435954588e-06,
      "loss": 0.3546,
      "num_input_tokens_seen": 14209464,
      "step": 21770
    },
    {
      "epoch": 12.839033018867925,
      "grad_norm": 3.985126256942749,
      "learning_rate": 3.4230438200726274e-06,
      "loss": 0.3767,
      "num_input_tokens_seen": 14212920,
      "step": 21775
    },
    {
      "epoch": 12.841981132075471,
      "grad_norm": 2.7161736488342285,
      "learning_rate": 3.4206026216966113e-06,
      "loss": 0.356,
      "num_input_tokens_seen": 14215480,
      "step": 21780
    },
    {
      "epoch": 12.84492924528302,
      "grad_norm": 2.8949382305145264,
      "learning_rate": 3.418161841472858e-06,
      "loss": 0.4151,
      "num_input_tokens_seen": 14218488,
      "step": 21785
    },
    {
      "epoch": 12.847877358490566,
      "grad_norm": 3.0132381916046143,
      "learning_rate": 3.4157214800475746e-06,
      "loss": 0.316,
      "num_input_tokens_seen": 14221752,
      "step": 21790
    },
    {
      "epoch": 12.850825471698114,
      "grad_norm": 2.1272871494293213,
      "learning_rate": 3.4132815380668577e-06,
      "loss": 0.2872,
      "num_input_tokens_seen": 14224888,
      "step": 21795
    },
    {
      "epoch": 12.85377358490566,
      "grad_norm": 3.0509471893310547,
      "learning_rate": 3.410842016176691e-06,
      "loss": 0.3791,
      "num_input_tokens_seen": 14228568,
      "step": 21800
    },
    {
      "epoch": 12.856721698113208,
      "grad_norm": 3.1193125247955322,
      "learning_rate": 3.4084029150229503e-06,
      "loss": 0.2852,
      "num_input_tokens_seen": 14232088,
      "step": 21805
    },
    {
      "epoch": 12.859669811320755,
      "grad_norm": 1.9920432567596436,
      "learning_rate": 3.4059642352513965e-06,
      "loss": 0.3437,
      "num_input_tokens_seen": 14235448,
      "step": 21810
    },
    {
      "epoch": 12.862617924528301,
      "grad_norm": 3.6852011680603027,
      "learning_rate": 3.4035259775076813e-06,
      "loss": 0.3702,
      "num_input_tokens_seen": 14238904,
      "step": 21815
    },
    {
      "epoch": 12.86556603773585,
      "grad_norm": 2.139373540878296,
      "learning_rate": 3.401088142437344e-06,
      "loss": 0.2566,
      "num_input_tokens_seen": 14242840,
      "step": 21820
    },
    {
      "epoch": 12.868514150943396,
      "grad_norm": 2.9665040969848633,
      "learning_rate": 3.398650730685813e-06,
      "loss": 0.2887,
      "num_input_tokens_seen": 14245496,
      "step": 21825
    },
    {
      "epoch": 12.871462264150944,
      "grad_norm": 1.9119702577590942,
      "learning_rate": 3.396213742898401e-06,
      "loss": 0.2586,
      "num_input_tokens_seen": 14250168,
      "step": 21830
    },
    {
      "epoch": 12.87441037735849,
      "grad_norm": 3.0877885818481445,
      "learning_rate": 3.3937771797203134e-06,
      "loss": 0.4301,
      "num_input_tokens_seen": 14253752,
      "step": 21835
    },
    {
      "epoch": 12.877358490566039,
      "grad_norm": 1.91194748878479,
      "learning_rate": 3.391341041796641e-06,
      "loss": 0.2607,
      "num_input_tokens_seen": 14257560,
      "step": 21840
    },
    {
      "epoch": 12.880306603773585,
      "grad_norm": 3.223383903503418,
      "learning_rate": 3.3889053297723585e-06,
      "loss": 0.2537,
      "num_input_tokens_seen": 14262008,
      "step": 21845
    },
    {
      "epoch": 12.883254716981131,
      "grad_norm": 2.230386972427368,
      "learning_rate": 3.3864700442923342e-06,
      "loss": 0.288,
      "num_input_tokens_seen": 14265560,
      "step": 21850
    },
    {
      "epoch": 12.88620283018868,
      "grad_norm": 6.485767364501953,
      "learning_rate": 3.384035186001318e-06,
      "loss": 0.3272,
      "num_input_tokens_seen": 14268536,
      "step": 21855
    },
    {
      "epoch": 12.889150943396226,
      "grad_norm": 6.27098274230957,
      "learning_rate": 3.381600755543953e-06,
      "loss": 0.309,
      "num_input_tokens_seen": 14270904,
      "step": 21860
    },
    {
      "epoch": 12.892099056603774,
      "grad_norm": 7.32874870300293,
      "learning_rate": 3.3791667535647615e-06,
      "loss": 0.3665,
      "num_input_tokens_seen": 14274232,
      "step": 21865
    },
    {
      "epoch": 12.89504716981132,
      "grad_norm": 4.632501125335693,
      "learning_rate": 3.3767331807081584e-06,
      "loss": 0.3601,
      "num_input_tokens_seen": 14278808,
      "step": 21870
    },
    {
      "epoch": 12.897995283018869,
      "grad_norm": 3.200711488723755,
      "learning_rate": 3.374300037618442e-06,
      "loss": 0.2633,
      "num_input_tokens_seen": 14281560,
      "step": 21875
    },
    {
      "epoch": 12.900943396226415,
      "grad_norm": 4.05991268157959,
      "learning_rate": 3.371867324939796e-06,
      "loss": 0.2684,
      "num_input_tokens_seen": 14284760,
      "step": 21880
    },
    {
      "epoch": 12.903891509433961,
      "grad_norm": 6.064416408538818,
      "learning_rate": 3.369435043316293e-06,
      "loss": 0.4136,
      "num_input_tokens_seen": 14287320,
      "step": 21885
    },
    {
      "epoch": 12.90683962264151,
      "grad_norm": 3.5634496212005615,
      "learning_rate": 3.36700319339189e-06,
      "loss": 0.3875,
      "num_input_tokens_seen": 14290840,
      "step": 21890
    },
    {
      "epoch": 12.909787735849056,
      "grad_norm": 2.4734292030334473,
      "learning_rate": 3.3645717758104286e-06,
      "loss": 0.3353,
      "num_input_tokens_seen": 14295736,
      "step": 21895
    },
    {
      "epoch": 12.912735849056604,
      "grad_norm": 2.804713726043701,
      "learning_rate": 3.3621407912156383e-06,
      "loss": 0.3002,
      "num_input_tokens_seen": 14299000,
      "step": 21900
    },
    {
      "epoch": 12.91568396226415,
      "grad_norm": 2.8870646953582764,
      "learning_rate": 3.3597102402511326e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 14302296,
      "step": 21905
    },
    {
      "epoch": 12.918632075471699,
      "grad_norm": 4.311373233795166,
      "learning_rate": 3.3572801235604093e-06,
      "loss": 0.302,
      "num_input_tokens_seen": 14305336,
      "step": 21910
    },
    {
      "epoch": 12.921580188679245,
      "grad_norm": 3.7751832008361816,
      "learning_rate": 3.3548504417868538e-06,
      "loss": 0.354,
      "num_input_tokens_seen": 14309240,
      "step": 21915
    },
    {
      "epoch": 12.924528301886792,
      "grad_norm": 2.933415651321411,
      "learning_rate": 3.352421195573734e-06,
      "loss": 0.3393,
      "num_input_tokens_seen": 14313368,
      "step": 21920
    },
    {
      "epoch": 12.92747641509434,
      "grad_norm": 3.7907328605651855,
      "learning_rate": 3.3499923855642026e-06,
      "loss": 0.496,
      "num_input_tokens_seen": 14316152,
      "step": 21925
    },
    {
      "epoch": 12.930424528301886,
      "grad_norm": 2.819135904312134,
      "learning_rate": 3.3475640124012986e-06,
      "loss": 0.3256,
      "num_input_tokens_seen": 14319032,
      "step": 21930
    },
    {
      "epoch": 12.933372641509434,
      "grad_norm": 1.99680495262146,
      "learning_rate": 3.345136076727945e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 14322264,
      "step": 21935
    },
    {
      "epoch": 12.93632075471698,
      "grad_norm": 3.8027737140655518,
      "learning_rate": 3.3427085791869453e-06,
      "loss": 0.4122,
      "num_input_tokens_seen": 14325528,
      "step": 21940
    },
    {
      "epoch": 12.939268867924529,
      "grad_norm": 2.7927749156951904,
      "learning_rate": 3.3402815204209926e-06,
      "loss": 0.3185,
      "num_input_tokens_seen": 14328600,
      "step": 21945
    },
    {
      "epoch": 12.942216981132075,
      "grad_norm": 2.7212982177734375,
      "learning_rate": 3.337854901072659e-06,
      "loss": 0.3109,
      "num_input_tokens_seen": 14331832,
      "step": 21950
    },
    {
      "epoch": 12.945165094339622,
      "grad_norm": 2.5636136531829834,
      "learning_rate": 3.3354287217844056e-06,
      "loss": 0.2806,
      "num_input_tokens_seen": 14334328,
      "step": 21955
    },
    {
      "epoch": 12.94811320754717,
      "grad_norm": 3.317197561264038,
      "learning_rate": 3.3330029831985712e-06,
      "loss": 0.3336,
      "num_input_tokens_seen": 14337816,
      "step": 21960
    },
    {
      "epoch": 12.951061320754716,
      "grad_norm": 2.771249771118164,
      "learning_rate": 3.330577685957382e-06,
      "loss": 0.3657,
      "num_input_tokens_seen": 14340536,
      "step": 21965
    },
    {
      "epoch": 12.954009433962264,
      "grad_norm": 2.747359275817871,
      "learning_rate": 3.3281528307029454e-06,
      "loss": 0.3736,
      "num_input_tokens_seen": 14343448,
      "step": 21970
    },
    {
      "epoch": 12.95695754716981,
      "grad_norm": 2.5416488647460938,
      "learning_rate": 3.325728418077251e-06,
      "loss": 0.3106,
      "num_input_tokens_seen": 14347320,
      "step": 21975
    },
    {
      "epoch": 12.959905660377359,
      "grad_norm": 3.9098031520843506,
      "learning_rate": 3.3233044487221744e-06,
      "loss": 0.3581,
      "num_input_tokens_seen": 14350392,
      "step": 21980
    },
    {
      "epoch": 12.962853773584905,
      "grad_norm": 3.4075942039489746,
      "learning_rate": 3.3208809232794715e-06,
      "loss": 0.356,
      "num_input_tokens_seen": 14354296,
      "step": 21985
    },
    {
      "epoch": 12.965801886792454,
      "grad_norm": 2.6556880474090576,
      "learning_rate": 3.3184578423907797e-06,
      "loss": 0.2802,
      "num_input_tokens_seen": 14357688,
      "step": 21990
    },
    {
      "epoch": 12.96875,
      "grad_norm": 2.916381597518921,
      "learning_rate": 3.3160352066976224e-06,
      "loss": 0.4164,
      "num_input_tokens_seen": 14360312,
      "step": 21995
    },
    {
      "epoch": 12.971698113207546,
      "grad_norm": 4.432518005371094,
      "learning_rate": 3.3136130168414003e-06,
      "loss": 0.4144,
      "num_input_tokens_seen": 14363288,
      "step": 22000
    },
    {
      "epoch": 12.974646226415095,
      "grad_norm": 2.3157541751861572,
      "learning_rate": 3.311191273463401e-06,
      "loss": 0.2744,
      "num_input_tokens_seen": 14367032,
      "step": 22005
    },
    {
      "epoch": 12.977594339622641,
      "grad_norm": 3.246169328689575,
      "learning_rate": 3.3087699772047908e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 14370296,
      "step": 22010
    },
    {
      "epoch": 12.98054245283019,
      "grad_norm": 3.3660190105438232,
      "learning_rate": 3.3063491287066164e-06,
      "loss": 0.4005,
      "num_input_tokens_seen": 14372792,
      "step": 22015
    },
    {
      "epoch": 12.983490566037736,
      "grad_norm": 2.3737194538116455,
      "learning_rate": 3.303928728609811e-06,
      "loss": 0.2704,
      "num_input_tokens_seen": 14375928,
      "step": 22020
    },
    {
      "epoch": 12.986438679245284,
      "grad_norm": 3.029210090637207,
      "learning_rate": 3.3015087775551835e-06,
      "loss": 0.3336,
      "num_input_tokens_seen": 14379608,
      "step": 22025
    },
    {
      "epoch": 12.98938679245283,
      "grad_norm": 3.567734956741333,
      "learning_rate": 3.299089276183427e-06,
      "loss": 0.3712,
      "num_input_tokens_seen": 14382872,
      "step": 22030
    },
    {
      "epoch": 12.992334905660378,
      "grad_norm": 2.584099292755127,
      "learning_rate": 3.2966702251351157e-06,
      "loss": 0.3521,
      "num_input_tokens_seen": 14386136,
      "step": 22035
    },
    {
      "epoch": 12.995283018867925,
      "grad_norm": 3.450352191925049,
      "learning_rate": 3.2942516250507035e-06,
      "loss": 0.3022,
      "num_input_tokens_seen": 14389912,
      "step": 22040
    },
    {
      "epoch": 12.998231132075471,
      "grad_norm": 3.222008228302002,
      "learning_rate": 3.2918334765705227e-06,
      "loss": 0.279,
      "num_input_tokens_seen": 14393368,
      "step": 22045
    },
    {
      "epoch": 13.00117924528302,
      "grad_norm": 2.2067952156066895,
      "learning_rate": 3.289415780334792e-06,
      "loss": 0.2082,
      "num_input_tokens_seen": 14396072,
      "step": 22050
    },
    {
      "epoch": 13.004127358490566,
      "grad_norm": 4.023777008056641,
      "learning_rate": 3.2869985369836067e-06,
      "loss": 0.3529,
      "num_input_tokens_seen": 14398920,
      "step": 22055
    },
    {
      "epoch": 13.007075471698114,
      "grad_norm": 1.790043830871582,
      "learning_rate": 3.2845817471569406e-06,
      "loss": 0.2676,
      "num_input_tokens_seen": 14403080,
      "step": 22060
    },
    {
      "epoch": 13.01002358490566,
      "grad_norm": 4.008598804473877,
      "learning_rate": 3.28216541149465e-06,
      "loss": 0.351,
      "num_input_tokens_seen": 14407176,
      "step": 22065
    },
    {
      "epoch": 13.012971698113208,
      "grad_norm": 4.423854827880859,
      "learning_rate": 3.2797495306364707e-06,
      "loss": 0.2045,
      "num_input_tokens_seen": 14409864,
      "step": 22070
    },
    {
      "epoch": 13.015919811320755,
      "grad_norm": 6.915042400360107,
      "learning_rate": 3.2773341052220174e-06,
      "loss": 0.442,
      "num_input_tokens_seen": 14413384,
      "step": 22075
    },
    {
      "epoch": 13.018867924528301,
      "grad_norm": 19.09061622619629,
      "learning_rate": 3.274919135890783e-06,
      "loss": 0.449,
      "num_input_tokens_seen": 14415912,
      "step": 22080
    },
    {
      "epoch": 13.02181603773585,
      "grad_norm": 3.322899103164673,
      "learning_rate": 3.2725046232821424e-06,
      "loss": 0.2272,
      "num_input_tokens_seen": 14418792,
      "step": 22085
    },
    {
      "epoch": 13.024764150943396,
      "grad_norm": 3.04266619682312,
      "learning_rate": 3.270090568035348e-06,
      "loss": 0.4411,
      "num_input_tokens_seen": 14421960,
      "step": 22090
    },
    {
      "epoch": 13.027712264150944,
      "grad_norm": 3.0039806365966797,
      "learning_rate": 3.2676769707895306e-06,
      "loss": 0.3193,
      "num_input_tokens_seen": 14425000,
      "step": 22095
    },
    {
      "epoch": 13.03066037735849,
      "grad_norm": 2.5538010597229004,
      "learning_rate": 3.2652638321837015e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 14429576,
      "step": 22100
    },
    {
      "epoch": 13.033608490566039,
      "grad_norm": 2.373823881149292,
      "learning_rate": 3.2628511528567497e-06,
      "loss": 0.3257,
      "num_input_tokens_seen": 14432168,
      "step": 22105
    },
    {
      "epoch": 13.036556603773585,
      "grad_norm": 2.758582353591919,
      "learning_rate": 3.2604389334474407e-06,
      "loss": 0.3248,
      "num_input_tokens_seen": 14435848,
      "step": 22110
    },
    {
      "epoch": 13.039504716981131,
      "grad_norm": 3.8507070541381836,
      "learning_rate": 3.2580271745944224e-06,
      "loss": 0.412,
      "num_input_tokens_seen": 14439336,
      "step": 22115
    },
    {
      "epoch": 13.04245283018868,
      "grad_norm": 5.218119144439697,
      "learning_rate": 3.255615876936217e-06,
      "loss": 0.2168,
      "num_input_tokens_seen": 14442568,
      "step": 22120
    },
    {
      "epoch": 13.045400943396226,
      "grad_norm": 3.8426151275634766,
      "learning_rate": 3.2532050411112248e-06,
      "loss": 0.2555,
      "num_input_tokens_seen": 14445736,
      "step": 22125
    },
    {
      "epoch": 13.048349056603774,
      "grad_norm": 3.670010566711426,
      "learning_rate": 3.2507946677577274e-06,
      "loss": 0.2493,
      "num_input_tokens_seen": 14450152,
      "step": 22130
    },
    {
      "epoch": 13.05129716981132,
      "grad_norm": 3.5470523834228516,
      "learning_rate": 3.2483847575138807e-06,
      "loss": 0.3562,
      "num_input_tokens_seen": 14453672,
      "step": 22135
    },
    {
      "epoch": 13.054245283018869,
      "grad_norm": 3.439145565032959,
      "learning_rate": 3.245975311017716e-06,
      "loss": 0.3402,
      "num_input_tokens_seen": 14456872,
      "step": 22140
    },
    {
      "epoch": 13.057193396226415,
      "grad_norm": 3.6502089500427246,
      "learning_rate": 3.2435663289071486e-06,
      "loss": 0.3406,
      "num_input_tokens_seen": 14460296,
      "step": 22145
    },
    {
      "epoch": 13.060141509433961,
      "grad_norm": 1.867048978805542,
      "learning_rate": 3.241157811819966e-06,
      "loss": 0.3162,
      "num_input_tokens_seen": 14463048,
      "step": 22150
    },
    {
      "epoch": 13.06308962264151,
      "grad_norm": 4.080491542816162,
      "learning_rate": 3.2387497603938327e-06,
      "loss": 0.3425,
      "num_input_tokens_seen": 14465864,
      "step": 22155
    },
    {
      "epoch": 13.066037735849056,
      "grad_norm": 7.849427700042725,
      "learning_rate": 3.2363421752662903e-06,
      "loss": 0.4108,
      "num_input_tokens_seen": 14468744,
      "step": 22160
    },
    {
      "epoch": 13.068985849056604,
      "grad_norm": 3.0759518146514893,
      "learning_rate": 3.233935057074759e-06,
      "loss": 0.2618,
      "num_input_tokens_seen": 14472360,
      "step": 22165
    },
    {
      "epoch": 13.07193396226415,
      "grad_norm": 3.8710527420043945,
      "learning_rate": 3.2315284064565324e-06,
      "loss": 0.3395,
      "num_input_tokens_seen": 14474824,
      "step": 22170
    },
    {
      "epoch": 13.074882075471699,
      "grad_norm": 2.691591739654541,
      "learning_rate": 3.2291222240487813e-06,
      "loss": 0.2309,
      "num_input_tokens_seen": 14477544,
      "step": 22175
    },
    {
      "epoch": 13.077830188679245,
      "grad_norm": 3.781956672668457,
      "learning_rate": 3.226716510488554e-06,
      "loss": 0.3505,
      "num_input_tokens_seen": 14481224,
      "step": 22180
    },
    {
      "epoch": 13.080778301886792,
      "grad_norm": 1.726909875869751,
      "learning_rate": 3.224311266412773e-06,
      "loss": 0.3061,
      "num_input_tokens_seen": 14483688,
      "step": 22185
    },
    {
      "epoch": 13.08372641509434,
      "grad_norm": 3.7889645099639893,
      "learning_rate": 3.2219064924582366e-06,
      "loss": 0.337,
      "num_input_tokens_seen": 14487112,
      "step": 22190
    },
    {
      "epoch": 13.086674528301886,
      "grad_norm": 2.6657702922821045,
      "learning_rate": 3.2195021892616197e-06,
      "loss": 0.3797,
      "num_input_tokens_seen": 14492104,
      "step": 22195
    },
    {
      "epoch": 13.089622641509434,
      "grad_norm": 2.5574259757995605,
      "learning_rate": 3.217098357459472e-06,
      "loss": 0.3244,
      "num_input_tokens_seen": 14496040,
      "step": 22200
    },
    {
      "epoch": 13.09257075471698,
      "grad_norm": 3.7597057819366455,
      "learning_rate": 3.214694997688217e-06,
      "loss": 0.4146,
      "num_input_tokens_seen": 14499432,
      "step": 22205
    },
    {
      "epoch": 13.095518867924529,
      "grad_norm": 3.8260746002197266,
      "learning_rate": 3.2122921105841572e-06,
      "loss": 0.3313,
      "num_input_tokens_seen": 14502664,
      "step": 22210
    },
    {
      "epoch": 13.098466981132075,
      "grad_norm": 2.786687135696411,
      "learning_rate": 3.2098896967834647e-06,
      "loss": 0.4458,
      "num_input_tokens_seen": 14506440,
      "step": 22215
    },
    {
      "epoch": 13.101415094339623,
      "grad_norm": 3.3483920097351074,
      "learning_rate": 3.2074877569221896e-06,
      "loss": 0.284,
      "num_input_tokens_seen": 14509160,
      "step": 22220
    },
    {
      "epoch": 13.10436320754717,
      "grad_norm": 3.296718120574951,
      "learning_rate": 3.205086291636257e-06,
      "loss": 0.413,
      "num_input_tokens_seen": 14512520,
      "step": 22225
    },
    {
      "epoch": 13.107311320754716,
      "grad_norm": 11.569144248962402,
      "learning_rate": 3.202685301561463e-06,
      "loss": 0.5629,
      "num_input_tokens_seen": 14517096,
      "step": 22230
    },
    {
      "epoch": 13.110259433962264,
      "grad_norm": 4.862022399902344,
      "learning_rate": 3.200284787333482e-06,
      "loss": 0.2871,
      "num_input_tokens_seen": 14519912,
      "step": 22235
    },
    {
      "epoch": 13.11320754716981,
      "grad_norm": 2.805330276489258,
      "learning_rate": 3.1978847495878595e-06,
      "loss": 0.2783,
      "num_input_tokens_seen": 14524680,
      "step": 22240
    },
    {
      "epoch": 13.116155660377359,
      "grad_norm": 3.4541144371032715,
      "learning_rate": 3.1954851889600176e-06,
      "loss": 0.3445,
      "num_input_tokens_seen": 14527912,
      "step": 22245
    },
    {
      "epoch": 13.119103773584905,
      "grad_norm": 5.086663722991943,
      "learning_rate": 3.1930861060852485e-06,
      "loss": 0.2862,
      "num_input_tokens_seen": 14531080,
      "step": 22250
    },
    {
      "epoch": 13.122051886792454,
      "grad_norm": 2.7891407012939453,
      "learning_rate": 3.1906875015987194e-06,
      "loss": 0.3712,
      "num_input_tokens_seen": 14534504,
      "step": 22255
    },
    {
      "epoch": 13.125,
      "grad_norm": 5.797956943511963,
      "learning_rate": 3.188289376135473e-06,
      "loss": 0.304,
      "num_input_tokens_seen": 14538056,
      "step": 22260
    },
    {
      "epoch": 13.127948113207546,
      "grad_norm": 4.2157158851623535,
      "learning_rate": 3.1858917303304213e-06,
      "loss": 0.3575,
      "num_input_tokens_seen": 14540968,
      "step": 22265
    },
    {
      "epoch": 13.130896226415095,
      "grad_norm": 3.279102325439453,
      "learning_rate": 3.1834945648183535e-06,
      "loss": 0.2763,
      "num_input_tokens_seen": 14544520,
      "step": 22270
    },
    {
      "epoch": 13.133844339622641,
      "grad_norm": 6.092158317565918,
      "learning_rate": 3.1810978802339283e-06,
      "loss": 0.3363,
      "num_input_tokens_seen": 14547176,
      "step": 22275
    },
    {
      "epoch": 13.13679245283019,
      "grad_norm": 3.129951000213623,
      "learning_rate": 3.1787016772116767e-06,
      "loss": 0.3015,
      "num_input_tokens_seen": 14552840,
      "step": 22280
    },
    {
      "epoch": 13.139740566037736,
      "grad_norm": 3.3574717044830322,
      "learning_rate": 3.1763059563860073e-06,
      "loss": 0.3367,
      "num_input_tokens_seen": 14555752,
      "step": 22285
    },
    {
      "epoch": 13.142688679245284,
      "grad_norm": 3.4803028106689453,
      "learning_rate": 3.1739107183911953e-06,
      "loss": 0.4115,
      "num_input_tokens_seen": 14558952,
      "step": 22290
    },
    {
      "epoch": 13.14563679245283,
      "grad_norm": 2.781121253967285,
      "learning_rate": 3.1715159638613898e-06,
      "loss": 0.382,
      "num_input_tokens_seen": 14563560,
      "step": 22295
    },
    {
      "epoch": 13.148584905660377,
      "grad_norm": 6.066686153411865,
      "learning_rate": 3.1691216934306134e-06,
      "loss": 0.3265,
      "num_input_tokens_seen": 14566984,
      "step": 22300
    },
    {
      "epoch": 13.151533018867925,
      "grad_norm": 4.6800665855407715,
      "learning_rate": 3.16672790773276e-06,
      "loss": 0.2904,
      "num_input_tokens_seen": 14569992,
      "step": 22305
    },
    {
      "epoch": 13.154481132075471,
      "grad_norm": 2.685800075531006,
      "learning_rate": 3.164334607401593e-06,
      "loss": 0.2653,
      "num_input_tokens_seen": 14573544,
      "step": 22310
    },
    {
      "epoch": 13.15742924528302,
      "grad_norm": 2.884449005126953,
      "learning_rate": 3.1619417930707506e-06,
      "loss": 0.3992,
      "num_input_tokens_seen": 14576584,
      "step": 22315
    },
    {
      "epoch": 13.160377358490566,
      "grad_norm": 3.444021463394165,
      "learning_rate": 3.1595494653737408e-06,
      "loss": 0.3906,
      "num_input_tokens_seen": 14579624,
      "step": 22320
    },
    {
      "epoch": 13.163325471698114,
      "grad_norm": 4.368343830108643,
      "learning_rate": 3.1571576249439408e-06,
      "loss": 0.4239,
      "num_input_tokens_seen": 14583304,
      "step": 22325
    },
    {
      "epoch": 13.16627358490566,
      "grad_norm": 3.0407683849334717,
      "learning_rate": 3.1547662724146e-06,
      "loss": 0.2929,
      "num_input_tokens_seen": 14586120,
      "step": 22330
    },
    {
      "epoch": 13.169221698113208,
      "grad_norm": 5.003282070159912,
      "learning_rate": 3.1523754084188436e-06,
      "loss": 0.2204,
      "num_input_tokens_seen": 14589544,
      "step": 22335
    },
    {
      "epoch": 13.172169811320755,
      "grad_norm": 2.7133677005767822,
      "learning_rate": 3.149985033589661e-06,
      "loss": 0.3502,
      "num_input_tokens_seen": 14593576,
      "step": 22340
    },
    {
      "epoch": 13.175117924528301,
      "grad_norm": 2.7527003288269043,
      "learning_rate": 3.147595148559912e-06,
      "loss": 0.4278,
      "num_input_tokens_seen": 14595976,
      "step": 22345
    },
    {
      "epoch": 13.17806603773585,
      "grad_norm": 2.8890938758850098,
      "learning_rate": 3.1452057539623328e-06,
      "loss": 0.2835,
      "num_input_tokens_seen": 14600200,
      "step": 22350
    },
    {
      "epoch": 13.181014150943396,
      "grad_norm": 3.151230812072754,
      "learning_rate": 3.142816850429523e-06,
      "loss": 0.342,
      "num_input_tokens_seen": 14603400,
      "step": 22355
    },
    {
      "epoch": 13.183962264150944,
      "grad_norm": 4.360023498535156,
      "learning_rate": 3.1404284385939552e-06,
      "loss": 0.3212,
      "num_input_tokens_seen": 14606280,
      "step": 22360
    },
    {
      "epoch": 13.18691037735849,
      "grad_norm": 2.9618966579437256,
      "learning_rate": 3.138040519087975e-06,
      "loss": 0.364,
      "num_input_tokens_seen": 14609224,
      "step": 22365
    },
    {
      "epoch": 13.189858490566039,
      "grad_norm": 3.9195735454559326,
      "learning_rate": 3.13565309254379e-06,
      "loss": 0.3287,
      "num_input_tokens_seen": 14612008,
      "step": 22370
    },
    {
      "epoch": 13.192806603773585,
      "grad_norm": 3.0627336502075195,
      "learning_rate": 3.1332661595934845e-06,
      "loss": 0.3505,
      "num_input_tokens_seen": 14614888,
      "step": 22375
    },
    {
      "epoch": 13.195754716981131,
      "grad_norm": 2.8627541065216064,
      "learning_rate": 3.130879720869008e-06,
      "loss": 0.4883,
      "num_input_tokens_seen": 14617960,
      "step": 22380
    },
    {
      "epoch": 13.19870283018868,
      "grad_norm": 2.579274892807007,
      "learning_rate": 3.1284937770021815e-06,
      "loss": 0.2341,
      "num_input_tokens_seen": 14620776,
      "step": 22385
    },
    {
      "epoch": 13.201650943396226,
      "grad_norm": 3.149773597717285,
      "learning_rate": 3.1261083286246916e-06,
      "loss": 0.3882,
      "num_input_tokens_seen": 14624392,
      "step": 22390
    },
    {
      "epoch": 13.204599056603774,
      "grad_norm": 4.622279644012451,
      "learning_rate": 3.1237233763680997e-06,
      "loss": 0.2625,
      "num_input_tokens_seen": 14628136,
      "step": 22395
    },
    {
      "epoch": 13.20754716981132,
      "grad_norm": 2.9745142459869385,
      "learning_rate": 3.1213389208638303e-06,
      "loss": 0.2881,
      "num_input_tokens_seen": 14632264,
      "step": 22400
    },
    {
      "epoch": 13.210495283018869,
      "grad_norm": 4.559311866760254,
      "learning_rate": 3.1189549627431757e-06,
      "loss": 0.3104,
      "num_input_tokens_seen": 14634952,
      "step": 22405
    },
    {
      "epoch": 13.213443396226415,
      "grad_norm": 2.2010486125946045,
      "learning_rate": 3.116571502637304e-06,
      "loss": 0.1907,
      "num_input_tokens_seen": 14637864,
      "step": 22410
    },
    {
      "epoch": 13.216391509433961,
      "grad_norm": 4.6809515953063965,
      "learning_rate": 3.1141885411772434e-06,
      "loss": 0.3162,
      "num_input_tokens_seen": 14640296,
      "step": 22415
    },
    {
      "epoch": 13.21933962264151,
      "grad_norm": 3.5405428409576416,
      "learning_rate": 3.111806078993893e-06,
      "loss": 0.2773,
      "num_input_tokens_seen": 14643688,
      "step": 22420
    },
    {
      "epoch": 13.222287735849056,
      "grad_norm": 2.863039493560791,
      "learning_rate": 3.1094241167180223e-06,
      "loss": 0.4029,
      "num_input_tokens_seen": 14646280,
      "step": 22425
    },
    {
      "epoch": 13.225235849056604,
      "grad_norm": 2.4853432178497314,
      "learning_rate": 3.1070426549802623e-06,
      "loss": 0.3263,
      "num_input_tokens_seen": 14650184,
      "step": 22430
    },
    {
      "epoch": 13.22818396226415,
      "grad_norm": 4.08460807800293,
      "learning_rate": 3.1046616944111196e-06,
      "loss": 0.3968,
      "num_input_tokens_seen": 14654152,
      "step": 22435
    },
    {
      "epoch": 13.231132075471699,
      "grad_norm": 5.702762603759766,
      "learning_rate": 3.1022812356409606e-06,
      "loss": 0.3581,
      "num_input_tokens_seen": 14656712,
      "step": 22440
    },
    {
      "epoch": 13.234080188679245,
      "grad_norm": 2.6994597911834717,
      "learning_rate": 3.0999012793000244e-06,
      "loss": 0.3072,
      "num_input_tokens_seen": 14659880,
      "step": 22445
    },
    {
      "epoch": 13.237028301886792,
      "grad_norm": 2.3515756130218506,
      "learning_rate": 3.097521826018414e-06,
      "loss": 0.2346,
      "num_input_tokens_seen": 14663752,
      "step": 22450
    },
    {
      "epoch": 13.23997641509434,
      "grad_norm": 3.334465503692627,
      "learning_rate": 3.0951428764260973e-06,
      "loss": 0.4994,
      "num_input_tokens_seen": 14666984,
      "step": 22455
    },
    {
      "epoch": 13.242924528301886,
      "grad_norm": 3.8413703441619873,
      "learning_rate": 3.092764431152915e-06,
      "loss": 0.3028,
      "num_input_tokens_seen": 14669896,
      "step": 22460
    },
    {
      "epoch": 13.245872641509434,
      "grad_norm": 2.1272010803222656,
      "learning_rate": 3.0903864908285693e-06,
      "loss": 0.2516,
      "num_input_tokens_seen": 14673704,
      "step": 22465
    },
    {
      "epoch": 13.24882075471698,
      "grad_norm": 4.125630855560303,
      "learning_rate": 3.088009056082629e-06,
      "loss": 0.3128,
      "num_input_tokens_seen": 14676680,
      "step": 22470
    },
    {
      "epoch": 13.251768867924529,
      "grad_norm": 5.0732550621032715,
      "learning_rate": 3.0856321275445324e-06,
      "loss": 0.3558,
      "num_input_tokens_seen": 14679656,
      "step": 22475
    },
    {
      "epoch": 13.254716981132075,
      "grad_norm": 13.07447338104248,
      "learning_rate": 3.0832557058435808e-06,
      "loss": 0.4397,
      "num_input_tokens_seen": 14681736,
      "step": 22480
    },
    {
      "epoch": 13.257665094339623,
      "grad_norm": 5.427025318145752,
      "learning_rate": 3.0808797916089405e-06,
      "loss": 0.3563,
      "num_input_tokens_seen": 14685256,
      "step": 22485
    },
    {
      "epoch": 13.26061320754717,
      "grad_norm": 3.1773502826690674,
      "learning_rate": 3.078504385469647e-06,
      "loss": 0.3046,
      "num_input_tokens_seen": 14688776,
      "step": 22490
    },
    {
      "epoch": 13.263561320754716,
      "grad_norm": 3.1019439697265625,
      "learning_rate": 3.076129488054599e-06,
      "loss": 0.3171,
      "num_input_tokens_seen": 14691720,
      "step": 22495
    },
    {
      "epoch": 13.266509433962264,
      "grad_norm": 4.894453048706055,
      "learning_rate": 3.0737550999925604e-06,
      "loss": 0.3176,
      "num_input_tokens_seen": 14694376,
      "step": 22500
    },
    {
      "epoch": 13.26945754716981,
      "grad_norm": 4.793021202087402,
      "learning_rate": 3.0713812219121604e-06,
      "loss": 0.3899,
      "num_input_tokens_seen": 14698376,
      "step": 22505
    },
    {
      "epoch": 13.272405660377359,
      "grad_norm": 4.297952651977539,
      "learning_rate": 3.0690078544418934e-06,
      "loss": 0.3128,
      "num_input_tokens_seen": 14701704,
      "step": 22510
    },
    {
      "epoch": 13.275353773584905,
      "grad_norm": 5.058052062988281,
      "learning_rate": 3.0666349982101198e-06,
      "loss": 0.4028,
      "num_input_tokens_seen": 14704936,
      "step": 22515
    },
    {
      "epoch": 13.278301886792454,
      "grad_norm": 2.7775187492370605,
      "learning_rate": 3.0642626538450627e-06,
      "loss": 0.3548,
      "num_input_tokens_seen": 14707976,
      "step": 22520
    },
    {
      "epoch": 13.28125,
      "grad_norm": 6.442705154418945,
      "learning_rate": 3.061890821974809e-06,
      "loss": 0.4021,
      "num_input_tokens_seen": 14710728,
      "step": 22525
    },
    {
      "epoch": 13.284198113207546,
      "grad_norm": 4.321109771728516,
      "learning_rate": 3.059519503227313e-06,
      "loss": 0.361,
      "num_input_tokens_seen": 14714472,
      "step": 22530
    },
    {
      "epoch": 13.287146226415095,
      "grad_norm": 8.554970741271973,
      "learning_rate": 3.057148698230393e-06,
      "loss": 0.2191,
      "num_input_tokens_seen": 14716840,
      "step": 22535
    },
    {
      "epoch": 13.290094339622641,
      "grad_norm": 3.132094383239746,
      "learning_rate": 3.0547784076117294e-06,
      "loss": 0.3205,
      "num_input_tokens_seen": 14719336,
      "step": 22540
    },
    {
      "epoch": 13.29304245283019,
      "grad_norm": 5.446914196014404,
      "learning_rate": 3.0524086319988635e-06,
      "loss": 0.3142,
      "num_input_tokens_seen": 14721896,
      "step": 22545
    },
    {
      "epoch": 13.295990566037736,
      "grad_norm": 3.332773208618164,
      "learning_rate": 3.0500393720192074e-06,
      "loss": 0.2795,
      "num_input_tokens_seen": 14725800,
      "step": 22550
    },
    {
      "epoch": 13.298938679245284,
      "grad_norm": 8.36528205871582,
      "learning_rate": 3.047670628300031e-06,
      "loss": 0.4061,
      "num_input_tokens_seen": 14729480,
      "step": 22555
    },
    {
      "epoch": 13.30188679245283,
      "grad_norm": 2.5486607551574707,
      "learning_rate": 3.0453024014684694e-06,
      "loss": 0.2783,
      "num_input_tokens_seen": 14732680,
      "step": 22560
    },
    {
      "epoch": 13.304834905660377,
      "grad_norm": 5.933586597442627,
      "learning_rate": 3.0429346921515225e-06,
      "loss": 0.3391,
      "num_input_tokens_seen": 14736488,
      "step": 22565
    },
    {
      "epoch": 13.307783018867925,
      "grad_norm": 7.017425537109375,
      "learning_rate": 3.04056750097605e-06,
      "loss": 0.3137,
      "num_input_tokens_seen": 14739976,
      "step": 22570
    },
    {
      "epoch": 13.310731132075471,
      "grad_norm": 2.932344436645508,
      "learning_rate": 3.0382008285687754e-06,
      "loss": 0.2779,
      "num_input_tokens_seen": 14742792,
      "step": 22575
    },
    {
      "epoch": 13.31367924528302,
      "grad_norm": 2.562885046005249,
      "learning_rate": 3.035834675556287e-06,
      "loss": 0.325,
      "num_input_tokens_seen": 14746824,
      "step": 22580
    },
    {
      "epoch": 13.316627358490566,
      "grad_norm": 3.401585817337036,
      "learning_rate": 3.0334690425650336e-06,
      "loss": 0.2529,
      "num_input_tokens_seen": 14750408,
      "step": 22585
    },
    {
      "epoch": 13.319575471698114,
      "grad_norm": 2.9116499423980713,
      "learning_rate": 3.031103930221325e-06,
      "loss": 0.3458,
      "num_input_tokens_seen": 14753352,
      "step": 22590
    },
    {
      "epoch": 13.32252358490566,
      "grad_norm": 3.032701015472412,
      "learning_rate": 3.028739339151338e-06,
      "loss": 0.2599,
      "num_input_tokens_seen": 14756264,
      "step": 22595
    },
    {
      "epoch": 13.325471698113208,
      "grad_norm": 2.722766876220703,
      "learning_rate": 3.0263752699811067e-06,
      "loss": 0.3395,
      "num_input_tokens_seen": 14759048,
      "step": 22600
    },
    {
      "epoch": 13.328419811320755,
      "grad_norm": 3.229226589202881,
      "learning_rate": 3.0240117233365267e-06,
      "loss": 0.2788,
      "num_input_tokens_seen": 14762248,
      "step": 22605
    },
    {
      "epoch": 13.331367924528301,
      "grad_norm": 2.351168155670166,
      "learning_rate": 3.0216486998433604e-06,
      "loss": 0.2963,
      "num_input_tokens_seen": 14765544,
      "step": 22610
    },
    {
      "epoch": 13.33431603773585,
      "grad_norm": 2.5547897815704346,
      "learning_rate": 3.0192862001272273e-06,
      "loss": 0.2308,
      "num_input_tokens_seen": 14768584,
      "step": 22615
    },
    {
      "epoch": 13.337264150943396,
      "grad_norm": 3.550908088684082,
      "learning_rate": 3.0169242248136066e-06,
      "loss": 0.3515,
      "num_input_tokens_seen": 14771496,
      "step": 22620
    },
    {
      "epoch": 13.340212264150944,
      "grad_norm": 3.6083719730377197,
      "learning_rate": 3.0145627745278457e-06,
      "loss": 0.3583,
      "num_input_tokens_seen": 14775016,
      "step": 22625
    },
    {
      "epoch": 13.34316037735849,
      "grad_norm": 1.5629523992538452,
      "learning_rate": 3.0122018498951478e-06,
      "loss": 0.4992,
      "num_input_tokens_seen": 14781096,
      "step": 22630
    },
    {
      "epoch": 13.346108490566039,
      "grad_norm": 7.021904468536377,
      "learning_rate": 3.0098414515405765e-06,
      "loss": 0.307,
      "num_input_tokens_seen": 14783528,
      "step": 22635
    },
    {
      "epoch": 13.349056603773585,
      "grad_norm": 3.5611305236816406,
      "learning_rate": 3.0074815800890576e-06,
      "loss": 0.3143,
      "num_input_tokens_seen": 14786728,
      "step": 22640
    },
    {
      "epoch": 13.352004716981131,
      "grad_norm": 4.927103042602539,
      "learning_rate": 3.005122236165378e-06,
      "loss": 0.3502,
      "num_input_tokens_seen": 14789480,
      "step": 22645
    },
    {
      "epoch": 13.35495283018868,
      "grad_norm": 2.151087522506714,
      "learning_rate": 3.0027634203941847e-06,
      "loss": 0.2423,
      "num_input_tokens_seen": 14792488,
      "step": 22650
    },
    {
      "epoch": 13.357900943396226,
      "grad_norm": 6.692821502685547,
      "learning_rate": 3.0004051333999816e-06,
      "loss": 0.3403,
      "num_input_tokens_seen": 14795368,
      "step": 22655
    },
    {
      "epoch": 13.360849056603774,
      "grad_norm": 5.832033634185791,
      "learning_rate": 2.998047375807139e-06,
      "loss": 0.3229,
      "num_input_tokens_seen": 14799112,
      "step": 22660
    },
    {
      "epoch": 13.36379716981132,
      "grad_norm": 7.336167812347412,
      "learning_rate": 2.995690148239881e-06,
      "loss": 0.2957,
      "num_input_tokens_seen": 14802920,
      "step": 22665
    },
    {
      "epoch": 13.366745283018869,
      "grad_norm": 2.8972487449645996,
      "learning_rate": 2.993333451322293e-06,
      "loss": 0.2759,
      "num_input_tokens_seen": 14805704,
      "step": 22670
    },
    {
      "epoch": 13.369693396226415,
      "grad_norm": 2.6393609046936035,
      "learning_rate": 2.9909772856783242e-06,
      "loss": 0.3963,
      "num_input_tokens_seen": 14808584,
      "step": 22675
    },
    {
      "epoch": 13.372641509433961,
      "grad_norm": 2.305396556854248,
      "learning_rate": 2.988621651931777e-06,
      "loss": 0.2733,
      "num_input_tokens_seen": 14813512,
      "step": 22680
    },
    {
      "epoch": 13.37558962264151,
      "grad_norm": 2.9063432216644287,
      "learning_rate": 2.986266550706315e-06,
      "loss": 0.3186,
      "num_input_tokens_seen": 14817096,
      "step": 22685
    },
    {
      "epoch": 13.378537735849056,
      "grad_norm": 11.045530319213867,
      "learning_rate": 2.9839119826254627e-06,
      "loss": 0.3839,
      "num_input_tokens_seen": 14819688,
      "step": 22690
    },
    {
      "epoch": 13.381485849056604,
      "grad_norm": 5.467602252960205,
      "learning_rate": 2.981557948312602e-06,
      "loss": 0.4517,
      "num_input_tokens_seen": 14822312,
      "step": 22695
    },
    {
      "epoch": 13.38443396226415,
      "grad_norm": 4.135236740112305,
      "learning_rate": 2.9792044483909733e-06,
      "loss": 0.3721,
      "num_input_tokens_seen": 14824744,
      "step": 22700
    },
    {
      "epoch": 13.387382075471699,
      "grad_norm": 3.174801826477051,
      "learning_rate": 2.9768514834836767e-06,
      "loss": 0.4582,
      "num_input_tokens_seen": 14828200,
      "step": 22705
    },
    {
      "epoch": 13.390330188679245,
      "grad_norm": 3.8671891689300537,
      "learning_rate": 2.9744990542136685e-06,
      "loss": 0.3289,
      "num_input_tokens_seen": 14831304,
      "step": 22710
    },
    {
      "epoch": 13.393278301886792,
      "grad_norm": 2.4437100887298584,
      "learning_rate": 2.9721471612037637e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 14835752,
      "step": 22715
    },
    {
      "epoch": 13.39622641509434,
      "grad_norm": 2.2491302490234375,
      "learning_rate": 2.9697958050766385e-06,
      "loss": 0.3673,
      "num_input_tokens_seen": 14838920,
      "step": 22720
    },
    {
      "epoch": 13.399174528301886,
      "grad_norm": 5.602293014526367,
      "learning_rate": 2.967444986454825e-06,
      "loss": 0.3679,
      "num_input_tokens_seen": 14842600,
      "step": 22725
    },
    {
      "epoch": 13.402122641509434,
      "grad_norm": 3.9482178688049316,
      "learning_rate": 2.9650947059607106e-06,
      "loss": 0.2756,
      "num_input_tokens_seen": 14845416,
      "step": 22730
    },
    {
      "epoch": 13.40507075471698,
      "grad_norm": 5.400810718536377,
      "learning_rate": 2.962744964216542e-06,
      "loss": 0.3059,
      "num_input_tokens_seen": 14847848,
      "step": 22735
    },
    {
      "epoch": 13.408018867924529,
      "grad_norm": 2.8984901905059814,
      "learning_rate": 2.960395761844425e-06,
      "loss": 0.3736,
      "num_input_tokens_seen": 14850856,
      "step": 22740
    },
    {
      "epoch": 13.410966981132075,
      "grad_norm": 4.835386276245117,
      "learning_rate": 2.95804709946632e-06,
      "loss": 0.4219,
      "num_input_tokens_seen": 14853704,
      "step": 22745
    },
    {
      "epoch": 13.413915094339623,
      "grad_norm": 2.659588575363159,
      "learning_rate": 2.9556989777040457e-06,
      "loss": 0.233,
      "num_input_tokens_seen": 14856872,
      "step": 22750
    },
    {
      "epoch": 13.41686320754717,
      "grad_norm": 5.155787944793701,
      "learning_rate": 2.9533513971792776e-06,
      "loss": 0.4429,
      "num_input_tokens_seen": 14859304,
      "step": 22755
    },
    {
      "epoch": 13.419811320754716,
      "grad_norm": 2.9190332889556885,
      "learning_rate": 2.9510043585135473e-06,
      "loss": 0.2641,
      "num_input_tokens_seen": 14862920,
      "step": 22760
    },
    {
      "epoch": 13.422759433962264,
      "grad_norm": 2.286158561706543,
      "learning_rate": 2.948657862328244e-06,
      "loss": 0.3865,
      "num_input_tokens_seen": 14867176,
      "step": 22765
    },
    {
      "epoch": 13.42570754716981,
      "grad_norm": 4.415739059448242,
      "learning_rate": 2.946311909244613e-06,
      "loss": 0.3741,
      "num_input_tokens_seen": 14870568,
      "step": 22770
    },
    {
      "epoch": 13.428655660377359,
      "grad_norm": 3.6429312229156494,
      "learning_rate": 2.9439664998837538e-06,
      "loss": 0.4991,
      "num_input_tokens_seen": 14873384,
      "step": 22775
    },
    {
      "epoch": 13.431603773584905,
      "grad_norm": 4.8949480056762695,
      "learning_rate": 2.941621634866626e-06,
      "loss": 0.3299,
      "num_input_tokens_seen": 14876648,
      "step": 22780
    },
    {
      "epoch": 13.434551886792454,
      "grad_norm": 2.4755232334136963,
      "learning_rate": 2.9392773148140406e-06,
      "loss": 0.3602,
      "num_input_tokens_seen": 14879976,
      "step": 22785
    },
    {
      "epoch": 13.4375,
      "grad_norm": 2.538907766342163,
      "learning_rate": 2.9369335403466676e-06,
      "loss": 0.293,
      "num_input_tokens_seen": 14882632,
      "step": 22790
    },
    {
      "epoch": 13.440448113207546,
      "grad_norm": 6.156617641448975,
      "learning_rate": 2.9345903120850318e-06,
      "loss": 0.5075,
      "num_input_tokens_seen": 14885480,
      "step": 22795
    },
    {
      "epoch": 13.443396226415095,
      "grad_norm": 2.482489585876465,
      "learning_rate": 2.932247630649512e-06,
      "loss": 0.3684,
      "num_input_tokens_seen": 14888328,
      "step": 22800
    },
    {
      "epoch": 13.446344339622641,
      "grad_norm": 3.7355949878692627,
      "learning_rate": 2.9299054966603424e-06,
      "loss": 0.3042,
      "num_input_tokens_seen": 14891688,
      "step": 22805
    },
    {
      "epoch": 13.44929245283019,
      "grad_norm": 1.738120675086975,
      "learning_rate": 2.927563910737613e-06,
      "loss": 0.29,
      "num_input_tokens_seen": 14894504,
      "step": 22810
    },
    {
      "epoch": 13.452240566037736,
      "grad_norm": 6.150461673736572,
      "learning_rate": 2.9252228735012722e-06,
      "loss": 0.2846,
      "num_input_tokens_seen": 14897128,
      "step": 22815
    },
    {
      "epoch": 13.455188679245284,
      "grad_norm": 3.227527379989624,
      "learning_rate": 2.9228823855711174e-06,
      "loss": 0.2392,
      "num_input_tokens_seen": 14902088,
      "step": 22820
    },
    {
      "epoch": 13.45813679245283,
      "grad_norm": 3.3321259021759033,
      "learning_rate": 2.920542447566802e-06,
      "loss": 0.3595,
      "num_input_tokens_seen": 14905064,
      "step": 22825
    },
    {
      "epoch": 13.461084905660377,
      "grad_norm": 4.2693305015563965,
      "learning_rate": 2.918203060107837e-06,
      "loss": 0.4227,
      "num_input_tokens_seen": 14908456,
      "step": 22830
    },
    {
      "epoch": 13.464033018867925,
      "grad_norm": 5.089859962463379,
      "learning_rate": 2.9158642238135813e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 14912456,
      "step": 22835
    },
    {
      "epoch": 13.466981132075471,
      "grad_norm": 5.6995015144348145,
      "learning_rate": 2.913525939303257e-06,
      "loss": 0.413,
      "num_input_tokens_seen": 14915656,
      "step": 22840
    },
    {
      "epoch": 13.46992924528302,
      "grad_norm": 4.336443901062012,
      "learning_rate": 2.9111882071959317e-06,
      "loss": 0.3186,
      "num_input_tokens_seen": 14918792,
      "step": 22845
    },
    {
      "epoch": 13.472877358490566,
      "grad_norm": 3.9792988300323486,
      "learning_rate": 2.908851028110532e-06,
      "loss": 0.3136,
      "num_input_tokens_seen": 14923112,
      "step": 22850
    },
    {
      "epoch": 13.475825471698114,
      "grad_norm": 5.973363399505615,
      "learning_rate": 2.906514402665834e-06,
      "loss": 0.2736,
      "num_input_tokens_seen": 14926632,
      "step": 22855
    },
    {
      "epoch": 13.47877358490566,
      "grad_norm": 2.341519355773926,
      "learning_rate": 2.9041783314804705e-06,
      "loss": 0.3132,
      "num_input_tokens_seen": 14929448,
      "step": 22860
    },
    {
      "epoch": 13.481721698113208,
      "grad_norm": 3.053133249282837,
      "learning_rate": 2.9018428151729238e-06,
      "loss": 0.3507,
      "num_input_tokens_seen": 14931752,
      "step": 22865
    },
    {
      "epoch": 13.484669811320755,
      "grad_norm": 3.11784029006958,
      "learning_rate": 2.899507854361537e-06,
      "loss": 0.2298,
      "num_input_tokens_seen": 14934376,
      "step": 22870
    },
    {
      "epoch": 13.487617924528301,
      "grad_norm": 3.128655195236206,
      "learning_rate": 2.8971734496644975e-06,
      "loss": 0.3315,
      "num_input_tokens_seen": 14937064,
      "step": 22875
    },
    {
      "epoch": 13.49056603773585,
      "grad_norm": 3.965510368347168,
      "learning_rate": 2.894839601699851e-06,
      "loss": 0.4637,
      "num_input_tokens_seen": 14940616,
      "step": 22880
    },
    {
      "epoch": 13.493514150943396,
      "grad_norm": 2.761537551879883,
      "learning_rate": 2.8925063110854923e-06,
      "loss": 0.361,
      "num_input_tokens_seen": 14944968,
      "step": 22885
    },
    {
      "epoch": 13.496462264150944,
      "grad_norm": 5.435567855834961,
      "learning_rate": 2.8901735784391683e-06,
      "loss": 0.4699,
      "num_input_tokens_seen": 14948488,
      "step": 22890
    },
    {
      "epoch": 13.49941037735849,
      "grad_norm": 3.0014123916625977,
      "learning_rate": 2.8878414043784844e-06,
      "loss": 0.4044,
      "num_input_tokens_seen": 14951176,
      "step": 22895
    },
    {
      "epoch": 13.502358490566039,
      "grad_norm": 3.4619250297546387,
      "learning_rate": 2.885509789520891e-06,
      "loss": 0.2662,
      "num_input_tokens_seen": 14953448,
      "step": 22900
    },
    {
      "epoch": 13.505306603773585,
      "grad_norm": 4.056725025177002,
      "learning_rate": 2.8831787344836926e-06,
      "loss": 0.3374,
      "num_input_tokens_seen": 14956232,
      "step": 22905
    },
    {
      "epoch": 13.508254716981131,
      "grad_norm": 2.493910789489746,
      "learning_rate": 2.880848239884049e-06,
      "loss": 0.3452,
      "num_input_tokens_seen": 14960072,
      "step": 22910
    },
    {
      "epoch": 13.51120283018868,
      "grad_norm": 1.4840056896209717,
      "learning_rate": 2.8785183063389667e-06,
      "loss": 0.2982,
      "num_input_tokens_seen": 14963944,
      "step": 22915
    },
    {
      "epoch": 13.514150943396226,
      "grad_norm": 5.421234130859375,
      "learning_rate": 2.876188934465306e-06,
      "loss": 0.3902,
      "num_input_tokens_seen": 14966888,
      "step": 22920
    },
    {
      "epoch": 13.517099056603774,
      "grad_norm": 5.5362548828125,
      "learning_rate": 2.8738601248797758e-06,
      "loss": 0.3042,
      "num_input_tokens_seen": 14969992,
      "step": 22925
    },
    {
      "epoch": 13.52004716981132,
      "grad_norm": 7.7347941398620605,
      "learning_rate": 2.8715318781989432e-06,
      "loss": 0.4694,
      "num_input_tokens_seen": 14972808,
      "step": 22930
    },
    {
      "epoch": 13.522995283018869,
      "grad_norm": 1.5919901132583618,
      "learning_rate": 2.869204195039219e-06,
      "loss": 0.2504,
      "num_input_tokens_seen": 14976264,
      "step": 22935
    },
    {
      "epoch": 13.525943396226415,
      "grad_norm": 3.2556636333465576,
      "learning_rate": 2.8668770760168673e-06,
      "loss": 0.3181,
      "num_input_tokens_seen": 14979912,
      "step": 22940
    },
    {
      "epoch": 13.528891509433961,
      "grad_norm": 3.1158010959625244,
      "learning_rate": 2.864550521748003e-06,
      "loss": 0.343,
      "num_input_tokens_seen": 14983176,
      "step": 22945
    },
    {
      "epoch": 13.53183962264151,
      "grad_norm": 1.6123318672180176,
      "learning_rate": 2.862224532848591e-06,
      "loss": 0.2512,
      "num_input_tokens_seen": 14986184,
      "step": 22950
    },
    {
      "epoch": 13.534787735849056,
      "grad_norm": 3.5667927265167236,
      "learning_rate": 2.8598991099344455e-06,
      "loss": 0.3083,
      "num_input_tokens_seen": 14989832,
      "step": 22955
    },
    {
      "epoch": 13.537735849056604,
      "grad_norm": 4.33881139755249,
      "learning_rate": 2.857574253621236e-06,
      "loss": 0.4403,
      "num_input_tokens_seen": 14992616,
      "step": 22960
    },
    {
      "epoch": 13.54068396226415,
      "grad_norm": 2.4982235431671143,
      "learning_rate": 2.855249964524476e-06,
      "loss": 0.4873,
      "num_input_tokens_seen": 14996424,
      "step": 22965
    },
    {
      "epoch": 13.543632075471699,
      "grad_norm": 5.1585469245910645,
      "learning_rate": 2.852926243259531e-06,
      "loss": 0.3503,
      "num_input_tokens_seen": 14999400,
      "step": 22970
    },
    {
      "epoch": 13.546580188679245,
      "grad_norm": 2.7232210636138916,
      "learning_rate": 2.850603090441617e-06,
      "loss": 0.3814,
      "num_input_tokens_seen": 15001928,
      "step": 22975
    },
    {
      "epoch": 13.549528301886792,
      "grad_norm": 5.4202094078063965,
      "learning_rate": 2.848280506685798e-06,
      "loss": 0.2914,
      "num_input_tokens_seen": 15005256,
      "step": 22980
    },
    {
      "epoch": 13.55247641509434,
      "grad_norm": 3.735093355178833,
      "learning_rate": 2.845958492606986e-06,
      "loss": 0.3507,
      "num_input_tokens_seen": 15009256,
      "step": 22985
    },
    {
      "epoch": 13.555424528301886,
      "grad_norm": 2.696908950805664,
      "learning_rate": 2.843637048819949e-06,
      "loss": 0.2759,
      "num_input_tokens_seen": 15012104,
      "step": 22990
    },
    {
      "epoch": 13.558372641509434,
      "grad_norm": 3.6791319847106934,
      "learning_rate": 2.8413161759392966e-06,
      "loss": 0.4352,
      "num_input_tokens_seen": 15015176,
      "step": 22995
    },
    {
      "epoch": 13.56132075471698,
      "grad_norm": 2.4970524311065674,
      "learning_rate": 2.8389958745794878e-06,
      "loss": 0.3515,
      "num_input_tokens_seen": 15018920,
      "step": 23000
    },
    {
      "epoch": 13.564268867924529,
      "grad_norm": 2.128286123275757,
      "learning_rate": 2.8366761453548366e-06,
      "loss": 0.3874,
      "num_input_tokens_seen": 15022536,
      "step": 23005
    },
    {
      "epoch": 13.567216981132075,
      "grad_norm": 3.6343510150909424,
      "learning_rate": 2.8343569888795e-06,
      "loss": 0.3192,
      "num_input_tokens_seen": 15025576,
      "step": 23010
    },
    {
      "epoch": 13.570165094339622,
      "grad_norm": 4.324986934661865,
      "learning_rate": 2.832038405767483e-06,
      "loss": 0.4975,
      "num_input_tokens_seen": 15032936,
      "step": 23015
    },
    {
      "epoch": 13.57311320754717,
      "grad_norm": 10.925804138183594,
      "learning_rate": 2.8297203966326397e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 15035336,
      "step": 23020
    },
    {
      "epoch": 13.576061320754716,
      "grad_norm": 5.401839256286621,
      "learning_rate": 2.8274029620886773e-06,
      "loss": 0.3654,
      "num_input_tokens_seen": 15038984,
      "step": 23025
    },
    {
      "epoch": 13.579009433962264,
      "grad_norm": 2.814840316772461,
      "learning_rate": 2.825086102749144e-06,
      "loss": 0.3568,
      "num_input_tokens_seen": 15042696,
      "step": 23030
    },
    {
      "epoch": 13.58195754716981,
      "grad_norm": 2.7198328971862793,
      "learning_rate": 2.822769819227438e-06,
      "loss": 0.3471,
      "num_input_tokens_seen": 15045512,
      "step": 23035
    },
    {
      "epoch": 13.584905660377359,
      "grad_norm": 5.235401153564453,
      "learning_rate": 2.8204541121368055e-06,
      "loss": 0.3416,
      "num_input_tokens_seen": 15048520,
      "step": 23040
    },
    {
      "epoch": 13.587853773584905,
      "grad_norm": 2.7315614223480225,
      "learning_rate": 2.8181389820903402e-06,
      "loss": 0.2729,
      "num_input_tokens_seen": 15051720,
      "step": 23045
    },
    {
      "epoch": 13.590801886792454,
      "grad_norm": 4.099570274353027,
      "learning_rate": 2.8158244297009814e-06,
      "loss": 0.3268,
      "num_input_tokens_seen": 15055304,
      "step": 23050
    },
    {
      "epoch": 13.59375,
      "grad_norm": 3.6875975131988525,
      "learning_rate": 2.8135104555815196e-06,
      "loss": 0.4602,
      "num_input_tokens_seen": 15058184,
      "step": 23055
    },
    {
      "epoch": 13.596698113207546,
      "grad_norm": 2.273271083831787,
      "learning_rate": 2.811197060344588e-06,
      "loss": 0.2435,
      "num_input_tokens_seen": 15061288,
      "step": 23060
    },
    {
      "epoch": 13.599646226415095,
      "grad_norm": 2.665348768234253,
      "learning_rate": 2.8088842446026677e-06,
      "loss": 0.2613,
      "num_input_tokens_seen": 15064488,
      "step": 23065
    },
    {
      "epoch": 13.602594339622641,
      "grad_norm": 7.053226470947266,
      "learning_rate": 2.806572008968087e-06,
      "loss": 0.2765,
      "num_input_tokens_seen": 15068680,
      "step": 23070
    },
    {
      "epoch": 13.60554245283019,
      "grad_norm": 2.7762746810913086,
      "learning_rate": 2.80426035405302e-06,
      "loss": 0.2525,
      "num_input_tokens_seen": 15071784,
      "step": 23075
    },
    {
      "epoch": 13.608490566037736,
      "grad_norm": 2.9505319595336914,
      "learning_rate": 2.8019492804694852e-06,
      "loss": 0.2494,
      "num_input_tokens_seen": 15074824,
      "step": 23080
    },
    {
      "epoch": 13.611438679245284,
      "grad_norm": 3.0616211891174316,
      "learning_rate": 2.799638788829354e-06,
      "loss": 0.304,
      "num_input_tokens_seen": 15077768,
      "step": 23085
    },
    {
      "epoch": 13.61438679245283,
      "grad_norm": 2.239469051361084,
      "learning_rate": 2.7973288797443367e-06,
      "loss": 0.2506,
      "num_input_tokens_seen": 15080712,
      "step": 23090
    },
    {
      "epoch": 13.617334905660378,
      "grad_norm": 2.850687026977539,
      "learning_rate": 2.7950195538259884e-06,
      "loss": 0.2947,
      "num_input_tokens_seen": 15083560,
      "step": 23095
    },
    {
      "epoch": 13.620283018867925,
      "grad_norm": 3.485285520553589,
      "learning_rate": 2.792710811685719e-06,
      "loss": 0.2852,
      "num_input_tokens_seen": 15087240,
      "step": 23100
    },
    {
      "epoch": 13.623231132075471,
      "grad_norm": 2.953608989715576,
      "learning_rate": 2.7904026539347743e-06,
      "loss": 0.2911,
      "num_input_tokens_seen": 15090440,
      "step": 23105
    },
    {
      "epoch": 13.62617924528302,
      "grad_norm": 3.5990209579467773,
      "learning_rate": 2.7880950811842507e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 15093992,
      "step": 23110
    },
    {
      "epoch": 13.629127358490566,
      "grad_norm": 5.496184349060059,
      "learning_rate": 2.785788094045085e-06,
      "loss": 0.4104,
      "num_input_tokens_seen": 15096904,
      "step": 23115
    },
    {
      "epoch": 13.632075471698114,
      "grad_norm": 3.197831630706787,
      "learning_rate": 2.7834816931280655e-06,
      "loss": 0.2685,
      "num_input_tokens_seen": 15100168,
      "step": 23120
    },
    {
      "epoch": 13.63502358490566,
      "grad_norm": 4.219309329986572,
      "learning_rate": 2.781175879043821e-06,
      "loss": 0.4806,
      "num_input_tokens_seen": 15102728,
      "step": 23125
    },
    {
      "epoch": 13.637971698113208,
      "grad_norm": 4.521852970123291,
      "learning_rate": 2.778870652402825e-06,
      "loss": 0.3423,
      "num_input_tokens_seen": 15105544,
      "step": 23130
    },
    {
      "epoch": 13.640919811320755,
      "grad_norm": 2.232077121734619,
      "learning_rate": 2.776566013815396e-06,
      "loss": 0.3124,
      "num_input_tokens_seen": 15109096,
      "step": 23135
    },
    {
      "epoch": 13.643867924528301,
      "grad_norm": 3.8576509952545166,
      "learning_rate": 2.774261963891698e-06,
      "loss": 0.3938,
      "num_input_tokens_seen": 15112840,
      "step": 23140
    },
    {
      "epoch": 13.64681603773585,
      "grad_norm": 3.055460214614868,
      "learning_rate": 2.771958503241735e-06,
      "loss": 0.3108,
      "num_input_tokens_seen": 15116200,
      "step": 23145
    },
    {
      "epoch": 13.649764150943396,
      "grad_norm": 3.8606085777282715,
      "learning_rate": 2.769655632475362e-06,
      "loss": 0.4542,
      "num_input_tokens_seen": 15118728,
      "step": 23150
    },
    {
      "epoch": 13.652712264150944,
      "grad_norm": 3.8641085624694824,
      "learning_rate": 2.7673533522022733e-06,
      "loss": 0.3012,
      "num_input_tokens_seen": 15121896,
      "step": 23155
    },
    {
      "epoch": 13.65566037735849,
      "grad_norm": 3.082353115081787,
      "learning_rate": 2.765051663032007e-06,
      "loss": 0.3006,
      "num_input_tokens_seen": 15124232,
      "step": 23160
    },
    {
      "epoch": 13.658608490566039,
      "grad_norm": 4.670108318328857,
      "learning_rate": 2.7627505655739446e-06,
      "loss": 0.249,
      "num_input_tokens_seen": 15126984,
      "step": 23165
    },
    {
      "epoch": 13.661556603773585,
      "grad_norm": 4.433457374572754,
      "learning_rate": 2.7604500604373097e-06,
      "loss": 0.391,
      "num_input_tokens_seen": 15130664,
      "step": 23170
    },
    {
      "epoch": 13.664504716981131,
      "grad_norm": 2.0076403617858887,
      "learning_rate": 2.7581501482311757e-06,
      "loss": 0.3589,
      "num_input_tokens_seen": 15133320,
      "step": 23175
    },
    {
      "epoch": 13.66745283018868,
      "grad_norm": 2.8566620349884033,
      "learning_rate": 2.7558508295644513e-06,
      "loss": 0.2966,
      "num_input_tokens_seen": 15136552,
      "step": 23180
    },
    {
      "epoch": 13.670400943396226,
      "grad_norm": 4.344300270080566,
      "learning_rate": 2.7535521050458922e-06,
      "loss": 0.2632,
      "num_input_tokens_seen": 15143208,
      "step": 23185
    },
    {
      "epoch": 13.673349056603774,
      "grad_norm": 3.4346835613250732,
      "learning_rate": 2.7512539752840926e-06,
      "loss": 0.2723,
      "num_input_tokens_seen": 15147240,
      "step": 23190
    },
    {
      "epoch": 13.67629716981132,
      "grad_norm": 3.64217209815979,
      "learning_rate": 2.748956440887497e-06,
      "loss": 0.4407,
      "num_input_tokens_seen": 15149736,
      "step": 23195
    },
    {
      "epoch": 13.679245283018869,
      "grad_norm": 6.360025882720947,
      "learning_rate": 2.7466595024643843e-06,
      "loss": 0.3704,
      "num_input_tokens_seen": 15152872,
      "step": 23200
    },
    {
      "epoch": 13.682193396226415,
      "grad_norm": 15.878799438476562,
      "learning_rate": 2.744363160622878e-06,
      "loss": 0.4685,
      "num_input_tokens_seen": 15155752,
      "step": 23205
    },
    {
      "epoch": 13.685141509433961,
      "grad_norm": 2.9222614765167236,
      "learning_rate": 2.742067415970948e-06,
      "loss": 0.3259,
      "num_input_tokens_seen": 15158600,
      "step": 23210
    },
    {
      "epoch": 13.68808962264151,
      "grad_norm": 4.539727210998535,
      "learning_rate": 2.739772269116402e-06,
      "loss": 0.3177,
      "num_input_tokens_seen": 15161608,
      "step": 23215
    },
    {
      "epoch": 13.691037735849056,
      "grad_norm": 2.921752452850342,
      "learning_rate": 2.7374777206668874e-06,
      "loss": 0.3805,
      "num_input_tokens_seen": 15164232,
      "step": 23220
    },
    {
      "epoch": 13.693985849056604,
      "grad_norm": 8.450603485107422,
      "learning_rate": 2.735183771229898e-06,
      "loss": 0.4193,
      "num_input_tokens_seen": 15167240,
      "step": 23225
    },
    {
      "epoch": 13.69693396226415,
      "grad_norm": 2.2667641639709473,
      "learning_rate": 2.732890421412765e-06,
      "loss": 0.2843,
      "num_input_tokens_seen": 15170568,
      "step": 23230
    },
    {
      "epoch": 13.699882075471699,
      "grad_norm": 2.742682695388794,
      "learning_rate": 2.7305976718226624e-06,
      "loss": 0.5135,
      "num_input_tokens_seen": 15175144,
      "step": 23235
    },
    {
      "epoch": 13.702830188679245,
      "grad_norm": 3.1275877952575684,
      "learning_rate": 2.728305523066609e-06,
      "loss": 0.3248,
      "num_input_tokens_seen": 15178664,
      "step": 23240
    },
    {
      "epoch": 13.705778301886792,
      "grad_norm": 4.941312789916992,
      "learning_rate": 2.726013975751458e-06,
      "loss": 0.2745,
      "num_input_tokens_seen": 15181160,
      "step": 23245
    },
    {
      "epoch": 13.70872641509434,
      "grad_norm": 2.426234722137451,
      "learning_rate": 2.723723030483908e-06,
      "loss": 0.3392,
      "num_input_tokens_seen": 15184392,
      "step": 23250
    },
    {
      "epoch": 13.711674528301886,
      "grad_norm": 3.2399232387542725,
      "learning_rate": 2.7214326878704953e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 15188520,
      "step": 23255
    },
    {
      "epoch": 13.714622641509434,
      "grad_norm": 3.921346426010132,
      "learning_rate": 2.7191429485175993e-06,
      "loss": 0.3898,
      "num_input_tokens_seen": 15191016,
      "step": 23260
    },
    {
      "epoch": 13.71757075471698,
      "grad_norm": 2.3347933292388916,
      "learning_rate": 2.716853813031435e-06,
      "loss": 0.2562,
      "num_input_tokens_seen": 15194248,
      "step": 23265
    },
    {
      "epoch": 13.720518867924529,
      "grad_norm": 2.8732569217681885,
      "learning_rate": 2.714565282018066e-06,
      "loss": 0.4399,
      "num_input_tokens_seen": 15197288,
      "step": 23270
    },
    {
      "epoch": 13.723466981132075,
      "grad_norm": 3.814807176589966,
      "learning_rate": 2.7122773560833877e-06,
      "loss": 0.2981,
      "num_input_tokens_seen": 15200904,
      "step": 23275
    },
    {
      "epoch": 13.726415094339622,
      "grad_norm": 3.655446767807007,
      "learning_rate": 2.709990035833139e-06,
      "loss": 0.2533,
      "num_input_tokens_seen": 15204072,
      "step": 23280
    },
    {
      "epoch": 13.72936320754717,
      "grad_norm": 2.7474193572998047,
      "learning_rate": 2.707703321872896e-06,
      "loss": 0.4192,
      "num_input_tokens_seen": 15207144,
      "step": 23285
    },
    {
      "epoch": 13.732311320754716,
      "grad_norm": 3.4528629779815674,
      "learning_rate": 2.705417214808079e-06,
      "loss": 0.3169,
