{
  "best_global_step": 10176,
  "best_metric": 0.4917214512825012,
  "best_model_checkpoint": "saves_multiple/prefix-tuning/llama-3-8b-instruct/train_codealpacapy_456_1765199214/checkpoint-10176",
  "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": 54.222721099853516,
      "learning_rate": 1.179245283018868e-08,
      "loss": 7.1049,
      "num_input_tokens_seen": 2752,
      "step": 5
    },
    {
      "epoch": 0.00589622641509434,
      "grad_norm": 52.61722946166992,
      "learning_rate": 2.6533018867924528e-08,
      "loss": 6.8578,
      "num_input_tokens_seen": 5760,
      "step": 10
    },
    {
      "epoch": 0.00884433962264151,
      "grad_norm": 77.26988983154297,
      "learning_rate": 4.127358490566038e-08,
      "loss": 6.5348,
      "num_input_tokens_seen": 9440,
      "step": 15
    },
    {
      "epoch": 0.01179245283018868,
      "grad_norm": 75.32003784179688,
      "learning_rate": 5.601415094339623e-08,
      "loss": 7.0069,
      "num_input_tokens_seen": 12448,
      "step": 20
    },
    {
      "epoch": 0.01474056603773585,
      "grad_norm": 56.44361114501953,
      "learning_rate": 7.075471698113208e-08,
      "loss": 6.7258,
      "num_input_tokens_seen": 15264,
      "step": 25
    },
    {
      "epoch": 0.01768867924528302,
      "grad_norm": 58.83096694946289,
      "learning_rate": 8.549528301886793e-08,
      "loss": 7.1352,
      "num_input_tokens_seen": 18240,
      "step": 30
    },
    {
      "epoch": 0.020636792452830188,
      "grad_norm": 52.42570877075195,
      "learning_rate": 1.0023584905660378e-07,
      "loss": 6.9864,
      "num_input_tokens_seen": 21024,
      "step": 35
    },
    {
      "epoch": 0.02358490566037736,
      "grad_norm": 68.35691833496094,
      "learning_rate": 1.1497641509433962e-07,
      "loss": 6.8129,
      "num_input_tokens_seen": 24320,
      "step": 40
    },
    {
      "epoch": 0.02653301886792453,
      "grad_norm": 69.62481689453125,
      "learning_rate": 1.297169811320755e-07,
      "loss": 6.923,
      "num_input_tokens_seen": 27488,
      "step": 45
    },
    {
      "epoch": 0.0294811320754717,
      "grad_norm": 52.665618896484375,
      "learning_rate": 1.4445754716981135e-07,
      "loss": 6.6817,
      "num_input_tokens_seen": 30560,
      "step": 50
    },
    {
      "epoch": 0.03242924528301887,
      "grad_norm": 60.46994400024414,
      "learning_rate": 1.591981132075472e-07,
      "loss": 6.5194,
      "num_input_tokens_seen": 33312,
      "step": 55
    },
    {
      "epoch": 0.03537735849056604,
      "grad_norm": 58.32202911376953,
      "learning_rate": 1.7393867924528304e-07,
      "loss": 6.531,
      "num_input_tokens_seen": 36320,
      "step": 60
    },
    {
      "epoch": 0.038325471698113206,
      "grad_norm": 55.93479919433594,
      "learning_rate": 1.886792452830189e-07,
      "loss": 6.4735,
      "num_input_tokens_seen": 39360,
      "step": 65
    },
    {
      "epoch": 0.041273584905660375,
      "grad_norm": 52.53193664550781,
      "learning_rate": 2.0341981132075473e-07,
      "loss": 6.2651,
      "num_input_tokens_seen": 43328,
      "step": 70
    },
    {
      "epoch": 0.044221698113207544,
      "grad_norm": 45.39249801635742,
      "learning_rate": 2.1816037735849058e-07,
      "loss": 6.398,
      "num_input_tokens_seen": 46976,
      "step": 75
    },
    {
      "epoch": 0.04716981132075472,
      "grad_norm": 53.14282989501953,
      "learning_rate": 2.3290094339622643e-07,
      "loss": 6.4309,
      "num_input_tokens_seen": 50208,
      "step": 80
    },
    {
      "epoch": 0.05011792452830189,
      "grad_norm": 48.709815979003906,
      "learning_rate": 2.476415094339623e-07,
      "loss": 6.0212,
      "num_input_tokens_seen": 52864,
      "step": 85
    },
    {
      "epoch": 0.05306603773584906,
      "grad_norm": 46.98457336425781,
      "learning_rate": 2.6238207547169815e-07,
      "loss": 6.2668,
      "num_input_tokens_seen": 56640,
      "step": 90
    },
    {
      "epoch": 0.05601415094339623,
      "grad_norm": 51.85078811645508,
      "learning_rate": 2.7712264150943397e-07,
      "loss": 5.7736,
      "num_input_tokens_seen": 59552,
      "step": 95
    },
    {
      "epoch": 0.0589622641509434,
      "grad_norm": 35.0634651184082,
      "learning_rate": 2.9186320754716984e-07,
      "loss": 5.9678,
      "num_input_tokens_seen": 63392,
      "step": 100
    },
    {
      "epoch": 0.061910377358490566,
      "grad_norm": 63.12975311279297,
      "learning_rate": 3.0660377358490567e-07,
      "loss": 6.0436,
      "num_input_tokens_seen": 66528,
      "step": 105
    },
    {
      "epoch": 0.06485849056603774,
      "grad_norm": 51.421791076660156,
      "learning_rate": 3.213443396226416e-07,
      "loss": 5.8762,
      "num_input_tokens_seen": 69632,
      "step": 110
    },
    {
      "epoch": 0.06780660377358491,
      "grad_norm": 68.058349609375,
      "learning_rate": 3.3608490566037736e-07,
      "loss": 5.6497,
      "num_input_tokens_seen": 73760,
      "step": 115
    },
    {
      "epoch": 0.07075471698113207,
      "grad_norm": 57.663169860839844,
      "learning_rate": 3.508254716981133e-07,
      "loss": 5.5807,
      "num_input_tokens_seen": 77088,
      "step": 120
    },
    {
      "epoch": 0.07370283018867925,
      "grad_norm": 80.65989685058594,
      "learning_rate": 3.6556603773584905e-07,
      "loss": 5.6997,
      "num_input_tokens_seen": 81312,
      "step": 125
    },
    {
      "epoch": 0.07665094339622641,
      "grad_norm": 52.963218688964844,
      "learning_rate": 3.80306603773585e-07,
      "loss": 5.7322,
      "num_input_tokens_seen": 84192,
      "step": 130
    },
    {
      "epoch": 0.07959905660377359,
      "grad_norm": 46.38023376464844,
      "learning_rate": 3.9504716981132075e-07,
      "loss": 5.0634,
      "num_input_tokens_seen": 87872,
      "step": 135
    },
    {
      "epoch": 0.08254716981132075,
      "grad_norm": 50.10238265991211,
      "learning_rate": 4.097877358490567e-07,
      "loss": 5.002,
      "num_input_tokens_seen": 90944,
      "step": 140
    },
    {
      "epoch": 0.08549528301886793,
      "grad_norm": 44.52301025390625,
      "learning_rate": 4.2452830188679244e-07,
      "loss": 5.163,
      "num_input_tokens_seen": 93824,
      "step": 145
    },
    {
      "epoch": 0.08844339622641509,
      "grad_norm": 40.37643051147461,
      "learning_rate": 4.3926886792452837e-07,
      "loss": 5.1964,
      "num_input_tokens_seen": 97280,
      "step": 150
    },
    {
      "epoch": 0.09139150943396226,
      "grad_norm": 45.784000396728516,
      "learning_rate": 4.5400943396226414e-07,
      "loss": 4.592,
      "num_input_tokens_seen": 100192,
      "step": 155
    },
    {
      "epoch": 0.09433962264150944,
      "grad_norm": 40.54258346557617,
      "learning_rate": 4.6875000000000006e-07,
      "loss": 4.6367,
      "num_input_tokens_seen": 103360,
      "step": 160
    },
    {
      "epoch": 0.0972877358490566,
      "grad_norm": 49.01432418823242,
      "learning_rate": 4.834905660377359e-07,
      "loss": 4.6398,
      "num_input_tokens_seen": 106176,
      "step": 165
    },
    {
      "epoch": 0.10023584905660378,
      "grad_norm": 35.23571014404297,
      "learning_rate": 4.982311320754717e-07,
      "loss": 4.4901,
      "num_input_tokens_seen": 109216,
      "step": 170
    },
    {
      "epoch": 0.10318396226415094,
      "grad_norm": 36.14580154418945,
      "learning_rate": 5.129716981132076e-07,
      "loss": 4.6312,
      "num_input_tokens_seen": 111936,
      "step": 175
    },
    {
      "epoch": 0.10613207547169812,
      "grad_norm": 42.39937973022461,
      "learning_rate": 5.277122641509435e-07,
      "loss": 4.4435,
      "num_input_tokens_seen": 114496,
      "step": 180
    },
    {
      "epoch": 0.10908018867924528,
      "grad_norm": 32.638999938964844,
      "learning_rate": 5.424528301886793e-07,
      "loss": 4.5376,
      "num_input_tokens_seen": 119008,
      "step": 185
    },
    {
      "epoch": 0.11202830188679246,
      "grad_norm": 36.270511627197266,
      "learning_rate": 5.571933962264151e-07,
      "loss": 4.3266,
      "num_input_tokens_seen": 122144,
      "step": 190
    },
    {
      "epoch": 0.11497641509433962,
      "grad_norm": 28.875019073486328,
      "learning_rate": 5.71933962264151e-07,
      "loss": 4.6784,
      "num_input_tokens_seen": 125568,
      "step": 195
    },
    {
      "epoch": 0.1179245283018868,
      "grad_norm": 42.33070373535156,
      "learning_rate": 5.866745283018868e-07,
      "loss": 3.7943,
      "num_input_tokens_seen": 128480,
      "step": 200
    },
    {
      "epoch": 0.12087264150943396,
      "grad_norm": 30.67900848388672,
      "learning_rate": 6.014150943396227e-07,
      "loss": 3.8278,
      "num_input_tokens_seen": 131200,
      "step": 205
    },
    {
      "epoch": 0.12382075471698113,
      "grad_norm": 27.169971466064453,
      "learning_rate": 6.161556603773585e-07,
      "loss": 4.1919,
      "num_input_tokens_seen": 134496,
      "step": 210
    },
    {
      "epoch": 0.1267688679245283,
      "grad_norm": 37.122535705566406,
      "learning_rate": 6.308962264150945e-07,
      "loss": 4.1177,
      "num_input_tokens_seen": 137152,
      "step": 215
    },
    {
      "epoch": 0.12971698113207547,
      "grad_norm": 33.89931869506836,
      "learning_rate": 6.456367924528302e-07,
      "loss": 3.89,
      "num_input_tokens_seen": 140576,
      "step": 220
    },
    {
      "epoch": 0.13266509433962265,
      "grad_norm": 35.99678039550781,
      "learning_rate": 6.603773584905661e-07,
      "loss": 3.3574,
      "num_input_tokens_seen": 147552,
      "step": 225
    },
    {
      "epoch": 0.13561320754716982,
      "grad_norm": 41.217838287353516,
      "learning_rate": 6.75117924528302e-07,
      "loss": 3.8011,
      "num_input_tokens_seen": 151040,
      "step": 230
    },
    {
      "epoch": 0.13856132075471697,
      "grad_norm": 33.13102722167969,
      "learning_rate": 6.898584905660379e-07,
      "loss": 4.497,
      "num_input_tokens_seen": 153888,
      "step": 235
    },
    {
      "epoch": 0.14150943396226415,
      "grad_norm": 31.191987991333008,
      "learning_rate": 7.045990566037736e-07,
      "loss": 3.7578,
      "num_input_tokens_seen": 157376,
      "step": 240
    },
    {
      "epoch": 0.14445754716981132,
      "grad_norm": 33.80820083618164,
      "learning_rate": 7.193396226415095e-07,
      "loss": 3.4508,
      "num_input_tokens_seen": 160096,
      "step": 245
    },
    {
      "epoch": 0.1474056603773585,
      "grad_norm": 29.554298400878906,
      "learning_rate": 7.340801886792454e-07,
      "loss": 4.2039,
      "num_input_tokens_seen": 163968,
      "step": 250
    },
    {
      "epoch": 0.15035377358490565,
      "grad_norm": 31.678449630737305,
      "learning_rate": 7.488207547169812e-07,
      "loss": 3.9289,
      "num_input_tokens_seen": 166720,
      "step": 255
    },
    {
      "epoch": 0.15330188679245282,
      "grad_norm": 21.113574981689453,
      "learning_rate": 7.63561320754717e-07,
      "loss": 3.5308,
      "num_input_tokens_seen": 170656,
      "step": 260
    },
    {
      "epoch": 0.15625,
      "grad_norm": 25.739238739013672,
      "learning_rate": 7.783018867924529e-07,
      "loss": 3.521,
      "num_input_tokens_seen": 173408,
      "step": 265
    },
    {
      "epoch": 0.15919811320754718,
      "grad_norm": 31.483016967773438,
      "learning_rate": 7.930424528301888e-07,
      "loss": 3.2204,
      "num_input_tokens_seen": 176800,
      "step": 270
    },
    {
      "epoch": 0.16214622641509435,
      "grad_norm": 24.661149978637695,
      "learning_rate": 8.077830188679246e-07,
      "loss": 3.3418,
      "num_input_tokens_seen": 179776,
      "step": 275
    },
    {
      "epoch": 0.1650943396226415,
      "grad_norm": 25.822702407836914,
      "learning_rate": 8.225235849056605e-07,
      "loss": 3.2372,
      "num_input_tokens_seen": 183040,
      "step": 280
    },
    {
      "epoch": 0.16804245283018868,
      "grad_norm": 24.649755477905273,
      "learning_rate": 8.372641509433963e-07,
      "loss": 3.2967,
      "num_input_tokens_seen": 186880,
      "step": 285
    },
    {
      "epoch": 0.17099056603773585,
      "grad_norm": 24.215253829956055,
      "learning_rate": 8.520047169811321e-07,
      "loss": 3.3929,
      "num_input_tokens_seen": 189600,
      "step": 290
    },
    {
      "epoch": 0.17393867924528303,
      "grad_norm": 31.711929321289062,
      "learning_rate": 8.66745283018868e-07,
      "loss": 3.4219,
      "num_input_tokens_seen": 193856,
      "step": 295
    },
    {
      "epoch": 0.17688679245283018,
      "grad_norm": 27.310096740722656,
      "learning_rate": 8.814858490566039e-07,
      "loss": 3.2748,
      "num_input_tokens_seen": 196480,
      "step": 300
    },
    {
      "epoch": 0.17983490566037735,
      "grad_norm": 19.583911895751953,
      "learning_rate": 8.962264150943397e-07,
      "loss": 3.3445,
      "num_input_tokens_seen": 199360,
      "step": 305
    },
    {
      "epoch": 0.18278301886792453,
      "grad_norm": 38.94401168823242,
      "learning_rate": 9.109669811320755e-07,
      "loss": 3.031,
      "num_input_tokens_seen": 202144,
      "step": 310
    },
    {
      "epoch": 0.1857311320754717,
      "grad_norm": 26.457304000854492,
      "learning_rate": 9.257075471698114e-07,
      "loss": 3.2587,
      "num_input_tokens_seen": 205280,
      "step": 315
    },
    {
      "epoch": 0.18867924528301888,
      "grad_norm": 29.977415084838867,
      "learning_rate": 9.404481132075473e-07,
      "loss": 3.123,
      "num_input_tokens_seen": 208128,
      "step": 320
    },
    {
      "epoch": 0.19162735849056603,
      "grad_norm": 17.921791076660156,
      "learning_rate": 9.551886792452833e-07,
      "loss": 3.1753,
      "num_input_tokens_seen": 211232,
      "step": 325
    },
    {
      "epoch": 0.1945754716981132,
      "grad_norm": 32.01469039916992,
      "learning_rate": 9.699292452830188e-07,
      "loss": 3.1799,
      "num_input_tokens_seen": 214560,
      "step": 330
    },
    {
      "epoch": 0.19752358490566038,
      "grad_norm": 19.83517074584961,
      "learning_rate": 9.846698113207548e-07,
      "loss": 2.9134,
      "num_input_tokens_seen": 218144,
      "step": 335
    },
    {
      "epoch": 0.20047169811320756,
      "grad_norm": 28.32101058959961,
      "learning_rate": 9.994103773584906e-07,
      "loss": 2.8138,
      "num_input_tokens_seen": 220608,
      "step": 340
    },
    {
      "epoch": 0.2034198113207547,
      "grad_norm": 21.481264114379883,
      "learning_rate": 1.0141509433962265e-06,
      "loss": 2.9892,
      "num_input_tokens_seen": 223616,
      "step": 345
    },
    {
      "epoch": 0.20636792452830188,
      "grad_norm": 19.38578987121582,
      "learning_rate": 1.0288915094339623e-06,
      "loss": 3.0692,
      "num_input_tokens_seen": 226816,
      "step": 350
    },
    {
      "epoch": 0.20931603773584906,
      "grad_norm": 40.563270568847656,
      "learning_rate": 1.043632075471698e-06,
      "loss": 3.1159,
      "num_input_tokens_seen": 229504,
      "step": 355
    },
    {
      "epoch": 0.21226415094339623,
      "grad_norm": 22.724910736083984,
      "learning_rate": 1.058372641509434e-06,
      "loss": 3.1408,
      "num_input_tokens_seen": 232800,
      "step": 360
    },
    {
      "epoch": 0.21521226415094338,
      "grad_norm": 22.565658569335938,
      "learning_rate": 1.07311320754717e-06,
      "loss": 2.7335,
      "num_input_tokens_seen": 235584,
      "step": 365
    },
    {
      "epoch": 0.21816037735849056,
      "grad_norm": 21.452585220336914,
      "learning_rate": 1.0878537735849056e-06,
      "loss": 2.9652,
      "num_input_tokens_seen": 239616,
      "step": 370
    },
    {
      "epoch": 0.22110849056603774,
      "grad_norm": 26.703571319580078,
      "learning_rate": 1.1025943396226416e-06,
      "loss": 2.7553,
      "num_input_tokens_seen": 242304,
      "step": 375
    },
    {
      "epoch": 0.2240566037735849,
      "grad_norm": 32.64052963256836,
      "learning_rate": 1.1173349056603773e-06,
      "loss": 2.7732,
      "num_input_tokens_seen": 246432,
      "step": 380
    },
    {
      "epoch": 0.2270047169811321,
      "grad_norm": 25.863054275512695,
      "learning_rate": 1.1320754716981133e-06,
      "loss": 3.0968,
      "num_input_tokens_seen": 249440,
      "step": 385
    },
    {
      "epoch": 0.22995283018867924,
      "grad_norm": 25.35382652282715,
      "learning_rate": 1.1468160377358493e-06,
      "loss": 2.5457,
      "num_input_tokens_seen": 254048,
      "step": 390
    },
    {
      "epoch": 0.2329009433962264,
      "grad_norm": 16.98879623413086,
      "learning_rate": 1.1615566037735849e-06,
      "loss": 2.4144,
      "num_input_tokens_seen": 256800,
      "step": 395
    },
    {
      "epoch": 0.2358490566037736,
      "grad_norm": 26.002470016479492,
      "learning_rate": 1.1762971698113208e-06,
      "loss": 2.2977,
      "num_input_tokens_seen": 259296,
      "step": 400
    },
    {
      "epoch": 0.23879716981132076,
      "grad_norm": 17.653837203979492,
      "learning_rate": 1.1910377358490568e-06,
      "loss": 2.5573,
      "num_input_tokens_seen": 262112,
      "step": 405
    },
    {
      "epoch": 0.2417452830188679,
      "grad_norm": 21.615575790405273,
      "learning_rate": 1.2057783018867926e-06,
      "loss": 2.5675,
      "num_input_tokens_seen": 265280,
      "step": 410
    },
    {
      "epoch": 0.2446933962264151,
      "grad_norm": 14.19434642791748,
      "learning_rate": 1.2205188679245284e-06,
      "loss": 2.4753,
      "num_input_tokens_seen": 268288,
      "step": 415
    },
    {
      "epoch": 0.24764150943396226,
      "grad_norm": 25.733150482177734,
      "learning_rate": 1.2352594339622641e-06,
      "loss": 2.7582,
      "num_input_tokens_seen": 274432,
      "step": 420
    },
    {
      "epoch": 0.2505896226415094,
      "grad_norm": 21.16981315612793,
      "learning_rate": 1.25e-06,
      "loss": 2.5861,
      "num_input_tokens_seen": 279136,
      "step": 425
    },
    {
      "epoch": 0.2535377358490566,
      "grad_norm": 29.296741485595703,
      "learning_rate": 1.264740566037736e-06,
      "loss": 3.0714,
      "num_input_tokens_seen": 283008,
      "step": 430
    },
    {
      "epoch": 0.25648584905660377,
      "grad_norm": 30.104623794555664,
      "learning_rate": 1.2794811320754718e-06,
      "loss": 2.439,
      "num_input_tokens_seen": 285728,
      "step": 435
    },
    {
      "epoch": 0.25943396226415094,
      "grad_norm": 17.618318557739258,
      "learning_rate": 1.2942216981132078e-06,
      "loss": 2.312,
      "num_input_tokens_seen": 288352,
      "step": 440
    },
    {
      "epoch": 0.2623820754716981,
      "grad_norm": 16.8856143951416,
      "learning_rate": 1.3089622641509436e-06,
      "loss": 2.4904,
      "num_input_tokens_seen": 291264,
      "step": 445
    },
    {
      "epoch": 0.2653301886792453,
      "grad_norm": 35.25967788696289,
      "learning_rate": 1.3237028301886792e-06,
      "loss": 2.6839,
      "num_input_tokens_seen": 294496,
      "step": 450
    },
    {
      "epoch": 0.26827830188679247,
      "grad_norm": 28.009521484375,
      "learning_rate": 1.3384433962264151e-06,
      "loss": 2.1994,
      "num_input_tokens_seen": 297536,
      "step": 455
    },
    {
      "epoch": 0.27122641509433965,
      "grad_norm": 20.93626594543457,
      "learning_rate": 1.353183962264151e-06,
      "loss": 2.6184,
      "num_input_tokens_seen": 300672,
      "step": 460
    },
    {
      "epoch": 0.27417452830188677,
      "grad_norm": 29.515880584716797,
      "learning_rate": 1.3679245283018869e-06,
      "loss": 2.5791,
      "num_input_tokens_seen": 303424,
      "step": 465
    },
    {
      "epoch": 0.27712264150943394,
      "grad_norm": 14.954526901245117,
      "learning_rate": 1.3826650943396229e-06,
      "loss": 2.1846,
      "num_input_tokens_seen": 307040,
      "step": 470
    },
    {
      "epoch": 0.2800707547169811,
      "grad_norm": 22.146657943725586,
      "learning_rate": 1.3974056603773586e-06,
      "loss": 2.3982,
      "num_input_tokens_seen": 310688,
      "step": 475
    },
    {
      "epoch": 0.2830188679245283,
      "grad_norm": 22.385478973388672,
      "learning_rate": 1.4121462264150946e-06,
      "loss": 2.4223,
      "num_input_tokens_seen": 313440,
      "step": 480
    },
    {
      "epoch": 0.28596698113207547,
      "grad_norm": 31.942729949951172,
      "learning_rate": 1.4268867924528304e-06,
      "loss": 2.2033,
      "num_input_tokens_seen": 316896,
      "step": 485
    },
    {
      "epoch": 0.28891509433962265,
      "grad_norm": 28.40995979309082,
      "learning_rate": 1.4416273584905664e-06,
      "loss": 2.161,
      "num_input_tokens_seen": 319456,
      "step": 490
    },
    {
      "epoch": 0.2918632075471698,
      "grad_norm": 21.830427169799805,
      "learning_rate": 1.456367924528302e-06,
      "loss": 2.0782,
      "num_input_tokens_seen": 322240,
      "step": 495
    },
    {
      "epoch": 0.294811320754717,
      "grad_norm": 19.030302047729492,
      "learning_rate": 1.4711084905660377e-06,
      "loss": 2.2208,
      "num_input_tokens_seen": 325984,
      "step": 500
    },
    {
      "epoch": 0.2977594339622642,
      "grad_norm": 22.816953659057617,
      "learning_rate": 1.4858490566037737e-06,
      "loss": 2.17,
      "num_input_tokens_seen": 329312,
      "step": 505
    },
    {
      "epoch": 0.3007075471698113,
      "grad_norm": 33.5879020690918,
      "learning_rate": 1.5005896226415096e-06,
      "loss": 2.0781,
      "num_input_tokens_seen": 332352,
      "step": 510
    },
    {
      "epoch": 0.30365566037735847,
      "grad_norm": 26.689653396606445,
      "learning_rate": 1.5153301886792454e-06,
      "loss": 2.642,
      "num_input_tokens_seen": 335008,
      "step": 515
    },
    {
      "epoch": 0.30660377358490565,
      "grad_norm": 34.094017028808594,
      "learning_rate": 1.5300707547169814e-06,
      "loss": 2.1205,
      "num_input_tokens_seen": 337376,
      "step": 520
    },
    {
      "epoch": 0.3095518867924528,
      "grad_norm": 16.514741897583008,
      "learning_rate": 1.5448113207547172e-06,
      "loss": 1.9965,
      "num_input_tokens_seen": 339776,
      "step": 525
    },
    {
      "epoch": 0.3125,
      "grad_norm": 18.866437911987305,
      "learning_rate": 1.5595518867924531e-06,
      "loss": 2.1677,
      "num_input_tokens_seen": 342336,
      "step": 530
    },
    {
      "epoch": 0.3154481132075472,
      "grad_norm": 20.757240295410156,
      "learning_rate": 1.574292452830189e-06,
      "loss": 2.0589,
      "num_input_tokens_seen": 345440,
      "step": 535
    },
    {
      "epoch": 0.31839622641509435,
      "grad_norm": 27.814619064331055,
      "learning_rate": 1.5890330188679245e-06,
      "loss": 2.307,
      "num_input_tokens_seen": 348608,
      "step": 540
    },
    {
      "epoch": 0.32134433962264153,
      "grad_norm": 16.26691436767578,
      "learning_rate": 1.6037735849056604e-06,
      "loss": 1.9491,
      "num_input_tokens_seen": 352992,
      "step": 545
    },
    {
      "epoch": 0.3242924528301887,
      "grad_norm": 18.54221534729004,
      "learning_rate": 1.6185141509433964e-06,
      "loss": 2.2303,
      "num_input_tokens_seen": 356768,
      "step": 550
    },
    {
      "epoch": 0.3272405660377358,
      "grad_norm": 19.25487518310547,
      "learning_rate": 1.6332547169811322e-06,
      "loss": 2.062,
      "num_input_tokens_seen": 359360,
      "step": 555
    },
    {
      "epoch": 0.330188679245283,
      "grad_norm": 16.693958282470703,
      "learning_rate": 1.6479952830188682e-06,
      "loss": 2.131,
      "num_input_tokens_seen": 363488,
      "step": 560
    },
    {
      "epoch": 0.3331367924528302,
      "grad_norm": 17.222332000732422,
      "learning_rate": 1.662735849056604e-06,
      "loss": 2.2984,
      "num_input_tokens_seen": 366688,
      "step": 565
    },
    {
      "epoch": 0.33608490566037735,
      "grad_norm": 17.67404556274414,
      "learning_rate": 1.67747641509434e-06,
      "loss": 1.9614,
      "num_input_tokens_seen": 369152,
      "step": 570
    },
    {
      "epoch": 0.33903301886792453,
      "grad_norm": 19.30936622619629,
      "learning_rate": 1.6922169811320757e-06,
      "loss": 1.8213,
      "num_input_tokens_seen": 372992,
      "step": 575
    },
    {
      "epoch": 0.3419811320754717,
      "grad_norm": 28.540185928344727,
      "learning_rate": 1.7069575471698112e-06,
      "loss": 2.168,
      "num_input_tokens_seen": 375680,
      "step": 580
    },
    {
      "epoch": 0.3449292452830189,
      "grad_norm": 19.80051612854004,
      "learning_rate": 1.7216981132075472e-06,
      "loss": 1.8592,
      "num_input_tokens_seen": 379808,
      "step": 585
    },
    {
      "epoch": 0.34787735849056606,
      "grad_norm": 15.090983390808105,
      "learning_rate": 1.736438679245283e-06,
      "loss": 2.003,
      "num_input_tokens_seen": 382752,
      "step": 590
    },
    {
      "epoch": 0.35082547169811323,
      "grad_norm": 17.44107437133789,
      "learning_rate": 1.751179245283019e-06,
      "loss": 1.9762,
      "num_input_tokens_seen": 386304,
      "step": 595
    },
    {
      "epoch": 0.35377358490566035,
      "grad_norm": 17.52680015563965,
      "learning_rate": 1.765919811320755e-06,
      "loss": 2.0236,
      "num_input_tokens_seen": 390592,
      "step": 600
    },
    {
      "epoch": 0.35672169811320753,
      "grad_norm": 19.416120529174805,
      "learning_rate": 1.7806603773584907e-06,
      "loss": 2.2317,
      "num_input_tokens_seen": 394304,
      "step": 605
    },
    {
      "epoch": 0.3596698113207547,
      "grad_norm": 15.273588180541992,
      "learning_rate": 1.7954009433962267e-06,
      "loss": 1.8942,
      "num_input_tokens_seen": 397664,
      "step": 610
    },
    {
      "epoch": 0.3626179245283019,
      "grad_norm": 16.087310791015625,
      "learning_rate": 1.8101415094339625e-06,
      "loss": 1.9195,
      "num_input_tokens_seen": 401632,
      "step": 615
    },
    {
      "epoch": 0.36556603773584906,
      "grad_norm": 22.126197814941406,
      "learning_rate": 1.8248820754716984e-06,
      "loss": 1.5036,
      "num_input_tokens_seen": 403904,
      "step": 620
    },
    {
      "epoch": 0.36851415094339623,
      "grad_norm": 13.83784294128418,
      "learning_rate": 1.839622641509434e-06,
      "loss": 1.6058,
      "num_input_tokens_seen": 406816,
      "step": 625
    },
    {
      "epoch": 0.3714622641509434,
      "grad_norm": 16.79551887512207,
      "learning_rate": 1.8543632075471698e-06,
      "loss": 1.8395,
      "num_input_tokens_seen": 411360,
      "step": 630
    },
    {
      "epoch": 0.3744103773584906,
      "grad_norm": 19.93268585205078,
      "learning_rate": 1.8691037735849057e-06,
      "loss": 1.9946,
      "num_input_tokens_seen": 415552,
      "step": 635
    },
    {
      "epoch": 0.37735849056603776,
      "grad_norm": 22.72307777404785,
      "learning_rate": 1.8838443396226417e-06,
      "loss": 1.7062,
      "num_input_tokens_seen": 420736,
      "step": 640
    },
    {
      "epoch": 0.3803066037735849,
      "grad_norm": 15.50692081451416,
      "learning_rate": 1.8985849056603775e-06,
      "loss": 1.7871,
      "num_input_tokens_seen": 423744,
      "step": 645
    },
    {
      "epoch": 0.38325471698113206,
      "grad_norm": 18.6272029876709,
      "learning_rate": 1.9133254716981133e-06,
      "loss": 2.0758,
      "num_input_tokens_seen": 427776,
      "step": 650
    },
    {
      "epoch": 0.38620283018867924,
      "grad_norm": 27.329578399658203,
      "learning_rate": 1.9280660377358494e-06,
      "loss": 1.6892,
      "num_input_tokens_seen": 431040,
      "step": 655
    },
    {
      "epoch": 0.3891509433962264,
      "grad_norm": 20.49376106262207,
      "learning_rate": 1.9428066037735852e-06,
      "loss": 1.4992,