      "num_input_tokens_seen": 15209832,
      "step": 23290
    },
    {
      "epoch": 13.735259433962264,
      "grad_norm": 2.9516336917877197,
      "learning_rate": 2.703131715243945e-06,
      "loss": 0.3322,
      "num_input_tokens_seen": 15212648,
      "step": 23295
    },
    {
      "epoch": 13.73820754716981,
      "grad_norm": 4.275572299957275,
      "learning_rate": 2.7008468237855855e-06,
      "loss": 0.278,
      "num_input_tokens_seen": 15216168,
      "step": 23300
    },
    {
      "epoch": 13.741155660377359,
      "grad_norm": 4.126182556152344,
      "learning_rate": 2.69856254103794e-06,
      "loss": 0.3287,
      "num_input_tokens_seen": 15218888,
      "step": 23305
    },
    {
      "epoch": 13.744103773584905,
      "grad_norm": 2.8749279975891113,
      "learning_rate": 2.6962788676057806e-06,
      "loss": 0.3739,
      "num_input_tokens_seen": 15221672,
      "step": 23310
    },
    {
      "epoch": 13.747051886792454,
      "grad_norm": 2.8239290714263916,
      "learning_rate": 2.69399580409372e-06,
      "loss": 0.3131,
      "num_input_tokens_seen": 15225352,
      "step": 23315
    },
    {
      "epoch": 13.75,
      "grad_norm": 2.651754379272461,
      "learning_rate": 2.6917133511062076e-06,
      "loss": 0.3032,
      "num_input_tokens_seen": 15228264,
      "step": 23320
    },
    {
      "epoch": 13.752948113207546,
      "grad_norm": 4.78381872177124,
      "learning_rate": 2.6894315092475342e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 15231560,
      "step": 23325
    },
    {
      "epoch": 13.755896226415095,
      "grad_norm": 4.6431684494018555,
      "learning_rate": 2.6871502791218245e-06,
      "loss": 0.2724,
      "num_input_tokens_seen": 15237192,
      "step": 23330
    },
    {
      "epoch": 13.758844339622641,
      "grad_norm": 3.291208028793335,
      "learning_rate": 2.684869661333048e-06,
      "loss": 0.4597,
      "num_input_tokens_seen": 15240200,
      "step": 23335
    },
    {
      "epoch": 13.76179245283019,
      "grad_norm": 3.223240852355957,
      "learning_rate": 2.6825896564850074e-06,
      "loss": 0.3341,
      "num_input_tokens_seen": 15243528,
      "step": 23340
    },
    {
      "epoch": 13.764740566037736,
      "grad_norm": 5.603666305541992,
      "learning_rate": 2.6803102651813416e-06,
      "loss": 0.3714,
      "num_input_tokens_seen": 15246120,
      "step": 23345
    },
    {
      "epoch": 13.767688679245284,
      "grad_norm": 2.7316973209381104,
      "learning_rate": 2.6780314880255307e-06,
      "loss": 0.2401,
      "num_input_tokens_seen": 15249896,
      "step": 23350
    },
    {
      "epoch": 13.77063679245283,
      "grad_norm": 3.015049934387207,
      "learning_rate": 2.675753325620891e-06,
      "loss": 0.3649,
      "num_input_tokens_seen": 15253576,
      "step": 23355
    },
    {
      "epoch": 13.773584905660378,
      "grad_norm": 3.319258451461792,
      "learning_rate": 2.6734757785705727e-06,
      "loss": 0.3845,
      "num_input_tokens_seen": 15257320,
      "step": 23360
    },
    {
      "epoch": 13.776533018867925,
      "grad_norm": 2.7273547649383545,
      "learning_rate": 2.6711988474775712e-06,
      "loss": 0.4991,
      "num_input_tokens_seen": 15260872,
      "step": 23365
    },
    {
      "epoch": 13.779481132075471,
      "grad_norm": 2.7415835857391357,
      "learning_rate": 2.668922532944711e-06,
      "loss": 0.2537,
      "num_input_tokens_seen": 15263976,
      "step": 23370
    },
    {
      "epoch": 13.78242924528302,
      "grad_norm": 3.642705202102661,
      "learning_rate": 2.6666468355746566e-06,
      "loss": 0.4049,
      "num_input_tokens_seen": 15267336,
      "step": 23375
    },
    {
      "epoch": 13.785377358490566,
      "grad_norm": 2.088280200958252,
      "learning_rate": 2.6643717559699073e-06,
      "loss": 0.3734,
      "num_input_tokens_seen": 15270312,
      "step": 23380
    },
    {
      "epoch": 13.788325471698114,
      "grad_norm": 3.0473570823669434,
      "learning_rate": 2.662097294732803e-06,
      "loss": 0.3298,
      "num_input_tokens_seen": 15273192,
      "step": 23385
    },
    {
      "epoch": 13.79127358490566,
      "grad_norm": 3.747938632965088,
      "learning_rate": 2.6598234524655165e-06,
      "loss": 0.3202,
      "num_input_tokens_seen": 15275624,
      "step": 23390
    },
    {
      "epoch": 13.794221698113208,
      "grad_norm": 2.1452693939208984,
      "learning_rate": 2.657550229770054e-06,
      "loss": 0.3269,
      "num_input_tokens_seen": 15278760,
      "step": 23395
    },
    {
      "epoch": 13.797169811320755,
      "grad_norm": 3.285217046737671,
      "learning_rate": 2.655277627248265e-06,
      "loss": 0.2838,
      "num_input_tokens_seen": 15283080,
      "step": 23400
    },
    {
      "epoch": 13.800117924528301,
      "grad_norm": 3.3791892528533936,
      "learning_rate": 2.65300564550183e-06,
      "loss": 0.3196,
      "num_input_tokens_seen": 15285768,
      "step": 23405
    },
    {
      "epoch": 13.80306603773585,
      "grad_norm": 4.2436089515686035,
      "learning_rate": 2.6507342851322647e-06,
      "loss": 0.3361,
      "num_input_tokens_seen": 15288616,
      "step": 23410
    },
    {
      "epoch": 13.806014150943396,
      "grad_norm": 4.9022932052612305,
      "learning_rate": 2.6484635467409233e-06,
      "loss": 0.3475,
      "num_input_tokens_seen": 15292264,
      "step": 23415
    },
    {
      "epoch": 13.808962264150944,
      "grad_norm": 3.644580841064453,
      "learning_rate": 2.64619343092899e-06,
      "loss": 0.3056,
      "num_input_tokens_seen": 15294664,
      "step": 23420
    },
    {
      "epoch": 13.81191037735849,
      "grad_norm": 4.194464683532715,
      "learning_rate": 2.643923938297492e-06,
      "loss": 0.3491,
      "num_input_tokens_seen": 15297288,
      "step": 23425
    },
    {
      "epoch": 13.814858490566039,
      "grad_norm": 4.232224941253662,
      "learning_rate": 2.6416550694472855e-06,
      "loss": 0.2772,
      "num_input_tokens_seen": 15300328,
      "step": 23430
    },
    {
      "epoch": 13.817806603773585,
      "grad_norm": 5.802033424377441,
      "learning_rate": 2.639386824979063e-06,
      "loss": 0.3236,
      "num_input_tokens_seen": 15303720,
      "step": 23435
    },
    {
      "epoch": 13.820754716981131,
      "grad_norm": 4.994040489196777,
      "learning_rate": 2.6371192054933525e-06,
      "loss": 0.3114,
      "num_input_tokens_seen": 15306824,
      "step": 23440
    },
    {
      "epoch": 13.82370283018868,
      "grad_norm": 4.669008731842041,
      "learning_rate": 2.634852211590516e-06,
      "loss": 0.3862,
      "num_input_tokens_seen": 15310472,
      "step": 23445
    },
    {
      "epoch": 13.826650943396226,
      "grad_norm": 3.0111196041107178,
      "learning_rate": 2.6325858438707473e-06,
      "loss": 0.3303,
      "num_input_tokens_seen": 15313896,
      "step": 23450
    },
    {
      "epoch": 13.829599056603774,
      "grad_norm": 2.1155176162719727,
      "learning_rate": 2.630320102934082e-06,
      "loss": 0.3356,
      "num_input_tokens_seen": 15316712,
      "step": 23455
    },
    {
      "epoch": 13.83254716981132,
      "grad_norm": 2.2951526641845703,
      "learning_rate": 2.628054989380382e-06,
      "loss": 0.2678,
      "num_input_tokens_seen": 15320360,
      "step": 23460
    },
    {
      "epoch": 13.835495283018869,
      "grad_norm": 4.330511093139648,
      "learning_rate": 2.625790503809346e-06,
      "loss": 0.4571,
      "num_input_tokens_seen": 15322952,
      "step": 23465
    },
    {
      "epoch": 13.838443396226415,
      "grad_norm": 2.4139559268951416,
      "learning_rate": 2.6235266468205067e-06,
      "loss": 0.3216,
      "num_input_tokens_seen": 15327080,
      "step": 23470
    },
    {
      "epoch": 13.841391509433961,
      "grad_norm": 5.090373992919922,
      "learning_rate": 2.621263419013227e-06,
      "loss": 0.4281,
      "num_input_tokens_seen": 15329640,
      "step": 23475
    },
    {
      "epoch": 13.84433962264151,
      "grad_norm": 5.41148567199707,
      "learning_rate": 2.619000820986711e-06,
      "loss": 0.2933,
      "num_input_tokens_seen": 15332232,
      "step": 23480
    },
    {
      "epoch": 13.847287735849056,
      "grad_norm": 2.7556118965148926,
      "learning_rate": 2.616738853339988e-06,
      "loss": 0.3308,
      "num_input_tokens_seen": 15335560,
      "step": 23485
    },
    {
      "epoch": 13.850235849056604,
      "grad_norm": 3.5513179302215576,
      "learning_rate": 2.614477516671926e-06,
      "loss": 0.3435,
      "num_input_tokens_seen": 15338536,
      "step": 23490
    },
    {
      "epoch": 13.85318396226415,
      "grad_norm": 8.474603652954102,
      "learning_rate": 2.612216811581223e-06,
      "loss": 0.3345,
      "num_input_tokens_seen": 15342216,
      "step": 23495
    },
    {
      "epoch": 13.856132075471699,
      "grad_norm": 5.50743293762207,
      "learning_rate": 2.6099567386664095e-06,
      "loss": 0.4646,
      "num_input_tokens_seen": 15345544,
      "step": 23500
    },
    {
      "epoch": 13.859080188679245,
      "grad_norm": 3.7315688133239746,
      "learning_rate": 2.60769729852585e-06,
      "loss": 0.3374,
      "num_input_tokens_seen": 15348296,
      "step": 23505
    },
    {
      "epoch": 13.862028301886792,
      "grad_norm": 5.208621501922607,
      "learning_rate": 2.6054384917577413e-06,
      "loss": 0.2977,
      "num_input_tokens_seen": 15351560,
      "step": 23510
    },
    {
      "epoch": 13.86497641509434,
      "grad_norm": 2.850694417953491,
      "learning_rate": 2.60318031896011e-06,
      "loss": 0.3796,
      "num_input_tokens_seen": 15354600,
      "step": 23515
    },
    {
      "epoch": 13.867924528301886,
      "grad_norm": 9.681855201721191,
      "learning_rate": 2.60092278073082e-06,
      "loss": 0.3804,
      "num_input_tokens_seen": 15357352,
      "step": 23520
    },
    {
      "epoch": 13.870872641509434,
      "grad_norm": 7.315431118011475,
      "learning_rate": 2.5986658776675644e-06,
      "loss": 0.5134,
      "num_input_tokens_seen": 15360872,
      "step": 23525
    },
    {
      "epoch": 13.87382075471698,
      "grad_norm": 2.3097472190856934,
      "learning_rate": 2.5964096103678666e-06,
      "loss": 0.254,
      "num_input_tokens_seen": 15363496,
      "step": 23530
    },
    {
      "epoch": 13.876768867924529,
      "grad_norm": 3.9536099433898926,
      "learning_rate": 2.5941539794290833e-06,
      "loss": 0.3577,
      "num_input_tokens_seen": 15366568,
      "step": 23535
    },
    {
      "epoch": 13.879716981132075,
      "grad_norm": 4.6044840812683105,
      "learning_rate": 2.5918989854484024e-06,
      "loss": 0.3424,
      "num_input_tokens_seen": 15370248,
      "step": 23540
    },
    {
      "epoch": 13.882665094339622,
      "grad_norm": 3.61196231842041,
      "learning_rate": 2.5896446290228417e-06,
      "loss": 0.2838,
      "num_input_tokens_seen": 15373352,
      "step": 23545
    },
    {
      "epoch": 13.88561320754717,
      "grad_norm": 2.695436954498291,
      "learning_rate": 2.5873909107492547e-06,
      "loss": 0.2772,
      "num_input_tokens_seen": 15376552,
      "step": 23550
    },
    {
      "epoch": 13.888561320754716,
      "grad_norm": 3.637519359588623,
      "learning_rate": 2.5851378312243224e-06,
      "loss": 0.4325,
      "num_input_tokens_seen": 15380232,
      "step": 23555
    },
    {
      "epoch": 13.891509433962264,
      "grad_norm": 2.0695247650146484,
      "learning_rate": 2.5828853910445572e-06,
      "loss": 0.3464,
      "num_input_tokens_seen": 15383592,
      "step": 23560
    },
    {
      "epoch": 13.89445754716981,
      "grad_norm": 5.051840305328369,
      "learning_rate": 2.5806335908063012e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 15385768,
      "step": 23565
    },
    {
      "epoch": 13.897405660377359,
      "grad_norm": 3.7241151332855225,
      "learning_rate": 2.5783824311057293e-06,
      "loss": 0.4094,
      "num_input_tokens_seen": 15389032,
      "step": 23570
    },
    {
      "epoch": 13.900353773584905,
      "grad_norm": 3.376349925994873,
      "learning_rate": 2.5761319125388433e-06,
      "loss": 0.3565,
      "num_input_tokens_seen": 15393000,
      "step": 23575
    },
    {
      "epoch": 13.903301886792454,
      "grad_norm": 4.668268203735352,
      "learning_rate": 2.57388203570148e-06,
      "loss": 0.3611,
      "num_input_tokens_seen": 15396648,
      "step": 23580
    },
    {
      "epoch": 13.90625,
      "grad_norm": 4.283264636993408,
      "learning_rate": 2.5716328011893055e-06,
      "loss": 0.4393,
      "num_input_tokens_seen": 15399496,
      "step": 23585
    },
    {
      "epoch": 13.909198113207546,
      "grad_norm": 2.6156091690063477,
      "learning_rate": 2.5693842095978127e-06,
      "loss": 0.3381,
      "num_input_tokens_seen": 15402504,
      "step": 23590
    },
    {
      "epoch": 13.912146226415095,
      "grad_norm": 4.564388751983643,
      "learning_rate": 2.567136261522325e-06,
      "loss": 0.2975,
      "num_input_tokens_seen": 15405896,
      "step": 23595
    },
    {
      "epoch": 13.915094339622641,
      "grad_norm": 3.2943637371063232,
      "learning_rate": 2.5648889575579985e-06,
      "loss": 0.3256,
      "num_input_tokens_seen": 15409640,
      "step": 23600
    },
    {
      "epoch": 13.91804245283019,
      "grad_norm": 2.9994988441467285,
      "learning_rate": 2.562642298299814e-06,
      "loss": 0.3153,
      "num_input_tokens_seen": 15412520,
      "step": 23605
    },
    {
      "epoch": 13.920990566037736,
      "grad_norm": 4.007449150085449,
      "learning_rate": 2.560396284342584e-06,
      "loss": 0.4431,
      "num_input_tokens_seen": 15415560,
      "step": 23610
    },
    {
      "epoch": 13.923938679245284,
      "grad_norm": 3.6312448978424072,
      "learning_rate": 2.558150916280954e-06,
      "loss": 0.3398,
      "num_input_tokens_seen": 15418216,
      "step": 23615
    },
    {
      "epoch": 13.92688679245283,
      "grad_norm": 4.2417449951171875,
      "learning_rate": 2.555906194709392e-06,
      "loss": 0.3278,
      "num_input_tokens_seen": 15421512,
      "step": 23620
    },
    {
      "epoch": 13.929834905660378,
      "grad_norm": 4.329840183258057,
      "learning_rate": 2.553662120222199e-06,
      "loss": 0.2963,
      "num_input_tokens_seen": 15424712,
      "step": 23625
    },
    {
      "epoch": 13.932783018867925,
      "grad_norm": 5.724375247955322,
      "learning_rate": 2.5514186934135026e-06,
      "loss": 0.3759,
      "num_input_tokens_seen": 15428360,
      "step": 23630
    },
    {
      "epoch": 13.935731132075471,
      "grad_norm": 1.7934311628341675,
      "learning_rate": 2.54917591487726e-06,
      "loss": 0.2503,
      "num_input_tokens_seen": 15432392,
      "step": 23635
    },
    {
      "epoch": 13.93867924528302,
      "grad_norm": 7.453405380249023,
      "learning_rate": 2.5469337852072547e-06,
      "loss": 0.3489,
      "num_input_tokens_seen": 15436136,
      "step": 23640
    },
    {
      "epoch": 13.941627358490566,
      "grad_norm": 4.376336574554443,
      "learning_rate": 2.5446923049971035e-06,
      "loss": 0.4321,
      "num_input_tokens_seen": 15438792,
      "step": 23645
    },
    {
      "epoch": 13.944575471698114,
      "grad_norm": 2.8678040504455566,
      "learning_rate": 2.5424514748402463e-06,
      "loss": 0.2789,
      "num_input_tokens_seen": 15441352,
      "step": 23650
    },
    {
      "epoch": 13.94752358490566,
      "grad_norm": 4.922937870025635,
      "learning_rate": 2.540211295329953e-06,
      "loss": 0.3534,
      "num_input_tokens_seen": 15443720,
      "step": 23655
    },
    {
      "epoch": 13.950471698113208,
      "grad_norm": 3.932076930999756,
      "learning_rate": 2.5379717670593197e-06,
      "loss": 0.3427,
      "num_input_tokens_seen": 15447016,
      "step": 23660
    },
    {
      "epoch": 13.953419811320755,
      "grad_norm": 3.239251136779785,
      "learning_rate": 2.53573289062127e-06,
      "loss": 0.3692,
      "num_input_tokens_seen": 15450024,
      "step": 23665
    },
    {
      "epoch": 13.956367924528301,
      "grad_norm": 3.143324851989746,
      "learning_rate": 2.5334946666085605e-06,
      "loss": 0.3301,
      "num_input_tokens_seen": 15453576,
      "step": 23670
    },
    {
      "epoch": 13.95931603773585,
      "grad_norm": 3.1578140258789062,
      "learning_rate": 2.531257095613766e-06,
      "loss": 0.2432,
      "num_input_tokens_seen": 15457480,
      "step": 23675
    },
    {
      "epoch": 13.962264150943396,
      "grad_norm": 3.1515250205993652,
      "learning_rate": 2.529020178229297e-06,
      "loss": 0.278,
      "num_input_tokens_seen": 15461000,
      "step": 23680
    },
    {
      "epoch": 13.965212264150944,
      "grad_norm": 3.1113359928131104,
      "learning_rate": 2.5267839150473846e-06,
      "loss": 0.3695,
      "num_input_tokens_seen": 15464712,
      "step": 23685
    },
    {
      "epoch": 13.96816037735849,
      "grad_norm": 4.312508583068848,
      "learning_rate": 2.5245483066600896e-06,
      "loss": 0.4236,
      "num_input_tokens_seen": 15468648,
      "step": 23690
    },
    {
      "epoch": 13.971108490566039,
      "grad_norm": 5.291144847869873,
      "learning_rate": 2.5223133536592996e-06,
      "loss": 0.2932,
      "num_input_tokens_seen": 15471240,
      "step": 23695
    },
    {
      "epoch": 13.974056603773585,
      "grad_norm": 3.8239948749542236,
      "learning_rate": 2.520079056636725e-06,
      "loss": 0.308,
      "num_input_tokens_seen": 15474824,
      "step": 23700
    },
    {
      "epoch": 13.977004716981131,
      "grad_norm": 4.0720295906066895,
      "learning_rate": 2.5178454161839106e-06,
      "loss": 0.2688,
      "num_input_tokens_seen": 15477608,
      "step": 23705
    },
    {
      "epoch": 13.97995283018868,
      "grad_norm": 2.5558931827545166,
      "learning_rate": 2.5156124328922195e-06,
      "loss": 0.3269,
      "num_input_tokens_seen": 15480872,
      "step": 23710
    },
    {
      "epoch": 13.982900943396226,
      "grad_norm": 7.178508281707764,
      "learning_rate": 2.513380107352844e-06,
      "loss": 0.3005,
      "num_input_tokens_seen": 15483656,
      "step": 23715
    },
    {
      "epoch": 13.985849056603774,
      "grad_norm": 2.667046546936035,
      "learning_rate": 2.5111484401568014e-06,
      "loss": 0.2763,
      "num_input_tokens_seen": 15487016,
      "step": 23720
    },
    {
      "epoch": 13.98879716981132,
      "grad_norm": 2.7007172107696533,
      "learning_rate": 2.508917431894936e-06,
      "loss": 0.3378,
      "num_input_tokens_seen": 15489992,
      "step": 23725
    },
    {
      "epoch": 13.991745283018869,
      "grad_norm": 2.756859540939331,
      "learning_rate": 2.5066870831579144e-06,
      "loss": 0.2797,
      "num_input_tokens_seen": 15492872,
      "step": 23730
    },
    {
      "epoch": 13.994693396226415,
      "grad_norm": 3.462620496749878,
      "learning_rate": 2.504457394536235e-06,
      "loss": 0.3308,
      "num_input_tokens_seen": 15495496,
      "step": 23735
    },
    {
      "epoch": 13.997641509433961,
      "grad_norm": 2.9258315563201904,
      "learning_rate": 2.502228366620216e-06,
      "loss": 0.2964,
      "num_input_tokens_seen": 15498760,
      "step": 23740
    },
    {
      "epoch": 14.0,
      "eval_loss": 0.5794597268104553,
      "eval_runtime": 19.2741,
      "eval_samples_per_second": 87.994,
      "eval_steps_per_second": 21.998,
      "num_input_tokens_seen": 15500632,
      "step": 23744
    },
    {
      "epoch": 14.00058962264151,
      "grad_norm": 2.4230945110321045,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.2296,
      "num_input_tokens_seen": 15501336,
      "step": 23745
    },
    {
      "epoch": 14.003537735849056,
      "grad_norm": 3.0880932807922363,
      "learning_rate": 2.497772295265561e-06,
      "loss": 0.4409,
      "num_input_tokens_seen": 15504728,
      "step": 23750
    },
    {
      "epoch": 14.006485849056604,
      "grad_norm": 4.411482810974121,
      "learning_rate": 2.4955452530066897e-06,
      "loss": 0.5093,
      "num_input_tokens_seen": 15508280,
      "step": 23755
    },
    {
      "epoch": 14.00943396226415,
      "grad_norm": 3.2562484741210938,
      "learning_rate": 2.4933188738130043e-06,
      "loss": 0.4395,
      "num_input_tokens_seen": 15511224,
      "step": 23760
    },
    {
      "epoch": 14.012382075471699,
      "grad_norm": 3.3255534172058105,
      "learning_rate": 2.49109315827395e-06,
      "loss": 0.2503,
      "num_input_tokens_seen": 15514008,
      "step": 23765
    },
    {
      "epoch": 14.015330188679245,
      "grad_norm": 1.890016794204712,
      "learning_rate": 2.4888681069787975e-06,
      "loss": 0.2367,
      "num_input_tokens_seen": 15517016,
      "step": 23770
    },
    {
      "epoch": 14.018278301886792,
      "grad_norm": 4.630746841430664,
      "learning_rate": 2.4866437205166353e-06,
      "loss": 0.2587,
      "num_input_tokens_seen": 15520632,
      "step": 23775
    },
    {
      "epoch": 14.02122641509434,
      "grad_norm": 3.211172342300415,
      "learning_rate": 2.4844199994763803e-06,
      "loss": 0.262,
      "num_input_tokens_seen": 15523832,
      "step": 23780
    },
    {
      "epoch": 14.024174528301886,
      "grad_norm": 3.8708865642547607,
      "learning_rate": 2.482196944446772e-06,
      "loss": 0.3154,
      "num_input_tokens_seen": 15526776,
      "step": 23785
    },
    {
      "epoch": 14.027122641509434,
      "grad_norm": 4.527602195739746,
      "learning_rate": 2.4799745560163736e-06,
      "loss": 0.3213,
      "num_input_tokens_seen": 15530328,
      "step": 23790
    },
    {
      "epoch": 14.03007075471698,
      "grad_norm": 2.807539463043213,
      "learning_rate": 2.4777528347735707e-06,
      "loss": 0.3617,
      "num_input_tokens_seen": 15533624,
      "step": 23795
    },
    {
      "epoch": 14.033018867924529,
      "grad_norm": 3.2349693775177,
      "learning_rate": 2.4755317813065766e-06,
      "loss": 0.3075,
      "num_input_tokens_seen": 15536696,
      "step": 23800
    },
    {
      "epoch": 14.035966981132075,
      "grad_norm": 4.578696250915527,
      "learning_rate": 2.4733113962034234e-06,
      "loss": 0.3541,
      "num_input_tokens_seen": 15539640,
      "step": 23805
    },
    {
      "epoch": 14.038915094339623,
      "grad_norm": 3.1331701278686523,
      "learning_rate": 2.4710916800519674e-06,
      "loss": 0.314,
      "num_input_tokens_seen": 15542648,
      "step": 23810
    },
    {
      "epoch": 14.04186320754717,
      "grad_norm": 4.700671672821045,
      "learning_rate": 2.4688726334398883e-06,
      "loss": 0.3291,
      "num_input_tokens_seen": 15545912,
      "step": 23815
    },
    {
      "epoch": 14.044811320754716,
      "grad_norm": 4.492247581481934,
      "learning_rate": 2.466654256954688e-06,
      "loss": 0.3219,
      "num_input_tokens_seen": 15549016,
      "step": 23820
    },
    {
      "epoch": 14.047759433962264,
      "grad_norm": 2.913160562515259,
      "learning_rate": 2.4644365511836895e-06,
      "loss": 0.2469,
      "num_input_tokens_seen": 15552216,
      "step": 23825
    },
    {
      "epoch": 14.05070754716981,
      "grad_norm": 3.5613784790039062,
      "learning_rate": 2.4622195167140432e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 15555640,
      "step": 23830
    },
    {
      "epoch": 14.053655660377359,
      "grad_norm": 4.012432098388672,
      "learning_rate": 2.4600031541327173e-06,
      "loss": 0.3689,
      "num_input_tokens_seen": 15558424,
      "step": 23835
    },
    {
      "epoch": 14.056603773584905,
      "grad_norm": 3.5178496837615967,
      "learning_rate": 2.457787464026503e-06,
      "loss": 0.2814,
      "num_input_tokens_seen": 15561112,
      "step": 23840
    },
    {
      "epoch": 14.059551886792454,
      "grad_norm": 2.2287051677703857,
      "learning_rate": 2.455572446982014e-06,
      "loss": 0.2921,
      "num_input_tokens_seen": 15564376,
      "step": 23845
    },
    {
      "epoch": 14.0625,
      "grad_norm": 3.907405376434326,
      "learning_rate": 2.453358103585686e-06,
      "loss": 0.3222,
      "num_input_tokens_seen": 15567928,
      "step": 23850
    },
    {
      "epoch": 14.065448113207546,
      "grad_norm": 4.563554286956787,
      "learning_rate": 2.4511444344237733e-06,
      "loss": 0.4102,
      "num_input_tokens_seen": 15574104,
      "step": 23855
    },
    {
      "epoch": 14.068396226415095,
      "grad_norm": 3.8091859817504883,
      "learning_rate": 2.4489314400823567e-06,
      "loss": 0.4085,
      "num_input_tokens_seen": 15577592,
      "step": 23860
    },
    {
      "epoch": 14.071344339622641,
      "grad_norm": 2.272507429122925,
      "learning_rate": 2.446719121147337e-06,
      "loss": 0.3375,
      "num_input_tokens_seen": 15580600,
      "step": 23865
    },
    {
      "epoch": 14.07429245283019,
      "grad_norm": 2.571584701538086,
      "learning_rate": 2.4445074782044347e-06,
      "loss": 0.2539,
      "num_input_tokens_seen": 15583224,
      "step": 23870
    },
    {
      "epoch": 14.077240566037736,
      "grad_norm": 3.227062463760376,
      "learning_rate": 2.442296511839191e-06,
      "loss": 0.3206,
      "num_input_tokens_seen": 15586264,
      "step": 23875
    },
    {
      "epoch": 14.080188679245284,
      "grad_norm": 2.988308906555176,
      "learning_rate": 2.4400862226369687e-06,
      "loss": 0.4368,
      "num_input_tokens_seen": 15589528,
      "step": 23880
    },
    {
      "epoch": 14.08313679245283,
      "grad_norm": 4.667728424072266,
      "learning_rate": 2.4378766111829514e-06,
      "loss": 0.3599,
      "num_input_tokens_seen": 15592216,
      "step": 23885
    },
    {
      "epoch": 14.086084905660377,
      "grad_norm": 4.95134162902832,
      "learning_rate": 2.435667678062142e-06,
      "loss": 0.3327,
      "num_input_tokens_seen": 15595672,
      "step": 23890
    },
    {
      "epoch": 14.089033018867925,
      "grad_norm": 3.828526258468628,
      "learning_rate": 2.4334594238593682e-06,
      "loss": 0.3362,
      "num_input_tokens_seen": 15599768,
      "step": 23895
    },
    {
      "epoch": 14.091981132075471,
      "grad_norm": 5.341838359832764,
      "learning_rate": 2.4312518491592727e-06,
      "loss": 0.3027,
      "num_input_tokens_seen": 15602872,
      "step": 23900
    },
    {
      "epoch": 14.09492924528302,
      "grad_norm": 6.476365089416504,
      "learning_rate": 2.429044954546322e-06,
      "loss": 0.3182,
      "num_input_tokens_seen": 15606936,
      "step": 23905
    },
    {
      "epoch": 14.097877358490566,
      "grad_norm": 3.139141321182251,
      "learning_rate": 2.426838740604799e-06,
      "loss": 0.2793,
      "num_input_tokens_seen": 15609752,
      "step": 23910
    },
    {
      "epoch": 14.100825471698114,
      "grad_norm": 4.3305816650390625,
      "learning_rate": 2.4246332079188066e-06,
      "loss": 0.2909,
      "num_input_tokens_seen": 15613720,
      "step": 23915
    },
    {
      "epoch": 14.10377358490566,
      "grad_norm": 4.007850170135498,
      "learning_rate": 2.4224283570722745e-06,
      "loss": 0.3562,
      "num_input_tokens_seen": 15616632,
      "step": 23920
    },
    {
      "epoch": 14.106721698113208,
      "grad_norm": 3.338759422302246,
      "learning_rate": 2.420224188648943e-06,
      "loss": 0.3721,
      "num_input_tokens_seen": 15620312,
      "step": 23925
    },
    {
      "epoch": 14.109669811320755,
      "grad_norm": 3.2355918884277344,
      "learning_rate": 2.418020703232376e-06,
      "loss": 0.2583,
      "num_input_tokens_seen": 15623736,
      "step": 23930
    },
    {
      "epoch": 14.112617924528301,
      "grad_norm": 5.2772111892700195,
      "learning_rate": 2.4158179014059556e-06,
      "loss": 0.3596,
      "num_input_tokens_seen": 15626936,
      "step": 23935
    },
    {
      "epoch": 14.11556603773585,
      "grad_norm": 5.276137351989746,
      "learning_rate": 2.413615783752883e-06,
      "loss": 0.3877,
      "num_input_tokens_seen": 15630040,
      "step": 23940
    },
    {
      "epoch": 14.118514150943396,
      "grad_norm": 6.82316780090332,
      "learning_rate": 2.4114143508561767e-06,
      "loss": 0.3231,
      "num_input_tokens_seen": 15633336,
      "step": 23945
    },
    {
      "epoch": 14.121462264150944,
      "grad_norm": 2.8025989532470703,
      "learning_rate": 2.4092136032986783e-06,
      "loss": 0.3425,
      "num_input_tokens_seen": 15636792,
      "step": 23950
    },
    {
      "epoch": 14.12441037735849,
      "grad_norm": 3.428359270095825,
      "learning_rate": 2.407013541663043e-06,
      "loss": 0.291,
      "num_input_tokens_seen": 15640696,
      "step": 23955
    },
    {
      "epoch": 14.127358490566039,
      "grad_norm": 3.9980931282043457,
      "learning_rate": 2.4048141665317493e-06,
      "loss": 0.3281,
      "num_input_tokens_seen": 15644312,
      "step": 23960
    },
    {
      "epoch": 14.130306603773585,
      "grad_norm": 2.9106061458587646,
      "learning_rate": 2.40261547848709e-06,
      "loss": 0.3335,
      "num_input_tokens_seen": 15646840,
      "step": 23965
    },
    {
      "epoch": 14.133254716981131,
      "grad_norm": 2.877027988433838,
      "learning_rate": 2.400417478111176e-06,
      "loss": 0.2666,
      "num_input_tokens_seen": 15649464,
      "step": 23970
    },
    {
      "epoch": 14.13620283018868,
      "grad_norm": 4.340320587158203,
      "learning_rate": 2.3982201659859387e-06,
      "loss": 0.2872,
      "num_input_tokens_seen": 15653432,
      "step": 23975
    },
    {
      "epoch": 14.139150943396226,
      "grad_norm": 4.135821342468262,
      "learning_rate": 2.3960235426931237e-06,
      "loss": 0.4049,
      "num_input_tokens_seen": 15657112,
      "step": 23980
    },
    {
      "epoch": 14.142099056603774,
      "grad_norm": 3.5019288063049316,
      "learning_rate": 2.3938276088143003e-06,
      "loss": 0.3375,
      "num_input_tokens_seen": 15660632,
      "step": 23985
    },
    {
      "epoch": 14.14504716981132,
      "grad_norm": 4.287090301513672,
      "learning_rate": 2.391632364930849e-06,
      "loss": 0.4826,
      "num_input_tokens_seen": 15663064,
      "step": 23990
    },
    {
      "epoch": 14.147995283018869,
      "grad_norm": 2.103794813156128,
      "learning_rate": 2.3894378116239706e-06,
      "loss": 0.357,
      "num_input_tokens_seen": 15666168,
      "step": 23995
    },
    {
      "epoch": 14.150943396226415,
      "grad_norm": 4.531780242919922,
      "learning_rate": 2.387243949474683e-06,
      "loss": 0.5006,
      "num_input_tokens_seen": 15669528,
      "step": 24000
    },
    {
      "epoch": 14.153891509433961,
      "grad_norm": 3.19777512550354,
      "learning_rate": 2.38505077906382e-06,
      "loss": 0.2869,
      "num_input_tokens_seen": 15673016,
      "step": 24005
    },
    {
      "epoch": 14.15683962264151,
      "grad_norm": 2.9102702140808105,
      "learning_rate": 2.382858300972031e-06,
      "loss": 0.2432,
      "num_input_tokens_seen": 15676696,
      "step": 24010
    },
    {
      "epoch": 14.159787735849056,
      "grad_norm": 2.849848747253418,
      "learning_rate": 2.380666515779788e-06,
      "loss": 0.4,
      "num_input_tokens_seen": 15680600,
      "step": 24015
    },
    {
      "epoch": 14.162735849056604,
      "grad_norm": 2.669362783432007,
      "learning_rate": 2.3784754240673734e-06,
      "loss": 0.2617,
      "num_input_tokens_seen": 15683800,
      "step": 24020
    },
    {
      "epoch": 14.16568396226415,
      "grad_norm": 3.906259059906006,
      "learning_rate": 2.3762850264148883e-06,
      "loss": 0.3724,
      "num_input_tokens_seen": 15686616,
      "step": 24025
    },
    {
      "epoch": 14.168632075471699,
      "grad_norm": 2.4967591762542725,
      "learning_rate": 2.374095323402251e-06,
      "loss": 0.2792,
      "num_input_tokens_seen": 15690872,
      "step": 24030
    },
    {
      "epoch": 14.171580188679245,
      "grad_norm": 5.160085678100586,
      "learning_rate": 2.371906315609193e-06,
      "loss": 0.3147,
      "num_input_tokens_seen": 15694104,
      "step": 24035
    },
    {
      "epoch": 14.174528301886792,
      "grad_norm": 3.1551268100738525,
      "learning_rate": 2.369718003615263e-06,
      "loss": 0.4198,
      "num_input_tokens_seen": 15697496,
      "step": 24040
    },
    {
      "epoch": 14.17747641509434,
      "grad_norm": 4.523599624633789,
      "learning_rate": 2.3675303879998284e-06,
      "loss": 0.3282,
      "num_input_tokens_seen": 15701016,
      "step": 24045
    },
    {
      "epoch": 14.180424528301886,
      "grad_norm": 6.506444454193115,
      "learning_rate": 2.365343469342068e-06,
      "loss": 0.4184,
      "num_input_tokens_seen": 15703512,
      "step": 24050
    },
    {
      "epoch": 14.183372641509434,
      "grad_norm": 3.0132546424865723,
      "learning_rate": 2.3631572482209803e-06,
      "loss": 0.2672,
      "num_input_tokens_seen": 15706936,
      "step": 24055
    },
    {
      "epoch": 14.18632075471698,
      "grad_norm": 3.9569203853607178,
      "learning_rate": 2.3609717252153752e-06,
      "loss": 0.3647,
      "num_input_tokens_seen": 15709560,
      "step": 24060
    },
    {
      "epoch": 14.189268867924529,
      "grad_norm": 2.8083577156066895,
      "learning_rate": 2.35878690090388e-06,
      "loss": 0.3087,
      "num_input_tokens_seen": 15712632,
      "step": 24065
    },
    {
      "epoch": 14.192216981132075,
      "grad_norm": 2.3957791328430176,
      "learning_rate": 2.356602775864935e-06,
      "loss": 0.3228,
      "num_input_tokens_seen": 15715800,
      "step": 24070
    },
    {
      "epoch": 14.195165094339623,
      "grad_norm": 2.199737310409546,
      "learning_rate": 2.354419350676796e-06,
      "loss": 0.348,
      "num_input_tokens_seen": 15718936,
      "step": 24075
    },
    {
      "epoch": 14.19811320754717,
      "grad_norm": 2.7082836627960205,
      "learning_rate": 2.3522366259175377e-06,
      "loss": 0.2849,
      "num_input_tokens_seen": 15722072,
      "step": 24080
    },
    {
      "epoch": 14.201061320754716,
      "grad_norm": 2.5783817768096924,
      "learning_rate": 2.350054602165044e-06,
      "loss": 0.3787,
      "num_input_tokens_seen": 15726008,
      "step": 24085
    },
    {
      "epoch": 14.204009433962264,
      "grad_norm": 3.0265262126922607,
      "learning_rate": 2.3478732799970143e-06,
      "loss": 0.2859,
      "num_input_tokens_seen": 15728504,
      "step": 24090
    },
    {
      "epoch": 14.20695754716981,
      "grad_norm": 5.609025955200195,
      "learning_rate": 2.3456926599909646e-06,
      "loss": 0.381,
      "num_input_tokens_seen": 15731768,
      "step": 24095
    },
    {
      "epoch": 14.209905660377359,
      "grad_norm": 4.55352783203125,
      "learning_rate": 2.343512742724222e-06,
      "loss": 0.3961,
      "num_input_tokens_seen": 15735416,
      "step": 24100
    },
    {
      "epoch": 14.212853773584905,
      "grad_norm": 4.563655376434326,
      "learning_rate": 2.341333528773928e-06,
      "loss": 0.3129,
      "num_input_tokens_seen": 15739896,
      "step": 24105
    },
    {
      "epoch": 14.215801886792454,
      "grad_norm": 4.843811988830566,
      "learning_rate": 2.3391550187170427e-06,
      "loss": 0.3178,
      "num_input_tokens_seen": 15742712,
      "step": 24110
    },
    {
      "epoch": 14.21875,
      "grad_norm": 6.251970291137695,
      "learning_rate": 2.336977213130333e-06,
      "loss": 0.3548,
      "num_input_tokens_seen": 15745528,
      "step": 24115
    },
    {
      "epoch": 14.221698113207546,
      "grad_norm": 5.0519304275512695,
      "learning_rate": 2.3348001125903837e-06,
      "loss": 0.4504,
      "num_input_tokens_seen": 15748536,
      "step": 24120
    },
    {
      "epoch": 14.224646226415095,
      "grad_norm": 2.9172656536102295,
      "learning_rate": 2.3326237176735905e-06,
      "loss": 0.2631,
      "num_input_tokens_seen": 15751000,
      "step": 24125
    },
    {
      "epoch": 14.227594339622641,
      "grad_norm": 3.3330605030059814,
      "learning_rate": 2.330448028956164e-06,
      "loss": 0.2291,
      "num_input_tokens_seen": 15754904,
      "step": 24130
    },
    {
      "epoch": 14.23054245283019,
      "grad_norm": 3.406487464904785,
      "learning_rate": 2.3282730470141255e-06,
      "loss": 0.3708,
      "num_input_tokens_seen": 15757912,
      "step": 24135
    },
    {
      "epoch": 14.233490566037736,
      "grad_norm": 5.141595363616943,
      "learning_rate": 2.3260987724233143e-06,
      "loss": 0.4114,
      "num_input_tokens_seen": 15760792,
      "step": 24140
    },
    {
      "epoch": 14.236438679245284,
      "grad_norm": 2.2629570960998535,
      "learning_rate": 2.323925205759374e-06,
      "loss": 0.3103,
      "num_input_tokens_seen": 15763704,
      "step": 24145
    },
    {
      "epoch": 14.23938679245283,
      "grad_norm": 1.752752423286438,
      "learning_rate": 2.3217523475977715e-06,
      "loss": 0.3204,
      "num_input_tokens_seen": 15767128,
      "step": 24150
    },
    {
      "epoch": 14.242334905660377,
      "grad_norm": 4.218230247497559,
      "learning_rate": 2.3195801985137773e-06,
      "loss": 0.4223,
      "num_input_tokens_seen": 15770904,
      "step": 24155
    },
    {
      "epoch": 14.245283018867925,
      "grad_norm": 3.371896505355835,
      "learning_rate": 2.317408759082478e-06,
      "loss": 0.2677,
      "num_input_tokens_seen": 15773528,
      "step": 24160
    },
    {
      "epoch": 14.248231132075471,
      "grad_norm": 3.2545254230499268,
      "learning_rate": 2.31523802987877e-06,
      "loss": 0.2553,
      "num_input_tokens_seen": 15776312,
      "step": 24165
    },
    {
      "epoch": 14.25117924528302,
      "grad_norm": 3.053255319595337,
      "learning_rate": 2.3130680114773637e-06,
      "loss": 0.3026,
      "num_input_tokens_seen": 15778680,
      "step": 24170
    },
    {
      "epoch": 14.254127358490566,
      "grad_norm": 4.017597675323486,
      "learning_rate": 2.310898704452782e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 15782072,
      "step": 24175
    },
    {
      "epoch": 14.257075471698114,
      "grad_norm": 3.984003782272339,
      "learning_rate": 2.3087301093793584e-06,
      "loss": 0.2779,
      "num_input_tokens_seen": 15784856,
      "step": 24180
    },
    {
      "epoch": 14.26002358490566,
      "grad_norm": 3.8263967037200928,
      "learning_rate": 2.306562226831237e-06,
      "loss": 0.3186,
      "num_input_tokens_seen": 15788024,
      "step": 24185
    },
    {
      "epoch": 14.262971698113208,
      "grad_norm": 3.95196795463562,
      "learning_rate": 2.304395057382374e-06,
      "loss": 0.2294,
      "num_input_tokens_seen": 15790648,
      "step": 24190
    },
    {
      "epoch": 14.265919811320755,
      "grad_norm": 2.1403892040252686,
      "learning_rate": 2.3022286016065354e-06,
      "loss": 0.3453,
      "num_input_tokens_seen": 15793656,
      "step": 24195
    },
    {
      "epoch": 14.268867924528301,
      "grad_norm": 2.7065579891204834,
      "learning_rate": 2.300062860077303e-06,
      "loss": 0.2681,
      "num_input_tokens_seen": 15796248,
      "step": 24200
    },
    {
      "epoch": 14.27181603773585,
      "grad_norm": 3.412202835083008,
      "learning_rate": 2.297897833368064e-06,
      "loss": 0.3628,
      "num_input_tokens_seen": 15799192,
      "step": 24205
    },
    {
      "epoch": 14.274764150943396,
      "grad_norm": 3.5632128715515137,
      "learning_rate": 2.2957335220520194e-06,
      "loss": 0.3291,
      "num_input_tokens_seen": 15801496,
      "step": 24210
    },
    {
      "epoch": 14.277712264150944,
      "grad_norm": 5.4536848068237305,
      "learning_rate": 2.293569926702179e-06,
      "loss": 0.2567,
      "num_input_tokens_seen": 15806520,
      "step": 24215
    },
    {
      "epoch": 14.28066037735849,
      "grad_norm": 2.7862112522125244,
      "learning_rate": 2.291407047891366e-06,
      "loss": 0.3222,
      "num_input_tokens_seen": 15810584,
      "step": 24220
    },
    {
      "epoch": 14.283608490566039,
      "grad_norm": 2.908254384994507,
      "learning_rate": 2.2892448861922075e-06,
      "loss": 0.3436,
      "num_input_tokens_seen": 15814072,
      "step": 24225
    },
    {
      "epoch": 14.286556603773585,
      "grad_norm": 5.162661552429199,
      "learning_rate": 2.2870834421771505e-06,
      "loss": 0.3707,
      "num_input_tokens_seen": 15816984,
      "step": 24230
    },
    {
      "epoch": 14.289504716981131,
      "grad_norm": 2.474459171295166,
      "learning_rate": 2.2849227164184433e-06,
      "loss": 0.313,
      "num_input_tokens_seen": 15820280,
      "step": 24235
    },
    {
      "epoch": 14.29245283018868,
      "grad_norm": 3.512907028198242,
      "learning_rate": 2.2827627094881473e-06,
      "loss": 0.4984,
      "num_input_tokens_seen": 15826072,
      "step": 24240
    },
    {
      "epoch": 14.295400943396226,
      "grad_norm": 3.002999782562256,
      "learning_rate": 2.2806034219581364e-06,
      "loss": 0.2882,
      "num_input_tokens_seen": 15831416,
      "step": 24245
    },
    {
      "epoch": 14.298349056603774,
      "grad_norm": 3.2839317321777344,
      "learning_rate": 2.278444854400089e-06,
      "loss": 0.4109,
      "num_input_tokens_seen": 15834200,
      "step": 24250
    },
    {
      "epoch": 14.30129716981132,
      "grad_norm": 3.258768081665039,
      "learning_rate": 2.276287007385496e-06,
      "loss": 0.3489,
      "num_input_tokens_seen": 15837016,
      "step": 24255
    },
    {
      "epoch": 14.304245283018869,
      "grad_norm": 3.3484063148498535,
      "learning_rate": 2.2741298814856542e-06,
      "loss": 0.2863,
      "num_input_tokens_seen": 15840760,
      "step": 24260
    },
    {
      "epoch": 14.307193396226415,
      "grad_norm": 3.0318691730499268,
      "learning_rate": 2.2719734772716763e-06,
      "loss": 0.2631,
      "num_input_tokens_seen": 15844184,
      "step": 24265
    },
    {
      "epoch": 14.310141509433961,
      "grad_norm": 2.8463633060455322,
      "learning_rate": 2.269817795314477e-06,
      "loss": 0.3773,
      "num_input_tokens_seen": 15848760,
      "step": 24270
    },
    {
      "epoch": 14.31308962264151,
      "grad_norm": 3.2096729278564453,
      "learning_rate": 2.2676628361847834e-06,
      "loss": 0.2671,
      "num_input_tokens_seen": 15851960,
      "step": 24275
    },
    {
      "epoch": 14.316037735849056,
      "grad_norm": 4.803024768829346,
      "learning_rate": 2.2655086004531296e-06,
      "loss": 0.3662,
      "num_input_tokens_seen": 15855960,
      "step": 24280
    },
    {
      "epoch": 14.318985849056604,
      "grad_norm": 3.032406806945801,
      "learning_rate": 2.2633550886898583e-06,
      "loss": 0.3505,
      "num_input_tokens_seen": 15858744,
      "step": 24285
    },
    {
      "epoch": 14.32193396226415,
      "grad_norm": 2.8560009002685547,
      "learning_rate": 2.26120230146512e-06,
      "loss": 0.3486,
      "num_input_tokens_seen": 15861752,
      "step": 24290
    },
    {
      "epoch": 14.324882075471699,
      "grad_norm": 3.820908784866333,
      "learning_rate": 2.2590502393488777e-06,
      "loss": 0.3813,
      "num_input_tokens_seen": 15865784,
      "step": 24295
    },
    {
      "epoch": 14.327830188679245,
      "grad_norm": 4.50192403793335,
      "learning_rate": 2.256898902910898e-06,
      "loss": 0.3559,
      "num_input_tokens_seen": 15868280,
      "step": 24300
    },
    {
      "epoch": 14.330778301886792,
      "grad_norm": 4.109908580780029,
      "learning_rate": 2.2547482927207548e-06,
      "loss": 0.3393,
      "num_input_tokens_seen": 15871672,
      "step": 24305
    },
    {
      "epoch": 14.33372641509434,
      "grad_norm": 2.7612059116363525,
      "learning_rate": 2.252598409347833e-06,
      "loss": 0.3198,
      "num_input_tokens_seen": 15874840,
      "step": 24310
    },
    {
      "epoch": 14.336674528301886,
      "grad_norm": 2.621295213699341,
      "learning_rate": 2.250449253361323e-06,
      "loss": 0.3886,
      "num_input_tokens_seen": 15878264,
      "step": 24315
    },
    {
      "epoch": 14.339622641509434,
      "grad_norm": 3.8720176219940186,
      "learning_rate": 2.2483008253302214e-06,
      "loss": 0.3586,
      "num_input_tokens_seen": 15881528,
      "step": 24320
    },
    {
      "epoch": 14.34257075471698,
      "grad_norm": 4.059556484222412,
      "learning_rate": 2.246153125823337e-06,
      "loss": 0.3412,
      "num_input_tokens_seen": 15885304,
      "step": 24325
    },
    {
      "epoch": 14.345518867924529,
      "grad_norm": 3.831134796142578,
      "learning_rate": 2.2440061554092813e-06,
      "loss": 0.4714,
      "num_input_tokens_seen": 15888408,
      "step": 24330
    },
    {
      "epoch": 14.348466981132075,
      "grad_norm": 3.310366153717041,
      "learning_rate": 2.2418599146564714e-06,
      "loss": 0.2547,
      "num_input_tokens_seen": 15892280,
      "step": 24335
    },
    {
      "epoch": 14.351415094339623,
      "grad_norm": 5.969259262084961,
      "learning_rate": 2.239714404133138e-06,
      "loss": 0.2886,
      "num_input_tokens_seen": 15895064,
      "step": 24340
    },
    {
      "epoch": 14.35436320754717,
      "grad_norm": 4.251812934875488,
      "learning_rate": 2.2375696244073126e-06,
      "loss": 0.3512,
      "num_input_tokens_seen": 15898232,
      "step": 24345
    },
    {
      "epoch": 14.357311320754716,
      "grad_norm": 3.0990238189697266,
      "learning_rate": 2.235425576046834e-06,
      "loss": 0.3115,
      "num_input_tokens_seen": 15900472,
      "step": 24350
    },
    {
      "epoch": 14.360259433962264,
      "grad_norm": 3.015045642852783,
      "learning_rate": 2.233282259619347e-06,
      "loss": 0.33,
      "num_input_tokens_seen": 15902808,
      "step": 24355
    },
    {
      "epoch": 14.36320754716981,
      "grad_norm": 2.810035228729248,
      "learning_rate": 2.231139675692308e-06,
      "loss": 0.3507,
      "num_input_tokens_seen": 15906136,
      "step": 24360
    },
    {
      "epoch": 14.366155660377359,
      "grad_norm": 3.0610291957855225,
      "learning_rate": 2.228997824832973e-06,
      "loss": 0.372,
      "num_input_tokens_seen": 15912728,
      "step": 24365
    },
    {
      "epoch": 14.369103773584905,
      "grad_norm": 2.4471137523651123,
      "learning_rate": 2.226856707608406e-06,
      "loss": 0.2313,
      "num_input_tokens_seen": 15917016,
      "step": 24370
    },
    {
      "epoch": 14.372051886792454,
      "grad_norm": 4.8992486000061035,
      "learning_rate": 2.2247163245854768e-06,
      "loss": 0.3072,
      "num_input_tokens_seen": 15919128,
      "step": 24375
    },
    {
      "epoch": 14.375,
      "grad_norm": 3.533554792404175,
      "learning_rate": 2.222576676330862e-06,
      "loss": 0.3135,
      "num_input_tokens_seen": 15921752,
      "step": 24380
    },
    {
      "epoch": 14.377948113207546,
      "grad_norm": 3.7241976261138916,
      "learning_rate": 2.2204377634110403e-06,
      "loss": 0.3074,
      "num_input_tokens_seen": 15925784,
      "step": 24385
    },
    {
      "epoch": 14.380896226415095,
      "grad_norm": 3.2024948596954346,
      "learning_rate": 2.218299586392301e-06,
      "loss": 0.3555,
      "num_input_tokens_seen": 15928632,
      "step": 24390
    },
    {
      "epoch": 14.383844339622641,
      "grad_norm": 1.9672027826309204,
      "learning_rate": 2.2161621458407355e-06,
      "loss": 0.3114,
      "num_input_tokens_seen": 15932216,
      "step": 24395
    },
    {
      "epoch": 14.38679245283019,
      "grad_norm": 2.917719602584839,
      "learning_rate": 2.2140254423222398e-06,
      "loss": 0.3,
      "num_input_tokens_seen": 15935800,
      "step": 24400
    },
    {
      "epoch": 14.389740566037736,
      "grad_norm": 3.368609666824341,
      "learning_rate": 2.2118894764025146e-06,
      "loss": 0.2226,
      "num_input_tokens_seen": 15938520,
      "step": 24405
    },
    {
      "epoch": 14.392688679245284,
      "grad_norm": 2.9019968509674072,
      "learning_rate": 2.2097542486470667e-06,
      "loss": 0.3389,
      "num_input_tokens_seen": 15941752,
      "step": 24410
    },
    {
      "epoch": 14.39563679245283,
      "grad_norm": 3.6766483783721924,
      "learning_rate": 2.207619759621205e-06,
      "loss": 0.3444,
      "num_input_tokens_seen": 15945272,
      "step": 24415
    },
    {
      "epoch": 14.398584905660377,
      "grad_norm": 3.259132146835327,
      "learning_rate": 2.205486009890049e-06,
      "loss": 0.3377,
      "num_input_tokens_seen": 15948952,
      "step": 24420
    },
    {
      "epoch": 14.401533018867925,
      "grad_norm": 4.039430618286133,
      "learning_rate": 2.2033530000185146e-06,
      "loss": 0.2825,
      "num_input_tokens_seen": 15954680,
      "step": 24425
    },
    {
      "epoch": 14.404481132075471,
      "grad_norm": 3.5588290691375732,
      "learning_rate": 2.2012207305713244e-06,
      "loss": 0.3256,
      "num_input_tokens_seen": 15956984,
      "step": 24430
    },
    {
      "epoch": 14.40742924528302,
      "grad_norm": 4.120483875274658,
      "learning_rate": 2.19908920211301e-06,
      "loss": 0.3601,
      "num_input_tokens_seen": 15959960,
      "step": 24435
    },
    {
      "epoch": 14.410377358490566,
      "grad_norm": 2.310478448867798,
      "learning_rate": 2.196958415207901e-06,
      "loss": 0.2407,
      "num_input_tokens_seen": 15963416,
      "step": 24440
    },
    {
      "epoch": 14.413325471698114,
      "grad_norm": 4.675232410430908,
      "learning_rate": 2.1948283704201312e-06,
      "loss": 0.3438,
      "num_input_tokens_seen": 15966488,
      "step": 24445
    },
    {
      "epoch": 14.41627358490566,
      "grad_norm": 2.6562023162841797,
      "learning_rate": 2.1926990683136383e-06,
      "loss": 0.3504,
      "num_input_tokens_seen": 15969784,
      "step": 24450
    },
    {
      "epoch": 14.419221698113208,
      "grad_norm": 3.6201164722442627,
      "learning_rate": 2.1905705094521685e-06,
      "loss": 0.3158,
      "num_input_tokens_seen": 15972696,
      "step": 24455
    },
    {
      "epoch": 14.422169811320755,
      "grad_norm": 2.400818109512329,
      "learning_rate": 2.1884426943992635e-06,
      "loss": 0.3193,
      "num_input_tokens_seen": 15975832,
      "step": 24460
    },
    {
      "epoch": 14.425117924528301,
      "grad_norm": 3.1438746452331543,
      "learning_rate": 2.1863156237182727e-06,
      "loss": 0.3011,
      "num_input_tokens_seen": 15978936,
      "step": 24465
    },
    {
      "epoch": 14.42806603773585,
      "grad_norm": 3.6834146976470947,
      "learning_rate": 2.1841892979723466e-06,
      "loss": 0.2822,
      "num_input_tokens_seen": 15981752,
      "step": 24470
    },
    {
      "epoch": 14.431014150943396,
      "grad_norm": 5.897389888763428,
      "learning_rate": 2.1820637177244375e-06,
      "loss": 0.3454,
      "num_input_tokens_seen": 15984536,
      "step": 24475
    },
    {
      "epoch": 14.433962264150944,
      "grad_norm": 2.9747421741485596,
      "learning_rate": 2.179938883537306e-06,
      "loss": 0.2998,
      "num_input_tokens_seen": 15987288,
      "step": 24480
    },
    {
      "epoch": 14.43691037735849,
      "grad_norm": 4.500246047973633,
      "learning_rate": 2.177814795973508e-06,
      "loss": 0.3096,
      "num_input_tokens_seen": 15990840,
      "step": 24485
    },
    {
      "epoch": 14.439858490566039,
      "grad_norm": 4.845057487487793,
      "learning_rate": 2.1756914555954064e-06,
      "loss": 0.2668,
      "num_input_tokens_seen": 15994424,
      "step": 24490
    },
    {
      "epoch": 14.442806603773585,
      "grad_norm": 1.8264859914779663,
      "learning_rate": 2.173568862965164e-06,
      "loss": 0.2608,
      "num_input_tokens_seen": 15997880,
      "step": 24495
    },
    {
      "epoch": 14.445754716981131,
      "grad_norm": 7.522373199462891,
      "learning_rate": 2.171447018644746e-06,
      "loss": 0.5477,
      "num_input_tokens_seen": 16000472,
      "step": 24500
    },
    {
      "epoch": 14.44870283018868,
      "grad_norm": 4.088736057281494,
      "learning_rate": 2.1693259231959186e-06,
      "loss": 0.3466,
      "num_input_tokens_seen": 16003512,
      "step": 24505
    },
    {
      "epoch": 14.451650943396226,
      "grad_norm": 6.272796630859375,
      "learning_rate": 2.1672055771802545e-06,
      "loss": 0.3931,
      "num_input_tokens_seen": 16006520,
      "step": 24510
    },
    {
      "epoch": 14.454599056603774,
      "grad_norm": 3.148322343826294,
      "learning_rate": 2.1650859811591224e-06,
      "loss": 0.265,
      "num_input_tokens_seen": 16009592,
      "step": 24515
    },
    {
      "epoch": 14.45754716981132,
      "grad_norm": 3.959523916244507,
      "learning_rate": 2.1629671356936943e-06,
      "loss": 0.3259,
      "num_input_tokens_seen": 16013656,
      "step": 24520
    },
    {
      "epoch": 14.460495283018869,
      "grad_norm": 3.443258285522461,
      "learning_rate": 2.1608490413449428e-06,
      "loss": 0.2691,
      "num_input_tokens_seen": 16016856,
      "step": 24525
    },
    {
      "epoch": 14.463443396226415,
      "grad_norm": 4.163712501525879,
      "learning_rate": 2.158731698673645e-06,
      "loss": 0.3788,
      "num_input_tokens_seen": 16019256,
      "step": 24530
    },
    {
      "epoch": 14.466391509433961,
      "grad_norm": 3.3735201358795166,
      "learning_rate": 2.1566151082403752e-06,
      "loss": 0.2687,
      "num_input_tokens_seen": 16022104,
      "step": 24535
    },
    {
      "epoch": 14.46933962264151,
      "grad_norm": 4.417233943939209,
      "learning_rate": 2.154499270605508e-06,
      "loss": 0.2921,
      "num_input_tokens_seen": 16025112,
      "step": 24540
    },
    {
      "epoch": 14.472287735849056,
      "grad_norm": 6.360416412353516,
      "learning_rate": 2.1523841863292243e-06,
      "loss": 0.3586,
      "num_input_tokens_seen": 16028088,
      "step": 24545
    },
    {
      "epoch": 14.475235849056604,
      "grad_norm": 2.3085317611694336,
      "learning_rate": 2.1502698559714998e-06,
      "loss": 0.3377,
      "num_input_tokens_seen": 16031640,
      "step": 24550
    },
    {
      "epoch": 14.47818396226415,
      "grad_norm": 1.9191153049468994,
      "learning_rate": 2.1481562800921125e-06,
      "loss": 0.3389,
      "num_input_tokens_seen": 16035064,
      "step": 24555
    },
    {
      "epoch": 14.481132075471699,
      "grad_norm": 5.830350875854492,
      "learning_rate": 2.146043459250641e-06,
      "loss": 0.3088,
      "num_input_tokens_seen": 16037848,
      "step": 24560
    },
    {
      "epoch": 14.484080188679245,
      "grad_norm": 4.165152549743652,
      "learning_rate": 2.1439313940064634e-06,
      "loss": 0.3072,
      "num_input_tokens_seen": 16041432,
      "step": 24565
    },
    {
      "epoch": 14.487028301886792,
      "grad_norm": 7.840921401977539,
      "learning_rate": 2.141820084918756e-06,
      "loss": 0.3589,
      "num_input_tokens_seen": 16045656,
      "step": 24570
    },
    {
      "epoch": 14.48997641509434,
      "grad_norm": 3.7904062271118164,
      "learning_rate": 2.1397095325465013e-06,
      "loss": 0.3401,
      "num_input_tokens_seen": 16048760,
      "step": 24575
    },
    {
      "epoch": 14.492924528301886,
      "grad_norm": 3.34272837638855,
      "learning_rate": 2.1375997374484754e-06,
      "loss": 0.3185,
      "num_input_tokens_seen": 16052792,
      "step": 24580
    },
    {
      "epoch": 14.495872641509434,
      "grad_norm": 4.443939208984375,
      "learning_rate": 2.1354907001832546e-06,
      "loss": 0.3174,
      "num_input_tokens_seen": 16055544,
      "step": 24585
    },
    {
      "epoch": 14.49882075471698,
      "grad_norm": 3.0289082527160645,
      "learning_rate": 2.133382421309217e-06,
      "loss": 0.2882,
      "num_input_tokens_seen": 16058168,
      "step": 24590
    },
    {
      "epoch": 14.501768867924529,
      "grad_norm": 3.923945188522339,
      "learning_rate": 2.131274901384537e-06,
      "loss": 0.3229,
      "num_input_tokens_seen": 16062072,
      "step": 24595
    },
    {
      "epoch": 14.504716981132075,
      "grad_norm": 3.5849239826202393,
      "learning_rate": 2.1291681409671896e-06,
      "loss": 0.3107,
      "num_input_tokens_seen": 16064824,
      "step": 24600
    },
    {
      "epoch": 14.507665094339622,
      "grad_norm": 17.40311050415039,
      "learning_rate": 2.12706214061495e-06,
      "loss": 0.302,
      "num_input_tokens_seen": 16067320,
      "step": 24605
    },
    {
      "epoch": 14.51061320754717,
      "grad_norm": 4.50157356262207,
      "learning_rate": 2.124956900885391e-06,
      "loss": 0.4105,
      "num_input_tokens_seen": 16070008,
      "step": 24610
    },
    {
      "epoch": 14.513561320754716,
      "grad_norm": 2.9883038997650146,
      "learning_rate": 2.1228524223358833e-06,
      "loss": 0.2902,
      "num_input_tokens_seen": 16072504,
      "step": 24615
    },
    {
      "epoch": 14.516509433962264,
      "grad_norm": 3.6989357471466064,
      "learning_rate": 2.120748705523595e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 16075896,
      "step": 24620
    },
    {
      "epoch": 14.51945754716981,
      "grad_norm": 4.705385684967041,
      "learning_rate": 2.1186457510054976e-06,
      "loss": 0.2418,
      "num_input_tokens_seen": 16078296,
      "step": 24625
    },
    {
      "epoch": 14.522405660377359,
      "grad_norm": 3.6937990188598633,
      "learning_rate": 2.116543559338355e-06,
      "loss": 0.3203,
      "num_input_tokens_seen": 16080792,
      "step": 24630
    },
    {
      "epoch": 14.525353773584905,
      "grad_norm": 3.274273633956909,
      "learning_rate": 2.1144421310787305e-06,
      "loss": 0.3166,
      "num_input_tokens_seen": 16084152,
      "step": 24635
    },
    {
      "epoch": 14.528301886792454,
      "grad_norm": 6.232609748840332,
      "learning_rate": 2.11234146678299e-06,
      "loss": 0.3757,
      "num_input_tokens_seen": 16087192,
      "step": 24640
    },
    {
      "epoch": 14.53125,
      "grad_norm": 4.3313889503479,
      "learning_rate": 2.1102415670072907e-06,
      "loss": 0.4327,
      "num_input_tokens_seen": 16090392,
      "step": 24645
    },
    {
      "epoch": 14.534198113207546,
      "grad_norm": 3.7667744159698486,
      "learning_rate": 2.108142432307591e-06,
      "loss": 0.4116,
      "num_input_tokens_seen": 16094200,
      "step": 24650
    },
    {
      "epoch": 14.537146226415095,
      "grad_norm": 2.474250078201294,
      "learning_rate": 2.1060440632396456e-06,
      "loss": 0.259,
      "num_input_tokens_seen": 16098008,
      "step": 24655
    },
    {
      "epoch": 14.540094339622641,
      "grad_norm": 3.3793652057647705,
      "learning_rate": 2.103946460359007e-06,
      "loss": 0.2354,
      "num_input_tokens_seen": 16100664,
      "step": 24660
    },
    {
      "epoch": 14.54304245283019,
      "grad_norm": 5.267772674560547,
      "learning_rate": 2.101849624221022e-06,
      "loss": 0.3667,
      "num_input_tokens_seen": 16103384,
      "step": 24665
    },
    {
      "epoch": 14.545990566037736,
      "grad_norm": 2.518512010574341,
      "learning_rate": 2.0997535553808417e-06,
      "loss": 0.2406,
      "num_input_tokens_seen": 16106488,
      "step": 24670
    },
    {
      "epoch": 14.548938679245284,
      "grad_norm": 2.0440850257873535,
      "learning_rate": 2.0976582543934064e-06,
      "loss": 0.2744,
      "num_input_tokens_seen": 16110360,
      "step": 24675
    },
    {
      "epoch": 14.55188679245283,
      "grad_norm": 3.3194093704223633,
      "learning_rate": 2.0955637218134573e-06,
      "loss": 0.2792,
      "num_input_tokens_seen": 16113880,
      "step": 24680
    },
    {
      "epoch": 14.554834905660378,
      "grad_norm": 2.589024543762207,
      "learning_rate": 2.09346995819553e-06,
      "loss": 0.2526,
      "num_input_tokens_seen": 16117080,
      "step": 24685
    },
    {
      "epoch": 14.557783018867925,
      "grad_norm": 5.041914939880371,
      "learning_rate": 2.0913769640939553e-06,
      "loss": 0.4293,
      "num_input_tokens_seen": 16120440,
      "step": 24690
    },
    {
      "epoch": 14.560731132075471,
      "grad_norm": 4.148814678192139,
      "learning_rate": 2.0892847400628674e-06,
      "loss": 0.4181,
      "num_input_tokens_seen": 16123864,
      "step": 24695
    },
    {
      "epoch": 14.56367924528302,
      "grad_norm": 6.190976619720459,
      "learning_rate": 2.0871932866561885e-06,
      "loss": 0.311,
      "num_input_tokens_seen": 16127032,
      "step": 24700
    },
    {
      "epoch": 14.566627358490566,
      "grad_norm": 2.834306478500366,
      "learning_rate": 2.0851026044276405e-06,
      "loss": 0.301,
      "num_input_tokens_seen": 16129528,
      "step": 24705
    },
    {
      "epoch": 14.569575471698114,
      "grad_norm": 6.893313884735107,
      "learning_rate": 2.083012693930741e-06,
      "loss": 0.3594,
      "num_input_tokens_seen": 16131864,
      "step": 24710
    },
    {
      "epoch": 14.57252358490566,
      "grad_norm": 2.034270763397217,
      "learning_rate": 2.0809235557188e-06,
      "loss": 0.2058,
      "num_input_tokens_seen": 16135992,
      "step": 24715
    },
    {
      "epoch": 14.575471698113208,
      "grad_norm": 2.03066086769104,
      "learning_rate": 2.0788351903449307e-06,
      "loss": 0.3389,
      "num_input_tokens_seen": 16139032,
      "step": 24720
    },
    {
      "epoch": 14.578419811320755,
      "grad_norm": 2.4219069480895996,
      "learning_rate": 2.0767475983620317e-06,
      "loss": 0.2686,
      "num_input_tokens_seen": 16142712,
      "step": 24725
    },
    {
      "epoch": 14.581367924528301,
      "grad_norm": 3.401991605758667,
      "learning_rate": 2.074660780322806e-06,
      "loss": 0.3082,
      "num_input_tokens_seen": 16146616,
      "step": 24730
    },
    {
      "epoch": 14.58431603773585,
      "grad_norm": 2.8044910430908203,
      "learning_rate": 2.0725747367797473e-06,
      "loss": 0.3198,
      "num_input_tokens_seen": 16149560,
      "step": 24735
    },
    {
      "epoch": 14.587264150943396,
      "grad_norm": 2.5029103755950928,
      "learning_rate": 2.070489468285143e-06,
      "loss": 0.3162,
      "num_input_tokens_seen": 16153656,
      "step": 24740
    },
    {
      "epoch": 14.590212264150944,
      "grad_norm": 3.0213730335235596,
      "learning_rate": 2.068404975391077e-06,
      "loss": 0.3303,
      "num_input_tokens_seen": 16157464,
      "step": 24745
    },
    {
      "epoch": 14.59316037735849,
      "grad_norm": 3.246367931365967,
      "learning_rate": 2.0663212586494293e-06,
      "loss": 0.372,
      "num_input_tokens_seen": 16160824,
      "step": 24750
    },
    {
      "epoch": 14.596108490566039,
      "grad_norm": 2.5751190185546875,
      "learning_rate": 2.064238318611869e-06,
      "loss": 0.2608,
      "num_input_tokens_seen": 16167384,
      "step": 24755
    },
    {
      "epoch": 14.599056603773585,
      "grad_norm": 1.9619578123092651,
      "learning_rate": 2.0621561558298693e-06,
      "loss": 0.2641,
      "num_input_tokens_seen": 16171224,
      "step": 24760
    },
    {
      "epoch": 14.602004716981131,
      "grad_norm": 3.241269588470459,
      "learning_rate": 2.0600747708546877e-06,
      "loss": 0.293,
      "num_input_tokens_seen": 16174200,
      "step": 24765
    },
    {
      "epoch": 14.60495283018868,
      "grad_norm": 2.0297579765319824,
      "learning_rate": 2.0579941642373814e-06,
      "loss": 0.2646,
      "num_input_tokens_seen": 16177816,
      "step": 24770
    },
    {
      "epoch": 14.607900943396226,
      "grad_norm": 3.2797138690948486,
      "learning_rate": 2.0559143365287993e-06,
      "loss": 0.4378,
      "num_input_tokens_seen": 16180952,
      "step": 24775
    },
    {
      "epoch": 14.610849056603774,
      "grad_norm": 3.4334628582000732,
      "learning_rate": 2.0538352882795846e-06,
      "loss": 0.3841,
      "num_input_tokens_seen": 16184120,
      "step": 24780
    },
    {
      "epoch": 14.61379716981132,
      "grad_norm": 5.2459211349487305,
      "learning_rate": 2.051757020040173e-06,
      "loss": 0.3101,
      "num_input_tokens_seen": 16187000,
      "step": 24785
    },
    {
      "epoch": 14.616745283018869,
      "grad_norm": 2.9022159576416016,
      "learning_rate": 2.0496795323607983e-06,
      "loss": 0.3104,
      "num_input_tokens_seen": 16189720,
      "step": 24790
    },
    {
      "epoch": 14.619693396226415,
      "grad_norm": 4.8147783279418945,
      "learning_rate": 2.0476028257914825e-06,
      "loss": 0.2769,
      "num_input_tokens_seen": 16193112,
      "step": 24795
    },
    {
      "epoch": 14.622641509433961,
      "grad_norm": 2.4804394245147705,
      "learning_rate": 2.0455269008820433e-06,
      "loss": 0.516,
      "num_input_tokens_seen": 16197304,
      "step": 24800
    },
    {
      "epoch": 14.62558962264151,
      "grad_norm": 3.421219825744629,
      "learning_rate": 2.0434517581820893e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 16200728,
      "step": 24805
    },
    {
      "epoch": 14.628537735849056,
      "grad_norm": 5.197613716125488,
      "learning_rate": 2.041377398241025e-06,
      "loss": 0.3165,
      "num_input_tokens_seen": 16203448,
      "step": 24810
    },
    {
      "epoch": 14.631485849056604,
      "grad_norm": 4.57867956161499,
      "learning_rate": 2.0393038216080433e-06,
      "loss": 0.3293,
      "num_input_tokens_seen": 16206584,
      "step": 24815
    },
    {
      "epoch": 14.63443396226415,
      "grad_norm": 16.227115631103516,
      "learning_rate": 2.037231028832135e-06,
      "loss": 0.5062,
      "num_input_tokens_seen": 16208984,
      "step": 24820
    },
    {
      "epoch": 14.637382075471699,
      "grad_norm": 3.1904683113098145,
      "learning_rate": 2.0351590204620823e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 16211832,
      "step": 24825
    },
    {
      "epoch": 14.640330188679245,
      "grad_norm": 4.230688095092773,
      "learning_rate": 2.033087797046457e-06,
      "loss": 0.3588,
      "num_input_tokens_seen": 16215160,
      "step": 24830
    },
    {
      "epoch": 14.643278301886792,
      "grad_norm": 2.4691450595855713,
      "learning_rate": 2.031017359133624e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 16218904,
      "step": 24835
    },
    {
      "epoch": 14.64622641509434,
      "grad_norm": 4.355981826782227,
      "learning_rate": 2.0289477072717406e-06,
      "loss": 0.2499,
      "num_input_tokens_seen": 16222552,
      "step": 24840
    },
    {
      "epoch": 14.649174528301886,
      "grad_norm": 4.029301643371582,
      "learning_rate": 2.026878842008756e-06,
      "loss": 0.2778,
      "num_input_tokens_seen": 16225496,
      "step": 24845
    },
    {
      "epoch": 14.652122641509434,
      "grad_norm": 3.060234308242798,
      "learning_rate": 2.0248107638924105e-06,
      "loss": 0.3617,
      "num_input_tokens_seen": 16228728,
      "step": 24850
    },
    {
      "epoch": 14.65507075471698,
      "grad_norm": 3.358344793319702,
      "learning_rate": 2.0227434734702386e-06,
      "loss": 0.3171,
      "num_input_tokens_seen": 16231960,
      "step": 24855
    },
    {
      "epoch": 14.658018867924529,
      "grad_norm": 2.5629827976226807,
      "learning_rate": 2.020676971289563e-06,
      "loss": 0.3514,
      "num_input_tokens_seen": 16236088,
      "step": 24860
    },
    {
      "epoch": 14.660966981132075,
      "grad_norm": 2.758998155593872,
      "learning_rate": 2.0186112578975005e-06,
      "loss": 0.3923,
      "num_input_tokens_seen": 16241336,
      "step": 24865
    },
    {
      "epoch": 14.663915094339622,
      "grad_norm": 3.4192328453063965,
      "learning_rate": 2.016546333840956e-06,
      "loss": 0.4374,
      "num_input_tokens_seen": 16244088,
      "step": 24870
    },
    {
      "epoch": 14.66686320754717,
      "grad_norm": 4.319182395935059,
      "learning_rate": 2.014482199666627e-06,
      "loss": 0.3159,
      "num_input_tokens_seen": 16247768,
      "step": 24875
    },
    {
      "epoch": 14.669811320754716,
      "grad_norm": 5.0578155517578125,
      "learning_rate": 2.0124188559210017e-06,
      "loss": 0.4157,
      "num_input_tokens_seen": 16250552,
      "step": 24880
    },
    {
      "epoch": 14.672759433962264,
      "grad_norm": 3.473557472229004,
      "learning_rate": 2.0103563031503613e-06,
      "loss": 0.3023,
      "num_input_tokens_seen": 16253880,
      "step": 24885
    },
    {
      "epoch": 14.67570754716981,
      "grad_norm": 3.932966470718384,
      "learning_rate": 2.0082945419007745e-06,
      "loss": 0.2616,
      "num_input_tokens_seen": 16256952,
      "step": 24890
    },
    {
      "epoch": 14.678655660377359,
      "grad_norm": 3.7742600440979004,
      "learning_rate": 2.0062335727181007e-06,
      "loss": 0.262,
      "num_input_tokens_seen": 16261112,
      "step": 24895
    },
    {
      "epoch": 14.681603773584905,
      "grad_norm": 4.404270648956299,
      "learning_rate": 2.004173396147992e-06,
      "loss": 0.2901,
      "num_input_tokens_seen": 16263832,
      "step": 24900
    },
    {
      "epoch": 14.684551886792454,
      "grad_norm": 3.263373613357544,
      "learning_rate": 2.0021140127358873e-06,
      "loss": 0.2831,
      "num_input_tokens_seen": 16267320,
      "step": 24905
    },
    {
      "epoch": 14.6875,
      "grad_norm": 7.989091396331787,
      "learning_rate": 2.0000554230270164e-06,
      "loss": 0.3113,
      "num_input_tokens_seen": 16270296,
      "step": 24910
    },
    {
      "epoch": 14.690448113207546,
      "grad_norm": 7.270644664764404,
      "learning_rate": 1.997997627566401e-06,
      "loss": 0.2193,
      "num_input_tokens_seen": 16273144,
      "step": 24915
    },
    {
      "epoch": 14.693396226415095,
      "grad_norm": 3.6785807609558105,
      "learning_rate": 1.9959406268988536e-06,
      "loss": 0.2916,
      "num_input_tokens_seen": 16276344,
      "step": 24920
    },
    {
      "epoch": 14.696344339622641,
      "grad_norm": 3.2129178047180176,
      "learning_rate": 1.9938844215689717e-06,
      "loss": 0.2714,
      "num_input_tokens_seen": 16279544,
      "step": 24925
    },
    {
      "epoch": 14.69929245283019,
      "grad_norm": 2.594600200653076,
      "learning_rate": 1.991829012121145e-06,
      "loss": 0.2246,
      "num_input_tokens_seen": 16283128,
      "step": 24930
    },
    {
      "epoch": 14.702240566037736,
      "grad_norm": 3.7569234371185303,
      "learning_rate": 1.989774399099552e-06,
      "loss": 0.3031,
      "num_input_tokens_seen": 16285880,
      "step": 24935
    },
    {
      "epoch": 14.705188679245284,
      "grad_norm": 7.433042049407959,
      "learning_rate": 1.98772058304816e-06,
      "loss": 0.3746,
      "num_input_tokens_seen": 16288472,
      "step": 24940
    },
    {
      "epoch": 14.70813679245283,
      "grad_norm": 5.312839508056641,
      "learning_rate": 1.9856675645107244e-06,
      "loss": 0.4514,
      "num_input_tokens_seen": 16291256,
      "step": 24945
    },
    {
      "epoch": 14.711084905660378,
      "grad_norm": 4.346865177154541,
      "learning_rate": 1.9836153440307936e-06,
      "loss": 0.2961,
      "num_input_tokens_seen": 16294872,
      "step": 24950
    },
    {
      "epoch": 14.714033018867925,
      "grad_norm": 5.7718281745910645,
      "learning_rate": 1.9815639221517002e-06,
      "loss": 0.269,
      "num_input_tokens_seen": 16297688,
      "step": 24955
    },
    {
      "epoch": 14.716981132075471,
      "grad_norm": 2.6061019897460938,
      "learning_rate": 1.9795132994165673e-06,
      "loss": 0.3559,
      "num_input_tokens_seen": 16301208,
      "step": 24960
    },
    {
      "epoch": 14.71992924528302,
      "grad_norm": 4.009645938873291,
      "learning_rate": 1.977463476368306e-06,
      "loss": 0.2998,
      "num_input_tokens_seen": 16303736,
      "step": 24965
    },
    {
      "epoch": 14.722877358490566,
      "grad_norm": 3.154634952545166,
      "learning_rate": 1.975414453549614e-06,
      "loss": 0.2754,
      "num_input_tokens_seen": 16306616,
      "step": 24970
    },
    {
      "epoch": 14.725825471698114,
      "grad_norm": 3.1322290897369385,
      "learning_rate": 1.9733662315029826e-06,
      "loss": 0.3963,
      "num_input_tokens_seen": 16310040,
      "step": 24975
    },
    {
      "epoch": 14.72877358490566,
      "grad_norm": 3.957031726837158,
      "learning_rate": 1.9713188107706856e-06,
      "loss": 0.3532,
      "num_input_tokens_seen": 16312856,
      "step": 24980
    },
    {
      "epoch": 14.731721698113208,
      "grad_norm": 2.846776008605957,
      "learning_rate": 1.969272191894786e-06,
      "loss": 0.2758,
      "num_input_tokens_seen": 16315640,
      "step": 24985
    },
    {
      "epoch": 14.734669811320755,
      "grad_norm": 3.068634271621704,
      "learning_rate": 1.967226375417135e-06,
      "loss": 0.297,
      "num_input_tokens_seen": 16318680,
      "step": 24990
    },
    {
      "epoch": 14.737617924528301,
      "grad_norm": 3.493964672088623,
      "learning_rate": 1.965181361879372e-06,
      "loss": 0.3126,
      "num_input_tokens_seen": 16321848,
      "step": 24995
    },
    {
      "epoch": 14.74056603773585,
      "grad_norm": 4.2359619140625,
      "learning_rate": 1.9631371518229214e-06,
      "loss": 0.2886,
      "num_input_tokens_seen": 16324600,
      "step": 25000
    },
    {
      "epoch": 14.743514150943396,
      "grad_norm": 2.6380057334899902,
      "learning_rate": 1.9610937457889975e-06,
      "loss": 0.2822,
      "num_input_tokens_seen": 16327896,
      "step": 25005
    },
    {
      "epoch": 14.746462264150944,
      "grad_norm": 4.02836275100708,
      "learning_rate": 1.9590511443186032e-06,
      "loss": 0.3042,
      "num_input_tokens_seen": 16331096,
      "step": 25010
    },
    {
      "epoch": 14.74941037735849,
      "grad_norm": 3.8817296028137207,
      "learning_rate": 1.9570093479525243e-06,
      "loss": 0.3561,
      "num_input_tokens_seen": 16333976,
      "step": 25015
    },
    {
      "epoch": 14.752358490566039,
      "grad_norm": 4.128470420837402,
      "learning_rate": 1.954968357231335e-06,
      "loss": 0.4548,
      "num_input_tokens_seen": 16337176,
      "step": 25020
    },
    {
      "epoch": 14.755306603773585,
      "grad_norm": 4.137679576873779,
      "learning_rate": 1.9529281726953964e-06,
      "loss": 0.3339,
      "num_input_tokens_seen": 16340312,
      "step": 25025
    },
    {
      "epoch": 14.758254716981131,
      "grad_norm": 7.715548515319824,
      "learning_rate": 1.9508887948848564e-06,
      "loss": 0.3169,
      "num_input_tokens_seen": 16344728,
      "step": 25030
    },
    {
      "epoch": 14.76120283018868,
      "grad_norm": 3.4236059188842773,
      "learning_rate": 1.9488502243396475e-06,
      "loss": 0.2614,
      "num_input_tokens_seen": 16347832,
      "step": 25035
    },
    {
      "epoch": 14.764150943396226,
      "grad_norm": 7.7521257400512695,
      "learning_rate": 1.946812461599492e-06,
      "loss": 0.2985,
      "num_input_tokens_seen": 16350584,
      "step": 25040
    },
    {
      "epoch": 14.767099056603774,
      "grad_norm": 3.282959222793579,
      "learning_rate": 1.944775507203897e-06,
      "loss": 0.2169,
      "num_input_tokens_seen": 16354104,
      "step": 25045
    },
    {
      "epoch": 14.77004716981132,
      "grad_norm": 2.107105016708374,
      "learning_rate": 1.942739361692153e-06,
      "loss": 0.2626,
      "num_input_tokens_seen": 16357048,
      "step": 25050
    },
    {
      "epoch": 14.772995283018869,
      "grad_norm": 4.1882524490356445,
      "learning_rate": 1.94070402560334e-06,
      "loss": 0.4593,
      "num_input_tokens_seen": 16361048,
      "step": 25055
    },
    {
      "epoch": 14.775943396226415,
      "grad_norm": 3.467486619949341,
      "learning_rate": 1.93866949947632e-06,
      "loss": 0.2883,
      "num_input_tokens_seen": 16363864,
      "step": 25060
    },
    {
      "epoch": 14.778891509433961,
      "grad_norm": 2.0222299098968506,
      "learning_rate": 1.9366357838497423e-06,
      "loss": 0.3646,
      "num_input_tokens_seen": 16367608,
      "step": 25065
    },
    {
      "epoch": 14.78183962264151,
      "grad_norm": 2.681932210922241,
      "learning_rate": 1.9346028792620454e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 16371928,
      "step": 25070
    },
    {
      "epoch": 14.784787735849056,
      "grad_norm": 3.1201298236846924,
      "learning_rate": 1.9325707862514464e-06,
      "loss": 0.3401,
      "num_input_tokens_seen": 16375544,
      "step": 25075
    },
    {
      "epoch": 14.787735849056604,
      "grad_norm": 2.4026060104370117,
      "learning_rate": 1.930539505355952e-06,
      "loss": 0.3298,
      "num_input_tokens_seen": 16378648,
      "step": 25080
    },
    {
      "epoch": 14.79068396226415,
      "grad_norm": 2.7981984615325928,
      "learning_rate": 1.9285090371133524e-06,
      "loss": 0.4165,
      "num_input_tokens_seen": 16381880,
      "step": 25085
    },
    {
      "epoch": 14.793632075471699,
      "grad_norm": 3.097672462463379,
      "learning_rate": 1.9264793820612228e-06,
      "loss": 0.4654,
      "num_input_tokens_seen": 16386296,
      "step": 25090
    },
    {
      "epoch": 14.796580188679245,
      "grad_norm": 2.9298319816589355,
      "learning_rate": 1.924450540736921e-06,
      "loss": 0.2619,
      "num_input_tokens_seen": 16389272,
      "step": 25095
    },
    {
      "epoch": 14.799528301886792,
      "grad_norm": 3.1842098236083984,
      "learning_rate": 1.922422513677593e-06,
      "loss": 0.3192,
      "num_input_tokens_seen": 16392664,
      "step": 25100
    },
    {
      "epoch": 14.80247641509434,
      "grad_norm": 2.408630132675171,
      "learning_rate": 1.9203953014201703e-06,
      "loss": 0.4344,
      "num_input_tokens_seen": 16396536,
      "step": 25105
    },
    {
      "epoch": 14.805424528301886,
      "grad_norm": 2.8250129222869873,
      "learning_rate": 1.918368904501364e-06,
      "loss": 0.2464,
      "num_input_tokens_seen": 16400952,
      "step": 25110
    },
    {
      "epoch": 14.808372641509434,
      "grad_norm": 4.00761604309082,
      "learning_rate": 1.9163433234576713e-06,
      "loss": 0.4455,
      "num_input_tokens_seen": 16403864,
      "step": 25115
    },
    {
      "epoch": 14.81132075471698,
      "grad_norm": 4.205718994140625,
      "learning_rate": 1.9143185588253733e-06,
      "loss": 0.3224,
      "num_input_tokens_seen": 16406648,
      "step": 25120
    },
    {
      "epoch": 14.814268867924529,
      "grad_norm": 2.6720190048217773,
      "learning_rate": 1.9122946111405354e-06,
      "loss": 0.3417,
      "num_input_tokens_seen": 16409368,
      "step": 25125
    },
    {
      "epoch": 14.817216981132075,
      "grad_norm": 4.602489948272705,
      "learning_rate": 1.910271480939005e-06,
      "loss": 0.2512,
      "num_input_tokens_seen": 16411832,
      "step": 25130
    },
    {
      "epoch": 14.820165094339622,
      "grad_norm": 4.8560791015625,
      "learning_rate": 1.9082491687564176e-06,
      "loss": 0.2818,
      "num_input_tokens_seen": 16416504,
      "step": 25135
    },
    {
      "epoch": 14.82311320754717,
      "grad_norm": 5.241753101348877,
      "learning_rate": 1.9062276751281872e-06,
      "loss": 0.3733,
      "num_input_tokens_seen": 16419000,
      "step": 25140
    },
    {
      "epoch": 14.826061320754716,
      "grad_norm": 1.7326551675796509,
      "learning_rate": 1.9042070005895136e-06,
      "loss": 0.4264,
      "num_input_tokens_seen": 16423064,
      "step": 25145
    },
    {
      "epoch": 14.829009433962264,
      "grad_norm": 4.018425464630127,
      "learning_rate": 1.9021871456753788e-06,
      "loss": 0.2879,
      "num_input_tokens_seen": 16426200,
      "step": 25150
    },
    {
      "epoch": 14.83195754716981,
      "grad_norm": 3.9319965839385986,
      "learning_rate": 1.9001681109205478e-06,
      "loss": 0.2652,
      "num_input_tokens_seen": 16429048,
      "step": 25155
    },
    {
      "epoch": 14.834905660377359,
      "grad_norm": 5.459967613220215,
      "learning_rate": 1.898149896859567e-06,
      "loss": 0.426,
      "num_input_tokens_seen": 16431768,
      "step": 25160
    },
    {
      "epoch": 14.837853773584905,
      "grad_norm": 1.9723478555679321,
      "learning_rate": 1.8961325040267714e-06,
      "loss": 0.2506,
      "num_input_tokens_seen": 16434904,
      "step": 25165
    },
    {
      "epoch": 14.840801886792454,
      "grad_norm": 3.8622987270355225,
      "learning_rate": 1.894115932956272e-06,
      "loss": 0.2935,
      "num_input_tokens_seen": 16437912,
      "step": 25170
    },
    {
      "epoch": 14.84375,
      "grad_norm": 3.6371262073516846,
      "learning_rate": 1.8921001841819652e-06,
      "loss": 0.3062,
      "num_input_tokens_seen": 16441208,
      "step": 25175
    },
    {
      "epoch": 14.846698113207546,
      "grad_norm": 7.932714939117432,
      "learning_rate": 1.8900852582375284e-06,
      "loss": 0.3655,
      "num_input_tokens_seen": 16444568,
      "step": 25180
    },
    {
      "epoch": 14.849646226415095,
      "grad_norm": 4.321752071380615,
      "learning_rate": 1.8880711556564214e-06,
      "loss": 0.2747,
      "num_input_tokens_seen": 16448216,
      "step": 25185
    },
    {
      "epoch": 14.852594339622641,
      "grad_norm": 3.3801183700561523,
      "learning_rate": 1.8860578769718891e-06,
      "loss": 0.3763,
      "num_input_tokens_seen": 16450712,
      "step": 25190
    },
    {
      "epoch": 14.85554245283019,
      "grad_norm": 5.338661193847656,
      "learning_rate": 1.8840454227169525e-06,
      "loss": 0.398,
      "num_input_tokens_seen": 16453880,
      "step": 25195
    },
    {
      "epoch": 14.858490566037736,
      "grad_norm": 6.127860069274902,
      "learning_rate": 1.882033793424421e-06,
      "loss": 0.3009,
      "num_input_tokens_seen": 16456856,
      "step": 25200
    },
    {
      "epoch": 14.861438679245284,
      "grad_norm": 2.8764519691467285,
      "learning_rate": 1.88002298962688e-06,
      "loss": 0.2878,
      "num_input_tokens_seen": 16459704,
      "step": 25205
    },
    {
      "epoch": 14.86438679245283,
      "grad_norm": 3.58858585357666,
      "learning_rate": 1.8780130118566996e-06,
      "loss": 0.3166,
      "num_input_tokens_seen": 16462680,
      "step": 25210
    },
    {
      "epoch": 14.867334905660378,
      "grad_norm": 2.39900279045105,
      "learning_rate": 1.876003860646029e-06,
      "loss": 0.2231,
      "num_input_tokens_seen": 16465528,
      "step": 25215
    },
    {
      "epoch": 14.870283018867925,
      "grad_norm": 4.4556379318237305,
      "learning_rate": 1.8739955365267997e-06,
      "loss": 0.319,
      "num_input_tokens_seen": 16468280,
      "step": 25220
    },
    {
      "epoch": 14.873231132075471,
      "grad_norm": 2.7291786670684814,
      "learning_rate": 1.8719880400307228e-06,
      "loss": 0.3685,
      "num_input_tokens_seen": 16471864,
      "step": 25225
    },
    {
      "epoch": 14.87617924528302,
      "grad_norm": 4.0243659019470215,
      "learning_rate": 1.869981371689295e-06,
      "loss": 0.336,
      "num_input_tokens_seen": 16475992,
      "step": 25230
    },
    {
      "epoch": 14.879127358490566,
      "grad_norm": 3.0323448181152344,
      "learning_rate": 1.867975532033789e-06,
      "loss": 0.3583,
      "num_input_tokens_seen": 16480120,
      "step": 25235
    },
    {
      "epoch": 14.882075471698114,
      "grad_norm": 1.9534499645233154,
      "learning_rate": 1.8659705215952589e-06,
      "loss": 0.301,
      "num_input_tokens_seen": 16483512,
      "step": 25240
    },
    {
      "epoch": 14.88502358490566,
      "grad_norm": 1.823805570602417,
      "learning_rate": 1.8639663409045405e-06,
      "loss": 0.3772,
      "num_input_tokens_seen": 16486296,
      "step": 25245
    },
    {
      "epoch": 14.887971698113208,
      "grad_norm": 2.388416290283203,
      "learning_rate": 1.8619629904922466e-06,
      "loss": 0.3067,
      "num_input_tokens_seen": 16490008,
      "step": 25250
    },
    {
      "epoch": 14.890919811320755,
      "grad_norm": 3.073296308517456,
      "learning_rate": 1.859960470888777e-06,
      "loss": 0.3912,
      "num_input_tokens_seen": 16493112,
      "step": 25255
    },
    {
      "epoch": 14.893867924528301,
      "grad_norm": 4.864343166351318,
      "learning_rate": 1.857958782624306e-06,
      "loss": 0.5101,
      "num_input_tokens_seen": 16495640,
      "step": 25260
    },
    {
      "epoch": 14.89681603773585,
      "grad_norm": 3.0970358848571777,
      "learning_rate": 1.8559579262287886e-06,
      "loss": 0.4065,
      "num_input_tokens_seen": 16498040,
      "step": 25265
    },
    {
      "epoch": 14.899764150943396,
      "grad_norm": 4.0576276779174805,
      "learning_rate": 1.8539579022319599e-06,
      "loss": 0.2834,
      "num_input_tokens_seen": 16501240,
      "step": 25270
    },
    {
      "epoch": 14.902712264150944,
      "grad_norm": 4.132444858551025,
      "learning_rate": 1.8519587111633357e-06,
      "loss": 0.3418,
      "num_input_tokens_seen": 16503896,
      "step": 25275
    },
    {
      "epoch": 14.90566037735849,
      "grad_norm": 4.599870681762695,
      "learning_rate": 1.8499603535522082e-06,
      "loss": 0.3834,
      "num_input_tokens_seen": 16506296,
      "step": 25280
    },
    {
      "epoch": 14.908608490566039,
      "grad_norm": 3.0442163944244385,
      "learning_rate": 1.8479628299276543e-06,
      "loss": 0.2442,
      "num_input_tokens_seen": 16509336,
      "step": 25285
    },
    {
      "epoch": 14.911556603773585,
      "grad_norm": 2.712388038635254,
      "learning_rate": 1.8459661408185241e-06,
      "loss": 0.2495,
      "num_input_tokens_seen": 16512920,
      "step": 25290
    },
    {
      "epoch": 14.914504716981131,
      "grad_norm": 6.849637508392334,
      "learning_rate": 1.8439702867534536e-06,
      "loss": 0.2916,
      "num_input_tokens_seen": 16515704,
      "step": 25295
    },
    {
      "epoch": 14.91745283018868,
      "grad_norm": 4.027571678161621,
      "learning_rate": 1.841975268260851e-06,
      "loss": 0.2867,
      "num_input_tokens_seen": 16518360,
      "step": 25300
    },
    {
      "epoch": 14.920400943396226,
      "grad_norm": 6.9468913078308105,
      "learning_rate": 1.8399810858689066e-06,
      "loss": 0.4697,
      "num_input_tokens_seen": 16520888,
      "step": 25305
    },
    {
      "epoch": 14.923349056603774,
      "grad_norm": 5.834597587585449,
      "learning_rate": 1.8379877401055884e-06,
      "loss": 0.278,
      "num_input_tokens_seen": 16523480,
      "step": 25310
    },
    {
      "epoch": 14.92629716981132,
      "grad_norm": 2.960740804672241,
      "learning_rate": 1.8359952314986418e-06,
      "loss": 0.4943,
      "num_input_tokens_seen": 16526648,
      "step": 25315
    },
    {
      "epoch": 14.929245283018869,
      "grad_norm": 3.694031000137329,
      "learning_rate": 1.8340035605755957e-06,
      "loss": 0.2614,
      "num_input_tokens_seen": 16529688,
      "step": 25320
    },
    {
      "epoch": 14.932193396226415,
      "grad_norm": 6.373254776000977,
      "learning_rate": 1.8320127278637518e-06,
      "loss": 0.3373,
      "num_input_tokens_seen": 16533240,
      "step": 25325
    },
    {
      "epoch": 14.935141509433961,
      "grad_norm": 6.575335502624512,
      "learning_rate": 1.830022733890191e-06,
      "loss": 0.4166,
      "num_input_tokens_seen": 16536344,
      "step": 25330
    },
    {
      "epoch": 14.93808962264151,
      "grad_norm": 6.3244123458862305,
      "learning_rate": 1.8280335791817733e-06,
      "loss": 0.2424,
      "num_input_tokens_seen": 16539224,
      "step": 25335
    },
    {
      "epoch": 14.941037735849056,
      "grad_norm": 2.961890459060669,
      "learning_rate": 1.826045264265136e-06,
      "loss": 0.2813,
      "num_input_tokens_seen": 16543224,
      "step": 25340
    },
    {
      "epoch": 14.943985849056604,
      "grad_norm": 5.144169807434082,
      "learning_rate": 1.8240577896666928e-06,
      "loss": 0.3727,
      "num_input_tokens_seen": 16546744,
      "step": 25345
    },
    {
      "epoch": 14.94693396226415,
      "grad_norm": 3.771613597869873,
      "learning_rate": 1.8220711559126382e-06,
      "loss": 0.3105,
      "num_input_tokens_seen": 16549336,
      "step": 25350
    },
    {
      "epoch": 14.949882075471699,
      "grad_norm": 3.303913116455078,
      "learning_rate": 1.8200853635289417e-06,
      "loss": 0.2301,
      "num_input_tokens_seen": 16552248,
      "step": 25355
    },
    {
      "epoch": 14.952830188679245,
      "grad_norm": 3.0048983097076416,
      "learning_rate": 1.81810041304135e-06,
      "loss": 0.2863,
      "num_input_tokens_seen": 16555800,
      "step": 25360
    },
    {
      "epoch": 14.955778301886792,
      "grad_norm": 4.158670425415039,
      "learning_rate": 1.8161163049753865e-06,
      "loss": 0.3241,
      "num_input_tokens_seen": 16558904,
      "step": 25365
    },
    {
      "epoch": 14.95872641509434,
      "grad_norm": 3.664205312728882,
      "learning_rate": 1.8141330398563533e-06,
      "loss": 0.3297,
      "num_input_tokens_seen": 16561880,
      "step": 25370
    },
    {
      "epoch": 14.961674528301886,
      "grad_norm": 3.264677047729492,
      "learning_rate": 1.8121506182093268e-06,
      "loss": 0.3352,
      "num_input_tokens_seen": 16565176,
      "step": 25375
    },
    {
      "epoch": 14.964622641509434,
      "grad_norm": 3.20595645904541,
      "learning_rate": 1.8101690405591643e-06,
      "loss": 0.3685,
      "num_input_tokens_seen": 16567864,
      "step": 25380
    },
    {
      "epoch": 14.96757075471698,
      "grad_norm": 1.9541380405426025,
      "learning_rate": 1.8081883074304945e-06,
      "loss": 0.2538,
      "num_input_tokens_seen": 16571256,
      "step": 25385
    },
    {
      "epoch": 14.970518867924529,
      "grad_norm": 1.7414108514785767,
      "learning_rate": 1.8062084193477275e-06,
      "loss": 0.2816,
      "num_input_tokens_seen": 16575224,
      "step": 25390
    },
    {
      "epoch": 14.973466981132075,
      "grad_norm": 4.276235580444336,
      "learning_rate": 1.804229376835046e-06,
      "loss": 0.2987,
      "num_input_tokens_seen": 16580056,
      "step": 25395
    },
    {
      "epoch": 14.976415094339622,
      "grad_norm": 2.6271822452545166,
      "learning_rate": 1.8022511804164105e-06,
      "loss": 0.3798,
      "num_input_tokens_seen": 16583448,
      "step": 25400
    },
    {
      "epoch": 14.97936320754717,
      "grad_norm": 2.8101799488067627,
      "learning_rate": 1.8002738306155559e-06,
      "loss": 0.3536,
      "num_input_tokens_seen": 16586488,
      "step": 25405
    },
    {
      "epoch": 14.982311320754716,
      "grad_norm": 6.60940408706665,
      "learning_rate": 1.7982973279559935e-06,
      "loss": 0.2914,
      "num_input_tokens_seen": 16589272,
      "step": 25410
    },
    {
      "epoch": 14.985259433962264,
      "grad_norm": 5.9747314453125,
      "learning_rate": 1.7963216729610134e-06,
      "loss": 0.2614,
      "num_input_tokens_seen": 16593368,
      "step": 25415
    },
    {
      "epoch": 14.98820754716981,
      "grad_norm": 3.487112045288086,
      "learning_rate": 1.7943468661536773e-06,
      "loss": 0.2487,
      "num_input_tokens_seen": 16596056,
      "step": 25420
    },
    {
      "epoch": 14.991155660377359,
      "grad_norm": 3.2089550495147705,
      "learning_rate": 1.7923729080568242e-06,
      "loss": 0.4671,
      "num_input_tokens_seen": 16599192,
      "step": 25425
    },
    {
      "epoch": 14.994103773584905,
      "grad_norm": 4.636631965637207,
      "learning_rate": 1.7903997991930683e-06,
      "loss": 0.3711,
      "num_input_tokens_seen": 16602360,
      "step": 25430
    },
    {
      "epoch": 14.997051886792454,
      "grad_norm": 3.2209982872009277,
      "learning_rate": 1.7884275400847972e-06,
      "loss": 0.2845,
      "num_input_tokens_seen": 16605816,
      "step": 25435
    },
    {
      "epoch": 15.0,
      "grad_norm": 9.4528226852417,
      "learning_rate": 1.786456131254175e-06,
      "loss": 0.2957,
      "num_input_tokens_seen": 16608128,
      "step": 25440
    },
    {
      "epoch": 15.002948113207546,
      "grad_norm": 4.454707145690918,
      "learning_rate": 1.784485573223143e-06,
      "loss": 0.3144,
      "num_input_tokens_seen": 16611392,
      "step": 25445
    },
    {
      "epoch": 15.005896226415095,
      "grad_norm": 2.239672899246216,
      "learning_rate": 1.782515866513414e-06,
      "loss": 0.3029,
      "num_input_tokens_seen": 16614944,
      "step": 25450
    },
    {
      "epoch": 15.008844339622641,
      "grad_norm": 2.878877878189087,
      "learning_rate": 1.7805470116464758e-06,
      "loss": 0.2815,
      "num_input_tokens_seen": 16618240,
      "step": 25455
    },
    {
      "epoch": 15.01179245283019,
      "grad_norm": 4.851008415222168,
      "learning_rate": 1.7785790091435911e-06,
      "loss": 0.29,
      "num_input_tokens_seen": 16620832,
      "step": 25460
    },
    {
      "epoch": 15.014740566037736,
      "grad_norm": 3.1711878776550293,
      "learning_rate": 1.776611859525796e-06,
      "loss": 0.3699,
      "num_input_tokens_seen": 16623616,
      "step": 25465
    },
    {
      "epoch": 15.017688679245284,
      "grad_norm": 3.8151462078094482,
      "learning_rate": 1.7746455633139042e-06,
      "loss": 0.3083,
      "num_input_tokens_seen": 16626560,
      "step": 25470
    },
    {
      "epoch": 15.02063679245283,
      "grad_norm": 4.4690632820129395,
      "learning_rate": 1.7726801210285005e-06,
      "loss": 0.4059,
      "num_input_tokens_seen": 16629632,
      "step": 25475
    },
    {
      "epoch": 15.023584905660377,
      "grad_norm": 2.7234582901000977,
      "learning_rate": 1.7707155331899418e-06,
      "loss": 0.3657,
      "num_input_tokens_seen": 16632576,
      "step": 25480
    },
    {
      "epoch": 15.026533018867925,
      "grad_norm": 4.251256465911865,
      "learning_rate": 1.7687518003183645e-06,
      "loss": 0.2627,
      "num_input_tokens_seen": 16637632,
      "step": 25485
    },
    {
      "epoch": 15.029481132075471,
      "grad_norm": 4.878654956817627,
      "learning_rate": 1.766788922933675e-06,
      "loss": 0.2495,
      "num_input_tokens_seen": 16640192,
      "step": 25490
    },
    {
      "epoch": 15.03242924528302,
      "grad_norm": 4.349775314331055,
      "learning_rate": 1.7648269015555514e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 16643200,
      "step": 25495
    },
    {
      "epoch": 15.035377358490566,
      "grad_norm": 2.7638635635375977,
      "learning_rate": 1.7628657367034474e-06,
      "loss": 0.2781,
      "num_input_tokens_seen": 16647584,
      "step": 25500
    },
    {
      "epoch": 15.038325471698114,
      "grad_norm": 4.539735317230225,
      "learning_rate": 1.7609054288965922e-06,
      "loss": 0.3434,
      "num_input_tokens_seen": 16650592,
      "step": 25505
    },
    {
      "epoch": 15.04127358490566,
      "grad_norm": 2.626659870147705,
      "learning_rate": 1.7589459786539847e-06,
      "loss": 0.2704,
      "num_input_tokens_seen": 16653824,
      "step": 25510
    },
    {
      "epoch": 15.044221698113208,
      "grad_norm": 3.5171616077423096,
      "learning_rate": 1.7569873864943975e-06,
      "loss": 0.3218,
      "num_input_tokens_seen": 16658624,
      "step": 25515
    },
    {
      "epoch": 15.047169811320755,
      "grad_norm": 3.7473433017730713,
      "learning_rate": 1.7550296529363764e-06,
      "loss": 0.2467,
      "num_input_tokens_seen": 16662144,
      "step": 25520
    },
    {
      "epoch": 15.050117924528301,
      "grad_norm": 1.9571799039840698,
      "learning_rate": 1.7530727784982393e-06,
      "loss": 0.3154,
      "num_input_tokens_seen": 16666208,
      "step": 25525
    },
    {
      "epoch": 15.05306603773585,
      "grad_norm": 3.209031820297241,
      "learning_rate": 1.7511167636980765e-06,
      "loss": 0.2863,
      "num_input_tokens_seen": 16669728,
      "step": 25530
    },
    {
      "epoch": 15.056014150943396,
      "grad_norm": 2.7215023040771484,
      "learning_rate": 1.7491616090537539e-06,
      "loss": 0.2888,
      "num_input_tokens_seen": 16673120,
      "step": 25535
    },
    {
      "epoch": 15.058962264150944,
      "grad_norm": 3.417344570159912,
      "learning_rate": 1.7472073150829056e-06,
      "loss": 0.3318,
      "num_input_tokens_seen": 16678144,
      "step": 25540
    },
    {
      "epoch": 15.06191037735849,
      "grad_norm": 4.313814163208008,
      "learning_rate": 1.745253882302939e-06,
      "loss": 0.3733,
      "num_input_tokens_seen": 16681312,
      "step": 25545
    },
    {
      "epoch": 15.064858490566039,
      "grad_norm": 3.0025970935821533,
      "learning_rate": 1.743301311231035e-06,
      "loss": 0.305,
      "num_input_tokens_seen": 16685216,
      "step": 25550
    },
    {
      "epoch": 15.067806603773585,
      "grad_norm": 4.73638916015625,
      "learning_rate": 1.7413496023841437e-06,
      "loss": 0.2366,
      "num_input_tokens_seen": 16688736,
      "step": 25555
    },
    {
      "epoch": 15.070754716981131,
      "grad_norm": 2.8894217014312744,
      "learning_rate": 1.7393987562789876e-06,
      "loss": 0.2316,
      "num_input_tokens_seen": 16692000,
      "step": 25560
    },
    {
      "epoch": 15.07370283018868,
      "grad_norm": 4.167801856994629,
      "learning_rate": 1.7374487734320655e-06,
      "loss": 0.3287,
      "num_input_tokens_seen": 16694912,
      "step": 25565
    },
    {
      "epoch": 15.076650943396226,
      "grad_norm": 3.0136146545410156,
      "learning_rate": 1.7354996543596408e-06,
      "loss": 0.2694,
      "num_input_tokens_seen": 16698016,
      "step": 25570
    },
    {
      "epoch": 15.079599056603774,
      "grad_norm": 2.9114134311676025,
      "learning_rate": 1.7335513995777504e-06,
      "loss": 0.2626,
      "num_input_tokens_seen": 16701088,
      "step": 25575
    },
    {
      "epoch": 15.08254716981132,
      "grad_norm": 3.192774772644043,
      "learning_rate": 1.7316040096022062e-06,
      "loss": 0.364,
      "num_input_tokens_seen": 16703552,
      "step": 25580
    },
    {
      "epoch": 15.085495283018869,
      "grad_norm": 4.4815673828125,
      "learning_rate": 1.7296574849485863e-06,
      "loss": 0.284,
      "num_input_tokens_seen": 16706848,
      "step": 25585
    },
    {
      "epoch": 15.088443396226415,
      "grad_norm": 4.820929527282715,
      "learning_rate": 1.7277118261322423e-06,
      "loss": 0.2834,
      "num_input_tokens_seen": 16709664,
      "step": 25590
    },
    {
      "epoch": 15.091391509433961,
      "grad_norm": 9.346332550048828,
      "learning_rate": 1.7257670336682925e-06,
      "loss": 0.2542,
      "num_input_tokens_seen": 16712896,
      "step": 25595
    },
    {
      "epoch": 15.09433962264151,
      "grad_norm": 5.682065486907959,
      "learning_rate": 1.7238231080716339e-06,
      "loss": 0.4803,
      "num_input_tokens_seen": 16715680,
      "step": 25600
    },
    {
      "epoch": 15.097287735849056,
      "grad_norm": 3.4871997833251953,
      "learning_rate": 1.721880049856927e-06,
      "loss": 0.3053,
      "num_input_tokens_seen": 16718848,
      "step": 25605
    },
    {
      "epoch": 15.100235849056604,
      "grad_norm": 7.554766654968262,
      "learning_rate": 1.7199378595386046e-06,
      "loss": 0.2962,
      "num_input_tokens_seen": 16721600,
      "step": 25610
    },
    {
      "epoch": 15.10318396226415,
      "grad_norm": 3.130730152130127,
      "learning_rate": 1.7179965376308705e-06,
      "loss": 0.2975,
      "num_input_tokens_seen": 16724224,
      "step": 25615
    },
    {
      "epoch": 15.106132075471699,
      "grad_norm": 3.5021908283233643,
      "learning_rate": 1.7160560846476976e-06,
      "loss": 0.2512,
      "num_input_tokens_seen": 16728096,
      "step": 25620
    },
    {
      "epoch": 15.109080188679245,
      "grad_norm": 2.9954609870910645,
      "learning_rate": 1.7141165011028277e-06,
      "loss": 0.2439,
      "num_input_tokens_seen": 16731168,
      "step": 25625
    },
    {
      "epoch": 15.112028301886792,
      "grad_norm": 2.9942305088043213,
      "learning_rate": 1.7121777875097767e-06,
      "loss": 0.2955,
      "num_input_tokens_seen": 16734720,
      "step": 25630
    },
    {
      "epoch": 15.11497641509434,
      "grad_norm": 3.5002331733703613,
      "learning_rate": 1.7102399443818268e-06,
      "loss": 0.2936,
      "num_input_tokens_seen": 16738496,
      "step": 25635
    },
    {
      "epoch": 15.117924528301886,
      "grad_norm": 4.55237340927124,
      "learning_rate": 1.7083029722320294e-06,
      "loss": 0.4011,
      "num_input_tokens_seen": 16742848,
      "step": 25640
    },
    {
      "epoch": 15.120872641509434,
      "grad_norm": 3.385624408721924,
      "learning_rate": 1.7063668715732063e-06,
      "loss": 0.3005,
      "num_input_tokens_seen": 16746240,
      "step": 25645
    },
    {
      "epoch": 15.12382075471698,
      "grad_norm": 2.2249433994293213,
      "learning_rate": 1.7044316429179492e-06,
      "loss": 0.3792,
      "num_input_tokens_seen": 16750528,
      "step": 25650
    },
    {
      "epoch": 15.126768867924529,
      "grad_norm": 2.7778255939483643,
      "learning_rate": 1.7024972867786155e-06,
      "loss": 0.2816,
      "num_input_tokens_seen": 16755680,
      "step": 25655
    },
    {
      "epoch": 15.129716981132075,
      "grad_norm": 2.397848606109619,
      "learning_rate": 1.7005638036673389e-06,
      "loss": 0.2923,
      "num_input_tokens_seen": 16758432,
      "step": 25660
    },
    {
      "epoch": 15.132665094339623,
      "grad_norm": 4.942976951599121,
      "learning_rate": 1.6986311940960148e-06,
      "loss": 0.3664,
      "num_input_tokens_seen": 16760896,
      "step": 25665
    },
    {
      "epoch": 15.13561320754717,
      "grad_norm": 5.957855224609375,
      "learning_rate": 1.696699458576308e-06,
      "loss": 0.3304,
      "num_input_tokens_seen": 16764032,
      "step": 25670
    },
    {
      "epoch": 15.138561320754716,
      "grad_norm": 4.301231384277344,
      "learning_rate": 1.6947685976196581e-06,
      "loss": 0.3979,
      "num_input_tokens_seen": 16766912,
      "step": 25675
    },
    {
      "epoch": 15.141509433962264,
      "grad_norm": 2.506430149078369,
      "learning_rate": 1.692838611737267e-06,
      "loss": 0.3337,
      "num_input_tokens_seen": 16770304,
      "step": 25680
    },
    {
      "epoch": 15.14445754716981,
      "grad_norm": 5.286166667938232,
      "learning_rate": 1.690909501440106e-06,
      "loss": 0.3284,
      "num_input_tokens_seen": 16772768,
      "step": 25685
    },
    {
      "epoch": 15.147405660377359,
      "grad_norm": 2.4631235599517822,
      "learning_rate": 1.688981267238915e-06,
      "loss": 0.2909,
      "num_input_tokens_seen": 16775936,
      "step": 25690
    },
    {
      "epoch": 15.150353773584905,
      "grad_norm": 6.819662570953369,
      "learning_rate": 1.687053909644204e-06,
      "loss": 0.3852,
      "num_input_tokens_seen": 16778784,
      "step": 25695
    },
    {
      "epoch": 15.153301886792454,
      "grad_norm": 5.26468563079834,
      "learning_rate": 1.685127429166249e-06,
      "loss": 0.3124,
      "num_input_tokens_seen": 16782176,
      "step": 25700
    },
    {
      "epoch": 15.15625,
      "grad_norm": 3.489720582962036,
      "learning_rate": 1.683201826315093e-06,
      "loss": 0.389,
      "num_input_tokens_seen": 16785344,
      "step": 25705
    },
    {
      "epoch": 15.159198113207546,
      "grad_norm": 3.3917508125305176,
      "learning_rate": 1.681277101600548e-06,
      "loss": 0.4336,
      "num_input_tokens_seen": 16788032,
      "step": 25710
    },
    {
      "epoch": 15.162146226415095,
      "grad_norm": 2.298631429672241,
      "learning_rate": 1.6793532555321939e-06,
      "loss": 0.3817,
      "num_input_tokens_seen": 16791648,
      "step": 25715
    },
    {
      "epoch": 15.165094339622641,
      "grad_norm": 3.780439853668213,
      "learning_rate": 1.6774302886193744e-06,
      "loss": 0.4318,
      "num_input_tokens_seen": 16794368,
      "step": 25720
    },
    {
      "epoch": 15.16804245283019,
      "grad_norm": 5.0596022605896,
      "learning_rate": 1.6755082013712076e-06,
      "loss": 0.3124,
      "num_input_tokens_seen": 16796640,
      "step": 25725
    },
    {
      "epoch": 15.170990566037736,
      "grad_norm": 4.355698108673096,
      "learning_rate": 1.6735869942965716e-06,
      "loss": 0.3717,
      "num_input_tokens_seen": 16799616,
      "step": 25730
    },
    {
      "epoch": 15.173938679245284,
      "grad_norm": 2.7480270862579346,
      "learning_rate": 1.6716666679041155e-06,
      "loss": 0.3702,
      "num_input_tokens_seen": 16802784,
      "step": 25735
    },
    {
      "epoch": 15.17688679245283,
      "grad_norm": 2.728898286819458,
      "learning_rate": 1.6697472227022533e-06,
      "loss": 0.2783,
      "num_input_tokens_seen": 16805568,
      "step": 25740
    },
    {
      "epoch": 15.179834905660377,
      "grad_norm": 2.5315985679626465,
      "learning_rate": 1.6678286591991644e-06,
      "loss": 0.2429,
      "num_input_tokens_seen": 16809312,
      "step": 25745
    },
    {
      "epoch": 15.182783018867925,
      "grad_norm": 8.58702564239502,
      "learning_rate": 1.665910977902801e-06,
      "loss": 0.4033,
      "num_input_tokens_seen": 16811328,
      "step": 25750
    },
    {
      "epoch": 15.185731132075471,
      "grad_norm": 3.6224279403686523,
      "learning_rate": 1.6639941793208747e-06,
      "loss": 0.357,
      "num_input_tokens_seen": 16814144,
      "step": 25755
    },
    {
      "epoch": 15.18867924528302,
      "grad_norm": 2.36954402923584,
      "learning_rate": 1.6620782639608674e-06,
      "loss": 0.2642,
      "num_input_tokens_seen": 16817248,
      "step": 25760
    },
    {
      "epoch": 15.191627358490566,
      "grad_norm": 2.8240365982055664,
      "learning_rate": 1.6601632323300231e-06,
      "loss": 0.3698,
      "num_input_tokens_seen": 16820800,
      "step": 25765
    },
    {
      "epoch": 15.194575471698114,
      "grad_norm": 3.197333574295044,
      "learning_rate": 1.6582490849353595e-06,
      "loss": 0.3494,
      "num_input_tokens_seen": 16823872,
      "step": 25770
    },
    {
      "epoch": 15.19752358490566,
      "grad_norm": 5.338634490966797,
      "learning_rate": 1.6563358222836523e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 16826560,
      "step": 25775
    },
    {
      "epoch": 15.200471698113208,
      "grad_norm": 3.3730363845825195,
      "learning_rate": 1.654423444881445e-06,
      "loss": 0.353,
      "num_input_tokens_seen": 16829920,
      "step": 25780
    },
    {
      "epoch": 15.203419811320755,
      "grad_norm": 6.791359901428223,
      "learning_rate": 1.652511953235051e-06,
      "loss": 0.4059,
      "num_input_tokens_seen": 16833280,
      "step": 25785
    },
    {
      "epoch": 15.206367924528301,
      "grad_norm": 4.881641864776611,
      "learning_rate": 1.650601347850544e-06,
      "loss": 0.2628,
      "num_input_tokens_seen": 16837088,
      "step": 25790
    },
    {
      "epoch": 15.20931603773585,
      "grad_norm": 5.139157772064209,
      "learning_rate": 1.6486916292337652e-06,
      "loss": 0.2907,
      "num_input_tokens_seen": 16841024,
      "step": 25795
    },
    {
      "epoch": 15.212264150943396,
      "grad_norm": 4.052705764770508,
      "learning_rate": 1.6467827978903212e-06,
      "loss": 0.285,
      "num_input_tokens_seen": 16843648,
      "step": 25800
    },
    {
      "epoch": 15.215212264150944,
      "grad_norm": 2.7996513843536377,
      "learning_rate": 1.6448748543255827e-06,
      "loss": 0.2227,
      "num_input_tokens_seen": 16846528,
      "step": 25805
    },
    {
      "epoch": 15.21816037735849,
      "grad_norm": 5.10759973526001,
      "learning_rate": 1.6429677990446845e-06,
      "loss": 0.2729,
      "num_input_tokens_seen": 16849152,
      "step": 25810
    },
    {
      "epoch": 15.221108490566039,
      "grad_norm": 4.511544704437256,
      "learning_rate": 1.6410616325525319e-06,
      "loss": 0.2593,
      "num_input_tokens_seen": 16853312,
      "step": 25815
    },
    {
      "epoch": 15.224056603773585,
      "grad_norm": 3.1804723739624023,
      "learning_rate": 1.6391563553537875e-06,
      "loss": 0.2888,
      "num_input_tokens_seen": 16855968,
      "step": 25820
    },
    {
      "epoch": 15.227004716981131,
      "grad_norm": 2.1584951877593994,
      "learning_rate": 1.6372519679528832e-06,
      "loss": 0.2297,
      "num_input_tokens_seen": 16858816,
      "step": 25825
    },
    {
      "epoch": 15.22995283018868,
      "grad_norm": 3.4638116359710693,
      "learning_rate": 1.6353484708540124e-06,
      "loss": 0.284,
      "num_input_tokens_seen": 16862240,
      "step": 25830
    },
    {
      "epoch": 15.232900943396226,
      "grad_norm": 3.020129919052124,
      "learning_rate": 1.633445864561135e-06,
      "loss": 0.216,
      "num_input_tokens_seen": 16865312,
      "step": 25835
    },
    {
      "epoch": 15.235849056603774,
      "grad_norm": 3.6519789695739746,
      "learning_rate": 1.6315441495779726e-06,
      "loss": 0.3221,
      "num_input_tokens_seen": 16867936,
      "step": 25840
    },
    {
      "epoch": 15.23879716981132,
      "grad_norm": 2.9586479663848877,
      "learning_rate": 1.6296433264080152e-06,
      "loss": 0.3484,
      "num_input_tokens_seen": 16871968,
      "step": 25845
    },
    {
      "epoch": 15.241745283018869,
      "grad_norm": 3.5766379833221436,
      "learning_rate": 1.627743395554513e-06,
      "loss": 0.2495,
      "num_input_tokens_seen": 16874464,
      "step": 25850
    },
    {
      "epoch": 15.244693396226415,
      "grad_norm": 3.1976966857910156,
      "learning_rate": 1.6258443575204802e-06,
      "loss": 0.3398,
      "num_input_tokens_seen": 16878304,
      "step": 25855
    },
    {
      "epoch": 15.247641509433961,
      "grad_norm": 4.662925720214844,
      "learning_rate": 1.6239462128086936e-06,
      "loss": 0.4084,
      "num_input_tokens_seen": 16880992,
      "step": 25860
    },
    {
      "epoch": 15.25058962264151,
      "grad_norm": 6.328319072723389,
      "learning_rate": 1.6220489619216988e-06,
      "loss": 0.305,
      "num_input_tokens_seen": 16884416,
      "step": 25865
    },
    {
      "epoch": 15.253537735849056,
      "grad_norm": 3.5702531337738037,
      "learning_rate": 1.6201526053618e-06,
      "loss": 0.3347,
      "num_input_tokens_seen": 16890112,
      "step": 25870
    },
    {
      "epoch": 15.256485849056604,
      "grad_norm": 2.707454204559326,
      "learning_rate": 1.6182571436310634e-06,
      "loss": 0.2463,
      "num_input_tokens_seen": 16893792,
      "step": 25875
    },
    {
      "epoch": 15.25943396226415,
      "grad_norm": 2.448408365249634,
      "learning_rate": 1.616362577231324e-06,
      "loss": 0.2625,
      "num_input_tokens_seen": 16897376,
      "step": 25880
    },
    {
      "epoch": 15.262382075471699,
      "grad_norm": 3.3578197956085205,
      "learning_rate": 1.614468906664175e-06,
      "loss": 0.2669,
      "num_input_tokens_seen": 16899968,
      "step": 25885
    },
    {
      "epoch": 15.265330188679245,
      "grad_norm": 2.9495012760162354,
      "learning_rate": 1.612576132430974e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 16902944,
      "step": 25890
    },
    {
      "epoch": 15.268278301886792,
      "grad_norm": 6.448451995849609,
      "learning_rate": 1.6106842550328406e-06,
      "loss": 0.2346,
      "num_input_tokens_seen": 16905664,
      "step": 25895
    },
    {
      "epoch": 15.27122641509434,
      "grad_norm": 2.8032243251800537,
      "learning_rate": 1.6087932749706582e-06,
      "loss": 0.2025,
      "num_input_tokens_seen": 16912096,
      "step": 25900
    },
    {
      "epoch": 15.274174528301886,
      "grad_norm": 3.229433059692383,
      "learning_rate": 1.6069031927450696e-06,
      "loss": 0.2872,
      "num_input_tokens_seen": 16915616,
      "step": 25905
    },
    {
      "epoch": 15.277122641509434,
      "grad_norm": 2.603994607925415,
      "learning_rate": 1.605014008856486e-06,
      "loss": 0.2822,
      "num_input_tokens_seen": 16919008,
      "step": 25910
    },
    {
      "epoch": 15.28007075471698,
      "grad_norm": 3.950531005859375,
      "learning_rate": 1.6031257238050745e-06,
      "loss": 0.4182,
      "num_input_tokens_seen": 16921568,
      "step": 25915
    },
    {
      "epoch": 15.283018867924529,
      "grad_norm": 2.3392112255096436,
      "learning_rate": 1.601238338090768e-06,
      "loss": 0.238,
      "num_input_tokens_seen": 16924128,
      "step": 25920
    },
    {
      "epoch": 15.285966981132075,
      "grad_norm": 2.9155807495117188,
      "learning_rate": 1.5993518522132595e-06,
      "loss": 0.2561,
      "num_input_tokens_seen": 16927776,
      "step": 25925
    },
    {
      "epoch": 15.288915094339623,
      "grad_norm": 4.762979030609131,
      "learning_rate": 1.5974662666720037e-06,
      "loss": 0.3429,
      "num_input_tokens_seen": 16930304,
      "step": 25930
    },
    {
      "epoch": 15.29186320754717,
      "grad_norm": 3.616849184036255,
      "learning_rate": 1.5955815819662162e-06,
      "loss": 0.2773,
      "num_input_tokens_seen": 16933920,
      "step": 25935
    },
    {
      "epoch": 15.294811320754716,
      "grad_norm": 1.9612195491790771,
      "learning_rate": 1.5936977985948788e-06,
      "loss": 0.2703,
      "num_input_tokens_seen": 16937280,
      "step": 25940
    },
    {
      "epoch": 15.297759433962264,
      "grad_norm": 5.351400852203369,
      "learning_rate": 1.5918149170567298e-06,
      "loss": 0.3939,
      "num_input_tokens_seen": 16939648,
      "step": 25945
    },
    {
      "epoch": 15.30070754716981,
      "grad_norm": 3.1368401050567627,
      "learning_rate": 1.5899329378502698e-06,
      "loss": 0.2569,
      "num_input_tokens_seen": 16943296,
      "step": 25950
    },
    {
      "epoch": 15.303655660377359,
      "grad_norm": 2.486192226409912,
      "learning_rate": 1.588051861473761e-06,
      "loss": 0.2147,
      "num_input_tokens_seen": 16946016,
      "step": 25955
    },
    {
      "epoch": 15.306603773584905,
      "grad_norm": 5.314333915710449,
      "learning_rate": 1.5861716884252253e-06,
      "loss": 0.3417,
      "num_input_tokens_seen": 16949248,
      "step": 25960
    },
    {
      "epoch": 15.309551886792454,
      "grad_norm": 4.597508430480957,
      "learning_rate": 1.5842924192024489e-06,
      "loss": 0.2476,
      "num_input_tokens_seen": 16951968,
      "step": 25965
    },
    {
      "epoch": 15.3125,
      "grad_norm": 3.496854305267334,
      "learning_rate": 1.5824140543029742e-06,
      "loss": 0.431,
      "num_input_tokens_seen": 16955360,
      "step": 25970
    },
    {
      "epoch": 15.315448113207546,
      "grad_norm": 4.190262317657471,
      "learning_rate": 1.5805365942241092e-06,
      "loss": 0.3109,
      "num_input_tokens_seen": 16959040,
      "step": 25975
    },
    {
      "epoch": 15.318396226415095,
      "grad_norm": 3.240661859512329,
      "learning_rate": 1.5786600394629181e-06,
      "loss": 0.4035,
      "num_input_tokens_seen": 16963456,
      "step": 25980
    },
    {
      "epoch": 15.321344339622641,
      "grad_norm": 3.06575345993042,
      "learning_rate": 1.5767843905162261e-06,
      "loss": 0.315,
      "num_input_tokens_seen": 16966528,
      "step": 25985
    },
    {
      "epoch": 15.32429245283019,
      "grad_norm": 3.030884027481079,
      "learning_rate": 1.5749096478806209e-06,
      "loss": 0.3156,
      "num_input_tokens_seen": 16969824,
      "step": 25990
    },
    {
      "epoch": 15.327240566037736,
      "grad_norm": 2.5016205310821533,
      "learning_rate": 1.5730358120524452e-06,
      "loss": 0.2571,
      "num_input_tokens_seen": 16973536,
      "step": 25995
    },
    {
      "epoch": 15.330188679245284,
      "grad_norm": 3.1649436950683594,
      "learning_rate": 1.5711628835278098e-06,
      "loss": 0.278,
      "num_input_tokens_seen": 16975808,
      "step": 26000
    },
    {
      "epoch": 15.33313679245283,
      "grad_norm": 2.832500696182251,
      "learning_rate": 1.5692908628025782e-06,
      "loss": 0.2894,
      "num_input_tokens_seen": 16979200,
      "step": 26005
    },
    {
      "epoch": 15.336084905660377,
      "grad_norm": 3.2382915019989014,
      "learning_rate": 1.5674197503723765e-06,
      "loss": 0.3037,
      "num_input_tokens_seen": 16982048,
      "step": 26010
    },
    {
      "epoch": 15.339033018867925,
      "grad_norm": 4.457133769989014,
      "learning_rate": 1.5655495467325893e-06,
      "loss": 0.2992,
      "num_input_tokens_seen": 16984608,
      "step": 26015
    },
    {
      "epoch": 15.341981132075471,
      "grad_norm": 4.746394157409668,
      "learning_rate": 1.5636802523783613e-06,
      "loss": 0.2037,
      "num_input_tokens_seen": 16987456,
      "step": 26020
    },
    {
      "epoch": 15.34492924528302,
      "grad_norm": 3.240077018737793,
      "learning_rate": 1.5618118678045947e-06,
      "loss": 0.3017,
      "num_input_tokens_seen": 16990048,
      "step": 26025
    },
    {
      "epoch": 15.347877358490566,
      "grad_norm": 3.805908441543579,
      "learning_rate": 1.5599443935059549e-06,
      "loss": 0.3415,
      "num_input_tokens_seen": 16993024,
      "step": 26030
    },
    {
      "epoch": 15.350825471698114,
      "grad_norm": 3.686028242111206,
      "learning_rate": 1.5580778299768635e-06,
      "loss": 0.2772,
      "num_input_tokens_seen": 16995552,
      "step": 26035
    },
    {
      "epoch": 15.35377358490566,
      "grad_norm": 4.051374912261963,
      "learning_rate": 1.5562121777114997e-06,
      "loss": 0.3273,
      "num_input_tokens_seen": 16998848,
      "step": 26040
    },
    {
      "epoch": 15.356721698113208,
      "grad_norm": 3.596458911895752,
      "learning_rate": 1.5543474372038043e-06,
      "loss": 0.3127,
      "num_input_tokens_seen": 17001792,
      "step": 26045
    },
    {
      "epoch": 15.359669811320755,
      "grad_norm": 4.27158784866333,
      "learning_rate": 1.5524836089474748e-06,
      "loss": 0.3338,
      "num_input_tokens_seen": 17004576,
      "step": 26050
    },
    {
      "epoch": 15.362617924528301,
      "grad_norm": 3.6558728218078613,
      "learning_rate": 1.5506206934359664e-06,
      "loss": 0.3955,
      "num_input_tokens_seen": 17008736,
      "step": 26055
    },
    {
      "epoch": 15.36556603773585,
      "grad_norm": 4.7534332275390625,
      "learning_rate": 1.5487586911624947e-06,
      "loss": 0.2797,
      "num_input_tokens_seen": 17012128,
      "step": 26060
    },
    {
      "epoch": 15.368514150943396,
      "grad_norm": 6.403425693511963,
      "learning_rate": 1.5468976026200355e-06,
      "loss": 0.3364,
      "num_input_tokens_seen": 17015584,
      "step": 26065
    },
    {
      "epoch": 15.371462264150944,
      "grad_norm": 4.322321891784668,
      "learning_rate": 1.5450374283013187e-06,
      "loss": 0.3649,
      "num_input_tokens_seen": 17019456,
      "step": 26070
    },
    {
      "epoch": 15.37441037735849,
      "grad_norm": 3.8460400104522705,
      "learning_rate": 1.5431781686988317e-06,
      "loss": 0.2437,
      "num_input_tokens_seen": 17022272,
      "step": 26075
    },
    {
      "epoch": 15.377358490566039,
      "grad_norm": 3.451488494873047,
      "learning_rate": 1.5413198243048233e-06,
      "loss": 0.3765,
      "num_input_tokens_seen": 17025856,
      "step": 26080
    },
    {
      "epoch": 15.380306603773585,
      "grad_norm": 3.2928130626678467,
      "learning_rate": 1.5394623956112974e-06,
      "loss": 0.3269,
      "num_input_tokens_seen": 17029024,
      "step": 26085
    },
    {
      "epoch": 15.383254716981131,
      "grad_norm": 3.556833505630493,
      "learning_rate": 1.537605883110015e-06,
      "loss": 0.4038,
      "num_input_tokens_seen": 17033056,
      "step": 26090
    },
    {
      "epoch": 15.38620283018868,
      "grad_norm": 4.7401814460754395,
      "learning_rate": 1.5357502872924984e-06,
      "loss": 0.361,
      "num_input_tokens_seen": 17035488,
      "step": 26095
    },
    {
      "epoch": 15.389150943396226,
      "grad_norm": 3.3929638862609863,
      "learning_rate": 1.5338956086500235e-06,
      "loss": 0.2865,
      "num_input_tokens_seen": 17039296,
      "step": 26100
    },
    {
      "epoch": 15.392099056603774,
      "grad_norm": 3.584487199783325,
      "learning_rate": 1.5320418476736237e-06,
      "loss": 0.3838,
      "num_input_tokens_seen": 17043104,
      "step": 26105
    },
    {
      "epoch": 15.39504716981132,
      "grad_norm": 4.780518054962158,
      "learning_rate": 1.5301890048540912e-06,
      "loss": 0.2337,
      "num_input_tokens_seen": 17046080,
      "step": 26110
    },
    {
      "epoch": 15.397995283018869,
      "grad_norm": 4.179786205291748,
      "learning_rate": 1.5283370806819743e-06,
      "loss": 0.3338,
      "num_input_tokens_seen": 17048768,
      "step": 26115
    },
    {
      "epoch": 15.400943396226415,
      "grad_norm": 3.493394374847412,
      "learning_rate": 1.5264860756475752e-06,
      "loss": 0.2568,
      "num_input_tokens_seen": 17051680,
      "step": 26120
    },
    {
      "epoch": 15.403891509433961,
      "grad_norm": 2.6219229698181152,
      "learning_rate": 1.5246359902409592e-06,
      "loss": 0.2933,
      "num_input_tokens_seen": 17054944,
      "step": 26125
    },
    {
      "epoch": 15.40683962264151,
      "grad_norm": 2.644037961959839,
      "learning_rate": 1.5227868249519423e-06,
      "loss": 0.3394,
      "num_input_tokens_seen": 17058048,
      "step": 26130
    },
    {
      "epoch": 15.409787735849056,
      "grad_norm": 2.4150314331054688,
      "learning_rate": 1.5209385802700999e-06,
      "loss": 0.2944,
      "num_input_tokens_seen": 17060960,
      "step": 26135
    },
    {
      "epoch": 15.412735849056604,
      "grad_norm": 2.8627138137817383,
      "learning_rate": 1.5190912566847626e-06,
      "loss": 0.2966,
      "num_input_tokens_seen": 17064992,
      "step": 26140
    },
    {
      "epoch": 15.41568396226415,
      "grad_norm": 5.699700832366943,
      "learning_rate": 1.5172448546850166e-06,
      "loss": 0.2344,
      "num_input_tokens_seen": 17067936,
      "step": 26145
    },
    {
      "epoch": 15.418632075471699,
      "grad_norm": 3.5310146808624268,
      "learning_rate": 1.515399374759704e-06,
      "loss": 0.2554,
      "num_input_tokens_seen": 17071168,
      "step": 26150
    },
    {
      "epoch": 15.421580188679245,
      "grad_norm": 5.196860313415527,
      "learning_rate": 1.513554817397424e-06,
      "loss": 0.282,
      "num_input_tokens_seen": 17074560,
      "step": 26155
    },
    {
      "epoch": 15.424528301886792,
      "grad_norm": 3.084752321243286,
      "learning_rate": 1.5117111830865338e-06,
      "loss": 0.3104,
      "num_input_tokens_seen": 17077312,
      "step": 26160
    },
    {
      "epoch": 15.42747641509434,
      "grad_norm": 3.71352481842041,
      "learning_rate": 1.509868472315142e-06,
      "loss": 0.2123,
      "num_input_tokens_seen": 17080608,
      "step": 26165
    },
    {
      "epoch": 15.430424528301886,
      "grad_norm": 2.066924810409546,
      "learning_rate": 1.508026685571113e-06,
      "loss": 0.2791,
      "num_input_tokens_seen": 17083072,
      "step": 26170
    },
    {
      "epoch": 15.433372641509434,
      "grad_norm": 3.460465669631958,
      "learning_rate": 1.506185823342069e-06,
      "loss": 0.2194,
      "num_input_tokens_seen": 17086208,
      "step": 26175
    },
    {
      "epoch": 15.43632075471698,
      "grad_norm": 2.629861831665039,
      "learning_rate": 1.504345886115386e-06,
      "loss": 0.3421,
      "num_input_tokens_seen": 17089312,
      "step": 26180
    },
    {
      "epoch": 15.439268867924529,
      "grad_norm": 3.4712040424346924,
      "learning_rate": 1.502506874378193e-06,
      "loss": 0.3095,
      "num_input_tokens_seen": 17092576,
      "step": 26185
    },
    {
      "epoch": 15.442216981132075,
      "grad_norm": 4.7518463134765625,
      "learning_rate": 1.5006687886173805e-06,
      "loss": 0.4259,
      "num_input_tokens_seen": 17096896,
      "step": 26190
    },
    {
      "epoch": 15.445165094339623,
      "grad_norm": 2.8150784969329834,
      "learning_rate": 1.498831629319587e-06,
      "loss": 0.2983,
      "num_input_tokens_seen": 17100320,
      "step": 26195
    },
    {
      "epoch": 15.44811320754717,
      "grad_norm": 3.150428533554077,
      "learning_rate": 1.4969953969712087e-06,
      "loss": 0.3975,
      "num_input_tokens_seen": 17103200,
      "step": 26200
    },
    {
      "epoch": 15.451061320754716,
      "grad_norm": 6.3504252433776855,
      "learning_rate": 1.4951600920583963e-06,
      "loss": 0.3394,
      "num_input_tokens_seen": 17106336,
      "step": 26205
    },
    {
      "epoch": 15.454009433962264,
      "grad_norm": 3.351583242416382,
      "learning_rate": 1.493325715067055e-06,
      "loss": 0.4025,
      "num_input_tokens_seen": 17109728,
      "step": 26210
    },
    {
      "epoch": 15.45695754716981,
      "grad_norm": 4.036856651306152,
      "learning_rate": 1.4914922664828417e-06,
      "loss": 0.3181,
      "num_input_tokens_seen": 17113152,
      "step": 26215
    },
    {
      "epoch": 15.459905660377359,
      "grad_norm": 4.360968589782715,
      "learning_rate": 1.4896597467911732e-06,
      "loss": 0.342,
      "num_input_tokens_seen": 17115904,
      "step": 26220
    },
    {
      "epoch": 15.462853773584905,
      "grad_norm": 2.939380645751953,
      "learning_rate": 1.4878281564772156e-06,
      "loss": 0.2785,
      "num_input_tokens_seen": 17118752,
      "step": 26225
    },
    {
      "epoch": 15.465801886792454,
      "grad_norm": 5.1072211265563965,
      "learning_rate": 1.4859974960258898e-06,
      "loss": 0.2652,
      "num_input_tokens_seen": 17121888,
      "step": 26230
    },
    {
      "epoch": 15.46875,
      "grad_norm": 3.8873965740203857,
      "learning_rate": 1.4841677659218723e-06,
      "loss": 0.349,
      "num_input_tokens_seen": 17124320,
      "step": 26235
    },
    {
      "epoch": 15.471698113207546,
      "grad_norm": 4.178337097167969,
      "learning_rate": 1.4823389666495886e-06,
      "loss": 0.3724,
      "num_input_tokens_seen": 17128480,
      "step": 26240
    },
    {
      "epoch": 15.474646226415095,
      "grad_norm": 4.621440887451172,
      "learning_rate": 1.4805110986932258e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 17132160,
      "step": 26245
    },
    {
      "epoch": 15.477594339622641,
      "grad_norm": 3.8071951866149902,
      "learning_rate": 1.4786841625367166e-06,
      "loss": 0.2759,
      "num_input_tokens_seen": 17135392,
      "step": 26250
    },
    {
      "epoch": 15.48054245283019,
      "grad_norm": 3.615647792816162,
      "learning_rate": 1.476858158663752e-06,
      "loss": 0.3144,
      "num_input_tokens_seen": 17139232,
      "step": 26255
    },
    {
      "epoch": 15.483490566037736,
      "grad_norm": 3.2508201599121094,
      "learning_rate": 1.4750330875577745e-06,
      "loss": 0.3538,
      "num_input_tokens_seen": 17142976,
      "step": 26260
    },
    {
      "epoch": 15.486438679245284,
      "grad_norm": 4.152102947235107,
      "learning_rate": 1.4732089497019787e-06,
      "loss": 0.2685,
      "num_input_tokens_seen": 17145824,
      "step": 26265
    },
    {
      "epoch": 15.48938679245283,
      "grad_norm": 2.2849650382995605,
      "learning_rate": 1.471385745579313e-06,
      "loss": 0.2811,
      "num_input_tokens_seen": 17148320,
      "step": 26270
    },
    {
      "epoch": 15.492334905660377,
      "grad_norm": 2.4169235229492188,
      "learning_rate": 1.4695634756724775e-06,
      "loss": 0.2491,
      "num_input_tokens_seen": 17151584,
      "step": 26275
    },
    {
      "epoch": 15.495283018867925,
      "grad_norm": 3.553770065307617,
      "learning_rate": 1.4677421404639281e-06,
      "loss": 0.3467,
      "num_input_tokens_seen": 17154496,
      "step": 26280
    },
    {
      "epoch": 15.498231132075471,
      "grad_norm": 4.313138484954834,
      "learning_rate": 1.4659217404358706e-06,
      "loss": 0.313,
      "num_input_tokens_seen": 17157376,
      "step": 26285
    },
    {
      "epoch": 15.50117924528302,
      "grad_norm": 3.95273494720459,
      "learning_rate": 1.4641022760702627e-06,
      "loss": 0.4681,
      "num_input_tokens_seen": 17161792,
      "step": 26290
    },
    {
      "epoch": 15.504127358490566,
      "grad_norm": 4.109541893005371,
      "learning_rate": 1.4622837478488172e-06,
      "loss": 0.2548,
      "num_input_tokens_seen": 17164992,
      "step": 26295
    },
    {
      "epoch": 15.507075471698114,
      "grad_norm": 3.1863913536071777,
      "learning_rate": 1.4604661562529953e-06,
      "loss": 0.422,
      "num_input_tokens_seen": 17168672,
      "step": 26300
    },
    {
      "epoch": 15.51002358490566,
      "grad_norm": 2.9035253524780273,
      "learning_rate": 1.4586495017640119e-06,
      "loss": 0.3129,
      "num_input_tokens_seen": 17171232,
      "step": 26305
    },
    {
      "epoch": 15.512971698113208,
      "grad_norm": 5.44696044921875,
      "learning_rate": 1.4568337848628366e-06,
      "loss": 0.3671,
      "num_input_tokens_seen": 17174080,
      "step": 26310
    },
    {
      "epoch": 15.515919811320755,
      "grad_norm": 4.756264686584473,
      "learning_rate": 1.4550190060301872e-06,
      "loss": 0.271,
      "num_input_tokens_seen": 17177728,
      "step": 26315
    },
    {
      "epoch": 15.518867924528301,
      "grad_norm": 3.962956190109253,
      "learning_rate": 1.4532051657465335e-06,
      "loss": 0.2997,
      "num_input_tokens_seen": 17180768,
      "step": 26320
    },
    {
      "epoch": 15.52181603773585,
      "grad_norm": 4.2770795822143555,
      "learning_rate": 1.4513922644920985e-06,
      "loss": 0.3333,
      "num_input_tokens_seen": 17183200,
      "step": 26325
    },
    {