      "num_input_tokens_seen": 434144,
      "step": 660
    },
    {
      "epoch": 0.3920990566037736,
      "grad_norm": 15.999664306640625,
      "learning_rate": 1.957547169811321e-06,
      "loss": 1.7418,
      "num_input_tokens_seen": 437600,
      "step": 665
    },
    {
      "epoch": 0.39504716981132076,
      "grad_norm": 14.668155670166016,
      "learning_rate": 1.9722877358490568e-06,
      "loss": 1.965,
      "num_input_tokens_seen": 441184,
      "step": 670
    },
    {
      "epoch": 0.39799528301886794,
      "grad_norm": 14.06435775756836,
      "learning_rate": 1.9870283018867925e-06,
      "loss": 1.6411,
      "num_input_tokens_seen": 444896,
      "step": 675
    },
    {
      "epoch": 0.4009433962264151,
      "grad_norm": 16.098995208740234,
      "learning_rate": 2.0017688679245283e-06,
      "loss": 1.6181,
      "num_input_tokens_seen": 447936,
      "step": 680
    },
    {
      "epoch": 0.40389150943396224,
      "grad_norm": 24.093454360961914,
      "learning_rate": 2.0165094339622645e-06,
      "loss": 1.6427,
      "num_input_tokens_seen": 451488,
      "step": 685
    },
    {
      "epoch": 0.4068396226415094,
      "grad_norm": 20.385398864746094,
      "learning_rate": 2.0312500000000002e-06,
      "loss": 1.4471,
      "num_input_tokens_seen": 454240,
      "step": 690
    },
    {
      "epoch": 0.4097877358490566,
      "grad_norm": 11.932026863098145,
      "learning_rate": 2.045990566037736e-06,
      "loss": 1.5753,
      "num_input_tokens_seen": 458240,
      "step": 695
    },
    {
      "epoch": 0.41273584905660377,
      "grad_norm": 15.874615669250488,
      "learning_rate": 2.0607311320754718e-06,
      "loss": 1.4879,
      "num_input_tokens_seen": 461376,
      "step": 700
    },
    {
      "epoch": 0.41568396226415094,
      "grad_norm": 17.140851974487305,
      "learning_rate": 2.075471698113208e-06,
      "loss": 1.2819,
      "num_input_tokens_seen": 464544,
      "step": 705
    },
    {
      "epoch": 0.4186320754716981,
      "grad_norm": 10.872191429138184,
      "learning_rate": 2.0902122641509437e-06,
      "loss": 1.5349,
      "num_input_tokens_seen": 467168,
      "step": 710
    },
    {
      "epoch": 0.4215801886792453,
      "grad_norm": 16.444684982299805,
      "learning_rate": 2.1049528301886795e-06,
      "loss": 1.4353,
      "num_input_tokens_seen": 470048,
      "step": 715
    },
    {
      "epoch": 0.42452830188679247,
      "grad_norm": 13.255218505859375,
      "learning_rate": 2.1196933962264153e-06,
      "loss": 1.4353,
      "num_input_tokens_seen": 473056,
      "step": 720
    },
    {
      "epoch": 0.42747641509433965,
      "grad_norm": 30.873544692993164,
      "learning_rate": 2.134433962264151e-06,
      "loss": 1.6527,
      "num_input_tokens_seen": 476288,
      "step": 725
    },
    {
      "epoch": 0.43042452830188677,
      "grad_norm": 27.139835357666016,
      "learning_rate": 2.149174528301887e-06,
      "loss": 1.2416,
      "num_input_tokens_seen": 479392,
      "step": 730
    },
    {
      "epoch": 0.43337264150943394,
      "grad_norm": 15.826363563537598,
      "learning_rate": 2.163915094339623e-06,
      "loss": 1.4099,
      "num_input_tokens_seen": 482560,
      "step": 735
    },
    {
      "epoch": 0.4363207547169811,
      "grad_norm": 16.070980072021484,
      "learning_rate": 2.1786556603773588e-06,
      "loss": 1.0659,
      "num_input_tokens_seen": 485568,
      "step": 740
    },
    {
      "epoch": 0.4392688679245283,
      "grad_norm": 9.828486442565918,
      "learning_rate": 2.1933962264150945e-06,
      "loss": 1.4024,
      "num_input_tokens_seen": 488032,
      "step": 745
    },
    {
      "epoch": 0.44221698113207547,
      "grad_norm": 16.718196868896484,
      "learning_rate": 2.2081367924528303e-06,
      "loss": 1.2506,
      "num_input_tokens_seen": 490688,
      "step": 750
    },
    {
      "epoch": 0.44516509433962265,
      "grad_norm": 16.405200958251953,
      "learning_rate": 2.2228773584905665e-06,
      "loss": 1.3233,
      "num_input_tokens_seen": 493568,
      "step": 755
    },
    {
      "epoch": 0.4481132075471698,
      "grad_norm": 16.074996948242188,
      "learning_rate": 2.237617924528302e-06,
      "loss": 1.5822,
      "num_input_tokens_seen": 496096,
      "step": 760
    },
    {
      "epoch": 0.451061320754717,
      "grad_norm": 11.769827842712402,
      "learning_rate": 2.252358490566038e-06,
      "loss": 1.3496,
      "num_input_tokens_seen": 499136,
      "step": 765
    },
    {
      "epoch": 0.4540094339622642,
      "grad_norm": 12.198741912841797,
      "learning_rate": 2.267099056603774e-06,
      "loss": 1.1394,
      "num_input_tokens_seen": 502048,
      "step": 770
    },
    {
      "epoch": 0.4569575471698113,
      "grad_norm": 10.2783203125,
      "learning_rate": 2.2818396226415096e-06,
      "loss": 1.0007,
      "num_input_tokens_seen": 505504,
      "step": 775
    },
    {
      "epoch": 0.45990566037735847,
      "grad_norm": 19.44044303894043,
      "learning_rate": 2.2965801886792453e-06,
      "loss": 1.1778,
      "num_input_tokens_seen": 508512,
      "step": 780
    },
    {
      "epoch": 0.46285377358490565,
      "grad_norm": 12.049741744995117,
      "learning_rate": 2.3113207547169815e-06,
      "loss": 0.8564,
      "num_input_tokens_seen": 511808,
      "step": 785
    },
    {
      "epoch": 0.4658018867924528,
      "grad_norm": 18.965778350830078,
      "learning_rate": 2.3260613207547173e-06,
      "loss": 1.0445,
      "num_input_tokens_seen": 514144,
      "step": 790
    },
    {
      "epoch": 0.46875,
      "grad_norm": 5.180530548095703,
      "learning_rate": 2.340801886792453e-06,
      "loss": 0.9787,
      "num_input_tokens_seen": 517120,
      "step": 795
    },
    {
      "epoch": 0.4716981132075472,
      "grad_norm": 14.623882293701172,
      "learning_rate": 2.355542452830189e-06,
      "loss": 0.9612,
      "num_input_tokens_seen": 519904,
      "step": 800
    },
    {
      "epoch": 0.47464622641509435,
      "grad_norm": 9.881180763244629,
      "learning_rate": 2.3702830188679246e-06,
      "loss": 0.9598,
      "num_input_tokens_seen": 523520,
      "step": 805
    },
    {
      "epoch": 0.47759433962264153,
      "grad_norm": 6.5216498374938965,
      "learning_rate": 2.3850235849056604e-06,
      "loss": 0.7343,
      "num_input_tokens_seen": 525952,
      "step": 810
    },
    {
      "epoch": 0.4805424528301887,
      "grad_norm": 13.199479103088379,
      "learning_rate": 2.3997641509433966e-06,
      "loss": 0.9348,
      "num_input_tokens_seen": 529504,
      "step": 815
    },
    {
      "epoch": 0.4834905660377358,
      "grad_norm": 9.821159362792969,
      "learning_rate": 2.4145047169811323e-06,
      "loss": 0.8783,
      "num_input_tokens_seen": 532960,
      "step": 820
    },
    {
      "epoch": 0.486438679245283,
      "grad_norm": 21.9212646484375,
      "learning_rate": 2.429245283018868e-06,
      "loss": 0.8925,
      "num_input_tokens_seen": 535744,
      "step": 825
    },
    {
      "epoch": 0.4893867924528302,
      "grad_norm": 18.933372497558594,
      "learning_rate": 2.443985849056604e-06,
      "loss": 0.8738,
      "num_input_tokens_seen": 539392,
      "step": 830
    },
    {
      "epoch": 0.49233490566037735,
      "grad_norm": 7.524964332580566,
      "learning_rate": 2.45872641509434e-06,
      "loss": 0.6848,
      "num_input_tokens_seen": 542912,
      "step": 835
    },
    {
      "epoch": 0.49528301886792453,
      "grad_norm": 7.029047966003418,
      "learning_rate": 2.473466981132076e-06,
      "loss": 0.7666,
      "num_input_tokens_seen": 547040,
      "step": 840
    },
    {
      "epoch": 0.4982311320754717,
      "grad_norm": 6.068385601043701,
      "learning_rate": 2.4882075471698116e-06,
      "loss": 0.7332,
      "num_input_tokens_seen": 550016,
      "step": 845
    },
    {
      "epoch": 0.5011792452830188,
      "grad_norm": 11.979207038879395,
      "learning_rate": 2.5029481132075474e-06,
      "loss": 0.7395,
      "num_input_tokens_seen": 552768,
      "step": 850
    },
    {
      "epoch": 0.504127358490566,
      "grad_norm": 8.341313362121582,
      "learning_rate": 2.517688679245283e-06,
      "loss": 0.9378,
      "num_input_tokens_seen": 556096,
      "step": 855
    },
    {
      "epoch": 0.5070754716981132,
      "grad_norm": 7.834328651428223,
      "learning_rate": 2.532429245283019e-06,
      "loss": 0.9479,
      "num_input_tokens_seen": 559712,
      "step": 860
    },
    {
      "epoch": 0.5100235849056604,
      "grad_norm": 6.522136688232422,
      "learning_rate": 2.547169811320755e-06,
      "loss": 0.8438,
      "num_input_tokens_seen": 563008,
      "step": 865
    },
    {
      "epoch": 0.5129716981132075,
      "grad_norm": 7.916810512542725,
      "learning_rate": 2.561910377358491e-06,
      "loss": 0.6522,
      "num_input_tokens_seen": 566592,
      "step": 870
    },
    {
      "epoch": 0.5159198113207547,
      "grad_norm": 9.186164855957031,
      "learning_rate": 2.5766509433962266e-06,
      "loss": 0.7146,
      "num_input_tokens_seen": 570144,
      "step": 875
    },
    {
      "epoch": 0.5188679245283019,
      "grad_norm": 5.8527984619140625,
      "learning_rate": 2.5913915094339624e-06,
      "loss": 0.9545,
      "num_input_tokens_seen": 573984,
      "step": 880
    },
    {
      "epoch": 0.5218160377358491,
      "grad_norm": 7.28742790222168,
      "learning_rate": 2.6061320754716986e-06,
      "loss": 0.8072,
      "num_input_tokens_seen": 576608,
      "step": 885
    },
    {
      "epoch": 0.5247641509433962,
      "grad_norm": 10.230030059814453,
      "learning_rate": 2.6208726415094343e-06,
      "loss": 0.7248,
      "num_input_tokens_seen": 579520,
      "step": 890
    },
    {
      "epoch": 0.5277122641509434,
      "grad_norm": 5.330428123474121,
      "learning_rate": 2.63561320754717e-06,
      "loss": 0.6788,
      "num_input_tokens_seen": 582976,
      "step": 895
    },
    {
      "epoch": 0.5306603773584906,
      "grad_norm": 11.809231758117676,
      "learning_rate": 2.650353773584906e-06,
      "loss": 0.7293,
      "num_input_tokens_seen": 586176,
      "step": 900
    },
    {
      "epoch": 0.5336084905660378,
      "grad_norm": 18.289920806884766,
      "learning_rate": 2.665094339622642e-06,
      "loss": 0.5789,
      "num_input_tokens_seen": 589088,
      "step": 905
    },
    {
      "epoch": 0.5365566037735849,
      "grad_norm": 8.130903244018555,
      "learning_rate": 2.679834905660378e-06,
      "loss": 0.6258,
      "num_input_tokens_seen": 592256,
      "step": 910
    },
    {
      "epoch": 0.5395047169811321,
      "grad_norm": 11.105206489562988,
      "learning_rate": 2.694575471698113e-06,
      "loss": 0.6574,
      "num_input_tokens_seen": 598336,
      "step": 915
    },
    {
      "epoch": 0.5424528301886793,
      "grad_norm": 6.774974822998047,
      "learning_rate": 2.709316037735849e-06,
      "loss": 0.7137,
      "num_input_tokens_seen": 602336,
      "step": 920
    },
    {
      "epoch": 0.5454009433962265,
      "grad_norm": 6.451699733734131,
      "learning_rate": 2.724056603773585e-06,
      "loss": 0.7011,
      "num_input_tokens_seen": 605120,
      "step": 925
    },
    {
      "epoch": 0.5483490566037735,
      "grad_norm": 5.600085258483887,
      "learning_rate": 2.738797169811321e-06,
      "loss": 0.8423,
      "num_input_tokens_seen": 608640,
      "step": 930
    },
    {
      "epoch": 0.5512971698113207,
      "grad_norm": 5.928076267242432,
      "learning_rate": 2.7535377358490567e-06,
      "loss": 0.8639,
      "num_input_tokens_seen": 611840,
      "step": 935
    },
    {
      "epoch": 0.5542452830188679,
      "grad_norm": 8.058202743530273,
      "learning_rate": 2.7682783018867925e-06,
      "loss": 0.6704,
      "num_input_tokens_seen": 615488,
      "step": 940
    },
    {
      "epoch": 0.5571933962264151,
      "grad_norm": 4.913827896118164,
      "learning_rate": 2.7830188679245286e-06,
      "loss": 0.7935,
      "num_input_tokens_seen": 618848,
      "step": 945
    },
    {
      "epoch": 0.5601415094339622,
      "grad_norm": 4.591134071350098,
      "learning_rate": 2.7977594339622644e-06,
      "loss": 0.6561,
      "num_input_tokens_seen": 622272,
      "step": 950
    },
    {
      "epoch": 0.5630896226415094,
      "grad_norm": 6.320178031921387,
      "learning_rate": 2.8125e-06,
      "loss": 0.5682,
      "num_input_tokens_seen": 625632,
      "step": 955
    },
    {
      "epoch": 0.5660377358490566,
      "grad_norm": 10.147370338439941,
      "learning_rate": 2.827240566037736e-06,
      "loss": 0.8593,
      "num_input_tokens_seen": 628448,
      "step": 960
    },
    {
      "epoch": 0.5689858490566038,
      "grad_norm": 6.732077121734619,
      "learning_rate": 2.841981132075472e-06,
      "loss": 0.727,
      "num_input_tokens_seen": 632736,
      "step": 965
    },
    {
      "epoch": 0.5719339622641509,
      "grad_norm": 3.9576258659362793,
      "learning_rate": 2.856721698113208e-06,
      "loss": 0.6668,
      "num_input_tokens_seen": 636576,
      "step": 970
    },
    {
      "epoch": 0.5748820754716981,
      "grad_norm": 7.329128742218018,
      "learning_rate": 2.8714622641509437e-06,
      "loss": 0.5873,
      "num_input_tokens_seen": 639840,
      "step": 975
    },
    {
      "epoch": 0.5778301886792453,
      "grad_norm": 8.611266136169434,
      "learning_rate": 2.8862028301886794e-06,
      "loss": 0.8116,
      "num_input_tokens_seen": 643712,
      "step": 980
    },
    {
      "epoch": 0.5807783018867925,
      "grad_norm": 7.204748153686523,
      "learning_rate": 2.9009433962264156e-06,
      "loss": 0.7136,
      "num_input_tokens_seen": 648000,
      "step": 985
    },
    {
      "epoch": 0.5837264150943396,
      "grad_norm": 7.753498077392578,
      "learning_rate": 2.9156839622641514e-06,
      "loss": 0.7514,
      "num_input_tokens_seen": 651488,
      "step": 990
    },
    {
      "epoch": 0.5866745283018868,
      "grad_norm": 6.6436591148376465,
      "learning_rate": 2.930424528301887e-06,
      "loss": 0.6084,
      "num_input_tokens_seen": 654688,
      "step": 995
    },
    {
      "epoch": 0.589622641509434,
      "grad_norm": 5.8500895500183105,
      "learning_rate": 2.9451650943396225e-06,
      "loss": 0.6281,
      "num_input_tokens_seen": 657440,
      "step": 1000
    },
    {
      "epoch": 0.5925707547169812,
      "grad_norm": 6.583737373352051,
      "learning_rate": 2.9599056603773587e-06,
      "loss": 0.7459,
      "num_input_tokens_seen": 660768,
      "step": 1005
    },
    {
      "epoch": 0.5955188679245284,
      "grad_norm": 6.009942531585693,
      "learning_rate": 2.9746462264150945e-06,
      "loss": 0.7491,
      "num_input_tokens_seen": 665536,
      "step": 1010
    },
    {
      "epoch": 0.5984669811320755,
      "grad_norm": 4.740694522857666,
      "learning_rate": 2.9893867924528302e-06,
      "loss": 0.7072,
      "num_input_tokens_seen": 668736,
      "step": 1015
    },
    {
      "epoch": 0.6014150943396226,
      "grad_norm": 7.021014213562012,
      "learning_rate": 3.004127358490566e-06,
      "loss": 0.5703,
      "num_input_tokens_seen": 671584,
      "step": 1020
    },
    {
      "epoch": 0.6043632075471698,
      "grad_norm": 7.415105819702148,
      "learning_rate": 3.018867924528302e-06,
      "loss": 0.5921,
      "num_input_tokens_seen": 674688,
      "step": 1025
    },
    {
      "epoch": 0.6073113207547169,
      "grad_norm": 10.266179084777832,
      "learning_rate": 3.033608490566038e-06,
      "loss": 0.7202,
      "num_input_tokens_seen": 679616,
      "step": 1030
    },
    {
      "epoch": 0.6102594339622641,
      "grad_norm": 6.04750919342041,
      "learning_rate": 3.0483490566037737e-06,
      "loss": 0.6464,
      "num_input_tokens_seen": 684288,
      "step": 1035
    },
    {
      "epoch": 0.6132075471698113,
      "grad_norm": 6.332635402679443,
      "learning_rate": 3.0630896226415095e-06,
      "loss": 0.7889,
      "num_input_tokens_seen": 687776,
      "step": 1040
    },
    {
      "epoch": 0.6161556603773585,
      "grad_norm": 12.109938621520996,
      "learning_rate": 3.0778301886792457e-06,
      "loss": 0.7147,
      "num_input_tokens_seen": 691264,
      "step": 1045
    },
    {
      "epoch": 0.6191037735849056,
      "grad_norm": 5.362137794494629,
      "learning_rate": 3.0925707547169815e-06,
      "loss": 0.5842,
      "num_input_tokens_seen": 694976,
      "step": 1050
    },
    {
      "epoch": 0.6220518867924528,
      "grad_norm": 13.0574951171875,
      "learning_rate": 3.1073113207547172e-06,
      "loss": 0.6791,
      "num_input_tokens_seen": 697472,
      "step": 1055
    },
    {
      "epoch": 0.625,
      "grad_norm": 4.749898433685303,
      "learning_rate": 3.122051886792453e-06,
      "loss": 0.8094,
      "num_input_tokens_seen": 701088,
      "step": 1060
    },
    {
      "epoch": 0.6279481132075472,
      "grad_norm": 91.45693969726562,
      "learning_rate": 3.136792452830189e-06,
      "loss": 0.7867,
      "num_input_tokens_seen": 704640,
      "step": 1065
    },
    {
      "epoch": 0.6308962264150944,
      "grad_norm": 5.956383228302002,
      "learning_rate": 3.151533018867925e-06,
      "loss": 0.5568,
      "num_input_tokens_seen": 707584,
      "step": 1070
    },
    {
      "epoch": 0.6338443396226415,
      "grad_norm": 9.517912864685059,
      "learning_rate": 3.1662735849056607e-06,
      "loss": 0.7206,
      "num_input_tokens_seen": 710560,
      "step": 1075
    },
    {
      "epoch": 0.6367924528301887,
      "grad_norm": 4.406369686126709,
      "learning_rate": 3.181014150943397e-06,
      "loss": 0.672,
      "num_input_tokens_seen": 715424,
      "step": 1080
    },
    {
      "epoch": 0.6397405660377359,
      "grad_norm": 5.649517059326172,
      "learning_rate": 3.1957547169811327e-06,
      "loss": 0.5535,
      "num_input_tokens_seen": 717952,
      "step": 1085
    },
    {
      "epoch": 0.6426886792452831,
      "grad_norm": 7.648313999176025,
      "learning_rate": 3.210495283018868e-06,
      "loss": 0.7721,
      "num_input_tokens_seen": 720608,
      "step": 1090
    },
    {
      "epoch": 0.6456367924528302,
      "grad_norm": 9.276202201843262,
      "learning_rate": 3.225235849056604e-06,
      "loss": 0.4886,
      "num_input_tokens_seen": 723264,
      "step": 1095
    },
    {
      "epoch": 0.6485849056603774,
      "grad_norm": 15.227643013000488,
      "learning_rate": 3.2399764150943396e-06,
      "loss": 0.537,
      "num_input_tokens_seen": 727968,
      "step": 1100
    },
    {
      "epoch": 0.6515330188679245,
      "grad_norm": 5.443690299987793,
      "learning_rate": 3.2547169811320758e-06,
      "loss": 0.6572,
      "num_input_tokens_seen": 731936,
      "step": 1105
    },
    {
      "epoch": 0.6544811320754716,
      "grad_norm": 7.382089138031006,
      "learning_rate": 3.2694575471698115e-06,
      "loss": 0.5512,
      "num_input_tokens_seen": 735200,
      "step": 1110
    },
    {
      "epoch": 0.6574292452830188,
      "grad_norm": 5.733156204223633,
      "learning_rate": 3.2841981132075473e-06,
      "loss": 0.6502,
      "num_input_tokens_seen": 738048,
      "step": 1115
    },
    {
      "epoch": 0.660377358490566,
      "grad_norm": 3.520015001296997,
      "learning_rate": 3.298938679245283e-06,
      "loss": 0.6195,
      "num_input_tokens_seen": 740672,
      "step": 1120
    },
    {
      "epoch": 0.6633254716981132,
      "grad_norm": 5.500694274902344,
      "learning_rate": 3.3136792452830192e-06,
      "loss": 0.7606,
      "num_input_tokens_seen": 744192,
      "step": 1125
    },
    {
      "epoch": 0.6662735849056604,
      "grad_norm": 1.3552123308181763,
      "learning_rate": 3.328419811320755e-06,
      "loss": 0.5875,
      "num_input_tokens_seen": 749184,
      "step": 1130
    },
    {
      "epoch": 0.6692216981132075,
      "grad_norm": 5.837432384490967,
      "learning_rate": 3.3431603773584908e-06,
      "loss": 0.6495,
      "num_input_tokens_seen": 751552,
      "step": 1135
    },
    {
      "epoch": 0.6721698113207547,
      "grad_norm": 4.1729631423950195,
      "learning_rate": 3.3579009433962266e-06,
      "loss": 0.661,
      "num_input_tokens_seen": 754752,
      "step": 1140
    },
    {
      "epoch": 0.6751179245283019,
      "grad_norm": 6.988410472869873,
      "learning_rate": 3.3726415094339627e-06,
      "loss": 0.5478,
      "num_input_tokens_seen": 757824,
      "step": 1145
    },
    {
      "epoch": 0.6780660377358491,
      "grad_norm": 9.738689422607422,
      "learning_rate": 3.3873820754716985e-06,
      "loss": 0.6251,
      "num_input_tokens_seen": 761088,
      "step": 1150
    },
    {
      "epoch": 0.6810141509433962,
      "grad_norm": 10.325052261352539,
      "learning_rate": 3.4021226415094343e-06,
      "loss": 0.5632,
      "num_input_tokens_seen": 764224,
      "step": 1155
    },
    {
      "epoch": 0.6839622641509434,
      "grad_norm": 6.405628204345703,
      "learning_rate": 3.4168632075471705e-06,
      "loss": 0.6473,
      "num_input_tokens_seen": 767488,
      "step": 1160
    },
    {
      "epoch": 0.6869103773584906,
      "grad_norm": 7.9999213218688965,
      "learning_rate": 3.4316037735849062e-06,
      "loss": 0.452,
      "num_input_tokens_seen": 770784,
      "step": 1165
    },
    {
      "epoch": 0.6898584905660378,
      "grad_norm": 7.728386402130127,
      "learning_rate": 3.446344339622642e-06,
      "loss": 0.6062,
      "num_input_tokens_seen": 773792,
      "step": 1170
    },
    {
      "epoch": 0.6928066037735849,
      "grad_norm": 7.399059295654297,
      "learning_rate": 3.4610849056603778e-06,
      "loss": 0.6429,
      "num_input_tokens_seen": 777216,
      "step": 1175
    },
    {
      "epoch": 0.6957547169811321,
      "grad_norm": 3.673644542694092,
      "learning_rate": 3.475825471698113e-06,
      "loss": 0.7303,
      "num_input_tokens_seen": 780544,
      "step": 1180
    },
    {
      "epoch": 0.6987028301886793,
      "grad_norm": 4.8195366859436035,
      "learning_rate": 3.4905660377358493e-06,
      "loss": 0.4687,
      "num_input_tokens_seen": 783264,
      "step": 1185
    },
    {
      "epoch": 0.7016509433962265,
      "grad_norm": 4.99150276184082,
      "learning_rate": 3.505306603773585e-06,
      "loss": 0.4716,
      "num_input_tokens_seen": 786688,
      "step": 1190
    },
    {
      "epoch": 0.7045990566037735,
      "grad_norm": 13.579989433288574,
      "learning_rate": 3.520047169811321e-06,
      "loss": 0.7469,
      "num_input_tokens_seen": 790112,
      "step": 1195
    },
    {
      "epoch": 0.7075471698113207,
      "grad_norm": 4.169748783111572,
      "learning_rate": 3.5347877358490566e-06,
      "loss": 0.7238,
      "num_input_tokens_seen": 792832,
      "step": 1200
    },
    {
      "epoch": 0.7104952830188679,
      "grad_norm": 4.344335079193115,
      "learning_rate": 3.549528301886793e-06,
      "loss": 0.5085,
      "num_input_tokens_seen": 795584,
      "step": 1205
    },
    {
      "epoch": 0.7134433962264151,
      "grad_norm": 4.301352500915527,
      "learning_rate": 3.5642688679245286e-06,
      "loss": 0.7549,
      "num_input_tokens_seen": 798848,
      "step": 1210
    },
    {
      "epoch": 0.7163915094339622,
      "grad_norm": 4.431910991668701,
      "learning_rate": 3.5790094339622643e-06,
      "loss": 0.6684,
      "num_input_tokens_seen": 802272,
      "step": 1215
    },
    {
      "epoch": 0.7193396226415094,
      "grad_norm": 5.362442970275879,
      "learning_rate": 3.59375e-06,
      "loss": 0.7879,
      "num_input_tokens_seen": 804960,
      "step": 1220
    },
    {
      "epoch": 0.7222877358490566,
      "grad_norm": 9.030864715576172,
      "learning_rate": 3.6084905660377363e-06,
      "loss": 0.5593,
      "num_input_tokens_seen": 807424,
      "step": 1225
    },
    {
      "epoch": 0.7252358490566038,
      "grad_norm": 11.150055885314941,
      "learning_rate": 3.623231132075472e-06,
      "loss": 0.6634,
      "num_input_tokens_seen": 809920,
      "step": 1230
    },
    {
      "epoch": 0.7281839622641509,
      "grad_norm": 5.471099853515625,
      "learning_rate": 3.637971698113208e-06,
      "loss": 0.5061,
      "num_input_tokens_seen": 812640,
      "step": 1235
    },
    {
      "epoch": 0.7311320754716981,
      "grad_norm": 6.307929039001465,
      "learning_rate": 3.652712264150944e-06,
      "loss": 0.7193,
      "num_input_tokens_seen": 817248,
      "step": 1240
    },
    {
      "epoch": 0.7340801886792453,
      "grad_norm": 9.318836212158203,
      "learning_rate": 3.66745283018868e-06,
      "loss": 0.7386,
      "num_input_tokens_seen": 819872,
      "step": 1245
    },
    {
      "epoch": 0.7370283018867925,
      "grad_norm": 6.369445323944092,
      "learning_rate": 3.6821933962264156e-06,
      "loss": 0.6575,
      "num_input_tokens_seen": 824256,
      "step": 1250
    },
    {
      "epoch": 0.7399764150943396,
      "grad_norm": 3.100823402404785,
      "learning_rate": 3.6969339622641513e-06,
      "loss": 0.6276,
      "num_input_tokens_seen": 827488,
      "step": 1255
    },
    {
      "epoch": 0.7429245283018868,
      "grad_norm": 8.468423843383789,
      "learning_rate": 3.7116745283018875e-06,
      "loss": 0.7224,
      "num_input_tokens_seen": 831488,
      "step": 1260
    },
    {
      "epoch": 0.745872641509434,
      "grad_norm": 6.153287887573242,
      "learning_rate": 3.726415094339623e-06,
      "loss": 0.6179,
      "num_input_tokens_seen": 834752,
      "step": 1265
    },
    {
      "epoch": 0.7488207547169812,
      "grad_norm": 6.040599346160889,
      "learning_rate": 3.7411556603773586e-06,
      "loss": 0.6258,
      "num_input_tokens_seen": 837216,
      "step": 1270
    },
    {
      "epoch": 0.7517688679245284,
      "grad_norm": 5.613821506500244,
      "learning_rate": 3.7558962264150944e-06,
      "loss": 0.7695,
      "num_input_tokens_seen": 841024,
      "step": 1275
    },
    {
      "epoch": 0.7547169811320755,
      "grad_norm": 5.608860015869141,
      "learning_rate": 3.77063679245283e-06,
      "loss": 0.7388,
      "num_input_tokens_seen": 843328,
      "step": 1280
    },
    {
      "epoch": 0.7576650943396226,
      "grad_norm": 4.624624729156494,
      "learning_rate": 3.7853773584905664e-06,
      "loss": 0.5092,
      "num_input_tokens_seen": 846912,
      "step": 1285
    },
    {
      "epoch": 0.7606132075471698,
      "grad_norm": 3.1260733604431152,
      "learning_rate": 3.800117924528302e-06,
      "loss": 0.6036,
      "num_input_tokens_seen": 849984,
      "step": 1290
    },
    {
      "epoch": 0.7635613207547169,
      "grad_norm": 5.125219345092773,
      "learning_rate": 3.814858490566038e-06,
      "loss": 0.6442,
      "num_input_tokens_seen": 853408,
      "step": 1295
    },
    {
      "epoch": 0.7665094339622641,
      "grad_norm": 2.7824866771698,
      "learning_rate": 3.829599056603774e-06,
      "loss": 0.5688,
      "num_input_tokens_seen": 856480,
      "step": 1300
    },
    {
      "epoch": 0.7694575471698113,
      "grad_norm": 4.030657768249512,
      "learning_rate": 3.8443396226415094e-06,
      "loss": 0.6107,
      "num_input_tokens_seen": 860320,
      "step": 1305
    },
    {
      "epoch": 0.7724056603773585,
      "grad_norm": 9.663349151611328,
      "learning_rate": 3.859080188679246e-06,
      "loss": 0.5536,
      "num_input_tokens_seen": 862912,
      "step": 1310
    },
    {
      "epoch": 0.7753537735849056,
      "grad_norm": 4.994947910308838,
      "learning_rate": 3.873820754716982e-06,
      "loss": 0.6512,
      "num_input_tokens_seen": 865952,
      "step": 1315
    },
    {
      "epoch": 0.7783018867924528,
      "grad_norm": 5.952736854553223,
      "learning_rate": 3.888561320754717e-06,
      "loss": 0.6134,
      "num_input_tokens_seen": 869376,
      "step": 1320
    },
    {
      "epoch": 0.78125,
      "grad_norm": 7.859884262084961,
      "learning_rate": 3.903301886792453e-06,
      "loss": 0.5273,
      "num_input_tokens_seen": 871968,
      "step": 1325
    },
    {
      "epoch": 0.7841981132075472,
      "grad_norm": 3.2229883670806885,
      "learning_rate": 3.9180424528301895e-06,
      "loss": 0.4988,
      "num_input_tokens_seen": 875872,
      "step": 1330
    },
    {
      "epoch": 0.7871462264150944,
      "grad_norm": 29.08355712890625,
      "learning_rate": 3.932783018867925e-06,
      "loss": 0.5674,
      "num_input_tokens_seen": 878496,
      "step": 1335
    },
    {
      "epoch": 0.7900943396226415,
      "grad_norm": 5.119779109954834,
      "learning_rate": 3.947523584905661e-06,
      "loss": 0.5671,
      "num_input_tokens_seen": 881088,
      "step": 1340
    },
    {
      "epoch": 0.7930424528301887,
      "grad_norm": 4.569984436035156,
      "learning_rate": 3.962264150943396e-06,
      "loss": 0.606,
      "num_input_tokens_seen": 884320,
      "step": 1345
    },
    {
      "epoch": 0.7959905660377359,
      "grad_norm": 2.7600409984588623,
      "learning_rate": 3.977004716981133e-06,
      "loss": 0.5859,
      "num_input_tokens_seen": 886464,
      "step": 1350
    },
    {
      "epoch": 0.7989386792452831,
      "grad_norm": 7.066147804260254,
      "learning_rate": 3.991745283018868e-06,
      "loss": 0.778,
      "num_input_tokens_seen": 889920,
      "step": 1355
    },
    {
      "epoch": 0.8018867924528302,
      "grad_norm": 6.630670547485352,
      "learning_rate": 4.006485849056604e-06,
      "loss": 0.625,
      "num_input_tokens_seen": 892448,
      "step": 1360
    },
    {
      "epoch": 0.8048349056603774,
      "grad_norm": 4.426787376403809,
      "learning_rate": 4.0212264150943395e-06,
      "loss": 0.5854,
      "num_input_tokens_seen": 895968,
      "step": 1365
    },
    {
      "epoch": 0.8077830188679245,
      "grad_norm": 5.518293857574463,
      "learning_rate": 4.035966981132076e-06,
      "loss": 0.6505,
      "num_input_tokens_seen": 899200,
      "step": 1370
    },
    {
      "epoch": 0.8107311320754716,
      "grad_norm": 6.248364448547363,
      "learning_rate": 4.050707547169812e-06,
      "loss": 0.6649,
      "num_input_tokens_seen": 901696,
      "step": 1375
    },
    {
      "epoch": 0.8136792452830188,
      "grad_norm": 6.829316139221191,
      "learning_rate": 4.065448113207547e-06,
      "loss": 0.4304,
      "num_input_tokens_seen": 904672,
      "step": 1380
    },
    {
      "epoch": 0.816627358490566,
      "grad_norm": 7.274075508117676,
      "learning_rate": 4.080188679245283e-06,
      "loss": 0.7747,
      "num_input_tokens_seen": 907328,
      "step": 1385
    },
    {
      "epoch": 0.8195754716981132,
      "grad_norm": 10.069584846496582,
      "learning_rate": 4.094929245283019e-06,
      "loss": 0.5602,
      "num_input_tokens_seen": 909920,
      "step": 1390
    },
    {
      "epoch": 0.8225235849056604,
      "grad_norm": 7.74467134475708,
      "learning_rate": 4.109669811320755e-06,
      "loss": 0.7597,
      "num_input_tokens_seen": 912608,
      "step": 1395
    },
    {
      "epoch": 0.8254716981132075,
      "grad_norm": 6.81733512878418,
      "learning_rate": 4.124410377358491e-06,
      "loss": 0.509,
      "num_input_tokens_seen": 915872,
      "step": 1400
    },
    {
      "epoch": 0.8284198113207547,
      "grad_norm": 6.004680156707764,
      "learning_rate": 4.1391509433962265e-06,
      "loss": 0.4469,
      "num_input_tokens_seen": 918400,
      "step": 1405
    },
    {
      "epoch": 0.8313679245283019,
      "grad_norm": 4.820994853973389,
      "learning_rate": 4.153891509433963e-06,
      "loss": 0.4976,
      "num_input_tokens_seen": 921312,
      "step": 1410
    },
    {
      "epoch": 0.8343160377358491,
      "grad_norm": 5.8090386390686035,
      "learning_rate": 4.168632075471699e-06,
      "loss": 0.5315,
      "num_input_tokens_seen": 925888,
      "step": 1415
    },
    {
      "epoch": 0.8372641509433962,
      "grad_norm": 4.7301530838012695,
      "learning_rate": 4.183372641509434e-06,
      "loss": 0.5742,
      "num_input_tokens_seen": 929792,
      "step": 1420
    },
    {
      "epoch": 0.8402122641509434,
      "grad_norm": 3.9643592834472656,
      "learning_rate": 4.19811320754717e-06,
      "loss": 0.4752,
      "num_input_tokens_seen": 932672,
      "step": 1425
    },
    {
      "epoch": 0.8431603773584906,
      "grad_norm": 6.6070146560668945,
      "learning_rate": 4.212853773584907e-06,
      "loss": 0.6862,
      "num_input_tokens_seen": 935552,
      "step": 1430
    },
    {
      "epoch": 0.8461084905660378,
      "grad_norm": 2.467632293701172,
      "learning_rate": 4.227594339622642e-06,
      "loss": 0.3899,
      "num_input_tokens_seen": 939456,
      "step": 1435
    },
    {
      "epoch": 0.8490566037735849,
      "grad_norm": 4.7261433601379395,
      "learning_rate": 4.242334905660378e-06,
      "loss": 0.6798,
      "num_input_tokens_seen": 943552,
      "step": 1440
    },
    {
      "epoch": 0.8520047169811321,
      "grad_norm": 3.0802769660949707,
      "learning_rate": 4.2570754716981135e-06,
      "loss": 0.5412,
      "num_input_tokens_seen": 946464,
      "step": 1445
    },
    {
      "epoch": 0.8549528301886793,
      "grad_norm": 3.9946064949035645,
      "learning_rate": 4.271816037735849e-06,
      "loss": 0.5545,
      "num_input_tokens_seen": 950816,
      "step": 1450
    },
    {
      "epoch": 0.8579009433962265,
      "grad_norm": 8.363245964050293,
      "learning_rate": 4.286556603773585e-06,
      "loss": 0.6894,
      "num_input_tokens_seen": 953216,
      "step": 1455
    },
    {
      "epoch": 0.8608490566037735,
      "grad_norm": 3.245978832244873,
      "learning_rate": 4.301297169811321e-06,
      "loss": 0.5793,
      "num_input_tokens_seen": 956064,
      "step": 1460
    },
    {
      "epoch": 0.8637971698113207,
      "grad_norm": 5.080158233642578,
      "learning_rate": 4.3160377358490565e-06,
      "loss": 0.7675,
      "num_input_tokens_seen": 959328,
      "step": 1465
    },
    {
      "epoch": 0.8667452830188679,
      "grad_norm": 3.148301601409912,
      "learning_rate": 4.330778301886793e-06,
      "loss": 0.4885,
      "num_input_tokens_seen": 962368,
      "step": 1470
    },
    {
      "epoch": 0.8696933962264151,
      "grad_norm": 4.756036281585693,
      "learning_rate": 4.345518867924529e-06,
      "loss": 0.668,
      "num_input_tokens_seen": 966080,
      "step": 1475
    },
    {
      "epoch": 0.8726415094339622,
      "grad_norm": 10.953571319580078,
      "learning_rate": 4.360259433962264e-06,
      "loss": 0.5523,
      "num_input_tokens_seen": 969344,
      "step": 1480
    },
    {
      "epoch": 0.8755896226415094,
      "grad_norm": 5.313912391662598,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 0.6765,
      "num_input_tokens_seen": 973056,
      "step": 1485
    },
    {
      "epoch": 0.8785377358490566,
      "grad_norm": 5.534646987915039,
      "learning_rate": 4.389740566037737e-06,
      "loss": 0.7078,
      "num_input_tokens_seen": 975936,
      "step": 1490
    },
    {
      "epoch": 0.8814858490566038,
      "grad_norm": 4.492164134979248,
      "learning_rate": 4.404481132075472e-06,
      "loss": 0.5228,
      "num_input_tokens_seen": 979584,
      "step": 1495
    },
    {
      "epoch": 0.8844339622641509,
      "grad_norm": 7.302002429962158,
      "learning_rate": 4.419221698113208e-06,
      "loss": 0.618,
      "num_input_tokens_seen": 982528,
      "step": 1500
    },
    {
      "epoch": 0.8873820754716981,
      "grad_norm": 3.1560916900634766,
      "learning_rate": 4.4339622641509435e-06,
      "loss": 0.549,
      "num_input_tokens_seen": 985504,
      "step": 1505
    },
    {
      "epoch": 0.8903301886792453,
      "grad_norm": 6.716491222381592,
      "learning_rate": 4.44870283018868e-06,
      "loss": 0.6405,
      "num_input_tokens_seen": 988224,
      "step": 1510
    },
    {
      "epoch": 0.8932783018867925,
      "grad_norm": 2.926795482635498,
      "learning_rate": 4.463443396226416e-06,
      "loss": 0.5217,
      "num_input_tokens_seen": 991424,
      "step": 1515
    },
    {
      "epoch": 0.8962264150943396,
      "grad_norm": 5.689585208892822,
      "learning_rate": 4.478183962264151e-06,
      "loss": 0.4717,
      "num_input_tokens_seen": 994784,
      "step": 1520
    },
    {
      "epoch": 0.8991745283018868,
      "grad_norm": 5.55499792098999,
      "learning_rate": 4.4929245283018875e-06,
      "loss": 0.636,
      "num_input_tokens_seen": 997600,
      "step": 1525
    },
    {
      "epoch": 0.902122641509434,
      "grad_norm": 2.786703109741211,
      "learning_rate": 4.507665094339623e-06,
      "loss": 0.4938,
      "num_input_tokens_seen": 1001024,
      "step": 1530
    },
    {
      "epoch": 0.9050707547169812,
      "grad_norm": 3.772987127304077,
      "learning_rate": 4.522405660377359e-06,
      "loss": 0.6038,
      "num_input_tokens_seen": 1003968,
      "step": 1535
    },
    {
      "epoch": 0.9080188679245284,
      "grad_norm": 3.227774143218994,
      "learning_rate": 4.537146226415094e-06,
      "loss": 0.7326,
      "num_input_tokens_seen": 1011200,
      "step": 1540
    },
    {
      "epoch": 0.9109669811320755,
      "grad_norm": 3.1589229106903076,
      "learning_rate": 4.5518867924528305e-06,
      "loss": 0.4789,
      "num_input_tokens_seen": 1018080,
      "step": 1545
    },
    {
      "epoch": 0.9139150943396226,
      "grad_norm": 6.886144638061523,
      "learning_rate": 4.566627358490566e-06,
      "loss": 0.5367,
      "num_input_tokens_seen": 1020832,
      "step": 1550
    },
    {
      "epoch": 0.9168632075471698,
      "grad_norm": 9.925893783569336,
      "learning_rate": 4.581367924528302e-06,
      "loss": 0.6011,
      "num_input_tokens_seen": 1023744,
      "step": 1555
    },
    {
      "epoch": 0.9198113207547169,
      "grad_norm": 3.9393069744110107,
      "learning_rate": 4.596108490566038e-06,
      "loss": 0.5486,
      "num_input_tokens_seen": 1028096,
      "step": 1560
    },
    {
      "epoch": 0.9227594339622641,
      "grad_norm": 3.165034055709839,
      "learning_rate": 4.610849056603774e-06,
      "loss": 0.672,
      "num_input_tokens_seen": 1031712,
      "step": 1565
    },
    {
      "epoch": 0.9257075471698113,
      "grad_norm": 2.696486711502075,
      "learning_rate": 4.62558962264151e-06,
      "loss": 0.5141,
      "num_input_tokens_seen": 1035744,
      "step": 1570
    },
    {
      "epoch": 0.9286556603773585,
      "grad_norm": 2.668795347213745,
      "learning_rate": 4.640330188679246e-06,
      "loss": 0.5329,
      "num_input_tokens_seen": 1039616,
      "step": 1575
    },
    {
      "epoch": 0.9316037735849056,
      "grad_norm": 4.632341384887695,
      "learning_rate": 4.655070754716981e-06,
      "loss": 0.6611,
      "num_input_tokens_seen": 1042496,
      "step": 1580
    },
    {
      "epoch": 0.9345518867924528,
      "grad_norm": 9.076416969299316,
      "learning_rate": 4.6698113207547175e-06,
      "loss": 0.6145,
      "num_input_tokens_seen": 1045568,
      "step": 1585
    },
    {
      "epoch": 0.9375,
      "grad_norm": 4.5063958168029785,
      "learning_rate": 4.684551886792454e-06,
      "loss": 0.5037,
      "num_input_tokens_seen": 1048480,
      "step": 1590
    },
    {
      "epoch": 0.9404481132075472,
      "grad_norm": 7.206030368804932,
      "learning_rate": 4.699292452830189e-06,
      "loss": 0.6527,
      "num_input_tokens_seen": 1050624,
      "step": 1595
    },
    {
      "epoch": 0.9433962264150944,
      "grad_norm": 3.055152416229248,
      "learning_rate": 4.714033018867925e-06,
      "loss": 0.7619,
      "num_input_tokens_seen": 1054624,
      "step": 1600
    },
    {
      "epoch": 0.9463443396226415,
      "grad_norm": 5.735093593597412,
      "learning_rate": 4.728773584905661e-06,
      "loss": 0.5878,
      "num_input_tokens_seen": 1057888,
      "step": 1605
    },
    {
      "epoch": 0.9492924528301887,
      "grad_norm": 3.8677663803100586,
      "learning_rate": 4.743514150943397e-06,
      "loss": 0.5211,
      "num_input_tokens_seen": 1060832,
      "step": 1610
    },
    {
      "epoch": 0.9522405660377359,
      "grad_norm": 4.158363342285156,
      "learning_rate": 4.758254716981133e-06,
      "loss": 0.7955,
      "num_input_tokens_seen": 1065088,
      "step": 1615
    },
    {
      "epoch": 0.9551886792452831,
      "grad_norm": 4.73134183883667,
      "learning_rate": 4.772995283018868e-06,
      "loss": 0.68,
      "num_input_tokens_seen": 1068864,
      "step": 1620
    },
    {
      "epoch": 0.9581367924528302,
      "grad_norm": 5.885763645172119,
      "learning_rate": 4.787735849056604e-06,
      "loss": 0.6414,
      "num_input_tokens_seen": 1071616,
      "step": 1625
    },
    {
      "epoch": 0.9610849056603774,
      "grad_norm": 5.330148696899414,
      "learning_rate": 4.80247641509434e-06,
      "loss": 0.6036,
      "num_input_tokens_seen": 1074560,
      "step": 1630
    },
    {
      "epoch": 0.9640330188679245,
      "grad_norm": 2.30084490776062,
      "learning_rate": 4.817216981132076e-06,
      "loss": 0.4249,
      "num_input_tokens_seen": 1079648,
      "step": 1635
    },
    {
      "epoch": 0.9669811320754716,
      "grad_norm": 2.898063898086548,
      "learning_rate": 4.831957547169811e-06,
      "loss": 0.5219,
      "num_input_tokens_seen": 1083520,
      "step": 1640
    },
    {
      "epoch": 0.9699292452830188,
      "grad_norm": 2.5638439655303955,
      "learning_rate": 4.8466981132075476e-06,
      "loss": 0.4732,
      "num_input_tokens_seen": 1086944,
      "step": 1645
    },
    {
      "epoch": 0.972877358490566,
      "grad_norm": 4.182637691497803,
      "learning_rate": 4.861438679245283e-06,
      "loss": 0.4876,
      "num_input_tokens_seen": 1090112,
      "step": 1650
    },
    {
      "epoch": 0.9758254716981132,
      "grad_norm": 3.797821521759033,
      "learning_rate": 4.876179245283019e-06,
      "loss": 0.6371,
      "num_input_tokens_seen": 1093152,
      "step": 1655
    },
    {
      "epoch": 0.9787735849056604,
      "grad_norm": 3.232534170150757,
      "learning_rate": 4.890919811320755e-06,
      "loss": 0.7638,
      "num_input_tokens_seen": 1095904,
      "step": 1660
    },
    {
      "epoch": 0.9817216981132075,
      "grad_norm": 24.084638595581055,
      "learning_rate": 4.905660377358491e-06,
      "loss": 0.598,
      "num_input_tokens_seen": 1099488,
      "step": 1665
    },
    {
      "epoch": 0.9846698113207547,
      "grad_norm": 5.937756538391113,
      "learning_rate": 4.920400943396227e-06,
      "loss": 0.4763,
      "num_input_tokens_seen": 1102112,
      "step": 1670
    },
    {
      "epoch": 0.9876179245283019,
      "grad_norm": 2.903235673904419,
      "learning_rate": 4.935141509433963e-06,
      "loss": 0.6022,
      "num_input_tokens_seen": 1107136,
      "step": 1675
    },
    {
      "epoch": 0.9905660377358491,
      "grad_norm": 3.5739521980285645,
      "learning_rate": 4.949882075471698e-06,
      "loss": 0.6062,
      "num_input_tokens_seen": 1109856,
      "step": 1680
    },
    {
      "epoch": 0.9935141509433962,
      "grad_norm": 4.700026035308838,
      "learning_rate": 4.9646226415094346e-06,
      "loss": 0.6143,
      "num_input_tokens_seen": 1112704,
      "step": 1685
    },
    {
      "epoch": 0.9964622641509434,
      "grad_norm": 4.1179890632629395,
      "learning_rate": 4.979363207547171e-06,
      "loss": 0.6408,
      "num_input_tokens_seen": 1115872,
      "step": 1690
    },
    {
      "epoch": 0.9994103773584906,
      "grad_norm": 3.357954502105713,
      "learning_rate": 4.994103773584906e-06,
      "loss": 0.5649,
      "num_input_tokens_seen": 1119136,
      "step": 1695
    },
    {
      "epoch": 1.0023584905660377,
      "grad_norm": 13.908370971679688,
      "learning_rate": 5.0088443396226414e-06,
      "loss": 0.5179,
      "num_input_tokens_seen": 1122416,
      "step": 1700
    },
    {
      "epoch": 1.005306603773585,
      "grad_norm": 5.964030742645264,
      "learning_rate": 5.023584905660378e-06,
      "loss": 0.3608,
      "num_input_tokens_seen": 1125744,
      "step": 1705
    },
    {
      "epoch": 1.008254716981132,
      "grad_norm": 4.8375420570373535,
      "learning_rate": 5.038325471698113e-06,
      "loss": 0.5221,
      "num_input_tokens_seen": 1128624,
      "step": 1710
    },
    {
      "epoch": 1.0112028301886793,
      "grad_norm": 2.7695775032043457,
      "learning_rate": 5.05306603773585e-06,
      "loss": 0.5201,
      "num_input_tokens_seen": 1132048,
      "step": 1715
    },
    {
      "epoch": 1.0141509433962264,
      "grad_norm": 3.7262253761291504,
      "learning_rate": 5.067806603773585e-06,
      "loss": 0.4196,
      "num_input_tokens_seen": 1135248,
      "step": 1720
    },
    {
      "epoch": 1.0170990566037736,
      "grad_norm": 4.333666801452637,
      "learning_rate": 5.0825471698113216e-06,
      "loss": 0.5721,
      "num_input_tokens_seen": 1138000,
      "step": 1725
    },
    {
      "epoch": 1.0200471698113207,
      "grad_norm": 3.417595386505127,
      "learning_rate": 5.097287735849057e-06,
      "loss": 0.6633,
      "num_input_tokens_seen": 1140752,
      "step": 1730
    },
    {
      "epoch": 1.022995283018868,
      "grad_norm": 4.6944098472595215,
      "learning_rate": 5.112028301886793e-06,
      "loss": 0.5353,
      "num_input_tokens_seen": 1142832,
      "step": 1735
    },
    {
      "epoch": 1.025943396226415,
      "grad_norm": 4.175461292266846,
      "learning_rate": 5.1267688679245284e-06,
      "loss": 0.6396,
      "num_input_tokens_seen": 1146704,
      "step": 1740
    },
    {
      "epoch": 1.0288915094339623,
      "grad_norm": 3.890413284301758,
      "learning_rate": 5.1415094339622655e-06,
      "loss": 0.5119,
      "num_input_tokens_seen": 1150704,
      "step": 1745
    },
    {
      "epoch": 1.0318396226415094,
      "grad_norm": 5.05267858505249,
      "learning_rate": 5.156250000000001e-06,
      "loss": 0.5417,
      "num_input_tokens_seen": 1153520,
      "step": 1750
    },
    {
      "epoch": 1.0347877358490567,
      "grad_norm": 4.987242221832275,
      "learning_rate": 5.170990566037736e-06,
      "loss": 0.5309,
      "num_input_tokens_seen": 1156656,
      "step": 1755
    },
    {
      "epoch": 1.0377358490566038,
      "grad_norm": 22.030105590820312,
      "learning_rate": 5.185731132075472e-06,
      "loss": 0.6719,
      "num_input_tokens_seen": 1160112,
      "step": 1760
    },
    {
      "epoch": 1.040683962264151,
      "grad_norm": 3.4976494312286377,
      "learning_rate": 5.200471698113208e-06,
      "loss": 0.6865,
      "num_input_tokens_seen": 1164336,
      "step": 1765
    },
    {
      "epoch": 1.0436320754716981,
      "grad_norm": 3.887812852859497,
      "learning_rate": 5.215212264150944e-06,
      "loss": 0.5965,
      "num_input_tokens_seen": 1168144,
      "step": 1770
    },
    {
      "epoch": 1.0465801886792452,
      "grad_norm": 3.0114710330963135,
      "learning_rate": 5.229952830188679e-06,
      "loss": 0.6306,
      "num_input_tokens_seen": 1172496,
      "step": 1775
    },
    {
      "epoch": 1.0495283018867925,
      "grad_norm": 6.255817890167236,
      "learning_rate": 5.2446933962264154e-06,
      "loss": 0.6169,
      "num_input_tokens_seen": 1176048,
      "step": 1780
    },
    {
      "epoch": 1.0524764150943395,
      "grad_norm": 2.976905345916748,
      "learning_rate": 5.259433962264151e-06,
      "loss": 0.5546,
      "num_input_tokens_seen": 1178896,
      "step": 1785
    },
    {
      "epoch": 1.0554245283018868,
      "grad_norm": 14.048487663269043,
      "learning_rate": 5.274174528301888e-06,
      "loss": 0.438,
      "num_input_tokens_seen": 1182192,
      "step": 1790
    },
    {
      "epoch": 1.0583726415094339,
      "grad_norm": 3.9571468830108643,
      "learning_rate": 5.288915094339623e-06,
      "loss": 0.5778,
      "num_input_tokens_seen": 1185104,
      "step": 1795
    },
    {
      "epoch": 1.0613207547169812,
      "grad_norm": 2.2025251388549805,
      "learning_rate": 5.303655660377359e-06,
      "loss": 0.7052,
      "num_input_tokens_seen": 1188592,
      "step": 1800
    },
    {
      "epoch": 1.0642688679245282,
      "grad_norm": 2.820408344268799,
      "learning_rate": 5.318396226415095e-06,
      "loss": 0.5746,
      "num_input_tokens_seen": 1192048,
      "step": 1805
    },
    {
      "epoch": 1.0672169811320755,
      "grad_norm": 3.058142900466919,
      "learning_rate": 5.333136792452831e-06,
      "loss": 0.5713,
      "num_input_tokens_seen": 1195984,
      "step": 1810
    },
    {
      "epoch": 1.0701650943396226,
      "grad_norm": 5.3654890060424805,
      "learning_rate": 5.347877358490566e-06,
      "loss": 0.6564,
      "num_input_tokens_seen": 1198800,
      "step": 1815
    },
    {
      "epoch": 1.0731132075471699,
      "grad_norm": 4.213716983795166,
      "learning_rate": 5.362617924528302e-06,
      "loss": 0.5165,
      "num_input_tokens_seen": 1202480,
      "step": 1820
    },
    {
      "epoch": 1.076061320754717,
      "grad_norm": 5.368893146514893,
      "learning_rate": 5.377358490566038e-06,
      "loss": 0.5944,
      "num_input_tokens_seen": 1206000,
      "step": 1825
    },
    {
      "epoch": 1.0790094339622642,
      "grad_norm": 5.8786187171936035,
      "learning_rate": 5.392099056603775e-06,
      "loss": 0.6458,
      "num_input_tokens_seen": 1208848,
      "step": 1830
    },
    {
      "epoch": 1.0819575471698113,
      "grad_norm": 2.806875705718994,
      "learning_rate": 5.40683962264151e-06,
      "loss": 0.4807,
      "num_input_tokens_seen": 1211856,
      "step": 1835
    },
    {
      "epoch": 1.0849056603773586,
      "grad_norm": 6.900170803070068,
      "learning_rate": 5.4215801886792455e-06,
      "loss": 0.5681,
      "num_input_tokens_seen": 1214576,
      "step": 1840
    },
    {
      "epoch": 1.0878537735849056,
      "grad_norm": 2.819911003112793,
      "learning_rate": 5.436320754716982e-06,
      "loss": 0.6066,
      "num_input_tokens_seen": 1218096,
      "step": 1845
    },
    {
      "epoch": 1.0908018867924527,
      "grad_norm": 3.4149374961853027,
      "learning_rate": 5.451061320754717e-06,
      "loss": 0.5001,
      "num_input_tokens_seen": 1221232,
      "step": 1850
    },
    {
      "epoch": 1.09375,
      "grad_norm": 2.5203027725219727,
      "learning_rate": 5.465801886792453e-06,
      "loss": 0.577,
      "num_input_tokens_seen": 1224112,
      "step": 1855
    },
    {
      "epoch": 1.0966981132075473,
      "grad_norm": 4.0001935958862305,
      "learning_rate": 5.4805424528301886e-06,
      "loss": 0.5246,
      "num_input_tokens_seen": 1226768,
      "step": 1860
    },
    {
      "epoch": 1.0996462264150944,
      "grad_norm": 6.391879081726074,
      "learning_rate": 5.495283018867925e-06,
      "loss": 0.5428,
      "num_input_tokens_seen": 1229488,
      "step": 1865
    },
    {
      "epoch": 1.1025943396226414,
      "grad_norm": 5.015661716461182,
      "learning_rate": 5.51002358490566e-06,
      "loss": 0.6491,
      "num_input_tokens_seen": 1233296,
      "step": 1870
    },
    {
      "epoch": 1.1055424528301887,
      "grad_norm": 3.5543301105499268,
      "learning_rate": 5.524764150943397e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 1236176,
      "step": 1875
    },
    {
      "epoch": 1.1084905660377358,
      "grad_norm": 2.2238500118255615,
      "learning_rate": 5.5395047169811325e-06,
      "loss": 0.5167,
      "num_input_tokens_seen": 1239472,
      "step": 1880
    },
    {
      "epoch": 1.111438679245283,
      "grad_norm": 7.680689334869385,
      "learning_rate": 5.554245283018869e-06,
      "loss": 0.711,
      "num_input_tokens_seen": 1242352,
      "step": 1885
    },
    {
      "epoch": 1.1143867924528301,
      "grad_norm": 2.91620135307312,
      "learning_rate": 5.568985849056604e-06,
      "loss": 0.6871,
      "num_input_tokens_seen": 1245456,
      "step": 1890
    },
    {
      "epoch": 1.1173349056603774,
      "grad_norm": 2.9253294467926025,
      "learning_rate": 5.58372641509434e-06,
      "loss": 0.6044,
      "num_input_tokens_seen": 1248432,
      "step": 1895
    },
    {
      "epoch": 1.1202830188679245,
      "grad_norm": 2.866159439086914,
      "learning_rate": 5.5984669811320755e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 1252176,
      "step": 1900
    },
    {
      "epoch": 1.1232311320754718,
      "grad_norm": 2.2031021118164062,
      "learning_rate": 5.613207547169813e-06,
      "loss": 0.4568,
      "num_input_tokens_seen": 1255312,
      "step": 1905
    },
    {
      "epoch": 1.1261792452830188,
      "grad_norm": 5.807628154754639,
      "learning_rate": 5.627948113207548e-06,
      "loss": 0.5332,
      "num_input_tokens_seen": 1258736,
      "step": 1910
    },
    {
      "epoch": 1.1291273584905661,
      "grad_norm": 6.226014137268066,
      "learning_rate": 5.642688679245284e-06,
      "loss": 0.6241,
      "num_input_tokens_seen": 1262224,
      "step": 1915
    },
    {
      "epoch": 1.1320754716981132,
      "grad_norm": 3.1039791107177734,
      "learning_rate": 5.6574292452830195e-06,
      "loss": 0.6054,
      "num_input_tokens_seen": 1265264,
      "step": 1920
    },
    {
      "epoch": 1.1350235849056605,
      "grad_norm": 5.4412384033203125,
      "learning_rate": 5.672169811320756e-06,
      "loss": 0.7396,
      "num_input_tokens_seen": 1267696,
      "step": 1925
    },
    {
      "epoch": 1.1379716981132075,
      "grad_norm": 3.0763511657714844,
      "learning_rate": 5.686910377358491e-06,
      "loss": 0.4089,
      "num_input_tokens_seen": 1271536,
      "step": 1930
    },
    {
      "epoch": 1.1409198113207548,
      "grad_norm": 3.9392354488372803,
      "learning_rate": 5.701650943396226e-06,
      "loss": 0.5402,
      "num_input_tokens_seen": 1274480,
      "step": 1935
    },
    {
      "epoch": 1.1438679245283019,
      "grad_norm": 3.276834726333618,
      "learning_rate": 5.7163915094339625e-06,
      "loss": 0.5844,
      "num_input_tokens_seen": 1277552,
      "step": 1940
    },
    {
      "epoch": 1.146816037735849,
      "grad_norm": 2.603140354156494,
      "learning_rate": 5.731132075471698e-06,
      "loss": 0.6108,
      "num_input_tokens_seen": 1281328,
      "step": 1945
    },
    {
      "epoch": 1.1497641509433962,
      "grad_norm": 2.821455478668213,
      "learning_rate": 5.745872641509435e-06,
      "loss": 0.4906,
      "num_input_tokens_seen": 1284016,
      "step": 1950
    },
    {
      "epoch": 1.1527122641509433,
      "grad_norm": 3.787261724472046,
      "learning_rate": 5.76061320754717e-06,
      "loss": 0.4547,
      "num_input_tokens_seen": 1287184,
      "step": 1955
    },
    {
      "epoch": 1.1556603773584906,
      "grad_norm": 5.147575378417969,
      "learning_rate": 5.7753537735849065e-06,
      "loss": 0.4502,
      "num_input_tokens_seen": 1290128,
      "step": 1960
    },
    {
      "epoch": 1.1586084905660377,
      "grad_norm": 4.2200727462768555,
      "learning_rate": 5.790094339622642e-06,
      "loss": 0.4668,
      "num_input_tokens_seen": 1292944,
      "step": 1965
    },
    {
      "epoch": 1.161556603773585,
      "grad_norm": 12.64889144897461,
      "learning_rate": 5.804834905660378e-06,
      "loss": 0.5097,
      "num_input_tokens_seen": 1295312,
      "step": 1970
    },
    {
      "epoch": 1.164504716981132,
      "grad_norm": 3.6878573894500732,
      "learning_rate": 5.819575471698113e-06,
      "loss": 0.5953,
      "num_input_tokens_seen": 1298704,
      "step": 1975
    },
    {
      "epoch": 1.1674528301886793,
      "grad_norm": 2.7699413299560547,
      "learning_rate": 5.8343160377358495e-06,
      "loss": 0.5009,
      "num_input_tokens_seen": 1301808,
      "step": 1980
    },
    {
      "epoch": 1.1704009433962264,
      "grad_norm": 2.697293519973755,