      "epoch": 15.524764150943396,
      "grad_norm": 4.2647809982299805,
      "learning_rate": 1.4495803027468552e-06,
      "loss": 0.2526,
      "num_input_tokens_seen": 17185888,
      "step": 26330
    },
    {
      "epoch": 15.527712264150944,
      "grad_norm": 3.7605979442596436,
      "learning_rate": 1.4477692809905263e-06,
      "loss": 0.2904,
      "num_input_tokens_seen": 17188960,
      "step": 26335
    },
    {
      "epoch": 15.53066037735849,
      "grad_norm": 3.663928508758545,
      "learning_rate": 1.4459591997025896e-06,
      "loss": 0.3754,
      "num_input_tokens_seen": 17193024,
      "step": 26340
    },
    {
      "epoch": 15.533608490566039,
      "grad_norm": 4.425746917724609,
      "learning_rate": 1.4441500593622737e-06,
      "loss": 0.3195,
      "num_input_tokens_seen": 17196672,
      "step": 26345
    },
    {
      "epoch": 15.536556603773585,
      "grad_norm": 5.2296857833862305,
      "learning_rate": 1.4423418604485539e-06,
      "loss": 0.3815,
      "num_input_tokens_seen": 17200192,
      "step": 26350
    },
    {
      "epoch": 15.539504716981131,
      "grad_norm": 3.1150588989257812,
      "learning_rate": 1.4405346034401597e-06,
      "loss": 0.3798,
      "num_input_tokens_seen": 17203616,
      "step": 26355
    },
    {
      "epoch": 15.54245283018868,
      "grad_norm": 3.5838539600372314,
      "learning_rate": 1.4387282888155695e-06,
      "loss": 0.2106,
      "num_input_tokens_seen": 17205920,
      "step": 26360
    },
    {
      "epoch": 15.545400943396226,
      "grad_norm": 3.058284282684326,
      "learning_rate": 1.436922917053013e-06,
      "loss": 0.3071,
      "num_input_tokens_seen": 17209152,
      "step": 26365
    },
    {
      "epoch": 15.548349056603774,
      "grad_norm": 4.9887542724609375,
      "learning_rate": 1.4351184886304686e-06,
      "loss": 0.3295,
      "num_input_tokens_seen": 17213056,
      "step": 26370
    },
    {
      "epoch": 15.55129716981132,
      "grad_norm": 4.2343926429748535,
      "learning_rate": 1.4333150040256699e-06,
      "loss": 0.346,
      "num_input_tokens_seen": 17216032,
      "step": 26375
    },
    {
      "epoch": 15.554245283018869,
      "grad_norm": 2.267852783203125,
      "learning_rate": 1.4315124637160954e-06,
      "loss": 0.2676,
      "num_input_tokens_seen": 17220640,
      "step": 26380
    },
    {
      "epoch": 15.557193396226415,
      "grad_norm": 3.4346415996551514,
      "learning_rate": 1.4297108681789752e-06,
      "loss": 0.3225,
      "num_input_tokens_seen": 17223520,
      "step": 26385
    },
    {
      "epoch": 15.560141509433961,
      "grad_norm": 6.319263458251953,
      "learning_rate": 1.4279102178912902e-06,
      "loss": 0.32,
      "num_input_tokens_seen": 17226112,
      "step": 26390
    },
    {
      "epoch": 15.56308962264151,
      "grad_norm": 2.5334126949310303,
      "learning_rate": 1.4261105133297693e-06,
      "loss": 0.4329,
      "num_input_tokens_seen": 17229120,
      "step": 26395
    },
    {
      "epoch": 15.566037735849056,
      "grad_norm": 2.146577835083008,
      "learning_rate": 1.4243117549708913e-06,
      "loss": 0.3536,
      "num_input_tokens_seen": 17232096,
      "step": 26400
    },
    {
      "epoch": 15.568985849056604,
      "grad_norm": 3.9969778060913086,
      "learning_rate": 1.422513943290888e-06,
      "loss": 0.2398,
      "num_input_tokens_seen": 17235744,
      "step": 26405
    },
    {
      "epoch": 15.57193396226415,
      "grad_norm": 3.6518540382385254,
      "learning_rate": 1.4207170787657365e-06,
      "loss": 0.3277,
      "num_input_tokens_seen": 17238656,
      "step": 26410
    },
    {
      "epoch": 15.574882075471699,
      "grad_norm": 3.3526155948638916,
      "learning_rate": 1.4189211618711646e-06,
      "loss": 0.3856,
      "num_input_tokens_seen": 17241952,
      "step": 26415
    },
    {
      "epoch": 15.577830188679245,
      "grad_norm": 3.049076557159424,
      "learning_rate": 1.417126193082648e-06,
      "loss": 0.2421,
      "num_input_tokens_seen": 17245152,
      "step": 26420
    },
    {
      "epoch": 15.580778301886792,
      "grad_norm": 3.4857053756713867,
      "learning_rate": 1.4153321728754133e-06,
      "loss": 0.2122,
      "num_input_tokens_seen": 17248416,
      "step": 26425
    },
    {
      "epoch": 15.58372641509434,
      "grad_norm": 4.722821235656738,
      "learning_rate": 1.4135391017244338e-06,
      "loss": 0.3111,
      "num_input_tokens_seen": 17251648,
      "step": 26430
    },
    {
      "epoch": 15.586674528301886,
      "grad_norm": 3.9381651878356934,
      "learning_rate": 1.4117469801044332e-06,
      "loss": 0.3405,
      "num_input_tokens_seen": 17255744,
      "step": 26435
    },
    {
      "epoch": 15.589622641509434,
      "grad_norm": 4.540798664093018,
      "learning_rate": 1.4099558084898862e-06,
      "loss": 0.3773,
      "num_input_tokens_seen": 17259040,
      "step": 26440
    },
    {
      "epoch": 15.59257075471698,
      "grad_norm": 2.6230900287628174,
      "learning_rate": 1.408165587355011e-06,
      "loss": 0.2728,
      "num_input_tokens_seen": 17262784,
      "step": 26445
    },
    {
      "epoch": 15.595518867924529,
      "grad_norm": 2.424760341644287,
      "learning_rate": 1.4063763171737766e-06,
      "loss": 0.3395,
      "num_input_tokens_seen": 17265504,
      "step": 26450
    },
    {
      "epoch": 15.598466981132075,
      "grad_norm": 2.973039388656616,
      "learning_rate": 1.4045879984198996e-06,
      "loss": 0.2791,
      "num_input_tokens_seen": 17269440,
      "step": 26455
    },
    {
      "epoch": 15.601415094339622,
      "grad_norm": 1.8335871696472168,
      "learning_rate": 1.4028006315668457e-06,
      "loss": 0.3203,
      "num_input_tokens_seen": 17272480,
      "step": 26460
    },
    {
      "epoch": 15.60436320754717,
      "grad_norm": 10.609156608581543,
      "learning_rate": 1.4010142170878261e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 17275808,
      "step": 26465
    },
    {
      "epoch": 15.607311320754716,
      "grad_norm": 3.8790392875671387,
      "learning_rate": 1.3992287554558042e-06,
      "loss": 0.3875,
      "num_input_tokens_seen": 17279424,
      "step": 26470
    },
    {
      "epoch": 15.610259433962264,
      "grad_norm": 2.501727342605591,
      "learning_rate": 1.3974442471434885e-06,
      "loss": 0.268,
      "num_input_tokens_seen": 17282080,
      "step": 26475
    },
    {
      "epoch": 15.61320754716981,
      "grad_norm": 5.29008150100708,
      "learning_rate": 1.395660692623334e-06,
      "loss": 0.3232,
      "num_input_tokens_seen": 17284608,
      "step": 26480
    },
    {
      "epoch": 15.616155660377359,
      "grad_norm": 3.5645017623901367,
      "learning_rate": 1.3938780923675454e-06,
      "loss": 0.312,
      "num_input_tokens_seen": 17287328,
      "step": 26485
    },
    {
      "epoch": 15.619103773584905,
      "grad_norm": 5.405980587005615,
      "learning_rate": 1.3920964468480718e-06,
      "loss": 0.3907,
      "num_input_tokens_seen": 17289856,
      "step": 26490
    },
    {
      "epoch": 15.622051886792454,
      "grad_norm": 4.0511884689331055,
      "learning_rate": 1.3903157565366143e-06,
      "loss": 0.2877,
      "num_input_tokens_seen": 17292736,
      "step": 26495
    },
    {
      "epoch": 15.625,
      "grad_norm": 3.301304340362549,
      "learning_rate": 1.3885360219046172e-06,
      "loss": 0.3608,
      "num_input_tokens_seen": 17295968,
      "step": 26500
    },
    {
      "epoch": 15.627948113207546,
      "grad_norm": 3.1807429790496826,
      "learning_rate": 1.386757243423273e-06,
      "loss": 0.3303,
      "num_input_tokens_seen": 17298144,
      "step": 26505
    },
    {
      "epoch": 15.630896226415095,
      "grad_norm": 3.662149429321289,
      "learning_rate": 1.384979421563521e-06,
      "loss": 0.3568,
      "num_input_tokens_seen": 17301184,
      "step": 26510
    },
    {
      "epoch": 15.633844339622641,
      "grad_norm": 5.271115303039551,
      "learning_rate": 1.3832025567960465e-06,
      "loss": 0.4353,
      "num_input_tokens_seen": 17303808,
      "step": 26515
    },
    {
      "epoch": 15.63679245283019,
      "grad_norm": 2.084275245666504,
      "learning_rate": 1.3814266495912815e-06,
      "loss": 0.3105,
      "num_input_tokens_seen": 17306784,
      "step": 26520
    },
    {
      "epoch": 15.639740566037736,
      "grad_norm": 3.6416585445404053,
      "learning_rate": 1.3796517004194078e-06,
      "loss": 0.3354,
      "num_input_tokens_seen": 17310432,
      "step": 26525
    },
    {
      "epoch": 15.642688679245284,
      "grad_norm": 4.203418731689453,
      "learning_rate": 1.3778777097503476e-06,
      "loss": 0.3413,
      "num_input_tokens_seen": 17313152,
      "step": 26530
    },
    {
      "epoch": 15.64563679245283,
      "grad_norm": 2.231419086456299,
      "learning_rate": 1.3761046780537757e-06,
      "loss": 0.2968,
      "num_input_tokens_seen": 17315680,
      "step": 26535
    },
    {
      "epoch": 15.648584905660378,
      "grad_norm": 2.2358946800231934,
      "learning_rate": 1.3743326057991086e-06,
      "loss": 0.1888,
      "num_input_tokens_seen": 17318880,
      "step": 26540
    },
    {
      "epoch": 15.651533018867925,
      "grad_norm": 3.609388828277588,
      "learning_rate": 1.3725614934555093e-06,
      "loss": 0.3633,
      "num_input_tokens_seen": 17321728,
      "step": 26545
    },
    {
      "epoch": 15.654481132075471,
      "grad_norm": 3.898571252822876,
      "learning_rate": 1.3707913414918882e-06,
      "loss": 0.2854,
      "num_input_tokens_seen": 17324896,
      "step": 26550
    },
    {
      "epoch": 15.65742924528302,
      "grad_norm": 3.0724971294403076,
      "learning_rate": 1.3690221503768996e-06,
      "loss": 0.2834,
      "num_input_tokens_seen": 17328448,
      "step": 26555
    },
    {
      "epoch": 15.660377358490566,
      "grad_norm": 3.8443989753723145,
      "learning_rate": 1.3672539205789465e-06,
      "loss": 0.4084,
      "num_input_tokens_seen": 17332224,
      "step": 26560
    },
    {
      "epoch": 15.663325471698114,
      "grad_norm": 4.084451198577881,
      "learning_rate": 1.3654866525661737e-06,
      "loss": 0.3564,
      "num_input_tokens_seen": 17334624,
      "step": 26565
    },
    {
      "epoch": 15.66627358490566,
      "grad_norm": 5.078795433044434,
      "learning_rate": 1.3637203468064741e-06,
      "loss": 0.4208,
      "num_input_tokens_seen": 17339936,
      "step": 26570
    },
    {
      "epoch": 15.669221698113208,
      "grad_norm": 7.502529621124268,
      "learning_rate": 1.3619550037674838e-06,
      "loss": 0.3612,
      "num_input_tokens_seen": 17342720,
      "step": 26575
    },
    {
      "epoch": 15.672169811320755,
      "grad_norm": 5.799961090087891,
      "learning_rate": 1.3601906239165857e-06,
      "loss": 0.3434,
      "num_input_tokens_seen": 17346464,
      "step": 26580
    },
    {
      "epoch": 15.675117924528301,
      "grad_norm": 3.802048444747925,
      "learning_rate": 1.3584272077209048e-06,
      "loss": 0.4204,
      "num_input_tokens_seen": 17349600,
      "step": 26585
    },
    {
      "epoch": 15.67806603773585,
      "grad_norm": 3.458008050918579,
      "learning_rate": 1.3566647556473168e-06,
      "loss": 0.2809,
      "num_input_tokens_seen": 17352512,
      "step": 26590
    },
    {
      "epoch": 15.681014150943396,
      "grad_norm": 3.2562687397003174,
      "learning_rate": 1.3549032681624363e-06,
      "loss": 0.3846,
      "num_input_tokens_seen": 17356896,
      "step": 26595
    },
    {
      "epoch": 15.683962264150944,
      "grad_norm": 3.19452166557312,
      "learning_rate": 1.3531427457326252e-06,
      "loss": 0.2189,
      "num_input_tokens_seen": 17360160,
      "step": 26600
    },
    {
      "epoch": 15.68691037735849,
      "grad_norm": 3.9227116107940674,
      "learning_rate": 1.3513831888239893e-06,
      "loss": 0.3953,
      "num_input_tokens_seen": 17363488,
      "step": 26605
    },
    {
      "epoch": 15.689858490566039,
      "grad_norm": 3.5639758110046387,
      "learning_rate": 1.3496245979023786e-06,
      "loss": 0.2436,
      "num_input_tokens_seen": 17367296,
      "step": 26610
    },
    {
      "epoch": 15.692806603773585,
      "grad_norm": 3.6830461025238037,
      "learning_rate": 1.3478669734333865e-06,
      "loss": 0.3612,
      "num_input_tokens_seen": 17371008,
      "step": 26615
    },
    {
      "epoch": 15.695754716981131,
      "grad_norm": 3.814260721206665,
      "learning_rate": 1.3461103158823546e-06,
      "loss": 0.3699,
      "num_input_tokens_seen": 17374304,
      "step": 26620
    },
    {
      "epoch": 15.69870283018868,
      "grad_norm": 4.493785381317139,
      "learning_rate": 1.3443546257143624e-06,
      "loss": 0.2904,
      "num_input_tokens_seen": 17377344,
      "step": 26625
    },
    {
      "epoch": 15.701650943396226,
      "grad_norm": 3.8498027324676514,
      "learning_rate": 1.3425999033942395e-06,
      "loss": 0.3278,
      "num_input_tokens_seen": 17380032,
      "step": 26630
    },
    {
      "epoch": 15.704599056603774,
      "grad_norm": 2.684558153152466,
      "learning_rate": 1.3408461493865549e-06,
      "loss": 0.3639,
      "num_input_tokens_seen": 17382336,
      "step": 26635
    },
    {
      "epoch": 15.70754716981132,
      "grad_norm": 4.0287394523620605,
      "learning_rate": 1.339093364155622e-06,
      "loss": 0.4187,
      "num_input_tokens_seen": 17385664,
      "step": 26640
    },
    {
      "epoch": 15.710495283018869,
      "grad_norm": 3.1829121112823486,
      "learning_rate": 1.3373415481654988e-06,
      "loss": 0.4071,
      "num_input_tokens_seen": 17389056,
      "step": 26645
    },
    {
      "epoch": 15.713443396226415,
      "grad_norm": 3.048290729522705,
      "learning_rate": 1.335590701879984e-06,
      "loss": 0.3444,
      "num_input_tokens_seen": 17391936,
      "step": 26650
    },
    {
      "epoch": 15.716391509433961,
      "grad_norm": 1.9633924961090088,
      "learning_rate": 1.3338408257626257e-06,
      "loss": 0.2407,
      "num_input_tokens_seen": 17396832,
      "step": 26655
    },
    {
      "epoch": 15.71933962264151,
      "grad_norm": 8.257603645324707,
      "learning_rate": 1.3320919202767086e-06,
      "loss": 0.2513,
      "num_input_tokens_seen": 17399232,
      "step": 26660
    },
    {
      "epoch": 15.722287735849056,
      "grad_norm": 9.309518814086914,
      "learning_rate": 1.3303439858852636e-06,
      "loss": 0.3838,
      "num_input_tokens_seen": 17402624,
      "step": 26665
    },
    {
      "epoch": 15.725235849056604,
      "grad_norm": 3.427581548690796,
      "learning_rate": 1.3285970230510636e-06,
      "loss": 0.3655,
      "num_input_tokens_seen": 17406016,
      "step": 26670
    },
    {
      "epoch": 15.72818396226415,
      "grad_norm": 2.9810760021209717,
      "learning_rate": 1.3268510322366246e-06,
      "loss": 0.3531,
      "num_input_tokens_seen": 17408928,
      "step": 26675
    },
    {
      "epoch": 15.731132075471699,
      "grad_norm": 3.1264753341674805,
      "learning_rate": 1.3251060139042038e-06,
      "loss": 0.2431,
      "num_input_tokens_seen": 17412096,
      "step": 26680
    },
    {
      "epoch": 15.734080188679245,
      "grad_norm": 4.51759147644043,
      "learning_rate": 1.3233619685158056e-06,
      "loss": 0.3417,
      "num_input_tokens_seen": 17414816,
      "step": 26685
    },
    {
      "epoch": 15.737028301886792,
      "grad_norm": 7.457120418548584,
      "learning_rate": 1.3216188965331712e-06,
      "loss": 0.3285,
      "num_input_tokens_seen": 17417376,
      "step": 26690
    },
    {
      "epoch": 15.73997641509434,
      "grad_norm": 4.285046100616455,
      "learning_rate": 1.3198767984177869e-06,
      "loss": 0.3344,
      "num_input_tokens_seen": 17420384,
      "step": 26695
    },
    {
      "epoch": 15.742924528301886,
      "grad_norm": 3.7005157470703125,
      "learning_rate": 1.3181356746308805e-06,
      "loss": 0.3501,
      "num_input_tokens_seen": 17423840,
      "step": 26700
    },
    {
      "epoch": 15.745872641509434,
      "grad_norm": 3.1018455028533936,
      "learning_rate": 1.3163955256334226e-06,
      "loss": 0.2717,
      "num_input_tokens_seen": 17426784,
      "step": 26705
    },
    {
      "epoch": 15.74882075471698,
      "grad_norm": 4.256278038024902,
      "learning_rate": 1.3146563518861227e-06,
      "loss": 0.3016,
      "num_input_tokens_seen": 17430080,
      "step": 26710
    },
    {
      "epoch": 15.751768867924529,
      "grad_norm": 2.8724923133850098,
      "learning_rate": 1.3129181538494384e-06,
      "loss": 0.2567,
      "num_input_tokens_seen": 17432768,
      "step": 26715
    },
    {
      "epoch": 15.754716981132075,
      "grad_norm": 2.3154377937316895,
      "learning_rate": 1.3111809319835622e-06,
      "loss": 0.3205,
      "num_input_tokens_seen": 17437504,
      "step": 26720
    },
    {
      "epoch": 15.757665094339622,
      "grad_norm": 3.688124895095825,
      "learning_rate": 1.3094446867484335e-06,
      "loss": 0.3815,
      "num_input_tokens_seen": 17440448,
      "step": 26725
    },
    {
      "epoch": 15.76061320754717,
      "grad_norm": 3.2079412937164307,
      "learning_rate": 1.3077094186037287e-06,
      "loss": 0.4823,
      "num_input_tokens_seen": 17443392,
      "step": 26730
    },
    {
      "epoch": 15.763561320754716,
      "grad_norm": 4.019981384277344,
      "learning_rate": 1.305975128008869e-06,
      "loss": 0.4009,
      "num_input_tokens_seen": 17445824,
      "step": 26735
    },
    {
      "epoch": 15.766509433962264,
      "grad_norm": 6.259704113006592,
      "learning_rate": 1.304241815423014e-06,
      "loss": 0.3692,
      "num_input_tokens_seen": 17448800,
      "step": 26740
    },
    {
      "epoch": 15.76945754716981,
      "grad_norm": 5.5584893226623535,
      "learning_rate": 1.3025094813050655e-06,
      "loss": 0.3118,
      "num_input_tokens_seen": 17451872,
      "step": 26745
    },
    {
      "epoch": 15.772405660377359,
      "grad_norm": 2.5257208347320557,
      "learning_rate": 1.3007781261136675e-06,
      "loss": 0.35,
      "num_input_tokens_seen": 17455584,
      "step": 26750
    },
    {
      "epoch": 15.775353773584905,
      "grad_norm": 3.5317041873931885,
      "learning_rate": 1.299047750307204e-06,
      "loss": 0.3432,
      "num_input_tokens_seen": 17459648,
      "step": 26755
    },
    {
      "epoch": 15.778301886792454,
      "grad_norm": 3.3328137397766113,
      "learning_rate": 1.297318354343799e-06,
      "loss": 0.2918,
      "num_input_tokens_seen": 17462528,
      "step": 26760
    },
    {
      "epoch": 15.78125,
      "grad_norm": 4.417114734649658,
      "learning_rate": 1.295589938681317e-06,
      "loss": 0.3034,
      "num_input_tokens_seen": 17466592,
      "step": 26765
    },
    {
      "epoch": 15.784198113207546,
      "grad_norm": 3.058224678039551,
      "learning_rate": 1.2938625037773628e-06,
      "loss": 0.3418,
      "num_input_tokens_seen": 17470240,
      "step": 26770
    },
    {
      "epoch": 15.787146226415095,
      "grad_norm": 3.829861640930176,
      "learning_rate": 1.2921360500892843e-06,
      "loss": 0.2628,
      "num_input_tokens_seen": 17474080,
      "step": 26775
    },
    {
      "epoch": 15.790094339622641,
      "grad_norm": 3.8211669921875,
      "learning_rate": 1.290410578074167e-06,
      "loss": 0.3553,
      "num_input_tokens_seen": 17476928,
      "step": 26780
    },
    {
      "epoch": 15.79304245283019,
      "grad_norm": 4.273032188415527,
      "learning_rate": 1.2886860881888362e-06,
      "loss": 0.3007,
      "num_input_tokens_seen": 17480256,
      "step": 26785
    },
    {
      "epoch": 15.795990566037736,
      "grad_norm": 4.894825458526611,
      "learning_rate": 1.2869625808898584e-06,
      "loss": 0.2926,
      "num_input_tokens_seen": 17483168,
      "step": 26790
    },
    {
      "epoch": 15.798938679245284,
      "grad_norm": 12.426753997802734,
      "learning_rate": 1.2852400566335398e-06,
      "loss": 0.5848,
      "num_input_tokens_seen": 17487360,
      "step": 26795
    },
    {
      "epoch": 15.80188679245283,
      "grad_norm": 4.626605033874512,
      "learning_rate": 1.2835185158759244e-06,
      "loss": 0.3643,
      "num_input_tokens_seen": 17491648,
      "step": 26800
    },
    {
      "epoch": 15.804834905660378,
      "grad_norm": 3.019444227218628,
      "learning_rate": 1.2817979590728009e-06,
      "loss": 0.3023,
      "num_input_tokens_seen": 17494464,
      "step": 26805
    },
    {
      "epoch": 15.807783018867925,
      "grad_norm": 4.013226509094238,
      "learning_rate": 1.2800783866796918e-06,
      "loss": 0.3423,
      "num_input_tokens_seen": 17497280,
      "step": 26810
    },
    {
      "epoch": 15.810731132075471,
      "grad_norm": 5.741434097290039,
      "learning_rate": 1.2783597991518604e-06,
      "loss": 0.363,
      "num_input_tokens_seen": 17500704,
      "step": 26815
    },
    {
      "epoch": 15.81367924528302,
      "grad_norm": 5.237295150756836,
      "learning_rate": 1.2766421969443131e-06,
      "loss": 0.3457,
      "num_input_tokens_seen": 17504192,
      "step": 26820
    },
    {
      "epoch": 15.816627358490566,
      "grad_norm": 3.333575963973999,
      "learning_rate": 1.274925580511791e-06,
      "loss": 0.2919,
      "num_input_tokens_seen": 17507808,
      "step": 26825
    },
    {
      "epoch": 15.819575471698114,
      "grad_norm": 3.4029648303985596,
      "learning_rate": 1.2732099503087757e-06,
      "loss": 0.1826,
      "num_input_tokens_seen": 17511488,
      "step": 26830
    },
    {
      "epoch": 15.82252358490566,
      "grad_norm": 4.97998571395874,
      "learning_rate": 1.2714953067894859e-06,
      "loss": 0.2992,
      "num_input_tokens_seen": 17514240,
      "step": 26835
    },
    {
      "epoch": 15.825471698113208,
      "grad_norm": 4.448131561279297,
      "learning_rate": 1.2697816504078847e-06,
      "loss": 0.2914,
      "num_input_tokens_seen": 17517024,
      "step": 26840
    },
    {
      "epoch": 15.828419811320755,
      "grad_norm": 3.664102077484131,
      "learning_rate": 1.2680689816176672e-06,
      "loss": 0.4058,
      "num_input_tokens_seen": 17519616,
      "step": 26845
    },
    {
      "epoch": 15.831367924528301,
      "grad_norm": 4.474094390869141,
      "learning_rate": 1.2663573008722707e-06,
      "loss": 0.3124,
      "num_input_tokens_seen": 17523072,
      "step": 26850
    },
    {
      "epoch": 15.83431603773585,
      "grad_norm": 6.806681156158447,
      "learning_rate": 1.2646466086248698e-06,
      "loss": 0.2789,
      "num_input_tokens_seen": 17525568,
      "step": 26855
    },
    {
      "epoch": 15.837264150943396,
      "grad_norm": 4.433311462402344,
      "learning_rate": 1.2629369053283779e-06,
      "loss": 0.3958,
      "num_input_tokens_seen": 17528480,
      "step": 26860
    },
    {
      "epoch": 15.840212264150944,
      "grad_norm": 3.9505839347839355,
      "learning_rate": 1.2612281914354452e-06,
      "loss": 0.2362,
      "num_input_tokens_seen": 17531168,
      "step": 26865
    },
    {
      "epoch": 15.84316037735849,
      "grad_norm": 3.1731743812561035,
      "learning_rate": 1.259520467398463e-06,
      "loss": 0.3053,
      "num_input_tokens_seen": 17534560,
      "step": 26870
    },
    {
      "epoch": 15.846108490566039,
      "grad_norm": 2.6175055503845215,
      "learning_rate": 1.2578137336695573e-06,
      "loss": 0.2452,
      "num_input_tokens_seen": 17537280,
      "step": 26875
    },
    {
      "epoch": 15.849056603773585,
      "grad_norm": 6.345149517059326,
      "learning_rate": 1.256107990700594e-06,
      "loss": 0.3069,
      "num_input_tokens_seen": 17540320,
      "step": 26880
    },
    {
      "epoch": 15.852004716981131,
      "grad_norm": 5.569468975067139,
      "learning_rate": 1.2544032389431753e-06,
      "loss": 0.3686,
      "num_input_tokens_seen": 17543264,
      "step": 26885
    },
    {
      "epoch": 15.85495283018868,
      "grad_norm": 3.3612802028656006,
      "learning_rate": 1.2526994788486418e-06,
      "loss": 0.245,
      "num_input_tokens_seen": 17546176,
      "step": 26890
    },
    {
      "epoch": 15.857900943396226,
      "grad_norm": 2.4688563346862793,
      "learning_rate": 1.2509967108680697e-06,
      "loss": 0.3591,
      "num_input_tokens_seen": 17549344,
      "step": 26895
    },
    {
      "epoch": 15.860849056603774,
      "grad_norm": 4.200671195983887,
      "learning_rate": 1.249294935452277e-06,
      "loss": 0.3506,
      "num_input_tokens_seen": 17553984,
      "step": 26900
    },
    {
      "epoch": 15.86379716981132,
      "grad_norm": 0.5015525817871094,
      "learning_rate": 1.247594153051815e-06,
      "loss": 0.1985,
      "num_input_tokens_seen": 17560768,
      "step": 26905
    },
    {
      "epoch": 15.866745283018869,
      "grad_norm": 5.61729621887207,
      "learning_rate": 1.2458943641169718e-06,
      "loss": 0.3016,
      "num_input_tokens_seen": 17563744,
      "step": 26910
    },
    {
      "epoch": 15.869693396226415,
      "grad_norm": 4.284752368927002,
      "learning_rate": 1.2441955690977758e-06,
      "loss": 0.2823,
      "num_input_tokens_seen": 17567168,
      "step": 26915
    },
    {
      "epoch": 15.872641509433961,
      "grad_norm": 5.747034072875977,
      "learning_rate": 1.2424977684439898e-06,
      "loss": 0.3532,
      "num_input_tokens_seen": 17570528,
      "step": 26920
    },
    {
      "epoch": 15.87558962264151,
      "grad_norm": 2.929811954498291,
      "learning_rate": 1.2408009626051137e-06,
      "loss": 0.3773,
      "num_input_tokens_seen": 17573440,
      "step": 26925
    },
    {
      "epoch": 15.878537735849056,
      "grad_norm": 4.433734893798828,
      "learning_rate": 1.2391051520303826e-06,
      "loss": 0.2489,
      "num_input_tokens_seen": 17576736,
      "step": 26930
    },
    {
      "epoch": 15.881485849056604,
      "grad_norm": 3.778148889541626,
      "learning_rate": 1.2374103371687723e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 17579936,
      "step": 26935
    },
    {
      "epoch": 15.88443396226415,
      "grad_norm": 6.346294403076172,
      "learning_rate": 1.2357165184689906e-06,
      "loss": 0.2868,
      "num_input_tokens_seen": 17583200,
      "step": 26940
    },
    {
      "epoch": 15.887382075471699,
      "grad_norm": 4.476423263549805,
      "learning_rate": 1.2340236963794845e-06,
      "loss": 0.2296,
      "num_input_tokens_seen": 17586144,
      "step": 26945
    },
    {
      "epoch": 15.890330188679245,
      "grad_norm": 6.744287967681885,
      "learning_rate": 1.232331871348435e-06,
      "loss": 0.3663,
      "num_input_tokens_seen": 17588992,
      "step": 26950
    },
    {
      "epoch": 15.893278301886792,
      "grad_norm": 3.5366709232330322,
      "learning_rate": 1.2306410438237603e-06,
      "loss": 0.388,
      "num_input_tokens_seen": 17591744,
      "step": 26955
    },
    {
      "epoch": 15.89622641509434,
      "grad_norm": 2.818878173828125,
      "learning_rate": 1.228951214253113e-06,
      "loss": 0.3444,
      "num_input_tokens_seen": 17595616,
      "step": 26960
    },
    {
      "epoch": 15.899174528301886,
      "grad_norm": 6.603791236877441,
      "learning_rate": 1.2272623830838854e-06,
      "loss": 0.3375,
      "num_input_tokens_seen": 17598880,
      "step": 26965
    },
    {
      "epoch": 15.902122641509434,
      "grad_norm": 3.752480983734131,
      "learning_rate": 1.2255745507632016e-06,
      "loss": 0.2687,
      "num_input_tokens_seen": 17602240,
      "step": 26970
    },
    {
      "epoch": 15.90507075471698,
      "grad_norm": 3.635101795196533,
      "learning_rate": 1.223887717737922e-06,
      "loss": 0.2864,
      "num_input_tokens_seen": 17605824,
      "step": 26975
    },
    {
      "epoch": 15.908018867924529,
      "grad_norm": 3.5107717514038086,
      "learning_rate": 1.2222018844546434e-06,
      "loss": 0.245,
      "num_input_tokens_seen": 17609248,
      "step": 26980
    },
    {
      "epoch": 15.910966981132075,
      "grad_norm": 2.665670156478882,
      "learning_rate": 1.2205170513596975e-06,
      "loss": 0.3283,
      "num_input_tokens_seen": 17612640,
      "step": 26985
    },
    {
      "epoch": 15.913915094339622,
      "grad_norm": 4.795198917388916,
      "learning_rate": 1.2188332188991493e-06,
      "loss": 0.2401,
      "num_input_tokens_seen": 17615680,
      "step": 26990
    },
    {
      "epoch": 15.91686320754717,
      "grad_norm": 5.135300636291504,
      "learning_rate": 1.217150387518804e-06,
      "loss": 0.3587,
      "num_input_tokens_seen": 17618656,
      "step": 26995
    },
    {
      "epoch": 15.919811320754716,
      "grad_norm": 7.929037570953369,
      "learning_rate": 1.2154685576641967e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 17621344,
      "step": 27000
    },
    {
      "epoch": 15.922759433962264,
      "grad_norm": 3.451627731323242,
      "learning_rate": 1.2137877297805972e-06,
      "loss": 0.2664,
      "num_input_tokens_seen": 17625088,
      "step": 27005
    },
    {
      "epoch": 15.92570754716981,
      "grad_norm": 4.957344055175781,
      "learning_rate": 1.2121079043130162e-06,
      "loss": 0.3389,
      "num_input_tokens_seen": 17632320,
      "step": 27010
    },
    {
      "epoch": 15.928655660377359,
      "grad_norm": 4.476447582244873,
      "learning_rate": 1.210429081706192e-06,
      "loss": 0.3491,
      "num_input_tokens_seen": 17635072,
      "step": 27015
    },
    {
      "epoch": 15.931603773584905,
      "grad_norm": 4.012531280517578,
      "learning_rate": 1.2087512624046005e-06,
      "loss": 0.4026,
      "num_input_tokens_seen": 17638304,
      "step": 27020
    },
    {
      "epoch": 15.934551886792454,
      "grad_norm": 3.8610475063323975,
      "learning_rate": 1.2070744468524503e-06,
      "loss": 0.3374,
      "num_input_tokens_seen": 17641216,
      "step": 27025
    },
    {
      "epoch": 15.9375,
      "grad_norm": 4.519809246063232,
      "learning_rate": 1.2053986354936887e-06,
      "loss": 0.2618,
      "num_input_tokens_seen": 17644576,
      "step": 27030
    },
    {
      "epoch": 15.940448113207546,
      "grad_norm": 5.832067966461182,
      "learning_rate": 1.2037238287719916e-06,
      "loss": 0.283,
      "num_input_tokens_seen": 17647648,
      "step": 27035
    },
    {
      "epoch": 15.943396226415095,
      "grad_norm": 3.2916293144226074,
      "learning_rate": 1.2020500271307721e-06,
      "loss": 0.3916,
      "num_input_tokens_seen": 17651040,
      "step": 27040
    },
    {
      "epoch": 15.946344339622641,
      "grad_norm": 3.110333204269409,
      "learning_rate": 1.200377231013176e-06,
      "loss": 0.4676,
      "num_input_tokens_seen": 17653728,
      "step": 27045
    },
    {
      "epoch": 15.94929245283019,
      "grad_norm": 4.068283557891846,
      "learning_rate": 1.1987054408620825e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 17656352,
      "step": 27050
    },
    {
      "epoch": 15.952240566037736,
      "grad_norm": 3.59666109085083,
      "learning_rate": 1.197034657120107e-06,
      "loss": 0.3539,
      "num_input_tokens_seen": 17662432,
      "step": 27055
    },
    {
      "epoch": 15.955188679245284,
      "grad_norm": 2.68633770942688,
      "learning_rate": 1.1953648802295964e-06,
      "loss": 0.2473,
      "num_input_tokens_seen": 17665888,
      "step": 27060
    },
    {
      "epoch": 15.95813679245283,
      "grad_norm": 5.497695446014404,
      "learning_rate": 1.1936961106326307e-06,
      "loss": 0.278,
      "num_input_tokens_seen": 17669056,
      "step": 27065
    },
    {
      "epoch": 15.961084905660378,
      "grad_norm": 5.573581695556641,
      "learning_rate": 1.1920283487710237e-06,
      "loss": 0.256,
      "num_input_tokens_seen": 17671200,
      "step": 27070
    },
    {
      "epoch": 15.964033018867925,
      "grad_norm": 4.499593257904053,
      "learning_rate": 1.1903615950863228e-06,
      "loss": 0.2659,
      "num_input_tokens_seen": 17673856,
      "step": 27075
    },
    {
      "epoch": 15.966981132075471,
      "grad_norm": 2.468602418899536,
      "learning_rate": 1.1886958500198076e-06,
      "loss": 0.3612,
      "num_input_tokens_seen": 17677280,
      "step": 27080
    },
    {
      "epoch": 15.96992924528302,
      "grad_norm": 3.609217405319214,
      "learning_rate": 1.1870311140124923e-06,
      "loss": 0.3786,
      "num_input_tokens_seen": 17680800,
      "step": 27085
    },
    {
      "epoch": 15.972877358490566,
      "grad_norm": 17.841140747070312,
      "learning_rate": 1.185367387505123e-06,
      "loss": 0.3165,
      "num_input_tokens_seen": 17683520,
      "step": 27090
    },
    {
      "epoch": 15.975825471698114,
      "grad_norm": 3.465113639831543,
      "learning_rate": 1.1837046709381783e-06,
      "loss": 0.6599,
      "num_input_tokens_seen": 17686272,
      "step": 27095
    },
    {
      "epoch": 15.97877358490566,
      "grad_norm": 4.342942237854004,
      "learning_rate": 1.1820429647518678e-06,
      "loss": 0.2871,
      "num_input_tokens_seen": 17689952,
      "step": 27100
    },
    {
      "epoch": 15.981721698113208,
      "grad_norm": 4.398411273956299,
      "learning_rate": 1.1803822693861377e-06,
      "loss": 0.3556,
      "num_input_tokens_seen": 17692672,
      "step": 27105
    },
    {
      "epoch": 15.984669811320755,
      "grad_norm": 2.412160634994507,
      "learning_rate": 1.1787225852806639e-06,
      "loss": 0.2598,
      "num_input_tokens_seen": 17695712,
      "step": 27110
    },
    {
      "epoch": 15.987617924528301,
      "grad_norm": 3.93445086479187,
      "learning_rate": 1.177063912874853e-06,
      "loss": 0.3386,
      "num_input_tokens_seen": 17698400,
      "step": 27115
    },
    {
      "epoch": 15.99056603773585,
      "grad_norm": 5.107640266418457,
      "learning_rate": 1.1754062526078487e-06,
      "loss": 0.537,
      "num_input_tokens_seen": 17701856,
      "step": 27120
    },
    {
      "epoch": 15.993514150943396,
      "grad_norm": 5.500094890594482,
      "learning_rate": 1.1737496049185215e-06,
      "loss": 0.3308,
      "num_input_tokens_seen": 17704864,
      "step": 27125
    },
    {
      "epoch": 15.996462264150944,
      "grad_norm": 4.875102519989014,
      "learning_rate": 1.172093970245477e-06,
      "loss": 0.3565,
      "num_input_tokens_seen": 17708480,
      "step": 27130
    },
    {
      "epoch": 15.99941037735849,
      "grad_norm": 2.8622496128082275,
      "learning_rate": 1.1704393490270516e-06,
      "loss": 0.3286,
      "num_input_tokens_seen": 17712032,
      "step": 27135
    },
    {
      "epoch": 16.0,
      "eval_loss": 0.6074872016906738,
      "eval_runtime": 19.3398,
      "eval_samples_per_second": 87.695,
      "eval_steps_per_second": 21.924,
      "num_input_tokens_seen": 17712104,
      "step": 27136
    },
    {
      "epoch": 16.00235849056604,
      "grad_norm": 2.9562137126922607,
      "learning_rate": 1.1687857417013126e-06,
      "loss": 0.2119,
      "num_input_tokens_seen": 17715048,
      "step": 27140
    },
    {
      "epoch": 16.005306603773583,
      "grad_norm": 3.408649206161499,
      "learning_rate": 1.1671331487060583e-06,
      "loss": 0.3081,
      "num_input_tokens_seen": 17717992,
      "step": 27145
    },
    {
      "epoch": 16.00825471698113,
      "grad_norm": 4.2913641929626465,
      "learning_rate": 1.1654815704788237e-06,
      "loss": 0.3316,
      "num_input_tokens_seen": 17720328,
      "step": 27150
    },
    {
      "epoch": 16.01120283018868,
      "grad_norm": 2.9438507556915283,
      "learning_rate": 1.1638310074568687e-06,
      "loss": 0.2589,
      "num_input_tokens_seen": 17723880,
      "step": 27155
    },
    {
      "epoch": 16.014150943396228,
      "grad_norm": 3.5376648902893066,
      "learning_rate": 1.162181460077188e-06,
      "loss": 0.2891,
      "num_input_tokens_seen": 17727944,
      "step": 27160
    },
    {
      "epoch": 16.017099056603772,
      "grad_norm": 2.6878323554992676,
      "learning_rate": 1.1605329287765056e-06,
      "loss": 0.2962,
      "num_input_tokens_seen": 17731848,
      "step": 27165
    },
    {
      "epoch": 16.02004716981132,
      "grad_norm": 6.028411388397217,
      "learning_rate": 1.1588854139912775e-06,
      "loss": 0.2633,
      "num_input_tokens_seen": 17734536,
      "step": 27170
    },
    {
      "epoch": 16.02299528301887,
      "grad_norm": 2.871877908706665,
      "learning_rate": 1.1572389161576886e-06,
      "loss": 0.2476,
      "num_input_tokens_seen": 17738312,
      "step": 27175
    },
    {
      "epoch": 16.025943396226417,
      "grad_norm": 3.234715461730957,
      "learning_rate": 1.15559343571166e-06,
      "loss": 0.4312,
      "num_input_tokens_seen": 17741480,
      "step": 27180
    },
    {
      "epoch": 16.02889150943396,
      "grad_norm": 6.845321178436279,
      "learning_rate": 1.153948973088837e-06,
      "loss": 0.3601,
      "num_input_tokens_seen": 17744520,
      "step": 27185
    },
    {
      "epoch": 16.03183962264151,
      "grad_norm": 3.866924524307251,
      "learning_rate": 1.1523055287245993e-06,
      "loss": 0.2473,
      "num_input_tokens_seen": 17747080,
      "step": 27190
    },
    {
      "epoch": 16.034787735849058,
      "grad_norm": 4.404837131500244,
      "learning_rate": 1.150663103054056e-06,
      "loss": 0.3196,
      "num_input_tokens_seen": 17750632,
      "step": 27195
    },
    {
      "epoch": 16.037735849056602,
      "grad_norm": 4.610869407653809,
      "learning_rate": 1.1490216965120438e-06,
      "loss": 0.2132,
      "num_input_tokens_seen": 17753864,
      "step": 27200
    },
    {
      "epoch": 16.04068396226415,
      "grad_norm": 3.881624937057495,
      "learning_rate": 1.147381309533136e-06,
      "loss": 0.3224,
      "num_input_tokens_seen": 17756744,
      "step": 27205
    },
    {
      "epoch": 16.0436320754717,
      "grad_norm": 3.3644521236419678,
      "learning_rate": 1.1457419425516287e-06,
      "loss": 0.2373,
      "num_input_tokens_seen": 17759720,
      "step": 27210
    },
    {
      "epoch": 16.046580188679247,
      "grad_norm": 2.6429624557495117,
      "learning_rate": 1.1441035960015544e-06,
      "loss": 0.3176,
      "num_input_tokens_seen": 17762792,
      "step": 27215
    },
    {
      "epoch": 16.04952830188679,
      "grad_norm": 5.130011558532715,
      "learning_rate": 1.1424662703166716e-06,
      "loss": 0.2959,
      "num_input_tokens_seen": 17765960,
      "step": 27220
    },
    {
      "epoch": 16.05247641509434,
      "grad_norm": 4.988994598388672,
      "learning_rate": 1.1408299659304684e-06,
      "loss": 0.4118,
      "num_input_tokens_seen": 17769160,
      "step": 27225
    },
    {
      "epoch": 16.055424528301888,
      "grad_norm": 4.97374963760376,
      "learning_rate": 1.1391946832761642e-06,
      "loss": 0.2657,
      "num_input_tokens_seen": 17772424,
      "step": 27230
    },
    {
      "epoch": 16.058372641509433,
      "grad_norm": 1.6416126489639282,
      "learning_rate": 1.137560422786706e-06,
      "loss": 0.2424,
      "num_input_tokens_seen": 17775336,
      "step": 27235
    },
    {
      "epoch": 16.06132075471698,
      "grad_norm": 3.584343194961548,
      "learning_rate": 1.1359271848947712e-06,
      "loss": 0.3095,
      "num_input_tokens_seen": 17778600,
      "step": 27240
    },
    {
      "epoch": 16.06426886792453,
      "grad_norm": 3.2108006477355957,
      "learning_rate": 1.1342949700327688e-06,
      "loss": 0.4226,
      "num_input_tokens_seen": 17782760,
      "step": 27245
    },
    {
      "epoch": 16.067216981132077,
      "grad_norm": 2.745878219604492,
      "learning_rate": 1.1326637786328332e-06,
      "loss": 0.4028,
      "num_input_tokens_seen": 17786952,
      "step": 27250
    },
    {
      "epoch": 16.07016509433962,
      "grad_norm": 3.7533814907073975,
      "learning_rate": 1.1310336111268293e-06,
      "loss": 0.3196,
      "num_input_tokens_seen": 17789544,
      "step": 27255
    },
    {
      "epoch": 16.07311320754717,
      "grad_norm": 2.7761082649230957,
      "learning_rate": 1.1294044679463517e-06,
      "loss": 0.3191,
      "num_input_tokens_seen": 17793576,
      "step": 27260
    },
    {
      "epoch": 16.076061320754718,
      "grad_norm": 2.255887269973755,
      "learning_rate": 1.1277763495227207e-06,
      "loss": 0.2853,
      "num_input_tokens_seen": 17797896,
      "step": 27265
    },
    {
      "epoch": 16.079009433962263,
      "grad_norm": 3.3480160236358643,
      "learning_rate": 1.1261492562869913e-06,
      "loss": 0.2075,
      "num_input_tokens_seen": 17800712,
      "step": 27270
    },
    {
      "epoch": 16.08195754716981,
      "grad_norm": 4.009567737579346,
      "learning_rate": 1.1245231886699415e-06,
      "loss": 0.2727,
      "num_input_tokens_seen": 17803592,
      "step": 27275
    },
    {
      "epoch": 16.08490566037736,
      "grad_norm": 3.303662061691284,
      "learning_rate": 1.12289814710208e-06,
      "loss": 0.3031,
      "num_input_tokens_seen": 17806920,
      "step": 27280
    },
    {
      "epoch": 16.087853773584907,
      "grad_norm": 3.079800605773926,
      "learning_rate": 1.1212741320136433e-06,
      "loss": 0.3875,
      "num_input_tokens_seen": 17810728,
      "step": 27285
    },
    {
      "epoch": 16.090801886792452,
      "grad_norm": 6.233506202697754,
      "learning_rate": 1.1196511438345963e-06,
      "loss": 0.3086,
      "num_input_tokens_seen": 17813928,
      "step": 27290
    },
    {
      "epoch": 16.09375,
      "grad_norm": 4.258075714111328,
      "learning_rate": 1.118029182994631e-06,
      "loss": 0.2685,
      "num_input_tokens_seen": 17817128,
      "step": 27295
    },
    {
      "epoch": 16.096698113207548,
      "grad_norm": 2.6861300468444824,
      "learning_rate": 1.1164082499231704e-06,
      "loss": 0.3804,
      "num_input_tokens_seen": 17820968,
      "step": 27300
    },
    {
      "epoch": 16.099646226415093,
      "grad_norm": 3.8812906742095947,
      "learning_rate": 1.114788345049364e-06,
      "loss": 0.266,
      "num_input_tokens_seen": 17823944,
      "step": 27305
    },
    {
      "epoch": 16.10259433962264,
      "grad_norm": 5.270434379577637,
      "learning_rate": 1.1131694688020872e-06,
      "loss": 0.4085,
      "num_input_tokens_seen": 17826824,
      "step": 27310
    },
    {
      "epoch": 16.10554245283019,
      "grad_norm": 3.719679832458496,
      "learning_rate": 1.1115516216099453e-06,
      "loss": 0.3681,
      "num_input_tokens_seen": 17829800,
      "step": 27315
    },
    {
      "epoch": 16.108490566037737,
      "grad_norm": 3.4988315105438232,
      "learning_rate": 1.1099348039012698e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 17832744,
      "step": 27320
    },
    {
      "epoch": 16.111438679245282,
      "grad_norm": 5.051823616027832,
      "learning_rate": 1.1083190161041202e-06,
      "loss": 0.3687,
      "num_input_tokens_seen": 17835240,
      "step": 27325
    },
    {
      "epoch": 16.11438679245283,
      "grad_norm": 4.022293567657471,
      "learning_rate": 1.1067042586462822e-06,
      "loss": 0.2662,
      "num_input_tokens_seen": 17838024,
      "step": 27330
    },
    {
      "epoch": 16.11733490566038,
      "grad_norm": 4.672703266143799,
      "learning_rate": 1.1050905319552718e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 17841032,
      "step": 27335
    },
    {
      "epoch": 16.120283018867923,
      "grad_norm": 3.8369011878967285,
      "learning_rate": 1.1034778364583293e-06,
      "loss": 0.2946,
      "num_input_tokens_seen": 17844968,
      "step": 27340
    },
    {
      "epoch": 16.12323113207547,
      "grad_norm": 4.505640506744385,
      "learning_rate": 1.1018661725824231e-06,
      "loss": 0.3007,
      "num_input_tokens_seen": 17847464,
      "step": 27345
    },
    {
      "epoch": 16.12617924528302,
      "grad_norm": 3.8580055236816406,
      "learning_rate": 1.100255540754247e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 17850504,
      "step": 27350
    },
    {
      "epoch": 16.129127358490567,
      "grad_norm": 3.8682162761688232,
      "learning_rate": 1.0986459414002244e-06,
      "loss": 0.3576,
      "num_input_tokens_seen": 17854248,
      "step": 27355
    },
    {
      "epoch": 16.132075471698112,
      "grad_norm": 3.619281053543091,
      "learning_rate": 1.0970373749465008e-06,
      "loss": 0.354,
      "num_input_tokens_seen": 17858600,
      "step": 27360
    },
    {
      "epoch": 16.13502358490566,
      "grad_norm": 4.133915424346924,
      "learning_rate": 1.095429841818954e-06,
      "loss": 0.4058,
      "num_input_tokens_seen": 17861928,
      "step": 27365
    },
    {
      "epoch": 16.13797169811321,
      "grad_norm": 6.480722427368164,
      "learning_rate": 1.093823342443185e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 17864584,
      "step": 27370
    },
    {
      "epoch": 16.140919811320753,
      "grad_norm": 8.885051727294922,
      "learning_rate": 1.0922178772445203e-06,
      "loss": 0.3409,
      "num_input_tokens_seen": 17867784,
      "step": 27375
    },
    {
      "epoch": 16.1438679245283,
      "grad_norm": 3.8647215366363525,
      "learning_rate": 1.0906134466480146e-06,
      "loss": 0.3639,
      "num_input_tokens_seen": 17871560,
      "step": 27380
    },
    {
      "epoch": 16.14681603773585,
      "grad_norm": 4.726562023162842,
      "learning_rate": 1.0890100510784473e-06,
      "loss": 0.354,
      "num_input_tokens_seen": 17874440,
      "step": 27385
    },
    {
      "epoch": 16.149764150943398,
      "grad_norm": 3.3339004516601562,
      "learning_rate": 1.0874076909603227e-06,
      "loss": 0.4718,
      "num_input_tokens_seen": 17880584,
      "step": 27390
    },
    {
      "epoch": 16.152712264150942,
      "grad_norm": 4.483731269836426,
      "learning_rate": 1.0858063667178747e-06,
      "loss": 0.3215,
      "num_input_tokens_seen": 17884360,
      "step": 27395
    },
    {
      "epoch": 16.15566037735849,
      "grad_norm": 4.394938945770264,
      "learning_rate": 1.0842060787750614e-06,
      "loss": 0.4204,
      "num_input_tokens_seen": 17887272,
      "step": 27400
    },
    {
      "epoch": 16.15860849056604,
      "grad_norm": 3.872849225997925,
      "learning_rate": 1.0826068275555652e-06,
      "loss": 0.2962,
      "num_input_tokens_seen": 17890888,
      "step": 27405
    },
    {
      "epoch": 16.161556603773583,
      "grad_norm": 3.1812868118286133,
      "learning_rate": 1.081008613482794e-06,
      "loss": 0.2733,
      "num_input_tokens_seen": 17894440,
      "step": 27410
    },
    {
      "epoch": 16.16450471698113,
      "grad_norm": 2.7634010314941406,
      "learning_rate": 1.079411436979883e-06,
      "loss": 0.2423,
      "num_input_tokens_seen": 17898184,
      "step": 27415
    },
    {
      "epoch": 16.16745283018868,
      "grad_norm": 3.4479756355285645,
      "learning_rate": 1.0778152984696905e-06,
      "loss": 0.318,
      "num_input_tokens_seen": 17901576,
      "step": 27420
    },
    {
      "epoch": 16.170400943396228,
      "grad_norm": 1.9659678936004639,
      "learning_rate": 1.0762201983747993e-06,
      "loss": 0.2908,
      "num_input_tokens_seen": 17905640,
      "step": 27425
    },
    {
      "epoch": 16.173349056603772,
      "grad_norm": 2.5028533935546875,
      "learning_rate": 1.0746261371175238e-06,
      "loss": 0.2446,
      "num_input_tokens_seen": 17909224,
      "step": 27430
    },
    {
      "epoch": 16.17629716981132,
      "grad_norm": 3.786068916320801,
      "learning_rate": 1.0730331151198953e-06,
      "loss": 0.2277,
      "num_input_tokens_seen": 17912232,
      "step": 27435
    },
    {
      "epoch": 16.17924528301887,
      "grad_norm": 4.085115909576416,
      "learning_rate": 1.0714411328036733e-06,
      "loss": 0.287,
      "num_input_tokens_seen": 17915464,
      "step": 27440
    },
    {
      "epoch": 16.182193396226417,
      "grad_norm": 2.496640682220459,
      "learning_rate": 1.0698501905903435e-06,
      "loss": 0.2722,
      "num_input_tokens_seen": 17918120,
      "step": 27445
    },
    {
      "epoch": 16.18514150943396,
      "grad_norm": 2.3432068824768066,
      "learning_rate": 1.0682602889011134e-06,
      "loss": 0.3386,
      "num_input_tokens_seen": 17921704,
      "step": 27450
    },
    {
      "epoch": 16.18808962264151,
      "grad_norm": 4.427986145019531,
      "learning_rate": 1.0666714281569152e-06,
      "loss": 0.225,
      "num_input_tokens_seen": 17924616,
      "step": 27455
    },
    {
      "epoch": 16.191037735849058,
      "grad_norm": 3.4739325046539307,
      "learning_rate": 1.0650836087784095e-06,
      "loss": 0.3371,
      "num_input_tokens_seen": 17927240,
      "step": 27460
    },
    {
      "epoch": 16.193985849056602,
      "grad_norm": 2.6365692615509033,
      "learning_rate": 1.0634968311859768e-06,
      "loss": 0.2921,
      "num_input_tokens_seen": 17929704,
      "step": 27465
    },
    {
      "epoch": 16.19693396226415,
      "grad_norm": 2.857795000076294,
      "learning_rate": 1.0619110957997237e-06,
      "loss": 0.3896,
      "num_input_tokens_seen": 17932392,
      "step": 27470
    },
    {
      "epoch": 16.1998820754717,
      "grad_norm": 2.2339065074920654,
      "learning_rate": 1.06032640303948e-06,
      "loss": 0.2726,
      "num_input_tokens_seen": 17936264,
      "step": 27475
    },
    {
      "epoch": 16.202830188679247,
      "grad_norm": 3.4672091007232666,
      "learning_rate": 1.0587427533248002e-06,
      "loss": 0.4424,
      "num_input_tokens_seen": 17939848,
      "step": 27480
    },
    {
      "epoch": 16.20577830188679,
      "grad_norm": 3.4370508193969727,
      "learning_rate": 1.057160147074961e-06,
      "loss": 0.2593,
      "num_input_tokens_seen": 17943080,
      "step": 27485
    },
    {
      "epoch": 16.20872641509434,
      "grad_norm": 3.60244083404541,
      "learning_rate": 1.0555785847089657e-06,
      "loss": 0.2887,
      "num_input_tokens_seen": 17945384,
      "step": 27490
    },
    {
      "epoch": 16.211674528301888,
      "grad_norm": 2.8007898330688477,
      "learning_rate": 1.0539980666455407e-06,
      "loss": 0.379,
      "num_input_tokens_seen": 17948488,
      "step": 27495
    },
    {
      "epoch": 16.214622641509433,
      "grad_norm": 3.984285593032837,
      "learning_rate": 1.052418593303134e-06,
      "loss": 0.3682,
      "num_input_tokens_seen": 17953000,
      "step": 27500
    },
    {
      "epoch": 16.21757075471698,
      "grad_norm": 3.0932931900024414,
      "learning_rate": 1.0508401650999178e-06,
      "loss": 0.2317,
      "num_input_tokens_seen": 17956648,
      "step": 27505
    },
    {
      "epoch": 16.22051886792453,
      "grad_norm": 5.290530204772949,
      "learning_rate": 1.0492627824537877e-06,
      "loss": 0.2588,
      "num_input_tokens_seen": 17959368,
      "step": 27510
    },
    {
      "epoch": 16.223466981132077,
      "grad_norm": 4.802031517028809,
      "learning_rate": 1.0476864457823626e-06,
      "loss": 0.2621,
      "num_input_tokens_seen": 17963016,
      "step": 27515
    },
    {
      "epoch": 16.22641509433962,
      "grad_norm": 8.34684944152832,
      "learning_rate": 1.0461111555029836e-06,
      "loss": 0.4282,
      "num_input_tokens_seen": 17965448,
      "step": 27520
    },
    {
      "epoch": 16.22936320754717,
      "grad_norm": 2.3559815883636475,
      "learning_rate": 1.0445369120327175e-06,
      "loss": 0.2906,
      "num_input_tokens_seen": 17969512,
      "step": 27525
    },
    {
      "epoch": 16.232311320754718,
      "grad_norm": 5.384519577026367,
      "learning_rate": 1.0429637157883516e-06,
      "loss": 0.4581,
      "num_input_tokens_seen": 17973000,
      "step": 27530
    },
    {
      "epoch": 16.235259433962263,
      "grad_norm": 2.6878702640533447,
      "learning_rate": 1.041391567186395e-06,
      "loss": 0.2866,
      "num_input_tokens_seen": 17976232,
      "step": 27535
    },
    {
      "epoch": 16.23820754716981,
      "grad_norm": 4.6357550621032715,
      "learning_rate": 1.0398204666430821e-06,
      "loss": 0.2283,
      "num_input_tokens_seen": 17978728,
      "step": 27540
    },
    {
      "epoch": 16.24115566037736,
      "grad_norm": 5.27734375,
      "learning_rate": 1.0382504145743667e-06,
      "loss": 0.229,
      "num_input_tokens_seen": 17982120,
      "step": 27545
    },
    {
      "epoch": 16.244103773584907,
      "grad_norm": 3.3389225006103516,
      "learning_rate": 1.0366814113959294e-06,
      "loss": 0.2887,
      "num_input_tokens_seen": 17985192,
      "step": 27550
    },
    {
      "epoch": 16.247051886792452,
      "grad_norm": 4.908576011657715,
      "learning_rate": 1.0351134575231697e-06,
      "loss": 0.2458,
      "num_input_tokens_seen": 17988456,
      "step": 27555
    },
    {
      "epoch": 16.25,
      "grad_norm": 2.8062827587127686,
      "learning_rate": 1.0335465533712098e-06,
      "loss": 0.2224,
      "num_input_tokens_seen": 17991272,
      "step": 27560
    },
    {
      "epoch": 16.252948113207548,
      "grad_norm": 4.088833332061768,
      "learning_rate": 1.031980699354894e-06,
      "loss": 0.2824,
      "num_input_tokens_seen": 17994792,
      "step": 27565
    },
    {
      "epoch": 16.255896226415093,
      "grad_norm": 3.451603651046753,
      "learning_rate": 1.03041589588879e-06,
      "loss": 0.2755,
      "num_input_tokens_seen": 17997992,
      "step": 27570
    },
    {
      "epoch": 16.25884433962264,
      "grad_norm": 3.2184457778930664,
      "learning_rate": 1.0288521433871834e-06,
      "loss": 0.2818,
      "num_input_tokens_seen": 18001256,
      "step": 27575
    },
    {
      "epoch": 16.26179245283019,
      "grad_norm": 3.632542610168457,
      "learning_rate": 1.0272894422640866e-06,
      "loss": 0.3305,
      "num_input_tokens_seen": 18004008,
      "step": 27580
    },
    {
      "epoch": 16.264740566037737,
      "grad_norm": 4.782447814941406,
      "learning_rate": 1.0257277929332332e-06,
      "loss": 0.2132,
      "num_input_tokens_seen": 18007176,
      "step": 27585
    },
    {
      "epoch": 16.267688679245282,
      "grad_norm": 1.9274590015411377,
      "learning_rate": 1.0241671958080745e-06,
      "loss": 0.3307,
      "num_input_tokens_seen": 18011240,
      "step": 27590
    },
    {
      "epoch": 16.27063679245283,
      "grad_norm": 4.54290246963501,
      "learning_rate": 1.0226076513017858e-06,
      "loss": 0.3755,
      "num_input_tokens_seen": 18014664,
      "step": 27595
    },
    {
      "epoch": 16.27358490566038,
      "grad_norm": 5.171290397644043,
      "learning_rate": 1.0210491598272625e-06,
      "loss": 0.2907,
      "num_input_tokens_seen": 18018536,
      "step": 27600
    },
    {
      "epoch": 16.276533018867923,
      "grad_norm": 4.232462406158447,
      "learning_rate": 1.0194917217971229e-06,
      "loss": 0.4067,
      "num_input_tokens_seen": 18022312,
      "step": 27605
    },
    {
      "epoch": 16.27948113207547,
      "grad_norm": 4.152926921844482,
      "learning_rate": 1.0179353376237038e-06,
      "loss": 0.2838,
      "num_input_tokens_seen": 18024968,
      "step": 27610
    },
    {
      "epoch": 16.28242924528302,
      "grad_norm": 2.1672232151031494,
      "learning_rate": 1.0163800077190672e-06,
      "loss": 0.2966,
      "num_input_tokens_seen": 18028104,
      "step": 27615
    },
    {
      "epoch": 16.285377358490567,
      "grad_norm": 2.8170223236083984,
      "learning_rate": 1.0148257324949916e-06,
      "loss": 0.3349,
      "num_input_tokens_seen": 18031080,
      "step": 27620
    },
    {
      "epoch": 16.288325471698112,
      "grad_norm": 3.1752476692199707,
      "learning_rate": 1.0132725123629783e-06,
      "loss": 0.2549,
      "num_input_tokens_seen": 18033832,
      "step": 27625
    },
    {
      "epoch": 16.29127358490566,
      "grad_norm": 4.458883762359619,
      "learning_rate": 1.0117203477342497e-06,
      "loss": 0.3281,
      "num_input_tokens_seen": 18036264,
      "step": 27630
    },
    {
      "epoch": 16.29422169811321,
      "grad_norm": 4.837848663330078,
      "learning_rate": 1.0101692390197477e-06,
      "loss": 0.3533,
      "num_input_tokens_seen": 18040616,
      "step": 27635
    },
    {
      "epoch": 16.297169811320753,
      "grad_norm": 3.146350145339966,
      "learning_rate": 1.0086191866301331e-06,
      "loss": 0.2777,
      "num_input_tokens_seen": 18044712,
      "step": 27640
    },
    {
      "epoch": 16.3001179245283,
      "grad_norm": 4.456493854522705,
      "learning_rate": 1.0070701909757918e-06,
      "loss": 0.2246,
      "num_input_tokens_seen": 18049256,
      "step": 27645
    },
    {
      "epoch": 16.30306603773585,
      "grad_norm": 3.4150750637054443,
      "learning_rate": 1.0055222524668267e-06,
      "loss": 0.383,
      "num_input_tokens_seen": 18053032,
      "step": 27650
    },
    {
      "epoch": 16.306014150943398,
      "grad_norm": 4.103562355041504,
      "learning_rate": 1.00397537151306e-06,
      "loss": 0.3418,
      "num_input_tokens_seen": 18055368,
      "step": 27655
    },
    {
      "epoch": 16.308962264150942,
      "grad_norm": 3.3517467975616455,
      "learning_rate": 1.002429548524036e-06,
      "loss": 0.4552,
      "num_input_tokens_seen": 18058856,
      "step": 27660
    },
    {
      "epoch": 16.31191037735849,
      "grad_norm": 4.0385050773620605,
      "learning_rate": 1.0008847839090175e-06,
      "loss": 0.2833,
      "num_input_tokens_seen": 18062280,
      "step": 27665
    },
    {
      "epoch": 16.31485849056604,
      "grad_norm": 3.517468214035034,
      "learning_rate": 9.993410780769862e-07,
      "loss": 0.294,
      "num_input_tokens_seen": 18065288,
      "step": 27670
    },
    {
      "epoch": 16.317806603773583,
      "grad_norm": 3.2257468700408936,
      "learning_rate": 9.977984314366463e-07,
      "loss": 0.3718,
      "num_input_tokens_seen": 18069352,
      "step": 27675
    },
    {
      "epoch": 16.32075471698113,
      "grad_norm": 4.003576278686523,
      "learning_rate": 9.962568443964216e-07,
      "loss": 0.2307,
      "num_input_tokens_seen": 18073224,
      "step": 27680
    },
    {
      "epoch": 16.32370283018868,
      "grad_norm": 3.413637638092041,
      "learning_rate": 9.947163173644524e-07,
      "loss": 0.4056,
      "num_input_tokens_seen": 18075880,
      "step": 27685
    },
    {
      "epoch": 16.326650943396228,
      "grad_norm": 3.6316475868225098,
      "learning_rate": 9.931768507486007e-07,
      "loss": 0.2761,
      "num_input_tokens_seen": 18078792,
      "step": 27690
    },
    {
      "epoch": 16.329599056603772,
      "grad_norm": 6.669281005859375,
      "learning_rate": 9.916384449564453e-07,
      "loss": 0.3088,
      "num_input_tokens_seen": 18082248,
      "step": 27695
    },
    {
      "epoch": 16.33254716981132,
      "grad_norm": 2.2219507694244385,
      "learning_rate": 9.90101100395287e-07,
      "loss": 0.2517,
      "num_input_tokens_seen": 18085384,
      "step": 27700
    },
    {
      "epoch": 16.33549528301887,
      "grad_norm": 2.369577407836914,
      "learning_rate": 9.885648174721428e-07,
      "loss": 0.332,
      "num_input_tokens_seen": 18088616,
      "step": 27705
    },
    {
      "epoch": 16.338443396226417,
      "grad_norm": 2.9724032878875732,
      "learning_rate": 9.870295965937532e-07,
      "loss": 0.2902,
      "num_input_tokens_seen": 18092808,
      "step": 27710
    },
    {
      "epoch": 16.34139150943396,
      "grad_norm": 3.9808011054992676,
      "learning_rate": 9.854954381665727e-07,
      "loss": 0.396,
      "num_input_tokens_seen": 18095656,
      "step": 27715
    },
    {
      "epoch": 16.34433962264151,
      "grad_norm": 4.299487590789795,
      "learning_rate": 9.83962342596776e-07,
      "loss": 0.2304,
      "num_input_tokens_seen": 18098120,
      "step": 27720
    },
    {
      "epoch": 16.347287735849058,
      "grad_norm": 2.4551925659179688,
      "learning_rate": 9.824303102902576e-07,
      "loss": 0.2218,
      "num_input_tokens_seen": 18101768,
      "step": 27725
    },
    {
      "epoch": 16.350235849056602,
      "grad_norm": 5.087007522583008,
      "learning_rate": 9.808993416526292e-07,
      "loss": 0.33,
      "num_input_tokens_seen": 18105352,
      "step": 27730
    },
    {
      "epoch": 16.35318396226415,
      "grad_norm": 4.335859298706055,
      "learning_rate": 9.793694370892204e-07,
      "loss": 0.2587,
      "num_input_tokens_seen": 18110600,
      "step": 27735
    },
    {
      "epoch": 16.3561320754717,
      "grad_norm": 3.5604605674743652,
      "learning_rate": 9.77840597005082e-07,
      "loss": 0.2693,
      "num_input_tokens_seen": 18114440,
      "step": 27740
    },
    {
      "epoch": 16.359080188679247,
      "grad_norm": 4.816673278808594,
      "learning_rate": 9.763128218049806e-07,
      "loss": 0.4437,
      "num_input_tokens_seen": 18117352,
      "step": 27745
    },
    {
      "epoch": 16.36202830188679,
      "grad_norm": 2.940211772918701,
      "learning_rate": 9.747861118934005e-07,
      "loss": 0.3237,
      "num_input_tokens_seen": 18120680,
      "step": 27750
    },
    {
      "epoch": 16.36497641509434,
      "grad_norm": 4.41790246963501,
      "learning_rate": 9.732604676745443e-07,
      "loss": 0.2719,
      "num_input_tokens_seen": 18123240,
      "step": 27755
    },
    {
      "epoch": 16.367924528301888,
      "grad_norm": 5.458648681640625,
      "learning_rate": 9.717358895523333e-07,
      "loss": 0.3171,
      "num_input_tokens_seen": 18125832,
      "step": 27760
    },
    {
      "epoch": 16.370872641509433,
      "grad_norm": 3.7510673999786377,
      "learning_rate": 9.702123779304074e-07,
      "loss": 0.2841,
      "num_input_tokens_seen": 18128872,
      "step": 27765
    },
    {
      "epoch": 16.37382075471698,
      "grad_norm": 3.298093795776367,
      "learning_rate": 9.686899332121203e-07,
      "loss": 0.2985,
      "num_input_tokens_seen": 18131688,
      "step": 27770
    },
    {
      "epoch": 16.37676886792453,
      "grad_norm": 1.7564769983291626,
      "learning_rate": 9.671685558005488e-07,
      "loss": 0.1736,
      "num_input_tokens_seen": 18135304,
      "step": 27775
    },
    {
      "epoch": 16.379716981132077,
      "grad_norm": 4.581481456756592,
      "learning_rate": 9.656482460984828e-07,
      "loss": 0.2442,
      "num_input_tokens_seen": 18138504,
      "step": 27780
    },
    {
      "epoch": 16.38266509433962,
      "grad_norm": 2.838304281234741,
      "learning_rate": 9.641290045084307e-07,
      "loss": 0.3456,
      "num_input_tokens_seen": 18141864,
      "step": 27785
    },
    {
      "epoch": 16.38561320754717,
      "grad_norm": 3.9666428565979004,
      "learning_rate": 9.626108314326182e-07,
      "loss": 0.2506,
      "num_input_tokens_seen": 18144872,
      "step": 27790
    },
    {
      "epoch": 16.388561320754718,
      "grad_norm": 3.7685439586639404,
      "learning_rate": 9.610937272729881e-07,
      "loss": 0.2807,
      "num_input_tokens_seen": 18149448,
      "step": 27795
    },
    {
      "epoch": 16.391509433962263,
      "grad_norm": 2.9581820964813232,
      "learning_rate": 9.595776924311996e-07,
      "loss": 0.4135,
      "num_input_tokens_seen": 18152968,
      "step": 27800
    },
    {
      "epoch": 16.39445754716981,
      "grad_norm": 8.78770923614502,
      "learning_rate": 9.580627273086313e-07,
      "loss": 0.3678,
      "num_input_tokens_seen": 18156264,
      "step": 27805
    },
    {
      "epoch": 16.39740566037736,
      "grad_norm": 5.531522274017334,
      "learning_rate": 9.565488323063754e-07,
      "loss": 0.2459,
      "num_input_tokens_seen": 18159112,
      "step": 27810
    },
    {
      "epoch": 16.400353773584907,
      "grad_norm": 4.561558723449707,
      "learning_rate": 9.55036007825243e-07,
      "loss": 0.3786,
      "num_input_tokens_seen": 18161800,
      "step": 27815
    },
    {
      "epoch": 16.403301886792452,
      "grad_norm": 4.129857063293457,
      "learning_rate": 9.535242542657602e-07,
      "loss": 0.3418,
      "num_input_tokens_seen": 18164808,
      "step": 27820
    },
    {
      "epoch": 16.40625,
      "grad_norm": 4.079511642456055,
      "learning_rate": 9.520135720281692e-07,
      "loss": 0.508,
      "num_input_tokens_seen": 18168072,
      "step": 27825
    },
    {
      "epoch": 16.409198113207548,
      "grad_norm": 5.035636901855469,
      "learning_rate": 9.505039615124318e-07,
      "loss": 0.2945,
      "num_input_tokens_seen": 18171464,
      "step": 27830
    },
    {
      "epoch": 16.412146226415093,
      "grad_norm": 3.3385486602783203,
      "learning_rate": 9.489954231182235e-07,
      "loss": 0.3909,
      "num_input_tokens_seen": 18174824,
      "step": 27835
    },
    {
      "epoch": 16.41509433962264,
      "grad_norm": 3.351029872894287,
      "learning_rate": 9.474879572449352e-07,
      "loss": 0.309,
      "num_input_tokens_seen": 18177672,
      "step": 27840
    },
    {
      "epoch": 16.41804245283019,
      "grad_norm": 4.073945045471191,