      "learning_rate": 5.849056603773585e-06,
      "loss": 0.6036,
      "num_input_tokens_seen": 1304560,
      "step": 1985
    },
    {
      "epoch": 1.1733490566037736,
      "grad_norm": 2.487804651260376,
      "learning_rate": 5.863797169811322e-06,
      "loss": 0.5975,
      "num_input_tokens_seen": 1307632,
      "step": 1990
    },
    {
      "epoch": 1.1762971698113207,
      "grad_norm": 7.155288219451904,
      "learning_rate": 5.878537735849057e-06,
      "loss": 0.5743,
      "num_input_tokens_seen": 1310576,
      "step": 1995
    },
    {
      "epoch": 1.179245283018868,
      "grad_norm": 3.3397135734558105,
      "learning_rate": 5.8932783018867934e-06,
      "loss": 0.5316,
      "num_input_tokens_seen": 1314544,
      "step": 2000
    },
    {
      "epoch": 1.182193396226415,
      "grad_norm": 3.8805930614471436,
      "learning_rate": 5.908018867924529e-06,
      "loss": 0.706,
      "num_input_tokens_seen": 1316912,
      "step": 2005
    },
    {
      "epoch": 1.1851415094339623,
      "grad_norm": 3.6034162044525146,
      "learning_rate": 5.922759433962265e-06,
      "loss": 0.4615,
      "num_input_tokens_seen": 1319952,
      "step": 2010
    },
    {
      "epoch": 1.1880896226415094,
      "grad_norm": 4.618619918823242,
      "learning_rate": 5.9375e-06,
      "loss": 0.5915,
      "num_input_tokens_seen": 1322992,
      "step": 2015
    },
    {
      "epoch": 1.1910377358490567,
      "grad_norm": 3.0141241550445557,
      "learning_rate": 5.952240566037736e-06,
      "loss": 0.5422,
      "num_input_tokens_seen": 1325936,
      "step": 2020
    },
    {
      "epoch": 1.1939858490566038,
      "grad_norm": 2.86138653755188,
      "learning_rate": 5.966981132075472e-06,
      "loss": 0.4733,
      "num_input_tokens_seen": 1329872,
      "step": 2025
    },
    {
      "epoch": 1.196933962264151,
      "grad_norm": 4.505765438079834,
      "learning_rate": 5.981721698113207e-06,
      "loss": 0.5504,
      "num_input_tokens_seen": 1334928,
      "step": 2030
    },
    {
      "epoch": 1.1998820754716981,
      "grad_norm": 2.7900173664093018,
      "learning_rate": 5.996462264150944e-06,
      "loss": 0.4764,
      "num_input_tokens_seen": 1337648,
      "step": 2035
    },
    {
      "epoch": 1.2028301886792452,
      "grad_norm": 1.807868480682373,
      "learning_rate": 6.01120283018868e-06,
      "loss": 0.7481,
      "num_input_tokens_seen": 1340592,
      "step": 2040
    },
    {
      "epoch": 1.2057783018867925,
      "grad_norm": 3.297664165496826,
      "learning_rate": 6.025943396226416e-06,
      "loss": 0.5213,
      "num_input_tokens_seen": 1344752,
      "step": 2045
    },
    {
      "epoch": 1.2087264150943395,
      "grad_norm": 2.691549777984619,
      "learning_rate": 6.040683962264151e-06,
      "loss": 0.572,
      "num_input_tokens_seen": 1347952,
      "step": 2050
    },
    {
      "epoch": 1.2116745283018868,
      "grad_norm": 4.4658355712890625,
      "learning_rate": 6.055424528301887e-06,
      "loss": 0.5808,
      "num_input_tokens_seen": 1350416,
      "step": 2055
    },
    {
      "epoch": 1.2146226415094339,
      "grad_norm": 7.120859146118164,
      "learning_rate": 6.070165094339623e-06,
      "loss": 0.6433,
      "num_input_tokens_seen": 1354160,
      "step": 2060
    },
    {
      "epoch": 1.2175707547169812,
      "grad_norm": 2.4757378101348877,
      "learning_rate": 6.08490566037736e-06,
      "loss": 0.5157,
      "num_input_tokens_seen": 1357104,
      "step": 2065
    },
    {
      "epoch": 1.2205188679245282,
      "grad_norm": 9.764918327331543,
      "learning_rate": 6.099646226415095e-06,
      "loss": 0.7669,
      "num_input_tokens_seen": 1360048,
      "step": 2070
    },
    {
      "epoch": 1.2234669811320755,
      "grad_norm": 5.175655841827393,
      "learning_rate": 6.114386792452831e-06,
      "loss": 0.4846,
      "num_input_tokens_seen": 1362960,
      "step": 2075
    },
    {
      "epoch": 1.2264150943396226,
      "grad_norm": 3.341864585876465,
      "learning_rate": 6.129127358490567e-06,
      "loss": 0.7337,
      "num_input_tokens_seen": 1366736,
      "step": 2080
    },
    {
      "epoch": 1.2293632075471699,
      "grad_norm": 3.3850133419036865,
      "learning_rate": 6.143867924528303e-06,
      "loss": 0.4757,
      "num_input_tokens_seen": 1369552,
      "step": 2085
    },
    {
      "epoch": 1.232311320754717,
      "grad_norm": 3.4951236248016357,
      "learning_rate": 6.158608490566038e-06,
      "loss": 0.5077,
      "num_input_tokens_seen": 1373008,
      "step": 2090
    },
    {
      "epoch": 1.2352594339622642,
      "grad_norm": 4.011752128601074,
      "learning_rate": 6.173349056603774e-06,
      "loss": 0.5869,
      "num_input_tokens_seen": 1376208,
      "step": 2095
    },
    {
      "epoch": 1.2382075471698113,
      "grad_norm": 30.416345596313477,
      "learning_rate": 6.18808962264151e-06,
      "loss": 0.6187,
      "num_input_tokens_seen": 1379248,
      "step": 2100
    },
    {
      "epoch": 1.2411556603773586,
      "grad_norm": 3.9827237129211426,
      "learning_rate": 6.202830188679245e-06,
      "loss": 0.4919,
      "num_input_tokens_seen": 1382192,
      "step": 2105
    },
    {
      "epoch": 1.2441037735849056,
      "grad_norm": 2.5855607986450195,
      "learning_rate": 6.217570754716982e-06,
      "loss": 0.4038,
      "num_input_tokens_seen": 1385552,
      "step": 2110
    },
    {
      "epoch": 1.2470518867924527,
      "grad_norm": 5.70745849609375,
      "learning_rate": 6.232311320754717e-06,
      "loss": 0.4879,
      "num_input_tokens_seen": 1389072,
      "step": 2115
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.305886268615723,
      "learning_rate": 6.2470518867924536e-06,
      "loss": 0.5443,
      "num_input_tokens_seen": 1392624,
      "step": 2120
    },
    {
      "epoch": 1.2529481132075473,
      "grad_norm": 4.0123291015625,
      "learning_rate": 6.261792452830189e-06,
      "loss": 0.6732,
      "num_input_tokens_seen": 1396880,
      "step": 2125
    },
    {
      "epoch": 1.2558962264150944,
      "grad_norm": 4.65297269821167,
      "learning_rate": 6.276533018867925e-06,
      "loss": 0.5951,
      "num_input_tokens_seen": 1399760,
      "step": 2130
    },
    {
      "epoch": 1.2588443396226414,
      "grad_norm": 2.181225061416626,
      "learning_rate": 6.2912735849056604e-06,
      "loss": 0.4292,
      "num_input_tokens_seen": 1404112,
      "step": 2135
    },
    {
      "epoch": 1.2617924528301887,
      "grad_norm": 3.2651619911193848,
      "learning_rate": 6.306014150943397e-06,
      "loss": 0.6067,
      "num_input_tokens_seen": 1407344,
      "step": 2140
    },
    {
      "epoch": 1.2647405660377358,
      "grad_norm": 10.28250503540039,
      "learning_rate": 6.320754716981132e-06,
      "loss": 0.4583,
      "num_input_tokens_seen": 1409808,
      "step": 2145
    },
    {
      "epoch": 1.267688679245283,
      "grad_norm": 2.498609781265259,
      "learning_rate": 6.335495283018869e-06,
      "loss": 0.4801,
      "num_input_tokens_seen": 1413744,
      "step": 2150
    },
    {
      "epoch": 1.2706367924528301,
      "grad_norm": 4.564564228057861,
      "learning_rate": 6.350235849056604e-06,
      "loss": 0.6338,
      "num_input_tokens_seen": 1416464,
      "step": 2155
    },
    {
      "epoch": 1.2735849056603774,
      "grad_norm": 2.365264654159546,
      "learning_rate": 6.3649764150943406e-06,
      "loss": 0.5422,
      "num_input_tokens_seen": 1419600,
      "step": 2160
    },
    {
      "epoch": 1.2765330188679245,
      "grad_norm": 2.3689560890197754,
      "learning_rate": 6.379716981132076e-06,
      "loss": 0.4637,
      "num_input_tokens_seen": 1422704,
      "step": 2165
    },
    {
      "epoch": 1.2794811320754718,
      "grad_norm": 3.973433256149292,
      "learning_rate": 6.394457547169812e-06,
      "loss": 0.6099,
      "num_input_tokens_seen": 1425328,
      "step": 2170
    },
    {
      "epoch": 1.2824292452830188,
      "grad_norm": 2.4362473487854004,
      "learning_rate": 6.4091981132075474e-06,
      "loss": 0.3928,
      "num_input_tokens_seen": 1428752,
      "step": 2175
    },
    {
      "epoch": 1.2853773584905661,
      "grad_norm": 3.010484457015991,
      "learning_rate": 6.423938679245284e-06,
      "loss": 0.5865,
      "num_input_tokens_seen": 1431728,
      "step": 2180
    },
    {
      "epoch": 1.2883254716981132,
      "grad_norm": 2.554955244064331,
      "learning_rate": 6.438679245283019e-06,
      "loss": 0.5842,
      "num_input_tokens_seen": 1434864,
      "step": 2185
    },
    {
      "epoch": 1.2912735849056602,
      "grad_norm": 4.47111177444458,
      "learning_rate": 6.453419811320756e-06,
      "loss": 0.5603,
      "num_input_tokens_seen": 1437456,
      "step": 2190
    },
    {
      "epoch": 1.2942216981132075,
      "grad_norm": 2.8106753826141357,
      "learning_rate": 6.468160377358491e-06,
      "loss": 0.5184,
      "num_input_tokens_seen": 1440848,
      "step": 2195
    },
    {
      "epoch": 1.2971698113207548,
      "grad_norm": 5.782093048095703,
      "learning_rate": 6.482900943396227e-06,
      "loss": 0.5226,
      "num_input_tokens_seen": 1443536,
      "step": 2200
    },
    {
      "epoch": 1.3001179245283019,
      "grad_norm": 2.9532227516174316,
      "learning_rate": 6.497641509433963e-06,
      "loss": 0.5002,
      "num_input_tokens_seen": 1446480,
      "step": 2205
    },
    {
      "epoch": 1.303066037735849,
      "grad_norm": 2.6872384548187256,
      "learning_rate": 6.512382075471698e-06,
      "loss": 0.4116,
      "num_input_tokens_seen": 1450896,
      "step": 2210
    },
    {
      "epoch": 1.3060141509433962,
      "grad_norm": 3.9611687660217285,
      "learning_rate": 6.5271226415094344e-06,
      "loss": 0.5683,
      "num_input_tokens_seen": 1453840,
      "step": 2215
    },
    {
      "epoch": 1.3089622641509435,
      "grad_norm": 4.857666492462158,
      "learning_rate": 6.54186320754717e-06,
      "loss": 0.4712,
      "num_input_tokens_seen": 1456816,
      "step": 2220
    },
    {
      "epoch": 1.3119103773584906,
      "grad_norm": 3.6266791820526123,
      "learning_rate": 6.556603773584907e-06,
      "loss": 0.4585,
      "num_input_tokens_seen": 1459920,
      "step": 2225
    },
    {
      "epoch": 1.3148584905660377,
      "grad_norm": 3.6602883338928223,
      "learning_rate": 6.571344339622641e-06,
      "loss": 0.558,
      "num_input_tokens_seen": 1462800,
      "step": 2230
    },
    {
      "epoch": 1.317806603773585,
      "grad_norm": 6.574439525604248,
      "learning_rate": 6.586084905660378e-06,
      "loss": 0.475,
      "num_input_tokens_seen": 1466608,
      "step": 2235
    },
    {
      "epoch": 1.320754716981132,
      "grad_norm": 4.640665054321289,
      "learning_rate": 6.600825471698114e-06,
      "loss": 0.4883,
      "num_input_tokens_seen": 1469648,
      "step": 2240
    },
    {
      "epoch": 1.3237028301886793,
      "grad_norm": 2.6839466094970703,
      "learning_rate": 6.61556603773585e-06,
      "loss": 0.4371,
      "num_input_tokens_seen": 1472848,
      "step": 2245
    },
    {
      "epoch": 1.3266509433962264,
      "grad_norm": 2.1694815158843994,
      "learning_rate": 6.630306603773585e-06,
      "loss": 0.3719,
      "num_input_tokens_seen": 1475696,
      "step": 2250
    },
    {
      "epoch": 1.3295990566037736,
      "grad_norm": 2.942765474319458,
      "learning_rate": 6.645047169811321e-06,
      "loss": 0.5327,
      "num_input_tokens_seen": 1478736,
      "step": 2255
    },
    {
      "epoch": 1.3325471698113207,
      "grad_norm": 2.990090847015381,
      "learning_rate": 6.659787735849057e-06,
      "loss": 0.4748,
      "num_input_tokens_seen": 1481232,
      "step": 2260
    },
    {
      "epoch": 1.335495283018868,
      "grad_norm": 5.8356194496154785,
      "learning_rate": 6.674528301886794e-06,
      "loss": 0.5118,
      "num_input_tokens_seen": 1483664,
      "step": 2265
    },
    {
      "epoch": 1.338443396226415,
      "grad_norm": 3.4210171699523926,
      "learning_rate": 6.689268867924529e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 1486480,
      "step": 2270
    },
    {
      "epoch": 1.3413915094339623,
      "grad_norm": 2.584010362625122,
      "learning_rate": 6.704009433962265e-06,
      "loss": 0.528,
      "num_input_tokens_seen": 1490448,
      "step": 2275
    },
    {
      "epoch": 1.3443396226415094,
      "grad_norm": 3.587372064590454,
      "learning_rate": 6.718750000000001e-06,
      "loss": 0.4612,
      "num_input_tokens_seen": 1493776,
      "step": 2280
    },
    {
      "epoch": 1.3472877358490565,
      "grad_norm": 2.4671006202697754,
      "learning_rate": 6.733490566037736e-06,
      "loss": 0.7333,
      "num_input_tokens_seen": 1496464,
      "step": 2285
    },
    {
      "epoch": 1.3502358490566038,
      "grad_norm": 2.1573379039764404,
      "learning_rate": 6.748231132075472e-06,
      "loss": 0.5952,
      "num_input_tokens_seen": 1500496,
      "step": 2290
    },
    {
      "epoch": 1.353183962264151,
      "grad_norm": 4.3435258865356445,
      "learning_rate": 6.7629716981132076e-06,
      "loss": 0.532,
      "num_input_tokens_seen": 1503888,
      "step": 2295
    },
    {
      "epoch": 1.3561320754716981,
      "grad_norm": 4.6127495765686035,
      "learning_rate": 6.777712264150944e-06,
      "loss": 0.5461,
      "num_input_tokens_seen": 1507632,
      "step": 2300
    },
    {
      "epoch": 1.3590801886792452,
      "grad_norm": 2.7768173217773438,
      "learning_rate": 6.792452830188679e-06,
      "loss": 0.4835,
      "num_input_tokens_seen": 1511088,
      "step": 2305
    },
    {
      "epoch": 1.3620283018867925,
      "grad_norm": 2.577155113220215,
      "learning_rate": 6.807193396226416e-06,
      "loss": 0.7306,
      "num_input_tokens_seen": 1517392,
      "step": 2310
    },
    {
      "epoch": 1.3649764150943398,
      "grad_norm": 8.828067779541016,
      "learning_rate": 6.8219339622641515e-06,
      "loss": 0.6151,
      "num_input_tokens_seen": 1520176,
      "step": 2315
    },
    {
      "epoch": 1.3679245283018868,
      "grad_norm": 3.852818489074707,
      "learning_rate": 6.836674528301888e-06,
      "loss": 0.5326,
      "num_input_tokens_seen": 1523088,
      "step": 2320
    },
    {
      "epoch": 1.3708726415094339,
      "grad_norm": 2.6593544483184814,
      "learning_rate": 6.851415094339623e-06,
      "loss": 0.4285,
      "num_input_tokens_seen": 1526736,
      "step": 2325
    },
    {
      "epoch": 1.3738207547169812,
      "grad_norm": 4.173259258270264,
      "learning_rate": 6.866155660377359e-06,
      "loss": 0.492,
      "num_input_tokens_seen": 1529488,
      "step": 2330
    },
    {
      "epoch": 1.3767688679245282,
      "grad_norm": 2.010636329650879,
      "learning_rate": 6.8808962264150946e-06,
      "loss": 0.3746,
      "num_input_tokens_seen": 1533616,
      "step": 2335
    },
    {
      "epoch": 1.3797169811320755,
      "grad_norm": 2.525909423828125,
      "learning_rate": 6.895636792452831e-06,
      "loss": 0.536,
      "num_input_tokens_seen": 1536432,
      "step": 2340
    },
    {
      "epoch": 1.3826650943396226,
      "grad_norm": 3.1138224601745605,
      "learning_rate": 6.910377358490566e-06,
      "loss": 0.3871,
      "num_input_tokens_seen": 1540432,
      "step": 2345
    },
    {
      "epoch": 1.3856132075471699,
      "grad_norm": 2.8702566623687744,
      "learning_rate": 6.925117924528303e-06,
      "loss": 0.7117,
      "num_input_tokens_seen": 1547184,
      "step": 2350
    },
    {
      "epoch": 1.388561320754717,
      "grad_norm": 2.8041229248046875,
      "learning_rate": 6.9398584905660385e-06,
      "loss": 0.5208,
      "num_input_tokens_seen": 1549872,
      "step": 2355
    },
    {
      "epoch": 1.3915094339622642,
      "grad_norm": 2.09570050239563,
      "learning_rate": 6.954599056603775e-06,
      "loss": 0.4235,
      "num_input_tokens_seen": 1553200,
      "step": 2360
    },
    {
      "epoch": 1.3944575471698113,
      "grad_norm": 3.4030659198760986,
      "learning_rate": 6.96933962264151e-06,
      "loss": 0.5073,
      "num_input_tokens_seen": 1555920,
      "step": 2365
    },
    {
      "epoch": 1.3974056603773586,
      "grad_norm": 4.678151607513428,
      "learning_rate": 6.984080188679245e-06,
      "loss": 0.5834,
      "num_input_tokens_seen": 1558736,
      "step": 2370
    },
    {
      "epoch": 1.4003537735849056,
      "grad_norm": 8.044450759887695,
      "learning_rate": 6.9988207547169815e-06,
      "loss": 0.4879,
      "num_input_tokens_seen": 1561968,
      "step": 2375
    },
    {
      "epoch": 1.4033018867924527,
      "grad_norm": 1.3745454549789429,
      "learning_rate": 7.013561320754717e-06,
      "loss": 0.4974,
      "num_input_tokens_seen": 1566096,
      "step": 2380
    },
    {
      "epoch": 1.40625,
      "grad_norm": 3.3225924968719482,
      "learning_rate": 7.028301886792454e-06,
      "loss": 0.5475,
      "num_input_tokens_seen": 1569744,
      "step": 2385
    },
    {
      "epoch": 1.4091981132075473,
      "grad_norm": 2.6871540546417236,
      "learning_rate": 7.043042452830188e-06,
      "loss": 0.5431,
      "num_input_tokens_seen": 1572656,
      "step": 2390
    },
    {
      "epoch": 1.4121462264150944,
      "grad_norm": 4.825122833251953,
      "learning_rate": 7.0577830188679255e-06,
      "loss": 0.4855,
      "num_input_tokens_seen": 1575664,
      "step": 2395
    },
    {
      "epoch": 1.4150943396226414,
      "grad_norm": 6.13507080078125,
      "learning_rate": 7.072523584905661e-06,
      "loss": 0.4392,
      "num_input_tokens_seen": 1579664,
      "step": 2400
    },
    {
      "epoch": 1.4180424528301887,
      "grad_norm": 2.3485279083251953,
      "learning_rate": 7.087264150943397e-06,
      "loss": 0.5357,
      "num_input_tokens_seen": 1583664,
      "step": 2405
    },
    {
      "epoch": 1.4209905660377358,
      "grad_norm": 3.297713279724121,
      "learning_rate": 7.102004716981132e-06,
      "loss": 0.4366,
      "num_input_tokens_seen": 1586320,
      "step": 2410
    },
    {
      "epoch": 1.423938679245283,
      "grad_norm": 1.8257416486740112,
      "learning_rate": 7.1167452830188685e-06,
      "loss": 0.3772,
      "num_input_tokens_seen": 1589456,
      "step": 2415
    },
    {
      "epoch": 1.4268867924528301,
      "grad_norm": 4.513745307922363,
      "learning_rate": 7.131485849056604e-06,
      "loss": 0.6385,
      "num_input_tokens_seen": 1592240,
      "step": 2420
    },
    {
      "epoch": 1.4298349056603774,
      "grad_norm": 2.3152623176574707,
      "learning_rate": 7.146226415094341e-06,
      "loss": 0.7068,
      "num_input_tokens_seen": 1594512,
      "step": 2425
    },
    {
      "epoch": 1.4327830188679245,
      "grad_norm": 2.936514377593994,
      "learning_rate": 7.160966981132076e-06,
      "loss": 0.5552,
      "num_input_tokens_seen": 1598032,
      "step": 2430
    },
    {
      "epoch": 1.4357311320754718,
      "grad_norm": 2.658808946609497,
      "learning_rate": 7.1757075471698125e-06,
      "loss": 0.408,
      "num_input_tokens_seen": 1601584,
      "step": 2435
    },
    {
      "epoch": 1.4386792452830188,
      "grad_norm": 2.593050718307495,
      "learning_rate": 7.190448113207548e-06,
      "loss": 0.5061,
      "num_input_tokens_seen": 1603888,
      "step": 2440
    },
    {
      "epoch": 1.4416273584905661,
      "grad_norm": 3.975923776626587,
      "learning_rate": 7.205188679245284e-06,
      "loss": 0.4003,
      "num_input_tokens_seen": 1610064,
      "step": 2445
    },
    {
      "epoch": 1.4445754716981132,
      "grad_norm": 3.7889389991760254,
      "learning_rate": 7.219929245283019e-06,
      "loss": 0.5983,
      "num_input_tokens_seen": 1613648,
      "step": 2450
    },
    {
      "epoch": 1.4475235849056602,
      "grad_norm": 4.751260757446289,
      "learning_rate": 7.2346698113207555e-06,
      "loss": 0.5458,
      "num_input_tokens_seen": 1617712,
      "step": 2455
    },
    {
      "epoch": 1.4504716981132075,
      "grad_norm": 2.4832839965820312,
      "learning_rate": 7.249410377358491e-06,
      "loss": 0.4777,
      "num_input_tokens_seen": 1621360,
      "step": 2460
    },
    {
      "epoch": 1.4534198113207548,
      "grad_norm": 6.365638732910156,
      "learning_rate": 7.264150943396226e-06,
      "loss": 0.5654,
      "num_input_tokens_seen": 1624592,
      "step": 2465
    },
    {
      "epoch": 1.4563679245283019,
      "grad_norm": 3.6393473148345947,
      "learning_rate": 7.278891509433963e-06,
      "loss": 0.3454,
      "num_input_tokens_seen": 1627536,
      "step": 2470
    },
    {
      "epoch": 1.459316037735849,
      "grad_norm": 2.9554598331451416,
      "learning_rate": 7.293632075471699e-06,
      "loss": 0.6499,
      "num_input_tokens_seen": 1630864,
      "step": 2475
    },
    {
      "epoch": 1.4622641509433962,
      "grad_norm": 2.5732383728027344,
      "learning_rate": 7.308372641509435e-06,
      "loss": 0.5216,
      "num_input_tokens_seen": 1637840,
      "step": 2480
    },
    {
      "epoch": 1.4652122641509435,
      "grad_norm": 3.4576008319854736,
      "learning_rate": 7.32311320754717e-06,
      "loss": 0.6997,
      "num_input_tokens_seen": 1640240,
      "step": 2485
    },
    {
      "epoch": 1.4681603773584906,
      "grad_norm": 3.3271079063415527,
      "learning_rate": 7.337853773584906e-06,
      "loss": 0.6019,
      "num_input_tokens_seen": 1643760,
      "step": 2490
    },
    {
      "epoch": 1.4711084905660377,
      "grad_norm": 3.1961793899536133,
      "learning_rate": 7.352594339622642e-06,
      "loss": 0.6693,
      "num_input_tokens_seen": 1647312,
      "step": 2495
    },
    {
      "epoch": 1.474056603773585,
      "grad_norm": 5.841244220733643,
      "learning_rate": 7.367334905660378e-06,
      "loss": 0.5773,
      "num_input_tokens_seen": 1652368,
      "step": 2500
    },
    {
      "epoch": 1.477004716981132,
      "grad_norm": 2.1236982345581055,
      "learning_rate": 7.382075471698113e-06,
      "loss": 0.5642,
      "num_input_tokens_seen": 1656944,
      "step": 2505
    },
    {
      "epoch": 1.4799528301886793,
      "grad_norm": 4.892528057098389,
      "learning_rate": 7.39681603773585e-06,
      "loss": 0.6037,
      "num_input_tokens_seen": 1659696,
      "step": 2510
    },
    {
      "epoch": 1.4829009433962264,
      "grad_norm": 3.8570189476013184,
      "learning_rate": 7.411556603773586e-06,
      "loss": 0.4081,
      "num_input_tokens_seen": 1662032,
      "step": 2515
    },
    {
      "epoch": 1.4858490566037736,
      "grad_norm": 2.083744764328003,
      "learning_rate": 7.426297169811322e-06,
      "loss": 0.465,
      "num_input_tokens_seen": 1668880,
      "step": 2520
    },
    {
      "epoch": 1.4887971698113207,
      "grad_norm": 1.6628202199935913,
      "learning_rate": 7.441037735849057e-06,
      "loss": 0.5302,
      "num_input_tokens_seen": 1671760,
      "step": 2525
    },
    {
      "epoch": 1.491745283018868,
      "grad_norm": 3.0010437965393066,
      "learning_rate": 7.455778301886793e-06,
      "loss": 0.6271,
      "num_input_tokens_seen": 1674704,
      "step": 2530
    },
    {
      "epoch": 1.494693396226415,
      "grad_norm": 3.750641345977783,
      "learning_rate": 7.470518867924529e-06,
      "loss": 0.4941,
      "num_input_tokens_seen": 1678768,
      "step": 2535
    },
    {
      "epoch": 1.4976415094339623,
      "grad_norm": 2.1739799976348877,
      "learning_rate": 7.485259433962266e-06,
      "loss": 0.6335,
      "num_input_tokens_seen": 1682672,
      "step": 2540
    },
    {
      "epoch": 1.5005896226415094,
      "grad_norm": 6.452185153961182,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.5537,
      "num_input_tokens_seen": 1685136,
      "step": 2545
    },
    {
      "epoch": 1.5035377358490565,
      "grad_norm": 2.450366258621216,
      "learning_rate": 7.5147405660377355e-06,
      "loss": 0.5876,
      "num_input_tokens_seen": 1687952,
      "step": 2550
    },
    {
      "epoch": 1.5064858490566038,
      "grad_norm": 2.1167068481445312,
      "learning_rate": 7.5294811320754726e-06,
      "loss": 0.6532,
      "num_input_tokens_seen": 1691408,
      "step": 2555
    },
    {
      "epoch": 1.509433962264151,
      "grad_norm": 2.3418920040130615,
      "learning_rate": 7.544221698113208e-06,
      "loss": 0.4903,
      "num_input_tokens_seen": 1694544,
      "step": 2560
    },
    {
      "epoch": 1.5123820754716981,
      "grad_norm": 8.44914722442627,
      "learning_rate": 7.558962264150944e-06,
      "loss": 0.6899,
      "num_input_tokens_seen": 1697872,
      "step": 2565
    },
    {
      "epoch": 1.5153301886792452,
      "grad_norm": 7.571895122528076,
      "learning_rate": 7.5737028301886795e-06,
      "loss": 0.4313,
      "num_input_tokens_seen": 1701168,
      "step": 2570
    },
    {
      "epoch": 1.5182783018867925,
      "grad_norm": 2.822213888168335,
      "learning_rate": 7.588443396226416e-06,
      "loss": 0.3597,
      "num_input_tokens_seen": 1704624,
      "step": 2575
    },
    {
      "epoch": 1.5212264150943398,
      "grad_norm": 4.7039008140563965,
      "learning_rate": 7.603183962264151e-06,
      "loss": 0.7222,
      "num_input_tokens_seen": 1707216,
      "step": 2580
    },
    {
      "epoch": 1.5241745283018868,
      "grad_norm": 3.9861202239990234,
      "learning_rate": 7.617924528301888e-06,
      "loss": 0.5891,
      "num_input_tokens_seen": 1710256,
      "step": 2585
    },
    {
      "epoch": 1.5271226415094339,
      "grad_norm": 2.801581621170044,
      "learning_rate": 7.632665094339623e-06,
      "loss": 0.4943,
      "num_input_tokens_seen": 1713168,
      "step": 2590
    },
    {
      "epoch": 1.5300707547169812,
      "grad_norm": 2.5071072578430176,
      "learning_rate": 7.64740566037736e-06,
      "loss": 0.5818,
      "num_input_tokens_seen": 1717584,
      "step": 2595
    },
    {
      "epoch": 1.5330188679245285,
      "grad_norm": 3.347670555114746,
      "learning_rate": 7.662146226415095e-06,
      "loss": 0.5867,
      "num_input_tokens_seen": 1721008,
      "step": 2600
    },
    {
      "epoch": 1.5359669811320755,
      "grad_norm": 2.055980682373047,
      "learning_rate": 7.676886792452832e-06,
      "loss": 0.5149,
      "num_input_tokens_seen": 1723824,
      "step": 2605
    },
    {
      "epoch": 1.5389150943396226,
      "grad_norm": 2.764608860015869,
      "learning_rate": 7.691627358490567e-06,
      "loss": 0.5639,
      "num_input_tokens_seen": 1726704,
      "step": 2610
    },
    {
      "epoch": 1.5418632075471699,
      "grad_norm": 2.548527240753174,
      "learning_rate": 7.706367924528303e-06,
      "loss": 0.5172,
      "num_input_tokens_seen": 1728944,
      "step": 2615
    },
    {
      "epoch": 1.544811320754717,
      "grad_norm": 3.7490572929382324,
      "learning_rate": 7.721108490566038e-06,
      "loss": 0.5789,
      "num_input_tokens_seen": 1732656,
      "step": 2620
    },
    {
      "epoch": 1.547759433962264,
      "grad_norm": 3.654243230819702,
      "learning_rate": 7.735849056603775e-06,
      "loss": 0.5201,
      "num_input_tokens_seen": 1736304,
      "step": 2625
    },
    {
      "epoch": 1.5507075471698113,
      "grad_norm": 2.4849588871002197,
      "learning_rate": 7.75058962264151e-06,
      "loss": 0.4645,
      "num_input_tokens_seen": 1738800,
      "step": 2630
    },
    {
      "epoch": 1.5536556603773586,
      "grad_norm": 1.584727168083191,
      "learning_rate": 7.765330188679246e-06,
      "loss": 0.6032,
      "num_input_tokens_seen": 1742864,
      "step": 2635
    },
    {
      "epoch": 1.5566037735849056,
      "grad_norm": 2.853015184402466,
      "learning_rate": 7.780070754716981e-06,
      "loss": 0.6604,
      "num_input_tokens_seen": 1745712,
      "step": 2640
    },
    {
      "epoch": 1.5595518867924527,
      "grad_norm": 1.5749913454055786,