      "learning_rate": 9.459815642916759e-07,
      "loss": 0.2923,
      "num_input_tokens_seen": 18180680,
      "step": 27845
    },
    {
      "epoch": 16.420990566037737,
      "grad_norm": 3.6125144958496094,
      "learning_rate": 9.444762446572692e-07,
      "loss": 0.1989,
      "num_input_tokens_seen": 18183176,
      "step": 27850
    },
    {
      "epoch": 16.423938679245282,
      "grad_norm": 1.9218395948410034,
      "learning_rate": 9.429719987402541e-07,
      "loss": 0.2629,
      "num_input_tokens_seen": 18186920,
      "step": 27855
    },
    {
      "epoch": 16.42688679245283,
      "grad_norm": 3.80049729347229,
      "learning_rate": 9.414688269388883e-07,
      "loss": 0.3354,
      "num_input_tokens_seen": 18190248,
      "step": 27860
    },
    {
      "epoch": 16.42983490566038,
      "grad_norm": 6.680559158325195,
      "learning_rate": 9.3996672965114e-07,
      "loss": 0.264,
      "num_input_tokens_seen": 18193320,
      "step": 27865
    },
    {
      "epoch": 16.432783018867923,
      "grad_norm": 3.0099081993103027,
      "learning_rate": 9.384657072747e-07,
      "loss": 0.2176,
      "num_input_tokens_seen": 18196776,
      "step": 27870
    },
    {
      "epoch": 16.43573113207547,
      "grad_norm": 4.875786781311035,
      "learning_rate": 9.369657602069676e-07,
      "loss": 0.3374,
      "num_input_tokens_seen": 18199496,
      "step": 27875
    },
    {
      "epoch": 16.43867924528302,
      "grad_norm": 3.948143243789673,
      "learning_rate": 9.354668888450608e-07,
      "loss": 0.371,
      "num_input_tokens_seen": 18201800,
      "step": 27880
    },
    {
      "epoch": 16.441627358490567,
      "grad_norm": 2.5716280937194824,
      "learning_rate": 9.339690935858125e-07,
      "loss": 0.3468,
      "num_input_tokens_seen": 18204648,
      "step": 27885
    },
    {
      "epoch": 16.444575471698112,
      "grad_norm": 4.36135721206665,
      "learning_rate": 9.324723748257697e-07,
      "loss": 0.3576,
      "num_input_tokens_seen": 18207528,
      "step": 27890
    },
    {
      "epoch": 16.44752358490566,
      "grad_norm": 3.9799201488494873,
      "learning_rate": 9.309767329611963e-07,
      "loss": 0.3242,
      "num_input_tokens_seen": 18210568,
      "step": 27895
    },
    {
      "epoch": 16.45047169811321,
      "grad_norm": 2.681002616882324,
      "learning_rate": 9.294821683880695e-07,
      "loss": 0.2337,
      "num_input_tokens_seen": 18213544,
      "step": 27900
    },
    {
      "epoch": 16.453419811320753,
      "grad_norm": 2.5427801609039307,
      "learning_rate": 9.279886815020816e-07,
      "loss": 0.2543,
      "num_input_tokens_seen": 18217352,
      "step": 27905
    },
    {
      "epoch": 16.4563679245283,
      "grad_norm": 3.7799453735351562,
      "learning_rate": 9.264962726986393e-07,
      "loss": 0.3855,
      "num_input_tokens_seen": 18220456,
      "step": 27910
    },
    {
      "epoch": 16.45931603773585,
      "grad_norm": 1.9801958799362183,
      "learning_rate": 9.250049423728652e-07,
      "loss": 0.2477,
      "num_input_tokens_seen": 18223880,
      "step": 27915
    },
    {
      "epoch": 16.462264150943398,
      "grad_norm": 4.047691345214844,
      "learning_rate": 9.235146909195936e-07,
      "loss": 0.3087,
      "num_input_tokens_seen": 18226856,
      "step": 27920
    },
    {
      "epoch": 16.465212264150942,
      "grad_norm": 4.105132579803467,
      "learning_rate": 9.220255187333771e-07,
      "loss": 0.2958,
      "num_input_tokens_seen": 18230248,
      "step": 27925
    },
    {
      "epoch": 16.46816037735849,
      "grad_norm": 7.007116317749023,
      "learning_rate": 9.205374262084798e-07,
      "loss": 0.3608,
      "num_input_tokens_seen": 18232136,
      "step": 27930
    },
    {
      "epoch": 16.47110849056604,
      "grad_norm": 3.514346122741699,
      "learning_rate": 9.190504137388806e-07,
      "loss": 0.4124,
      "num_input_tokens_seen": 18236552,
      "step": 27935
    },
    {
      "epoch": 16.474056603773583,
      "grad_norm": 4.259034633636475,
      "learning_rate": 9.175644817182722e-07,
      "loss": 0.3101,
      "num_input_tokens_seen": 18239240,
      "step": 27940
    },
    {
      "epoch": 16.47700471698113,
      "grad_norm": 2.2055740356445312,
      "learning_rate": 9.16079630540061e-07,
      "loss": 0.3018,
      "num_input_tokens_seen": 18242088,
      "step": 27945
    },
    {
      "epoch": 16.47995283018868,
      "grad_norm": 3.444841146469116,
      "learning_rate": 9.145958605973676e-07,
      "loss": 0.3046,
      "num_input_tokens_seen": 18244936,
      "step": 27950
    },
    {
      "epoch": 16.482900943396228,
      "grad_norm": 4.489775657653809,
      "learning_rate": 9.131131722830289e-07,
      "loss": 0.3508,
      "num_input_tokens_seen": 18248520,
      "step": 27955
    },
    {
      "epoch": 16.485849056603772,
      "grad_norm": 4.1201491355896,
      "learning_rate": 9.116315659895892e-07,
      "loss": 0.387,
      "num_input_tokens_seen": 18251752,
      "step": 27960
    },
    {
      "epoch": 16.48879716981132,
      "grad_norm": 3.685603618621826,
      "learning_rate": 9.10151042109314e-07,
      "loss": 0.2943,
      "num_input_tokens_seen": 18255144,
      "step": 27965
    },
    {
      "epoch": 16.49174528301887,
      "grad_norm": 4.379791736602783,
      "learning_rate": 9.086716010341767e-07,
      "loss": 0.3224,
      "num_input_tokens_seen": 18258376,
      "step": 27970
    },
    {
      "epoch": 16.494693396226417,
      "grad_norm": 3.7092583179473877,
      "learning_rate": 9.071932431558655e-07,
      "loss": 0.23,
      "num_input_tokens_seen": 18261128,
      "step": 27975
    },
    {
      "epoch": 16.49764150943396,
      "grad_norm": 2.9440362453460693,
      "learning_rate": 9.057159688657824e-07,
      "loss": 0.2526,
      "num_input_tokens_seen": 18264360,
      "step": 27980
    },
    {
      "epoch": 16.50058962264151,
      "grad_norm": 3.211334466934204,
      "learning_rate": 9.042397785550405e-07,
      "loss": 0.2936,
      "num_input_tokens_seen": 18267528,
      "step": 27985
    },
    {
      "epoch": 16.503537735849058,
      "grad_norm": 2.7018685340881348,
      "learning_rate": 9.027646726144707e-07,
      "loss": 0.299,
      "num_input_tokens_seen": 18270952,
      "step": 27990
    },
    {
      "epoch": 16.506485849056602,
      "grad_norm": 5.471151351928711,
      "learning_rate": 9.012906514346115e-07,
      "loss": 0.2675,
      "num_input_tokens_seen": 18274472,
      "step": 27995
    },
    {
      "epoch": 16.50943396226415,
      "grad_norm": 3.833176612854004,
      "learning_rate": 8.99817715405717e-07,
      "loss": 0.2104,
      "num_input_tokens_seen": 18278568,
      "step": 28000
    },
    {
      "epoch": 16.5123820754717,
      "grad_norm": 4.792274475097656,
      "learning_rate": 8.983458649177529e-07,
      "loss": 0.226,
      "num_input_tokens_seen": 18280936,
      "step": 28005
    },
    {
      "epoch": 16.515330188679247,
      "grad_norm": 3.603153944015503,
      "learning_rate": 8.968751003603982e-07,
      "loss": 0.3768,
      "num_input_tokens_seen": 18283944,
      "step": 28010
    },
    {
      "epoch": 16.51827830188679,
      "grad_norm": 9.184059143066406,
      "learning_rate": 8.95405422123043e-07,
      "loss": 0.3887,
      "num_input_tokens_seen": 18287144,
      "step": 28015
    },
    {
      "epoch": 16.52122641509434,
      "grad_norm": 2.5323679447174072,
      "learning_rate": 8.939368305947932e-07,
      "loss": 0.4066,
      "num_input_tokens_seen": 18291080,
      "step": 28020
    },
    {
      "epoch": 16.524174528301888,
      "grad_norm": 3.2594141960144043,
      "learning_rate": 8.92469326164464e-07,
      "loss": 0.281,
      "num_input_tokens_seen": 18293768,
      "step": 28025
    },
    {
      "epoch": 16.527122641509433,
      "grad_norm": 3.4369442462921143,
      "learning_rate": 8.910029092205829e-07,
      "loss": 0.1902,
      "num_input_tokens_seen": 18296040,
      "step": 28030
    },
    {
      "epoch": 16.53007075471698,
      "grad_norm": 2.4512956142425537,
      "learning_rate": 8.895375801513906e-07,
      "loss": 0.1956,
      "num_input_tokens_seen": 18300552,
      "step": 28035
    },
    {
      "epoch": 16.53301886792453,
      "grad_norm": 2.2350974082946777,
      "learning_rate": 8.880733393448377e-07,
      "loss": 0.266,
      "num_input_tokens_seen": 18304040,
      "step": 28040
    },
    {
      "epoch": 16.535966981132077,
      "grad_norm": 3.849118232727051,
      "learning_rate": 8.866101871885907e-07,
      "loss": 0.3377,
      "num_input_tokens_seen": 18307016,
      "step": 28045
    },
    {
      "epoch": 16.53891509433962,
      "grad_norm": 3.746448040008545,
      "learning_rate": 8.851481240700249e-07,
      "loss": 0.3493,
      "num_input_tokens_seen": 18310344,
      "step": 28050
    },
    {
      "epoch": 16.54186320754717,
      "grad_norm": 3.6305148601531982,
      "learning_rate": 8.836871503762257e-07,
      "loss": 0.2163,
      "num_input_tokens_seen": 18313704,
      "step": 28055
    },
    {
      "epoch": 16.544811320754718,
      "grad_norm": 5.9658331871032715,
      "learning_rate": 8.822272664939946e-07,
      "loss": 0.297,
      "num_input_tokens_seen": 18317032,
      "step": 28060
    },
    {
      "epoch": 16.547759433962263,
      "grad_norm": 5.061834812164307,
      "learning_rate": 8.80768472809842e-07,
      "loss": 0.3069,
      "num_input_tokens_seen": 18320008,
      "step": 28065
    },
    {
      "epoch": 16.55070754716981,
      "grad_norm": 4.820870876312256,
      "learning_rate": 8.793107697099884e-07,
      "loss": 0.5205,
      "num_input_tokens_seen": 18322888,
      "step": 28070
    },
    {
      "epoch": 16.55365566037736,
      "grad_norm": 5.755917072296143,
      "learning_rate": 8.778541575803673e-07,
      "loss": 0.3003,
      "num_input_tokens_seen": 18325608,
      "step": 28075
    },
    {
      "epoch": 16.556603773584907,
      "grad_norm": 2.8354055881500244,
      "learning_rate": 8.763986368066241e-07,
      "loss": 0.2158,
      "num_input_tokens_seen": 18328616,
      "step": 28080
    },
    {
      "epoch": 16.559551886792452,
      "grad_norm": 3.812364339828491,
      "learning_rate": 8.749442077741138e-07,
      "loss": 0.2683,
      "num_input_tokens_seen": 18331496,
      "step": 28085
    },
    {
      "epoch": 16.5625,
      "grad_norm": 3.0752851963043213,
      "learning_rate": 8.734908708679024e-07,
      "loss": 0.4226,
      "num_input_tokens_seen": 18334440,
      "step": 28090
    },
    {
      "epoch": 16.565448113207548,
      "grad_norm": 8.289811134338379,
      "learning_rate": 8.72038626472767e-07,
      "loss": 0.3538,
      "num_input_tokens_seen": 18337384,
      "step": 28095
    },
    {
      "epoch": 16.568396226415093,
      "grad_norm": 3.749800205230713,
      "learning_rate": 8.705874749731962e-07,
      "loss": 0.3107,
      "num_input_tokens_seen": 18340520,
      "step": 28100
    },
    {
      "epoch": 16.57134433962264,
      "grad_norm": 3.3201005458831787,
      "learning_rate": 8.691374167533867e-07,
      "loss": 0.3541,
      "num_input_tokens_seen": 18343656,
      "step": 28105
    },
    {
      "epoch": 16.57429245283019,
      "grad_norm": 7.969613075256348,
      "learning_rate": 8.6768845219725e-07,
      "loss": 0.2594,
      "num_input_tokens_seen": 18347176,
      "step": 28110
    },
    {
      "epoch": 16.577240566037737,
      "grad_norm": 5.6708292961120605,
      "learning_rate": 8.662405816884056e-07,
      "loss": 0.3494,
      "num_input_tokens_seen": 18350792,
      "step": 28115
    },
    {
      "epoch": 16.580188679245282,
      "grad_norm": 2.214641571044922,
      "learning_rate": 8.647938056101824e-07,
      "loss": 0.1981,
      "num_input_tokens_seen": 18354408,
      "step": 28120
    },
    {
      "epoch": 16.58313679245283,
      "grad_norm": 3.3968143463134766,
      "learning_rate": 8.63348124345621e-07,
      "loss": 0.3991,
      "num_input_tokens_seen": 18357352,
      "step": 28125
    },
    {
      "epoch": 16.58608490566038,
      "grad_norm": 2.846489191055298,
      "learning_rate": 8.619035382774716e-07,
      "loss": 0.2869,
      "num_input_tokens_seen": 18360136,
      "step": 28130
    },
    {
      "epoch": 16.589033018867923,
      "grad_norm": 3.7617170810699463,
      "learning_rate": 8.60460047788193e-07,
      "loss": 0.2447,
      "num_input_tokens_seen": 18363080,
      "step": 28135
    },
    {
      "epoch": 16.59198113207547,
      "grad_norm": 3.938455820083618,
      "learning_rate": 8.590176532599587e-07,
      "loss": 0.2602,
      "num_input_tokens_seen": 18367528,
      "step": 28140
    },
    {
      "epoch": 16.59492924528302,
      "grad_norm": 4.964748859405518,
      "learning_rate": 8.575763550746475e-07,
      "loss": 0.276,
      "num_input_tokens_seen": 18370856,
      "step": 28145
    },
    {
      "epoch": 16.597877358490567,
      "grad_norm": 4.684011936187744,
      "learning_rate": 8.56136153613848e-07,
      "loss": 0.4144,
      "num_input_tokens_seen": 18373864,
      "step": 28150
    },
    {
      "epoch": 16.600825471698112,
      "grad_norm": 3.855015277862549,
      "learning_rate": 8.546970492588619e-07,
      "loss": 0.2785,
      "num_input_tokens_seen": 18377544,
      "step": 28155
    },
    {
      "epoch": 16.60377358490566,
      "grad_norm": 3.310488224029541,
      "learning_rate": 8.532590423906973e-07,
      "loss": 0.3432,
      "num_input_tokens_seen": 18380680,
      "step": 28160
    },
    {
      "epoch": 16.60672169811321,
      "grad_norm": 3.829721689224243,
      "learning_rate": 8.518221333900728e-07,
      "loss": 0.2575,
      "num_input_tokens_seen": 18383560,
      "step": 28165
    },
    {
      "epoch": 16.609669811320753,
      "grad_norm": 3.348219156265259,
      "learning_rate": 8.503863226374148e-07,
      "loss": 0.3153,
      "num_input_tokens_seen": 18386312,
      "step": 28170
    },
    {
      "epoch": 16.6126179245283,
      "grad_norm": 4.839154243469238,
      "learning_rate": 8.489516105128632e-07,
      "loss": 0.263,
      "num_input_tokens_seen": 18389768,
      "step": 28175
    },
    {
      "epoch": 16.61556603773585,
      "grad_norm": 6.341581344604492,
      "learning_rate": 8.475179973962621e-07,
      "loss": 0.3698,
      "num_input_tokens_seen": 18393096,
      "step": 28180
    },
    {
      "epoch": 16.618514150943398,
      "grad_norm": 3.94744610786438,
      "learning_rate": 8.460854836671678e-07,
      "loss": 0.2733,
      "num_input_tokens_seen": 18396136,
      "step": 28185
    },
    {
      "epoch": 16.621462264150942,
      "grad_norm": 5.274620532989502,
      "learning_rate": 8.446540697048445e-07,
      "loss": 0.2841,
      "num_input_tokens_seen": 18399016,
      "step": 28190
    },
    {
      "epoch": 16.62441037735849,
      "grad_norm": 4.905893802642822,
      "learning_rate": 8.432237558882639e-07,
      "loss": 0.3294,
      "num_input_tokens_seen": 18402792,
      "step": 28195
    },
    {
      "epoch": 16.62735849056604,
      "grad_norm": 5.490345478057861,
      "learning_rate": 8.417945425961083e-07,
      "loss": 0.2806,
      "num_input_tokens_seen": 18405736,
      "step": 28200
    },
    {
      "epoch": 16.630306603773583,
      "grad_norm": 4.3853044509887695,
      "learning_rate": 8.403664302067688e-07,
      "loss": 0.383,
      "num_input_tokens_seen": 18408968,
      "step": 28205
    },
    {
      "epoch": 16.63325471698113,
      "grad_norm": 3.4984960556030273,
      "learning_rate": 8.389394190983446e-07,
      "loss": 0.2602,
      "num_input_tokens_seen": 18412200,
      "step": 28210
    },
    {
      "epoch": 16.63620283018868,
      "grad_norm": 3.7801666259765625,
      "learning_rate": 8.37513509648642e-07,
      "loss": 0.2409,
      "num_input_tokens_seen": 18415336,
      "step": 28215
    },
    {
      "epoch": 16.639150943396228,
      "grad_norm": 5.987263202667236,
      "learning_rate": 8.360887022351771e-07,
      "loss": 0.4233,
      "num_input_tokens_seen": 18418728,
      "step": 28220
    },
    {
      "epoch": 16.642099056603772,
      "grad_norm": 2.0957329273223877,
      "learning_rate": 8.346649972351739e-07,
      "loss": 0.3023,
      "num_input_tokens_seen": 18425896,
      "step": 28225
    },
    {
      "epoch": 16.64504716981132,
      "grad_norm": 3.4638280868530273,
      "learning_rate": 8.33242395025563e-07,
      "loss": 0.3672,
      "num_input_tokens_seen": 18428552,
      "step": 28230
    },
    {
      "epoch": 16.64799528301887,
      "grad_norm": 3.376800298690796,
      "learning_rate": 8.318208959829871e-07,
      "loss": 0.3192,
      "num_input_tokens_seen": 18431176,
      "step": 28235
    },
    {
      "epoch": 16.650943396226417,
      "grad_norm": 4.078437328338623,
      "learning_rate": 8.304005004837929e-07,
      "loss": 0.3292,
      "num_input_tokens_seen": 18434760,
      "step": 28240
    },
    {
      "epoch": 16.65389150943396,
      "grad_norm": 2.1647026538848877,
      "learning_rate": 8.289812089040344e-07,
      "loss": 0.3402,
      "num_input_tokens_seen": 18437896,
      "step": 28245
    },
    {
      "epoch": 16.65683962264151,
      "grad_norm": 3.9181089401245117,
      "learning_rate": 8.275630216194785e-07,
      "loss": 0.291,
      "num_input_tokens_seen": 18441064,
      "step": 28250
    },
    {
      "epoch": 16.659787735849058,
      "grad_norm": 2.9407787322998047,
      "learning_rate": 8.261459390055948e-07,
      "loss": 0.3331,
      "num_input_tokens_seen": 18443560,
      "step": 28255
    },
    {
      "epoch": 16.662735849056602,
      "grad_norm": 5.153355598449707,
      "learning_rate": 8.24729961437562e-07,
      "loss": 0.3333,
      "num_input_tokens_seen": 18446632,
      "step": 28260
    },
    {
      "epoch": 16.66568396226415,
      "grad_norm": 2.584152936935425,
      "learning_rate": 8.233150892902653e-07,
      "loss": 0.3638,
      "num_input_tokens_seen": 18450088,
      "step": 28265
    },
    {
      "epoch": 16.6686320754717,
      "grad_norm": 2.558051347732544,
      "learning_rate": 8.219013229383005e-07,
      "loss": 0.2796,
      "num_input_tokens_seen": 18454152,
      "step": 28270
    },
    {
      "epoch": 16.671580188679247,
      "grad_norm": 4.4236555099487305,
      "learning_rate": 8.204886627559666e-07,
      "loss": 0.3369,
      "num_input_tokens_seen": 18456840,
      "step": 28275
    },
    {
      "epoch": 16.67452830188679,
      "grad_norm": 2.3827240467071533,
      "learning_rate": 8.190771091172722e-07,
      "loss": 0.2469,
      "num_input_tokens_seen": 18460168,
      "step": 28280
    },
    {
      "epoch": 16.67747641509434,
      "grad_norm": 7.2898406982421875,
      "learning_rate": 8.176666623959323e-07,
      "loss": 0.5131,
      "num_input_tokens_seen": 18463464,
      "step": 28285
    },
    {
      "epoch": 16.680424528301888,
      "grad_norm": 5.622857570648193,
      "learning_rate": 8.162573229653681e-07,
      "loss": 0.2696,
      "num_input_tokens_seen": 18466952,
      "step": 28290
    },
    {
      "epoch": 16.683372641509433,
      "grad_norm": 4.583674907684326,
      "learning_rate": 8.148490911987073e-07,
      "loss": 0.2215,
      "num_input_tokens_seen": 18469448,
      "step": 28295
    },
    {
      "epoch": 16.68632075471698,
      "grad_norm": 5.918069839477539,
      "learning_rate": 8.134419674687876e-07,
      "loss": 0.4376,
      "num_input_tokens_seen": 18473000,
      "step": 28300
    },
    {
      "epoch": 16.68926886792453,
      "grad_norm": 4.1532511711120605,
      "learning_rate": 8.120359521481502e-07,
      "loss": 0.3218,
      "num_input_tokens_seen": 18475848,
      "step": 28305
    },
    {
      "epoch": 16.692216981132077,
      "grad_norm": 3.7391672134399414,
      "learning_rate": 8.106310456090438e-07,
      "loss": 0.2893,
      "num_input_tokens_seen": 18479848,
      "step": 28310
    },
    {
      "epoch": 16.69516509433962,
      "grad_norm": 3.1562418937683105,
      "learning_rate": 8.092272482234231e-07,
      "loss": 0.3681,
      "num_input_tokens_seen": 18484264,
      "step": 28315
    },
    {
      "epoch": 16.69811320754717,
      "grad_norm": 3.536468982696533,
      "learning_rate": 8.078245603629486e-07,
      "loss": 0.2736,
      "num_input_tokens_seen": 18487080,
      "step": 28320
    },
    {
      "epoch": 16.701061320754718,
      "grad_norm": 2.602874517440796,
      "learning_rate": 8.0642298239899e-07,
      "loss": 0.2139,
      "num_input_tokens_seen": 18490248,
      "step": 28325
    },
    {
      "epoch": 16.704009433962263,
      "grad_norm": 2.5799262523651123,
      "learning_rate": 8.050225147026202e-07,
      "loss": 0.3513,
      "num_input_tokens_seen": 18494504,
      "step": 28330
    },
    {
      "epoch": 16.70695754716981,
      "grad_norm": 7.087551593780518,
      "learning_rate": 8.03623157644619e-07,
      "loss": 0.4131,
      "num_input_tokens_seen": 18498120,
      "step": 28335
    },
    {
      "epoch": 16.70990566037736,
      "grad_norm": 5.125203609466553,
      "learning_rate": 8.022249115954728e-07,
      "loss": 0.2537,
      "num_input_tokens_seen": 18501768,
      "step": 28340
    },
    {
      "epoch": 16.712853773584907,
      "grad_norm": 3.553215980529785,
      "learning_rate": 8.008277769253709e-07,
      "loss": 0.2565,
      "num_input_tokens_seen": 18504136,
      "step": 28345
    },
    {
      "epoch": 16.715801886792452,
      "grad_norm": 5.535646438598633,
      "learning_rate": 7.994317540042135e-07,
      "loss": 0.3364,
      "num_input_tokens_seen": 18507208,
      "step": 28350
    },
    {
      "epoch": 16.71875,
      "grad_norm": 3.334515333175659,
      "learning_rate": 7.980368432016017e-07,
      "loss": 0.1977,
      "num_input_tokens_seen": 18510472,
      "step": 28355
    },
    {
      "epoch": 16.721698113207548,
      "grad_norm": 4.659951686859131,
      "learning_rate": 7.966430448868461e-07,
      "loss": 0.313,
      "num_input_tokens_seen": 18513064,
      "step": 28360
    },
    {
      "epoch": 16.724646226415093,
      "grad_norm": 3.0508806705474854,
      "learning_rate": 7.952503594289601e-07,
      "loss": 0.3984,
      "num_input_tokens_seen": 18515624,
      "step": 28365
    },
    {
      "epoch": 16.72759433962264,
      "grad_norm": 6.148863315582275,
      "learning_rate": 7.93858787196663e-07,
      "loss": 0.3171,
      "num_input_tokens_seen": 18518888,
      "step": 28370
    },
    {
      "epoch": 16.73054245283019,
      "grad_norm": 7.815550327301025,
      "learning_rate": 7.92468328558379e-07,
      "loss": 0.3995,
      "num_input_tokens_seen": 18521416,
      "step": 28375
    },
    {
      "epoch": 16.733490566037737,
      "grad_norm": 4.067500591278076,
      "learning_rate": 7.910789838822386e-07,
      "loss": 0.392,
      "num_input_tokens_seen": 18524712,
      "step": 28380
    },
    {
      "epoch": 16.736438679245282,
      "grad_norm": 5.509260177612305,
      "learning_rate": 7.89690753536076e-07,
      "loss": 0.4172,
      "num_input_tokens_seen": 18527560,
      "step": 28385
    },
    {
      "epoch": 16.73938679245283,
      "grad_norm": 4.308719635009766,
      "learning_rate": 7.883036378874326e-07,
      "loss": 0.3907,
      "num_input_tokens_seen": 18530696,
      "step": 28390
    },
    {
      "epoch": 16.74233490566038,
      "grad_norm": 3.7203173637390137,
      "learning_rate": 7.86917637303552e-07,
      "loss": 0.2807,
      "num_input_tokens_seen": 18534280,
      "step": 28395
    },
    {
      "epoch": 16.745283018867923,
      "grad_norm": 3.3045477867126465,
      "learning_rate": 7.855327521513851e-07,
      "loss": 0.3088,
      "num_input_tokens_seen": 18536648,
      "step": 28400
    },
    {
      "epoch": 16.74823113207547,
      "grad_norm": 3.9346742630004883,
      "learning_rate": 7.841489827975851e-07,
      "loss": 0.2986,
      "num_input_tokens_seen": 18539496,
      "step": 28405
    },
    {
      "epoch": 16.75117924528302,
      "grad_norm": 6.151522636413574,
      "learning_rate": 7.827663296085109e-07,
      "loss": 0.3336,
      "num_input_tokens_seen": 18541896,
      "step": 28410
    },
    {
      "epoch": 16.754127358490567,
      "grad_norm": 4.613429069519043,
      "learning_rate": 7.813847929502255e-07,
      "loss": 0.336,
      "num_input_tokens_seen": 18544840,
      "step": 28415
    },
    {
      "epoch": 16.757075471698112,
      "grad_norm": 1.8258585929870605,
      "learning_rate": 7.800043731884982e-07,
      "loss": 0.2786,
      "num_input_tokens_seen": 18548808,
      "step": 28420
    },
    {
      "epoch": 16.76002358490566,
      "grad_norm": 2.6319079399108887,
      "learning_rate": 7.786250706888005e-07,
      "loss": 0.3584,
      "num_input_tokens_seen": 18552040,
      "step": 28425
    },
    {
      "epoch": 16.76297169811321,
      "grad_norm": 2.326843500137329,
      "learning_rate": 7.772468858163085e-07,
      "loss": 0.2194,
      "num_input_tokens_seen": 18555176,
      "step": 28430
    },
    {
      "epoch": 16.765919811320753,
      "grad_norm": 4.060580730438232,
      "learning_rate": 7.758698189359026e-07,
      "loss": 0.2222,
      "num_input_tokens_seen": 18561032,
      "step": 28435
    },
    {
      "epoch": 16.7688679245283,
      "grad_norm": 4.190971374511719,
      "learning_rate": 7.744938704121658e-07,
      "loss": 0.2306,
      "num_input_tokens_seen": 18563848,
      "step": 28440
    },
    {
      "epoch": 16.77181603773585,
      "grad_norm": 2.9407505989074707,
      "learning_rate": 7.731190406093892e-07,
      "loss": 0.2519,
      "num_input_tokens_seen": 18567144,
      "step": 28445
    },
    {
      "epoch": 16.774764150943398,
      "grad_norm": 3.611067533493042,
      "learning_rate": 7.717453298915617e-07,
      "loss": 0.526,
      "num_input_tokens_seen": 18569768,
      "step": 28450
    },
    {
      "epoch": 16.777712264150942,
      "grad_norm": 3.9767723083496094,
      "learning_rate": 7.703727386223825e-07,
      "loss": 0.2791,
      "num_input_tokens_seen": 18572328,
      "step": 28455
    },
    {
      "epoch": 16.78066037735849,
      "grad_norm": 5.371555805206299,
      "learning_rate": 7.690012671652491e-07,
      "loss": 0.3235,
      "num_input_tokens_seen": 18575272,
      "step": 28460
    },
    {
      "epoch": 16.78360849056604,
      "grad_norm": 3.3217289447784424,
      "learning_rate": 7.676309158832651e-07,
      "loss": 0.3406,
      "num_input_tokens_seen": 18578152,
      "step": 28465
    },
    {
      "epoch": 16.786556603773583,
      "grad_norm": 4.040096282958984,
      "learning_rate": 7.662616851392362e-07,
      "loss": 0.3584,
      "num_input_tokens_seen": 18581608,
      "step": 28470
    },
    {
      "epoch": 16.78950471698113,
      "grad_norm": 7.39586877822876,
      "learning_rate": 7.648935752956732e-07,
      "loss": 0.5609,
      "num_input_tokens_seen": 18585512,
      "step": 28475
    },
    {
      "epoch": 16.79245283018868,
      "grad_norm": 2.883486270904541,
      "learning_rate": 7.635265867147867e-07,
      "loss": 0.3372,
      "num_input_tokens_seen": 18587912,
      "step": 28480
    },
    {
      "epoch": 16.795400943396228,
      "grad_norm": 3.3772594928741455,
      "learning_rate": 7.621607197584963e-07,
      "loss": 0.3827,
      "num_input_tokens_seen": 18591464,
      "step": 28485
    },
    {
      "epoch": 16.798349056603772,
      "grad_norm": 3.912789821624756,
      "learning_rate": 7.607959747884186e-07,
      "loss": 0.454,
      "num_input_tokens_seen": 18594312,
      "step": 28490
    },
    {
      "epoch": 16.80129716981132,
      "grad_norm": 4.989803314208984,
      "learning_rate": 7.594323521658769e-07,
      "loss": 0.3148,
      "num_input_tokens_seen": 18597256,
      "step": 28495
    },
    {
      "epoch": 16.80424528301887,
      "grad_norm": 3.8708603382110596,
      "learning_rate": 7.580698522518958e-07,
      "loss": 0.3437,
      "num_input_tokens_seen": 18600168,
      "step": 28500
    },
    {
      "epoch": 16.807193396226417,
      "grad_norm": 4.615939140319824,
      "learning_rate": 7.567084754072035e-07,
      "loss": 0.2778,
      "num_input_tokens_seen": 18603304,
      "step": 28505
    },
    {
      "epoch": 16.81014150943396,
      "grad_norm": 6.80972146987915,
      "learning_rate": 7.553482219922282e-07,
      "loss": 0.2765,
      "num_input_tokens_seen": 18606312,
      "step": 28510
    },
    {
      "epoch": 16.81308962264151,
      "grad_norm": 2.9495906829833984,
      "learning_rate": 7.539890923671061e-07,
      "loss": 0.2633,
      "num_input_tokens_seen": 18609576,
      "step": 28515
    },
    {
      "epoch": 16.816037735849058,
      "grad_norm": 5.903151988983154,
      "learning_rate": 7.526310868916708e-07,
      "loss": 0.2989,
      "num_input_tokens_seen": 18612904,
      "step": 28520
    },
    {
      "epoch": 16.818985849056602,
      "grad_norm": 2.3333492279052734,
      "learning_rate": 7.512742059254602e-07,
      "loss": 0.2481,
      "num_input_tokens_seen": 18616616,
      "step": 28525
    },
    {
      "epoch": 16.82193396226415,
      "grad_norm": 3.5993716716766357,
      "learning_rate": 7.499184498277151e-07,
      "loss": 0.2846,
      "num_input_tokens_seen": 18620008,
      "step": 28530
    },
    {
      "epoch": 16.8248820754717,
      "grad_norm": 1.9620471000671387,
      "learning_rate": 7.485638189573758e-07,
      "loss": 0.2309,
      "num_input_tokens_seen": 18622984,
      "step": 28535
    },
    {
      "epoch": 16.827830188679247,
      "grad_norm": 3.4604578018188477,
      "learning_rate": 7.472103136730891e-07,
      "loss": 0.3069,
      "num_input_tokens_seen": 18626024,
      "step": 28540
    },
    {
      "epoch": 16.83077830188679,
      "grad_norm": 4.1467084884643555,
      "learning_rate": 7.458579343331996e-07,
      "loss": 0.4178,
      "num_input_tokens_seen": 18629576,
      "step": 28545
    },
    {
      "epoch": 16.83372641509434,
      "grad_norm": 4.977681636810303,
      "learning_rate": 7.445066812957569e-07,
      "loss": 0.26,
      "num_input_tokens_seen": 18632744,
      "step": 28550
    },
    {
      "epoch": 16.836674528301888,
      "grad_norm": 3.6723358631134033,
      "learning_rate": 7.43156554918511e-07,
      "loss": 0.3094,
      "num_input_tokens_seen": 18635784,
      "step": 28555
    },
    {
      "epoch": 16.839622641509433,
      "grad_norm": 5.035487651824951,
      "learning_rate": 7.418075555589132e-07,
      "loss": 0.3556,
      "num_input_tokens_seen": 18639464,
      "step": 28560
    },
    {
      "epoch": 16.84257075471698,
      "grad_norm": 4.815647602081299,
      "learning_rate": 7.404596835741168e-07,
      "loss": 0.4945,
      "num_input_tokens_seen": 18642760,
      "step": 28565
    },
    {
      "epoch": 16.84551886792453,
      "grad_norm": 12.751043319702148,
      "learning_rate": 7.391129393209751e-07,
      "loss": 0.3026,
      "num_input_tokens_seen": 18646536,
      "step": 28570
    },
    {
      "epoch": 16.848466981132077,
      "grad_norm": 4.579164981842041,
      "learning_rate": 7.377673231560478e-07,
      "loss": 0.2912,
      "num_input_tokens_seen": 18649704,
      "step": 28575
    },
    {
      "epoch": 16.85141509433962,
      "grad_norm": 4.935809135437012,
      "learning_rate": 7.364228354355907e-07,
      "loss": 0.3859,
      "num_input_tokens_seen": 18652424,
      "step": 28580
    },
    {
      "epoch": 16.85436320754717,
      "grad_norm": 4.620152473449707,
      "learning_rate": 7.350794765155627e-07,
      "loss": 0.3358,
      "num_input_tokens_seen": 18655368,
      "step": 28585
    },
    {
      "epoch": 16.857311320754718,
      "grad_norm": 8.465409278869629,
      "learning_rate": 7.337372467516246e-07,
      "loss": 0.3491,
      "num_input_tokens_seen": 18658344,
      "step": 28590
    },
    {
      "epoch": 16.860259433962263,
      "grad_norm": 2.5904970169067383,
      "learning_rate": 7.323961464991369e-07,
      "loss": 0.3281,
      "num_input_tokens_seen": 18661448,
      "step": 28595
    },
    {
      "epoch": 16.86320754716981,
      "grad_norm": 2.090069532394409,
      "learning_rate": 7.310561761131601e-07,
      "loss": 0.2802,
      "num_input_tokens_seen": 18664648,
      "step": 28600
    },
    {
      "epoch": 16.86615566037736,
      "grad_norm": 5.940135955810547,
      "learning_rate": 7.297173359484605e-07,
      "loss": 0.3634,
      "num_input_tokens_seen": 18667400,
      "step": 28605
    },
    {
      "epoch": 16.869103773584907,
      "grad_norm": 3.509368896484375,
      "learning_rate": 7.283796263595e-07,
      "loss": 0.2503,
      "num_input_tokens_seen": 18670440,
      "step": 28610
    },
    {
      "epoch": 16.872051886792452,
      "grad_norm": 4.251274585723877,
      "learning_rate": 7.270430477004431e-07,
      "loss": 0.522,
      "num_input_tokens_seen": 18673032,
      "step": 28615
    },
    {
      "epoch": 16.875,
      "grad_norm": 4.0522894859313965,
      "learning_rate": 7.257076003251545e-07,
      "loss": 0.447,
      "num_input_tokens_seen": 18677192,
      "step": 28620
    },
    {
      "epoch": 16.877948113207548,
      "grad_norm": 2.482048511505127,
      "learning_rate": 7.243732845871998e-07,
      "loss": 0.2254,
      "num_input_tokens_seen": 18682280,
      "step": 28625
    },
    {
      "epoch": 16.880896226415093,
      "grad_norm": 2.4756174087524414,
      "learning_rate": 7.230401008398441e-07,
      "loss": 0.3509,
      "num_input_tokens_seen": 18685736,
      "step": 28630
    },
    {
      "epoch": 16.88384433962264,
      "grad_norm": 2.6911680698394775,
      "learning_rate": 7.217080494360546e-07,
      "loss": 0.334,
      "num_input_tokens_seen": 18689032,
      "step": 28635
    },
    {
      "epoch": 16.88679245283019,
      "grad_norm": 3.798809051513672,
      "learning_rate": 7.20377130728498e-07,
      "loss": 0.3288,
      "num_input_tokens_seen": 18692552,
      "step": 28640
    },
    {
      "epoch": 16.889740566037737,
      "grad_norm": 4.084588050842285,
      "learning_rate": 7.190473450695407e-07,
      "loss": 0.28,
      "num_input_tokens_seen": 18695912,
      "step": 28645
    },
    {
      "epoch": 16.892688679245282,
      "grad_norm": 2.770679473876953,
      "learning_rate": 7.177186928112484e-07,
      "loss": 0.289,
      "num_input_tokens_seen": 18698728,
      "step": 28650
    },
    {
      "epoch": 16.89563679245283,
      "grad_norm": 3.297852039337158,
      "learning_rate": 7.163911743053876e-07,
      "loss": 0.3824,
      "num_input_tokens_seen": 18701864,
      "step": 28655
    },
    {
      "epoch": 16.89858490566038,
      "grad_norm": 5.296316146850586,
      "learning_rate": 7.150647899034252e-07,
      "loss": 0.2998,
      "num_input_tokens_seen": 18704424,
      "step": 28660
    },
    {
      "epoch": 16.901533018867923,
      "grad_norm": 4.755214214324951,
      "learning_rate": 7.13739539956525e-07,
      "loss": 0.3218,
      "num_input_tokens_seen": 18707720,
      "step": 28665
    },
    {
      "epoch": 16.90448113207547,
      "grad_norm": 4.040946006774902,
      "learning_rate": 7.124154248155562e-07,
      "loss": 0.3398,
      "num_input_tokens_seen": 18713576,
      "step": 28670
    },
    {
      "epoch": 16.90742924528302,
      "grad_norm": 3.970780849456787,
      "learning_rate": 7.110924448310813e-07,
      "loss": 0.3325,
      "num_input_tokens_seen": 18716232,
      "step": 28675
    },
    {
      "epoch": 16.910377358490567,
      "grad_norm": 2.820988893508911,
      "learning_rate": 7.097706003533666e-07,
      "loss": 0.2167,
      "num_input_tokens_seen": 18719432,
      "step": 28680
    },
    {
      "epoch": 16.913325471698112,
      "grad_norm": 2.992216110229492,
      "learning_rate": 7.084498917323751e-07,
      "loss": 0.3636,
      "num_input_tokens_seen": 18722664,
      "step": 28685
    },
    {
      "epoch": 16.91627358490566,
      "grad_norm": 3.8460693359375,
      "learning_rate": 7.071303193177698e-07,
      "loss": 0.3266,
      "num_input_tokens_seen": 18725992,
      "step": 28690
    },
    {
      "epoch": 16.91922169811321,
      "grad_norm": 2.9601423740386963,
      "learning_rate": 7.058118834589133e-07,
      "loss": 0.3559,
      "num_input_tokens_seen": 18728520,
      "step": 28695
    },
    {
      "epoch": 16.922169811320753,
      "grad_norm": 3.2146620750427246,
      "learning_rate": 7.044945845048684e-07,
      "loss": 0.3259,
      "num_input_tokens_seen": 18732168,
      "step": 28700
    },
    {
      "epoch": 16.9251179245283,
      "grad_norm": 4.605044364929199,
      "learning_rate": 7.031784228043948e-07,
      "loss": 0.2902,
      "num_input_tokens_seen": 18736264,
      "step": 28705
    },
    {
      "epoch": 16.92806603773585,
      "grad_norm": 2.1182103157043457,
      "learning_rate": 7.01863398705952e-07,
      "loss": 0.2599,
      "num_input_tokens_seen": 18739144,
      "step": 28710
    },
    {
      "epoch": 16.931014150943398,
      "grad_norm": 4.951053142547607,
      "learning_rate": 7.005495125576983e-07,
      "loss": 0.3527,
      "num_input_tokens_seen": 18745096,
      "step": 28715
    },
    {
      "epoch": 16.933962264150942,
      "grad_norm": 3.7235867977142334,
      "learning_rate": 6.99236764707491e-07,
      "loss": 0.2285,
      "num_input_tokens_seen": 18747432,
      "step": 28720
    },
    {
      "epoch": 16.93691037735849,
      "grad_norm": 4.073049545288086,
      "learning_rate": 6.979251555028843e-07,
      "loss": 0.2431,
      "num_input_tokens_seen": 18751080,
      "step": 28725
    },
    {
      "epoch": 16.93985849056604,
      "grad_norm": 4.604541301727295,
      "learning_rate": 6.966146852911332e-07,
      "loss": 0.2945,
      "num_input_tokens_seen": 18753672,
      "step": 28730
    },
    {
      "epoch": 16.942806603773583,
      "grad_norm": 3.618743419647217,
      "learning_rate": 6.953053544191923e-07,
      "loss": 0.2463,
      "num_input_tokens_seen": 18757032,
      "step": 28735
    },
    {
      "epoch": 16.94575471698113,
      "grad_norm": 3.5476527214050293,
      "learning_rate": 6.939971632337111e-07,
      "loss": 0.3649,
      "num_input_tokens_seen": 18760744,
      "step": 28740
    },
    {
      "epoch": 16.94870283018868,
      "grad_norm": 5.173311233520508,
      "learning_rate": 6.926901120810387e-07,
      "loss": 0.3169,
      "num_input_tokens_seen": 18764072,
      "step": 28745
    },
    {
      "epoch": 16.951650943396228,
      "grad_norm": 3.281554937362671,
      "learning_rate": 6.91384201307222e-07,
      "loss": 0.2927,
      "num_input_tokens_seen": 18767496,
      "step": 28750
    },
    {
      "epoch": 16.954599056603772,
      "grad_norm": 3.33485746383667,
      "learning_rate": 6.900794312580078e-07,
      "loss": 0.4488,
      "num_input_tokens_seen": 18771240,
      "step": 28755
    },
    {
      "epoch": 16.95754716981132,
      "grad_norm": 4.5220842361450195,
      "learning_rate": 6.887758022788377e-07,
      "loss": 0.2602,
      "num_input_tokens_seen": 18774312,
      "step": 28760
    },
    {
      "epoch": 16.96049528301887,
      "grad_norm": 2.8788657188415527,
      "learning_rate": 6.874733147148549e-07,
      "loss": 0.2412,
      "num_input_tokens_seen": 18779272,
      "step": 28765
    },
    {
      "epoch": 16.963443396226417,
      "grad_norm": 3.564028024673462,
      "learning_rate": 6.861719689108987e-07,
      "loss": 0.311,
      "num_input_tokens_seen": 18782408,
      "step": 28770
    },
    {
      "epoch": 16.96639150943396,
      "grad_norm": 3.9503748416900635,
      "learning_rate": 6.84871765211505e-07,
      "loss": 0.3793,
      "num_input_tokens_seen": 18785096,
      "step": 28775
    },
    {
      "epoch": 16.96933962264151,
      "grad_norm": 4.933022499084473,
      "learning_rate": 6.835727039609086e-07,
      "loss": 0.3615,
      "num_input_tokens_seen": 18788680,
      "step": 28780
    },
    {
      "epoch": 16.972287735849058,
      "grad_norm": 3.2679145336151123,
      "learning_rate": 6.822747855030415e-07,
      "loss": 0.38,
      "num_input_tokens_seen": 18791592,
      "step": 28785
    },
    {
      "epoch": 16.975235849056602,
      "grad_norm": 3.2902326583862305,
      "learning_rate": 6.809780101815322e-07,
      "loss": 0.4038,
      "num_input_tokens_seen": 18794248,
      "step": 28790
    },
    {
      "epoch": 16.97818396226415,
      "grad_norm": 3.654723644256592,
      "learning_rate": 6.796823783397099e-07,
      "loss": 0.2689,
      "num_input_tokens_seen": 18797032,
      "step": 28795
    },
    {
      "epoch": 16.9811320754717,
      "grad_norm": 4.483514308929443,
      "learning_rate": 6.783878903205976e-07,
      "loss": 0.3077,
      "num_input_tokens_seen": 18800552,
      "step": 28800
    },
    {
      "epoch": 16.984080188679247,
      "grad_norm": 7.218395233154297,
      "learning_rate": 6.77094546466916e-07,
      "loss": 0.3135,
      "num_input_tokens_seen": 18803304,
      "step": 28805
    },
    {
      "epoch": 16.98702830188679,
      "grad_norm": 4.23872709274292,
      "learning_rate": 6.758023471210845e-07,
      "loss": 0.2349,
      "num_input_tokens_seen": 18806440,
      "step": 28810
    },
    {
      "epoch": 16.98997641509434,
      "grad_norm": 3.07650089263916,
      "learning_rate": 6.745112926252162e-07,
      "loss": 0.3146,
      "num_input_tokens_seen": 18809512,
      "step": 28815
    },
    {
      "epoch": 16.992924528301888,
      "grad_norm": 2.9675638675689697,
      "learning_rate": 6.732213833211265e-07,
      "loss": 0.3026,
      "num_input_tokens_seen": 18813544,
      "step": 28820
    },
    {
      "epoch": 16.995872641509433,
      "grad_norm": 3.477440118789673,
      "learning_rate": 6.719326195503218e-07,
      "loss": 0.3043,
      "num_input_tokens_seen": 18816200,
      "step": 28825
    },
    {
      "epoch": 16.99882075471698,
      "grad_norm": 6.5422821044921875,
      "learning_rate": 6.706450016540094e-07,
      "loss": 0.3289,
      "num_input_tokens_seen": 18818120,
      "step": 28830
    },
    {
      "epoch": 17.00176886792453,
      "grad_norm": 5.277149677276611,
      "learning_rate": 6.69358529973092e-07,
      "loss": 0.4684,
      "num_input_tokens_seen": 18820560,
      "step": 28835
    },
    {
      "epoch": 17.004716981132077,
      "grad_norm": 2.618361711502075,
      "learning_rate": 6.680732048481681e-07,
      "loss": 0.2389,
      "num_input_tokens_seen": 18823376,
      "step": 28840
    },
    {
      "epoch": 17.00766509433962,
      "grad_norm": 6.985800266265869,
      "learning_rate": 6.667890266195321e-07,
      "loss": 0.2828,
      "num_input_tokens_seen": 18826320,
      "step": 28845
    },
    {
      "epoch": 17.01061320754717,
      "grad_norm": 3.6116225719451904,
      "learning_rate": 6.655059956271759e-07,
      "loss": 0.4073,
      "num_input_tokens_seen": 18829392,
      "step": 28850
    },
    {
      "epoch": 17.013561320754718,
      "grad_norm": 3.4441182613372803,
      "learning_rate": 6.642241122107884e-07,
      "loss": 0.2946,
      "num_input_tokens_seen": 18832880,
      "step": 28855
    },
    {
      "epoch": 17.016509433962263,
      "grad_norm": 5.60723352432251,
      "learning_rate": 6.629433767097537e-07,
      "loss": 0.4036,
      "num_input_tokens_seen": 18835600,
      "step": 28860
    },
    {
      "epoch": 17.01945754716981,
      "grad_norm": 5.501210689544678,
      "learning_rate": 6.616637894631517e-07,
      "loss": 0.2604,
      "num_input_tokens_seen": 18838384,
      "step": 28865
    },
    {
      "epoch": 17.02240566037736,
      "grad_norm": 2.9047465324401855,
      "learning_rate": 6.603853508097591e-07,
      "loss": 0.3213,
      "num_input_tokens_seen": 18841360,
      "step": 28870
    },
    {
      "epoch": 17.025353773584907,
      "grad_norm": 4.209240436553955,
      "learning_rate": 6.591080610880468e-07,
      "loss": 0.2987,
      "num_input_tokens_seen": 18844464,
      "step": 28875
    },
    {
      "epoch": 17.028301886792452,
      "grad_norm": 3.9276926517486572,
      "learning_rate": 6.578319206361828e-07,
      "loss": 0.2957,
      "num_input_tokens_seen": 18847632,
      "step": 28880
    },
    {
      "epoch": 17.03125,
      "grad_norm": 4.21993350982666,
      "learning_rate": 6.565569297920327e-07,
      "loss": 0.2527,
      "num_input_tokens_seen": 18850832,
      "step": 28885
    },
    {
      "epoch": 17.034198113207548,
      "grad_norm": 5.8264479637146,
      "learning_rate": 6.552830888931544e-07,
      "loss": 0.3308,
      "num_input_tokens_seen": 18855088,
      "step": 28890
    },
    {
      "epoch": 17.037146226415093,
      "grad_norm": 4.929085731506348,
      "learning_rate": 6.540103982768031e-07,
      "loss": 0.3567,
      "num_input_tokens_seen": 18859024,
      "step": 28895
    },
    {
      "epoch": 17.04009433962264,
      "grad_norm": 4.483932971954346,
      "learning_rate": 6.527388582799293e-07,
      "loss": 0.3168,
      "num_input_tokens_seen": 18862768,
      "step": 28900
    },
    {
      "epoch": 17.04304245283019,
      "grad_norm": 3.3115944862365723,
      "learning_rate": 6.514684692391782e-07,
      "loss": 0.336,
      "num_input_tokens_seen": 18866448,
      "step": 28905
    },
    {
      "epoch": 17.045990566037737,
      "grad_norm": 2.215034246444702,
      "learning_rate": 6.501992314908895e-07,
      "loss": 0.28,
      "num_input_tokens_seen": 18869200,
      "step": 28910
    },
    {
      "epoch": 17.048938679245282,
      "grad_norm": 4.784584999084473,
      "learning_rate": 6.489311453711017e-07,
      "loss": 0.2487,
      "num_input_tokens_seen": 18871792,
      "step": 28915
    },
    {
      "epoch": 17.05188679245283,
      "grad_norm": 3.3275740146636963,
      "learning_rate": 6.476642112155457e-07,
      "loss": 0.3703,
      "num_input_tokens_seen": 18875824,
      "step": 28920
    },
    {
      "epoch": 17.05483490566038,
      "grad_norm": 3.625434637069702,
      "learning_rate": 6.463984293596476e-07,
      "loss": 0.3144,
      "num_input_tokens_seen": 18878384,
      "step": 28925
    },
    {
      "epoch": 17.057783018867923,
      "grad_norm": 7.646711349487305,
      "learning_rate": 6.451338001385282e-07,
      "loss": 0.2724,
      "num_input_tokens_seen": 18881584,
      "step": 28930
    },
    {
      "epoch": 17.06073113207547,
      "grad_norm": 3.476174831390381,
      "learning_rate": 6.438703238870037e-07,
      "loss": 0.3482,
      "num_input_tokens_seen": 18885808,
      "step": 28935
    },
    {
      "epoch": 17.06367924528302,
      "grad_norm": 5.961153507232666,
      "learning_rate": 6.426080009395846e-07,
      "loss": 0.2442,
      "num_input_tokens_seen": 18889680,
      "step": 28940
    },
    {
      "epoch": 17.066627358490567,
      "grad_norm": 3.243980884552002,
      "learning_rate": 6.413468316304755e-07,
      "loss": 0.3201,
      "num_input_tokens_seen": 18893200,
      "step": 28945
    },
    {
      "epoch": 17.069575471698112,
      "grad_norm": 3.2043697834014893,
      "learning_rate": 6.400868162935786e-07,
      "loss": 0.298,
      "num_input_tokens_seen": 18896496,
      "step": 28950
    },
    {
      "epoch": 17.07252358490566,
      "grad_norm": 5.6146321296691895,
      "learning_rate": 6.388279552624877e-07,
      "loss": 0.3733,
      "num_input_tokens_seen": 18900528,
      "step": 28955
    },
    {
      "epoch": 17.07547169811321,
      "grad_norm": 3.801034927368164,
      "learning_rate": 6.37570248870491e-07,
      "loss": 0.3959,
      "num_input_tokens_seen": 18903664,
      "step": 28960
    },
    {
      "epoch": 17.078419811320753,
      "grad_norm": 2.3925743103027344,
      "learning_rate": 6.363136974505718e-07,
      "loss": 0.2677,
      "num_input_tokens_seen": 18907216,
      "step": 28965
    },
    {
      "epoch": 17.0813679245283,
      "grad_norm": 4.378587245941162,
      "learning_rate": 6.350583013354078e-07,
      "loss": 0.3265,
      "num_input_tokens_seen": 18910992,
      "step": 28970
    },
    {
      "epoch": 17.08431603773585,
      "grad_norm": 3.958256483078003,
      "learning_rate": 6.338040608573693e-07,
      "loss": 0.3975,
      "num_input_tokens_seen": 18914736,
      "step": 28975
    },
    {
      "epoch": 17.087264150943398,
      "grad_norm": 3.798232078552246,
      "learning_rate": 6.325509763485238e-07,
      "loss": 0.4356,
      "num_input_tokens_seen": 18917680,
      "step": 28980
    },
    {
      "epoch": 17.090212264150942,
      "grad_norm": 3.3746109008789062,
      "learning_rate": 6.312990481406301e-07,
      "loss": 0.422,
      "num_input_tokens_seen": 18921104,
      "step": 28985
    },
    {
      "epoch": 17.09316037735849,
      "grad_norm": 8.453032493591309,
      "learning_rate": 6.300482765651411e-07,
      "loss": 0.3194,
      "num_input_tokens_seen": 18923632,
      "step": 28990
    },
    {
      "epoch": 17.09610849056604,
      "grad_norm": 4.32177734375,
      "learning_rate": 6.28798661953205e-07,
      "loss": 0.3847,
      "num_input_tokens_seen": 18927440,
      "step": 28995
    },
    {
      "epoch": 17.099056603773583,
      "grad_norm": 2.9792327880859375,
      "learning_rate": 6.275502046356618e-07,
      "loss": 0.3914,
      "num_input_tokens_seen": 18930896,
      "step": 29000
    },
    {
      "epoch": 17.10200471698113,
      "grad_norm": 5.194839954376221,
      "learning_rate": 6.263029049430447e-07,
      "loss": 0.4015,
      "num_input_tokens_seen": 18933680,
      "step": 29005
    },
    {
      "epoch": 17.10495283018868,
      "grad_norm": 3.6355044841766357,
      "learning_rate": 6.250567632055832e-07,
      "loss": 0.2719,
      "num_input_tokens_seen": 18937200,
      "step": 29010
    },
    {
      "epoch": 17.107900943396228,
      "grad_norm": 3.861161470413208,
      "learning_rate": 6.238117797532e-07,
      "loss": 0.2425,
      "num_input_tokens_seen": 18940976,
      "step": 29015
    },
    {
      "epoch": 17.110849056603772,
      "grad_norm": 3.0211374759674072,
      "learning_rate": 6.225679549155083e-07,
      "loss": 0.2221,
      "num_input_tokens_seen": 18943472,
      "step": 29020
    },
    {
      "epoch": 17.11379716981132,
      "grad_norm": 2.980250358581543,
      "learning_rate": 6.213252890218163e-07,
      "loss": 0.4864,
      "num_input_tokens_seen": 18946608,
      "step": 29025
    },
    {
      "epoch": 17.11674528301887,
      "grad_norm": 4.8874592781066895,
      "learning_rate": 6.200837824011247e-07,
      "loss": 0.2447,
      "num_input_tokens_seen": 18949232,
      "step": 29030
    },
    {
      "epoch": 17.119693396226417,
      "grad_norm": 2.636491060256958,
      "learning_rate": 6.188434353821282e-07,
      "loss": 0.283,
      "num_input_tokens_seen": 18953040,
      "step": 29035
    },
    {
      "epoch": 17.12264150943396,
      "grad_norm": 3.269869327545166,
      "learning_rate": 6.176042482932132e-07,
      "loss": 0.3229,
      "num_input_tokens_seen": 18956144,
      "step": 29040
    },
    {
      "epoch": 17.12558962264151,
      "grad_norm": 3.9316980838775635,
      "learning_rate": 6.163662214624616e-07,
      "loss": 0.3138,
      "num_input_tokens_seen": 18959312,
      "step": 29045
    },
    {
      "epoch": 17.128537735849058,
      "grad_norm": 3.5140061378479004,
      "learning_rate": 6.151293552176451e-07,
      "loss": 0.29,
      "num_input_tokens_seen": 18962384,
      "step": 29050
    },
    {
      "epoch": 17.131485849056602,
      "grad_norm": 2.257054090499878,
      "learning_rate": 6.138936498862291e-07,
      "loss": 0.3089,
      "num_input_tokens_seen": 18965936,
      "step": 29055
    },
    {
      "epoch": 17.13443396226415,
      "grad_norm": 4.3367414474487305,
      "learning_rate": 6.126591057953729e-07,
      "loss": 0.3179,
      "num_input_tokens_seen": 18969488,
      "step": 29060
    },
    {
      "epoch": 17.1373820754717,
      "grad_norm": 7.230766296386719,
      "learning_rate": 6.114257232719267e-07,
      "loss": 0.2565,
      "num_input_tokens_seen": 18971792,
      "step": 29065
    },
    {
      "epoch": 17.140330188679247,
      "grad_norm": 2.9478418827056885,
      "learning_rate": 6.101935026424332e-07,
      "loss": 0.411,
      "num_input_tokens_seen": 18975312,
      "step": 29070
    },
    {
      "epoch": 17.14327830188679,
      "grad_norm": 3.457930088043213,
      "learning_rate": 6.089624442331293e-07,
      "loss": 0.3314,
      "num_input_tokens_seen": 18977968,
      "step": 29075
    },
    {
      "epoch": 17.14622641509434,
      "grad_norm": 4.382983207702637,
      "learning_rate": 6.077325483699432e-07,
      "loss": 0.296,
      "num_input_tokens_seen": 18980432,
      "step": 29080
    },
    {
      "epoch": 17.149174528301888,
      "grad_norm": 3.730106830596924,
      "learning_rate": 6.065038153784947e-07,
      "loss": 0.237,
      "num_input_tokens_seen": 18983216,
      "step": 29085
    },
    {
      "epoch": 17.152122641509433,
      "grad_norm": 3.6993281841278076,
      "learning_rate": 6.052762455840955e-07,
      "loss": 0.2683,
      "num_input_tokens_seen": 18986672,
      "step": 29090
    },
    {
      "epoch": 17.15507075471698,
      "grad_norm": 3.766350746154785,
      "learning_rate": 6.040498393117494e-07,
      "loss": 0.2812,
      "num_input_tokens_seen": 18990000,
      "step": 29095
    },
    {
      "epoch": 17.15801886792453,
      "grad_norm": 2.8253731727600098,
      "learning_rate": 6.028245968861551e-07,
      "loss": 0.2802,
      "num_input_tokens_seen": 18994000,
      "step": 29100
    },
    {
      "epoch": 17.160966981132077,
      "grad_norm": 4.212186336517334,
      "learning_rate": 6.016005186316987e-07,
      "loss": 0.3364,
      "num_input_tokens_seen": 18997136,
      "step": 29105
    },
    {
      "epoch": 17.16391509433962,
      "grad_norm": 2.9638659954071045,
      "learning_rate": 6.003776048724614e-07,
      "loss": 0.3032,
      "num_input_tokens_seen": 19000208,
      "step": 29110
    },
    {
      "epoch": 17.16686320754717,
      "grad_norm": 3.2039759159088135,
      "learning_rate": 5.991558559322152e-07,
      "loss": 0.246,
      "num_input_tokens_seen": 19002736,
      "step": 29115
    },
    {
      "epoch": 17.169811320754718,
      "grad_norm": 5.007462024688721,
      "learning_rate": 5.979352721344223e-07,
      "loss": 0.3158,
      "num_input_tokens_seen": 19006480,
      "step": 29120
    },
    {
      "epoch": 17.172759433962263,
      "grad_norm": 3.2318761348724365,
      "learning_rate": 5.967158538022383e-07,
      "loss": 0.2779,
      "num_input_tokens_seen": 19010416,
      "step": 29125
    },
    {
      "epoch": 17.17570754716981,
      "grad_norm": 6.452537536621094,
      "learning_rate": 5.954976012585078e-07,
      "loss": 0.2406,
      "num_input_tokens_seen": 19013552,
      "step": 29130
    },
    {
      "epoch": 17.17865566037736,
      "grad_norm": 2.605916976928711,
      "learning_rate": 5.942805148257713e-07,
      "loss": 0.2935,
      "num_input_tokens_seen": 19017328,
      "step": 29135
    },
    {
      "epoch": 17.181603773584907,
      "grad_norm": 3.168342113494873,
      "learning_rate": 5.930645948262553e-07,
      "loss": 0.4511,
      "num_input_tokens_seen": 19020816,
      "step": 29140
    },
    {
      "epoch": 17.184551886792452,
      "grad_norm": 6.5007405281066895,
      "learning_rate": 5.918498415818813e-07,
      "loss": 0.2451,
      "num_input_tokens_seen": 19023408,
      "step": 29145
    },
    {
      "epoch": 17.1875,
      "grad_norm": 3.4187076091766357,
      "learning_rate": 5.906362554142592e-07,
      "loss": 0.271,
      "num_input_tokens_seen": 19026416,
      "step": 29150
    },
    {
      "epoch": 17.190448113207548,
      "grad_norm": 4.388454437255859,
      "learning_rate": 5.894238366446925e-07,
      "loss": 0.2569,
      "num_input_tokens_seen": 19029168,
      "step": 29155
    },
    {
      "epoch": 17.193396226415093,
      "grad_norm": 2.6591086387634277,
      "learning_rate": 5.882125855941723e-07,
      "loss": 0.2518,
      "num_input_tokens_seen": 19032912,
      "step": 29160
    },
    {
      "epoch": 17.19634433962264,
      "grad_norm": 5.403634548187256,
      "learning_rate": 5.870025025833842e-07,
      "loss": 0.3503,
      "num_input_tokens_seen": 19035472,
      "step": 29165
    },
    {
      "epoch": 17.19929245283019,
      "grad_norm": 3.2672014236450195,
      "learning_rate": 5.857935879327031e-07,
      "loss": 0.2846,
      "num_input_tokens_seen": 19039408,
      "step": 29170
    },
    {
      "epoch": 17.202240566037737,
      "grad_norm": 3.047598123550415,
      "learning_rate": 5.845858419621936e-07,
      "loss": 0.298,
      "num_input_tokens_seen": 19042384,
      "step": 29175
    },
    {
      "epoch": 17.205188679245282,
      "grad_norm": 4.28211784362793,
      "learning_rate": 5.83379264991612e-07,
      "loss": 0.3781,
      "num_input_tokens_seen": 19045648,
      "step": 29180
    },
    {
      "epoch": 17.20813679245283,
      "grad_norm": 5.4828267097473145,
      "learning_rate": 5.821738573404046e-07,
      "loss": 0.4946,
      "num_input_tokens_seen": 19049136,
      "step": 29185
    },
    {
      "epoch": 17.21108490566038,
      "grad_norm": 3.1514363288879395,
      "learning_rate": 5.80969619327707e-07,
      "loss": 0.2932,
      "num_input_tokens_seen": 19051728,
      "step": 29190
    },
    {
      "epoch": 17.214033018867923,
      "grad_norm": 2.9034767150878906,
      "learning_rate": 5.797665512723488e-07,
      "loss": 0.2857,
      "num_input_tokens_seen": 19055728,
      "step": 29195
    },
    {
      "epoch": 17.21698113207547,
      "grad_norm": 6.199516773223877,
      "learning_rate": 5.785646534928452e-07,
      "loss": 0.3064,
      "num_input_tokens_seen": 19058832,
      "step": 29200
    },
    {
      "epoch": 17.21992924528302,
      "grad_norm": 3.9082412719726562,
      "learning_rate": 5.77363926307406e-07,
      "loss": 0.2681,
      "num_input_tokens_seen": 19062608,
      "step": 29205
    },
    {
      "epoch": 17.222877358490567,
      "grad_norm": 5.118161201477051,
      "learning_rate": 5.761643700339281e-07,
      "loss": 0.297,
      "num_input_tokens_seen": 19066320,
      "step": 29210
    },
    {
      "epoch": 17.225825471698112,
      "grad_norm": 3.933835744857788,
      "learning_rate": 5.749659849899985e-07,
      "loss": 0.3395,
      "num_input_tokens_seen": 19069136,
      "step": 29215
    },
    {
      "epoch": 17.22877358490566,
      "grad_norm": 3.4229798316955566,
      "learning_rate": 5.737687714928953e-07,
      "loss": 0.3699,
      "num_input_tokens_seen": 19072496,
      "step": 29220
    },
    {
      "epoch": 17.23172169811321,
      "grad_norm": 3.1838982105255127,
      "learning_rate": 5.725727298595846e-07,
      "loss": 0.3469,
      "num_input_tokens_seen": 19075440,
      "step": 29225
    },
    {
      "epoch": 17.234669811320753,
      "grad_norm": 3.5164546966552734,
      "learning_rate": 5.71377860406726e-07,
      "loss": 0.3471,
      "num_input_tokens_seen": 19079728,
      "step": 29230
    },
    {
      "epoch": 17.2376179245283,
      "grad_norm": 5.103067874908447,
      "learning_rate": 5.701841634506655e-07,
      "loss": 0.3067,
      "num_input_tokens_seen": 19082640,
      "step": 29235
    },
    {
      "epoch": 17.24056603773585,
      "grad_norm": 4.745849132537842,
      "learning_rate": 5.689916393074391e-07,
      "loss": 0.3328,
      "num_input_tokens_seen": 19085456,
      "step": 29240
    },
    {
      "epoch": 17.243514150943398,
      "grad_norm": 11.054007530212402,
      "learning_rate": 5.678002882927725e-07,
      "loss": 0.4058,
      "num_input_tokens_seen": 19088304,
      "step": 29245
    },
    {
      "epoch": 17.246462264150942,
      "grad_norm": 4.894809246063232,
      "learning_rate": 5.666101107220811e-07,
      "loss": 0.2134,
      "num_input_tokens_seen": 19090672,
      "step": 29250
    },
    {
      "epoch": 17.24941037735849,
      "grad_norm": 3.7762036323547363,
      "learning_rate": 5.654211069104693e-07,
      "loss": 0.2583,
      "num_input_tokens_seen": 19094000,
      "step": 29255
    },
    {
      "epoch": 17.25235849056604,
      "grad_norm": 3.5437495708465576,
      "learning_rate": 5.642332771727321e-07,
      "loss": 0.2633,
      "num_input_tokens_seen": 19097040,
      "step": 29260
    },
    {
      "epoch": 17.255306603773583,
      "grad_norm": 4.022183418273926,
      "learning_rate": 5.630466218233521e-07,
      "loss": 0.4488,
      "num_input_tokens_seen": 19099984,
      "step": 29265
    },
    {
      "epoch": 17.25825471698113,
      "grad_norm": 7.098259449005127,
      "learning_rate": 5.618611411765007e-07,
      "loss": 0.3237,
      "num_input_tokens_seen": 19102576,
      "step": 29270
    },
    {
      "epoch": 17.26120283018868,
      "grad_norm": 3.7024929523468018,
      "learning_rate": 5.606768355460401e-07,
      "loss": 0.2747,
      "num_input_tokens_seen": 19105648,
      "step": 29275
    },
    {
      "epoch": 17.264150943396228,
      "grad_norm": 4.162961959838867,
      "learning_rate": 5.594937052455191e-07,
      "loss": 0.2793,
      "num_input_tokens_seen": 19108432,
      "step": 29280
    },
    {
      "epoch": 17.267099056603772,
      "grad_norm": 8.350800514221191,
      "learning_rate": 5.583117505881764e-07,
      "loss": 0.2991,
      "num_input_tokens_seen": 19112496,
      "step": 29285
    },
    {
      "epoch": 17.27004716981132,
      "grad_norm": 3.2989964485168457,
      "learning_rate": 5.571309718869417e-07,
      "loss": 0.2687,
      "num_input_tokens_seen": 19116528,
      "step": 29290
    },
    {
      "epoch": 17.27299528301887,
      "grad_norm": 5.076319694519043,
      "learning_rate": 5.559513694544282e-07,
      "loss": 0.2841,
      "num_input_tokens_seen": 19119664,
      "step": 29295
    },
    {
      "epoch": 17.275943396226417,
      "grad_norm": 3.0870983600616455,
      "learning_rate": 5.547729436029442e-07,
      "loss": 0.3022,
      "num_input_tokens_seen": 19122320,
      "step": 29300
    },
    {
      "epoch": 17.27889150943396,
      "grad_norm": 3.8009376525878906,
      "learning_rate": 5.535956946444809e-07,
      "loss": 0.3635,
      "num_input_tokens_seen": 19126096,
      "step": 29305
    },
    {
      "epoch": 17.28183962264151,
      "grad_norm": 4.083861827850342,
      "learning_rate": 5.524196228907203e-07,
      "loss": 0.351,
      "num_input_tokens_seen": 19130608,
      "step": 29310
    },
    {
      "epoch": 17.284787735849058,
      "grad_norm": 4.0996575355529785,
      "learning_rate": 5.512447286530326e-07,
      "loss": 0.4131,
      "num_input_tokens_seen": 19134832,
      "step": 29315
    },
    {
      "epoch": 17.287735849056602,
      "grad_norm": 6.757421970367432,
      "learning_rate": 5.500710122424746e-07,
      "loss": 0.334,
      "num_input_tokens_seen": 19138064,
      "step": 29320
    },
    {
      "epoch": 17.29068396226415,
      "grad_norm": 9.738259315490723,
      "learning_rate": 5.488984739697961e-07,
      "loss": 0.2652,
      "num_input_tokens_seen": 19140528,
      "step": 29325
    },
    {
      "epoch": 17.2936320754717,
      "grad_norm": 4.88520622253418,
      "learning_rate": 5.477271141454294e-07,
      "loss": 0.3746,
      "num_input_tokens_seen": 19143600,
      "step": 29330
    },
    {
      "epoch": 17.296580188679247,
      "grad_norm": 2.0883452892303467,
      "learning_rate": 5.465569330794974e-07,
      "loss": 0.251,
      "num_input_tokens_seen": 19146960,
      "step": 29335
    },
    {
      "epoch": 17.29952830188679,
      "grad_norm": 3.2387006282806396,
      "learning_rate": 5.453879310818105e-07,
      "loss": 0.3181,
      "num_input_tokens_seen": 19149488,
      "step": 29340
    },
    {
      "epoch": 17.30247641509434,
      "grad_norm": 3.0843491554260254,
      "learning_rate": 5.442201084618664e-07,
      "loss": 0.2449,
      "num_input_tokens_seen": 19152464,
      "step": 29345
    },
    {
      "epoch": 17.305424528301888,
      "grad_norm": 3.8614251613616943,
      "learning_rate": 5.430534655288528e-07,
      "loss": 0.2897,
      "num_input_tokens_seen": 19155856,
      "step": 29350
    },
    {
      "epoch": 17.308372641509433,
      "grad_norm": 5.18664026260376,
      "learning_rate": 5.418880025916428e-07,
      "loss": 0.2634,
      "num_input_tokens_seen": 19158704,
      "step": 29355
    },
    {
      "epoch": 17.31132075471698,
      "grad_norm": 3.4826254844665527,
      "learning_rate": 5.407237199587973e-07,
      "loss": 0.2969,
      "num_input_tokens_seen": 19162064,
      "step": 29360
    },
    {
      "epoch": 17.31426886792453,
      "grad_norm": 3.203946590423584,
      "learning_rate": 5.395606179385654e-07,
      "loss": 0.4221,
      "num_input_tokens_seen": 19165968,
      "step": 29365
    },
    {
      "epoch": 17.317216981132077,
      "grad_norm": 3.0634963512420654,
      "learning_rate": 5.383986968388833e-07,
      "loss": 0.2409,
      "num_input_tokens_seen": 19169424,
      "step": 29370