      "learning_rate": 7.794811320754716e-06,
      "loss": 0.498,
      "num_input_tokens_seen": 1748848,
      "step": 2645
    },
    {
      "epoch": 1.5625,
      "grad_norm": 2.302140235900879,
      "learning_rate": 7.809551886792453e-06,
      "loss": 0.6063,
      "num_input_tokens_seen": 1751632,
      "step": 2650
    },
    {
      "epoch": 1.5654481132075473,
      "grad_norm": 2.8924572467803955,
      "learning_rate": 7.824292452830189e-06,
      "loss": 0.5786,
      "num_input_tokens_seen": 1754256,
      "step": 2655
    },
    {
      "epoch": 1.5683962264150944,
      "grad_norm": 2.7571394443511963,
      "learning_rate": 7.839033018867926e-06,
      "loss": 0.4408,
      "num_input_tokens_seen": 1758256,
      "step": 2660
    },
    {
      "epoch": 1.5713443396226414,
      "grad_norm": 2.3545072078704834,
      "learning_rate": 7.853773584905661e-06,
      "loss": 0.449,
      "num_input_tokens_seen": 1761104,
      "step": 2665
    },
    {
      "epoch": 1.5742924528301887,
      "grad_norm": 5.708223819732666,
      "learning_rate": 7.868514150943397e-06,
      "loss": 0.6445,
      "num_input_tokens_seen": 1764368,
      "step": 2670
    },
    {
      "epoch": 1.577240566037736,
      "grad_norm": 5.253635883331299,
      "learning_rate": 7.883254716981132e-06,
      "loss": 0.6282,
      "num_input_tokens_seen": 1767600,
      "step": 2675
    },
    {
      "epoch": 1.580188679245283,
      "grad_norm": 4.910017013549805,
      "learning_rate": 7.897995283018869e-06,
      "loss": 0.5946,
      "num_input_tokens_seen": 1770704,
      "step": 2680
    },
    {
      "epoch": 1.5831367924528301,
      "grad_norm": 2.2701804637908936,
      "learning_rate": 7.912735849056604e-06,
      "loss": 0.4877,
      "num_input_tokens_seen": 1774192,
      "step": 2685
    },
    {
      "epoch": 1.5860849056603774,
      "grad_norm": 2.352328062057495,
      "learning_rate": 7.927476415094341e-06,
      "loss": 0.5043,
      "num_input_tokens_seen": 1777104,
      "step": 2690
    },
    {
      "epoch": 1.5890330188679245,
      "grad_norm": 1.751832365989685,
      "learning_rate": 7.942216981132077e-06,
      "loss": 0.4547,
      "num_input_tokens_seen": 1781040,
      "step": 2695
    },
    {
      "epoch": 1.5919811320754715,
      "grad_norm": 2.155081272125244,
      "learning_rate": 7.956957547169812e-06,
      "loss": 0.4374,
      "num_input_tokens_seen": 1784240,
      "step": 2700
    },
    {
      "epoch": 1.5949292452830188,
      "grad_norm": 1.4876408576965332,
      "learning_rate": 7.971698113207547e-06,
      "loss": 0.5109,
      "num_input_tokens_seen": 1788016,
      "step": 2705
    },
    {
      "epoch": 1.5978773584905661,
      "grad_norm": 3.913264513015747,
      "learning_rate": 7.986438679245284e-06,
      "loss": 0.6303,
      "num_input_tokens_seen": 1791280,
      "step": 2710
    },
    {
      "epoch": 1.6008254716981132,
      "grad_norm": 3.5021796226501465,
      "learning_rate": 8.00117924528302e-06,
      "loss": 0.4833,
      "num_input_tokens_seen": 1794384,
      "step": 2715
    },
    {
      "epoch": 1.6037735849056602,
      "grad_norm": 4.358457088470459,
      "learning_rate": 8.015919811320757e-06,
      "loss": 0.7109,
      "num_input_tokens_seen": 1797872,
      "step": 2720
    },
    {
      "epoch": 1.6067216981132075,
      "grad_norm": 4.717020511627197,
      "learning_rate": 8.030660377358492e-06,
      "loss": 0.5164,
      "num_input_tokens_seen": 1800976,
      "step": 2725
    },
    {
      "epoch": 1.6096698113207548,
      "grad_norm": 7.1886396408081055,
      "learning_rate": 8.045400943396227e-06,
      "loss": 0.4662,
      "num_input_tokens_seen": 1804368,
      "step": 2730
    },
    {
      "epoch": 1.6126179245283019,
      "grad_norm": 3.3873188495635986,
      "learning_rate": 8.060141509433963e-06,
      "loss": 0.4143,
      "num_input_tokens_seen": 1807856,
      "step": 2735
    },
    {
      "epoch": 1.615566037735849,
      "grad_norm": 1.8609068393707275,
      "learning_rate": 8.074882075471698e-06,
      "loss": 0.5368,
      "num_input_tokens_seen": 1810896,
      "step": 2740
    },
    {
      "epoch": 1.6185141509433962,
      "grad_norm": 4.164821624755859,
      "learning_rate": 8.089622641509435e-06,
      "loss": 0.4909,
      "num_input_tokens_seen": 1814864,
      "step": 2745
    },
    {
      "epoch": 1.6214622641509435,
      "grad_norm": 4.435966968536377,
      "learning_rate": 8.10436320754717e-06,
      "loss": 0.5272,
      "num_input_tokens_seen": 1818544,
      "step": 2750
    },
    {
      "epoch": 1.6244103773584906,
      "grad_norm": 2.6753389835357666,
      "learning_rate": 8.119103773584906e-06,
      "loss": 0.6445,
      "num_input_tokens_seen": 1821296,
      "step": 2755
    },
    {
      "epoch": 1.6273584905660377,
      "grad_norm": 8.43822956085205,
      "learning_rate": 8.133844339622641e-06,
      "loss": 0.4564,
      "num_input_tokens_seen": 1823984,
      "step": 2760
    },
    {
      "epoch": 1.630306603773585,
      "grad_norm": 4.250800132751465,
      "learning_rate": 8.148584905660378e-06,
      "loss": 0.6902,
      "num_input_tokens_seen": 1826384,
      "step": 2765
    },
    {
      "epoch": 1.6332547169811322,
      "grad_norm": 3.2443556785583496,
      "learning_rate": 8.163325471698114e-06,
      "loss": 0.5771,
      "num_input_tokens_seen": 1830768,
      "step": 2770
    },
    {
      "epoch": 1.6362028301886793,
      "grad_norm": 4.230912208557129,
      "learning_rate": 8.17806603773585e-06,
      "loss": 0.5018,
      "num_input_tokens_seen": 1833168,
      "step": 2775
    },
    {
      "epoch": 1.6391509433962264,
      "grad_norm": 2.6462976932525635,
      "learning_rate": 8.192806603773586e-06,
      "loss": 0.5226,
      "num_input_tokens_seen": 1835664,
      "step": 2780
    },
    {
      "epoch": 1.6420990566037736,
      "grad_norm": 14.54749584197998,
      "learning_rate": 8.207547169811321e-06,
      "loss": 0.6979,
      "num_input_tokens_seen": 1839440,
      "step": 2785
    },
    {
      "epoch": 1.6450471698113207,
      "grad_norm": 3.943333864212036,
      "learning_rate": 8.222287735849057e-06,
      "loss": 0.7172,
      "num_input_tokens_seen": 1842800,
      "step": 2790
    },
    {
      "epoch": 1.6479952830188678,
      "grad_norm": 2.6584460735321045,
      "learning_rate": 8.237028301886794e-06,
      "loss": 0.4863,
      "num_input_tokens_seen": 1845872,
      "step": 2795
    },
    {
      "epoch": 1.650943396226415,
      "grad_norm": 3.1197404861450195,
      "learning_rate": 8.251768867924529e-06,
      "loss": 0.5963,
      "num_input_tokens_seen": 1848688,
      "step": 2800
    },
    {
      "epoch": 1.6538915094339623,
      "grad_norm": 4.877439975738525,
      "learning_rate": 8.266509433962266e-06,
      "loss": 0.5444,
      "num_input_tokens_seen": 1852784,
      "step": 2805
    },
    {
      "epoch": 1.6568396226415094,
      "grad_norm": 2.2218098640441895,
      "learning_rate": 8.281250000000001e-06,
      "loss": 0.4602,
      "num_input_tokens_seen": 1855568,
      "step": 2810
    },
    {
      "epoch": 1.6597877358490565,
      "grad_norm": 4.642866134643555,
      "learning_rate": 8.295990566037737e-06,
      "loss": 0.6375,
      "num_input_tokens_seen": 1859216,
      "step": 2815
    },
    {
      "epoch": 1.6627358490566038,
      "grad_norm": 2.14306902885437,
      "learning_rate": 8.310731132075472e-06,
      "loss": 0.6615,
      "num_input_tokens_seen": 1862832,
      "step": 2820
    },
    {
      "epoch": 1.665683962264151,
      "grad_norm": 5.428103446960449,
      "learning_rate": 8.325471698113207e-06,
      "loss": 0.5801,
      "num_input_tokens_seen": 1865840,
      "step": 2825
    },
    {
      "epoch": 1.6686320754716981,
      "grad_norm": 4.68764066696167,
      "learning_rate": 8.340212264150944e-06,
      "loss": 0.4777,
      "num_input_tokens_seen": 1869328,
      "step": 2830
    },
    {
      "epoch": 1.6715801886792452,
      "grad_norm": 3.1852357387542725,
      "learning_rate": 8.35495283018868e-06,
      "loss": 0.5575,
      "num_input_tokens_seen": 1874288,
      "step": 2835
    },
    {
      "epoch": 1.6745283018867925,
      "grad_norm": 3.0576438903808594,
      "learning_rate": 8.369693396226415e-06,
      "loss": 0.5495,
      "num_input_tokens_seen": 1876784,
      "step": 2840
    },
    {
      "epoch": 1.6774764150943398,
      "grad_norm": 2.2294256687164307,
      "learning_rate": 8.38443396226415e-06,
      "loss": 0.5281,
      "num_input_tokens_seen": 1879504,
      "step": 2845
    },
    {
      "epoch": 1.6804245283018868,
      "grad_norm": 4.603205680847168,
      "learning_rate": 8.399174528301888e-06,
      "loss": 0.4791,
      "num_input_tokens_seen": 1884016,
      "step": 2850
    },
    {
      "epoch": 1.6833726415094339,
      "grad_norm": 3.1160526275634766,
      "learning_rate": 8.413915094339623e-06,
      "loss": 0.5521,
      "num_input_tokens_seen": 1887568,
      "step": 2855
    },
    {
      "epoch": 1.6863207547169812,
      "grad_norm": 5.0818023681640625,
      "learning_rate": 8.42865566037736e-06,
      "loss": 0.6056,
      "num_input_tokens_seen": 1890736,
      "step": 2860
    },
    {
      "epoch": 1.6892688679245285,
      "grad_norm": 3.5554003715515137,
      "learning_rate": 8.443396226415095e-06,
      "loss": 0.5741,
      "num_input_tokens_seen": 1894384,
      "step": 2865
    },
    {
      "epoch": 1.6922169811320755,
      "grad_norm": 2.7747068405151367,
      "learning_rate": 8.45813679245283e-06,
      "loss": 0.4658,
      "num_input_tokens_seen": 1897456,
      "step": 2870
    },
    {
      "epoch": 1.6951650943396226,
      "grad_norm": 2.2472503185272217,
      "learning_rate": 8.472877358490566e-06,
      "loss": 0.5902,
      "num_input_tokens_seen": 1900624,
      "step": 2875
    },
    {
      "epoch": 1.6981132075471699,
      "grad_norm": 6.759567737579346,
      "learning_rate": 8.487617924528303e-06,
      "loss": 0.4154,
      "num_input_tokens_seen": 1903600,
      "step": 2880
    },
    {
      "epoch": 1.701061320754717,
      "grad_norm": 2.111589193344116,
      "learning_rate": 8.502358490566038e-06,
      "loss": 0.6079,
      "num_input_tokens_seen": 1906544,
      "step": 2885
    },
    {
      "epoch": 1.704009433962264,
      "grad_norm": 2.7130184173583984,
      "learning_rate": 8.517099056603775e-06,
      "loss": 0.6543,
      "num_input_tokens_seen": 1909744,
      "step": 2890
    },
    {
      "epoch": 1.7069575471698113,
      "grad_norm": 1.4910885095596313,
      "learning_rate": 8.53183962264151e-06,
      "loss": 0.5449,
      "num_input_tokens_seen": 1913616,
      "step": 2895
    },
    {
      "epoch": 1.7099056603773586,
      "grad_norm": 4.704848289489746,
      "learning_rate": 8.546580188679246e-06,
      "loss": 0.4876,
      "num_input_tokens_seen": 1916560,
      "step": 2900
    },
    {
      "epoch": 1.7128537735849056,
      "grad_norm": 2.1430678367614746,
      "learning_rate": 8.561320754716981e-06,
      "loss": 0.7924,
      "num_input_tokens_seen": 1919088,
      "step": 2905
    },
    {
      "epoch": 1.7158018867924527,
      "grad_norm": 3.0918450355529785,
      "learning_rate": 8.576061320754717e-06,
      "loss": 0.7876,
      "num_input_tokens_seen": 1922640,
      "step": 2910
    },
    {
      "epoch": 1.71875,
      "grad_norm": 3.782632827758789,
      "learning_rate": 8.590801886792454e-06,
      "loss": 0.4375,
      "num_input_tokens_seen": 1925392,
      "step": 2915
    },
    {
      "epoch": 1.7216981132075473,
      "grad_norm": 3.000537395477295,
      "learning_rate": 8.605542452830189e-06,
      "loss": 0.7063,
      "num_input_tokens_seen": 1928464,
      "step": 2920
    },
    {
      "epoch": 1.7246462264150944,
      "grad_norm": 3.8343188762664795,
      "learning_rate": 8.620283018867926e-06,
      "loss": 0.6664,
      "num_input_tokens_seen": 1931600,
      "step": 2925
    },
    {
      "epoch": 1.7275943396226414,
      "grad_norm": 1.4098401069641113,
      "learning_rate": 8.635023584905662e-06,
      "loss": 0.6571,
      "num_input_tokens_seen": 1934864,
      "step": 2930
    },
    {
      "epoch": 1.7305424528301887,
      "grad_norm": 2.0239765644073486,
      "learning_rate": 8.649764150943397e-06,
      "loss": 0.5229,
      "num_input_tokens_seen": 1938096,
      "step": 2935
    },
    {
      "epoch": 1.733490566037736,
      "grad_norm": 1.8730028867721558,
      "learning_rate": 8.664504716981132e-06,
      "loss": 0.5641,
      "num_input_tokens_seen": 1941328,
      "step": 2940
    },
    {
      "epoch": 1.736438679245283,
      "grad_norm": 2.6406800746917725,
      "learning_rate": 8.67924528301887e-06,
      "loss": 0.6303,
      "num_input_tokens_seen": 1944976,
      "step": 2945
    },
    {
      "epoch": 1.7393867924528301,
      "grad_norm": 4.340983867645264,
      "learning_rate": 8.693985849056605e-06,
      "loss": 0.4712,
      "num_input_tokens_seen": 1947376,
      "step": 2950
    },
    {
      "epoch": 1.7423349056603774,
      "grad_norm": 2.405297040939331,
      "learning_rate": 8.70872641509434e-06,
      "loss": 0.4929,
      "num_input_tokens_seen": 1950576,
      "step": 2955
    },
    {
      "epoch": 1.7452830188679245,
      "grad_norm": 3.072497844696045,
      "learning_rate": 8.723466981132075e-06,
      "loss": 0.5411,
      "num_input_tokens_seen": 1954448,
      "step": 2960
    },
    {
      "epoch": 1.7482311320754715,
      "grad_norm": 1.56428062915802,
      "learning_rate": 8.738207547169812e-06,
      "loss": 0.5475,
      "num_input_tokens_seen": 1957968,
      "step": 2965
    },
    {
      "epoch": 1.7511792452830188,
      "grad_norm": 3.3503153324127197,
      "learning_rate": 8.752948113207548e-06,
      "loss": 0.5321,
      "num_input_tokens_seen": 1960176,
      "step": 2970
    },
    {
      "epoch": 1.7541273584905661,
      "grad_norm": 2.014655828475952,
      "learning_rate": 8.767688679245285e-06,
      "loss": 0.5657,
      "num_input_tokens_seen": 1964048,
      "step": 2975
    },
    {
      "epoch": 1.7570754716981132,
      "grad_norm": 4.355897426605225,
      "learning_rate": 8.78242924528302e-06,
      "loss": 0.604,
      "num_input_tokens_seen": 1967568,
      "step": 2980
    },
    {
      "epoch": 1.7600235849056602,
      "grad_norm": 3.4178993701934814,
      "learning_rate": 8.797169811320755e-06,
      "loss": 0.5514,
      "num_input_tokens_seen": 1970448,
      "step": 2985
    },
    {
      "epoch": 1.7629716981132075,
      "grad_norm": 1.568340539932251,
      "learning_rate": 8.81191037735849e-06,
      "loss": 0.4758,
      "num_input_tokens_seen": 1974160,
      "step": 2990
    },
    {
      "epoch": 1.7659198113207548,
      "grad_norm": 2.496394395828247,
      "learning_rate": 8.826650943396226e-06,
      "loss": 0.7704,
      "num_input_tokens_seen": 1977264,
      "step": 2995
    },
    {
      "epoch": 1.7688679245283019,
      "grad_norm": 3.024653434753418,
      "learning_rate": 8.841391509433963e-06,
      "loss": 0.5609,
      "num_input_tokens_seen": 1980880,
      "step": 3000
    },
    {
      "epoch": 1.771816037735849,
      "grad_norm": 4.412405490875244,
      "learning_rate": 8.856132075471698e-06,
      "loss": 0.5336,
      "num_input_tokens_seen": 1983440,
      "step": 3005
    },
    {
      "epoch": 1.7747641509433962,
      "grad_norm": 1.6027352809906006,
      "learning_rate": 8.870872641509435e-06,
      "loss": 0.4548,
      "num_input_tokens_seen": 1986960,
      "step": 3010
    },
    {
      "epoch": 1.7777122641509435,
      "grad_norm": 2.797729969024658,
      "learning_rate": 8.88561320754717e-06,
      "loss": 0.5591,
      "num_input_tokens_seen": 1989968,
      "step": 3015
    },
    {
      "epoch": 1.7806603773584906,
      "grad_norm": 2.425788164138794,
      "learning_rate": 8.900353773584906e-06,
      "loss": 0.733,
      "num_input_tokens_seen": 1992912,
      "step": 3020
    },
    {
      "epoch": 1.7836084905660377,
      "grad_norm": 2.2217931747436523,
      "learning_rate": 8.915094339622642e-06,
      "loss": 0.5463,
      "num_input_tokens_seen": 1995856,
      "step": 3025
    },
    {
      "epoch": 1.786556603773585,
      "grad_norm": 3.9583733081817627,
      "learning_rate": 8.929834905660379e-06,
      "loss": 0.5287,
      "num_input_tokens_seen": 1998928,
      "step": 3030
    },
    {
      "epoch": 1.7895047169811322,
      "grad_norm": 2.6888885498046875,
      "learning_rate": 8.944575471698114e-06,
      "loss": 0.6037,
      "num_input_tokens_seen": 2001776,
      "step": 3035
    },
    {
      "epoch": 1.7924528301886793,
      "grad_norm": 2.0285441875457764,
      "learning_rate": 8.959316037735851e-06,
      "loss": 0.6202,
      "num_input_tokens_seen": 2005104,
      "step": 3040
    },
    {
      "epoch": 1.7954009433962264,
      "grad_norm": 2.2529139518737793,
      "learning_rate": 8.974056603773586e-06,
      "loss": 0.4392,
      "num_input_tokens_seen": 2009296,
      "step": 3045
    },
    {
      "epoch": 1.7983490566037736,
      "grad_norm": 1.4532564878463745,
      "learning_rate": 8.988797169811322e-06,
      "loss": 0.5464,
      "num_input_tokens_seen": 2013840,
      "step": 3050
    },
    {
      "epoch": 1.8012971698113207,
      "grad_norm": 2.448000431060791,
      "learning_rate": 9.003537735849057e-06,
      "loss": 0.4425,
      "num_input_tokens_seen": 2017392,
      "step": 3055
    },
    {
      "epoch": 1.8042452830188678,
      "grad_norm": 2.4874463081359863,
      "learning_rate": 9.018278301886794e-06,
      "loss": 0.6254,
      "num_input_tokens_seen": 2020368,
      "step": 3060
    },
    {
      "epoch": 1.807193396226415,
      "grad_norm": 3.264829635620117,
      "learning_rate": 9.03301886792453e-06,
      "loss": 0.5859,
      "num_input_tokens_seen": 2023760,
      "step": 3065
    },
    {
      "epoch": 1.8101415094339623,
      "grad_norm": 3.4850151538848877,
      "learning_rate": 9.047759433962265e-06,
      "loss": 0.4831,
      "num_input_tokens_seen": 2026288,
      "step": 3070
    },
    {
      "epoch": 1.8130896226415094,
      "grad_norm": 2.919337511062622,
      "learning_rate": 9.0625e-06,
      "loss": 0.5683,
      "num_input_tokens_seen": 2028720,
      "step": 3075
    },
    {
      "epoch": 1.8160377358490565,
      "grad_norm": 2.1497690677642822,
      "learning_rate": 9.077240566037735e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 2033520,
      "step": 3080
    },
    {
      "epoch": 1.8189858490566038,
      "grad_norm": 2.0660948753356934,
      "learning_rate": 9.091981132075472e-06,
      "loss": 0.5524,
      "num_input_tokens_seen": 2038320,
      "step": 3085
    },
    {
      "epoch": 1.821933962264151,
      "grad_norm": 1.7834588289260864,
      "learning_rate": 9.106721698113208e-06,
      "loss": 0.5314,
      "num_input_tokens_seen": 2041008,
      "step": 3090
    },
    {
      "epoch": 1.8248820754716981,
      "grad_norm": 1.9613697528839111,
      "learning_rate": 9.121462264150945e-06,
      "loss": 0.5212,
      "num_input_tokens_seen": 2044592,
      "step": 3095
    },
    {
      "epoch": 1.8278301886792452,
      "grad_norm": 3.5046775341033936,
      "learning_rate": 9.13620283018868e-06,
      "loss": 0.4562,
      "num_input_tokens_seen": 2047888,
      "step": 3100
    },
    {
      "epoch": 1.8307783018867925,
      "grad_norm": 2.8478949069976807,
      "learning_rate": 9.150943396226416e-06,
      "loss": 0.6087,
      "num_input_tokens_seen": 2050928,
      "step": 3105
    },
    {
      "epoch": 1.8337264150943398,
      "grad_norm": 1.637563943862915,
      "learning_rate": 9.165683962264151e-06,
      "loss": 0.4487,
      "num_input_tokens_seen": 2054416,
      "step": 3110
    },
    {
      "epoch": 1.8366745283018868,
      "grad_norm": 7.1760573387146,
      "learning_rate": 9.180424528301888e-06,
      "loss": 0.5342,
      "num_input_tokens_seen": 2057552,
      "step": 3115
    },
    {
      "epoch": 1.8396226415094339,
      "grad_norm": 3.505415916442871,
      "learning_rate": 9.195165094339623e-06,
      "loss": 0.5337,
      "num_input_tokens_seen": 2059920,
      "step": 3120
    },
    {
      "epoch": 1.8425707547169812,
      "grad_norm": 2.9641242027282715,
      "learning_rate": 9.20990566037736e-06,
      "loss": 0.4585,
      "num_input_tokens_seen": 2063792,
      "step": 3125
    },
    {
      "epoch": 1.8455188679245285,
      "grad_norm": 2.6536765098571777,
      "learning_rate": 9.224646226415096e-06,
      "loss": 0.5091,
      "num_input_tokens_seen": 2066512,
      "step": 3130
    },
    {
      "epoch": 1.8484669811320755,
      "grad_norm": 2.0200655460357666,
      "learning_rate": 9.239386792452831e-06,
      "loss": 0.5233,
      "num_input_tokens_seen": 2069904,
      "step": 3135
    },
    {
      "epoch": 1.8514150943396226,
      "grad_norm": 3.896906852722168,
      "learning_rate": 9.254127358490566e-06,
      "loss": 0.5258,
      "num_input_tokens_seen": 2073008,
      "step": 3140
    },
    {
      "epoch": 1.8543632075471699,
      "grad_norm": 1.817622184753418,
      "learning_rate": 9.268867924528303e-06,
      "loss": 0.4825,
      "num_input_tokens_seen": 2075600,
      "step": 3145
    },
    {
      "epoch": 1.857311320754717,
      "grad_norm": 4.97838830947876,
      "learning_rate": 9.283608490566039e-06,
      "loss": 0.4515,
      "num_input_tokens_seen": 2078800,
      "step": 3150
    },
    {
      "epoch": 1.860259433962264,
      "grad_norm": 4.210000038146973,
      "learning_rate": 9.298349056603774e-06,
      "loss": 0.565,
      "num_input_tokens_seen": 2081936,
      "step": 3155
    },
    {
      "epoch": 1.8632075471698113,
      "grad_norm": 3.07875657081604,
      "learning_rate": 9.31308962264151e-06,
      "loss": 0.5596,
      "num_input_tokens_seen": 2084592,
      "step": 3160
    },
    {
      "epoch": 1.8661556603773586,
      "grad_norm": 3.6212480068206787,
      "learning_rate": 9.327830188679245e-06,
      "loss": 0.4971,
      "num_input_tokens_seen": 2088496,
      "step": 3165
    },
    {
      "epoch": 1.8691037735849056,
      "grad_norm": 1.9862090349197388,
      "learning_rate": 9.342570754716982e-06,
      "loss": 0.5197,
      "num_input_tokens_seen": 2091472,
      "step": 3170
    },
    {
      "epoch": 1.8720518867924527,
      "grad_norm": 4.634102821350098,
      "learning_rate": 9.357311320754717e-06,
      "loss": 0.579,
      "num_input_tokens_seen": 2094224,
      "step": 3175
    },
    {
      "epoch": 1.875,
      "grad_norm": 3.712751865386963,
      "learning_rate": 9.372051886792454e-06,
      "loss": 0.596,
      "num_input_tokens_seen": 2098256,
      "step": 3180
    },
    {
      "epoch": 1.8779481132075473,
      "grad_norm": 2.3242452144622803,
      "learning_rate": 9.38679245283019e-06,
      "loss": 0.5862,
      "num_input_tokens_seen": 2101584,
      "step": 3185
    },
    {
      "epoch": 1.8808962264150944,
      "grad_norm": 3.013577461242676,
      "learning_rate": 9.401533018867925e-06,
      "loss": 0.5773,
      "num_input_tokens_seen": 2104592,
      "step": 3190
    },
    {
      "epoch": 1.8838443396226414,
      "grad_norm": 1.9774895906448364,
      "learning_rate": 9.41627358490566e-06,
      "loss": 0.4231,
      "num_input_tokens_seen": 2108144,
      "step": 3195
    },
    {
      "epoch": 1.8867924528301887,
      "grad_norm": 2.1005117893218994,
      "learning_rate": 9.431014150943397e-06,
      "loss": 0.3617,
      "num_input_tokens_seen": 2112304,
      "step": 3200
    },
    {
      "epoch": 1.889740566037736,
      "grad_norm": 2.526937246322632,
      "learning_rate": 9.445754716981133e-06,
      "loss": 0.5277,
      "num_input_tokens_seen": 2115312,
      "step": 3205
    },
    {
      "epoch": 1.892688679245283,
      "grad_norm": 2.6452088356018066,
      "learning_rate": 9.46049528301887e-06,
      "loss": 0.4618,
      "num_input_tokens_seen": 2118576,
      "step": 3210
    },
    {
      "epoch": 1.8956367924528301,
      "grad_norm": 3.5076706409454346,
      "learning_rate": 9.475235849056605e-06,
      "loss": 0.7152,
      "num_input_tokens_seen": 2122416,
      "step": 3215
    },
    {
      "epoch": 1.8985849056603774,
      "grad_norm": 2.707148790359497,
      "learning_rate": 9.48997641509434e-06,
      "loss": 0.4489,
      "num_input_tokens_seen": 2125136,
      "step": 3220
    },
    {
      "epoch": 1.9015330188679245,
      "grad_norm": 1.7831107378005981,
      "learning_rate": 9.504716981132076e-06,
      "loss": 0.5092,
      "num_input_tokens_seen": 2128112,
      "step": 3225
    },
    {
      "epoch": 1.9044811320754715,
      "grad_norm": 2.2815866470336914,
      "learning_rate": 9.519457547169813e-06,
      "loss": 0.6295,
      "num_input_tokens_seen": 2131728,
      "step": 3230
    },
    {
      "epoch": 1.9074292452830188,
      "grad_norm": 1.6930468082427979,
      "learning_rate": 9.534198113207548e-06,
      "loss": 0.3722,
      "num_input_tokens_seen": 2135024,
      "step": 3235
    },
    {
      "epoch": 1.9103773584905661,
      "grad_norm": 1.7304255962371826,
      "learning_rate": 9.548938679245285e-06,
      "loss": 0.5673,
      "num_input_tokens_seen": 2138448,
      "step": 3240
    },
    {
      "epoch": 1.9133254716981132,
      "grad_norm": 3.7182111740112305,
      "learning_rate": 9.56367924528302e-06,
      "loss": 0.5578,
      "num_input_tokens_seen": 2142032,
      "step": 3245
    },
    {
      "epoch": 1.9162735849056602,
      "grad_norm": 1.7988394498825073,
      "learning_rate": 9.578419811320756e-06,
      "loss": 0.5426,
      "num_input_tokens_seen": 2145008,
      "step": 3250
    },
    {
      "epoch": 1.9192216981132075,
      "grad_norm": 3.94294810295105,
      "learning_rate": 9.593160377358491e-06,
      "loss": 0.4345,
      "num_input_tokens_seen": 2147568,
      "step": 3255
    },
    {
      "epoch": 1.9221698113207548,
      "grad_norm": 6.033405303955078,
      "learning_rate": 9.607900943396226e-06,
      "loss": 0.6536,
      "num_input_tokens_seen": 2150608,
      "step": 3260
    },
    {
      "epoch": 1.9251179245283019,
      "grad_norm": 3.6172943115234375,
      "learning_rate": 9.622641509433963e-06,
      "loss": 0.5344,
      "num_input_tokens_seen": 2154192,
      "step": 3265
    },
    {
      "epoch": 1.928066037735849,
      "grad_norm": 2.1095807552337646,
      "learning_rate": 9.637382075471699e-06,
      "loss": 0.4811,
      "num_input_tokens_seen": 2156976,
      "step": 3270
    },
    {
      "epoch": 1.9310141509433962,
      "grad_norm": 1.899039387702942,
      "learning_rate": 9.652122641509434e-06,
      "loss": 0.4355,
      "num_input_tokens_seen": 2160432,
      "step": 3275
    },
    {
      "epoch": 1.9339622641509435,
      "grad_norm": 6.3099446296691895,
      "learning_rate": 9.66686320754717e-06,
      "loss": 0.4871,
      "num_input_tokens_seen": 2163408,
      "step": 3280
    },
    {
      "epoch": 1.9369103773584906,
      "grad_norm": 1.9237180948257446,
      "learning_rate": 9.681603773584907e-06,
      "loss": 0.4767,
      "num_input_tokens_seen": 2166768,
      "step": 3285
    },
    {
      "epoch": 1.9398584905660377,
      "grad_norm": 3.0543997287750244,
      "learning_rate": 9.696344339622642e-06,
      "loss": 0.4676,
      "num_input_tokens_seen": 2169456,
      "step": 3290
    },
    {
      "epoch": 1.942806603773585,
      "grad_norm": 1.3310855627059937,
      "learning_rate": 9.711084905660379e-06,
      "loss": 0.4803,
      "num_input_tokens_seen": 2174352,
      "step": 3295
    },
    {
      "epoch": 1.9457547169811322,
      "grad_norm": 4.219992637634277,
      "learning_rate": 9.725825471698114e-06,
      "loss": 0.5593,
      "num_input_tokens_seen": 2177584,
      "step": 3300
    },
    {
      "epoch": 1.9487028301886793,
      "grad_norm": 2.526475191116333,
      "learning_rate": 9.74056603773585e-06,
      "loss": 0.5101,
      "num_input_tokens_seen": 2180528,
      "step": 3305
    },
    {
      "epoch": 1.9516509433962264,
      "grad_norm": 2.525611639022827,
      "learning_rate": 9.755306603773585e-06,
      "loss": 0.5028,
      "num_input_tokens_seen": 2183888,
      "step": 3310
    },
    {
      "epoch": 1.9545990566037736,
      "grad_norm": 2.737025260925293,
      "learning_rate": 9.770047169811322e-06,
      "loss": 0.5164,
      "num_input_tokens_seen": 2187536,
      "step": 3315
    },
    {
      "epoch": 1.9575471698113207,
      "grad_norm": 3.864600896835327,
      "learning_rate": 9.784787735849057e-06,
      "loss": 0.5586,
      "num_input_tokens_seen": 2190864,
      "step": 3320
    },
    {
      "epoch": 1.9604952830188678,
      "grad_norm": 4.561527252197266,
      "learning_rate": 9.799528301886794e-06,
      "loss": 0.4895,
      "num_input_tokens_seen": 2193872,
      "step": 3325
    },
    {
      "epoch": 1.963443396226415,
      "grad_norm": 2.0018696784973145,
      "learning_rate": 9.81426886792453e-06,
      "loss": 0.4372,
      "num_input_tokens_seen": 2197744,
      "step": 3330
    },
    {
      "epoch": 1.9663915094339623,
      "grad_norm": 1.7424472570419312,
      "learning_rate": 9.829009433962265e-06,
      "loss": 0.4577,
      "num_input_tokens_seen": 2201648,
      "step": 3335
    },
    {
      "epoch": 1.9693396226415094,
      "grad_norm": 2.6151909828186035,
      "learning_rate": 9.84375e-06,
      "loss": 0.5065,
      "num_input_tokens_seen": 2205808,
      "step": 3340
    },
    {
      "epoch": 1.9722877358490565,
      "grad_norm": 5.248737335205078,
      "learning_rate": 9.858490566037736e-06,
      "loss": 0.6834,
      "num_input_tokens_seen": 2208656,
      "step": 3345
    },
    {
      "epoch": 1.9752358490566038,
      "grad_norm": 1.7748459577560425,
      "learning_rate": 9.873231132075473e-06,
      "loss": 0.563,
      "num_input_tokens_seen": 2212048,
      "step": 3350
    },
    {
      "epoch": 1.978183962264151,
      "grad_norm": 2.3101630210876465,
      "learning_rate": 9.887971698113208e-06,
      "loss": 0.5711,
      "num_input_tokens_seen": 2215344,
      "step": 3355
    },
    {
      "epoch": 1.9811320754716981,
      "grad_norm": 1.7551658153533936,
      "learning_rate": 9.902712264150945e-06,
      "loss": 0.3759,
      "num_input_tokens_seen": 2219504,
      "step": 3360
    },
    {
      "epoch": 1.9840801886792452,
      "grad_norm": 1.940865159034729,
      "learning_rate": 9.917452830188679e-06,
      "loss": 0.4917,
      "num_input_tokens_seen": 2222352,
      "step": 3365
    },
    {
      "epoch": 1.9870283018867925,
      "grad_norm": 2.858696222305298,
      "learning_rate": 9.932193396226416e-06,
      "loss": 0.3843,
      "num_input_tokens_seen": 2225520,
      "step": 3370
    },
    {
      "epoch": 1.9899764150943398,
      "grad_norm": 1.8749135732650757,
      "learning_rate": 9.946933962264151e-06,
      "loss": 0.6343,
      "num_input_tokens_seen": 2228528,
      "step": 3375
    },
    {
      "epoch": 1.9929245283018868,
      "grad_norm": 1.7372334003448486,
      "learning_rate": 9.961674528301888e-06,
      "loss": 0.4117,
      "num_input_tokens_seen": 2233136,
      "step": 3380
    },
    {
      "epoch": 1.9958726415094339,
      "grad_norm": 3.473111867904663,
      "learning_rate": 9.976415094339624e-06,
      "loss": 0.4315,
      "num_input_tokens_seen": 2235952,
      "step": 3385
    },
    {
      "epoch": 1.9988207547169812,
      "grad_norm": 2.2239444255828857,
      "learning_rate": 9.991155660377359e-06,
      "loss": 0.39,
      "num_input_tokens_seen": 2239280,
      "step": 3390
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.5126349329948425,
      "eval_runtime": 18.358,
      "eval_samples_per_second": 92.385,
      "eval_steps_per_second": 23.096,
      "num_input_tokens_seen": 2240840,
      "step": 3392
    },
    {
      "epoch": 2.0017688679245285,
      "grad_norm": 1.7303496599197388,
      "learning_rate": 9.999999894098275e-06,
      "loss": 0.5249,
      "num_input_tokens_seen": 2242888,
      "step": 3395
    },
    {
      "epoch": 2.0047169811320753,
      "grad_norm": 2.2698607444763184,
      "learning_rate": 9.99999870270391e-06,
      "loss": 0.5551,
      "num_input_tokens_seen": 2245768,
      "step": 3400
    },
    {
      "epoch": 2.0076650943396226,
      "grad_norm": 2.9583072662353516,
      "learning_rate": 9.999996187538341e-06,
      "loss": 0.3755,
      "num_input_tokens_seen": 2248360,
      "step": 3405
    },
    {
      "epoch": 2.01061320754717,
      "grad_norm": 2.121467113494873,
      "learning_rate": 9.999992348602233e-06,
      "loss": 0.5116,
      "num_input_tokens_seen": 2251432,
      "step": 3410
    },
    {
      "epoch": 2.013561320754717,
      "grad_norm": 4.143191814422607,
      "learning_rate": 9.999987185896598e-06,
      "loss": 0.5466,
      "num_input_tokens_seen": 2254792,
      "step": 3415
    },
    {
      "epoch": 2.016509433962264,
      "grad_norm": 1.85322904586792,
      "learning_rate": 9.99998069942281e-06,
      "loss": 0.6024,
      "num_input_tokens_seen": 2257928,
      "step": 3420
    },
    {
      "epoch": 2.0194575471698113,
      "grad_norm": 1.961767554283142,
      "learning_rate": 9.999972889182583e-06,
      "loss": 0.5609,
      "num_input_tokens_seen": 2260488,
      "step": 3425
    },
    {
      "epoch": 2.0224056603773586,
      "grad_norm": 3.6571621894836426,
      "learning_rate": 9.999963755177984e-06,
      "loss": 0.4909,
      "num_input_tokens_seen": 2263784,
      "step": 3430
    },
    {
      "epoch": 2.025353773584906,
      "grad_norm": 3.3070151805877686,
      "learning_rate": 9.999953297411434e-06,
      "loss": 0.4853,
      "num_input_tokens_seen": 2266792,
      "step": 3435
    },
    {
      "epoch": 2.0283018867924527,
      "grad_norm": 2.3486104011535645,
      "learning_rate": 9.999941515885699e-06,
      "loss": 0.3852,
      "num_input_tokens_seen": 2269832,
      "step": 3440
    },
    {
      "epoch": 2.03125,
      "grad_norm": 3.421314001083374,
      "learning_rate": 9.999928410603897e-06,
      "loss": 0.4318,
      "num_input_tokens_seen": 2273928,
      "step": 3445
    },
    {
      "epoch": 2.0341981132075473,
      "grad_norm": 3.6625282764434814,
      "learning_rate": 9.999913981569502e-06,
      "loss": 0.5244,
      "num_input_tokens_seen": 2276552,
      "step": 3450
    },
    {
      "epoch": 2.037146226415094,
      "grad_norm": 2.341811180114746,
      "learning_rate": 9.999898228786332e-06,
      "loss": 0.5453,
      "num_input_tokens_seen": 2279272,
      "step": 3455
    },
    {
      "epoch": 2.0400943396226414,
      "grad_norm": 1.826859951019287,
      "learning_rate": 9.999881152258557e-06,
      "loss": 0.5779,
      "num_input_tokens_seen": 2282216,
      "step": 3460
    },
    {
      "epoch": 2.0430424528301887,
      "grad_norm": 3.9800732135772705,
      "learning_rate": 9.999862751990697e-06,
      "loss": 0.5276,
      "num_input_tokens_seen": 2285480,
      "step": 3465
    },
    {
      "epoch": 2.045990566037736,
      "grad_norm": 2.662952423095703,
      "learning_rate": 9.999843027987628e-06,
      "loss": 0.5144,
      "num_input_tokens_seen": 2289160,
      "step": 3470
    },
    {
      "epoch": 2.048938679245283,
      "grad_norm": 2.3676702976226807,
      "learning_rate": 9.999821980254567e-06,
      "loss": 0.5571,
      "num_input_tokens_seen": 2292520,
      "step": 3475
    },
    {
      "epoch": 2.05188679245283,
      "grad_norm": 2.6496920585632324,
      "learning_rate": 9.99979960879709e-06,
      "loss": 0.3955,
      "num_input_tokens_seen": 2295144,
      "step": 3480
    },
    {
      "epoch": 2.0548349056603774,
      "grad_norm": 12.361777305603027,
      "learning_rate": 9.99977591362112e-06,
      "loss": 0.6016,
      "num_input_tokens_seen": 2297992,
      "step": 3485
    },
    {
      "epoch": 2.0577830188679247,
      "grad_norm": 2.133758783340454,
      "learning_rate": 9.999750894732927e-06,
      "loss": 0.3958,
      "num_input_tokens_seen": 2301640,
      "step": 3490
    },
    {
      "epoch": 2.0607311320754715,
      "grad_norm": 1.6844228506088257,
      "learning_rate": 9.999724552139136e-06,
      "loss": 0.4512,
      "num_input_tokens_seen": 2304456,
      "step": 3495
    },
    {
      "epoch": 2.063679245283019,
      "grad_norm": 1.891971468925476,
      "learning_rate": 9.999696885846724e-06,
      "loss": 0.5275,
      "num_input_tokens_seen": 2307304,
      "step": 3500
    },
    {
      "epoch": 2.066627358490566,
      "grad_norm": 2.008331537246704,
      "learning_rate": 9.999667895863012e-06,
      "loss": 0.5665,
      "num_input_tokens_seen": 2310920,
      "step": 3505
    },
    {
      "epoch": 2.0695754716981134,
      "grad_norm": 3.4308035373687744,
      "learning_rate": 9.99963758219568e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 2315048,
      "step": 3510
    },
    {
      "epoch": 2.0725235849056602,
      "grad_norm": 1.6369047164916992,
      "learning_rate": 9.999605944852749e-06,
      "loss": 0.4154,
      "num_input_tokens_seen": 2318056,
      "step": 3515
    },
    {
      "epoch": 2.0754716981132075,
      "grad_norm": 2.4952547550201416,
      "learning_rate": 9.999572983842599e-06,
      "loss": 0.4955,
      "num_input_tokens_seen": 2320936,
      "step": 3520
    },
    {
      "epoch": 2.078419811320755,
      "grad_norm": 2.305112838745117,
      "learning_rate": 9.999538699173951e-06,
      "loss": 0.5123,
      "num_input_tokens_seen": 2324424,
      "step": 3525
    },
    {
      "epoch": 2.081367924528302,
      "grad_norm": 1.9047130346298218,
      "learning_rate": 9.99950309085589e-06,
      "loss": 0.592,
      "num_input_tokens_seen": 2328232,
      "step": 3530
    },
    {
      "epoch": 2.084316037735849,
      "grad_norm": 2.199582576751709,
      "learning_rate": 9.999466158897835e-06,
      "loss": 0.5512,
      "num_input_tokens_seen": 2331336,
      "step": 3535
    },
    {
      "epoch": 2.0872641509433962,
      "grad_norm": 1.7395168542861938,
      "learning_rate": 9.999427903309569e-06,
      "loss": 0.6292,
      "num_input_tokens_seen": 2334632,
      "step": 3540
    },
    {
      "epoch": 2.0902122641509435,
      "grad_norm": 3.2136826515197754,
      "learning_rate": 9.99938832410122e-06,
      "loss": 0.6109,
      "num_input_tokens_seen": 2337032,
      "step": 3545
    },
    {
      "epoch": 2.0931603773584904,
      "grad_norm": 4.786601543426514,
      "learning_rate": 9.999347421283267e-06,
      "loss": 0.5749,
      "num_input_tokens_seen": 2339880,
      "step": 3550
    },
    {
      "epoch": 2.0961084905660377,
      "grad_norm": 2.594761848449707,
      "learning_rate": 9.999305194866538e-06,
      "loss": 0.5289,
      "num_input_tokens_seen": 2344168,
      "step": 3555
    },
    {
      "epoch": 2.099056603773585,
      "grad_norm": 5.254917621612549,
      "learning_rate": 9.99926164486221e-06,
      "loss": 0.6447,
      "num_input_tokens_seen": 2347336,
      "step": 3560
    },
    {
      "epoch": 2.1020047169811322,
      "grad_norm": 2.8885111808776855,
      "learning_rate": 9.99921677128182e-06,
      "loss": 0.5079,
      "num_input_tokens_seen": 2350376,
      "step": 3565
    },
    {
      "epoch": 2.104952830188679,
      "grad_norm": 2.2943599224090576,
      "learning_rate": 9.99917057413724e-06,
      "loss": 0.4816,
      "num_input_tokens_seen": 2353672,
      "step": 3570
    },
    {
      "epoch": 2.1079009433962264,
      "grad_norm": 2.2593648433685303,
      "learning_rate": 9.99912305344071e-06,
      "loss": 0.4366,
      "num_input_tokens_seen": 2356968,
      "step": 3575
    },
    {
      "epoch": 2.1108490566037736,
      "grad_norm": 3.3187105655670166,
      "learning_rate": 9.999074209204803e-06,
      "loss": 0.5367,
      "num_input_tokens_seen": 2361640,
      "step": 3580
    },
    {
      "epoch": 2.113797169811321,
      "grad_norm": 1.873623013496399,
      "learning_rate": 9.999024041442455e-06,
      "loss": 0.4306,
      "num_input_tokens_seen": 2365736,
      "step": 3585
    },
    {
      "epoch": 2.1167452830188678,
      "grad_norm": 2.2563114166259766,
      "learning_rate": 9.998972550166948e-06,
      "loss": 0.5415,
      "num_input_tokens_seen": 2369992,
      "step": 3590
    },
    {
      "epoch": 2.119693396226415,
      "grad_norm": 2.586265802383423,
      "learning_rate": 9.998919735391915e-06,
      "loss": 0.583,
      "num_input_tokens_seen": 2372712,
      "step": 3595
    },
    {
      "epoch": 2.1226415094339623,
      "grad_norm": 2.155385732650757,
      "learning_rate": 9.998865597131336e-06,
      "loss": 0.5088,
      "num_input_tokens_seen": 2375688,
      "step": 3600
    },
    {
      "epoch": 2.1255896226415096,
      "grad_norm": 1.7584561109542847,
      "learning_rate": 9.998810135399545e-06,
      "loss": 0.7655,
      "num_input_tokens_seen": 2379208,
      "step": 3605
    },
    {
      "epoch": 2.1285377358490565,
      "grad_norm": 2.719663619995117,
      "learning_rate": 9.99875335021123e-06,
      "loss": 0.703,
      "num_input_tokens_seen": 2382248,
      "step": 3610
    },
    {
      "epoch": 2.1314858490566038,
      "grad_norm": 1.7762134075164795,
      "learning_rate": 9.998695241581423e-06,
      "loss": 0.4949,
      "num_input_tokens_seen": 2384680,
      "step": 3615
    },
    {
      "epoch": 2.134433962264151,
      "grad_norm": 12.621379852294922,
      "learning_rate": 9.998635809525504e-06,
      "loss": 0.5446,
      "num_input_tokens_seen": 2387720,
      "step": 3620
    },
    {
      "epoch": 2.137382075471698,
      "grad_norm": 2.1943585872650146,
      "learning_rate": 9.998575054059212e-06,
      "loss": 0.419,
      "num_input_tokens_seen": 2390984,
      "step": 3625
    },
    {
      "epoch": 2.140330188679245,
      "grad_norm": 2.075981855392456,
      "learning_rate": 9.998512975198633e-06,
      "loss": 0.4044,
      "num_input_tokens_seen": 2394088,
      "step": 3630
    },
    {
      "epoch": 2.1432783018867925,
      "grad_norm": 2.0441489219665527,
      "learning_rate": 9.998449572960202e-06,
      "loss": 0.3272,
      "num_input_tokens_seen": 2397448,
      "step": 3635
    },
    {
      "epoch": 2.1462264150943398,
      "grad_norm": 2.8592276573181152,
      "learning_rate": 9.998384847360705e-06,
      "loss": 0.4501,
      "num_input_tokens_seen": 2401544,
      "step": 3640
    },
    {
      "epoch": 2.1491745283018866,
      "grad_norm": 3.271057605743408,
      "learning_rate": 9.998318798417276e-06,
      "loss": 0.5206,
      "num_input_tokens_seen": 2405384,
      "step": 3645
    },
    {
      "epoch": 2.152122641509434,
      "grad_norm": 2.1515817642211914,
      "learning_rate": 9.998251426147403e-06,
      "loss": 0.5323,
      "num_input_tokens_seen": 2409416,
      "step": 3650
    },
    {
      "epoch": 2.155070754716981,
      "grad_norm": 2.106083631515503,
      "learning_rate": 9.998182730568927e-06,
      "loss": 0.5034,
      "num_input_tokens_seen": 2411784,
      "step": 3655
    },
    {
      "epoch": 2.1580188679245285,
      "grad_norm": 3.4970130920410156,
      "learning_rate": 9.998112711700028e-06,
      "loss": 0.5168,
      "num_input_tokens_seen": 2414056,
      "step": 3660
    },
    {
      "epoch": 2.1609669811320753,
      "grad_norm": 5.957014083862305,
      "learning_rate": 9.99804136955925e-06,
      "loss": 0.5445,
      "num_input_tokens_seen": 2417032,
      "step": 3665
    },
    {
      "epoch": 2.1639150943396226,
      "grad_norm": 2.179136276245117,
      "learning_rate": 9.99796870416548e-06,
      "loss": 0.4939,
      "num_input_tokens_seen": 2420712,
      "step": 3670
    },
    {
      "epoch": 2.16686320754717,
      "grad_norm": 4.403555393218994,
      "learning_rate": 9.997894715537953e-06,
      "loss": 0.4394,
      "num_input_tokens_seen": 2424008,
      "step": 3675
    },
    {
      "epoch": 2.169811320754717,
      "grad_norm": 2.499314069747925,
      "learning_rate": 9.997819403696263e-06,
      "loss": 0.5627,
      "num_input_tokens_seen": 2426632,
      "step": 3680
    },
    {
      "epoch": 2.172759433962264,
      "grad_norm": 2.7302334308624268,
      "learning_rate": 9.997742768660345e-06,
      "loss": 0.4672,
      "num_input_tokens_seen": 2430216,
      "step": 3685
    },
    {
      "epoch": 2.1757075471698113,
      "grad_norm": 1.8167957067489624,
      "learning_rate": 9.99766481045049e-06,
      "loss": 0.4919,
      "num_input_tokens_seen": 2433064,
      "step": 3690
    },
    {
      "epoch": 2.1786556603773586,
      "grad_norm": 4.698959827423096,
      "learning_rate": 9.997585529087338e-06,
      "loss": 0.4946,
      "num_input_tokens_seen": 2436168,
      "step": 3695
    },
    {
      "epoch": 2.1816037735849054,
      "grad_norm": 6.5109782218933105,
      "learning_rate": 9.997504924591878e-06,
      "loss": 0.5926,
      "num_input_tokens_seen": 2438984,
      "step": 3700
    },
    {
      "epoch": 2.1845518867924527,
      "grad_norm": 2.352698564529419,
      "learning_rate": 9.99742299698545e-06,
      "loss": 0.4573,
      "num_input_tokens_seen": 2445704,
      "step": 3705
    },
    {
      "epoch": 2.1875,
      "grad_norm": 4.09055757522583,
      "learning_rate": 9.997339746289749e-06,
      "loss": 0.6285,
      "num_input_tokens_seen": 2449416,
      "step": 3710
    },
    {
      "epoch": 2.1904481132075473,
      "grad_norm": 2.0476622581481934,
      "learning_rate": 9.997255172526812e-06,
      "loss": 0.9688,
      "num_input_tokens_seen": 2451912,
      "step": 3715
    },
    {
      "epoch": 2.1933962264150946,
      "grad_norm": 3.584923505783081,
      "learning_rate": 9.99716927571903e-06,
      "loss": 0.5414,
      "num_input_tokens_seen": 2454216,
      "step": 3720
    },
    {
      "epoch": 2.1963443396226414,
      "grad_norm": 2.489267587661743,
      "learning_rate": 9.997082055889147e-06,
      "loss": 0.4932,
      "num_input_tokens_seen": 2457192,
      "step": 3725
    },
    {
      "epoch": 2.1992924528301887,
      "grad_norm": 2.852328300476074,
      "learning_rate": 9.996993513060252e-06,
      "loss": 0.5475,
      "num_input_tokens_seen": 2460520,
      "step": 3730
    },
    {
      "epoch": 2.202240566037736,
      "grad_norm": 3.823976755142212,
      "learning_rate": 9.996903647255789e-06,
      "loss": 0.6302,
      "num_input_tokens_seen": 2463784,
      "step": 3735
    },
    {
      "epoch": 2.205188679245283,
      "grad_norm": 2.0159411430358887,
      "learning_rate": 9.99681245849955e-06,
      "loss": 0.4811,
      "num_input_tokens_seen": 2466984,
      "step": 3740
    },
    {
      "epoch": 2.20813679245283,
      "grad_norm": 1.6007546186447144,
      "learning_rate": 9.996719946815679e-06,
      "loss": 0.5141,
      "num_input_tokens_seen": 2469864,
      "step": 3745
    },
    {
      "epoch": 2.2110849056603774,
      "grad_norm": 2.8939871788024902,
      "learning_rate": 9.996626112228665e-06,
      "loss": 0.4908,
      "num_input_tokens_seen": 2472936,
      "step": 3750
    },
    {
      "epoch": 2.2140330188679247,
      "grad_norm": 2.2986795902252197,
      "learning_rate": 9.996530954763355e-06,
      "loss": 0.6395,
      "num_input_tokens_seen": 2475688,
      "step": 3755
    },
    {
      "epoch": 2.2169811320754715,
      "grad_norm": 1.8638129234313965,
      "learning_rate": 9.99643447444494e-06,
      "loss": 0.5129,
      "num_input_tokens_seen": 2479176,
      "step": 3760
    },
    {
      "epoch": 2.219929245283019,
      "grad_norm": 2.012836456298828,
      "learning_rate": 9.996336671298965e-06,
      "loss": 0.4446,
      "num_input_tokens_seen": 2482184,
      "step": 3765
    },
    {
      "epoch": 2.222877358490566,
      "grad_norm": 3.2771873474121094,
      "learning_rate": 9.996237545351323e-06,
      "loss": 0.628,
      "num_input_tokens_seen": 2485320,
      "step": 3770
    },
    {
      "epoch": 2.2258254716981134,
      "grad_norm": 2.191920757293701,
      "learning_rate": 9.996137096628259e-06,
      "loss": 0.4482,
      "num_input_tokens_seen": 2489576,
      "step": 3775
    },
    {
      "epoch": 2.2287735849056602,
      "grad_norm": 4.106274127960205,
      "learning_rate": 9.996035325156366e-06,
      "loss": 0.5175,
      "num_input_tokens_seen": 2492680,
      "step": 3780
    },
    {
      "epoch": 2.2317216981132075,
      "grad_norm": 2.0458126068115234,
      "learning_rate": 9.995932230962589e-06,
      "loss": 0.5551,
      "num_input_tokens_seen": 2495656,
      "step": 3785
    },
    {
      "epoch": 2.234669811320755,
      "grad_norm": 3.3118228912353516,
      "learning_rate": 9.995827814074223e-06,
      "loss": 0.5522,
      "num_input_tokens_seen": 2498536,
      "step": 3790
    },
    {
      "epoch": 2.237617924528302,
      "grad_norm": 1.5962865352630615,
      "learning_rate": 9.995722074518913e-06,
      "loss": 0.4882,
      "num_input_tokens_seen": 2501320,
      "step": 3795
    },
    {
      "epoch": 2.240566037735849,
      "grad_norm": 4.463668346405029,
      "learning_rate": 9.99561501232465e-06,
      "loss": 0.4298,
      "num_input_tokens_seen": 2505352,
      "step": 3800
    },
    {
      "epoch": 2.2435141509433962,
      "grad_norm": 2.5580005645751953,
      "learning_rate": 9.995506627519786e-06,
      "loss": 0.5105,
      "num_input_tokens_seen": 2509096,
      "step": 3805
    },
    {
      "epoch": 2.2464622641509435,
      "grad_norm": 2.028205394744873,
      "learning_rate": 9.995396920133012e-06,
      "loss": 0.4181,
      "num_input_tokens_seen": 2512232,
      "step": 3810
    },
    {
      "epoch": 2.2494103773584904,
      "grad_norm": 1.1195536851882935,
      "learning_rate": 9.995285890193373e-06,
      "loss": 0.4455,
      "num_input_tokens_seen": 2515208,
      "step": 3815
    },
    {
      "epoch": 2.2523584905660377,
      "grad_norm": 1.3056161403656006,
      "learning_rate": 9.995173537730267e-06,
      "loss": 0.6918,
      "num_input_tokens_seen": 2519432,
      "step": 3820
    },
    {
      "epoch": 2.255306603773585,
      "grad_norm": 1.1498873233795166,
      "learning_rate": 9.99505986277344e-06,
      "loss": 0.4126,
      "num_input_tokens_seen": 2522792,
      "step": 3825
    },
    {
      "epoch": 2.2582547169811322,
      "grad_norm": 4.129491329193115,
      "learning_rate": 9.994944865352986e-06,
      "loss": 0.5137,
      "num_input_tokens_seen": 2526088,
      "step": 3830
    },
    {
      "epoch": 2.261202830188679,
      "grad_norm": 3.9379968643188477,
      "learning_rate": 9.994828545499351e-06,
      "loss": 0.5356,
      "num_input_tokens_seen": 2528744,
      "step": 3835
    },
    {
      "epoch": 2.2641509433962264,
      "grad_norm": 1.828855037689209,
      "learning_rate": 9.994710903243334e-06,
      "loss": 0.5342,
      "num_input_tokens_seen": 2531720,
      "step": 3840
    },
    {
      "epoch": 2.2670990566037736,
      "grad_norm": 2.0305917263031006,
      "learning_rate": 9.994591938616079e-06,
      "loss": 0.4574,
      "num_input_tokens_seen": 2534632,
      "step": 3845
    },
    {
      "epoch": 2.270047169811321,
      "grad_norm": 2.9462578296661377,
      "learning_rate": 9.994471651649082e-06,
      "loss": 0.3777,
      "num_input_tokens_seen": 2537160,
      "step": 3850
    },
    {
      "epoch": 2.2729952830188678,
      "grad_norm": 1.898756504058838,
      "learning_rate": 9.99435004237419e-06,
      "loss": 0.4324,
      "num_input_tokens_seen": 2541032,
      "step": 3855
    },
    {
      "epoch": 2.275943396226415,
      "grad_norm": 1.7668652534484863,
      "learning_rate": 9.9942271108236e-06,
      "loss": 0.4278,
      "num_input_tokens_seen": 2544488,
      "step": 3860
    },
    {
      "epoch": 2.2788915094339623,
      "grad_norm": 3.6266613006591797,
      "learning_rate": 9.994102857029859e-06,
      "loss": 0.5759,
      "num_input_tokens_seen": 2547944,
      "step": 3865
    },
    {
      "epoch": 2.2818396226415096,
      "grad_norm": 2.334014654159546,
      "learning_rate": 9.993977281025862e-06,
      "loss": 0.572,
      "num_input_tokens_seen": 2550824,
      "step": 3870
    },
    {
      "epoch": 2.2847877358490565,
      "grad_norm": 2.4741313457489014,
      "learning_rate": 9.993850382844858e-06,
      "loss": 0.5236,
      "num_input_tokens_seen": 2553352,
      "step": 3875
    },
    {
      "epoch": 2.2877358490566038,
      "grad_norm": 3.927119016647339,
      "learning_rate": 9.993722162520443e-06,
      "loss": 0.4871,
      "num_input_tokens_seen": 2556232,
      "step": 3880
    },
    {
      "epoch": 2.290683962264151,
      "grad_norm": 2.129019260406494,
      "learning_rate": 9.993592620086564e-06,
      "loss": 0.4901,
      "num_input_tokens_seen": 2558376,
      "step": 3885
    },
    {
      "epoch": 2.293632075471698,
      "grad_norm": 1.8753594160079956,
      "learning_rate": 9.993461755577518e-06,
      "loss": 0.4081,
      "num_input_tokens_seen": 2562984,
      "step": 3890
    },
    {
      "epoch": 2.296580188679245,
      "grad_norm": 3.6495864391326904,
      "learning_rate": 9.99332956902795e-06,
      "loss": 0.4689,
      "num_input_tokens_seen": 2565032,
      "step": 3895
    },
    {
      "epoch": 2.2995283018867925,
      "grad_norm": 1.466093897819519,
      "learning_rate": 9.993196060472859e-06,
      "loss": 0.4981,
      "num_input_tokens_seen": 2569064,
      "step": 3900
    },
    {
      "epoch": 2.3024764150943398,
      "grad_norm": 1.6233388185501099,
      "learning_rate": 9.993061229947591e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 2572200,
      "step": 3905
    },
    {
      "epoch": 2.3054245283018866,
      "grad_norm": 2.978595018386841,
      "learning_rate": 9.992925077487845e-06,
      "loss": 0.691,
      "num_input_tokens_seen": 2576328,
      "step": 3910
    },
    {
      "epoch": 2.308372641509434,
      "grad_norm": 4.070454120635986,
      "learning_rate": 9.992787603129666e-06,
      "loss": 0.3389,
      "num_input_tokens_seen": 2579528,
      "step": 3915
    },
    {
      "epoch": 2.311320754716981,
      "grad_norm": 1.493847131729126,
      "learning_rate": 9.99264880690945e-06,
      "loss": 0.7044,
      "num_input_tokens_seen": 2582920,
      "step": 3920
    },
    {
      "epoch": 2.3142688679245285,
      "grad_norm": 1.646718978881836,
      "learning_rate": 9.992508688863947e-06,
      "loss": 0.496,
      "num_input_tokens_seen": 2585256,
      "step": 3925
    },
    {
      "epoch": 2.3172169811320753,
      "grad_norm": 2.461080312728882,
      "learning_rate": 9.99236724903025e-06,
      "loss": 0.4545,
      "num_input_tokens_seen": 2588008,
      "step": 3930
    },
    {
      "epoch": 2.3201650943396226,
      "grad_norm": 4.3974761962890625,
      "learning_rate": 9.992224487445809e-06,
      "loss": 0.5569,
      "num_input_tokens_seen": 2591304,
      "step": 3935
    },
    {
      "epoch": 2.32311320754717,
      "grad_norm": 4.087823867797852,
      "learning_rate": 9.99208040414842e-06,
      "loss": 0.5325,
      "num_input_tokens_seen": 2594312,
      "step": 3940
    },
    {
      "epoch": 2.326061320754717,
      "grad_norm": 1.8751938343048096,
      "learning_rate": 9.99193499917623e-06,
      "loss": 0.5032,
      "num_input_tokens_seen": 2597832,
      "step": 3945
    },
    {
      "epoch": 2.329009433962264,
      "grad_norm": 3.942800760269165,
      "learning_rate": 9.991788272567735e-06,
      "loss": 0.4063,
      "num_input_tokens_seen": 2601704,
      "step": 3950
    },
    {
      "epoch": 2.3319575471698113,
      "grad_norm": 1.4229016304016113,
      "learning_rate": 9.991640224361781e-06,
      "loss": 0.388,
      "num_input_tokens_seen": 2606184,
      "step": 3955
    },
    {
      "epoch": 2.3349056603773586,
      "grad_norm": 2.075326681137085,