    },
    {
      "epoch": 17.32016509433962,
      "grad_norm": 4.479260444641113,
      "learning_rate": 5.372379569673736e-07,
      "loss": 0.2774,
      "num_input_tokens_seen": 19172176,
      "step": 29375
    },
    {
      "epoch": 17.32311320754717,
      "grad_norm": 4.5797224044799805,
      "learning_rate": 5.360783986313495e-07,
      "loss": 0.2571,
      "num_input_tokens_seen": 19175216,
      "step": 29380
    },
    {
      "epoch": 17.326061320754718,
      "grad_norm": 7.678421974182129,
      "learning_rate": 5.349200221378076e-07,
      "loss": 0.3781,
      "num_input_tokens_seen": 19177744,
      "step": 29385
    },
    {
      "epoch": 17.329009433962263,
      "grad_norm": 3.3629536628723145,
      "learning_rate": 5.33762827793432e-07,
      "loss": 0.2195,
      "num_input_tokens_seen": 19181552,
      "step": 29390
    },
    {
      "epoch": 17.33195754716981,
      "grad_norm": 4.92515754699707,
      "learning_rate": 5.326068159045978e-07,
      "loss": 0.3001,
      "num_input_tokens_seen": 19184272,
      "step": 29395
    },
    {
      "epoch": 17.33490566037736,
      "grad_norm": 3.416231632232666,
      "learning_rate": 5.314519867773621e-07,
      "loss": 0.2516,
      "num_input_tokens_seen": 19186736,
      "step": 29400
    },
    {
      "epoch": 17.337853773584907,
      "grad_norm": 4.624024391174316,
      "learning_rate": 5.302983407174711e-07,
      "loss": 0.2914,
      "num_input_tokens_seen": 19190032,
      "step": 29405
    },
    {
      "epoch": 17.340801886792452,
      "grad_norm": 5.584841728210449,
      "learning_rate": 5.291458780303572e-07,
      "loss": 0.3048,
      "num_input_tokens_seen": 19192560,
      "step": 29410
    },
    {
      "epoch": 17.34375,
      "grad_norm": 4.910147666931152,
      "learning_rate": 5.279945990211411e-07,
      "loss": 0.2656,
      "num_input_tokens_seen": 19195664,
      "step": 29415
    },
    {
      "epoch": 17.346698113207548,
      "grad_norm": 5.374019622802734,
      "learning_rate": 5.26844503994628e-07,
      "loss": 0.2001,
      "num_input_tokens_seen": 19198512,
      "step": 29420
    },
    {
      "epoch": 17.349646226415093,
      "grad_norm": 3.2921152114868164,
      "learning_rate": 5.25695593255311e-07,
      "loss": 0.2826,
      "num_input_tokens_seen": 19201296,
      "step": 29425
    },
    {
      "epoch": 17.35259433962264,
      "grad_norm": 2.220937490463257,
      "learning_rate": 5.24547867107369e-07,
      "loss": 0.3581,
      "num_input_tokens_seen": 19204496,
      "step": 29430
    },
    {
      "epoch": 17.35554245283019,
      "grad_norm": 3.2003278732299805,
      "learning_rate": 5.234013258546672e-07,
      "loss": 0.352,
      "num_input_tokens_seen": 19207280,
      "step": 29435
    },
    {
      "epoch": 17.358490566037737,
      "grad_norm": 3.3672468662261963,
      "learning_rate": 5.222559698007563e-07,
      "loss": 0.3048,
      "num_input_tokens_seen": 19210640,
      "step": 29440
    },
    {
      "epoch": 17.361438679245282,
      "grad_norm": 4.007109642028809,
      "learning_rate": 5.211117992488763e-07,
      "loss": 0.3254,
      "num_input_tokens_seen": 19213680,
      "step": 29445
    },
    {
      "epoch": 17.36438679245283,
      "grad_norm": 8.020753860473633,
      "learning_rate": 5.199688145019505e-07,
      "loss": 0.3811,
      "num_input_tokens_seen": 19216336,
      "step": 29450
    },
    {
      "epoch": 17.36733490566038,
      "grad_norm": 2.816948413848877,
      "learning_rate": 5.188270158625891e-07,
      "loss": 0.2949,
      "num_input_tokens_seen": 19219600,
      "step": 29455
    },
    {
      "epoch": 17.370283018867923,
      "grad_norm": 6.062239170074463,
      "learning_rate": 5.176864036330875e-07,
      "loss": 0.4705,
      "num_input_tokens_seen": 19222000,
      "step": 29460
    },
    {
      "epoch": 17.37323113207547,
      "grad_norm": 3.4085443019866943,
      "learning_rate": 5.165469781154287e-07,
      "loss": 0.2213,
      "num_input_tokens_seen": 19226320,
      "step": 29465
    },
    {
      "epoch": 17.37617924528302,
      "grad_norm": 3.9273881912231445,
      "learning_rate": 5.154087396112789e-07,
      "loss": 0.2652,
      "num_input_tokens_seen": 19230448,
      "step": 29470
    },
    {
      "epoch": 17.379127358490567,
      "grad_norm": 2.818270683288574,
      "learning_rate": 5.142716884219939e-07,
      "loss": 0.2501,
      "num_input_tokens_seen": 19233392,
      "step": 29475
    },
    {
      "epoch": 17.382075471698112,
      "grad_norm": 7.058661460876465,
      "learning_rate": 5.131358248486118e-07,
      "loss": 0.2874,
      "num_input_tokens_seen": 19236432,
      "step": 29480
    },
    {
      "epoch": 17.38502358490566,
      "grad_norm": 4.268390655517578,
      "learning_rate": 5.120011491918564e-07,
      "loss": 0.2896,
      "num_input_tokens_seen": 19240496,
      "step": 29485
    },
    {
      "epoch": 17.38797169811321,
      "grad_norm": 4.086554527282715,
      "learning_rate": 5.108676617521402e-07,
      "loss": 0.338,
      "num_input_tokens_seen": 19242960,
      "step": 29490
    },
    {
      "epoch": 17.390919811320753,
      "grad_norm": 2.0774359703063965,
      "learning_rate": 5.097353628295571e-07,
      "loss": 0.3103,
      "num_input_tokens_seen": 19246032,
      "step": 29495
    },
    {
      "epoch": 17.3938679245283,
      "grad_norm": 6.527927398681641,
      "learning_rate": 5.086042527238893e-07,
      "loss": 0.3332,
      "num_input_tokens_seen": 19249488,
      "step": 29500
    },
    {
      "epoch": 17.39681603773585,
      "grad_norm": 6.069390773773193,
      "learning_rate": 5.074743317346009e-07,
      "loss": 0.3523,
      "num_input_tokens_seen": 19252784,
      "step": 29505
    },
    {
      "epoch": 17.399764150943398,
      "grad_norm": 3.542381525039673,
      "learning_rate": 5.063456001608458e-07,
      "loss": 0.5155,
      "num_input_tokens_seen": 19256080,
      "step": 29510
    },
    {
      "epoch": 17.402712264150942,
      "grad_norm": 4.585089683532715,
      "learning_rate": 5.052180583014599e-07,
      "loss": 0.2626,
      "num_input_tokens_seen": 19258736,
      "step": 29515
    },
    {
      "epoch": 17.40566037735849,
      "grad_norm": 3.1628377437591553,
      "learning_rate": 5.04091706454965e-07,
      "loss": 0.2374,
      "num_input_tokens_seen": 19261072,
      "step": 29520
    },
    {
      "epoch": 17.40860849056604,
      "grad_norm": 4.301898956298828,
      "learning_rate": 5.029665449195665e-07,
      "loss": 0.326,
      "num_input_tokens_seen": 19263632,
      "step": 29525
    },
    {
      "epoch": 17.411556603773583,
      "grad_norm": 1.9644662141799927,
      "learning_rate": 5.018425739931559e-07,
      "loss": 0.2177,
      "num_input_tokens_seen": 19266320,
      "step": 29530
    },
    {
      "epoch": 17.41450471698113,
      "grad_norm": 5.495102405548096,
      "learning_rate": 5.007197939733099e-07,
      "loss": 0.2709,
      "num_input_tokens_seen": 19270320,
      "step": 29535
    },
    {
      "epoch": 17.41745283018868,
      "grad_norm": 3.2260141372680664,
      "learning_rate": 4.995982051572895e-07,
      "loss": 0.2668,
      "num_input_tokens_seen": 19274032,
      "step": 29540
    },
    {
      "epoch": 17.420400943396228,
      "grad_norm": 4.1187872886657715,
      "learning_rate": 4.984778078420405e-07,
      "loss": 0.2502,
      "num_input_tokens_seen": 19277520,
      "step": 29545
    },
    {
      "epoch": 17.423349056603772,
      "grad_norm": 3.1905806064605713,
      "learning_rate": 4.973586023241917e-07,
      "loss": 0.3827,
      "num_input_tokens_seen": 19280336,
      "step": 29550
    },
    {
      "epoch": 17.42629716981132,
      "grad_norm": 7.913694381713867,
      "learning_rate": 4.962405889000588e-07,
      "loss": 0.3391,
      "num_input_tokens_seen": 19283472,
      "step": 29555
    },
    {
      "epoch": 17.42924528301887,
      "grad_norm": 3.9036004543304443,
      "learning_rate": 4.951237678656396e-07,
      "loss": 0.3236,
      "num_input_tokens_seen": 19287024,
      "step": 29560
    },
    {
      "epoch": 17.432193396226417,
      "grad_norm": 3.0127153396606445,
      "learning_rate": 4.940081395166174e-07,
      "loss": 0.3463,
      "num_input_tokens_seen": 19290544,
      "step": 29565
    },
    {
      "epoch": 17.43514150943396,
      "grad_norm": 3.3962459564208984,
      "learning_rate": 4.928937041483606e-07,
      "loss": 0.3387,
      "num_input_tokens_seen": 19295184,
      "step": 29570
    },
    {
      "epoch": 17.43808962264151,
      "grad_norm": 1.5757339000701904,
      "learning_rate": 4.917804620559202e-07,
      "loss": 0.1971,
      "num_input_tokens_seen": 19298160,
      "step": 29575
    },
    {
      "epoch": 17.441037735849058,
      "grad_norm": 4.236644744873047,
      "learning_rate": 4.906684135340317e-07,
      "loss": 0.4083,
      "num_input_tokens_seen": 19300976,
      "step": 29580
    },
    {
      "epoch": 17.443985849056602,
      "grad_norm": 3.140695810317993,
      "learning_rate": 4.89557558877114e-07,
      "loss": 0.457,
      "num_input_tokens_seen": 19304048,
      "step": 29585
    },
    {
      "epoch": 17.44693396226415,
      "grad_norm": 4.475461959838867,
      "learning_rate": 4.884478983792728e-07,
      "loss": 0.2452,
      "num_input_tokens_seen": 19307248,
      "step": 29590
    },
    {
      "epoch": 17.4498820754717,
      "grad_norm": 3.385934591293335,
      "learning_rate": 4.873394323342939e-07,
      "loss": 0.2956,
      "num_input_tokens_seen": 19310608,
      "step": 29595
    },
    {
      "epoch": 17.452830188679247,
      "grad_norm": 4.398652076721191,
      "learning_rate": 4.86232161035648e-07,
      "loss": 0.2721,
      "num_input_tokens_seen": 19313584,
      "step": 29600
    },
    {
      "epoch": 17.45577830188679,
      "grad_norm": 3.378872871398926,
      "learning_rate": 4.851260847764916e-07,
      "loss": 0.301,
      "num_input_tokens_seen": 19317968,
      "step": 29605
    },
    {
      "epoch": 17.45872641509434,
      "grad_norm": 2.934174060821533,
      "learning_rate": 4.840212038496622e-07,
      "loss": 0.317,
      "num_input_tokens_seen": 19321104,
      "step": 29610
    },
    {
      "epoch": 17.461674528301888,
      "grad_norm": 4.517685890197754,
      "learning_rate": 4.82917518547682e-07,
      "loss": 0.331,
      "num_input_tokens_seen": 19324656,
      "step": 29615
    },
    {
      "epoch": 17.464622641509433,
      "grad_norm": 3.247096061706543,
      "learning_rate": 4.81815029162756e-07,
      "loss": 0.3208,
      "num_input_tokens_seen": 19327920,
      "step": 29620
    },
    {
      "epoch": 17.46757075471698,
      "grad_norm": 4.3389787673950195,
      "learning_rate": 4.807137359867725e-07,
      "loss": 0.291,
      "num_input_tokens_seen": 19330800,
      "step": 29625
    },
    {
      "epoch": 17.47051886792453,
      "grad_norm": 3.516417980194092,
      "learning_rate": 4.79613639311306e-07,
      "loss": 0.2731,
      "num_input_tokens_seen": 19334320,
      "step": 29630
    },
    {
      "epoch": 17.473466981132077,
      "grad_norm": 3.692194938659668,
      "learning_rate": 4.785147394276096e-07,
      "loss": 0.2686,
      "num_input_tokens_seen": 19337648,
      "step": 29635
    },
    {
      "epoch": 17.47641509433962,
      "grad_norm": 3.881232261657715,
      "learning_rate": 4.774170366266223e-07,
      "loss": 0.3165,
      "num_input_tokens_seen": 19340976,
      "step": 29640
    },
    {
      "epoch": 17.47936320754717,
      "grad_norm": 4.988696098327637,
      "learning_rate": 4.763205311989666e-07,
      "loss": 0.4306,
      "num_input_tokens_seen": 19344080,
      "step": 29645
    },
    {
      "epoch": 17.482311320754718,
      "grad_norm": 2.917496681213379,
      "learning_rate": 4.752252234349458e-07,
      "loss": 0.4194,
      "num_input_tokens_seen": 19348816,
      "step": 29650
    },
    {
      "epoch": 17.485259433962263,
      "grad_norm": 2.9181694984436035,
      "learning_rate": 4.7413111362454634e-07,
      "loss": 0.2636,
      "num_input_tokens_seen": 19351824,
      "step": 29655
    },
    {
      "epoch": 17.48820754716981,
      "grad_norm": 2.9263641834259033,
      "learning_rate": 4.7303820205744143e-07,
      "loss": 0.311,
      "num_input_tokens_seen": 19356368,
      "step": 29660
    },
    {
      "epoch": 17.49115566037736,
      "grad_norm": 5.925682067871094,
      "learning_rate": 4.7194648902298303e-07,
      "loss": 0.3681,
      "num_input_tokens_seen": 19359536,
      "step": 29665
    },
    {
      "epoch": 17.494103773584907,
      "grad_norm": 3.42815899848938,
      "learning_rate": 4.7085597481020594e-07,
      "loss": 0.2564,
      "num_input_tokens_seen": 19362832,
      "step": 29670
    },
    {
      "epoch": 17.497051886792452,
      "grad_norm": 3.28584623336792,
      "learning_rate": 4.697666597078293e-07,
      "loss": 0.4519,
      "num_input_tokens_seen": 19365264,
      "step": 29675
    },
    {
      "epoch": 17.5,
      "grad_norm": 3.416677236557007,
      "learning_rate": 4.6867854400425237e-07,
      "loss": 0.276,
      "num_input_tokens_seen": 19368048,
      "step": 29680
    },
    {
      "epoch": 17.502948113207548,
      "grad_norm": 3.6103389263153076,
      "learning_rate": 4.6759162798756084e-07,
      "loss": 0.3494,
      "num_input_tokens_seen": 19371248,
      "step": 29685
    },
    {
      "epoch": 17.505896226415093,
      "grad_norm": 3.540623188018799,
      "learning_rate": 4.6650591194551895e-07,
      "loss": 0.3489,
      "num_input_tokens_seen": 19373904,
      "step": 29690
    },
    {
      "epoch": 17.50884433962264,
      "grad_norm": 4.782562732696533,
      "learning_rate": 4.654213961655757e-07,
      "loss": 0.2702,
      "num_input_tokens_seen": 19377168,
      "step": 29695
    },
    {
      "epoch": 17.51179245283019,
      "grad_norm": 4.321146011352539,
      "learning_rate": 4.6433808093486075e-07,
      "loss": 0.2467,
      "num_input_tokens_seen": 19380560,
      "step": 29700
    },
    {
      "epoch": 17.514740566037737,
      "grad_norm": 3.540022850036621,
      "learning_rate": 4.63255966540187e-07,
      "loss": 0.3421,
      "num_input_tokens_seen": 19383728,
      "step": 29705
    },
    {
      "epoch": 17.517688679245282,
      "grad_norm": 4.388162136077881,
      "learning_rate": 4.62175053268048e-07,
      "loss": 0.3978,
      "num_input_tokens_seen": 19387248,
      "step": 29710
    },
    {
      "epoch": 17.52063679245283,
      "grad_norm": 4.140903472900391,
      "learning_rate": 4.6109534140462045e-07,
      "loss": 0.3003,
      "num_input_tokens_seen": 19389968,
      "step": 29715
    },
    {
      "epoch": 17.52358490566038,
      "grad_norm": 2.79817271232605,
      "learning_rate": 4.6001683123576226e-07,
      "loss": 0.2801,
      "num_input_tokens_seen": 19393232,
      "step": 29720
    },
    {
      "epoch": 17.526533018867923,
      "grad_norm": 7.1996917724609375,
      "learning_rate": 4.589395230470145e-07,
      "loss": 0.2981,
      "num_input_tokens_seen": 19396016,
      "step": 29725
    },
    {
      "epoch": 17.52948113207547,
      "grad_norm": 3.436506748199463,
      "learning_rate": 4.578634171235996e-07,
      "loss": 0.272,
      "num_input_tokens_seen": 19399248,
      "step": 29730
    },
    {
      "epoch": 17.53242924528302,
      "grad_norm": 5.760735034942627,
      "learning_rate": 4.567885137504202e-07,
      "loss": 0.4159,
      "num_input_tokens_seen": 19402544,
      "step": 29735
    },
    {
      "epoch": 17.535377358490567,
      "grad_norm": 5.711773872375488,
      "learning_rate": 4.55714813212062e-07,
      "loss": 0.2357,
      "num_input_tokens_seen": 19405264,
      "step": 29740
    },
    {
      "epoch": 17.538325471698112,
      "grad_norm": 4.368917465209961,
      "learning_rate": 4.5464231579279206e-07,
      "loss": 0.319,
      "num_input_tokens_seen": 19408720,
      "step": 29745
    },
    {
      "epoch": 17.54127358490566,
      "grad_norm": 1.6674206256866455,
      "learning_rate": 4.535710217765571e-07,
      "loss": 0.264,
      "num_input_tokens_seen": 19412304,
      "step": 29750
    },
    {
      "epoch": 17.54422169811321,
      "grad_norm": 3.3602254390716553,
      "learning_rate": 4.5250093144698913e-07,
      "loss": 0.2706,
      "num_input_tokens_seen": 19415216,
      "step": 29755
    },
    {
      "epoch": 17.547169811320753,
      "grad_norm": 4.324750900268555,
      "learning_rate": 4.514320450873988e-07,
      "loss": 0.2453,
      "num_input_tokens_seen": 19419600,
      "step": 29760
    },
    {
      "epoch": 17.5501179245283,
      "grad_norm": 6.401323318481445,
      "learning_rate": 4.503643629807769e-07,
      "loss": 0.2474,
      "num_input_tokens_seen": 19421872,
      "step": 29765
    },
    {
      "epoch": 17.55306603773585,
      "grad_norm": 2.8764748573303223,
      "learning_rate": 4.4929788540979844e-07,
      "loss": 0.3223,
      "num_input_tokens_seen": 19425136,
      "step": 29770
    },
    {
      "epoch": 17.556014150943398,
      "grad_norm": 8.648366928100586,
      "learning_rate": 4.4823261265681596e-07,
      "loss": 0.3028,
      "num_input_tokens_seen": 19430288,
      "step": 29775
    },
    {
      "epoch": 17.558962264150942,
      "grad_norm": 2.7787599563598633,
      "learning_rate": 4.471685450038671e-07,
      "loss": 0.3121,
      "num_input_tokens_seen": 19433392,
      "step": 29780
    },
    {
      "epoch": 17.56191037735849,
      "grad_norm": 3.8126487731933594,
      "learning_rate": 4.4610568273266706e-07,
      "loss": 0.2639,
      "num_input_tokens_seen": 19436752,
      "step": 29785
    },
    {
      "epoch": 17.56485849056604,
      "grad_norm": 2.481271982192993,
      "learning_rate": 4.450440261246142e-07,
      "loss": 0.291,
      "num_input_tokens_seen": 19440112,
      "step": 29790
    },
    {
      "epoch": 17.567806603773583,
      "grad_norm": 7.814016342163086,
      "learning_rate": 4.439835754607863e-07,
      "loss": 0.2637,
      "num_input_tokens_seen": 19443088,
      "step": 29795
    },
    {
      "epoch": 17.57075471698113,
      "grad_norm": 2.29601788520813,
      "learning_rate": 4.429243310219422e-07,
      "loss": 0.3194,
      "num_input_tokens_seen": 19447056,
      "step": 29800
    },
    {
      "epoch": 17.57370283018868,
      "grad_norm": 3.193683385848999,
      "learning_rate": 4.418662930885215e-07,
      "loss": 0.3207,
      "num_input_tokens_seen": 19450640,
      "step": 29805
    },
    {
      "epoch": 17.576650943396228,
      "grad_norm": 4.817287921905518,
      "learning_rate": 4.408094619406439e-07,
      "loss": 0.2822,
      "num_input_tokens_seen": 19453840,
      "step": 29810
    },
    {
      "epoch": 17.579599056603772,
      "grad_norm": 4.155791282653809,
      "learning_rate": 4.3975383785810954e-07,
      "loss": 0.2301,
      "num_input_tokens_seen": 19457424,
      "step": 29815
    },
    {
      "epoch": 17.58254716981132,
      "grad_norm": 3.9995663166046143,
      "learning_rate": 4.3869942112040096e-07,
      "loss": 0.3121,
      "num_input_tokens_seen": 19463376,
      "step": 29820
    },
    {
      "epoch": 17.58549528301887,
      "grad_norm": 3.8244709968566895,
      "learning_rate": 4.3764621200667936e-07,
      "loss": 0.3803,
      "num_input_tokens_seen": 19466736,
      "step": 29825
    },
    {
      "epoch": 17.588443396226417,
      "grad_norm": 4.089743614196777,
      "learning_rate": 4.365942107957849e-07,
      "loss": 0.2282,
      "num_input_tokens_seen": 19469168,
      "step": 29830
    },
    {
      "epoch": 17.59139150943396,
      "grad_norm": 8.447809219360352,
      "learning_rate": 4.35543417766241e-07,
      "loss": 0.2505,
      "num_input_tokens_seen": 19472688,
      "step": 29835
    },
    {
      "epoch": 17.59433962264151,
      "grad_norm": 7.290267467498779,
      "learning_rate": 4.3449383319624785e-07,
      "loss": 0.3478,
      "num_input_tokens_seen": 19476016,
      "step": 29840
    },
    {
      "epoch": 17.597287735849058,
      "grad_norm": 3.0142695903778076,
      "learning_rate": 4.3344545736368926e-07,
      "loss": 0.3952,
      "num_input_tokens_seen": 19479248,
      "step": 29845
    },
    {
      "epoch": 17.600235849056602,
      "grad_norm": 3.7160372734069824,
      "learning_rate": 4.323982905461266e-07,
      "loss": 0.3353,
      "num_input_tokens_seen": 19483152,
      "step": 29850
    },
    {
      "epoch": 17.60318396226415,
      "grad_norm": 3.0339345932006836,
      "learning_rate": 4.313523330208019e-07,
      "loss": 0.3586,
      "num_input_tokens_seen": 19486864,
      "step": 29855
    },
    {
      "epoch": 17.6061320754717,
      "grad_norm": 3.8709583282470703,
      "learning_rate": 4.303075850646371e-07,
      "loss": 0.2687,
      "num_input_tokens_seen": 19489680,
      "step": 29860
    },
    {
      "epoch": 17.609080188679247,
      "grad_norm": 4.574602127075195,
      "learning_rate": 4.2926404695423305e-07,
      "loss": 0.3526,
      "num_input_tokens_seen": 19492080,
      "step": 29865
    },
    {
      "epoch": 17.61202830188679,
      "grad_norm": 3.695984125137329,
      "learning_rate": 4.282217189658705e-07,
      "loss": 0.313,
      "num_input_tokens_seen": 19495184,
      "step": 29870
    },
    {
      "epoch": 17.61497641509434,
      "grad_norm": 4.408347129821777,
      "learning_rate": 4.27180601375512e-07,
      "loss": 0.354,
      "num_input_tokens_seen": 19498160,
      "step": 29875
    },
    {
      "epoch": 17.617924528301888,
      "grad_norm": 3.8960680961608887,
      "learning_rate": 4.2614069445879646e-07,
      "loss": 0.3712,
      "num_input_tokens_seen": 19501968,
      "step": 29880
    },
    {
      "epoch": 17.620872641509433,
      "grad_norm": 4.106062889099121,
      "learning_rate": 4.251019984910448e-07,
      "loss": 0.4004,
      "num_input_tokens_seen": 19505072,
      "step": 29885
    },
    {
      "epoch": 17.62382075471698,
      "grad_norm": 5.152200698852539,
      "learning_rate": 4.2406451374725597e-07,
      "loss": 0.3436,
      "num_input_tokens_seen": 19508016,
      "step": 29890
    },
    {
      "epoch": 17.62676886792453,
      "grad_norm": 2.682178020477295,
      "learning_rate": 4.2302824050210855e-07,
      "loss": 0.2946,
      "num_input_tokens_seen": 19510640,
      "step": 29895
    },
    {
      "epoch": 17.629716981132077,
      "grad_norm": 3.4632980823516846,
      "learning_rate": 4.2199317902995974e-07,
      "loss": 0.2977,
      "num_input_tokens_seen": 19515120,
      "step": 29900
    },
    {
      "epoch": 17.63266509433962,
      "grad_norm": 5.079971790313721,
      "learning_rate": 4.209593296048459e-07,
      "loss": 0.2225,
      "num_input_tokens_seen": 19517616,
      "step": 29905
    },
    {
      "epoch": 17.63561320754717,
      "grad_norm": 3.9023067951202393,
      "learning_rate": 4.1992669250048524e-07,
      "loss": 0.3369,
      "num_input_tokens_seen": 19520976,
      "step": 29910
    },
    {
      "epoch": 17.638561320754718,
      "grad_norm": 4.456829071044922,
      "learning_rate": 4.188952679902719e-07,
      "loss": 0.4101,
      "num_input_tokens_seen": 19523984,
      "step": 29915
    },
    {
      "epoch": 17.641509433962263,
      "grad_norm": 3.4006335735321045,
      "learning_rate": 4.178650563472797e-07,
      "loss": 0.3525,
      "num_input_tokens_seen": 19527600,
      "step": 29920
    },
    {
      "epoch": 17.64445754716981,
      "grad_norm": 2.8123559951782227,
      "learning_rate": 4.168360578442615e-07,
      "loss": 0.1958,
      "num_input_tokens_seen": 19531088,
      "step": 29925
    },
    {
      "epoch": 17.64740566037736,
      "grad_norm": 4.697080135345459,
      "learning_rate": 4.1580827275365e-07,
      "loss": 0.4181,
      "num_input_tokens_seen": 19534096,
      "step": 29930
    },
    {
      "epoch": 17.650353773584907,
      "grad_norm": 5.454930782318115,
      "learning_rate": 4.147817013475536e-07,
      "loss": 0.2538,
      "num_input_tokens_seen": 19537104,
      "step": 29935
    },
    {
      "epoch": 17.653301886792452,
      "grad_norm": 3.5876967906951904,
      "learning_rate": 4.1375634389776375e-07,
      "loss": 0.364,
      "num_input_tokens_seen": 19539792,
      "step": 29940
    },
    {
      "epoch": 17.65625,
      "grad_norm": 3.8499343395233154,
      "learning_rate": 4.127322006757478e-07,
      "loss": 0.3605,
      "num_input_tokens_seen": 19542736,
      "step": 29945
    },
    {
      "epoch": 17.659198113207548,
      "grad_norm": 2.883835554122925,
      "learning_rate": 4.1170927195265163e-07,
      "loss": 0.3084,
      "num_input_tokens_seen": 19545808,
      "step": 29950
    },
    {
      "epoch": 17.662146226415093,
      "grad_norm": 3.778754949569702,
      "learning_rate": 4.1068755799930026e-07,
      "loss": 0.3281,
      "num_input_tokens_seen": 19548560,
      "step": 29955
    },
    {
      "epoch": 17.66509433962264,
      "grad_norm": 6.076083660125732,
      "learning_rate": 4.096670590861962e-07,
      "loss": 0.2985,
      "num_input_tokens_seen": 19551152,
      "step": 29960
    },
    {
      "epoch": 17.66804245283019,
      "grad_norm": 3.78519344329834,
      "learning_rate": 4.086477754835211e-07,
      "loss": 0.2938,
      "num_input_tokens_seen": 19554096,
      "step": 29965
    },
    {
      "epoch": 17.670990566037737,
      "grad_norm": 3.2133800983428955,
      "learning_rate": 4.0762970746113517e-07,
      "loss": 0.2838,
      "num_input_tokens_seen": 19557712,
      "step": 29970
    },
    {
      "epoch": 17.673938679245282,
      "grad_norm": 3.8399910926818848,
      "learning_rate": 4.0661285528857676e-07,
      "loss": 0.3082,
      "num_input_tokens_seen": 19560752,
      "step": 29975
    },
    {
      "epoch": 17.67688679245283,
      "grad_norm": 3.958176612854004,
      "learning_rate": 4.0559721923506155e-07,
      "loss": 0.269,
      "num_input_tokens_seen": 19563952,
      "step": 29980
    },
    {
      "epoch": 17.67983490566038,
      "grad_norm": 5.1226325035095215,
      "learning_rate": 4.045827995694834e-07,
      "loss": 0.3325,
      "num_input_tokens_seen": 19566832,
      "step": 29985
    },
    {
      "epoch": 17.682783018867923,
      "grad_norm": 4.479787826538086,
      "learning_rate": 4.035695965604142e-07,
      "loss": 0.3582,
      "num_input_tokens_seen": 19569840,
      "step": 29990
    },
    {
      "epoch": 17.68573113207547,
      "grad_norm": 2.767713785171509,
      "learning_rate": 4.0255761047610365e-07,
      "loss": 0.2413,
      "num_input_tokens_seen": 19572944,
      "step": 29995
    },
    {
      "epoch": 17.68867924528302,
      "grad_norm": 6.384271144866943,
      "learning_rate": 4.0154684158447864e-07,
      "loss": 0.3416,
      "num_input_tokens_seen": 19576880,
      "step": 30000
    },
    {
      "epoch": 17.691627358490567,
      "grad_norm": 5.059001922607422,
      "learning_rate": 4.0053729015314623e-07,
      "loss": 0.2505,
      "num_input_tokens_seen": 19580240,
      "step": 30005
    },
    {
      "epoch": 17.694575471698112,
      "grad_norm": 5.936522483825684,
      "learning_rate": 3.9952895644938926e-07,
      "loss": 0.3053,
      "num_input_tokens_seen": 19583280,
      "step": 30010
    },
    {
      "epoch": 17.69752358490566,
      "grad_norm": 4.144394874572754,
      "learning_rate": 3.985218407401681e-07,
      "loss": 0.3301,
      "num_input_tokens_seen": 19586736,
      "step": 30015
    },
    {
      "epoch": 17.70047169811321,
      "grad_norm": 3.082507848739624,
      "learning_rate": 3.975159432921205e-07,
      "loss": 0.2569,
      "num_input_tokens_seen": 19590544,
      "step": 30020
    },
    {
      "epoch": 17.703419811320753,
      "grad_norm": 3.03660249710083,
      "learning_rate": 3.9651126437156294e-07,
      "loss": 0.2829,
      "num_input_tokens_seen": 19594160,
      "step": 30025
    },
    {
      "epoch": 17.7063679245283,
      "grad_norm": 5.3290581703186035,
      "learning_rate": 3.9550780424448653e-07,
      "loss": 0.2393,
      "num_input_tokens_seen": 19597296,
      "step": 30030
    },
    {
      "epoch": 17.70931603773585,
      "grad_norm": 2.8234241008758545,
      "learning_rate": 3.9450556317656487e-07,
      "loss": 0.3034,
      "num_input_tokens_seen": 19600976,
      "step": 30035
    },
    {
      "epoch": 17.712264150943398,
      "grad_norm": 5.621461391448975,
      "learning_rate": 3.935045414331434e-07,
      "loss": 0.3863,
      "num_input_tokens_seen": 19605008,
      "step": 30040
    },
    {
      "epoch": 17.715212264150942,
      "grad_norm": 2.3685083389282227,
      "learning_rate": 3.925047392792475e-07,
      "loss": 0.3222,
      "num_input_tokens_seen": 19608944,
      "step": 30045
    },
    {
      "epoch": 17.71816037735849,
      "grad_norm": 7.218699932098389,
      "learning_rate": 3.9150615697957917e-07,
      "loss": 0.2717,
      "num_input_tokens_seen": 19612016,
      "step": 30050
    },
    {
      "epoch": 17.72110849056604,
      "grad_norm": 3.70358943939209,
      "learning_rate": 3.9050879479851753e-07,
      "loss": 0.3504,
      "num_input_tokens_seen": 19615632,
      "step": 30055
    },
    {
      "epoch": 17.724056603773583,
      "grad_norm": 6.316326141357422,
      "learning_rate": 3.89512653000117e-07,
      "loss": 0.3382,
      "num_input_tokens_seen": 19618160,
      "step": 30060
    },
    {
      "epoch": 17.72700471698113,
      "grad_norm": 3.5059216022491455,
      "learning_rate": 3.8851773184811203e-07,
      "loss": 0.3233,
      "num_input_tokens_seen": 19620784,
      "step": 30065
    },
    {
      "epoch": 17.72995283018868,
      "grad_norm": 6.4526543617248535,
      "learning_rate": 3.8752403160591255e-07,
      "loss": 0.2703,
      "num_input_tokens_seen": 19623408,
      "step": 30070
    },
    {
      "epoch": 17.732900943396228,
      "grad_norm": 2.48557448387146,
      "learning_rate": 3.8653155253660477e-07,
      "loss": 0.3463,
      "num_input_tokens_seen": 19626832,
      "step": 30075
    },
    {
      "epoch": 17.735849056603772,
      "grad_norm": 3.0013647079467773,
      "learning_rate": 3.8554029490295073e-07,
      "loss": 0.2681,
      "num_input_tokens_seen": 19630224,
      "step": 30080
    },
    {
      "epoch": 17.73879716981132,
      "grad_norm": 2.65704345703125,
      "learning_rate": 3.8455025896739164e-07,
      "loss": 0.2114,
      "num_input_tokens_seen": 19632656,
      "step": 30085
    },
    {
      "epoch": 17.74174528301887,
      "grad_norm": 4.569005012512207,
      "learning_rate": 3.8356144499204215e-07,
      "loss": 0.3769,
      "num_input_tokens_seen": 19636400,
      "step": 30090
    },
    {
      "epoch": 17.744693396226417,
      "grad_norm": 3.1655654907226562,
      "learning_rate": 3.8257385323869576e-07,
      "loss": 0.3341,
      "num_input_tokens_seen": 19639440,
      "step": 30095
    },
    {
      "epoch": 17.74764150943396,
      "grad_norm": 2.605433940887451,
      "learning_rate": 3.815874839688222e-07,
      "loss": 0.2892,
      "num_input_tokens_seen": 19643536,
      "step": 30100
    },
    {
      "epoch": 17.75058962264151,
      "grad_norm": 4.5944390296936035,
      "learning_rate": 3.8060233744356634e-07,
      "loss": 0.2481,
      "num_input_tokens_seen": 19646224,
      "step": 30105
    },
    {
      "epoch": 17.753537735849058,
      "grad_norm": 3.501011371612549,
      "learning_rate": 3.796184139237502e-07,
      "loss": 0.2878,
      "num_input_tokens_seen": 19648848,
      "step": 30110
    },
    {
      "epoch": 17.756485849056602,
      "grad_norm": 3.2684073448181152,
      "learning_rate": 3.7863571366987206e-07,
      "loss": 0.2424,
      "num_input_tokens_seen": 19652560,
      "step": 30115
    },
    {
      "epoch": 17.75943396226415,
      "grad_norm": 3.3470709323883057,
      "learning_rate": 3.776542369421049e-07,
      "loss": 0.2921,
      "num_input_tokens_seen": 19655440,
      "step": 30120
    },
    {
      "epoch": 17.7623820754717,
      "grad_norm": 6.413597583770752,
      "learning_rate": 3.766739840003003e-07,
      "loss": 0.2471,
      "num_input_tokens_seen": 19657936,
      "step": 30125
    },
    {
      "epoch": 17.765330188679247,
      "grad_norm": 3.7638256549835205,
      "learning_rate": 3.756949551039835e-07,
      "loss": 0.3031,
      "num_input_tokens_seen": 19662544,
      "step": 30130
    },
    {
      "epoch": 17.76827830188679,
      "grad_norm": 4.620702743530273,
      "learning_rate": 3.7471715051235757e-07,
      "loss": 0.2436,
      "num_input_tokens_seen": 19666416,
      "step": 30135
    },
    {
      "epoch": 17.77122641509434,
      "grad_norm": 5.6916913986206055,
      "learning_rate": 3.7374057048429947e-07,
      "loss": 0.3651,
      "num_input_tokens_seen": 19669040,
      "step": 30140
    },
    {
      "epoch": 17.774174528301888,
      "grad_norm": 3.1047825813293457,
      "learning_rate": 3.7276521527836396e-07,
      "loss": 0.4039,
      "num_input_tokens_seen": 19673648,
      "step": 30145
    },
    {
      "epoch": 17.777122641509433,
      "grad_norm": 7.431186676025391,
      "learning_rate": 3.717910851527784e-07,
      "loss": 0.2903,
      "num_input_tokens_seen": 19676528,
      "step": 30150
    },
    {
      "epoch": 17.78007075471698,
      "grad_norm": 5.141274452209473,
      "learning_rate": 3.708181803654498e-07,
      "loss": 0.244,
      "num_input_tokens_seen": 19679696,
      "step": 30155
    },
    {
      "epoch": 17.78301886792453,
      "grad_norm": 3.8741374015808105,
      "learning_rate": 3.6984650117395993e-07,
      "loss": 0.2425,
      "num_input_tokens_seen": 19682512,
      "step": 30160
    },
    {
      "epoch": 17.785966981132077,
      "grad_norm": 1.933734655380249,
      "learning_rate": 3.688760478355635e-07,
      "loss": 0.2804,
      "num_input_tokens_seen": 19685552,
      "step": 30165
    },
    {
      "epoch": 17.78891509433962,
      "grad_norm": 5.843873977661133,
      "learning_rate": 3.679068206071923e-07,
      "loss": 0.3399,
      "num_input_tokens_seen": 19688048,
      "step": 30170
    },
    {
      "epoch": 17.79186320754717,
      "grad_norm": 4.484386920928955,
      "learning_rate": 3.669388197454532e-07,
      "loss": 0.2963,
      "num_input_tokens_seen": 19690960,
      "step": 30175
    },
    {
      "epoch": 17.794811320754718,
      "grad_norm": 3.630345582962036,
      "learning_rate": 3.6597204550662956e-07,
      "loss": 0.2896,
      "num_input_tokens_seen": 19693936,
      "step": 30180
    },
    {
      "epoch": 17.797759433962263,
      "grad_norm": 4.386622428894043,
      "learning_rate": 3.650064981466772e-07,
      "loss": 0.29,
      "num_input_tokens_seen": 19697072,
      "step": 30185
    },
    {
      "epoch": 17.80070754716981,
      "grad_norm": 3.656722068786621,
      "learning_rate": 3.640421779212311e-07,
      "loss": 0.2891,
      "num_input_tokens_seen": 19699696,
      "step": 30190
    },
    {
      "epoch": 17.80365566037736,
      "grad_norm": 5.5976972579956055,
      "learning_rate": 3.630790850855986e-07,
      "loss": 0.323,
      "num_input_tokens_seen": 19702928,
      "step": 30195
    },
    {
      "epoch": 17.806603773584907,
      "grad_norm": 4.749702453613281,
      "learning_rate": 3.62117219894762e-07,
      "loss": 0.2993,
      "num_input_tokens_seen": 19706064,
      "step": 30200
    },
    {
      "epoch": 17.809551886792452,
      "grad_norm": 7.9056782722473145,
      "learning_rate": 3.611565826033797e-07,
      "loss": 0.4681,
      "num_input_tokens_seen": 19709520,
      "step": 30205
    },
    {
      "epoch": 17.8125,
      "grad_norm": 4.5009002685546875,
      "learning_rate": 3.6019717346578445e-07,
      "loss": 0.3719,
      "num_input_tokens_seen": 19716112,
      "step": 30210
    },
    {
      "epoch": 17.815448113207548,
      "grad_norm": 8.460150718688965,
      "learning_rate": 3.5923899273598293e-07,
      "loss": 0.3702,
      "num_input_tokens_seen": 19719056,
      "step": 30215
    },
    {
      "epoch": 17.818396226415093,
      "grad_norm": 3.733607530593872,
      "learning_rate": 3.582820406676596e-07,
      "loss": 0.2669,
      "num_input_tokens_seen": 19721456,
      "step": 30220
    },
    {
      "epoch": 17.82134433962264,
      "grad_norm": 6.31386137008667,
      "learning_rate": 3.5732631751417056e-07,
      "loss": 0.3417,
      "num_input_tokens_seen": 19724432,
      "step": 30225
    },
    {
      "epoch": 17.82429245283019,
      "grad_norm": 3.8544058799743652,
      "learning_rate": 3.563718235285485e-07,
      "loss": 0.356,
      "num_input_tokens_seen": 19728496,
      "step": 30230
    },
    {
      "epoch": 17.827240566037737,
      "grad_norm": 5.316858768463135,
      "learning_rate": 3.5541855896349844e-07,
      "loss": 0.2916,
      "num_input_tokens_seen": 19731120,
      "step": 30235
    },
    {
      "epoch": 17.830188679245282,
      "grad_norm": 4.305591583251953,
      "learning_rate": 3.544665240714018e-07,
      "loss": 0.2883,
      "num_input_tokens_seen": 19734192,
      "step": 30240
    },
    {
      "epoch": 17.83313679245283,
      "grad_norm": 5.590890884399414,
      "learning_rate": 3.535157191043137e-07,
      "loss": 0.4068,
      "num_input_tokens_seen": 19737104,
      "step": 30245
    },
    {
      "epoch": 17.83608490566038,
      "grad_norm": 5.575997352600098,
      "learning_rate": 3.5256614431396385e-07,
      "loss": 0.3138,
      "num_input_tokens_seen": 19742480,
      "step": 30250
    },
    {
      "epoch": 17.839033018867923,
      "grad_norm": 4.585902690887451,
      "learning_rate": 3.516177999517578e-07,
      "loss": 0.2504,
      "num_input_tokens_seen": 19745040,
      "step": 30255
    },
    {
      "epoch": 17.84198113207547,
      "grad_norm": 2.8222391605377197,
      "learning_rate": 3.50670686268772e-07,
      "loss": 0.2715,
      "num_input_tokens_seen": 19747664,
      "step": 30260
    },
    {
      "epoch": 17.84492924528302,
      "grad_norm": 5.196811199188232,
      "learning_rate": 3.497248035157602e-07,
      "loss": 0.3268,
      "num_input_tokens_seen": 19751184,
      "step": 30265
    },
    {
      "epoch": 17.847877358490567,
      "grad_norm": 4.497313022613525,
      "learning_rate": 3.4878015194314773e-07,
      "loss": 0.4222,
      "num_input_tokens_seen": 19754800,
      "step": 30270
    },
    {
      "epoch": 17.850825471698112,
      "grad_norm": 3.775904655456543,
      "learning_rate": 3.4783673180103617e-07,
      "loss": 0.3148,
      "num_input_tokens_seen": 19757904,
      "step": 30275
    },
    {
      "epoch": 17.85377358490566,
      "grad_norm": 2.69215989112854,
      "learning_rate": 3.468945433391985e-07,
      "loss": 0.5146,
      "num_input_tokens_seen": 19761392,
      "step": 30280
    },
    {
      "epoch": 17.85672169811321,
      "grad_norm": 5.309360980987549,
      "learning_rate": 3.459535868070851e-07,
      "loss": 0.2667,
      "num_input_tokens_seen": 19763568,
      "step": 30285
    },
    {
      "epoch": 17.859669811320753,
      "grad_norm": 7.470308303833008,
      "learning_rate": 3.450138624538174e-07,
      "loss": 0.3641,
      "num_input_tokens_seen": 19766704,
      "step": 30290
    },
    {
      "epoch": 17.8626179245283,
      "grad_norm": 3.211538553237915,
      "learning_rate": 3.440753705281913e-07,
      "loss": 0.2482,
      "num_input_tokens_seen": 19769904,
      "step": 30295
    },
    {
      "epoch": 17.86556603773585,
      "grad_norm": 5.239882469177246,
      "learning_rate": 3.4313811127867693e-07,
      "loss": 0.3732,
      "num_input_tokens_seen": 19772720,
      "step": 30300
    },
    {
      "epoch": 17.868514150943398,
      "grad_norm": 3.390812397003174,
      "learning_rate": 3.4220208495341745e-07,
      "loss": 0.3046,
      "num_input_tokens_seen": 19776144,
      "step": 30305
    },
    {
      "epoch": 17.871462264150942,
      "grad_norm": 2.789263963699341,
      "learning_rate": 3.412672918002291e-07,
      "loss": 0.4016,
      "num_input_tokens_seen": 19779824,
      "step": 30310
    },
    {
      "epoch": 17.87441037735849,
      "grad_norm": 3.523062229156494,
      "learning_rate": 3.403337320666045e-07,
      "loss": 0.236,
      "num_input_tokens_seen": 19782832,
      "step": 30315
    },
    {
      "epoch": 17.87735849056604,
      "grad_norm": 3.138392925262451,
      "learning_rate": 3.394014059997064e-07,
      "loss": 0.2286,
      "num_input_tokens_seen": 19786192,
      "step": 30320
    },
    {
      "epoch": 17.880306603773583,
      "grad_norm": 4.906579494476318,
      "learning_rate": 3.3847031384637185e-07,
      "loss": 0.3561,
      "num_input_tokens_seen": 19790736,
      "step": 30325
    },
    {
      "epoch": 17.88325471698113,
      "grad_norm": 3.9168195724487305,
      "learning_rate": 3.3754045585311147e-07,
      "loss": 0.3541,
      "num_input_tokens_seen": 19793584,
      "step": 30330
    },
    {
      "epoch": 17.88620283018868,
      "grad_norm": 2.9901039600372314,
      "learning_rate": 3.366118322661094e-07,
      "loss": 0.2574,
      "num_input_tokens_seen": 19798832,
      "step": 30335
    },
    {
      "epoch": 17.889150943396228,
      "grad_norm": 3.9961998462677,
      "learning_rate": 3.3568444333122283e-07,
      "loss": 0.3633,
      "num_input_tokens_seen": 19801936,
      "step": 30340
    },
    {
      "epoch": 17.892099056603772,
      "grad_norm": 4.012005805969238,
      "learning_rate": 3.347582892939816e-07,
      "loss": 0.3435,
      "num_input_tokens_seen": 19805648,
      "step": 30345
    },
    {
      "epoch": 17.89504716981132,
      "grad_norm": 7.200450897216797,
      "learning_rate": 3.338333703995905e-07,
      "loss": 0.4421,
      "num_input_tokens_seen": 19808240,
      "step": 30350
    },
    {
      "epoch": 17.89799528301887,
      "grad_norm": 4.785140514373779,
      "learning_rate": 3.329096868929238e-07,
      "loss": 0.3599,
      "num_input_tokens_seen": 19811184,
      "step": 30355
    },
    {
      "epoch": 17.900943396226417,
      "grad_norm": 5.302978515625,
      "learning_rate": 3.319872390185325e-07,
      "loss": 0.2981,
      "num_input_tokens_seen": 19814064,
      "step": 30360
    },
    {
      "epoch": 17.90389150943396,
      "grad_norm": 6.354378700256348,
      "learning_rate": 3.3106602702063727e-07,
      "loss": 0.3339,
      "num_input_tokens_seen": 19816976,
      "step": 30365
    },
    {
      "epoch": 17.90683962264151,
      "grad_norm": 2.7377803325653076,
      "learning_rate": 3.3014605114313316e-07,
      "loss": 0.2165,
      "num_input_tokens_seen": 19820240,
      "step": 30370
    },
    {
      "epoch": 17.909787735849058,
      "grad_norm": 2.375612735748291,
      "learning_rate": 3.2922731162958744e-07,
      "loss": 0.2161,
      "num_input_tokens_seen": 19822608,
      "step": 30375
    },
    {
      "epoch": 17.912735849056602,
      "grad_norm": 3.5113744735717773,
      "learning_rate": 3.2830980872324114e-07,
      "loss": 0.289,
      "num_input_tokens_seen": 19825744,
      "step": 30380
    },
    {
      "epoch": 17.91568396226415,
      "grad_norm": 1.9386707544326782,
      "learning_rate": 3.2739354266700775e-07,
      "loss": 0.1966,
      "num_input_tokens_seen": 19829264,
      "step": 30385
    },
    {
      "epoch": 17.9186320754717,
      "grad_norm": 4.5147247314453125,
      "learning_rate": 3.264785137034709e-07,
      "loss": 0.2513,
      "num_input_tokens_seen": 19831728,
      "step": 30390
    },
    {
      "epoch": 17.921580188679247,
      "grad_norm": 3.4204776287078857,
      "learning_rate": 3.2556472207488977e-07,
      "loss": 0.2735,
      "num_input_tokens_seen": 19835664,
      "step": 30395
    },
    {
      "epoch": 17.92452830188679,
      "grad_norm": 2.9941046237945557,
      "learning_rate": 3.246521680231934e-07,
      "loss": 0.3221,
      "num_input_tokens_seen": 19838672,
      "step": 30400
    },
    {
      "epoch": 17.92747641509434,
      "grad_norm": 3.2318358421325684,
      "learning_rate": 3.2374085178998594e-07,
      "loss": 0.3461,
      "num_input_tokens_seen": 19841808,
      "step": 30405
    },
    {
      "epoch": 17.930424528301888,
      "grad_norm": 3.9075708389282227,
      "learning_rate": 3.2283077361654145e-07,
      "loss": 0.272,
      "num_input_tokens_seen": 19844528,
      "step": 30410
    },
    {
      "epoch": 17.933372641509433,
      "grad_norm": 3.9072682857513428,
      "learning_rate": 3.2192193374380677e-07,
      "loss": 0.2614,
      "num_input_tokens_seen": 19849360,
      "step": 30415
    },
    {
      "epoch": 17.93632075471698,
      "grad_norm": 4.664844989776611,
      "learning_rate": 3.210143324124021e-07,
      "loss": 0.2683,
      "num_input_tokens_seen": 19856400,
      "step": 30420
    },
    {
      "epoch": 17.93926886792453,
      "grad_norm": 4.720576763153076,
      "learning_rate": 3.2010796986261805e-07,
      "loss": 0.4036,
      "num_input_tokens_seen": 19860848,
      "step": 30425
    },
    {
      "epoch": 17.942216981132077,
      "grad_norm": 4.2414398193359375,
      "learning_rate": 3.1920284633441713e-07,
      "loss": 0.2543,
      "num_input_tokens_seen": 19863600,
      "step": 30430
    },
    {
      "epoch": 17.94516509433962,
      "grad_norm": 4.348722457885742,
      "learning_rate": 3.1829896206743704e-07,
      "loss": 0.2435,
      "num_input_tokens_seen": 19866320,
      "step": 30435
    },
    {
      "epoch": 17.94811320754717,
      "grad_norm": 3.7772605419158936,
      "learning_rate": 3.173963173009825e-07,
      "loss": 0.2821,
      "num_input_tokens_seen": 19869040,
      "step": 30440
    },
    {
      "epoch": 17.951061320754718,
      "grad_norm": 4.74680233001709,
      "learning_rate": 3.164949122740352e-07,
      "loss": 0.3873,
      "num_input_tokens_seen": 19871920,
      "step": 30445
    },
    {
      "epoch": 17.954009433962263,
      "grad_norm": 4.48148250579834,
      "learning_rate": 3.1559474722524406e-07,
      "loss": 0.3104,
      "num_input_tokens_seen": 19874704,
      "step": 30450
    },
    {
      "epoch": 17.95695754716981,
      "grad_norm": 3.523751974105835,
      "learning_rate": 3.146958223929325e-07,
      "loss": 0.2786,
      "num_input_tokens_seen": 19877520,
      "step": 30455
    },
    {
      "epoch": 17.95990566037736,
      "grad_norm": 3.61141300201416,
      "learning_rate": 3.1379813801509454e-07,
      "loss": 0.452,
      "num_input_tokens_seen": 19881840,
      "step": 30460
    },
    {
      "epoch": 17.962853773584907,
      "grad_norm": 2.4816577434539795,
      "learning_rate": 3.1290169432939556e-07,
      "loss": 0.2294,
      "num_input_tokens_seen": 19885712,
      "step": 30465
    },
    {
      "epoch": 17.965801886792452,
      "grad_norm": 6.697802543640137,
      "learning_rate": 3.120064915731735e-07,
      "loss": 0.3337,
      "num_input_tokens_seen": 19889328,
      "step": 30470
    },
    {
      "epoch": 17.96875,
      "grad_norm": 2.0659384727478027,
      "learning_rate": 3.1111252998343723e-07,
      "loss": 0.2596,
      "num_input_tokens_seen": 19892944,
      "step": 30475
    },
    {
      "epoch": 17.971698113207548,
      "grad_norm": 4.429078102111816,
      "learning_rate": 3.102198097968662e-07,
      "loss": 0.3538,
      "num_input_tokens_seen": 19896144,
      "step": 30480
    },
    {
      "epoch": 17.974646226415093,
      "grad_norm": 4.173407077789307,
      "learning_rate": 3.093283312498124e-07,
      "loss": 0.2861,
      "num_input_tokens_seen": 19899056,
      "step": 30485
    },
    {
      "epoch": 17.97759433962264,
      "grad_norm": 3.0966267585754395,
      "learning_rate": 3.084380945782989e-07,
      "loss": 0.3174,
      "num_input_tokens_seen": 19902064,
      "step": 30490
    },
    {
      "epoch": 17.98054245283019,
      "grad_norm": 3.1534929275512695,
      "learning_rate": 3.0754910001801866e-07,
      "loss": 0.2133,
      "num_input_tokens_seen": 19905616,
      "step": 30495
    },
    {
      "epoch": 17.983490566037737,
      "grad_norm": 5.631502628326416,
      "learning_rate": 3.0666134780433786e-07,
      "loss": 0.2998,
      "num_input_tokens_seen": 19908336,
      "step": 30500
    },
    {
      "epoch": 17.986438679245282,
      "grad_norm": 7.465195655822754,
      "learning_rate": 3.0577483817229306e-07,
      "loss": 0.3673,
      "num_input_tokens_seen": 19911152,
      "step": 30505
    },
    {
      "epoch": 17.98938679245283,
      "grad_norm": 4.607506275177002,
      "learning_rate": 3.0488957135659023e-07,
      "loss": 0.3242,
      "num_input_tokens_seen": 19914640,
      "step": 30510
    },
    {
      "epoch": 17.99233490566038,
      "grad_norm": 3.0459094047546387,
      "learning_rate": 3.040055475916087e-07,
      "loss": 0.2564,
      "num_input_tokens_seen": 19917936,
      "step": 30515
    },
    {
      "epoch": 17.995283018867923,
      "grad_norm": 3.427809953689575,
      "learning_rate": 3.0312276711139675e-07,
      "loss": 0.328,
      "num_input_tokens_seen": 19921776,
      "step": 30520
    },
    {
      "epoch": 17.99823113207547,
      "grad_norm": 2.3180325031280518,
      "learning_rate": 3.0224123014967353e-07,
      "loss": 0.3527,
      "num_input_tokens_seen": 19924848,
      "step": 30525
    },
    {
      "epoch": 18.0,
      "eval_loss": 0.6273570656776428,
      "eval_runtime": 19.3797,
      "eval_samples_per_second": 87.514,
      "eval_steps_per_second": 21.879,
      "num_input_tokens_seen": 19926608,
      "step": 30528
    },
    {
      "epoch": 18.00117924528302,
      "grad_norm": 5.795991897583008,
      "learning_rate": 3.013609369398324e-07,
      "loss": 0.2776,
      "num_input_tokens_seen": 19927888,
      "step": 30530
    },
    {
      "epoch": 18.004127358490567,
      "grad_norm": 3.457123041152954,
      "learning_rate": 3.004818877149318e-07,
      "loss": 0.2073,
      "num_input_tokens_seen": 19930608,
      "step": 30535
    },
    {
      "epoch": 18.007075471698112,
      "grad_norm": 2.05129075050354,
      "learning_rate": 2.9960408270770624e-07,
      "loss": 0.3069,
      "num_input_tokens_seen": 19933872,
      "step": 30540
    },
    {
      "epoch": 18.01002358490566,
      "grad_norm": 4.635818004608154,
      "learning_rate": 2.9872752215055755e-07,
      "loss": 0.3925,
      "num_input_tokens_seen": 19936944,
      "step": 30545
    },
    {
      "epoch": 18.01297169811321,
      "grad_norm": 5.408751010894775,
      "learning_rate": 2.9785220627555844e-07,
      "loss": 0.3283,
      "num_input_tokens_seen": 19939536,
      "step": 30550
    },
    {
      "epoch": 18.015919811320753,
      "grad_norm": 4.959076404571533,
      "learning_rate": 2.9697813531445295e-07,
      "loss": 0.3485,
      "num_input_tokens_seen": 19942512,
      "step": 30555
    },
    {
      "epoch": 18.0188679245283,
      "grad_norm": 4.398247241973877,
      "learning_rate": 2.9610530949865433e-07,
      "loss": 0.2613,
      "num_input_tokens_seen": 19945840,
      "step": 30560
    },
    {
      "epoch": 18.02181603773585,
      "grad_norm": 3.203413486480713,
      "learning_rate": 2.952337290592483e-07,
      "loss": 0.3541,
      "num_input_tokens_seen": 19948688,
      "step": 30565
    },
    {
      "epoch": 18.024764150943398,
      "grad_norm": 3.4848127365112305,
      "learning_rate": 2.9436339422698913e-07,
      "loss": 0.2613,
      "num_input_tokens_seen": 19952464,
      "step": 30570
    },
    {
      "epoch": 18.027712264150942,
      "grad_norm": 3.749217987060547,
      "learning_rate": 2.934943052323008e-07,
      "loss": 0.359,
      "num_input_tokens_seen": 19956112,
      "step": 30575
    },
    {
      "epoch": 18.03066037735849,
      "grad_norm": 4.3835768699646,
      "learning_rate": 2.926264623052799e-07,
      "loss": 0.2984,
      "num_input_tokens_seen": 19959472,
      "step": 30580
    },
    {
      "epoch": 18.03360849056604,
      "grad_norm": 4.386301040649414,
      "learning_rate": 2.9175986567569036e-07,
      "loss": 0.2322,
      "num_input_tokens_seen": 19962224,
      "step": 30585
    },
    {
      "epoch": 18.036556603773583,
      "grad_norm": 7.671792507171631,
      "learning_rate": 2.9089451557296755e-07,
      "loss": 0.259,
      "num_input_tokens_seen": 19965008,
      "step": 30590
    },
    {
      "epoch": 18.03950471698113,
      "grad_norm": 3.2223095893859863,
      "learning_rate": 2.9003041222621706e-07,
      "loss": 0.4284,
      "num_input_tokens_seen": 19968624,
      "step": 30595
    },
    {
      "epoch": 18.04245283018868,
      "grad_norm": 3.1935136318206787,
      "learning_rate": 2.8916755586421375e-07,
      "loss": 0.3152,
      "num_input_tokens_seen": 19971728,
      "step": 30600
    },
    {
      "epoch": 18.045400943396228,
      "grad_norm": 2.774658441543579,
      "learning_rate": 2.883059467154031e-07,
      "loss": 0.2578,
      "num_input_tokens_seen": 19974768,
      "step": 30605
    },
    {
      "epoch": 18.048349056603772,
      "grad_norm": 3.2750120162963867,
      "learning_rate": 2.8744558500789887e-07,
      "loss": 0.3173,
      "num_input_tokens_seen": 19978384,
      "step": 30610
    },
    {
      "epoch": 18.05129716981132,
      "grad_norm": 5.315641403198242,
      "learning_rate": 2.8658647096948546e-07,
      "loss": 0.4031,
      "num_input_tokens_seen": 19981872,
      "step": 30615
    },
    {
      "epoch": 18.05424528301887,
      "grad_norm": 7.541603088378906,
      "learning_rate": 2.8572860482761813e-07,
      "loss": 0.2739,
      "num_input_tokens_seen": 19985232,
      "step": 30620
    },
    {
      "epoch": 18.057193396226417,
      "grad_norm": 3.4834625720977783,
      "learning_rate": 2.8487198680942017e-07,
      "loss": 0.5049,
      "num_input_tokens_seen": 19987824,
      "step": 30625
    },
    {
      "epoch": 18.06014150943396,
      "grad_norm": 2.477468490600586,
      "learning_rate": 2.840166171416836e-07,
      "loss": 0.2201,
      "num_input_tokens_seen": 19991184,
      "step": 30630
    },
    {
      "epoch": 18.06308962264151,
      "grad_norm": 3.4069912433624268,
      "learning_rate": 2.8316249605087386e-07,
      "loss": 0.2935,
      "num_input_tokens_seen": 19994672,
      "step": 30635
    },
    {
      "epoch": 18.066037735849058,
      "grad_norm": 4.839682579040527,
      "learning_rate": 2.823096237631212e-07,
      "loss": 0.2354,
      "num_input_tokens_seen": 19997744,
      "step": 30640
    },
    {
      "epoch": 18.068985849056602,
      "grad_norm": 3.7567875385284424,
      "learning_rate": 2.814580005042283e-07,
      "loss": 0.2568,
      "num_input_tokens_seen": 20001904,
      "step": 30645
    },
    {
      "epoch": 18.07193396226415,
      "grad_norm": 3.836216449737549,
      "learning_rate": 2.8060762649966435e-07,
      "loss": 0.2637,
      "num_input_tokens_seen": 20004624,
      "step": 30650
    },
    {
      "epoch": 18.0748820754717,
      "grad_norm": 5.0879082679748535,
      "learning_rate": 2.797585019745713e-07,
      "loss": 0.3086,
      "num_input_tokens_seen": 20007824,
      "step": 30655
    },
    {
      "epoch": 18.077830188679247,
      "grad_norm": 3.0640363693237305,
      "learning_rate": 2.789106271537584e-07,
      "loss": 0.2621,
      "num_input_tokens_seen": 20010896,
      "step": 30660
    },
    {
      "epoch": 18.08077830188679,
      "grad_norm": 3.1409871578216553,
      "learning_rate": 2.780640022617037e-07,
      "loss": 0.1847,
      "num_input_tokens_seen": 20014160,
      "step": 30665
    },
    {
      "epoch": 18.08372641509434,
      "grad_norm": 2.538684844970703,
      "learning_rate": 2.772186275225547e-07,
      "loss": 0.2586,
      "num_input_tokens_seen": 20017968,
      "step": 30670
    },
    {
      "epoch": 18.086674528301888,
      "grad_norm": 3.700963258743286,
      "learning_rate": 2.7637450316012836e-07,
      "loss": 0.2762,
      "num_input_tokens_seen": 20020752,
      "step": 30675
    },
    {
      "epoch": 18.089622641509433,
      "grad_norm": 5.596766948699951,
      "learning_rate": 2.755316293979088e-07,
      "loss": 0.2624,
      "num_input_tokens_seen": 20024688,
      "step": 30680
    },
    {
      "epoch": 18.09257075471698,
      "grad_norm": 4.245615482330322,
      "learning_rate": 2.7469000645905295e-07,
      "loss": 0.3607,
      "num_input_tokens_seen": 20028016,
      "step": 30685
    },
    {
      "epoch": 18.09551886792453,
      "grad_norm": 4.443362236022949,
      "learning_rate": 2.738496345663827e-07,
      "loss": 0.3788,
      "num_input_tokens_seen": 20031248,
      "step": 30690
    },
    {
      "epoch": 18.098466981132077,
      "grad_norm": 4.10257625579834,
      "learning_rate": 2.7301051394239e-07,
      "loss": 0.2775,
      "num_input_tokens_seen": 20034128,
      "step": 30695
    },
    {
      "epoch": 18.10141509433962,
      "grad_norm": 2.2175498008728027,
      "learning_rate": 2.72172644809236e-07,
      "loss": 0.1811,
      "num_input_tokens_seen": 20038032,
      "step": 30700
    },
    {
      "epoch": 18.10436320754717,
      "grad_norm": 3.892087697982788,
      "learning_rate": 2.7133602738875e-07,
      "loss": 0.3259,
      "num_input_tokens_seen": 20040880,
      "step": 30705
    },
    {
      "epoch": 18.107311320754718,
      "grad_norm": 4.485229969024658,
      "learning_rate": 2.7050066190242976e-07,
      "loss": 0.3329,
      "num_input_tokens_seen": 20044720,
      "step": 30710
    },
    {
      "epoch": 18.110259433962263,
      "grad_norm": 3.0180797576904297,
      "learning_rate": 2.696665485714428e-07,
      "loss": 0.2293,
      "num_input_tokens_seen": 20047888,
      "step": 30715
    },
    {
      "epoch": 18.11320754716981,
      "grad_norm": 3.1001508235931396,
      "learning_rate": 2.6883368761662367e-07,
      "loss": 0.2933,
      "num_input_tokens_seen": 20050896,
      "step": 30720
    },
    {
      "epoch": 18.11615566037736,
      "grad_norm": 4.274606704711914,
      "learning_rate": 2.680020792584759e-07,
      "loss": 0.321,
      "num_input_tokens_seen": 20053648,
      "step": 30725
    },
    {
      "epoch": 18.119103773584907,
      "grad_norm": 6.149851322174072,
      "learning_rate": 2.6717172371717113e-07,
      "loss": 0.2552,
      "num_input_tokens_seen": 20056368,
      "step": 30730
    },
    {
      "epoch": 18.122051886792452,
      "grad_norm": 5.6037468910217285,
      "learning_rate": 2.663426212125503e-07,
      "loss": 0.3099,
      "num_input_tokens_seen": 20058672,
      "step": 30735
    },
    {
      "epoch": 18.125,
      "grad_norm": 3.153165817260742,
      "learning_rate": 2.655147719641216e-07,
      "loss": 0.4385,
      "num_input_tokens_seen": 20061936,
      "step": 30740
    },
    {
      "epoch": 18.127948113207548,
      "grad_norm": 4.360229969024658,
      "learning_rate": 2.646881761910602e-07,
      "loss": 0.2967,
      "num_input_tokens_seen": 20064560,
      "step": 30745
    },
    {
      "epoch": 18.130896226415093,
      "grad_norm": 3.8156898021698,
      "learning_rate": 2.638628341122135e-07,
      "loss": 0.3224,
      "num_input_tokens_seen": 20068112,
      "step": 30750
    },
    {
      "epoch": 18.13384433962264,
      "grad_norm": 6.548239707946777,
      "learning_rate": 2.6303874594609314e-07,
      "loss": 0.2529,
      "num_input_tokens_seen": 20070960,
      "step": 30755
    },
    {
      "epoch": 18.13679245283019,
      "grad_norm": 5.625372886657715,
      "learning_rate": 2.622159119108797e-07,
      "loss": 0.2808,
      "num_input_tokens_seen": 20074224,
      "step": 30760
    },
    {
      "epoch": 18.139740566037737,
      "grad_norm": 2.3498785495758057,
      "learning_rate": 2.6139433222442226e-07,
      "loss": 0.2716,
      "num_input_tokens_seen": 20077648,
      "step": 30765
    },
    {
      "epoch": 18.142688679245282,
      "grad_norm": 3.6622071266174316,
      "learning_rate": 2.6057400710423787e-07,
      "loss": 0.181,
      "num_input_tokens_seen": 20081104,
      "step": 30770
    },
    {
      "epoch": 18.14563679245283,
      "grad_norm": 3.964223623275757,
      "learning_rate": 2.5975493676751004e-07,
      "loss": 0.3386,
      "num_input_tokens_seen": 20083728,
      "step": 30775
    },
    {
      "epoch": 18.14858490566038,
      "grad_norm": 2.6674911975860596,
      "learning_rate": 2.589371214310926e-07,
      "loss": 0.3307,
      "num_input_tokens_seen": 20086832,
      "step": 30780
    },
    {
      "epoch": 18.151533018867923,
      "grad_norm": 3.323371410369873,
      "learning_rate": 2.581205613115051e-07,
      "loss": 0.2589,
      "num_input_tokens_seen": 20090992,
      "step": 30785
    },
    {
      "epoch": 18.15448113207547,
      "grad_norm": 4.968255519866943,
      "learning_rate": 2.573052566249357e-07,
      "loss": 0.3204,
      "num_input_tokens_seen": 20093776,
      "step": 30790
    },
    {
      "epoch": 18.15742924528302,
      "grad_norm": 3.695032835006714,
      "learning_rate": 2.5649120758723945e-07,
      "loss": 0.3867,
      "num_input_tokens_seen": 20097232,
      "step": 30795
    },
    {
      "epoch": 18.160377358490567,
      "grad_norm": 2.4645185470581055,
      "learning_rate": 2.5567841441393906e-07,
      "loss": 0.2212,
      "num_input_tokens_seen": 20101328,
      "step": 30800
    },
    {
      "epoch": 18.163325471698112,
      "grad_norm": 3.7863998413085938,
      "learning_rate": 2.548668773202245e-07,
      "loss": 0.2554,
      "num_input_tokens_seen": 20103952,
      "step": 30805
    },
    {
      "epoch": 18.16627358490566,
      "grad_norm": 3.025130033493042,
      "learning_rate": 2.5405659652095573e-07,
      "loss": 0.365,
      "num_input_tokens_seen": 20106736,
      "step": 30810
    },
    {
      "epoch": 18.16922169811321,
      "grad_norm": 3.5622360706329346,
      "learning_rate": 2.5324757223065655e-07,
      "loss": 0.2938,
      "num_input_tokens_seen": 20109808,
      "step": 30815
    },
    {
      "epoch": 18.172169811320753,
      "grad_norm": 5.007580757141113,
      "learning_rate": 2.524398046635207e-07,
      "loss": 0.5024,
      "num_input_tokens_seen": 20113648,
      "step": 30820
    },
    {
      "epoch": 18.1751179245283,
      "grad_norm": 2.71625018119812,
      "learning_rate": 2.51633294033406e-07,
      "loss": 0.2344,
      "num_input_tokens_seen": 20117328,
      "step": 30825
    },
    {
      "epoch": 18.17806603773585,
      "grad_norm": 3.0770347118377686,
      "learning_rate": 2.5082804055384214e-07,
      "loss": 0.27,
      "num_input_tokens_seen": 20120368,
      "step": 30830
    },
    {
      "epoch": 18.181014150943398,
      "grad_norm": 4.538959503173828,
      "learning_rate": 2.50024044438022e-07,
      "loss": 0.3116,
      "num_input_tokens_seen": 20122928,
      "step": 30835
    },
    {
      "epoch": 18.183962264150942,
      "grad_norm": 3.5060131549835205,
      "learning_rate": 2.492213058988069e-07,
      "loss": 0.2658,
      "num_input_tokens_seen": 20126384,
      "step": 30840
    },
    {
      "epoch": 18.18691037735849,
      "grad_norm": 3.048906087875366,
      "learning_rate": 2.4841982514872633e-07,
      "loss": 0.1827,
      "num_input_tokens_seen": 20129488,
      "step": 30845
    },
    {
      "epoch": 18.18985849056604,
      "grad_norm": 6.240050792694092,
      "learning_rate": 2.4761960239997497e-07,
      "loss": 0.3116,
      "num_input_tokens_seen": 20132176,
      "step": 30850
    },
    {
      "epoch": 18.192806603773583,
      "grad_norm": 3.786304473876953,
      "learning_rate": 2.4682063786441556e-07,
      "loss": 0.3685,
      "num_input_tokens_seen": 20135696,
      "step": 30855
    },
    {
      "epoch": 18.19575471698113,
      "grad_norm": 3.792616367340088,
      "learning_rate": 2.460229317535778e-07,
      "loss": 0.3455,
      "num_input_tokens_seen": 20138576,
      "step": 30860
    },
    {
      "epoch": 18.19870283018868,
      "grad_norm": 2.999385356903076,
      "learning_rate": 2.4522648427865725e-07,
      "loss": 0.3234,
      "num_input_tokens_seen": 20142256,
      "step": 30865
    },
    {
      "epoch": 18.201650943396228,
      "grad_norm": 2.3591747283935547,
      "learning_rate": 2.444312956505163e-07,
      "loss": 0.285,
      "num_input_tokens_seen": 20145808,
      "step": 30870
    },
    {
      "epoch": 18.204599056603772,
      "grad_norm": 2.9386510848999023,
      "learning_rate": 2.4363736607968537e-07,
      "loss": 0.4278,
      "num_input_tokens_seen": 20148432,
      "step": 30875
    },
    {
      "epoch": 18.20754716981132,
      "grad_norm": 5.070833683013916,
      "learning_rate": 2.428446957763608e-07,
      "loss": 0.2909,
      "num_input_tokens_seen": 20150800,
      "step": 30880
    },
    {
      "epoch": 18.21049528301887,
      "grad_norm": 3.2415924072265625,
      "learning_rate": 2.4205328495040535e-07,
      "loss": 0.2139,
      "num_input_tokens_seen": 20155728,
      "step": 30885
    },
    {
      "epoch": 18.213443396226417,
      "grad_norm": 3.8197202682495117,
      "learning_rate": 2.412631338113486e-07,
      "loss": 0.3497,
      "num_input_tokens_seen": 20158608,