      "learning_rate": 9.991490854597563e-06,
      "loss": 0.4533,
      "num_input_tokens_seen": 2609992,
      "step": 3960
    },
    {
      "epoch": 2.3378537735849054,
      "grad_norm": 2.0213751792907715,
      "learning_rate": 9.991340163314632e-06,
      "loss": 0.5082,
      "num_input_tokens_seen": 2613000,
      "step": 3965
    },
    {
      "epoch": 2.3408018867924527,
      "grad_norm": 3.008556842803955,
      "learning_rate": 9.991188150552878e-06,
      "loss": 0.5198,
      "num_input_tokens_seen": 2615400,
      "step": 3970
    },
    {
      "epoch": 2.34375,
      "grad_norm": 2.0121102333068848,
      "learning_rate": 9.991034816352553e-06,
      "loss": 0.7398,
      "num_input_tokens_seen": 2618056,
      "step": 3975
    },
    {
      "epoch": 2.3466981132075473,
      "grad_norm": 2.3988847732543945,
      "learning_rate": 9.990880160754248e-06,
      "loss": 0.5908,
      "num_input_tokens_seen": 2621288,
      "step": 3980
    },
    {
      "epoch": 2.3496462264150946,
      "grad_norm": 3.6705634593963623,
      "learning_rate": 9.990724183798914e-06,
      "loss": 0.5206,
      "num_input_tokens_seen": 2624424,
      "step": 3985
    },
    {
      "epoch": 2.3525943396226414,
      "grad_norm": 5.486850738525391,
      "learning_rate": 9.990566885527841e-06,
      "loss": 0.5438,
      "num_input_tokens_seen": 2628360,
      "step": 3990
    },
    {
      "epoch": 2.3555424528301887,
      "grad_norm": 1.515970230102539,
      "learning_rate": 9.99040826598268e-06,
      "loss": 0.45,
      "num_input_tokens_seen": 2631432,
      "step": 3995
    },
    {
      "epoch": 2.358490566037736,
      "grad_norm": 1.2861870527267456,
      "learning_rate": 9.990248325205418e-06,
      "loss": 0.4399,
      "num_input_tokens_seen": 2635528,
      "step": 4000
    },
    {
      "epoch": 2.361438679245283,
      "grad_norm": 1.4380439519882202,
      "learning_rate": 9.990087063238408e-06,
      "loss": 0.5747,
      "num_input_tokens_seen": 2639880,
      "step": 4005
    },
    {
      "epoch": 2.36438679245283,
      "grad_norm": 1.5695492029190063,
      "learning_rate": 9.989924480124342e-06,
      "loss": 0.4146,
      "num_input_tokens_seen": 2644360,
      "step": 4010
    },
    {
      "epoch": 2.3673349056603774,
      "grad_norm": 1.554626703262329,
      "learning_rate": 9.989760575906264e-06,
      "loss": 0.5246,
      "num_input_tokens_seen": 2647048,
      "step": 4015
    },
    {
      "epoch": 2.3702830188679247,
      "grad_norm": 2.625765562057495,
      "learning_rate": 9.98959535062757e-06,
      "loss": 0.4869,
      "num_input_tokens_seen": 2650440,
      "step": 4020
    },
    {
      "epoch": 2.3732311320754715,
      "grad_norm": 1.8809514045715332,
      "learning_rate": 9.989428804332002e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 2654472,
      "step": 4025
    },
    {
      "epoch": 2.376179245283019,
      "grad_norm": 3.0745849609375,
      "learning_rate": 9.989260937063656e-06,
      "loss": 0.5308,
      "num_input_tokens_seen": 2661640,
      "step": 4030
    },
    {
      "epoch": 2.379127358490566,
      "grad_norm": 2.4265127182006836,
      "learning_rate": 9.989091748866973e-06,
      "loss": 0.3831,
      "num_input_tokens_seen": 2664680,
      "step": 4035
    },
    {
      "epoch": 2.3820754716981134,
      "grad_norm": 3.9562931060791016,
      "learning_rate": 9.988921239786748e-06,
      "loss": 0.5771,
      "num_input_tokens_seen": 2667624,
      "step": 4040
    },
    {
      "epoch": 2.3850235849056602,
      "grad_norm": 3.7344202995300293,
      "learning_rate": 9.988749409868124e-06,
      "loss": 0.387,
      "num_input_tokens_seen": 2671464,
      "step": 4045
    },
    {
      "epoch": 2.3879716981132075,
      "grad_norm": 3.008343458175659,
      "learning_rate": 9.988576259156593e-06,
      "loss": 0.677,
      "num_input_tokens_seen": 2674760,
      "step": 4050
    },
    {
      "epoch": 2.390919811320755,
      "grad_norm": 1.556858777999878,
      "learning_rate": 9.988401787697996e-06,
      "loss": 0.4774,
      "num_input_tokens_seen": 2678088,
      "step": 4055
    },
    {
      "epoch": 2.393867924528302,
      "grad_norm": 1.2803210020065308,
      "learning_rate": 9.98822599553853e-06,
      "loss": 0.3826,
      "num_input_tokens_seen": 2681864,
      "step": 4060
    },
    {
      "epoch": 2.396816037735849,
      "grad_norm": 2.5849153995513916,
      "learning_rate": 9.988048882724732e-06,
      "loss": 0.4208,
      "num_input_tokens_seen": 2684488,
      "step": 4065
    },
    {
      "epoch": 2.3997641509433962,
      "grad_norm": 2.002643585205078,
      "learning_rate": 9.987870449303497e-06,
      "loss": 0.4784,
      "num_input_tokens_seen": 2687624,
      "step": 4070
    },
    {
      "epoch": 2.4027122641509435,
      "grad_norm": 2.7580080032348633,
      "learning_rate": 9.98769069532206e-06,
      "loss": 0.4295,
      "num_input_tokens_seen": 2691400,
      "step": 4075
    },
    {
      "epoch": 2.4056603773584904,
      "grad_norm": 1.1368303298950195,
      "learning_rate": 9.98750962082802e-06,
      "loss": 0.4634,
      "num_input_tokens_seen": 2695464,
      "step": 4080
    },
    {
      "epoch": 2.4086084905660377,
      "grad_norm": 1.5278986692428589,
      "learning_rate": 9.987327225869312e-06,
      "loss": 0.5222,
      "num_input_tokens_seen": 2698216,
      "step": 4085
    },
    {
      "epoch": 2.411556603773585,
      "grad_norm": 2.005452871322632,
      "learning_rate": 9.987143510494225e-06,
      "loss": 0.4439,
      "num_input_tokens_seen": 2701928,
      "step": 4090
    },
    {
      "epoch": 2.4145047169811322,
      "grad_norm": 3.6434128284454346,
      "learning_rate": 9.9869584747514e-06,
      "loss": 0.4705,
      "num_input_tokens_seen": 2704136,
      "step": 4095
    },
    {
      "epoch": 2.417452830188679,
      "grad_norm": 1.485411286354065,
      "learning_rate": 9.986772118689828e-06,
      "loss": 0.5592,
      "num_input_tokens_seen": 2707560,
      "step": 4100
    },
    {
      "epoch": 2.4204009433962264,
      "grad_norm": 1.6012200117111206,
      "learning_rate": 9.986584442358845e-06,
      "loss": 0.4479,
      "num_input_tokens_seen": 2711208,
      "step": 4105
    },
    {
      "epoch": 2.4233490566037736,
      "grad_norm": 2.5406899452209473,
      "learning_rate": 9.98639544580814e-06,
      "loss": 0.4091,
      "num_input_tokens_seen": 2714024,
      "step": 4110
    },
    {
      "epoch": 2.426297169811321,
      "grad_norm": 1.7301311492919922,
      "learning_rate": 9.98620512908775e-06,
      "loss": 0.4409,
      "num_input_tokens_seen": 2717288,
      "step": 4115
    },
    {
      "epoch": 2.4292452830188678,
      "grad_norm": 25.11029624938965,
      "learning_rate": 9.986013492248064e-06,
      "loss": 0.5263,
      "num_input_tokens_seen": 2720072,
      "step": 4120
    },
    {
      "epoch": 2.432193396226415,
      "grad_norm": 1.5920137166976929,
      "learning_rate": 9.985820535339817e-06,
      "loss": 0.4058,
      "num_input_tokens_seen": 2723048,
      "step": 4125
    },
    {
      "epoch": 2.4351415094339623,
      "grad_norm": 2.7993741035461426,
      "learning_rate": 9.985626258414093e-06,
      "loss": 0.5037,
      "num_input_tokens_seen": 2726184,
      "step": 4130
    },
    {
      "epoch": 2.4380896226415096,
      "grad_norm": 3.084409475326538,
      "learning_rate": 9.985430661522333e-06,
      "loss": 0.4961,
      "num_input_tokens_seen": 2728968,
      "step": 4135
    },
    {
      "epoch": 2.4410377358490565,
      "grad_norm": 3.027163505554199,
      "learning_rate": 9.98523374471632e-06,
      "loss": 0.4729,
      "num_input_tokens_seen": 2732520,
      "step": 4140
    },
    {
      "epoch": 2.4439858490566038,
      "grad_norm": 9.03763484954834,
      "learning_rate": 9.985035508048186e-06,
      "loss": 0.5134,
      "num_input_tokens_seen": 2735624,
      "step": 4145
    },
    {
      "epoch": 2.446933962264151,
      "grad_norm": 4.393592357635498,
      "learning_rate": 9.984835951570417e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 2738536,
      "step": 4150
    },
    {
      "epoch": 2.449882075471698,
      "grad_norm": 1.9219294786453247,
      "learning_rate": 9.984635075335847e-06,
      "loss": 0.4454,
      "num_input_tokens_seen": 2741480,
      "step": 4155
    },
    {
      "epoch": 2.452830188679245,
      "grad_norm": 1.3858346939086914,
      "learning_rate": 9.984432879397659e-06,
      "loss": 0.5333,
      "num_input_tokens_seen": 2745480,
      "step": 4160
    },
    {
      "epoch": 2.4557783018867925,
      "grad_norm": 2.187110185623169,
      "learning_rate": 9.984229363809383e-06,
      "loss": 0.4846,
      "num_input_tokens_seen": 2749224,
      "step": 4165
    },
    {
      "epoch": 2.4587264150943398,
      "grad_norm": 5.269033908843994,
      "learning_rate": 9.984024528624904e-06,
      "loss": 0.5522,
      "num_input_tokens_seen": 2752392,
      "step": 4170
    },
    {
      "epoch": 2.4616745283018866,
      "grad_norm": 1.687891960144043,
      "learning_rate": 9.98381837389845e-06,
      "loss": 0.5109,
      "num_input_tokens_seen": 2755624,
      "step": 4175
    },
    {
      "epoch": 2.464622641509434,
      "grad_norm": 1.7017974853515625,
      "learning_rate": 9.983610899684601e-06,
      "loss": 0.4611,
      "num_input_tokens_seen": 2759560,
      "step": 4180
    },
    {
      "epoch": 2.467570754716981,
      "grad_norm": 1.7831724882125854,
      "learning_rate": 9.983402106038292e-06,
      "loss": 0.6268,
      "num_input_tokens_seen": 2762664,
      "step": 4185
    },
    {
      "epoch": 2.4705188679245285,
      "grad_norm": 1.8225523233413696,
      "learning_rate": 9.983191993014793e-06,
      "loss": 0.5182,
      "num_input_tokens_seen": 2766760,
      "step": 4190
    },
    {
      "epoch": 2.4734669811320753,
      "grad_norm": 2.9301183223724365,
      "learning_rate": 9.982980560669742e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 2769832,
      "step": 4195
    },
    {
      "epoch": 2.4764150943396226,
      "grad_norm": 2.6149425506591797,
      "learning_rate": 9.98276780905911e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 2772488,
      "step": 4200
    },
    {
      "epoch": 2.47936320754717,
      "grad_norm": 1.4796416759490967,
      "learning_rate": 9.982553738239225e-06,
      "loss": 0.5507,
      "num_input_tokens_seen": 2776008,
      "step": 4205
    },
    {
      "epoch": 2.482311320754717,
      "grad_norm": 1.2229875326156616,
      "learning_rate": 9.982338348266766e-06,
      "loss": 0.4561,
      "num_input_tokens_seen": 2779560,
      "step": 4210
    },
    {
      "epoch": 2.485259433962264,
      "grad_norm": 3.134162187576294,
      "learning_rate": 9.982121639198756e-06,
      "loss": 0.4362,
      "num_input_tokens_seen": 2782152,
      "step": 4215
    },
    {
      "epoch": 2.4882075471698113,
      "grad_norm": 2.455576181411743,
      "learning_rate": 9.98190361109257e-06,
      "loss": 0.6411,
      "num_input_tokens_seen": 2785352,
      "step": 4220
    },
    {
      "epoch": 2.4911556603773586,
      "grad_norm": 2.387749433517456,
      "learning_rate": 9.981684264005934e-06,
      "loss": 0.5097,
      "num_input_tokens_seen": 2788456,
      "step": 4225
    },
    {
      "epoch": 2.4941037735849054,
      "grad_norm": 3.23799991607666,
      "learning_rate": 9.981463597996917e-06,
      "loss": 0.4753,
      "num_input_tokens_seen": 2792232,
      "step": 4230
    },
    {
      "epoch": 2.4970518867924527,
      "grad_norm": 2.3035972118377686,
      "learning_rate": 9.981241613123944e-06,
      "loss": 0.4779,
      "num_input_tokens_seen": 2796008,
      "step": 4235
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.2560789585113525,
      "learning_rate": 9.981018309445785e-06,
      "loss": 0.5288,
      "num_input_tokens_seen": 2798792,
      "step": 4240
    },
    {
      "epoch": 2.5029481132075473,
      "grad_norm": 2.1265058517456055,
      "learning_rate": 9.980793687021564e-06,
      "loss": 0.4292,
      "num_input_tokens_seen": 2802344,
      "step": 4245
    },
    {
      "epoch": 2.5058962264150946,
      "grad_norm": 1.7708169221878052,
      "learning_rate": 9.980567745910746e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 2805288,
      "step": 4250
    },
    {
      "epoch": 2.5088443396226414,
      "grad_norm": 3.2203383445739746,
      "learning_rate": 9.980340486173155e-06,
      "loss": 0.3831,
      "num_input_tokens_seen": 2808648,
      "step": 4255
    },
    {
      "epoch": 2.5117924528301887,
      "grad_norm": 1.4237184524536133,
      "learning_rate": 9.980111907868954e-06,
      "loss": 0.3639,
      "num_input_tokens_seen": 2811496,
      "step": 4260
    },
    {
      "epoch": 2.514740566037736,
      "grad_norm": 1.8509389162063599,
      "learning_rate": 9.979882011058662e-06,
      "loss": 0.5316,
      "num_input_tokens_seen": 2814024,
      "step": 4265
    },
    {
      "epoch": 2.517688679245283,
      "grad_norm": 2.129279375076294,
      "learning_rate": 9.979650795803146e-06,
      "loss": 0.5004,
      "num_input_tokens_seen": 2817896,
      "step": 4270
    },
    {
      "epoch": 2.52063679245283,
      "grad_norm": 1.507406234741211,
      "learning_rate": 9.979418262163621e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 2821416,
      "step": 4275
    },
    {
      "epoch": 2.5235849056603774,
      "grad_norm": 2.6295461654663086,
      "learning_rate": 9.979184410201652e-06,
      "loss": 0.4974,
      "num_input_tokens_seen": 2824456,
      "step": 4280
    },
    {
      "epoch": 2.5265330188679247,
      "grad_norm": 2.2110230922698975,
      "learning_rate": 9.97894923997915e-06,
      "loss": 0.3977,
      "num_input_tokens_seen": 2827496,
      "step": 4285
    },
    {
      "epoch": 2.5294811320754715,
      "grad_norm": 5.1190972328186035,
      "learning_rate": 9.97871275155838e-06,
      "loss": 0.5493,
      "num_input_tokens_seen": 2830024,
      "step": 4290
    },
    {
      "epoch": 2.532429245283019,
      "grad_norm": 0.9987030625343323,
      "learning_rate": 9.978474945001949e-06,
      "loss": 0.4841,
      "num_input_tokens_seen": 2833416,
      "step": 4295
    },
    {
      "epoch": 2.535377358490566,
      "grad_norm": 1.6966310739517212,
      "learning_rate": 9.978235820372822e-06,
      "loss": 0.3385,
      "num_input_tokens_seen": 2836648,
      "step": 4300
    },
    {
      "epoch": 2.538325471698113,
      "grad_norm": 2.156578540802002,
      "learning_rate": 9.977995377734307e-06,
      "loss": 0.398,
      "num_input_tokens_seen": 2839176,
      "step": 4305
    },
    {
      "epoch": 2.5412735849056602,
      "grad_norm": 1.7088420391082764,
      "learning_rate": 9.977753617150061e-06,
      "loss": 0.4788,
      "num_input_tokens_seen": 2842600,
      "step": 4310
    },
    {
      "epoch": 2.5442216981132075,
      "grad_norm": 1.546927571296692,
      "learning_rate": 9.977510538684094e-06,
      "loss": 0.5259,
      "num_input_tokens_seen": 2845576,
      "step": 4315
    },
    {
      "epoch": 2.547169811320755,
      "grad_norm": 3.8812224864959717,
      "learning_rate": 9.977266142400757e-06,
      "loss": 0.3801,
      "num_input_tokens_seen": 2848360,
      "step": 4320
    },
    {
      "epoch": 2.550117924528302,
      "grad_norm": 1.6892043352127075,
      "learning_rate": 9.977020428364759e-06,
      "loss": 0.5354,
      "num_input_tokens_seen": 2851624,
      "step": 4325
    },
    {
      "epoch": 2.553066037735849,
      "grad_norm": 2.320016860961914,
      "learning_rate": 9.976773396641154e-06,
      "loss": 0.4972,
      "num_input_tokens_seen": 2855720,
      "step": 4330
    },
    {
      "epoch": 2.5560141509433962,
      "grad_norm": 5.302420139312744,
      "learning_rate": 9.976525047295342e-06,
      "loss": 0.5182,
      "num_input_tokens_seen": 2858856,
      "step": 4335
    },
    {
      "epoch": 2.5589622641509435,
      "grad_norm": 1.4778472185134888,
      "learning_rate": 9.976275380393077e-06,
      "loss": 0.474,
      "num_input_tokens_seen": 2861960,
      "step": 4340
    },
    {
      "epoch": 2.5619103773584904,
      "grad_norm": 2.552856922149658,
      "learning_rate": 9.976024396000459e-06,
      "loss": 0.5053,
      "num_input_tokens_seen": 2865160,
      "step": 4345
    },
    {
      "epoch": 2.5648584905660377,
      "grad_norm": 2.2262771129608154,
      "learning_rate": 9.975772094183935e-06,
      "loss": 0.4972,
      "num_input_tokens_seen": 2869352,
      "step": 4350
    },
    {
      "epoch": 2.567806603773585,
      "grad_norm": 2.12382173538208,
      "learning_rate": 9.975518475010306e-06,
      "loss": 0.3867,
      "num_input_tokens_seen": 2872360,
      "step": 4355
    },
    {
      "epoch": 2.5707547169811322,
      "grad_norm": 2.982560396194458,
      "learning_rate": 9.975263538546717e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 2875528,
      "step": 4360
    },
    {
      "epoch": 2.5737028301886795,
      "grad_norm": 1.7153925895690918,
      "learning_rate": 9.975007284860664e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 2879944,
      "step": 4365
    },
    {
      "epoch": 2.5766509433962264,
      "grad_norm": 2.330193519592285,
      "learning_rate": 9.974749714019993e-06,
      "loss": 0.537,
      "num_input_tokens_seen": 2882696,
      "step": 4370
    },
    {
      "epoch": 2.5795990566037736,
      "grad_norm": 2.2447235584259033,
      "learning_rate": 9.974490826092894e-06,
      "loss": 0.4226,
      "num_input_tokens_seen": 2886408,
      "step": 4375
    },
    {
      "epoch": 2.5825471698113205,
      "grad_norm": 2.0150246620178223,
      "learning_rate": 9.974230621147907e-06,
      "loss": 0.3779,
      "num_input_tokens_seen": 2890408,
      "step": 4380
    },
    {
      "epoch": 2.5854952830188678,
      "grad_norm": 3.2024435997009277,
      "learning_rate": 9.973969099253928e-06,
      "loss": 0.4925,
      "num_input_tokens_seen": 2893512,
      "step": 4385
    },
    {
      "epoch": 2.588443396226415,
      "grad_norm": 1.7686220407485962,
      "learning_rate": 9.973706260480194e-06,
      "loss": 0.404,
      "num_input_tokens_seen": 2896712,
      "step": 4390
    },
    {
      "epoch": 2.5913915094339623,
      "grad_norm": 3.298051357269287,
      "learning_rate": 9.97344210489629e-06,
      "loss": 0.4259,
      "num_input_tokens_seen": 2899048,
      "step": 4395
    },
    {
      "epoch": 2.5943396226415096,
      "grad_norm": 1.6972724199295044,
      "learning_rate": 9.973176632572158e-06,
      "loss": 0.404,
      "num_input_tokens_seen": 2902248,
      "step": 4400
    },
    {
      "epoch": 2.5972877358490565,
      "grad_norm": 1.8643391132354736,
      "learning_rate": 9.972909843578076e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 2906184,
      "step": 4405
    },
    {
      "epoch": 2.6002358490566038,
      "grad_norm": 2.334367513656616,
      "learning_rate": 9.972641737984681e-06,
      "loss": 0.3976,
      "num_input_tokens_seen": 2910152,
      "step": 4410
    },
    {
      "epoch": 2.603183962264151,
      "grad_norm": 1.4338387250900269,
      "learning_rate": 9.972372315862956e-06,
      "loss": 0.483,
      "num_input_tokens_seen": 2913736,
      "step": 4415
    },
    {
      "epoch": 2.606132075471698,
      "grad_norm": 2.110466718673706,
      "learning_rate": 9.97210157728423e-06,
      "loss": 0.4722,
      "num_input_tokens_seen": 2917096,
      "step": 4420
    },
    {
      "epoch": 2.609080188679245,
      "grad_norm": 1.5699862241744995,
      "learning_rate": 9.971829522320185e-06,
      "loss": 0.6214,
      "num_input_tokens_seen": 2921128,
      "step": 4425
    },
    {
      "epoch": 2.6120283018867925,
      "grad_norm": 1.9804028272628784,
      "learning_rate": 9.971556151042843e-06,
      "loss": 0.568,
      "num_input_tokens_seen": 2924904,
      "step": 4430
    },
    {
      "epoch": 2.6149764150943398,
      "grad_norm": 2.005406141281128,
      "learning_rate": 9.971281463524588e-06,
      "loss": 0.4269,
      "num_input_tokens_seen": 2927976,
      "step": 4435
    },
    {
      "epoch": 2.617924528301887,
      "grad_norm": 2.3433732986450195,
      "learning_rate": 9.971005459838136e-06,
      "loss": 0.5383,
      "num_input_tokens_seen": 2931176,
      "step": 4440
    },
    {
      "epoch": 2.620872641509434,
      "grad_norm": 1.6776050329208374,
      "learning_rate": 9.970728140056567e-06,
      "loss": 0.5067,
      "num_input_tokens_seen": 2934952,
      "step": 4445
    },
    {
      "epoch": 2.623820754716981,
      "grad_norm": 1.4501911401748657,
      "learning_rate": 9.9704495042533e-06,
      "loss": 0.599,
      "num_input_tokens_seen": 2938408,
      "step": 4450
    },
    {
      "epoch": 2.6267688679245285,
      "grad_norm": 1.9769806861877441,
      "learning_rate": 9.970169552502105e-06,
      "loss": 0.4809,
      "num_input_tokens_seen": 2941768,
      "step": 4455
    },
    {
      "epoch": 2.6297169811320753,
      "grad_norm": 5.118475914001465,
      "learning_rate": 9.969888284877102e-06,
      "loss": 0.4068,
      "num_input_tokens_seen": 2944840,
      "step": 4460
    },
    {
      "epoch": 2.6326650943396226,
      "grad_norm": 2.871044635772705,
      "learning_rate": 9.969605701452757e-06,
      "loss": 0.509,
      "num_input_tokens_seen": 2948168,
      "step": 4465
    },
    {
      "epoch": 2.63561320754717,
      "grad_norm": 2.7484588623046875,
      "learning_rate": 9.969321802303882e-06,
      "loss": 0.4817,
      "num_input_tokens_seen": 2951528,
      "step": 4470
    },
    {
      "epoch": 2.638561320754717,
      "grad_norm": 2.1320464611053467,
      "learning_rate": 9.969036587505644e-06,
      "loss": 0.6412,
      "num_input_tokens_seen": 2954696,
      "step": 4475
    },
    {
      "epoch": 2.641509433962264,
      "grad_norm": 3.958263635635376,
      "learning_rate": 9.968750057133555e-06,
      "loss": 0.593,
      "num_input_tokens_seen": 2958248,
      "step": 4480
    },
    {
      "epoch": 2.6444575471698113,
      "grad_norm": 1.9238780736923218,
      "learning_rate": 9.968462211263474e-06,
      "loss": 0.588,
      "num_input_tokens_seen": 2961256,
      "step": 4485
    },
    {
      "epoch": 2.6474056603773586,
      "grad_norm": 3.292086124420166,
      "learning_rate": 9.96817304997161e-06,
      "loss": 0.4894,
      "num_input_tokens_seen": 2963816,
      "step": 4490
    },
    {
      "epoch": 2.6503537735849054,
      "grad_norm": 3.2834489345550537,
      "learning_rate": 9.967882573334519e-06,
      "loss": 0.5316,
      "num_input_tokens_seen": 2966728,
      "step": 4495
    },
    {
      "epoch": 2.6533018867924527,
      "grad_norm": 1.2596665620803833,
      "learning_rate": 9.967590781429106e-06,
      "loss": 0.5928,
      "num_input_tokens_seen": 2970376,
      "step": 4500
    },
    {
      "epoch": 2.65625,
      "grad_norm": 1.8423634767532349,
      "learning_rate": 9.967297674332625e-06,
      "loss": 0.6376,
      "num_input_tokens_seen": 2972840,
      "step": 4505
    },
    {
      "epoch": 2.6591981132075473,
      "grad_norm": 0.9945535063743591,
      "learning_rate": 9.967003252122675e-06,
      "loss": 0.4021,
      "num_input_tokens_seen": 2976360,
      "step": 4510
    },
    {
      "epoch": 2.6621462264150946,
      "grad_norm": 2.490900754928589,
      "learning_rate": 9.96670751487721e-06,
      "loss": 0.6066,
      "num_input_tokens_seen": 2978888,
      "step": 4515
    },
    {
      "epoch": 2.6650943396226414,
      "grad_norm": 3.6023502349853516,
      "learning_rate": 9.966410462674525e-06,
      "loss": 0.4737,
      "num_input_tokens_seen": 2981608,
      "step": 4520
    },
    {
      "epoch": 2.6680424528301887,
      "grad_norm": 1.72983717918396,
      "learning_rate": 9.966112095593264e-06,
      "loss": 0.4,
      "num_input_tokens_seen": 2984648,
      "step": 4525
    },
    {
      "epoch": 2.670990566037736,
      "grad_norm": 2.7108218669891357,
      "learning_rate": 9.965812413712425e-06,
      "loss": 0.4497,
      "num_input_tokens_seen": 2988168,
      "step": 4530
    },
    {
      "epoch": 2.673938679245283,
      "grad_norm": 1.808093786239624,
      "learning_rate": 9.965511417111346e-06,
      "loss": 0.504,
      "num_input_tokens_seen": 2991240,
      "step": 4535
    },
    {
      "epoch": 2.67688679245283,
      "grad_norm": 2.5136258602142334,
      "learning_rate": 9.96520910586972e-06,
      "loss": 0.5166,
      "num_input_tokens_seen": 2993800,
      "step": 4540
    },
    {
      "epoch": 2.6798349056603774,
      "grad_norm": 3.0255253314971924,
      "learning_rate": 9.964905480067585e-06,
      "loss": 0.7833,
      "num_input_tokens_seen": 2997032,
      "step": 4545
    },
    {
      "epoch": 2.6827830188679247,
      "grad_norm": 1.6684508323669434,
      "learning_rate": 9.964600539785328e-06,
      "loss": 0.535,
      "num_input_tokens_seen": 3000552,
      "step": 4550
    },
    {
      "epoch": 2.6857311320754715,
      "grad_norm": 1.751739501953125,
      "learning_rate": 9.96429428510368e-06,
      "loss": 0.4483,
      "num_input_tokens_seen": 3004168,
      "step": 4555
    },
    {
      "epoch": 2.688679245283019,
      "grad_norm": 6.161697864532471,
      "learning_rate": 9.963986716103724e-06,
      "loss": 0.6757,
      "num_input_tokens_seen": 3006440,
      "step": 4560
    },
    {
      "epoch": 2.691627358490566,
      "grad_norm": 2.1169791221618652,
      "learning_rate": 9.963677832866893e-06,
      "loss": 0.4026,
      "num_input_tokens_seen": 3009736,
      "step": 4565
    },
    {
      "epoch": 2.694575471698113,
      "grad_norm": 1.9320697784423828,
      "learning_rate": 9.963367635474962e-06,
      "loss": 0.5164,
      "num_input_tokens_seen": 3012616,
      "step": 4570
    },
    {
      "epoch": 2.6975235849056602,
      "grad_norm": 2.2091867923736572,
      "learning_rate": 9.96305612401006e-06,
      "loss": 0.5754,
      "num_input_tokens_seen": 3015784,
      "step": 4575
    },
    {
      "epoch": 2.7004716981132075,
      "grad_norm": 1.2282915115356445,
      "learning_rate": 9.96274329855466e-06,
      "loss": 0.4224,
      "num_input_tokens_seen": 3019048,
      "step": 4580
    },
    {
      "epoch": 2.703419811320755,
      "grad_norm": 1.590360403060913,
      "learning_rate": 9.962429159191583e-06,
      "loss": 0.4968,
      "num_input_tokens_seen": 3021896,
      "step": 4585
    },
    {
      "epoch": 2.706367924528302,
      "grad_norm": 2.58952260017395,
      "learning_rate": 9.962113706003997e-06,
      "loss": 0.4146,
      "num_input_tokens_seen": 3024904,
      "step": 4590
    },
    {
      "epoch": 2.709316037735849,
      "grad_norm": 2.1218931674957275,
      "learning_rate": 9.961796939075424e-06,
      "loss": 0.5484,
      "num_input_tokens_seen": 3027720,
      "step": 4595
    },
    {
      "epoch": 2.7122641509433962,
      "grad_norm": 2.763819932937622,
      "learning_rate": 9.961478858489728e-06,
      "loss": 0.5156,
      "num_input_tokens_seen": 3030248,
      "step": 4600
    },
    {
      "epoch": 2.7152122641509435,
      "grad_norm": 2.269247055053711,
      "learning_rate": 9.961159464331119e-06,
      "loss": 0.6219,
      "num_input_tokens_seen": 3034088,
      "step": 4605
    },
    {
      "epoch": 2.7181603773584904,
      "grad_norm": 2.165090322494507,
      "learning_rate": 9.960838756684161e-06,
      "loss": 0.438,
      "num_input_tokens_seen": 3037064,
      "step": 4610
    },
    {
      "epoch": 2.7211084905660377,
      "grad_norm": 1.1724038124084473,
      "learning_rate": 9.960516735633764e-06,
      "loss": 0.4455,
      "num_input_tokens_seen": 3040200,
      "step": 4615
    },
    {
      "epoch": 2.724056603773585,
      "grad_norm": 1.6792476177215576,
      "learning_rate": 9.960193401265181e-06,
      "loss": 0.6147,