      "step": 30890
    },
    {
      "epoch": 18.21639150943396,
      "grad_norm": 4.628184795379639,
      "learning_rate": 2.404742425683848e-07,
      "loss": 0.4422,
      "num_input_tokens_seen": 20161904,
      "step": 30895
    },
    {
      "epoch": 18.21933962264151,
      "grad_norm": 3.5616936683654785,
      "learning_rate": 2.3968661143037864e-07,
      "loss": 0.4067,
      "num_input_tokens_seen": 20165296,
      "step": 30900
    },
    {
      "epoch": 18.222287735849058,
      "grad_norm": 4.433199882507324,
      "learning_rate": 2.3890024060585823e-07,
      "loss": 0.4141,
      "num_input_tokens_seen": 20168400,
      "step": 30905
    },
    {
      "epoch": 18.225235849056602,
      "grad_norm": 4.060523509979248,
      "learning_rate": 2.3811513030301826e-07,
      "loss": 0.349,
      "num_input_tokens_seen": 20172080,
      "step": 30910
    },
    {
      "epoch": 18.22818396226415,
      "grad_norm": 4.877480506896973,
      "learning_rate": 2.373312807297201e-07,
      "loss": 0.2551,
      "num_input_tokens_seen": 20175088,
      "step": 30915
    },
    {
      "epoch": 18.2311320754717,
      "grad_norm": 3.3634192943573,
      "learning_rate": 2.3654869209349007e-07,
      "loss": 0.2623,
      "num_input_tokens_seen": 20178096,
      "step": 30920
    },
    {
      "epoch": 18.234080188679247,
      "grad_norm": 3.5076520442962646,
      "learning_rate": 2.357673646015246e-07,
      "loss": 0.253,
      "num_input_tokens_seen": 20180848,
      "step": 30925
    },
    {
      "epoch": 18.23702830188679,
      "grad_norm": 3.9768006801605225,
      "learning_rate": 2.3498729846068103e-07,
      "loss": 0.2387,
      "num_input_tokens_seen": 20183536,
      "step": 30930
    },
    {
      "epoch": 18.23997641509434,
      "grad_norm": 5.699945449829102,
      "learning_rate": 2.342084938774869e-07,
      "loss": 0.5658,
      "num_input_tokens_seen": 20186000,
      "step": 30935
    },
    {
      "epoch": 18.242924528301888,
      "grad_norm": 7.519649982452393,
      "learning_rate": 2.334309510581334e-07,
      "loss": 0.3096,
      "num_input_tokens_seen": 20188240,
      "step": 30940
    },
    {
      "epoch": 18.245872641509433,
      "grad_norm": 3.784212589263916,
      "learning_rate": 2.3265467020847864e-07,
      "loss": 0.3215,
      "num_input_tokens_seen": 20190512,
      "step": 30945
    },
    {
      "epoch": 18.24882075471698,
      "grad_norm": 3.8966903686523438,
      "learning_rate": 2.31879651534046e-07,
      "loss": 0.3134,
      "num_input_tokens_seen": 20193040,
      "step": 30950
    },
    {
      "epoch": 18.25176886792453,
      "grad_norm": 4.3160176277160645,
      "learning_rate": 2.311058952400247e-07,
      "loss": 0.4113,
      "num_input_tokens_seen": 20196432,
      "step": 30955
    },
    {
      "epoch": 18.254716981132077,
      "grad_norm": 3.0872766971588135,
      "learning_rate": 2.3033340153127026e-07,
      "loss": 0.3499,
      "num_input_tokens_seen": 20198800,
      "step": 30960
    },
    {
      "epoch": 18.25766509433962,
      "grad_norm": 2.901895523071289,
      "learning_rate": 2.295621706123041e-07,
      "loss": 0.4237,
      "num_input_tokens_seen": 20202384,
      "step": 30965
    },
    {
      "epoch": 18.26061320754717,
      "grad_norm": 2.930816888809204,
      "learning_rate": 2.287922026873135e-07,
      "loss": 0.3298,
      "num_input_tokens_seen": 20206832,
      "step": 30970
    },
    {
      "epoch": 18.263561320754718,
      "grad_norm": 6.388247013092041,
      "learning_rate": 2.2802349796014923e-07,
      "loss": 0.2324,
      "num_input_tokens_seen": 20209904,
      "step": 30975
    },
    {
      "epoch": 18.266509433962263,
      "grad_norm": 3.3831570148468018,
      "learning_rate": 2.2725605663433013e-07,
      "loss": 0.1808,
      "num_input_tokens_seen": 20213712,
      "step": 30980
    },
    {
      "epoch": 18.26945754716981,
      "grad_norm": 3.8719987869262695,
      "learning_rate": 2.264898789130393e-07,
      "loss": 0.2718,
      "num_input_tokens_seen": 20216976,
      "step": 30985
    },
    {
      "epoch": 18.27240566037736,
      "grad_norm": 2.7047226428985596,
      "learning_rate": 2.2572496499912554e-07,
      "loss": 0.3002,
      "num_input_tokens_seen": 20219408,
      "step": 30990
    },
    {
      "epoch": 18.275353773584907,
      "grad_norm": 2.986130475997925,
      "learning_rate": 2.2496131509510354e-07,
      "loss": 0.3919,
      "num_input_tokens_seen": 20226672,
      "step": 30995
    },
    {
      "epoch": 18.278301886792452,
      "grad_norm": 3.546712875366211,
      "learning_rate": 2.2419892940315268e-07,
      "loss": 0.2783,
      "num_input_tokens_seen": 20230448,
      "step": 31000
    },
    {
      "epoch": 18.28125,
      "grad_norm": 4.5099778175354,
      "learning_rate": 2.2343780812511819e-07,
      "loss": 0.3248,
      "num_input_tokens_seen": 20233936,
      "step": 31005
    },
    {
      "epoch": 18.284198113207548,
      "grad_norm": 1.9003348350524902,
      "learning_rate": 2.2267795146250936e-07,
      "loss": 0.3385,
      "num_input_tokens_seen": 20237232,
      "step": 31010
    },
    {
      "epoch": 18.287146226415093,
      "grad_norm": 4.365650177001953,
      "learning_rate": 2.2191935961650146e-07,
      "loss": 0.3543,
      "num_input_tokens_seen": 20240112,
      "step": 31015
    },
    {
      "epoch": 18.29009433962264,
      "grad_norm": 5.9216413497924805,
      "learning_rate": 2.2116203278793603e-07,
      "loss": 0.4193,
      "num_input_tokens_seen": 20243856,
      "step": 31020
    },
    {
      "epoch": 18.29304245283019,
      "grad_norm": 1.963316559791565,
      "learning_rate": 2.2040597117731766e-07,
      "loss": 0.3104,
      "num_input_tokens_seen": 20246704,
      "step": 31025
    },
    {
      "epoch": 18.295990566037737,
      "grad_norm": 3.0110862255096436,
      "learning_rate": 2.1965117498481793e-07,
      "loss": 0.1606,
      "num_input_tokens_seen": 20252592,
      "step": 31030
    },
    {
      "epoch": 18.298938679245282,
      "grad_norm": 3.420036792755127,
      "learning_rate": 2.188976444102714e-07,
      "loss": 0.3302,
      "num_input_tokens_seen": 20256976,
      "step": 31035
    },
    {
      "epoch": 18.30188679245283,
      "grad_norm": 2.4606235027313232,
      "learning_rate": 2.181453796531796e-07,
      "loss": 0.3509,
      "num_input_tokens_seen": 20260464,
      "step": 31040
    },
    {
      "epoch": 18.30483490566038,
      "grad_norm": 5.958358287811279,
      "learning_rate": 2.1739438091270658e-07,
      "loss": 0.2147,
      "num_input_tokens_seen": 20262864,
      "step": 31045
    },
    {
      "epoch": 18.307783018867923,
      "grad_norm": 4.798393249511719,
      "learning_rate": 2.1664464838768329e-07,
      "loss": 0.2624,
      "num_input_tokens_seen": 20265744,
      "step": 31050
    },
    {
      "epoch": 18.31073113207547,
      "grad_norm": 4.676506042480469,
      "learning_rate": 2.1589618227660426e-07,
      "loss": 0.2706,
      "num_input_tokens_seen": 20268272,
      "step": 31055
    },
    {
      "epoch": 18.31367924528302,
      "grad_norm": 5.121920108795166,
      "learning_rate": 2.151489827776293e-07,
      "loss": 0.3173,
      "num_input_tokens_seen": 20271088,
      "step": 31060
    },
    {
      "epoch": 18.316627358490567,
      "grad_norm": 3.1759066581726074,
      "learning_rate": 2.1440305008858298e-07,
      "loss": 0.3216,
      "num_input_tokens_seen": 20273680,
      "step": 31065
    },
    {
      "epoch": 18.319575471698112,
      "grad_norm": 3.0045599937438965,
      "learning_rate": 2.1365838440695397e-07,
      "loss": 0.2761,
      "num_input_tokens_seen": 20277232,
      "step": 31070
    },
    {
      "epoch": 18.32252358490566,
      "grad_norm": 6.1052093505859375,
      "learning_rate": 2.129149859298957e-07,
      "loss": 0.4049,
      "num_input_tokens_seen": 20280816,
      "step": 31075
    },
    {
      "epoch": 18.32547169811321,
      "grad_norm": 3.8784468173980713,
      "learning_rate": 2.1217285485422622e-07,
      "loss": 0.2688,
      "num_input_tokens_seen": 20283920,
      "step": 31080
    },
    {
      "epoch": 18.328419811320753,
      "grad_norm": 6.382956027984619,
      "learning_rate": 2.114319913764268e-07,
      "loss": 0.3839,
      "num_input_tokens_seen": 20287472,
      "step": 31085
    },
    {
      "epoch": 18.3313679245283,
      "grad_norm": 3.877950429916382,
      "learning_rate": 2.10692395692646e-07,
      "loss": 0.2511,
      "num_input_tokens_seen": 20290448,
      "step": 31090
    },
    {
      "epoch": 18.33431603773585,
      "grad_norm": 4.273087024688721,
      "learning_rate": 2.0995406799869444e-07,
      "loss": 0.2557,
      "num_input_tokens_seen": 20294448,
      "step": 31095
    },
    {
      "epoch": 18.337264150943398,
      "grad_norm": 2.5536012649536133,
      "learning_rate": 2.0921700849004743e-07,
      "loss": 0.307,
      "num_input_tokens_seen": 20298256,
      "step": 31100
    },
    {
      "epoch": 18.340212264150942,
      "grad_norm": 5.228384017944336,
      "learning_rate": 2.084812173618439e-07,
      "loss": 0.2463,
      "num_input_tokens_seen": 20301456,
      "step": 31105
    },
    {
      "epoch": 18.34316037735849,
      "grad_norm": 3.6189115047454834,
      "learning_rate": 2.0774669480888853e-07,
      "loss": 0.2524,
      "num_input_tokens_seen": 20304144,
      "step": 31110
    },
    {
      "epoch": 18.34610849056604,
      "grad_norm": 4.273752212524414,
      "learning_rate": 2.0701344102564912e-07,
      "loss": 0.2215,
      "num_input_tokens_seen": 20306480,
      "step": 31115
    },
    {
      "epoch": 18.349056603773583,
      "grad_norm": 4.518650531768799,
      "learning_rate": 2.062814562062576e-07,
      "loss": 0.2812,
      "num_input_tokens_seen": 20309072,
      "step": 31120
    },
    {
      "epoch": 18.35200471698113,
      "grad_norm": 2.799726724624634,
      "learning_rate": 2.0555074054451063e-07,
      "loss": 0.4939,
      "num_input_tokens_seen": 20311856,
      "step": 31125
    },
    {
      "epoch": 18.35495283018868,
      "grad_norm": 2.913865327835083,
      "learning_rate": 2.0482129423386843e-07,
      "loss": 0.3563,
      "num_input_tokens_seen": 20315600,
      "step": 31130
    },
    {
      "epoch": 18.357900943396228,
      "grad_norm": 4.299365997314453,
      "learning_rate": 2.040931174674543e-07,
      "loss": 0.2881,
      "num_input_tokens_seen": 20318480,
      "step": 31135
    },
    {
      "epoch": 18.360849056603772,
      "grad_norm": 4.572053909301758,
      "learning_rate": 2.0336621043805682e-07,
      "loss": 0.1799,
      "num_input_tokens_seen": 20321008,
      "step": 31140
    },
    {
      "epoch": 18.36379716981132,
      "grad_norm": 5.818107604980469,
      "learning_rate": 2.0264057333812704e-07,
      "loss": 0.2926,
      "num_input_tokens_seen": 20323728,
      "step": 31145
    },
    {
      "epoch": 18.36674528301887,
      "grad_norm": 2.7387173175811768,
      "learning_rate": 2.0191620635978127e-07,
      "loss": 0.2844,
      "num_input_tokens_seen": 20327504,
      "step": 31150
    },
    {
      "epoch": 18.369693396226417,
      "grad_norm": 2.655416965484619,
      "learning_rate": 2.0119310969479833e-07,
      "loss": 0.3416,
      "num_input_tokens_seen": 20331088,
      "step": 31155
    },
    {
      "epoch": 18.37264150943396,
      "grad_norm": 3.6547698974609375,
      "learning_rate": 2.004712835346212e-07,
      "loss": 0.2262,
      "num_input_tokens_seen": 20334352,
      "step": 31160
    },
    {
      "epoch": 18.37558962264151,
      "grad_norm": 4.643865585327148,
      "learning_rate": 1.99750728070357e-07,
      "loss": 0.2664,
      "num_input_tokens_seen": 20337456,
      "step": 31165
    },
    {
      "epoch": 18.378537735849058,
      "grad_norm": 4.551769733428955,
      "learning_rate": 1.9903144349277536e-07,
      "loss": 0.3164,
      "num_input_tokens_seen": 20341072,
      "step": 31170
    },
    {
      "epoch": 18.381485849056602,
      "grad_norm": 6.5776567459106445,
      "learning_rate": 1.983134299923095e-07,
      "loss": 0.2525,
      "num_input_tokens_seen": 20343632,
      "step": 31175
    },
    {
      "epoch": 18.38443396226415,
      "grad_norm": 4.07740592956543,
      "learning_rate": 1.9759668775905737e-07,
      "loss": 0.3141,
      "num_input_tokens_seen": 20346928,
      "step": 31180
    },
    {
      "epoch": 18.3873820754717,
      "grad_norm": 3.0550296306610107,
      "learning_rate": 1.9688121698277995e-07,
      "loss": 0.2668,
      "num_input_tokens_seen": 20350032,
      "step": 31185
    },
    {
      "epoch": 18.390330188679247,
      "grad_norm": 3.922240734100342,
      "learning_rate": 1.9616701785290015e-07,
      "loss": 0.2537,
      "num_input_tokens_seen": 20353200,
      "step": 31190
    },
    {
      "epoch": 18.39327830188679,
      "grad_norm": 4.638723373413086,
      "learning_rate": 1.954540905585056e-07,
      "loss": 0.4038,
      "num_input_tokens_seen": 20356080,
      "step": 31195
    },
    {
      "epoch": 18.39622641509434,
      "grad_norm": 4.057901859283447,
      "learning_rate": 1.9474243528834757e-07,
      "loss": 0.2868,
      "num_input_tokens_seen": 20359280,
      "step": 31200
    },
    {
      "epoch": 18.399174528301888,
      "grad_norm": 2.1456689834594727,
      "learning_rate": 1.9403205223083866e-07,
      "loss": 0.2359,
      "num_input_tokens_seen": 20362448,
      "step": 31205
    },
    {
      "epoch": 18.402122641509433,
      "grad_norm": 3.055636167526245,
      "learning_rate": 1.9332294157405619e-07,
      "loss": 0.3469,
      "num_input_tokens_seen": 20365904,
      "step": 31210
    },
    {
      "epoch": 18.40507075471698,
      "grad_norm": 5.299835205078125,
      "learning_rate": 1.926151035057411e-07,
      "loss": 0.2711,
      "num_input_tokens_seen": 20368880,
      "step": 31215
    },
    {
      "epoch": 18.40801886792453,
      "grad_norm": 4.883914470672607,
      "learning_rate": 1.9190853821329626e-07,
      "loss": 0.2884,
      "num_input_tokens_seen": 20371920,
      "step": 31220
    },
    {
      "epoch": 18.410966981132077,
      "grad_norm": 3.625037431716919,
      "learning_rate": 1.9120324588378757e-07,
      "loss": 0.4157,
      "num_input_tokens_seen": 20375440,
      "step": 31225
    },
    {
      "epoch": 18.41391509433962,
      "grad_norm": 4.024853706359863,
      "learning_rate": 1.9049922670394461e-07,
      "loss": 0.3652,
      "num_input_tokens_seen": 20378992,
      "step": 31230
    },
    {
      "epoch": 18.41686320754717,
      "grad_norm": 3.7979280948638916,
      "learning_rate": 1.897964808601588e-07,
      "loss": 0.2763,
      "num_input_tokens_seen": 20381616,
      "step": 31235
    },
    {
      "epoch": 18.419811320754718,
      "grad_norm": 3.5593016147613525,
      "learning_rate": 1.8909500853848517e-07,
      "loss": 0.2763,
      "num_input_tokens_seen": 20384880,
      "step": 31240
    },
    {
      "epoch": 18.422759433962263,
      "grad_norm": 4.1576032638549805,
      "learning_rate": 1.8839480992464243e-07,
      "loss": 0.4847,
      "num_input_tokens_seen": 20388048,
      "step": 31245
    },
    {
      "epoch": 18.42570754716981,
      "grad_norm": 2.9888906478881836,
      "learning_rate": 1.8769588520401005e-07,
      "loss": 0.3937,
      "num_input_tokens_seen": 20391824,
      "step": 31250
    },
    {
      "epoch": 18.42865566037736,
      "grad_norm": 8.12476634979248,
      "learning_rate": 1.8699823456163279e-07,
      "loss": 0.2707,
      "num_input_tokens_seen": 20395280,
      "step": 31255
    },
    {
      "epoch": 18.431603773584907,
      "grad_norm": 9.915144920349121,
      "learning_rate": 1.8630185818221514e-07,
      "loss": 0.3827,
      "num_input_tokens_seen": 20398320,
      "step": 31260
    },
    {
      "epoch": 18.434551886792452,
      "grad_norm": 5.705941677093506,
      "learning_rate": 1.856067562501268e-07,
      "loss": 0.3208,
      "num_input_tokens_seen": 20400944,
      "step": 31265
    },
    {
      "epoch": 18.4375,
      "grad_norm": 5.286247253417969,
      "learning_rate": 1.8491292894939837e-07,
      "loss": 0.2665,
      "num_input_tokens_seen": 20404112,
      "step": 31270
    },
    {
      "epoch": 18.440448113207548,
      "grad_norm": 2.721579074859619,
      "learning_rate": 1.8422037646372405e-07,
      "loss": 0.2769,
      "num_input_tokens_seen": 20408048,
      "step": 31275
    },
    {
      "epoch": 18.443396226415093,
      "grad_norm": 2.894845485687256,
      "learning_rate": 1.8352909897645989e-07,
      "loss": 0.3773,
      "num_input_tokens_seen": 20411792,
      "step": 31280
    },
    {
      "epoch": 18.44634433962264,
      "grad_norm": 2.38879656791687,
      "learning_rate": 1.8283909667062448e-07,
      "loss": 0.3554,
      "num_input_tokens_seen": 20414896,
      "step": 31285
    },
    {
      "epoch": 18.44929245283019,
      "grad_norm": 3.0960772037506104,
      "learning_rate": 1.82150369728899e-07,
      "loss": 0.2903,
      "num_input_tokens_seen": 20417744,
      "step": 31290
    },
    {
      "epoch": 18.452240566037737,
      "grad_norm": 3.889052629470825,
      "learning_rate": 1.814629183336275e-07,
      "loss": 0.274,
      "num_input_tokens_seen": 20421552,
      "step": 31295
    },
    {
      "epoch": 18.455188679245282,
      "grad_norm": 4.070337772369385,
      "learning_rate": 1.807767426668139e-07,
      "loss": 0.3764,
      "num_input_tokens_seen": 20424400,
      "step": 31300
    },
    {
      "epoch": 18.45813679245283,
      "grad_norm": 6.076757907867432,
      "learning_rate": 1.8009184291012783e-07,
      "loss": 0.2896,
      "num_input_tokens_seen": 20427600,
      "step": 31305
    },
    {
      "epoch": 18.46108490566038,
      "grad_norm": 3.4834136962890625,
      "learning_rate": 1.7940821924489926e-07,
      "loss": 0.3118,
      "num_input_tokens_seen": 20431568,
      "step": 31310
    },
    {
      "epoch": 18.464033018867923,
      "grad_norm": 3.16585636138916,
      "learning_rate": 1.7872587185212009e-07,
      "loss": 0.3353,
      "num_input_tokens_seen": 20434704,
      "step": 31315
    },
    {
      "epoch": 18.46698113207547,
      "grad_norm": 4.02031946182251,
      "learning_rate": 1.7804480091244524e-07,
      "loss": 0.3823,
      "num_input_tokens_seen": 20437968,
      "step": 31320
    },
    {
      "epoch": 18.46992924528302,
      "grad_norm": 3.884927272796631,
      "learning_rate": 1.7736500660619104e-07,
      "loss": 0.3554,
      "num_input_tokens_seen": 20441872,
      "step": 31325
    },
    {
      "epoch": 18.472877358490567,
      "grad_norm": 5.149443626403809,
      "learning_rate": 1.766864891133352e-07,
      "loss": 0.3027,
      "num_input_tokens_seen": 20445072,
      "step": 31330
    },
    {
      "epoch": 18.475825471698112,
      "grad_norm": 5.718039512634277,
      "learning_rate": 1.7600924861351843e-07,
      "loss": 0.3749,
      "num_input_tokens_seen": 20451920,
      "step": 31335
    },
    {
      "epoch": 18.47877358490566,
      "grad_norm": 3.3054327964782715,
      "learning_rate": 1.7533328528604398e-07,
      "loss": 0.2864,
      "num_input_tokens_seen": 20456656,
      "step": 31340
    },
    {
      "epoch": 18.48172169811321,
      "grad_norm": 5.711171627044678,
      "learning_rate": 1.746585993098754e-07,
      "loss": 0.2848,
      "num_input_tokens_seen": 20459600,
      "step": 31345
    },
    {
      "epoch": 18.484669811320753,
      "grad_norm": 3.836273193359375,
      "learning_rate": 1.7398519086363864e-07,
      "loss": 0.2667,
      "num_input_tokens_seen": 20462768,
      "step": 31350
    },
    {
      "epoch": 18.4876179245283,
      "grad_norm": 2.85617995262146,
      "learning_rate": 1.733130601256211e-07,
      "loss": 0.2421,
      "num_input_tokens_seen": 20465488,
      "step": 31355
    },
    {
      "epoch": 18.49056603773585,
      "grad_norm": 3.5198850631713867,
      "learning_rate": 1.7264220727377323e-07,
      "loss": 0.5042,
      "num_input_tokens_seen": 20469264,
      "step": 31360
    },
    {
      "epoch": 18.493514150943398,
      "grad_norm": 2.750945568084717,
      "learning_rate": 1.7197263248570517e-07,
      "loss": 0.2789,
      "num_input_tokens_seen": 20472176,
      "step": 31365
    },
    {
      "epoch": 18.496462264150942,
      "grad_norm": 4.666244983673096,
      "learning_rate": 1.7130433593869124e-07,
      "loss": 0.3625,
      "num_input_tokens_seen": 20476048,
      "step": 31370
    },
    {
      "epoch": 18.49941037735849,
      "grad_norm": 3.435650587081909,
      "learning_rate": 1.706373178096643e-07,
      "loss": 0.2588,
      "num_input_tokens_seen": 20479888,
      "step": 31375
    },
    {
      "epoch": 18.50235849056604,
      "grad_norm": 4.002604007720947,
      "learning_rate": 1.6997157827522092e-07,
      "loss": 0.3334,
      "num_input_tokens_seen": 20482896,
      "step": 31380
    },
    {
      "epoch": 18.505306603773583,
      "grad_norm": 3.4299371242523193,
      "learning_rate": 1.6930711751161843e-07,
      "loss": 0.2593,
      "num_input_tokens_seen": 20486352,
      "step": 31385
    },
    {
      "epoch": 18.50825471698113,
      "grad_norm": 3.40209698677063,
      "learning_rate": 1.6864393569477556e-07,
      "loss": 0.2878,
      "num_input_tokens_seen": 20489456,
      "step": 31390
    },
    {
      "epoch": 18.51120283018868,
      "grad_norm": 4.260105133056641,
      "learning_rate": 1.6798203300027295e-07,
      "loss": 0.2706,
      "num_input_tokens_seen": 20493744,
      "step": 31395
    },
    {
      "epoch": 18.514150943396228,
      "grad_norm": 2.823831558227539,
      "learning_rate": 1.6732140960335152e-07,
      "loss": 0.4022,
      "num_input_tokens_seen": 20496400,
      "step": 31400
    },
    {
      "epoch": 18.517099056603772,
      "grad_norm": 4.569741249084473,
      "learning_rate": 1.666620656789153e-07,
      "loss": 0.411,
      "num_input_tokens_seen": 20498768,
      "step": 31405
    },
    {
      "epoch": 18.52004716981132,
      "grad_norm": 7.168444633483887,
      "learning_rate": 1.660040014015274e-07,
      "loss": 0.2382,
      "num_input_tokens_seen": 20501584,
      "step": 31410
    },
    {
      "epoch": 18.52299528301887,
      "grad_norm": 3.0427677631378174,
      "learning_rate": 1.6534721694541344e-07,
      "loss": 0.1969,
      "num_input_tokens_seen": 20504144,
      "step": 31415
    },
    {
      "epoch": 18.525943396226417,
      "grad_norm": 2.8198704719543457,
      "learning_rate": 1.6469171248445993e-07,
      "loss": 0.3683,
      "num_input_tokens_seen": 20507728,
      "step": 31420
    },
    {
      "epoch": 18.52889150943396,
      "grad_norm": 5.25468111038208,
      "learning_rate": 1.6403748819221464e-07,
      "loss": 0.2864,
      "num_input_tokens_seen": 20511472,
      "step": 31425
    },
    {
      "epoch": 18.53183962264151,
      "grad_norm": 3.428649663925171,
      "learning_rate": 1.6338454424188632e-07,
      "loss": 0.2558,
      "num_input_tokens_seen": 20514448,
      "step": 31430
    },
    {
      "epoch": 18.534787735849058,
      "grad_norm": 3.517892360687256,
      "learning_rate": 1.6273288080634442e-07,
      "loss": 0.3448,
      "num_input_tokens_seen": 20517584,
      "step": 31435
    },
    {
      "epoch": 18.537735849056602,
      "grad_norm": 2.573636293411255,
      "learning_rate": 1.6208249805811982e-07,
      "loss": 0.3152,
      "num_input_tokens_seen": 20521456,
      "step": 31440
    },
    {
      "epoch": 18.54068396226415,
      "grad_norm": 3.259951114654541,
      "learning_rate": 1.6143339616940423e-07,
      "loss": 0.1984,
      "num_input_tokens_seen": 20524336,
      "step": 31445
    },
    {
      "epoch": 18.5436320754717,
      "grad_norm": 3.234400749206543,
      "learning_rate": 1.6078557531205018e-07,
      "loss": 0.2552,
      "num_input_tokens_seen": 20527856,
      "step": 31450
    },
    {
      "epoch": 18.546580188679247,
      "grad_norm": 3.0910327434539795,
      "learning_rate": 1.601390356575705e-07,
      "loss": 0.3231,
      "num_input_tokens_seen": 20532080,
      "step": 31455
    },
    {
      "epoch": 18.54952830188679,
      "grad_norm": 2.6766817569732666,
      "learning_rate": 1.5949377737713988e-07,
      "loss": 0.3175,
      "num_input_tokens_seen": 20536560,
      "step": 31460
    },
    {
      "epoch": 18.55247641509434,
      "grad_norm": 4.010452747344971,
      "learning_rate": 1.5884980064159338e-07,
      "loss": 0.273,
      "num_input_tokens_seen": 20539472,
      "step": 31465
    },
    {
      "epoch": 18.555424528301888,
      "grad_norm": 3.956282377243042,
      "learning_rate": 1.5820710562142627e-07,
      "loss": 0.3347,
      "num_input_tokens_seen": 20543728,
      "step": 31470
    },
    {
      "epoch": 18.558372641509433,
      "grad_norm": 3.55106258392334,
      "learning_rate": 1.575656924867952e-07,
      "loss": 0.325,
      "num_input_tokens_seen": 20547248,
      "step": 31475
    },
    {
      "epoch": 18.56132075471698,
      "grad_norm": 2.6266252994537354,
      "learning_rate": 1.5692556140751658e-07,
      "loss": 0.3612,
      "num_input_tokens_seen": 20550608,
      "step": 31480
    },
    {
      "epoch": 18.56426886792453,
      "grad_norm": 5.347181797027588,
      "learning_rate": 1.5628671255306706e-07,
      "loss": 0.2816,
      "num_input_tokens_seen": 20553744,
      "step": 31485
    },
    {
      "epoch": 18.567216981132077,
      "grad_norm": 7.649124622344971,
      "learning_rate": 1.556491460925863e-07,
      "loss": 0.2924,
      "num_input_tokens_seen": 20556016,
      "step": 31490
    },
    {
      "epoch": 18.57016509433962,
      "grad_norm": 5.071175575256348,
      "learning_rate": 1.550128621948721e-07,
      "loss": 0.2443,
      "num_input_tokens_seen": 20559184,
      "step": 31495
    },
    {
      "epoch": 18.57311320754717,
      "grad_norm": 5.96088981628418,
      "learning_rate": 1.5437786102838413e-07,
      "loss": 0.3123,
      "num_input_tokens_seen": 20561904,
      "step": 31500
    },
    {
      "epoch": 18.576061320754718,
      "grad_norm": 3.1630208492279053,
      "learning_rate": 1.5374414276124017e-07,
      "loss": 0.3126,
      "num_input_tokens_seen": 20566448,
      "step": 31505
    },
    {
      "epoch": 18.579009433962263,
      "grad_norm": 3.5068132877349854,
      "learning_rate": 1.5311170756122095e-07,
      "loss": 0.2926,
      "num_input_tokens_seen": 20568880,
      "step": 31510
    },
    {
      "epoch": 18.58195754716981,
      "grad_norm": 3.951460361480713,
      "learning_rate": 1.5248055559576647e-07,
      "loss": 0.3251,
      "num_input_tokens_seen": 20571888,
      "step": 31515
    },
    {
      "epoch": 18.58490566037736,
      "grad_norm": 5.022000312805176,
      "learning_rate": 1.5185068703197526e-07,
      "loss": 0.4509,
      "num_input_tokens_seen": 20574768,
      "step": 31520
    },
    {
      "epoch": 18.587853773584907,
      "grad_norm": 3.242913007736206,
      "learning_rate": 1.5122210203661004e-07,
      "loss": 0.2427,
      "num_input_tokens_seen": 20578032,
      "step": 31525
    },
    {
      "epoch": 18.590801886792452,
      "grad_norm": 3.5637128353118896,
      "learning_rate": 1.505948007760899e-07,
      "loss": 0.3167,
      "num_input_tokens_seen": 20580208,
      "step": 31530
    },
    {
      "epoch": 18.59375,
      "grad_norm": 3.633031129837036,
      "learning_rate": 1.4996878341649647e-07,
      "loss": 0.4487,
      "num_input_tokens_seen": 20583024,
      "step": 31535
    },
    {
      "epoch": 18.596698113207548,
      "grad_norm": 4.795530319213867,
      "learning_rate": 1.493440501235699e-07,
      "loss": 0.3554,
      "num_input_tokens_seen": 20585744,
      "step": 31540
    },
    {
      "epoch": 18.599646226415093,
      "grad_norm": 3.434183120727539,
      "learning_rate": 1.487206010627118e-07,
      "loss": 0.308,
      "num_input_tokens_seen": 20589296,
      "step": 31545
    },
    {
      "epoch": 18.60259433962264,
      "grad_norm": 2.3632426261901855,
      "learning_rate": 1.4809843639898124e-07,
      "loss": 0.3127,
      "num_input_tokens_seen": 20592400,
      "step": 31550
    },
    {
      "epoch": 18.60554245283019,
      "grad_norm": 3.9533703327178955,
      "learning_rate": 1.4747755629710093e-07,
      "loss": 0.1978,
      "num_input_tokens_seen": 20595024,
      "step": 31555
    },
    {
      "epoch": 18.608490566037737,
      "grad_norm": 2.5510294437408447,
      "learning_rate": 1.4685796092145045e-07,
      "loss": 0.2828,
      "num_input_tokens_seen": 20598544,
      "step": 31560
    },
    {
      "epoch": 18.611438679245282,
      "grad_norm": 5.0588788986206055,
      "learning_rate": 1.4623965043607135e-07,
      "loss": 0.3113,
      "num_input_tokens_seen": 20601456,
      "step": 31565
    },
    {
      "epoch": 18.61438679245283,
      "grad_norm": 3.6721224784851074,
      "learning_rate": 1.4562262500466273e-07,
      "loss": 0.3825,
      "num_input_tokens_seen": 20604336,
      "step": 31570
    },
    {
      "epoch": 18.61733490566038,
      "grad_norm": 3.8494021892547607,
      "learning_rate": 1.4500688479058556e-07,
      "loss": 0.2636,
      "num_input_tokens_seen": 20607184,
      "step": 31575
    },
    {
      "epoch": 18.620283018867923,
      "grad_norm": 3.2143402099609375,
      "learning_rate": 1.4439242995685943e-07,
      "loss": 0.2943,
      "num_input_tokens_seen": 20610896,
      "step": 31580
    },
    {
      "epoch": 18.62323113207547,
      "grad_norm": 4.953895568847656,
      "learning_rate": 1.4377926066616364e-07,
      "loss": 0.3603,
      "num_input_tokens_seen": 20613936,
      "step": 31585
    },
    {
      "epoch": 18.62617924528302,
      "grad_norm": 5.222681999206543,
      "learning_rate": 1.4316737708083783e-07,
      "loss": 0.3972,
      "num_input_tokens_seen": 20617136,
      "step": 31590
    },
    {
      "epoch": 18.629127358490567,
      "grad_norm": 3.253453254699707,
      "learning_rate": 1.4255677936288127e-07,
      "loss": 0.3487,
      "num_input_tokens_seen": 20620304,
      "step": 31595
    },
    {
      "epoch": 18.632075471698112,
      "grad_norm": 6.208869934082031,
      "learning_rate": 1.4194746767395184e-07,
      "loss": 0.3254,
      "num_input_tokens_seen": 20623728,
      "step": 31600
    },
    {
      "epoch": 18.63502358490566,
      "grad_norm": 2.1416375637054443,
      "learning_rate": 1.4133944217536722e-07,
      "loss": 0.2835,
      "num_input_tokens_seen": 20627088,
      "step": 31605
    },
    {
      "epoch": 18.63797169811321,
      "grad_norm": 3.497950553894043,
      "learning_rate": 1.4073270302810471e-07,
      "loss": 0.2135,
      "num_input_tokens_seen": 20631088,
      "step": 31610
    },
    {
      "epoch": 18.640919811320753,
      "grad_norm": 4.315011024475098,
      "learning_rate": 1.4012725039280084e-07,
      "loss": 0.2733,
      "num_input_tokens_seen": 20634032,
      "step": 31615
    },
    {
      "epoch": 18.6438679245283,
      "grad_norm": 3.0566840171813965,
      "learning_rate": 1.3952308442975292e-07,
      "loss": 0.3314,
      "num_input_tokens_seen": 20637104,
      "step": 31620
    },
    {
      "epoch": 18.64681603773585,
      "grad_norm": 3.422818183898926,
      "learning_rate": 1.3892020529891637e-07,
      "loss": 0.2342,
      "num_input_tokens_seen": 20640368,
      "step": 31625
    },
    {
      "epoch": 18.649764150943398,
      "grad_norm": 6.178411483764648,
      "learning_rate": 1.3831861315990514e-07,
      "loss": 0.3761,
      "num_input_tokens_seen": 20643056,
      "step": 31630
    },
    {
      "epoch": 18.652712264150942,
      "grad_norm": 2.749654531478882,
      "learning_rate": 1.377183081719935e-07,
      "loss": 0.3219,
      "num_input_tokens_seen": 20647376,
      "step": 31635
    },
    {
      "epoch": 18.65566037735849,
      "grad_norm": 2.8359181880950928,
      "learning_rate": 1.3711929049411544e-07,
      "loss": 0.2729,
      "num_input_tokens_seen": 20650864,
      "step": 31640
    },
    {
      "epoch": 18.65860849056604,
      "grad_norm": 2.852304220199585,
      "learning_rate": 1.365215602848624e-07,
      "loss": 0.3813,
      "num_input_tokens_seen": 20653712,
      "step": 31645
    },
    {
      "epoch": 18.661556603773583,
      "grad_norm": 4.442895889282227,
      "learning_rate": 1.3592511770248727e-07,
      "loss": 0.3763,
      "num_input_tokens_seen": 20656528,
      "step": 31650
    },
    {
      "epoch": 18.66450471698113,
      "grad_norm": 5.450018882751465,
      "learning_rate": 1.3532996290490041e-07,
      "loss": 0.2725,
      "num_input_tokens_seen": 20659440,
      "step": 31655
    },
    {
      "epoch": 18.66745283018868,
      "grad_norm": 3.923981189727783,
      "learning_rate": 1.347360960496713e-07,
      "loss": 0.2856,
      "num_input_tokens_seen": 20662096,
      "step": 31660
    },
    {
      "epoch": 18.670400943396228,
      "grad_norm": 2.3668906688690186,
      "learning_rate": 1.3414351729402862e-07,
      "loss": 0.289,
      "num_input_tokens_seen": 20665808,
      "step": 31665
    },
    {
      "epoch": 18.673349056603772,
      "grad_norm": 3.667651653289795,
      "learning_rate": 1.3355222679486025e-07,
      "loss": 0.3186,
      "num_input_tokens_seen": 20668848,
      "step": 31670
    },
    {
      "epoch": 18.67629716981132,
      "grad_norm": 2.1343894004821777,
      "learning_rate": 1.3296222470871367e-07,
      "loss": 0.2851,
      "num_input_tokens_seen": 20671760,
      "step": 31675
    },
    {
      "epoch": 18.67924528301887,
      "grad_norm": 3.0044713020324707,
      "learning_rate": 1.3237351119179287e-07,
      "loss": 0.2815,
      "num_input_tokens_seen": 20675472,
      "step": 31680
    },
    {
      "epoch": 18.682193396226417,
      "grad_norm": 3.3570733070373535,
      "learning_rate": 1.3178608639996425e-07,
      "loss": 0.2789,
      "num_input_tokens_seen": 20678128,
      "step": 31685
    },
    {
      "epoch": 18.68514150943396,
      "grad_norm": 3.7691731452941895,
      "learning_rate": 1.3119995048874957e-07,
      "loss": 0.3361,
      "num_input_tokens_seen": 20681360,
      "step": 31690
    },
    {
      "epoch": 18.68808962264151,
      "grad_norm": 7.246938705444336,
      "learning_rate": 1.3061510361333186e-07,
      "loss": 0.3677,
      "num_input_tokens_seen": 20684816,
      "step": 31695
    },
    {
      "epoch": 18.691037735849058,
      "grad_norm": 4.411552429199219,
      "learning_rate": 1.3003154592855116e-07,
      "loss": 0.2802,
      "num_input_tokens_seen": 20689232,
      "step": 31700
    },
    {
      "epoch": 18.693985849056602,
      "grad_norm": 2.6474556922912598,
      "learning_rate": 1.2944927758890668e-07,
      "loss": 0.2515,
      "num_input_tokens_seen": 20691664,
      "step": 31705
    },
    {
      "epoch": 18.69693396226415,
      "grad_norm": 5.467953205108643,
      "learning_rate": 1.2886829874855733e-07,
      "loss": 0.2726,
      "num_input_tokens_seen": 20694896,
      "step": 31710
    },
    {
      "epoch": 18.6998820754717,
      "grad_norm": 2.5844666957855225,
      "learning_rate": 1.2828860956131894e-07,
      "loss": 0.2433,
      "num_input_tokens_seen": 20699312,
      "step": 31715
    },
    {
      "epoch": 18.702830188679247,
      "grad_norm": 3.1489176750183105,
      "learning_rate": 1.2771021018066765e-07,
      "loss": 0.2558,
      "num_input_tokens_seen": 20703056,
      "step": 31720
    },
    {
      "epoch": 18.70577830188679,
      "grad_norm": 3.6816179752349854,
      "learning_rate": 1.271331007597365e-07,
      "loss": 0.2757,
      "num_input_tokens_seen": 20705520,
      "step": 31725
    },
    {
      "epoch": 18.70872641509434,
      "grad_norm": 2.58306622505188,
      "learning_rate": 1.2655728145131774e-07,
      "loss": 0.3836,
      "num_input_tokens_seen": 20708784,
      "step": 31730
    },
    {
      "epoch": 18.711674528301888,
      "grad_norm": 2.371227264404297,
      "learning_rate": 1.2598275240786105e-07,
      "loss": 0.3546,
      "num_input_tokens_seen": 20712176,
      "step": 31735
    },
    {
      "epoch": 18.714622641509433,
      "grad_norm": 3.0417706966400146,
      "learning_rate": 1.254095137814776e-07,
      "loss": 0.3474,
      "num_input_tokens_seen": 20715152,
      "step": 31740
    },
    {
      "epoch": 18.71757075471698,
      "grad_norm": 3.8357276916503906,
      "learning_rate": 1.2483756572393368e-07,
      "loss": 0.2912,
      "num_input_tokens_seen": 20718480,
      "step": 31745
    },
    {
      "epoch": 18.72051886792453,
      "grad_norm": 3.006688356399536,
      "learning_rate": 1.242669083866549e-07,
      "loss": 0.2832,
      "num_input_tokens_seen": 20720944,
      "step": 31750
    },
    {
      "epoch": 18.723466981132077,
      "grad_norm": 3.0669965744018555,
      "learning_rate": 1.2369754192072537e-07,
      "loss": 0.3728,
      "num_input_tokens_seen": 20723856,
      "step": 31755
    },
    {
      "epoch": 18.72641509433962,
      "grad_norm": 4.026971340179443,
      "learning_rate": 1.231294664768873e-07,
      "loss": 0.3143,
      "num_input_tokens_seen": 20726928,
      "step": 31760
    },
    {
      "epoch": 18.72936320754717,
      "grad_norm": 3.7815651893615723,
      "learning_rate": 1.225626822055409e-07,
      "loss": 0.3305,
      "num_input_tokens_seen": 20730192,
      "step": 31765
    },
    {
      "epoch": 18.732311320754718,
      "grad_norm": 3.2977066040039062,
      "learning_rate": 1.2199718925674508e-07,
      "loss": 0.2117,
      "num_input_tokens_seen": 20735376,
      "step": 31770
    },
    {
      "epoch": 18.735259433962263,
      "grad_norm": 2.773209571838379,
      "learning_rate": 1.2143298778021616e-07,
      "loss": 0.2889,
      "num_input_tokens_seen": 20739888,
      "step": 31775
    },
    {
      "epoch": 18.73820754716981,
      "grad_norm": 4.491086959838867,
      "learning_rate": 1.2087007792532967e-07,
      "loss": 0.334,
      "num_input_tokens_seen": 20742640,
      "step": 31780
    },
    {
      "epoch": 18.74115566037736,
      "grad_norm": 2.7369511127471924,
      "learning_rate": 1.203084598411175e-07,
      "loss": 0.2551,
      "num_input_tokens_seen": 20745808,
      "step": 31785
    },
    {
      "epoch": 18.744103773584907,
      "grad_norm": 4.866786003112793,
      "learning_rate": 1.1974813367627124e-07,
      "loss": 0.2814,
      "num_input_tokens_seen": 20748592,
      "step": 31790
    },
    {
      "epoch": 18.747051886792452,
      "grad_norm": 2.6254589557647705,
      "learning_rate": 1.1918909957913949e-07,
      "loss": 0.3338,
      "num_input_tokens_seen": 20752048,
      "step": 31795
    },
    {
      "epoch": 18.75,
      "grad_norm": 3.3751771450042725,
      "learning_rate": 1.1863135769772827e-07,
      "loss": 0.3979,
      "num_input_tokens_seen": 20758544,
      "step": 31800
    },
    {
      "epoch": 18.752948113207548,
      "grad_norm": 2.5983822345733643,
      "learning_rate": 1.1807490817970279e-07,
      "loss": 0.2111,
      "num_input_tokens_seen": 20762672,
      "step": 31805
    },
    {
      "epoch": 18.755896226415093,
      "grad_norm": 4.350729465484619,
      "learning_rate": 1.1751975117238578e-07,
      "loss": 0.2503,
      "num_input_tokens_seen": 20765776,
      "step": 31810
    },
    {
      "epoch": 18.75884433962264,
      "grad_norm": 2.7183430194854736,
      "learning_rate": 1.1696588682275633e-07,
      "loss": 0.2976,
      "num_input_tokens_seen": 20768528,
      "step": 31815
    },
    {
      "epoch": 18.76179245283019,
      "grad_norm": 5.421873092651367,
      "learning_rate": 1.1641331527745325e-07,
      "loss": 0.3963,
      "num_input_tokens_seen": 20775248,
      "step": 31820
    },
    {
      "epoch": 18.764740566037737,
      "grad_norm": 3.816894054412842,
      "learning_rate": 1.1586203668277229e-07,
      "loss": 0.2465,
      "num_input_tokens_seen": 20778736,
      "step": 31825
    },
    {
      "epoch": 18.767688679245282,
      "grad_norm": 3.8145275115966797,
      "learning_rate": 1.1531205118466615e-07,
      "loss": 0.3572,
      "num_input_tokens_seen": 20782128,
      "step": 31830
    },
    {
      "epoch": 18.77063679245283,
      "grad_norm": 4.744506359100342,
      "learning_rate": 1.1476335892874669e-07,
      "loss": 0.3461,
      "num_input_tokens_seen": 20785424,
      "step": 31835
    },
    {
      "epoch": 18.77358490566038,
      "grad_norm": 4.102356910705566,
      "learning_rate": 1.1421596006028157e-07,
      "loss": 0.4265,
      "num_input_tokens_seen": 20788976,
      "step": 31840
    },
    {
      "epoch": 18.776533018867923,
      "grad_norm": 6.971306324005127,
      "learning_rate": 1.1366985472419823e-07,
      "loss": 0.2755,
      "num_input_tokens_seen": 20791536,
      "step": 31845
    },
    {
      "epoch": 18.77948113207547,
      "grad_norm": 4.049649715423584,
      "learning_rate": 1.1312504306507987e-07,
      "loss": 0.2372,
      "num_input_tokens_seen": 20795568,
      "step": 31850
    },
    {
      "epoch": 18.78242924528302,
      "grad_norm": 5.1576247215271,
      "learning_rate": 1.1258152522716725e-07,
      "loss": 0.3405,
      "num_input_tokens_seen": 20798832,
      "step": 31855
    },
    {
      "epoch": 18.785377358490567,
      "grad_norm": 4.836615562438965,
      "learning_rate": 1.1203930135435914e-07,
      "loss": 0.2894,
      "num_input_tokens_seen": 20802352,
      "step": 31860
    },
    {
      "epoch": 18.788325471698112,
      "grad_norm": 3.9058573246002197,
      "learning_rate": 1.1149837159021238e-07,
      "loss": 0.3813,
      "num_input_tokens_seen": 20806160,
      "step": 31865
    },
    {
      "epoch": 18.79127358490566,
      "grad_norm": 3.538270950317383,
      "learning_rate": 1.1095873607793961e-07,
      "loss": 0.2949,
      "num_input_tokens_seen": 20809424,
      "step": 31870
    },
    {
      "epoch": 18.79422169811321,
      "grad_norm": 3.828787326812744,
      "learning_rate": 1.1042039496041212e-07,
      "loss": 0.3164,
      "num_input_tokens_seen": 20812624,
      "step": 31875
    },
    {
      "epoch": 18.797169811320753,
      "grad_norm": 4.499998092651367,
      "learning_rate": 1.0988334838015812e-07,
      "loss": 0.2514,
      "num_input_tokens_seen": 20815728,
      "step": 31880
    },
    {
      "epoch": 18.8001179245283,
      "grad_norm": 2.882920026779175,
      "learning_rate": 1.0934759647936333e-07,
      "loss": 0.3727,
      "num_input_tokens_seen": 20819664,
      "step": 31885
    },
    {
      "epoch": 18.80306603773585,
      "grad_norm": 3.671429395675659,
      "learning_rate": 1.0881313939986926e-07,
      "loss": 0.2504,
      "num_input_tokens_seen": 20823280,
      "step": 31890
    },
    {
      "epoch": 18.806014150943398,
      "grad_norm": 4.965364456176758,
      "learning_rate": 1.0827997728317662e-07,
      "loss": 0.2794,
      "num_input_tokens_seen": 20825840,
      "step": 31895
    },
    {
      "epoch": 18.808962264150942,
      "grad_norm": 7.650140762329102,
      "learning_rate": 1.0774811027044196e-07,
      "loss": 0.3608,
      "num_input_tokens_seen": 20828816,
      "step": 31900
    },
    {
      "epoch": 18.81191037735849,
      "grad_norm": 4.418237209320068,
      "learning_rate": 1.0721753850247984e-07,
      "loss": 0.224,
      "num_input_tokens_seen": 20831280,
      "step": 31905
    },
    {
      "epoch": 18.81485849056604,
      "grad_norm": 4.494792461395264,
      "learning_rate": 1.0668826211976124e-07,
      "loss": 0.1971,
      "num_input_tokens_seen": 20834256,
      "step": 31910
    },
    {
      "epoch": 18.817806603773583,
      "grad_norm": 3.804464340209961,
      "learning_rate": 1.0616028126241407e-07,
      "loss": 0.2567,
      "num_input_tokens_seen": 20837264,
      "step": 31915
    },
    {
      "epoch": 18.82075471698113,
      "grad_norm": 4.1853461265563965,
      "learning_rate": 1.0563359607022372e-07,
      "loss": 0.2677,
      "num_input_tokens_seen": 20840208,
      "step": 31920
    },
    {
      "epoch": 18.82370283018868,
      "grad_norm": 6.727773666381836,
      "learning_rate": 1.05108206682632e-07,
      "loss": 0.2378,
      "num_input_tokens_seen": 20844336,
      "step": 31925
    },
    {
      "epoch": 18.826650943396228,
      "grad_norm": 1.8606854677200317,
      "learning_rate": 1.0458411323873874e-07,
      "loss": 0.2943,
      "num_input_tokens_seen": 20847728,
      "step": 31930
    },
    {
      "epoch": 18.829599056603772,
      "grad_norm": 3.531827926635742,
      "learning_rate": 1.0406131587729962e-07,
      "loss": 0.2771,
      "num_input_tokens_seen": 20850928,
      "step": 31935
    },
    {
      "epoch": 18.83254716981132,
      "grad_norm": 5.268754482269287,
      "learning_rate": 1.035398147367278e-07,
      "loss": 0.3013,
      "num_input_tokens_seen": 20853968,
      "step": 31940
    },
    {
      "epoch": 18.83549528301887,
      "grad_norm": 4.493147850036621,
      "learning_rate": 1.030196099550923e-07,
      "loss": 0.2883,
      "num_input_tokens_seen": 20858064,
      "step": 31945
    },
    {
      "epoch": 18.838443396226417,
      "grad_norm": 4.69831657409668,
      "learning_rate": 1.0250070167011905e-07,
      "loss": 0.2785,
      "num_input_tokens_seen": 20860208,
      "step": 31950
    },
    {
      "epoch": 18.84139150943396,
      "grad_norm": 2.76827073097229,
      "learning_rate": 1.0198309001919315e-07,
      "loss": 0.2631,
      "num_input_tokens_seen": 20862864,
      "step": 31955
    },
    {
      "epoch": 18.84433962264151,
      "grad_norm": 5.016300201416016,
      "learning_rate": 1.0146677513935277e-07,
      "loss": 0.3585,
      "num_input_tokens_seen": 20865968,
      "step": 31960
    },
    {
      "epoch": 18.847287735849058,
      "grad_norm": 7.613326549530029,
      "learning_rate": 1.0095175716729578e-07,
      "loss": 0.275,
      "num_input_tokens_seen": 20869200,
      "step": 31965
    },
    {
      "epoch": 18.850235849056602,
      "grad_norm": 4.24086856842041,
      "learning_rate": 1.004380362393742e-07,
      "loss": 0.3183,
      "num_input_tokens_seen": 20871888,
      "step": 31970
    },
    {
      "epoch": 18.85318396226415,
      "grad_norm": 4.29726505279541,
      "learning_rate": 9.99256124915987e-08,
      "loss": 0.3647,
      "num_input_tokens_seen": 20874704,
      "step": 31975
    },
    {
      "epoch": 18.8561320754717,
      "grad_norm": 3.991434335708618,
      "learning_rate": 9.941448605963577e-08,
      "loss": 0.4206,
      "num_input_tokens_seen": 20878160,
      "step": 31980
    },
    {
      "epoch": 18.859080188679247,
      "grad_norm": 3.2095234394073486,
      "learning_rate": 9.890465707880715e-08,
      "loss": 0.2586,
      "num_input_tokens_seen": 20880656,
      "step": 31985
    },
    {
      "epoch": 18.86202830188679,
      "grad_norm": 3.644953489303589,
      "learning_rate": 9.839612568409374e-08,
      "loss": 0.3039,
      "num_input_tokens_seen": 20883760,
      "step": 31990
    },
    {
      "epoch": 18.86497641509434,
      "grad_norm": 3.309516429901123,
      "learning_rate": 9.788889201013119e-08,
      "loss": 0.2815,
      "num_input_tokens_seen": 20886864,
      "step": 31995
    },
    {
      "epoch": 18.867924528301888,
      "grad_norm": 4.729465484619141,
      "learning_rate": 9.738295619121097e-08,
      "loss": 0.3522,
      "num_input_tokens_seen": 20889552,
      "step": 32000
    },
    {
      "epoch": 18.870872641509433,
      "grad_norm": 3.4798028469085693,
      "learning_rate": 9.687831836128203e-08,
      "loss": 0.2596,
      "num_input_tokens_seen": 20893360,
      "step": 32005
    },
    {
      "epoch": 18.87382075471698,
      "grad_norm": 5.643940448760986,
      "learning_rate": 9.637497865395029e-08,
      "loss": 0.2454,
      "num_input_tokens_seen": 20896208,
      "step": 32010
    },
    {
      "epoch": 18.87676886792453,
      "grad_norm": 6.753708362579346,
      "learning_rate": 9.587293720247526e-08,
      "loss": 0.3027,
      "num_input_tokens_seen": 20898896,
      "step": 32015
    },
    {
      "epoch": 18.879716981132077,
      "grad_norm": 5.177978992462158,
      "learning_rate": 9.537219413977672e-08,
      "loss": 0.2769,
      "num_input_tokens_seen": 20902704,
      "step": 32020
    },
    {
      "epoch": 18.88266509433962,
      "grad_norm": 3.4996254444122314,
      "learning_rate": 9.487274959842696e-08,
      "loss": 0.2813,
      "num_input_tokens_seen": 20905712,
      "step": 32025
    },
    {
      "epoch": 18.88561320754717,
      "grad_norm": 4.770746231079102,
      "learning_rate": 9.437460371065687e-08,
      "loss": 0.3012,
      "num_input_tokens_seen": 20908176,
      "step": 32030
    },
    {
      "epoch": 18.888561320754718,
      "grad_norm": 3.8718643188476562,
      "learning_rate": 9.387775660835263e-08,
      "loss": 0.2973,
      "num_input_tokens_seen": 20911760,
      "step": 32035
    },
    {
      "epoch": 18.891509433962263,
      "grad_norm": 7.138391494750977,
      "learning_rate": 9.338220842305678e-08,
      "loss": 0.2477,
      "num_input_tokens_seen": 20914256,
      "step": 32040
    },
    {
      "epoch": 18.89445754716981,
      "grad_norm": 4.266496658325195,
      "learning_rate": 9.288795928596661e-08,
      "loss": 0.2981,
      "num_input_tokens_seen": 20917904,
      "step": 32045
    },
    {
      "epoch": 18.89740566037736,
      "grad_norm": 6.928383827209473,
      "learning_rate": 9.239500932793854e-08,
      "loss": 0.2732,
      "num_input_tokens_seen": 20922288,
      "step": 32050
    },
    {
      "epoch": 18.900353773584907,
      "grad_norm": 3.0919578075408936,
      "learning_rate": 9.190335867948263e-08,
      "loss": 0.4326,
      "num_input_tokens_seen": 20925328,
      "step": 32055
    },
    {
      "epoch": 18.903301886792452,
      "grad_norm": 3.1570050716400146,
      "learning_rate": 9.141300747076476e-08,
      "loss": 0.4076,
      "num_input_tokens_seen": 20928432,
      "step": 32060
    },
    {
      "epoch": 18.90625,
      "grad_norm": 3.9997150897979736,
      "learning_rate": 9.092395583160773e-08,
      "loss": 0.2425,
      "num_input_tokens_seen": 20932496,
      "step": 32065
    },
    {
      "epoch": 18.909198113207548,
      "grad_norm": 5.04506254196167,
      "learning_rate": 9.043620389149021e-08,
      "loss": 0.1986,
      "num_input_tokens_seen": 20935184,
      "step": 32070
    },
    {
      "epoch": 18.912146226415093,
      "grad_norm": 3.264798879623413,
      "learning_rate": 8.994975177954723e-08,
      "loss": 0.3418,
      "num_input_tokens_seen": 20938288,
      "step": 32075
    },
    {
      "epoch": 18.91509433962264,
      "grad_norm": 3.250704050064087,
      "learning_rate": 8.946459962456855e-08,
      "loss": 0.3366,
      "num_input_tokens_seen": 20941648,
      "step": 32080
    },
    {
      "epoch": 18.91804245283019,
      "grad_norm": 4.3506317138671875,
      "learning_rate": 8.89807475550003e-08,
      "loss": 0.3754,
      "num_input_tokens_seen": 20945008,
      "step": 32085
    },
    {
      "epoch": 18.920990566037737,
      "grad_norm": 4.307066440582275,
      "learning_rate": 8.849819569894447e-08,
      "loss": 0.3541,
      "num_input_tokens_seen": 20947792,
      "step": 32090
    },
    {
      "epoch": 18.923938679245282,
      "grad_norm": 2.9909331798553467,
      "learning_rate": 8.801694418415884e-08,
      "loss": 0.2469,
      "num_input_tokens_seen": 20950576,
      "step": 32095
    },
    {
      "epoch": 18.92688679245283,
      "grad_norm": 4.034100532531738,
      "learning_rate": 8.753699313805708e-08,
      "loss": 0.3381,
      "num_input_tokens_seen": 20953840,
      "step": 32100
    },
    {
      "epoch": 18.92983490566038,
      "grad_norm": 4.447424411773682,
      "learning_rate": 8.705834268770753e-08,
      "loss": 0.2294,
      "num_input_tokens_seen": 20956592,
      "step": 32105
    },
    {
      "epoch": 18.932783018867923,
      "grad_norm": 4.128286838531494,
      "learning_rate": 8.65809929598349e-08,
      "loss": 0.3171,
      "num_input_tokens_seen": 20959088,
      "step": 32110
    },
    {
      "epoch": 18.93573113207547,
      "grad_norm": 2.827216148376465,
      "learning_rate": 8.610494408082037e-08,
      "loss": 0.2669,
      "num_input_tokens_seen": 20962576,
      "step": 32115
    },
    {
      "epoch": 18.93867924528302,
      "grad_norm": 1.4719778299331665,
      "learning_rate": 8.563019617669977e-08,
      "loss": 0.2132,
      "num_input_tokens_seen": 20966896,
      "step": 32120
    },
    {
      "epoch": 18.941627358490567,
      "grad_norm": 5.397637367248535,
      "learning_rate": 8.51567493731642e-08,
      "loss": 0.3379,
      "num_input_tokens_seen": 20970608,
      "step": 32125
    },
    {
      "epoch": 18.944575471698112,
      "grad_norm": 4.547111988067627,
      "learning_rate": 8.468460379556176e-08,
      "loss": 0.2447,
      "num_input_tokens_seen": 20973616,
      "step": 32130
    },
    {
      "epoch": 18.94752358490566,
      "grad_norm": 5.0956244468688965,
      "learning_rate": 8.421375956889355e-08,
      "loss": 0.3532,
      "num_input_tokens_seen": 20976272,
      "step": 32135
    },
    {
      "epoch": 18.95047169811321,
      "grad_norm": 3.890078067779541,
      "learning_rate": 8.374421681781819e-08,
      "loss": 0.2378,
      "num_input_tokens_seen": 20979248,
      "step": 32140
    },
    {
      "epoch": 18.953419811320753,
      "grad_norm": 2.777700662612915,
      "learning_rate": 8.327597566665013e-08,
      "loss": 0.4016,
      "num_input_tokens_seen": 20982448,
      "step": 32145
    },
    {
      "epoch": 18.9563679245283,
      "grad_norm": 3.621309995651245,
      "learning_rate": 8.280903623935688e-08,
      "loss": 0.4091,
      "num_input_tokens_seen": 20987120,
      "step": 32150
    },
    {
      "epoch": 18.95931603773585,
      "grad_norm": 3.2641937732696533,
      "learning_rate": 8.234339865956342e-08,
      "loss": 0.2368,
      "num_input_tokens_seen": 20990192,
      "step": 32155
    },
    {
      "epoch": 18.962264150943398,
      "grad_norm": 5.773024082183838,
      "learning_rate": 8.187906305054838e-08,
      "loss": 0.2239,
      "num_input_tokens_seen": 20993264,
      "step": 32160
    },
    {
      "epoch": 18.965212264150942,
      "grad_norm": 4.664160251617432,
      "learning_rate": 8.141602953524841e-08,
      "loss": 0.2199,
      "num_input_tokens_seen": 20995984,
      "step": 32165
    },
    {
      "epoch": 18.96816037735849,
      "grad_norm": 5.944611072540283,
      "learning_rate": 8.095429823625212e-08,
      "loss": 0.3432,
      "num_input_tokens_seen": 20998672,
      "step": 32170
    },
    {
      "epoch": 18.97110849056604,
      "grad_norm": 4.054778575897217,
      "learning_rate": 8.04938692758045e-08,
      "loss": 0.2767,
      "num_input_tokens_seen": 21002256,
      "step": 32175
    },
    {
      "epoch": 18.974056603773583,
      "grad_norm": 4.914705276489258,
      "learning_rate": 8.003474277580803e-08,
      "loss": 0.344,
      "num_input_tokens_seen": 21005616,
      "step": 32180
    },
    {
      "epoch": 18.97700471698113,
      "grad_norm": 3.643676519393921,
      "learning_rate": 7.95769188578166e-08,
      "loss": 0.3489,
      "num_input_tokens_seen": 21008976,
      "step": 32185
    },
    {
      "epoch": 18.97995283018868,
      "grad_norm": 4.52269172668457,
      "learning_rate": 7.912039764304213e-08,
      "loss": 0.4761,
      "num_input_tokens_seen": 21011728,
      "step": 32190
    },
    {
      "epoch": 18.982900943396228,
      "grad_norm": 5.998030662536621,
      "learning_rate": 7.866517925235017e-08,
      "loss": 0.3162,
      "num_input_tokens_seen": 21015312,
      "step": 32195
    },
    {
      "epoch": 18.985849056603772,
      "grad_norm": 5.424129009246826,
      "learning_rate": 7.821126380626154e-08,
      "loss": 0.2954,
      "num_input_tokens_seen": 21018160,
      "step": 32200
    },
    {
      "epoch": 18.98879716981132,
      "grad_norm": 4.437142372131348,
      "learning_rate": 7.775865142495286e-08,
      "loss": 0.2075,
      "num_input_tokens_seen": 21020976,
      "step": 32205
    },
    {
      "epoch": 18.99174528301887,
      "grad_norm": 4.372691631317139,
      "learning_rate": 7.730734222825442e-08,
      "loss": 0.3183,
      "num_input_tokens_seen": 21024240,
      "step": 32210
    },
    {
      "epoch": 18.994693396226417,
      "grad_norm": 4.8950910568237305,
      "learning_rate": 7.68573363356534e-08,
      "loss": 0.1867,
      "num_input_tokens_seen": 21028400,
      "step": 32215
    },
    {
      "epoch": 18.99764150943396,
      "grad_norm": 4.0425896644592285,
      "learning_rate": 7.640863386629005e-08,
      "loss": 0.2273,
      "num_input_tokens_seen": 21032048,
      "step": 32220
    },
    {
      "epoch": 19.00058962264151,
      "grad_norm": 3.6532318592071533,
      "learning_rate": 7.59612349389599e-08,
      "loss": 0.3804,
      "num_input_tokens_seen": 21034792,
      "step": 32225
    },
    {
      "epoch": 19.003537735849058,
      "grad_norm": 4.5206074714660645,
      "learning_rate": 7.551513967211433e-08,
      "loss": 0.356,
      "num_input_tokens_seen": 21037768,
      "step": 32230
    },
    {
      "epoch": 19.006485849056602,
      "grad_norm": 4.272274017333984,
      "learning_rate": 7.507034818385883e-08,
      "loss": 0.2647,
      "num_input_tokens_seen": 21040552,
      "step": 32235
    },
    {
      "epoch": 19.00943396226415,
      "grad_norm": 4.421605587005615,
      "learning_rate": 7.462686059195423e-08,
      "loss": 0.2797,
      "num_input_tokens_seen": 21043368,
      "step": 32240
    },
    {
      "epoch": 19.0123820754717,
      "grad_norm": 4.233822822570801,
      "learning_rate": 7.418467701381548e-08,
      "loss": 0.3193,
      "num_input_tokens_seen": 21045928,
      "step": 32245
    },
    {
      "epoch": 19.015330188679247,
      "grad_norm": 5.534940242767334,
      "learning_rate": 7.374379756651285e-08,
      "loss": 0.2731,
      "num_input_tokens_seen": 21048616,
      "step": 32250
    },
    {
      "epoch": 19.01827830188679,
      "grad_norm": 5.380414009094238,
      "learning_rate": 7.330422236677015e-08,
      "loss": 0.2961,
      "num_input_tokens_seen": 21051304,
      "step": 32255
    },
    {
      "epoch": 19.02122641509434,
      "grad_norm": 3.872614860534668,
      "learning_rate": 7.286595153096765e-08,
      "loss": 0.4085,
      "num_input_tokens_seen": 21054856,
      "step": 32260
    },
    {
      "epoch": 19.024174528301888,
      "grad_norm": 2.8566131591796875,
      "learning_rate": 7.242898517513864e-08,
      "loss": 0.3108,
      "num_input_tokens_seen": 21057832,
      "step": 32265
    },
    {
      "epoch": 19.027122641509433,
      "grad_norm": 5.40977668762207,
      "learning_rate": 7.199332341497333e-08,
      "loss": 0.2392,
      "num_input_tokens_seen": 21060968,
      "step": 32270
    },
    {
      "epoch": 19.03007075471698,
      "grad_norm": 2.8786494731903076,
      "learning_rate": 7.155896636581394e-08,
      "loss": 0.2418,
      "num_input_tokens_seen": 21064232,
      "step": 32275
    },
    {
      "epoch": 19.03301886792453,
      "grad_norm": 5.631832599639893,
      "learning_rate": 7.112591414265901e-08,
      "loss": 0.2453,
      "num_input_tokens_seen": 21066792,
      "step": 32280
    },
    {
      "epoch": 19.035966981132077,
      "grad_norm": 3.814474582672119,
      "learning_rate": 7.069416686016018e-08,
      "loss": 0.4086,
      "num_input_tokens_seen": 21069288,
      "step": 32285
    },
    {
      "epoch": 19.03891509433962,
      "grad_norm": 2.784569263458252,
      "learning_rate": 7.026372463262488e-08,
      "loss": 0.2938,
      "num_input_tokens_seen": 21073544,
      "step": 32290
    },
    {
      "epoch": 19.04186320754717,
      "grad_norm": 3.023380756378174,
      "learning_rate": 6.983458757401418e-08,
      "loss": 0.307,
      "num_input_tokens_seen": 21077096,
      "step": 32295
    },
    {
      "epoch": 19.044811320754718,
      "grad_norm": 2.9062161445617676,
      "learning_rate": 6.940675579794443e-08,
      "loss": 0.3095,
      "num_input_tokens_seen": 21080104,
      "step": 32300
    },
    {
      "epoch": 19.047759433962263,
      "grad_norm": 6.451761722564697,
      "learning_rate": 6.898022941768612e-08,
      "loss": 0.2692,
      "num_input_tokens_seen": 21083304,
      "step": 32305
    },
    {
      "epoch": 19.05070754716981,
      "grad_norm": 3.206260919570923,
      "learning_rate": 6.855500854616337e-08,
      "loss": 0.3282,
      "num_input_tokens_seen": 21086920,
      "step": 32310
    },
    {
      "epoch": 19.05365566037736,
      "grad_norm": 2.1237645149230957,
      "learning_rate": 6.813109329595557e-08,
      "loss": 0.3242,
      "num_input_tokens_seen": 21090952,
      "step": 32315
    },
    {
      "epoch": 19.056603773584907,
      "grad_norm": 6.356142044067383,
      "learning_rate": 6.770848377929573e-08,
      "loss": 0.3587,
      "num_input_tokens_seen": 21095336,
      "step": 32320
    },
    {
      "epoch": 19.059551886792452,
      "grad_norm": 4.190639972686768,
      "learning_rate": 6.728718010807156e-08,
      "loss": 0.3279,
      "num_input_tokens_seen": 21098120,
      "step": 32325
    },
    {
      "epoch": 19.0625,
      "grad_norm": 3.594177484512329,
      "learning_rate": 6.68671823938255e-08,
      "loss": 0.2359,
      "num_input_tokens_seen": 21101096,
      "step": 32330
    },
    {
      "epoch": 19.065448113207548,
      "grad_norm": 5.312868118286133,
      "learning_rate": 6.644849074775361e-08,
      "loss": 0.3963,
      "num_input_tokens_seen": 21104072,
      "step": 32335
    },
    {
      "epoch": 19.068396226415093,
      "grad_norm": 2.4226672649383545,
      "learning_rate": 6.603110528070667e-08,
      "loss": 0.277,
      "num_input_tokens_seen": 21107176,
      "step": 32340
    },
    {
      "epoch": 19.07134433962264,
      "grad_norm": 5.041682720184326,
      "learning_rate": 6.561502610318849e-08,
      "loss": 0.2613,
      "num_input_tokens_seen": 21110056,
      "step": 32345
    },
    {
      "epoch": 19.07429245283019,
      "grad_norm": 3.1307733058929443,
      "learning_rate": 6.520025332535762e-08,
      "loss": 0.2424,
      "num_input_tokens_seen": 21113416,
      "step": 32350
    },
    {
      "epoch": 19.077240566037737,
      "grad_norm": 4.361947536468506,
      "learning_rate": 6.47867870570279e-08,
      "loss": 0.2071,
      "num_input_tokens_seen": 21116648,
      "step": 32355
    },
    {
      "epoch": 19.080188679245282,
      "grad_norm": 2.14378023147583,
      "learning_rate": 6.437462740766564e-08,
      "loss": 0.2245,
      "num_input_tokens_seen": 21120488,
      "step": 32360
    },
    {
      "epoch": 19.08313679245283,
      "grad_norm": 2.909815788269043,
      "learning_rate": 6.396377448639246e-08,
      "loss": 0.2974,
      "num_input_tokens_seen": 21123688,
      "step": 32365
    },
    {
      "epoch": 19.08608490566038,
      "grad_norm": 3.7488393783569336,
      "learning_rate": 6.3554228401983e-08,
      "loss": 0.2927,
      "num_input_tokens_seen": 21126760,
      "step": 32370
    },
    {
      "epoch": 19.089033018867923,
      "grad_norm": 15.224458694458008,
      "learning_rate": 6.314598926286663e-08,
      "loss": 0.2894,
      "num_input_tokens_seen": 21129224,
      "step": 32375
    },
    {
      "epoch": 19.09198113207547,
      "grad_norm": 3.416590929031372,
      "learning_rate": 6.273905717712637e-08,
      "loss": 0.2526,
      "num_input_tokens_seen": 21132072,
      "step": 32380
    },
    {
      "epoch": 19.09492924528302,
      "grad_norm": 2.687267303466797,
      "learning_rate": 6.233343225249933e-08,
      "loss": 0.3588,
      "num_input_tokens_seen": 21135592,
      "step": 32385
    },
    {
      "epoch": 19.097877358490567,
      "grad_norm": 3.487229347229004,
      "learning_rate": 6.192911459637519e-08,
      "loss": 0.4374,
      "num_input_tokens_seen": 21138824,
      "step": 32390
    },
    {
      "epoch": 19.100825471698112,
      "grad_norm": 3.9031264781951904,
      "learning_rate": 6.152610431580052e-08,
      "loss": 0.2376,
      "num_input_tokens_seen": 21141896,
      "step": 32395
    },
    {
      "epoch": 19.10377358490566,
      "grad_norm": 4.132111072540283,
      "learning_rate": 6.112440151747389e-08,
      "loss": 0.2266,
      "num_input_tokens_seen": 21145448,
      "step": 32400
    },
    {
      "epoch": 19.10672169811321,
      "grad_norm": 3.3005239963531494,
      "learning_rate": 6.072400630774689e-08,
      "loss": 0.3065,
      "num_input_tokens_seen": 21148616,
      "step": 32405
    },
    {
      "epoch": 19.109669811320753,
      "grad_norm": 3.4781055450439453,
      "learning_rate": 6.032491879262637e-08,
      "loss": 0.3904,
      "num_input_tokens_seen": 21151816,
      "step": 32410
    },
    {
      "epoch": 19.1126179245283,
      "grad_norm": 4.325366497039795,
      "learning_rate": 5.99271390777717e-08,
      "loss": 0.2691,
      "num_input_tokens_seen": 21155656,
      "step": 32415
    },
    {
      "epoch": 19.11556603773585,
      "grad_norm": 3.52805495262146,
      "learning_rate": 5.953066726849865e-08,
      "loss": 0.2806,
      "num_input_tokens_seen": 21159144,
      "step": 32420
    },
    {
      "epoch": 19.118514150943398,
      "grad_norm": 5.7765045166015625,
      "learning_rate": 5.913550346977326e-08,
      "loss": 0.2503,
      "num_input_tokens_seen": 21164616,
      "step": 32425
    },
    {
      "epoch": 19.121462264150942,
      "grad_norm": 2.5773887634277344,
      "learning_rate": 5.874164778621683e-08,
      "loss": 0.4119,
      "num_input_tokens_seen": 21167176,
      "step": 32430
    },
    {
      "epoch": 19.12441037735849,
      "grad_norm": 5.0169267654418945,
      "learning_rate": 5.834910032210539e-08,
      "loss": 0.3008,
      "num_input_tokens_seen": 21169736,
      "step": 32435
    },
    {
      "epoch": 19.12735849056604,
      "grad_norm": 6.32297945022583,
      "learning_rate": 5.795786118136693e-08,
      "loss": 0.361,
      "num_input_tokens_seen": 21172520,
      "step": 32440
    },
    {
      "epoch": 19.130306603773583,
      "grad_norm": 5.741361141204834,
      "learning_rate": 5.756793046758302e-08,
      "loss": 0.3234,
      "num_input_tokens_seen": 21175048,
      "step": 32445
    },
    {
      "epoch": 19.13325471698113,
      "grad_norm": 4.130574703216553,
      "learning_rate": 5.7179308283990544e-08,
      "loss": 0.3188,
      "num_input_tokens_seen": 21177672,
      "step": 32450
    },
    {
      "epoch": 19.13620283018868,
      "grad_norm": 3.929359197616577,
      "learning_rate": 5.679199473347885e-08,
      "loss": 0.3806,
      "num_input_tokens_seen": 21180712,
      "step": 32455
    },
    {
      "epoch": 19.139150943396228,
      "grad_norm": 4.547061443328857,
      "learning_rate": 5.6405989918590366e-08,
      "loss": 0.4888,
      "num_input_tokens_seen": 21184872,
      "step": 32460
    },
    {
      "epoch": 19.142099056603772,
      "grad_norm": 9.877241134643555,
      "learning_rate": 5.6021293941522225e-08,
      "loss": 0.3319,
      "num_input_tokens_seen": 21187784,
      "step": 32465
    },
    {
      "epoch": 19.14504716981132,
      "grad_norm": 4.080076694488525,
      "learning_rate": 5.563790690412352e-08,
      "loss": 0.2975,
      "num_input_tokens_seen": 21191656,
      "step": 32470
    },
    {
      "epoch": 19.14799528301887,
      "grad_norm": 2.6291699409484863,
      "learning_rate": 5.525582890789805e-08,
      "loss": 0.2556,
      "num_input_tokens_seen": 21194792,
      "step": 32475
    },
    {
      "epoch": 19.150943396226417,
      "grad_norm": 4.920339107513428,
      "learning_rate": 5.4875060054002115e-08,
      "loss": 0.3273,
      "num_input_tokens_seen": 21197640,
      "step": 32480
    },
    {
      "epoch": 19.15389150943396,
      "grad_norm": 3.7789671421051025,
      "learning_rate": 5.4495600443246755e-08,
      "loss": 0.3259,
      "num_input_tokens_seen": 21202056,
      "step": 32485
    },
    {
      "epoch": 19.15683962264151,
      "grad_norm": 5.435826301574707,
      "learning_rate": 5.411745017609493e-08,
      "loss": 0.4404,
      "num_input_tokens_seen": 21205192,
      "step": 32490
    },
    {
      "epoch": 19.159787735849058,
      "grad_norm": 2.7826404571533203,
      "learning_rate": 5.374060935266434e-08,
      "loss": 0.1705,
      "num_input_tokens_seen": 21207880,
      "step": 32495
    },
    {
      "epoch": 19.162735849056602,
      "grad_norm": 4.484762668609619,
      "learning_rate": 5.3365078072724065e-08,
      "loss": 0.2481,
      "num_input_tokens_seen": 21211112,
      "step": 32500
    },
    {
      "epoch": 19.16568396226415,
      "grad_norm": 3.878443479537964,
      "learning_rate": 5.299085643569846e-08,
      "loss": 0.306,
      "num_input_tokens_seen": 21214184,
      "step": 32505
    },
    {
      "epoch": 19.1686320754717,
      "grad_norm": 3.3356823921203613,
      "learning_rate": 5.261794454066327e-08,
      "loss": 0.4544,
      "num_input_tokens_seen": 21217128,
      "step": 32510
    },
    {
      "epoch": 19.171580188679247,
      "grad_norm": 3.0034141540527344,
      "learning_rate": 5.224634248635008e-08,
      "loss": 0.2095,
      "num_input_tokens_seen": 21220616,
      "step": 32515
    },
    {
      "epoch": 19.17452830188679,
      "grad_norm": 5.759814739227295,
      "learning_rate": 5.187605037114129e-08,
      "loss": 0.2812,
      "num_input_tokens_seen": 21224168,
      "step": 32520
    },
    {
      "epoch": 19.17747641509434,
      "grad_norm": 3.05263352394104,
      "learning_rate": 5.15070682930735e-08,
      "loss": 0.3054,
      "num_input_tokens_seen": 21227944,
      "step": 32525
    },
    {
      "epoch": 19.180424528301888,
      "grad_norm": 3.8854377269744873,
      "learning_rate": 5.113939634983578e-08,
      "loss": 0.3258,
      "num_input_tokens_seen": 21230184,
      "step": 32530
    },
    {
      "epoch": 19.183372641509433,
      "grad_norm": 3.7161104679107666,
      "learning_rate": 5.077303463877192e-08,
      "loss": 0.266,
      "num_input_tokens_seen": 21233768,
      "step": 32535
    },
    {
      "epoch": 19.18632075471698,
      "grad_norm": 4.1021833419799805,
      "learning_rate": 5.040798325687601e-08,
      "loss": 0.4034,
      "num_input_tokens_seen": 21237224,
      "step": 32540
    },
    {
      "epoch": 19.18926886792453,
      "grad_norm": 2.2263023853302,
      "learning_rate": 5.004424230079852e-08,
      "loss": 0.2871,
      "num_input_tokens_seen": 21240392,
      "step": 32545
    },
    {
      "epoch": 19.192216981132077,
      "grad_norm": 4.2272467613220215,
      "learning_rate": 4.968181186684129e-08,
      "loss": 0.2167,
      "num_input_tokens_seen": 21243880,
      "step": 32550
    },
    {
      "epoch": 19.19516509433962,
      "grad_norm": 3.266697406768799,
      "learning_rate": 4.932069205095924e-08,
      "loss": 0.3156,
      "num_input_tokens_seen": 21247592,
      "step": 32555
    },
    {
      "epoch": 19.19811320754717,
      "grad_norm": 5.634979724884033,
      "learning_rate": 4.896088294875978e-08,
      "loss": 0.3853,
      "num_input_tokens_seen": 21250056,
      "step": 32560
    },
    {
      "epoch": 19.201061320754718,
      "grad_norm": 5.005356311798096,
      "learning_rate": 4.8602384655505044e-08,
      "loss": 0.3021,
      "num_input_tokens_seen": 21252872,
      "step": 32565
    },
    {
      "epoch": 19.204009433962263,
      "grad_norm": 4.0302324295043945,
      "learning_rate": 4.824519726610744e-08,
      "loss": 0.3121,
      "num_input_tokens_seen": 21256904,
      "step": 32570
    },
    {
      "epoch": 19.20695754716981,
      "grad_norm": 4.050544738769531,
      "learning_rate": 4.7889320875135206e-08,
      "loss": 0.2636,
      "num_input_tokens_seen": 21259848,
      "step": 32575
    },
    {
      "epoch": 19.20990566037736,
      "grad_norm": 15.176353454589844,
      "learning_rate": 4.753475557680742e-08,
      "loss": 0.3051,
      "num_input_tokens_seen": 21262760,
      "step": 32580
    },
    {
      "epoch": 19.212853773584907,
      "grad_norm": 5.9407057762146,
      "learning_rate": 4.718150146499734e-08,
      "loss": 0.2632,
      "num_input_tokens_seen": 21265928,
      "step": 32585
    },
    {
      "epoch": 19.215801886792452,
      "grad_norm": 5.718442440032959,
      "learning_rate": 4.682955863323013e-08,
      "loss": 0.2068,
      "num_input_tokens_seen": 21268936,
      "step": 32590
    },
    {
      "epoch": 19.21875,
      "grad_norm": 3.8035826683044434,
      "learning_rate": 4.6478927174684606e-08,
      "loss": 0.2969,
      "num_input_tokens_seen": 21271912,
      "step": 32595
    },
    {
      "epoch": 19.221698113207548,
      "grad_norm": 9.988144874572754,
      "learning_rate": 4.612960718219095e-08,
      "loss": 0.273,
      "num_input_tokens_seen": 21274536,
      "step": 32600
    },
    {
      "epoch": 19.224646226415093,
      "grad_norm": 3.817896604537964,
      "learning_rate": 4.578159874823407e-08,
      "loss": 0.2829,
      "num_input_tokens_seen": 21278792,
      "step": 32605
    },
    {
      "epoch": 19.22759433962264,
      "grad_norm": 2.4605555534362793,
      "learning_rate": 4.5434901964950264e-08,
      "loss": 0.2358,
      "num_input_tokens_seen": 21281544,
      "step": 32610
    },
    {
      "epoch": 19.23054245283019,
      "grad_norm": 3.4008195400238037,
      "learning_rate": 4.508951692412944e-08,
      "loss": 0.4405,
      "num_input_tokens_seen": 21284712,
      "step": 32615
    },
    {
      "epoch": 19.233490566037737,
      "grad_norm": 3.319037675857544,
      "learning_rate": 4.4745443717213455e-08,
      "loss": 0.2929,
      "num_input_tokens_seen": 21288232,
      "step": 32620
    },
    {
      "epoch": 19.236438679245282,
      "grad_norm": 3.9437336921691895,
      "learning_rate": 4.4402682435296666e-08,
      "loss": 0.2574,
      "num_input_tokens_seen": 21292040,
      "step": 32625
    },
    {
      "epoch": 19.23938679245283,
      "grad_norm": 3.434069871902466,
      "learning_rate": 4.406123316912758e-08,
      "loss": 0.3215,
      "num_input_tokens_seen": 21295528,
      "step": 32630
    },
    {
      "epoch": 19.24233490566038,
      "grad_norm": 2.705620050430298,
      "learning_rate": 4.372109600910612e-08,
      "loss": 0.1895,
      "num_input_tokens_seen": 21298216,
      "step": 32635
    },
    {
      "epoch": 19.245283018867923,
      "grad_norm": 2.950105667114258,
      "learning_rate": 4.338227104528414e-08,
      "loss": 0.2654,
      "num_input_tokens_seen": 21301352,
      "step": 32640
    },
    {
      "epoch": 19.24823113207547,
      "grad_norm": 4.302958965301514,
      "learning_rate": 4.304475836736821e-08,
      "loss": 0.4293,
      "num_input_tokens_seen": 21303720,
      "step": 32645
    },
    {
      "epoch": 19.25117924528302,
      "grad_norm": 7.830981254577637,
      "learning_rate": 4.27085580647163e-08,
      "loss": 0.405,
      "num_input_tokens_seen": 21306760,
      "step": 32650
    },
    {
      "epoch": 19.254127358490567,
      "grad_norm": 3.535897970199585,
      "learning_rate": 4.237367022633776e-08,
      "loss": 0.342,
      "num_input_tokens_seen": 21311752,
      "step": 32655
    },
    {
      "epoch": 19.257075471698112,
      "grad_norm": 4.188340187072754,
      "learning_rate": 4.204009494089612e-08,
      "loss": 0.2638,
      "num_input_tokens_seen": 21314696,
      "step": 32660
    },
    {
      "epoch": 19.26002358490566,
      "grad_norm": 5.6307454109191895,
      "learning_rate": 4.170783229670739e-08,
      "loss": 0.2289,
      "num_input_tokens_seen": 21317960,
      "step": 32665
    },
    {
      "epoch": 19.26297169811321,
      "grad_norm": 4.13720703125,
      "learning_rate": 4.137688238173898e-08,
      "loss": 0.2801,
      "num_input_tokens_seen": 21322024,
      "step": 32670
    },
    {
      "epoch": 19.265919811320753,
      "grad_norm": 2.883984327316284,
      "learning_rate": 4.104724528361137e-08,
      "loss": 0.2995,
      "num_input_tokens_seen": 21325384,
      "step": 32675
    },
    {
      "epoch": 19.2688679245283,
      "grad_norm": 2.080756664276123,
      "learning_rate": 4.071892108959752e-08,
      "loss": 0.3396,
      "num_input_tokens_seen": 21328680,
      "step": 32680
    },
    {
      "epoch": 19.27181603773585,
      "grad_norm": 4.280841827392578,
      "learning_rate": 4.039190988662234e-08,
      "loss": 0.2063,
      "num_input_tokens_seen": 21331432,
      "step": 32685
    },
    {
      "epoch": 19.274764150943398,
      "grad_norm": 3.937467575073242,
      "learning_rate": 4.006621176126435e-08,
      "loss": 0.2236,
      "num_input_tokens_seen": 21334728,
      "step": 32690
    },
    {
      "epoch": 19.277712264150942,
      "grad_norm": 3.7968103885650635,
      "learning_rate": 3.974182679975236e-08,
      "loss": 0.3599,
      "num_input_tokens_seen": 21338376,
      "step": 32695
    },
    {
      "epoch": 19.28066037735849,
      "grad_norm": 3.42622971534729,
      "learning_rate": 3.941875508796933e-08,
      "loss": 0.2961,
      "num_input_tokens_seen": 21341032,
      "step": 32700
    },
    {
      "epoch": 19.28360849056604,
      "grad_norm": 4.56218957901001,
      "learning_rate": 3.909699671145017e-08,
      "loss": 0.281,
      "num_input_tokens_seen": 21344264,
      "step": 32705
    },
    {
      "epoch": 19.286556603773583,
      "grad_norm": 3.310807704925537,
      "learning_rate": 3.87765517553812e-08,
      "loss": 0.3551,
      "num_input_tokens_seen": 21346920,
      "step": 32710
    },
    {
      "epoch": 19.28950471698113,
      "grad_norm": 4.824003219604492,
      "learning_rate": 3.8457420304601756e-08,
      "loss": 0.3673,
      "num_input_tokens_seen": 21349896,
      "step": 32715
    },
    {
      "epoch": 19.29245283018868,
      "grad_norm": 4.598348140716553,
      "learning_rate": 3.813960244360371e-08,
      "loss": 0.2812,
      "num_input_tokens_seen": 21352808,
      "step": 32720
    },
    {
      "epoch": 19.295400943396228,
      "grad_norm": 4.563314914703369,
      "learning_rate": 3.7823098256529744e-08,
      "loss": 0.3668,
      "num_input_tokens_seen": 21356136,
      "step": 32725
    },
    {
      "epoch": 19.298349056603772,
      "grad_norm": 3.006699800491333,
      "learning_rate": 3.750790782717673e-08,
      "loss": 0.3403,
      "num_input_tokens_seen": 21360808,
      "step": 32730
    },
    {
      "epoch": 19.30129716981132,
      "grad_norm": 3.948866128921509,
      "learning_rate": 3.719403123899179e-08,
      "loss": 0.2883,
      "num_input_tokens_seen": 21363464,
      "step": 32735
    },
    {
      "epoch": 19.30424528301887,
      "grad_norm": 2.479295492172241,
      "learning_rate": 3.688146857507624e-08,
      "loss": 0.2235,
      "num_input_tokens_seen": 21368104,
      "step": 32740
    },
    {
      "epoch": 19.307193396226417,
      "grad_norm": 5.731419086456299,
      "learning_rate": 3.657021991818166e-08,
      "loss": 0.4012,
      "num_input_tokens_seen": 21370824,
      "step": 32745
    },
    {
      "epoch": 19.31014150943396,
      "grad_norm": 6.35030460357666,
      "learning_rate": 3.626028535071213e-08,
      "loss": 0.2299,
      "num_input_tokens_seen": 21373512,
      "step": 32750
    },
    {
      "epoch": 19.31308962264151,
      "grad_norm": 2.27323055267334,
      "learning_rate": 3.59516649547248e-08,
      "loss": 0.2486,
      "num_input_tokens_seen": 21376552,
      "step": 32755
    },
    {
      "epoch": 19.316037735849058,
      "grad_norm": 5.354336738586426,
      "learning_rate": 3.564435881192818e-08,
      "loss": 0.2959,
      "num_input_tokens_seen": 21379368,
      "step": 32760
    },
    {
      "epoch": 19.318985849056602,
      "grad_norm": 3.7711942195892334,
      "learning_rate": 3.5338367003682763e-08,
      "loss": 0.2732,
      "num_input_tokens_seen": 21382888,
      "step": 32765
    },
    {
      "epoch": 19.32193396226415,
      "grad_norm": 3.641944408416748,
      "learning_rate": 3.5033689611000954e-08,
      "loss": 0.2127,
      "num_input_tokens_seen": 21385768,
      "step": 32770
    },
    {
      "epoch": 19.3248820754717,
      "grad_norm": 5.2400736808776855,
      "learning_rate": 3.473032671454768e-08,
      "loss": 0.4212,
      "num_input_tokens_seen": 21388904,
      "step": 32775
    },
    {
      "epoch": 19.327830188679247,
      "grad_norm": 4.567250728607178,
      "learning_rate": 3.44282783946398e-08,
      "loss": 0.2825,
      "num_input_tokens_seen": 21391624,
      "step": 32780
    },
    {
      "epoch": 19.33077830188679,
      "grad_norm": 1.8661001920700073,
      "learning_rate": 3.4127544731245575e-08,
      "loss": 0.3603,
      "num_input_tokens_seen": 21395912,
      "step": 32785
    },
    {
      "epoch": 19.33372641509434,
      "grad_norm": 3.29093861579895,
      "learning_rate": 3.38281258039852e-08,
      "loss": 0.1849,
      "num_input_tokens_seen": 21399496,
      "step": 32790
    },
    {
      "epoch": 19.336674528301888,
      "grad_norm": 6.583939075469971,
      "learning_rate": 3.353002169213193e-08,
      "loss": 0.303,
      "num_input_tokens_seen": 21402088,
      "step": 32795
    },
    {
      "epoch": 19.339622641509433,
      "grad_norm": 4.505652904510498,
      "learning_rate": 3.32332324746093e-08,
      "loss": 0.4208,
      "num_input_tokens_seen": 21404936,
      "step": 32800
    },
    {
      "epoch": 19.34257075471698,
      "grad_norm": 3.436218500137329,
      "learning_rate": 3.2937758229994455e-08,
      "loss": 0.3071,
      "num_input_tokens_seen": 21408680,
      "step": 32805
    },
    {
      "epoch": 19.34551886792453,
      "grad_norm": 4.795374870300293,
      "learning_rate": 3.2643599036514815e-08,
      "loss": 0.2711,
      "num_input_tokens_seen": 21412072,
      "step": 32810
    },
    {
      "epoch": 19.348466981132077,
      "grad_norm": 4.39146614074707,
      "learning_rate": 3.2350754972050316e-08,
      "loss": 0.2704,
      "num_input_tokens_seen": 21415016,
      "step": 32815
    },
    {
      "epoch": 19.35141509433962,
      "grad_norm": 2.3008618354797363,
      "learning_rate": 3.2059226114132815e-08,
      "loss": 0.2713,
      "num_input_tokens_seen": 21417928,
      "step": 32820
    },
    {
      "epoch": 19.35436320754717,
      "grad_norm": 2.805619955062866,
      "learning_rate": 3.1769012539945575e-08,
      "loss": 0.2296,
      "num_input_tokens_seen": 21420136,
      "step": 32825
    },
    {
      "epoch": 19.357311320754718,
      "grad_norm": 6.54548978805542,
      "learning_rate": 3.1480114326324364e-08,
      "loss": 0.3328,
      "num_input_tokens_seen": 21422408,
      "step": 32830
    },
    {
      "epoch": 19.360259433962263,
      "grad_norm": 6.361226558685303,
      "learning_rate": 3.1192531549756325e-08,
      "loss": 0.2794,
      "num_input_tokens_seen": 21425352,
      "step": 32835
    },
    {
      "epoch": 19.36320754716981,
      "grad_norm": 4.07490873336792,
      "learning_rate": 3.090626428638e-08,
      "loss": 0.4428,
      "num_input_tokens_seen": 21428392,
      "step": 32840
    },
    {
      "epoch": 19.36615566037736,
      "grad_norm": 3.5838065147399902,
      "learning_rate": 3.062131261198531e-08,
      "loss": 0.33,
      "num_input_tokens_seen": 21431272,
      "step": 32845
    },
    {
      "epoch": 19.369103773584907,
      "grad_norm": 4.106979846954346,
      "learning_rate": 3.033767660201525e-08,
      "loss": 0.3193,
      "num_input_tokens_seen": 21434248,
      "step": 32850
    },
    {
      "epoch": 19.372051886792452,
      "grad_norm": 2.625016450881958,
      "learning_rate": 3.005535633156309e-08,
      "loss": 0.1988,
      "num_input_tokens_seen": 21437320,
      "step": 32855
    },
    {
      "epoch": 19.375,
      "grad_norm": 2.410367488861084,
      "learning_rate": 2.977435187537514e-08,
      "loss": 0.2124,
      "num_input_tokens_seen": 21440552,
      "step": 32860
    },
    {
      "epoch": 19.377948113207548,
      "grad_norm": 3.2358736991882324,
      "learning_rate": 2.949466330784745e-08,
      "loss": 0.2473,
      "num_input_tokens_seen": 21444488,
      "step": 32865
    },
    {
      "epoch": 19.380896226415093,
      "grad_norm": 6.545119285583496,
      "learning_rate": 2.921629070302967e-08,
      "loss": 0.3404,
      "num_input_tokens_seen": 21447304,
      "step": 32870
    },
    {
      "epoch": 19.38384433962264,
      "grad_norm": 5.443304061889648,
      "learning_rate": 2.893923413462174e-08,
      "loss": 0.3005,
      "num_input_tokens_seen": 21450152,
      "step": 32875
    },
    {
      "epoch": 19.38679245283019,
      "grad_norm": 4.565801620483398,
      "learning_rate": 2.866349367597554e-08,
      "loss": 0.3742,
      "num_input_tokens_seen": 21452744,
      "step": 32880
    },
    {
      "epoch": 19.389740566037737,
      "grad_norm": 4.493298053741455,
      "learning_rate": 2.8389069400094893e-08,
      "loss": 0.3459,
      "num_input_tokens_seen": 21457224,
      "step": 32885
    },
    {
      "epoch": 19.392688679245282,
      "grad_norm": 6.825876712799072,
      "learning_rate": 2.811596137963446e-08,
      "loss": 0.2872,
      "num_input_tokens_seen": 21460616,
      "step": 32890
    },
    {
      "epoch": 19.39563679245283,
      "grad_norm": 3.8724637031555176,
      "learning_rate": 2.7844169686900844e-08,
      "loss": 0.2755,
      "num_input_tokens_seen": 21464488,
      "step": 32895
    },
    {
      "epoch": 19.39858490566038,
      "grad_norm": 2.525320291519165,
      "learning_rate": 2.7573694393852047e-08,
      "loss": 0.2681,
      "num_input_tokens_seen": 21467720,
      "step": 32900
    },
    {
      "epoch": 19.401533018867923,
      "grad_norm": 5.503297805786133,
      "learning_rate": 2.7304535572098e-08,
      "loss": 0.3072,
      "num_input_tokens_seen": 21470600,
      "step": 32905
    },
    {
      "epoch": 19.40448113207547,
      "grad_norm": 2.876127243041992,
      "learning_rate": 2.703669329289893e-08,
      "loss": 0.253,
      "num_input_tokens_seen": 21473992,
      "step": 32910
    },
    {
      "epoch": 19.40742924528302,
      "grad_norm": 5.287059783935547,
      "learning_rate": 2.6770167627167554e-08,
      "loss": 0.4588,
      "num_input_tokens_seen": 21477672,
      "step": 32915
    },
    {
      "epoch": 19.410377358490567,
      "grad_norm": 4.27646017074585,
      "learning_rate": 2.6504958645467426e-08,
      "loss": 0.3745,
      "num_input_tokens_seen": 21480648,
      "step": 32920
    },
    {
      "epoch": 19.413325471698112,
      "grad_norm": 4.470993995666504,
      "learning_rate": 2.6241066418014605e-08,
      "loss": 0.2767,
      "num_input_tokens_seen": 21483208,
      "step": 32925
    },
    {
      "epoch": 19.41627358490566,
      "grad_norm": 2.758554697036743,
      "learning_rate": 2.5978491014674866e-08,
      "loss": 0.1813,
      "num_input_tokens_seen": 21486504,
      "step": 32930
    },
    {
      "epoch": 19.41922169811321,
      "grad_norm": 3.8388237953186035,
      "learning_rate": 2.571723250496705e-08,
      "loss": 0.2385,
      "num_input_tokens_seen": 21490024,
      "step": 32935
    },
    {
      "epoch": 19.422169811320753,
      "grad_norm": 3.5145936012268066,
      "learning_rate": 2.5457290958059155e-08,
      "loss": 0.2529,
      "num_input_tokens_seen": 21493704,
      "step": 32940
    },
    {
      "epoch": 19.4251179245283,
      "grad_norm": 4.431590557098389,
      "learning_rate": 2.51986664427728e-08,
      "loss": 0.2939,
      "num_input_tokens_seen": 21497448,
      "step": 32945
    },
    {
      "epoch": 19.42806603773585,
      "grad_norm": 9.322731018066406,
      "learning_rate": 2.4941359027579883e-08,
      "loss": 0.474,
      "num_input_tokens_seen": 21500520,
      "step": 32950
    },
    {
      "epoch": 19.431014150943398,
      "grad_norm": 11.131851196289062,
      "learning_rate": 2.468536878060368e-08,
      "loss": 0.2568,
      "num_input_tokens_seen": 21503784,
      "step": 32955
    },
    {
      "epoch": 19.433962264150942,
      "grad_norm": 6.309096813201904,
      "learning_rate": 2.443069576961832e-08,
      "loss": 0.3477,
      "num_input_tokens_seen": 21507048,
      "step": 32960
    },
    {
      "epoch": 19.43691037735849,
      "grad_norm": 3.2915539741516113,
      "learning_rate": 2.4177340062049304e-08,
      "loss": 0.265,
      "num_input_tokens_seen": 21509512,
      "step": 32965
    },
    {
      "epoch": 19.43985849056604,
      "grad_norm": 4.698244094848633,
      "learning_rate": 2.3925301724974647e-08,
      "loss": 0.3565,
      "num_input_tokens_seen": 21512232,
      "step": 32970
    },
    {
      "epoch": 19.442806603773583,
      "grad_norm": 6.325228214263916,
      "learning_rate": 2.367458082512153e-08,
      "loss": 0.2231,
      "num_input_tokens_seen": 21515464,
      "step": 32975
    },
    {
      "epoch": 19.44575471698113,
      "grad_norm": 4.202943801879883,
      "learning_rate": 2.3425177428870737e-08,
      "loss": 0.2759,
      "num_input_tokens_seen": 21518312,
      "step": 32980
    },
    {
      "epoch": 19.44870283018868,
      "grad_norm": 5.13239860534668,
      "learning_rate": 2.3177091602251677e-08,
      "loss": 0.2931,
      "num_input_tokens_seen": 21521576,
      "step": 32985
    },
    {
      "epoch": 19.451650943396228,
      "grad_norm": 4.273693084716797,
      "learning_rate": 2.2930323410946254e-08,
      "loss": 0.2695,
      "num_input_tokens_seen": 21525640,
      "step": 32990
    },
    {
      "epoch": 19.454599056603772,
      "grad_norm": 3.2863235473632812,
      "learning_rate": 2.2684872920287758e-08,
      "loss": 0.3224,
      "num_input_tokens_seen": 21529608,
      "step": 32995
    },
    {
      "epoch": 19.45754716981132,
      "grad_norm": 4.616123676300049,
      "learning_rate": 2.2440740195260323e-08,
      "loss": 0.3279,
      "num_input_tokens_seen": 21532840,
      "step": 33000
    },
    {
      "epoch": 19.46049528301887,
      "grad_norm": 5.251607894897461,
      "learning_rate": 2.219792530049891e-08,
      "loss": 0.4382,
      "num_input_tokens_seen": 21538696,
      "step": 33005
    },
    {
      "epoch": 19.463443396226417,
      "grad_norm": 3.545989513397217,
      "learning_rate": 2.1956428300290434e-08,
      "loss": 0.2382,
      "num_input_tokens_seen": 21542216,
      "step": 33010
    },
    {
      "epoch": 19.46639150943396,
      "grad_norm": 4.449833869934082,
      "learning_rate": 2.1716249258570966e-08,
      "loss": 0.3048,
      "num_input_tokens_seen": 21544808,
      "step": 33015
    },
    {
      "epoch": 19.46933962264151,
      "grad_norm": 3.825828790664673,
      "learning_rate": 2.1477388238930196e-08,
      "loss": 0.2526,
      "num_input_tokens_seen": 21548552,
      "step": 33020
    },
    {
      "epoch": 19.472287735849058,
      "grad_norm": 6.2495317459106445,
      "learning_rate": 2.1239845304606988e-08,
      "loss": 0.3218,
      "num_input_tokens_seen": 21551272,
      "step": 33025
    },
    {
      "epoch": 19.475235849056602,
      "grad_norm": 4.540937900543213,
      "learning_rate": 2.100362051849214e-08,
      "loss": 0.3461,
      "num_input_tokens_seen": 21554952,
      "step": 33030
    },
    {
      "epoch": 19.47818396226415,
      "grad_norm": 3.5415186882019043,
      "learning_rate": 2.076871394312674e-08,
      "loss": 0.2865,
      "num_input_tokens_seen": 21558952,
      "step": 33035
    },
    {
      "epoch": 19.4811320754717,
      "grad_norm": 3.3420169353485107,
      "learning_rate": 2.0535125640703813e-08,
      "loss": 0.3412,
      "num_input_tokens_seen": 21561800,
      "step": 33040
    },
    {
      "epoch": 19.484080188679247,
      "grad_norm": 3.171975612640381,
      "learning_rate": 2.0302855673066667e-08,
      "loss": 0.2964,
      "num_input_tokens_seen": 21564616,
      "step": 33045
    },
    {
      "epoch": 19.48702830188679,
      "grad_norm": 4.28542423248291,
      "learning_rate": 2.0071904101710004e-08,
      "loss": 0.271,
      "num_input_tokens_seen": 21567240,
      "step": 33050
    },
    {
      "epoch": 19.48997641509434,
      "grad_norm": 6.73660945892334,
      "learning_rate": 1.98422709877788e-08,
      "loss": 0.4516,
      "num_input_tokens_seen": 21569896,
      "step": 33055
    },
    {
      "epoch": 19.492924528301888,
      "grad_norm": 11.652873039245605,
      "learning_rate": 1.961395639206942e-08,
      "loss": 0.3562,
      "num_input_tokens_seen": 21573064,
      "step": 33060
    },
    {
      "epoch": 19.495872641509433,
      "grad_norm": 3.7151145935058594,
      "learning_rate": 1.9386960375029628e-08,
      "loss": 0.201,
      "num_input_tokens_seen": 21576232,
      "step": 33065
    },
    {
      "epoch": 19.49882075471698,
      "grad_norm": 3.0851712226867676,
      "learning_rate": 1.9161282996757458e-08,
      "loss": 0.3357,
      "num_input_tokens_seen": 21579208,
      "step": 33070
    },
    {
      "epoch": 19.50176886792453,
      "grad_norm": 2.995922803878784,
      "learning_rate": 1.8936924317001225e-08,
      "loss": 0.3204,
      "num_input_tokens_seen": 21582504,
      "step": 33075
    },
    {
      "epoch": 19.504716981132077,
      "grad_norm": 1.9775701761245728,
      "learning_rate": 1.8713884395162308e-08,
      "loss": 0.179,
      "num_input_tokens_seen": 21585416,
      "step": 33080
    },
    {
      "epoch": 19.50766509433962,
      "grad_norm": 5.540274620056152,
      "learning_rate": 1.8492163290290132e-08,
      "loss": 0.4274,
      "num_input_tokens_seen": 21588488,
      "step": 33085
    },
    {
      "epoch": 19.51061320754717,
      "grad_norm": 3.431436538696289,
      "learning_rate": 1.827176106108719e-08,
      "loss": 0.3146,
      "num_input_tokens_seen": 21591592,
      "step": 33090
    },
    {
      "epoch": 19.513561320754718,
      "grad_norm": 3.6508395671844482,
      "learning_rate": 1.8052677765905137e-08,
      "loss": 0.3905,
      "num_input_tokens_seen": 21595432,
      "step": 33095
    },
    {
      "epoch": 19.516509433962263,
      "grad_norm": 5.253223896026611,
      "learning_rate": 1.783491346274757e-08,
      "loss": 0.3075,
      "num_input_tokens_seen": 21597864,
      "step": 33100
    },
    {
      "epoch": 19.51945754716981,
      "grad_norm": 4.272646427154541,
      "learning_rate": 1.7618468209268936e-08,
      "loss": 0.2955,
      "num_input_tokens_seen": 21600712,
      "step": 33105
    },
    {
      "epoch": 19.52240566037736,
      "grad_norm": 1.939947247505188,
      "learning_rate": 1.7403342062773943e-08,
      "loss": 0.2128,
      "num_input_tokens_seen": 21606216,
      "step": 33110
    },
    {
      "epoch": 19.525353773584907,
      "grad_norm": 5.6585822105407715,
      "learning_rate": 1.718953508021759e-08,
      "loss": 0.3772,
      "num_input_tokens_seen": 21609768,
      "step": 33115
    },
    {
      "epoch": 19.528301886792452,
      "grad_norm": 2.951385259628296,
      "learning_rate": 1.6977047318206262e-08,
      "loss": 0.4107,
      "num_input_tokens_seen": 21612200,
      "step": 33120
    },
    {
      "epoch": 19.53125,
      "grad_norm": 2.8376553058624268,
      "learning_rate": 1.676587883299774e-08,
      "loss": 0.3996,
      "num_input_tokens_seen": 21615752,
      "step": 33125
    },
    {
      "epoch": 19.534198113207548,
      "grad_norm": 3.900975227355957,
      "learning_rate": 1.655602968049952e-08,
      "loss": 0.2302,
      "num_input_tokens_seen": 21618888,
      "step": 33130
    },
    {
      "epoch": 19.537146226415093,
      "grad_norm": 3.894360065460205,
      "learning_rate": 1.634749991626938e-08,
      "loss": 0.3375,
      "num_input_tokens_seen": 21621352,
      "step": 33135
    },
    {
      "epoch": 19.54009433962264,
      "grad_norm": 3.01920223236084,
      "learning_rate": 1.6140289595517056e-08,
      "loss": 0.3306,
      "num_input_tokens_seen": 21625960,
      "step": 33140
    },
    {
      "epoch": 19.54304245283019,
      "grad_norm": 3.068626880645752,
      "learning_rate": 1.5934398773102545e-08,
      "loss": 0.3775,
      "num_input_tokens_seen": 21629480,
      "step": 33145
    },
    {
      "epoch": 19.545990566037737,
      "grad_norm": 4.055754661560059,
      "learning_rate": 1.5729827503536133e-08,
      "loss": 0.325,
      "num_input_tokens_seen": 21633160,
      "step": 33150
    },
    {
      "epoch": 19.548938679245282,
      "grad_norm": 3.890103340148926,
      "learning_rate": 1.5526575840978942e-08,
      "loss": 0.2864,
      "num_input_tokens_seen": 21635752,
      "step": 33155
    },
    {
      "epoch": 19.55188679245283,
      "grad_norm": 2.6376302242279053,
      "learning_rate": 1.532464383924237e-08,
      "loss": 0.1994,
      "num_input_tokens_seen": 21638792,
      "step": 33160
    },
    {
      "epoch": 19.55483490566038,
      "grad_norm": 3.577603816986084,
      "learning_rate": 1.5124031551789208e-08,
      "loss": 0.3359,
      "num_input_tokens_seen": 21641672,
      "step": 33165
    },
    {
      "epoch": 19.557783018867923,
      "grad_norm": 4.259702205657959,
      "learning_rate": 1.4924739031732527e-08,
      "loss": 0.3339,
      "num_input_tokens_seen": 21644232,
      "step": 33170
    },
    {
      "epoch": 19.56073113207547,
      "grad_norm": 4.122569561004639,
      "learning_rate": 1.4726766331835118e-08,
      "loss": 0.3654,
      "num_input_tokens_seen": 21647624,
      "step": 33175
    },
    {
      "epoch": 19.56367924528302,
      "grad_norm": 3.83862566947937,
      "learning_rate": 1.4530113504512278e-08,
      "loss": 0.2071,
      "num_input_tokens_seen": 21650152,
      "step": 33180
    },
    {
      "epoch": 19.566627358490567,
      "grad_norm": 4.257011890411377,
      "learning_rate": 1.4334780601827914e-08,
      "loss": 0.3733,
      "num_input_tokens_seen": 21653608,
      "step": 33185
    },
    {
      "epoch": 19.569575471698112,
      "grad_norm": 4.108937740325928,
      "learning_rate": 1.4140767675497325e-08,
      "loss": 0.3154,
      "num_input_tokens_seen": 21658120,
      "step": 33190
    },
    {
      "epoch": 19.57252358490566,
      "grad_norm": 1.9480382204055786,
      "learning_rate": 1.394807477688609e-08,
      "loss": 0.1911,
      "num_input_tokens_seen": 21661480,
      "step": 33195
    },
    {
      "epoch": 19.57547169811321,
      "grad_norm": 4.602447509765625,
      "learning_rate": 1.3756701957011177e-08,
      "loss": 0.2926,
      "num_input_tokens_seen": 21664456,
      "step": 33200
    },
    {
      "epoch": 19.578419811320753,
      "grad_norm": 4.1927170753479,
      "learning_rate": 1.3566649266538723e-08,
      "loss": 0.2349,
      "num_input_tokens_seen": 21667592,
      "step": 33205
    },
    {
      "epoch": 19.5813679245283,
      "grad_norm": 2.297913074493408,
      "learning_rate": 1.3377916755786257e-08,
      "loss": 0.3306,
      "num_input_tokens_seen": 21670536,
      "step": 33210
    },
    {
      "epoch": 19.58431603773585,
      "grad_norm": 4.387843132019043,
      "learning_rate": 1.3190504474721588e-08,
      "loss": 0.2972,
      "num_input_tokens_seen": 21673256,
      "step": 33215
    },
    {
      "epoch": 19.587264150943398,
      "grad_norm": 3.084247589111328,
      "learning_rate": 1.3004412472962802e-08,
      "loss": 0.2841,
      "num_input_tokens_seen": 21676424,
      "step": 33220
    },
    {
      "epoch": 19.590212264150942,
      "grad_norm": 6.051823139190674,
      "learning_rate": 1.2819640799778266e-08,
      "loss": 0.2338,
      "num_input_tokens_seen": 21679528,
      "step": 33225
    },
    {
      "epoch": 19.59316037735849,
      "grad_norm": 4.355128288269043,
      "learning_rate": 1.2636189504087737e-08,
      "loss": 0.3074,
      "num_input_tokens_seen": 21682088,
      "step": 33230
    },
    {
      "epoch": 19.59610849056604,
      "grad_norm": 3.2457728385925293,
      "learning_rate": 1.2454058634460142e-08,
      "loss": 0.281,
      "num_input_tokens_seen": 21685192,
      "step": 33235
    },
    {
      "epoch": 19.599056603773583,
      "grad_norm": 5.141165733337402,
      "learning_rate": 1.2273248239115798e-08,
      "loss": 0.3018,
      "num_input_tokens_seen": 21687912,
      "step": 33240
    },
    {
      "epoch": 19.60200471698113,
      "grad_norm": 2.468722105026245,
      "learning_rate": 1.2093758365924746e-08,
      "loss": 0.2699,
      "num_input_tokens_seen": 21691560,
      "step": 33245
    },
    {
      "epoch": 19.60495283018868,
      "grad_norm": 2.7395548820495605,
      "learning_rate": 1.1915589062408417e-08,
      "loss": 0.3169,
      "num_input_tokens_seen": 21694696,
      "step": 33250
    },
    {
      "epoch": 19.607900943396228,
      "grad_norm": 3.670421600341797,
      "learning_rate": 1.1738740375736301e-08,
      "loss": 0.2779,
      "num_input_tokens_seen": 21698344,
      "step": 33255
    },
    {
      "epoch": 19.610849056603772,
      "grad_norm": 6.9125823974609375,
      "learning_rate": 1.15632123527315e-08,
      "loss": 0.2563,
      "num_input_tokens_seen": 21702024,
      "step": 33260
    },
    {
      "epoch": 19.61379716981132,
      "grad_norm": 4.731331825256348,
      "learning_rate": 1.1389005039865176e-08,
      "loss": 0.2604,
      "num_input_tokens_seen": 21705736,
      "step": 33265
    },
    {
      "epoch": 19.61674528301887,
      "grad_norm": 3.6765620708465576,
      "learning_rate": 1.1216118483259875e-08,
      "loss": 0.3332,
      "num_input_tokens_seen": 21708808,
      "step": 33270
    },
    {
      "epoch": 19.619693396226417,
      "grad_norm": 4.913275241851807,
      "learning_rate": 1.1044552728687319e-08,
      "loss": 0.3127,
      "num_input_tokens_seen": 21711112,
      "step": 33275
    },
    {
      "epoch": 19.62264150943396,
      "grad_norm": 3.6895456314086914,
      "learning_rate": 1.0874307821570618e-08,
      "loss": 0.3657,
      "num_input_tokens_seen": 21714600,
      "step": 33280
    },
    {
      "epoch": 19.62558962264151,
      "grad_norm": 4.182900428771973,
      "learning_rate": 1.0705383806982606e-08,
      "loss": 0.2756,
      "num_input_tokens_seen": 21717992,
      "step": 33285
    },
    {
      "epoch": 19.628537735849058,
      "grad_norm": 4.212728977203369,
      "learning_rate": 1.0537780729646952e-08,
      "loss": 0.1905,
      "num_input_tokens_seen": 21720392,
      "step": 33290
    },
    {
      "epoch": 19.631485849056602,
      "grad_norm": 4.146754264831543,
      "learning_rate": 1.0371498633937605e-08,
      "loss": 0.357,
      "num_input_tokens_seen": 21723528,
      "step": 33295
    },
    {
      "epoch": 19.63443396226415,
      "grad_norm": 3.1314327716827393,
      "learning_rate": 1.0206537563877683e-08,
      "loss": 0.2156,
      "num_input_tokens_seen": 21726600,
      "step": 33300
    },
    {
      "epoch": 19.6373820754717,
      "grad_norm": 3.277869462966919,
      "learning_rate": 1.0042897563141695e-08,
      "loss": 0.2274,
      "num_input_tokens_seen": 21729608,
      "step": 33305
    },
    {
      "epoch": 19.640330188679247,
      "grad_norm": 2.9126152992248535,
      "learning_rate": 9.88057867505443e-09,
      "loss": 0.3439,
      "num_input_tokens_seen": 21732904,
      "step": 33310
    },
    {
      "epoch": 19.64327830188679,
      "grad_norm": 3.525317668914795,
      "learning_rate": 9.7195809425904e-09,
      "loss": 0.4144,
      "num_input_tokens_seen": 21735496,
      "step": 33315
    },
    {
      "epoch": 19.64622641509434,
      "grad_norm": 5.068065643310547,
      "learning_rate": 9.559904408373844e-09,
      "loss": 0.2932,
      "num_input_tokens_seen": 21738024,
      "step": 33320
    },
    {
      "epoch": 19.649174528301888,
      "grad_norm": 2.2403738498687744,
      "learning_rate": 9.401549114680387e-09,
      "loss": 0.2104,
      "num_input_tokens_seen": 21742280,
      "step": 33325
    },
    {
      "epoch": 19.652122641509433,
      "grad_norm": 4.431077480316162,
      "learning_rate": 9.244515103434826e-09,
      "loss": 0.2711,
      "num_input_tokens_seen": 21745416,
      "step": 33330
    },
    {
      "epoch": 19.65507075471698,
      "grad_norm": 4.923156261444092,
      "learning_rate": 9.08880241621335e-09,
      "loss": 0.2808,
      "num_input_tokens_seen": 21750344,
      "step": 33335
    },
    {
      "epoch": 19.65801886792453,
      "grad_norm": 3.161891222000122,
      "learning_rate": 8.934411094240758e-09,
      "loss": 0.3335,
      "num_input_tokens_seen": 21754152,
      "step": 33340
    },
    {
      "epoch": 19.660966981132077,
      "grad_norm": 3.94423770904541,
      "learning_rate": 8.781341178393244e-09,
      "loss": 0.3372,
      "num_input_tokens_seen": 21757800,
      "step": 33345
    },
    {
      "epoch": 19.66391509433962,
      "grad_norm": 3.609478235244751,
      "learning_rate": 8.629592709196167e-09,
      "loss": 0.3472,
      "num_input_tokens_seen": 21760808,
      "step": 33350
    },
    {
      "epoch": 19.66686320754717,
      "grad_norm": 3.5832736492156982,
      "learning_rate": 8.479165726826277e-09,
      "loss": 0.2646,
      "num_input_tokens_seen": 21763848,
      "step": 33355
    },
    {
      "epoch": 19.669811320754718,
      "grad_norm": 3.6949477195739746,
      "learning_rate": 8.330060271109496e-09,
      "loss": 0.3371,
      "num_input_tokens_seen": 21766856,
      "step": 33360
    },
    {
      "epoch": 19.672759433962263,
      "grad_norm": 2.791074514389038,
      "learning_rate": 8.18227638152258e-09,
      "loss": 0.2962,
      "num_input_tokens_seen": 21770088,
      "step": 33365
    },
    {
      "epoch": 19.67570754716981,
      "grad_norm": 4.624428749084473,
      "learning_rate": 8.035814097191452e-09,
      "loss": 0.3431,
      "num_input_tokens_seen": 21773128,
      "step": 33370
    },
    {
      "epoch": 19.67865566037736,
      "grad_norm": 5.59794282913208,
      "learning_rate": 7.890673456892317e-09,
      "loss": 0.294,
      "num_input_tokens_seen": 21776296,
      "step": 33375
    },
    {
      "epoch": 19.681603773584907,
      "grad_norm": 4.726974010467529,
      "learning_rate": 7.746854499052215e-09,
      "loss": 0.2832,
      "num_input_tokens_seen": 21779176,
      "step": 33380
    },
    {
      "epoch": 19.684551886792452,
      "grad_norm": 4.822123050689697,
      "learning_rate": 7.604357261747907e-09,
      "loss": 0.4126,
      "num_input_tokens_seen": 21782632,
      "step": 33385
    },
    {
      "epoch": 19.6875,
      "grad_norm": 3.692781686782837,
      "learning_rate": 7.463181782705886e-09,
      "loss": 0.3576,
      "num_input_tokens_seen": 21785512,
      "step": 33390
    },
    {
      "epoch": 19.690448113207548,
      "grad_norm": 2.5595858097076416,
      "learning_rate": 7.3233280993034726e-09,
      "loss": 0.3427,
      "num_input_tokens_seen": 21789128,
      "step": 33395
    },
    {
      "epoch": 19.693396226415093,
      "grad_norm": 3.8874311447143555,
      "learning_rate": 7.184796248567161e-09,
      "loss": 0.3668,
      "num_input_tokens_seen": 21792168,
      "step": 33400
    },
    {
      "epoch": 19.69634433962264,
      "grad_norm": 4.511477470397949,
      "learning_rate": 7.047586267173723e-09,
      "loss": 0.2823,
      "num_input_tokens_seen": 21794888,
      "step": 33405
    },
    {
      "epoch": 19.69929245283019,
      "grad_norm": 3.34980845451355,
      "learning_rate": 6.9116981914502114e-09,
      "loss": 0.301,
      "num_input_tokens_seen": 21798152,
      "step": 33410
    },
    {
      "epoch": 19.702240566037737,
      "grad_norm": 12.337843894958496,
      "learning_rate": 6.7771320573734036e-09,
      "loss": 0.4092,
      "num_input_tokens_seen": 21801000,
      "step": 33415
    },
    {
      "epoch": 19.705188679245282,
      "grad_norm": 8.585694313049316,
      "learning_rate": 6.6438879005709114e-09,
      "loss": 0.2337,
      "num_input_tokens_seen": 21804392,
      "step": 33420
    },
    {
      "epoch": 19.70813679245283,
      "grad_norm": 6.219948768615723,
      "learning_rate": 6.511965756318961e-09,
      "loss": 0.398,
      "num_input_tokens_seen": 21807304,
      "step": 33425
    },
    {
      "epoch": 19.71108490566038,
      "grad_norm": 2.3407206535339355,
      "learning_rate": 6.381365659545169e-09,
      "loss": 0.2455,
      "num_input_tokens_seen": 21812008,
      "step": 33430
    },
    {
      "epoch": 19.714033018867923,
      "grad_norm": 5.603320598602295,
      "learning_rate": 6.252087644825766e-09,
      "loss": 0.2908,
      "num_input_tokens_seen": 21814696,
      "step": 33435
    },
    {
      "epoch": 19.71698113207547,
      "grad_norm": 4.824310302734375,
      "learning_rate": 6.124131746388373e-09,
      "loss": 0.2589,
      "num_input_tokens_seen": 21818344,
      "step": 33440
    },
    {
      "epoch": 19.71992924528302,
      "grad_norm": 3.7015762329101562,
      "learning_rate": 5.997497998109225e-09,
      "loss": 0.2833,
      "num_input_tokens_seen": 21821480,
      "step": 33445
    },
    {
      "epoch": 19.722877358490567,
      "grad_norm": 4.04315710067749,
      "learning_rate": 5.8721864335153925e-09,
      "loss": 0.2268,
      "num_input_tokens_seen": 21827240,
      "step": 33450
    },
    {
      "epoch": 19.725825471698112,
      "grad_norm": 3.7808432579040527,
      "learning_rate": 5.748197085784224e-09,
      "loss": 0.1984,
      "num_input_tokens_seen": 21830440,
      "step": 33455
    },
    {
      "epoch": 19.72877358490566,
      "grad_norm": 2.934755325317383,
      "learning_rate": 5.62552998774113e-09,
      "loss": 0.3141,
      "num_input_tokens_seen": 21833320,
      "step": 33460
    },
    {
      "epoch": 19.73172169811321,
      "grad_norm": 3.226236343383789,
      "learning_rate": 5.504185171864018e-09,
      "loss": 0.337,
      "num_input_tokens_seen": 21837352,
      "step": 33465
    },
    {
      "epoch": 19.734669811320753,
      "grad_norm": 5.003249168395996,
      "learning_rate": 5.384162670278858e-09,
      "loss": 0.3542,
      "num_input_tokens_seen": 21840104,
      "step": 33470
    },
    {
      "epoch": 19.7376179245283,
      "grad_norm": 3.75529146194458,
      "learning_rate": 5.265462514762454e-09,
      "loss": 0.279,
      "num_input_tokens_seen": 21842856,
      "step": 33475
    },
    {
      "epoch": 19.74056603773585,
      "grad_norm": 4.079929828643799,
      "learning_rate": 5.148084736740777e-09,
      "loss": 0.3237,
      "num_input_tokens_seen": 21845992,
      "step": 33480
    },
    {
      "epoch": 19.743514150943398,
      "grad_norm": 2.646021604537964,
      "learning_rate": 5.032029367290081e-09,
      "loss": 0.311,
      "num_input_tokens_seen": 21849576,
      "step": 33485
    },
    {
      "epoch": 19.746462264150942,
      "grad_norm": 2.7977280616760254,
      "learning_rate": 4.917296437136898e-09,
      "loss": 0.2966,
      "num_input_tokens_seen": 21852808,
      "step": 33490
    },
    {
      "epoch": 19.74941037735849,
      "grad_norm": 3.1355819702148438,
      "learning_rate": 4.8038859766569305e-09,
      "loss": 0.3706,
      "num_input_tokens_seen": 21855720,
      "step": 33495
    },
    {
      "epoch": 19.75235849056604,
      "grad_norm": 4.428485870361328,
      "learning_rate": 4.691798015876714e-09,
      "loss": 0.3162,
      "num_input_tokens_seen": 21858248,
      "step": 33500
    },
    {
      "epoch": 19.755306603773583,
      "grad_norm": 2.8598434925079346,
      "learning_rate": 4.581032584470846e-09,
      "loss": 0.3158,
      "num_input_tokens_seen": 21861928,
      "step": 33505
    },
    {
      "epoch": 19.75825471698113,
      "grad_norm": 0.6737237572669983,
      "learning_rate": 4.471589711766422e-09,
      "loss": 0.2533,
      "num_input_tokens_seen": 21868200,
      "step": 33510
    },
    {
      "epoch": 19.76120283018868,
      "grad_norm": 6.102258205413818,
      "learning_rate": 4.363469426737487e-09,
      "loss": 0.31,
      "num_input_tokens_seen": 21871336,
      "step": 33515
    },
    {
      "epoch": 19.764150943396228,
      "grad_norm": 3.0342588424682617,
      "learning_rate": 4.256671758010588e-09,
      "loss": 0.2765,
      "num_input_tokens_seen": 21874088,
      "step": 33520
    },
    {
      "epoch": 19.767099056603772,
      "grad_norm": 3.785327434539795,
      "learning_rate": 4.151196733859775e-09,
      "loss": 0.2339,
      "num_input_tokens_seen": 21876936,
      "step": 33525
    },
    {
      "epoch": 19.77004716981132,
      "grad_norm": 2.918762445449829,
      "learning_rate": 4.047044382211041e-09,
      "loss": 0.2932,
      "num_input_tokens_seen": 21880104,
      "step": 33530
    },
    {
      "epoch": 19.77299528301887,
      "grad_norm": 1.7151894569396973,
      "learning_rate": 3.94421473063844e-09,
      "loss": 0.3409,
      "num_input_tokens_seen": 21883528,
      "step": 33535
    },
    {
      "epoch": 19.775943396226417,
      "grad_norm": 3.7225520610809326,
      "learning_rate": 3.842707806366863e-09,
      "loss": 0.5135,
      "num_input_tokens_seen": 21887752,
      "step": 33540
    },
    {
      "epoch": 19.77889150943396,
      "grad_norm": 10.884625434875488,
      "learning_rate": 3.742523636270368e-09,
      "loss": 0.4784,
      "num_input_tokens_seen": 21891240,
      "step": 33545
    },
    {
      "epoch": 19.78183962264151,
      "grad_norm": 6.8832783699035645,
      "learning_rate": 3.6436622468738468e-09,
      "loss": 0.3121,
      "num_input_tokens_seen": 21893864,
      "step": 33550
    },
    {
      "epoch": 19.784787735849058,
      "grad_norm": 3.7381439208984375,
      "learning_rate": 3.546123664350254e-09,
      "loss": 0.3719,
      "num_input_tokens_seen": 21896904,
      "step": 33555
    },
    {
      "epoch": 19.787735849056602,
      "grad_norm": 3.070403575897217,
      "learning_rate": 3.449907914524486e-09,
      "loss": 0.347,
      "num_input_tokens_seen": 21899400,
      "step": 33560
    },
    {
      "epoch": 19.79068396226415,
      "grad_norm": 3.017155408859253,
      "learning_rate": 3.355015022869501e-09,
      "loss": 0.2557,
      "num_input_tokens_seen": 21902408,
      "step": 33565
    },
    {
      "epoch": 19.7936320754717,
      "grad_norm": 3.618273973464966,
      "learning_rate": 3.261445014508535e-09,
      "loss": 0.3006,
      "num_input_tokens_seen": 21905992,
      "step": 33570
    },
    {
      "epoch": 19.796580188679247,
      "grad_norm": 3.5126900672912598,
      "learning_rate": 3.1691979142145503e-09,
      "loss": 0.3445,
      "num_input_tokens_seen": 21910920,
      "step": 33575
    },
    {
      "epoch": 19.79952830188679,
      "grad_norm": 2.4283151626586914,
      "learning_rate": 3.078273746410787e-09,
      "loss": 0.2323,
      "num_input_tokens_seen": 21914984,
      "step": 33580
    },
    {
      "epoch": 19.80247641509434,
      "grad_norm": 3.705442190170288,
      "learning_rate": 2.988672535169657e-09,
      "loss": 0.3341,
      "num_input_tokens_seen": 21917800,
      "step": 33585
    },
    {
      "epoch": 19.805424528301888,
      "grad_norm": 1.9761531352996826,
      "learning_rate": 2.9003943042127393e-09,
      "loss": 0.2924,
      "num_input_tokens_seen": 21920680,
      "step": 33590
    },
    {
      "epoch": 19.808372641509433,
      "grad_norm": 3.415003776550293,
      "learning_rate": 2.8134390769135598e-09,
      "loss": 0.3244,
      "num_input_tokens_seen": 21923880,
      "step": 33595
    },
    {
      "epoch": 19.81132075471698,
      "grad_norm": 5.928948879241943,
      "learning_rate": 2.7278068762925935e-09,
      "loss": 0.2789,
      "num_input_tokens_seen": 21926632,
      "step": 33600
    },
    {
      "epoch": 19.81426886792453,
      "grad_norm": 3.166869640350342,
      "learning_rate": 2.6434977250217043e-09,
      "loss": 0.3285,
      "num_input_tokens_seen": 21929544,
      "step": 33605
    },
    {
      "epoch": 19.817216981132077,
      "grad_norm": 4.835379600524902,
      "learning_rate": 2.560511645422481e-09,
      "loss": 0.2893,
      "num_input_tokens_seen": 21932424,
      "step": 33610
    },
    {
      "epoch": 19.82016509433962,
      "grad_norm": 3.599423408508301,
      "learning_rate": 2.4788486594656825e-09,
      "loss": 0.3389,
      "num_input_tokens_seen": 21935848,
      "step": 33615
    },
    {
      "epoch": 19.82311320754717,
      "grad_norm": 2.914738893508911,
      "learning_rate": 2.398508788771792e-09,
      "loss": 0.4151,
      "num_input_tokens_seen": 21939240,
      "step": 33620
    },
    {
      "epoch": 19.826061320754718,
      "grad_norm": 3.976651668548584,
      "learning_rate": 2.3194920546110166e-09,
      "loss": 0.2993,
      "num_input_tokens_seen": 21942312,
      "step": 33625
    },
    {
      "epoch": 19.829009433962263,
      "grad_norm": 7.003410816192627,
      "learning_rate": 2.2417984779032896e-09,
      "loss": 0.3233,
      "num_input_tokens_seen": 21945800,
      "step": 33630
    },
    {
      "epoch": 19.83195754716981,
      "grad_norm": 2.878335952758789,
      "learning_rate": 2.1654280792193782e-09,
      "loss": 0.3665,
      "num_input_tokens_seen": 21949544,
      "step": 33635
    },
    {
      "epoch": 19.83490566037736,
      "grad_norm": 4.359111309051514,
      "learning_rate": 2.0903808787769987e-09,
      "loss": 0.3524,
      "num_input_tokens_seen": 21952808,
      "step": 33640
    },
    {
      "epoch": 19.837853773584907,
      "grad_norm": 3.0404725074768066,
      "learning_rate": 2.0166568964463673e-09,
      "loss": 0.2926,
      "num_input_tokens_seen": 21960040,
      "step": 33645
    },
    {
      "epoch": 19.840801886792452,
      "grad_norm": 3.753020763397217,
      "learning_rate": 1.9442561517463153e-09,
      "loss": 0.2284,
      "num_input_tokens_seen": 21963784,
      "step": 33650
    },
    {
      "epoch": 19.84375,
      "grad_norm": 3.1365296840667725,
      "learning_rate": 1.8731786638442886e-09,
      "loss": 0.335,
      "num_input_tokens_seen": 21967144,
      "step": 33655
    },
    {
      "epoch": 19.846698113207548,
      "grad_norm": 3.0995407104492188,
      "learning_rate": 1.8034244515591214e-09,
      "loss": 0.3539,
      "num_input_tokens_seen": 21970248,
      "step": 33660
    },
    {
      "epoch": 19.849646226415093,
      "grad_norm": 4.749851226806641,
      "learning_rate": 1.7349935333582646e-09,
      "loss": 0.2825,
      "num_input_tokens_seen": 21973448,
      "step": 33665
    },
    {
      "epoch": 19.85259433962264,
      "grad_norm": 5.330183982849121,
      "learning_rate": 1.6678859273594471e-09,
      "loss": 0.2435,
      "num_input_tokens_seen": 21976360,
      "step": 33670
    },
    {
      "epoch": 19.85554245283019,
      "grad_norm": 4.068929195404053,
      "learning_rate": 1.6021016513295683e-09,
      "loss": 0.301,
      "num_input_tokens_seen": 21979208,
      "step": 33675
    },
    {
      "epoch": 19.858490566037737,
      "grad_norm": 3.971526861190796,
      "learning_rate": 1.5376407226846968e-09,
      "loss": 0.3385,
      "num_input_tokens_seen": 21981672,
      "step": 33680
    },
    {
      "epoch": 19.861438679245282,
      "grad_norm": 3.8497467041015625,
      "learning_rate": 1.4745031584917357e-09,
      "loss": 0.2773,
      "num_input_tokens_seen": 21983944,
      "step": 33685
    },
    {
      "epoch": 19.86438679245283,
      "grad_norm": 4.032742500305176,
      "learning_rate": 1.4126889754667583e-09,
      "loss": 0.2815,
      "num_input_tokens_seen": 21987208,
      "step": 33690
    },
    {
      "epoch": 19.86733490566038,
      "grad_norm": 4.820897579193115,
      "learning_rate": 1.3521981899750069e-09,
      "loss": 0.3957,
      "num_input_tokens_seen": 21991912,
      "step": 33695
    },
    {
      "epoch": 19.870283018867923,
      "grad_norm": 2.5630223751068115,
      "learning_rate": 1.293030818032004e-09,
      "loss": 0.3726,
      "num_input_tokens_seen": 21995976,
      "step": 33700
    },
    {
      "epoch": 19.87323113207547,
      "grad_norm": 3.7156996726989746,
      "learning_rate": 1.2351868753018858e-09,
      "loss": 0.3954,
      "num_input_tokens_seen": 21999144,
      "step": 33705
    },
    {
      "epoch": 19.87617924528302,
      "grad_norm": 2.6448280811309814,
      "learning_rate": 1.1786663770996242e-09,
      "loss": 0.189,
      "num_input_tokens_seen": 22002216,
      "step": 33710
    },
    {
      "epoch": 19.879127358490567,
      "grad_norm": 2.757309675216675,
      "learning_rate": 1.1234693383893602e-09,
      "loss": 0.3457,
      "num_input_tokens_seen": 22007016,
      "step": 33715
    },
    {
      "epoch": 19.882075471698112,
      "grad_norm": 4.485562801361084,
      "learning_rate": 1.0695957737844043e-09,
      "loss": 0.3566,
      "num_input_tokens_seen": 22010440,
      "step": 33720
    },
    {
      "epoch": 19.88502358490566,
      "grad_norm": 3.5489165782928467,
      "learning_rate": 1.0170456975483467e-09,
      "loss": 0.2645,
      "num_input_tokens_seen": 22014088,
      "step": 33725
    },
    {
      "epoch": 19.88797169811321,
      "grad_norm": 3.215893030166626,
      "learning_rate": 9.658191235933922e-10,
      "loss": 0.3112,
      "num_input_tokens_seen": 22017352,
      "step": 33730
    },
    {
      "epoch": 19.890919811320753,
      "grad_norm": 2.998401641845703,
      "learning_rate": 9.159160654825805e-10,
      "loss": 0.295,
      "num_input_tokens_seen": 22021192,
      "step": 33735
    },
    {
      "epoch": 19.8938679245283,
      "grad_norm": 7.281124114990234,
      "learning_rate": 8.673365364281205e-10,
      "loss": 0.2125,
      "num_input_tokens_seen": 22023784,
      "step": 33740
    },
    {
      "epoch": 19.89681603773585,
      "grad_norm": 2.0302274227142334,
      "learning_rate": 8.200805492913911e-10,
      "loss": 0.288,
      "num_input_tokens_seen": 22027176,
      "step": 33745
    },
    {
      "epoch": 19.899764150943398,
      "grad_norm": 5.196775436401367,
      "learning_rate": 7.741481165834952e-10,
      "loss": 0.3316,
      "num_input_tokens_seen": 22030248,
      "step": 33750
    },
    {
      "epoch": 19.902712264150942,
      "grad_norm": 5.006543159484863,
      "learning_rate": 7.29539250465261e-10,
      "loss": 0.3139,
      "num_input_tokens_seen": 22033384,
      "step": 33755
    },
    {
      "epoch": 19.90566037735849,
      "grad_norm": 3.58197021484375,
      "learning_rate": 6.862539627472409e-10,
      "loss": 0.3287,
      "num_input_tokens_seen": 22037128,
      "step": 33760
    },
    {
      "epoch": 19.90860849056604,
      "grad_norm": 5.2429962158203125,
      "learning_rate": 6.442922648897121e-10,
      "loss": 0.35,
      "num_input_tokens_seen": 22040264,
      "step": 33765
    },
    {
      "epoch": 19.911556603773583,
      "grad_norm": 4.104638576507568,
      "learning_rate": 6.036541680015662e-10,
      "loss": 0.3984,
      "num_input_tokens_seen": 22042472,
      "step": 33770
    },
    {
      "epoch": 19.91450471698113,
      "grad_norm": 2.1150755882263184,
      "learning_rate": 5.643396828419745e-10,
      "loss": 0.3069,
      "num_input_tokens_seen": 22046088,
      "step": 33775
    },
    {
      "epoch": 19.91745283018868,
      "grad_norm": 3.7125000953674316,
      "learning_rate": 5.26348819819833e-10,
      "loss": 0.3025,
      "num_input_tokens_seen": 22049192,
      "step": 33780
    },
    {
      "epoch": 19.920400943396228,
      "grad_norm": 3.397738218307495,
      "learning_rate": 4.896815889937622e-10,
      "loss": 0.2736,
      "num_input_tokens_seen": 22052776,
      "step": 33785
    },
    {
      "epoch": 19.923349056603772,
      "grad_norm": 5.0815277099609375,
      "learning_rate": 4.543380000704423e-10,
      "loss": 0.3318,
      "num_input_tokens_seen": 22055752,
      "step": 33790
    },
    {
      "epoch": 19.92629716981132,
      "grad_norm": 5.774856090545654,
      "learning_rate": 4.203180624084979e-10,
      "loss": 0.2691,
      "num_input_tokens_seen": 22059048,
      "step": 33795
    },
    {
      "epoch": 19.92924528301887,
      "grad_norm": 4.3554606437683105,
      "learning_rate": 3.876217850146136e-10,
      "loss": 0.3122,
      "num_input_tokens_seen": 22063176,
      "step": 33800
    },
    {
      "epoch": 19.932193396226417,
      "grad_norm": 2.9344284534454346,
      "learning_rate": 3.56249176544643e-10,
      "loss": 0.3053,
      "num_input_tokens_seen": 22066952,
      "step": 33805
    },
    {
      "epoch": 19.93514150943396,
      "grad_norm": 2.6180758476257324,
      "learning_rate": 3.262002453047197e-10,
      "loss": 0.3078,
      "num_input_tokens_seen": 22070312,
      "step": 33810
    },
    {
      "epoch": 19.93808962264151,
      "grad_norm": 4.800815582275391,
      "learning_rate": 2.974749992512571e-10,
      "loss": 0.4239,
      "num_input_tokens_seen": 22073608,
      "step": 33815
    },
    {
      "epoch": 19.941037735849058,
      "grad_norm": 5.486624717712402,
      "learning_rate": 2.700734459881726e-10,
      "loss": 0.2968,
      "num_input_tokens_seen": 22076264,
      "step": 33820
    },
    {
      "epoch": 19.943985849056602,
      "grad_norm": 4.669373512268066,
      "learning_rate": 2.439955927713289e-10,
      "loss": 0.3224,
      "num_input_tokens_seen": 22079624,
      "step": 33825
    },
    {
      "epoch": 19.94693396226415,
      "grad_norm": 3.429659366607666,
      "learning_rate": 2.1924144650409263e-10,
      "loss": 0.3369,
      "num_input_tokens_seen": 22084136,
      "step": 33830
    },
    {
      "epoch": 19.9498820754717,
      "grad_norm": 4.6284871101379395,
      "learning_rate": 1.9581101374066546e-10,
      "loss": 0.3427,
      "num_input_tokens_seen": 22086984,
      "step": 33835
    },
    {
      "epoch": 19.952830188679247,
      "grad_norm": 3.6983838081359863,
      "learning_rate": 1.7370430068441858e-10,
      "loss": 0.3402,
      "num_input_tokens_seen": 22090760,
      "step": 33840
    },
    {
      "epoch": 19.95577830188679,
      "grad_norm": 2.2568986415863037,
      "learning_rate": 1.529213131878926e-10,
      "loss": 0.2809,
      "num_input_tokens_seen": 22093544,
      "step": 33845
    },
    {
      "epoch": 19.95872641509434,
      "grad_norm": 3.906728506088257,
      "learning_rate": 1.3346205675335288e-10,
      "loss": 0.2686,
      "num_input_tokens_seen": 22097032,
      "step": 33850
    },
    {
      "epoch": 19.961674528301888,
      "grad_norm": 3.609347343444824,
      "learning_rate": 1.1532653653334447e-10,
      "loss": 0.2918,
      "num_input_tokens_seen": 22100904,
      "step": 33855
    },
    {
      "epoch": 19.964622641509433,
      "grad_norm": 3.7154812812805176,
      "learning_rate": 9.85147573284717e-11,
      "loss": 0.2894,
      "num_input_tokens_seen": 22104136,
      "step": 33860
    },
    {
      "epoch": 19.96757075471698,
      "grad_norm": 3.174069404602051,
      "learning_rate": 8.302672359072894e-11,
      "loss": 0.3854,
      "num_input_tokens_seen": 22107496,
      "step": 33865
    },
    {
      "epoch": 19.97051886792453,
      "grad_norm": 4.42632532119751,
      "learning_rate": 6.886243941961468e-11,
      "loss": 0.339,
      "num_input_tokens_seen": 22109960,
      "step": 33870
    },
    {
      "epoch": 19.973466981132077,
      "grad_norm": 4.359560966491699,
      "learning_rate": 5.602190856601741e-11,
      "loss": 0.2939,
      "num_input_tokens_seen": 22113384,
      "step": 33875
    },
    {
      "epoch": 19.97641509433962,
      "grad_norm": 3.5203444957733154,
      "learning_rate": 4.450513442888493e-11,
      "loss": 0.2258,
      "num_input_tokens_seen": 22116328,
      "step": 33880
    },
    {
      "epoch": 19.97936320754717,
      "grad_norm": 5.701579570770264,
      "learning_rate": 3.4312120057999886e-11,
      "loss": 0.249,
      "num_input_tokens_seen": 22119368,
      "step": 33885
    },
    {
      "epoch": 19.982311320754718,
      "grad_norm": 4.176498889923096,
      "learning_rate": 2.5442868151204224e-11,
      "loss": 0.2913,
      "num_input_tokens_seen": 22121960,
      "step": 33890
    },
    {
      "epoch": 19.985259433962263,
      "grad_norm": 3.7649621963500977,
      "learning_rate": 1.7897381057729867e-11,
      "loss": 0.4277,
      "num_input_tokens_seen": 22124904,
      "step": 33895
    },
    {
      "epoch": 19.98820754716981,
      "grad_norm": 4.336522579193115,
      "learning_rate": 1.1675660773757813e-11,
      "loss": 0.2248,
      "num_input_tokens_seen": 22127944,
      "step": 33900
    },
    {
      "epoch": 19.99115566037736,
      "grad_norm": 2.063112258911133,
      "learning_rate": 6.777708947969253e-12,
      "loss": 0.2733,
      "num_input_tokens_seen": 22130952,
      "step": 33905
    },
    {
      "epoch": 19.994103773584907,
      "grad_norm": 5.496767520904541,
      "learning_rate": 3.2035268765495674e-12,
      "loss": 0.2144,
      "num_input_tokens_seen": 22133416,
      "step": 33910
    },
    {
      "epoch": 19.997051886792452,
      "grad_norm": 3.0440354347229004,
      "learning_rate": 9.531155059638863e-13,
      "loss": 0.3743,
      "num_input_tokens_seen": 22136744,
      "step": 33915
    },
    {
      "epoch": 20.0,
      "grad_norm": 4.857840538024902,
      "learning_rate": 2.6475431291750342e-14,
      "loss": 0.2493,
      "num_input_tokens_seen": 22139032,
      "step": 33920
    },
    {
      "epoch": 20.0,
      "eval_loss": 0.6312955617904663,
      "eval_runtime": 19.3725,
      "eval_samples_per_second": 87.547,
      "eval_steps_per_second": 21.887,
      "num_input_tokens_seen": 22139032,
      "step": 33920
    },
    {
      "epoch": 20.0,
      "num_input_tokens_seen": 22139032,
      "step": 33920,
      "total_flos": 9.96910604825985e+17,
      "train_loss": 0.4624784426517644,
      "train_runtime": 3944.015,
      "train_samples_per_second": 34.386,
      "train_steps_per_second": 8.6
    }
  ],
  "logging_steps": 5,
  "max_steps": 33920,
  "num_input_tokens_seen": 22139032,
  "num_train_epochs": 20,
  "save_steps": 3392,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.96910604825985e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}