      "num_input_tokens_seen": 3043912,
      "step": 4620
    },
    {
      "epoch": 2.7270047169811322,
      "grad_norm": 2.0050954818725586,
      "learning_rate": 9.959868753664018e-06,
      "loss": 0.566,
      "num_input_tokens_seen": 3050568,
      "step": 4625
    },
    {
      "epoch": 2.7299528301886795,
      "grad_norm": 0.825964629650116,
      "learning_rate": 9.959542792916227e-06,
      "loss": 0.437,
      "num_input_tokens_seen": 3054856,
      "step": 4630
    },
    {
      "epoch": 2.7329009433962264,
      "grad_norm": 2.035921573638916,
      "learning_rate": 9.959215519108108e-06,
      "loss": 0.4886,
      "num_input_tokens_seen": 3057736,
      "step": 4635
    },
    {
      "epoch": 2.7358490566037736,
      "grad_norm": 2.0604443550109863,
      "learning_rate": 9.958886932326306e-06,
      "loss": 0.5056,
      "num_input_tokens_seen": 3061384,
      "step": 4640
    },
    {
      "epoch": 2.7387971698113205,
      "grad_norm": 1.5592554807662964,
      "learning_rate": 9.958557032657817e-06,
      "loss": 0.5473,
      "num_input_tokens_seen": 3065288,
      "step": 4645
    },
    {
      "epoch": 2.7417452830188678,
      "grad_norm": 3.6112709045410156,
      "learning_rate": 9.958225820189984e-06,
      "loss": 0.5462,
      "num_input_tokens_seen": 3068136,
      "step": 4650
    },
    {
      "epoch": 2.744693396226415,
      "grad_norm": 1.7489184141159058,
      "learning_rate": 9.957893295010495e-06,
      "loss": 0.4807,
      "num_input_tokens_seen": 3070856,
      "step": 4655
    },
    {
      "epoch": 2.7476415094339623,
      "grad_norm": 6.506619453430176,
      "learning_rate": 9.957559457207391e-06,
      "loss": 0.4584,
      "num_input_tokens_seen": 3073448,
      "step": 4660
    },
    {
      "epoch": 2.7505896226415096,
      "grad_norm": 1.7130005359649658,
      "learning_rate": 9.957224306869053e-06,
      "loss": 0.5709,
      "num_input_tokens_seen": 3077256,
      "step": 4665
    },
    {
      "epoch": 2.7535377358490565,
      "grad_norm": 1.4348260164260864,
      "learning_rate": 9.956887844084216e-06,
      "loss": 0.4162,
      "num_input_tokens_seen": 3080840,
      "step": 4670
    },
    {
      "epoch": 2.7564858490566038,
      "grad_norm": 3.450330972671509,
      "learning_rate": 9.956550068941958e-06,
      "loss": 0.4649,
      "num_input_tokens_seen": 3083784,
      "step": 4675
    },
    {
      "epoch": 2.759433962264151,
      "grad_norm": 3.3960325717926025,
      "learning_rate": 9.95621098153171e-06,
      "loss": 0.5056,
      "num_input_tokens_seen": 3086888,
      "step": 4680
    },
    {
      "epoch": 2.762382075471698,
      "grad_norm": 1.3809740543365479,
      "learning_rate": 9.955870581943243e-06,
      "loss": 0.483,
      "num_input_tokens_seen": 3090024,
      "step": 4685
    },
    {
      "epoch": 2.765330188679245,
      "grad_norm": 1.5376495122909546,
      "learning_rate": 9.955528870266681e-06,
      "loss": 0.3514,
      "num_input_tokens_seen": 3094056,
      "step": 4690
    },
    {
      "epoch": 2.7682783018867925,
      "grad_norm": 2.124464511871338,
      "learning_rate": 9.955185846592495e-06,
      "loss": 0.4796,
      "num_input_tokens_seen": 3096904,
      "step": 4695
    },
    {
      "epoch": 2.7712264150943398,
      "grad_norm": 2.4395151138305664,
      "learning_rate": 9.9548415110115e-06,
      "loss": 0.3768,
      "num_input_tokens_seen": 3100232,
      "step": 4700
    },
    {
      "epoch": 2.774174528301887,
      "grad_norm": 2.1577095985412598,
      "learning_rate": 9.95449586361486e-06,
      "loss": 0.4237,
      "num_input_tokens_seen": 3102888,
      "step": 4705
    },
    {
      "epoch": 2.777122641509434,
      "grad_norm": 1.566810965538025,
      "learning_rate": 9.954148904494085e-06,
      "loss": 0.4489,
      "num_input_tokens_seen": 3111336,
      "step": 4710
    },
    {
      "epoch": 2.780070754716981,
      "grad_norm": 1.6114418506622314,
      "learning_rate": 9.95380063374104e-06,
      "loss": 0.4105,
      "num_input_tokens_seen": 3114440,
      "step": 4715
    },
    {
      "epoch": 2.7830188679245285,
      "grad_norm": 4.0832929611206055,
      "learning_rate": 9.953451051447927e-06,
      "loss": 0.6385,
      "num_input_tokens_seen": 3117896,
      "step": 4720
    },
    {
      "epoch": 2.7859669811320753,
      "grad_norm": 2.3294880390167236,
      "learning_rate": 9.953100157707299e-06,
      "loss": 0.4559,
      "num_input_tokens_seen": 3120776,
      "step": 4725
    },
    {
      "epoch": 2.7889150943396226,
      "grad_norm": 3.9324138164520264,
      "learning_rate": 9.952747952612056e-06,
      "loss": 0.4768,
      "num_input_tokens_seen": 3123880,
      "step": 4730
    },
    {
      "epoch": 2.79186320754717,
      "grad_norm": 1.5330997705459595,
      "learning_rate": 9.952394436255451e-06,
      "loss": 0.4645,
      "num_input_tokens_seen": 3128040,
      "step": 4735
    },
    {
      "epoch": 2.794811320754717,
      "grad_norm": 1.4479578733444214,
      "learning_rate": 9.952039608731072e-06,
      "loss": 0.4071,
      "num_input_tokens_seen": 3131176,
      "step": 4740
    },
    {
      "epoch": 2.797759433962264,
      "grad_norm": 2.3454015254974365,
      "learning_rate": 9.951683470132868e-06,
      "loss": 0.5153,
      "num_input_tokens_seen": 3134024,
      "step": 4745
    },
    {
      "epoch": 2.8007075471698113,
      "grad_norm": 3.487772226333618,
      "learning_rate": 9.951326020555122e-06,
      "loss": 0.7633,
      "num_input_tokens_seen": 3137832,
      "step": 4750
    },
    {
      "epoch": 2.8036556603773586,
      "grad_norm": 2.5226452350616455,
      "learning_rate": 9.950967260092473e-06,
      "loss": 0.4172,
      "num_input_tokens_seen": 3140744,
      "step": 4755
    },
    {
      "epoch": 2.8066037735849054,
      "grad_norm": 3.895587921142578,
      "learning_rate": 9.950607188839905e-06,
      "loss": 0.3426,
      "num_input_tokens_seen": 3145000,
      "step": 4760
    },
    {
      "epoch": 2.8095518867924527,
      "grad_norm": 1.326190710067749,
      "learning_rate": 9.950245806892749e-06,
      "loss": 0.4432,
      "num_input_tokens_seen": 3148232,
      "step": 4765
    },
    {
      "epoch": 2.8125,
      "grad_norm": 1.6042728424072266,
      "learning_rate": 9.94988311434668e-06,
      "loss": 0.4699,
      "num_input_tokens_seen": 3151272,
      "step": 4770
    },
    {
      "epoch": 2.8154481132075473,
      "grad_norm": 2.171055793762207,
      "learning_rate": 9.949519111297723e-06,
      "loss": 0.623,
      "num_input_tokens_seen": 3153480,
      "step": 4775
    },
    {
      "epoch": 2.8183962264150946,
      "grad_norm": 2.0972671508789062,
      "learning_rate": 9.949153797842252e-06,
      "loss": 0.4967,
      "num_input_tokens_seen": 3156360,
      "step": 4780
    },
    {
      "epoch": 2.8213443396226414,
      "grad_norm": 2.2818446159362793,
      "learning_rate": 9.948787174076982e-06,
      "loss": 0.469,
      "num_input_tokens_seen": 3161448,
      "step": 4785
    },
    {
      "epoch": 2.8242924528301887,
      "grad_norm": 2.3294730186462402,
      "learning_rate": 9.948419240098978e-06,
      "loss": 0.5256,
      "num_input_tokens_seen": 3164040,
      "step": 4790
    },
    {
      "epoch": 2.827240566037736,
      "grad_norm": 1.7556889057159424,
      "learning_rate": 9.948049996005657e-06,
      "loss": 0.4954,
      "num_input_tokens_seen": 3166920,
      "step": 4795
    },
    {
      "epoch": 2.830188679245283,
      "grad_norm": 5.244544982910156,
      "learning_rate": 9.947679441894773e-06,
      "loss": 0.4903,
      "num_input_tokens_seen": 3169352,
      "step": 4800
    },
    {
      "epoch": 2.83313679245283,
      "grad_norm": 2.484682083129883,
      "learning_rate": 9.947307577864433e-06,
      "loss": 0.5239,
      "num_input_tokens_seen": 3173608,
      "step": 4805
    },
    {
      "epoch": 2.8360849056603774,
      "grad_norm": 1.371943473815918,
      "learning_rate": 9.94693440401309e-06,
      "loss": 0.5856,
      "num_input_tokens_seen": 3177000,
      "step": 4810
    },
    {
      "epoch": 2.8390330188679247,
      "grad_norm": 2.280512571334839,
      "learning_rate": 9.946559920439545e-06,
      "loss": 0.4848,
      "num_input_tokens_seen": 3180712,
      "step": 4815
    },
    {
      "epoch": 2.8419811320754715,
      "grad_norm": 4.084051609039307,
      "learning_rate": 9.946184127242942e-06,
      "loss": 0.49,
      "num_input_tokens_seen": 3183368,
      "step": 4820
    },
    {
      "epoch": 2.844929245283019,
      "grad_norm": 1.5862314701080322,
      "learning_rate": 9.945807024522774e-06,
      "loss": 0.7982,
      "num_input_tokens_seen": 3187560,
      "step": 4825
    },
    {
      "epoch": 2.847877358490566,
      "grad_norm": 1.4235109090805054,
      "learning_rate": 9.945428612378881e-06,
      "loss": 0.5629,
      "num_input_tokens_seen": 3191144,
      "step": 4830
    },
    {
      "epoch": 2.850825471698113,
      "grad_norm": 2.975647449493408,
      "learning_rate": 9.94504889091145e-06,
      "loss": 0.6127,
      "num_input_tokens_seen": 3194120,
      "step": 4835
    },
    {
      "epoch": 2.8537735849056602,
      "grad_norm": 2.699789047241211,
      "learning_rate": 9.944667860221013e-06,
      "loss": 0.5936,
      "num_input_tokens_seen": 3197800,
      "step": 4840
    },
    {
      "epoch": 2.8567216981132075,
      "grad_norm": 1.0659258365631104,
      "learning_rate": 9.944285520408448e-06,
      "loss": 0.4644,
      "num_input_tokens_seen": 3200936,
      "step": 4845
    },
    {
      "epoch": 2.859669811320755,
      "grad_norm": 1.3656165599822998,
      "learning_rate": 9.943901871574984e-06,
      "loss": 0.5222,
      "num_input_tokens_seen": 3204104,
      "step": 4850
    },
    {
      "epoch": 2.862617924528302,
      "grad_norm": 2.0803208351135254,
      "learning_rate": 9.943516913822192e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 3206536,
      "step": 4855
    },
    {
      "epoch": 2.865566037735849,
      "grad_norm": 2.3221771717071533,
      "learning_rate": 9.943130647251994e-06,
      "loss": 0.4468,
      "num_input_tokens_seen": 3209576,
      "step": 4860
    },
    {
      "epoch": 2.8685141509433962,
      "grad_norm": 2.3049073219299316,
      "learning_rate": 9.94274307196665e-06,
      "loss": 0.5257,
      "num_input_tokens_seen": 3212136,
      "step": 4865
    },
    {
      "epoch": 2.8714622641509435,
      "grad_norm": 1.7633936405181885,
      "learning_rate": 9.942354188068778e-06,
      "loss": 0.6088,
      "num_input_tokens_seen": 3215240,
      "step": 4870
    },
    {
      "epoch": 2.8744103773584904,
      "grad_norm": 6.412152290344238,
      "learning_rate": 9.941963995661333e-06,
      "loss": 0.595,
      "num_input_tokens_seen": 3218184,
      "step": 4875
    },
    {
      "epoch": 2.8773584905660377,
      "grad_norm": 2.0356063842773438,
      "learning_rate": 9.941572494847622e-06,
      "loss": 0.6327,
      "num_input_tokens_seen": 3221928,
      "step": 4880
    },
    {
      "epoch": 2.880306603773585,
      "grad_norm": 2.5995659828186035,
      "learning_rate": 9.941179685731297e-06,
      "loss": 0.4573,
      "num_input_tokens_seen": 3224840,
      "step": 4885
    },
    {
      "epoch": 2.8832547169811322,
      "grad_norm": 2.1424007415771484,
      "learning_rate": 9.940785568416354e-06,
      "loss": 0.3782,
      "num_input_tokens_seen": 3228008,
      "step": 4890
    },
    {
      "epoch": 2.8862028301886795,
      "grad_norm": 2.438298463821411,
      "learning_rate": 9.940390143007137e-06,
      "loss": 0.449,
      "num_input_tokens_seen": 3230408,
      "step": 4895
    },
    {
      "epoch": 2.8891509433962264,
      "grad_norm": 2.19368577003479,
      "learning_rate": 9.939993409608339e-06,
      "loss": 0.5554,
      "num_input_tokens_seen": 3233640,
      "step": 4900
    },
    {
      "epoch": 2.8920990566037736,
      "grad_norm": 1.8119715452194214,
      "learning_rate": 9.939595368324996e-06,
      "loss": 0.4095,
      "num_input_tokens_seen": 3236072,
      "step": 4905
    },
    {
      "epoch": 2.8950471698113205,
      "grad_norm": 3.13248610496521,
      "learning_rate": 9.93919601926249e-06,
      "loss": 0.4188,
      "num_input_tokens_seen": 3238472,
      "step": 4910
    },
    {
      "epoch": 2.8979952830188678,
      "grad_norm": 1.353373408317566,
      "learning_rate": 9.938795362526552e-06,
      "loss": 0.3847,
      "num_input_tokens_seen": 3241544,
      "step": 4915
    },
    {
      "epoch": 2.900943396226415,
      "grad_norm": 2.0244503021240234,
      "learning_rate": 9.938393398223255e-06,
      "loss": 0.5639,
      "num_input_tokens_seen": 3244456,
      "step": 4920
    },
    {
      "epoch": 2.9038915094339623,
      "grad_norm": 2.009319543838501,
      "learning_rate": 9.937990126459024e-06,
      "loss": 0.4046,
      "num_input_tokens_seen": 3247784,
      "step": 4925
    },
    {
      "epoch": 2.9068396226415096,
      "grad_norm": 3.247314453125,
      "learning_rate": 9.937585547340624e-06,
      "loss": 0.5231,
      "num_input_tokens_seen": 3251848,
      "step": 4930
    },
    {
      "epoch": 2.9097877358490565,
      "grad_norm": 1.9397766590118408,
      "learning_rate": 9.937179660975174e-06,
      "loss": 0.4796,
      "num_input_tokens_seen": 3254600,
      "step": 4935
    },
    {
      "epoch": 2.9127358490566038,
      "grad_norm": 3.238943099975586,
      "learning_rate": 9.936772467470127e-06,
      "loss": 0.4688,
      "num_input_tokens_seen": 3258280,
      "step": 4940
    },
    {
      "epoch": 2.915683962264151,
      "grad_norm": 3.769240140914917,
      "learning_rate": 9.936363966933294e-06,
      "loss": 0.4604,
      "num_input_tokens_seen": 3261448,
      "step": 4945
    },
    {
      "epoch": 2.918632075471698,
      "grad_norm": 1.6465041637420654,
      "learning_rate": 9.935954159472828e-06,
      "loss": 0.4396,
      "num_input_tokens_seen": 3263944,
      "step": 4950
    },
    {
      "epoch": 2.921580188679245,
      "grad_norm": 2.143521547317505,
      "learning_rate": 9.935543045197222e-06,
      "loss": 0.5498,
      "num_input_tokens_seen": 3266952,
      "step": 4955
    },
    {
      "epoch": 2.9245283018867925,
      "grad_norm": 1.6058701276779175,
      "learning_rate": 9.935130624215326e-06,
      "loss": 0.4095,
      "num_input_tokens_seen": 3269640,
      "step": 4960
    },
    {
      "epoch": 2.9274764150943398,
      "grad_norm": 1.711782693862915,
      "learning_rate": 9.934716896636329e-06,
      "loss": 0.4913,
      "num_input_tokens_seen": 3273032,
      "step": 4965
    },
    {
      "epoch": 2.930424528301887,
      "grad_norm": 1.9224025011062622,
      "learning_rate": 9.934301862569764e-06,
      "loss": 0.3139,
      "num_input_tokens_seen": 3275560,
      "step": 4970
    },
    {
      "epoch": 2.933372641509434,
      "grad_norm": 1.8773002624511719,
      "learning_rate": 9.933885522125517e-06,
      "loss": 0.5789,
      "num_input_tokens_seen": 3279272,
      "step": 4975
    },
    {
      "epoch": 2.936320754716981,
      "grad_norm": 1.231673002243042,
      "learning_rate": 9.933467875413813e-06,
      "loss": 0.4948,
      "num_input_tokens_seen": 3283176,
      "step": 4980
    },
    {
      "epoch": 2.9392688679245285,
      "grad_norm": 5.456884860992432,
      "learning_rate": 9.933048922545227e-06,
      "loss": 0.6969,
      "num_input_tokens_seen": 3285832,
      "step": 4985
    },
    {
      "epoch": 2.9422169811320753,
      "grad_norm": 2.2592976093292236,
      "learning_rate": 9.932628663630679e-06,
      "loss": 0.4719,
      "num_input_tokens_seen": 3289704,
      "step": 4990
    },
    {
      "epoch": 2.9451650943396226,
      "grad_norm": 1.1453009843826294,
      "learning_rate": 9.932207098781432e-06,
      "loss": 0.5436,
      "num_input_tokens_seen": 3293000,
      "step": 4995
    },
    {
      "epoch": 2.94811320754717,
      "grad_norm": 2.3213272094726562,
      "learning_rate": 9.931784228109102e-06,
      "loss": 0.442,
      "num_input_tokens_seen": 3295528,
      "step": 5000
    },
    {
      "epoch": 2.951061320754717,
      "grad_norm": 1.387251853942871,
      "learning_rate": 9.93136005172564e-06,
      "loss": 0.5258,
      "num_input_tokens_seen": 3298696,
      "step": 5005
    },
    {
      "epoch": 2.954009433962264,
      "grad_norm": 2.003241539001465,
      "learning_rate": 9.930934569743354e-06,
      "loss": 0.4759,
      "num_input_tokens_seen": 3301832,
      "step": 5010
    },
    {
      "epoch": 2.9569575471698113,
      "grad_norm": 2.6539113521575928,
      "learning_rate": 9.930507782274888e-06,
      "loss": 0.5014,
      "num_input_tokens_seen": 3304456,
      "step": 5015
    },
    {
      "epoch": 2.9599056603773586,
      "grad_norm": 1.6749892234802246,
      "learning_rate": 9.930079689433236e-06,
      "loss": 0.5144,
      "num_input_tokens_seen": 3307720,
      "step": 5020
    },
    {
      "epoch": 2.9628537735849054,
      "grad_norm": 1.7611149549484253,
      "learning_rate": 9.92965029133174e-06,
      "loss": 0.3842,
      "num_input_tokens_seen": 3313096,
      "step": 5025
    },
    {
      "epoch": 2.9658018867924527,
      "grad_norm": 1.3233667612075806,
      "learning_rate": 9.929219588084084e-06,
      "loss": 0.7789,
      "num_input_tokens_seen": 3316552,
      "step": 5030
    },
    {
      "epoch": 2.96875,
      "grad_norm": 1.7890528440475464,
      "learning_rate": 9.9287875798043e-06,
      "loss": 0.4895,
      "num_input_tokens_seen": 3320200,
      "step": 5035
    },
    {
      "epoch": 2.9716981132075473,
      "grad_norm": 3.683420181274414,
      "learning_rate": 9.92835426660676e-06,
      "loss": 0.4348,
      "num_input_tokens_seen": 3322984,
      "step": 5040
    },
    {
      "epoch": 2.9746462264150946,
      "grad_norm": 1.5043342113494873,
      "learning_rate": 9.927919648606188e-06,
      "loss": 0.5042,
      "num_input_tokens_seen": 3327304,
      "step": 5045
    },
    {
      "epoch": 2.9775943396226414,
      "grad_norm": 1.133007526397705,
      "learning_rate": 9.927483725917652e-06,
      "loss": 0.3553,
      "num_input_tokens_seen": 3330856,
      "step": 5050
    },
    {
      "epoch": 2.9805424528301887,
      "grad_norm": 2.1597917079925537,
      "learning_rate": 9.927046498656562e-06,
      "loss": 0.5841,
      "num_input_tokens_seen": 3334632,
      "step": 5055
    },
    {
      "epoch": 2.983490566037736,
      "grad_norm": 6.557418346405029,
      "learning_rate": 9.926607966938679e-06,
      "loss": 0.631,
      "num_input_tokens_seen": 3339048,
      "step": 5060
    },
    {
      "epoch": 2.986438679245283,
      "grad_norm": 2.1250457763671875,
      "learning_rate": 9.926168130880103e-06,
      "loss": 0.3796,
      "num_input_tokens_seen": 3341960,
      "step": 5065
    },
    {
      "epoch": 2.98938679245283,
      "grad_norm": 1.889841914176941,
      "learning_rate": 9.925726990597283e-06,
      "loss": 0.4301,
      "num_input_tokens_seen": 3345448,
      "step": 5070
    },
    {
      "epoch": 2.9923349056603774,
      "grad_norm": 1.9022197723388672,
      "learning_rate": 9.925284546207015e-06,
      "loss": 0.5687,
      "num_input_tokens_seen": 3348168,
      "step": 5075
    },
    {
      "epoch": 2.9952830188679247,
      "grad_norm": 1.4927198886871338,
      "learning_rate": 9.924840797826436e-06,
      "loss": 0.6291,
      "num_input_tokens_seen": 3352584,
      "step": 5080
    },
    {
      "epoch": 2.9982311320754715,
      "grad_norm": 1.7284501791000366,
      "learning_rate": 9.924395745573029e-06,
      "loss": 0.4345,
      "num_input_tokens_seen": 3356264,
      "step": 5085
    },
    {
      "epoch": 3.001179245283019,
      "grad_norm": 2.574808120727539,
      "learning_rate": 9.923949389564629e-06,
      "loss": 0.6127,
      "num_input_tokens_seen": 3358648,
      "step": 5090
    },
    {
      "epoch": 3.004127358490566,
      "grad_norm": 1.250864028930664,
      "learning_rate": 9.923501729919404e-06,
      "loss": 0.4606,
      "num_input_tokens_seen": 3362520,
      "step": 5095
    },
    {
      "epoch": 3.0070754716981134,
      "grad_norm": 1.8709083795547485,
      "learning_rate": 9.923052766755878e-06,
      "loss": 0.4932,
      "num_input_tokens_seen": 3366776,
      "step": 5100
    },
    {
      "epoch": 3.0100235849056602,
      "grad_norm": 3.279412031173706,
      "learning_rate": 9.922602500192914e-06,
      "loss": 0.6105,
      "num_input_tokens_seen": 3370776,
      "step": 5105
    },
    {
      "epoch": 3.0129716981132075,
      "grad_norm": 3.466262102127075,
      "learning_rate": 9.922150930349725e-06,
      "loss": 0.5598,
      "num_input_tokens_seen": 3374040,
      "step": 5110
    },
    {
      "epoch": 3.015919811320755,
      "grad_norm": 2.249790906906128,
      "learning_rate": 9.921698057345863e-06,
      "loss": 0.5882,
      "num_input_tokens_seen": 3376696,
      "step": 5115
    },
    {
      "epoch": 3.018867924528302,
      "grad_norm": 2.4296271800994873,
      "learning_rate": 9.921243881301229e-06,
      "loss": 0.57,
      "num_input_tokens_seen": 3380312,
      "step": 5120
    },
    {
      "epoch": 3.021816037735849,
      "grad_norm": 1.2398099899291992,
      "learning_rate": 9.920788402336068e-06,
      "loss": 0.3741,
      "num_input_tokens_seen": 3383384,
      "step": 5125
    },
    {
      "epoch": 3.0247641509433962,
      "grad_norm": 1.2645957469940186,
      "learning_rate": 9.92033162057097e-06,
      "loss": 0.5019,
      "num_input_tokens_seen": 3386008,
      "step": 5130
    },
    {
      "epoch": 3.0277122641509435,
      "grad_norm": 2.0354270935058594,
      "learning_rate": 9.919873536126869e-06,
      "loss": 0.4807,
      "num_input_tokens_seen": 3389048,
      "step": 5135
    },
    {
      "epoch": 3.0306603773584904,
      "grad_norm": 1.5833359956741333,
      "learning_rate": 9.919414149125046e-06,
      "loss": 0.3965,
      "num_input_tokens_seen": 3392024,
      "step": 5140
    },
    {
      "epoch": 3.0336084905660377,
      "grad_norm": 1.7305680513381958,
      "learning_rate": 9.918953459687126e-06,
      "loss": 0.4001,
      "num_input_tokens_seen": 3395864,
      "step": 5145
    },
    {
      "epoch": 3.036556603773585,
      "grad_norm": 1.4488825798034668,
      "learning_rate": 9.918491467935078e-06,
      "loss": 0.4424,
      "num_input_tokens_seen": 3399288,
      "step": 5150
    },
    {
      "epoch": 3.0395047169811322,
      "grad_norm": 1.3241454362869263,
      "learning_rate": 9.918028173991218e-06,
      "loss": 0.4176,
      "num_input_tokens_seen": 3401720,
      "step": 5155
    },
    {
      "epoch": 3.042452830188679,
      "grad_norm": 1.383872389793396,
      "learning_rate": 9.917563577978202e-06,
      "loss": 0.5423,
      "num_input_tokens_seen": 3405336,
      "step": 5160
    },
    {
      "epoch": 3.0454009433962264,
      "grad_norm": 1.6932015419006348,
      "learning_rate": 9.917097680019035e-06,
      "loss": 0.4517,
      "num_input_tokens_seen": 3408440,
      "step": 5165
    },
    {
      "epoch": 3.0483490566037736,
      "grad_norm": 2.5172946453094482,
      "learning_rate": 9.916630480237066e-06,
      "loss": 0.507,
      "num_input_tokens_seen": 3411480,
      "step": 5170
    },
    {
      "epoch": 3.051297169811321,
      "grad_norm": 2.2656972408294678,
      "learning_rate": 9.916161978755988e-06,
      "loss": 0.3844,
      "num_input_tokens_seen": 3415128,
      "step": 5175
    },
    {
      "epoch": 3.0542452830188678,
      "grad_norm": 1.0464344024658203,
      "learning_rate": 9.915692175699838e-06,
      "loss": 0.4293,
      "num_input_tokens_seen": 3418936,
      "step": 5180
    },
    {
      "epoch": 3.057193396226415,
      "grad_norm": 2.120067596435547,
      "learning_rate": 9.915221071193e-06,
      "loss": 0.6226,
      "num_input_tokens_seen": 3421912,
      "step": 5185
    },
    {
      "epoch": 3.0601415094339623,
      "grad_norm": 1.2411586046218872,
      "learning_rate": 9.914748665360199e-06,
      "loss": 0.4832,
      "num_input_tokens_seen": 3426008,
      "step": 5190
    },
    {
      "epoch": 3.0630896226415096,
      "grad_norm": 1.2993457317352295,
      "learning_rate": 9.914274958326507e-06,
      "loss": 0.5451,
      "num_input_tokens_seen": 3429656,
      "step": 5195
    },
    {
      "epoch": 3.0660377358490565,
      "grad_norm": 2.153761386871338,
      "learning_rate": 9.913799950217341e-06,
      "loss": 0.4867,
      "num_input_tokens_seen": 3432760,
      "step": 5200
    },
    {
      "epoch": 3.0689858490566038,
      "grad_norm": 2.3473877906799316,
      "learning_rate": 9.91332364115846e-06,
      "loss": 0.4476,
      "num_input_tokens_seen": 3435800,
      "step": 5205
    },
    {
      "epoch": 3.071933962264151,
      "grad_norm": 4.432081699371338,
      "learning_rate": 9.912846031275972e-06,
      "loss": 0.4721,
      "num_input_tokens_seen": 3439064,
      "step": 5210
    },
    {
      "epoch": 3.074882075471698,
      "grad_norm": 1.9132925271987915,
      "learning_rate": 9.912367120696322e-06,
      "loss": 0.5042,
      "num_input_tokens_seen": 3441976,
      "step": 5215
    },
    {
      "epoch": 3.077830188679245,
      "grad_norm": 2.0753188133239746,
      "learning_rate": 9.911886909546307e-06,
      "loss": 0.4671,
      "num_input_tokens_seen": 3445048,
      "step": 5220
    },
    {
      "epoch": 3.0807783018867925,
      "grad_norm": 2.092656135559082,
      "learning_rate": 9.911405397953063e-06,
      "loss": 0.4114,
      "num_input_tokens_seen": 3448664,
      "step": 5225
    },
    {
      "epoch": 3.0837264150943398,
      "grad_norm": 2.623739242553711,
      "learning_rate": 9.910922586044073e-06,
      "loss": 0.3941,
      "num_input_tokens_seen": 3452152,
      "step": 5230
    },
    {
      "epoch": 3.0866745283018866,
      "grad_norm": 1.8942844867706299,
      "learning_rate": 9.910438473947163e-06,
      "loss": 0.4591,
      "num_input_tokens_seen": 3455640,
      "step": 5235
    },
    {
      "epoch": 3.089622641509434,
      "grad_norm": 5.13803768157959,
      "learning_rate": 9.909953061790506e-06,
      "loss": 0.4339,
      "num_input_tokens_seen": 3457976,
      "step": 5240
    },
    {
      "epoch": 3.092570754716981,
      "grad_norm": 1.2253117561340332,
      "learning_rate": 9.909466349702613e-06,
      "loss": 0.5002,
      "num_input_tokens_seen": 3461240,
      "step": 5245
    },
    {
      "epoch": 3.0955188679245285,
      "grad_norm": 1.4781394004821777,
      "learning_rate": 9.908978337812348e-06,
      "loss": 0.615,
      "num_input_tokens_seen": 3464216,
      "step": 5250
    },
    {
      "epoch": 3.0984669811320753,
      "grad_norm": 1.3525395393371582,
      "learning_rate": 9.908489026248909e-06,
      "loss": 0.5009,
      "num_input_tokens_seen": 3466392,
      "step": 5255
    },
    {
      "epoch": 3.1014150943396226,
      "grad_norm": 6.740045547485352,
      "learning_rate": 9.907998415141846e-06,
      "loss": 0.5223,
      "num_input_tokens_seen": 3469464,
      "step": 5260
    },
    {
      "epoch": 3.10436320754717,
      "grad_norm": 1.5601493120193481,
      "learning_rate": 9.907506504621052e-06,
      "loss": 0.5,
      "num_input_tokens_seen": 3473688,
      "step": 5265
    },
    {
      "epoch": 3.107311320754717,
      "grad_norm": 4.9317240715026855,
      "learning_rate": 9.907013294816759e-06,
      "loss": 0.5134,
      "num_input_tokens_seen": 3476376,
      "step": 5270
    },
    {
      "epoch": 3.110259433962264,
      "grad_norm": 2.9300546646118164,
      "learning_rate": 9.906518785859548e-06,
      "loss": 0.3746,
      "num_input_tokens_seen": 3480824,
      "step": 5275
    },
    {
      "epoch": 3.1132075471698113,
      "grad_norm": 2.5729615688323975,
      "learning_rate": 9.906022977880344e-06,
      "loss": 0.5306,
      "num_input_tokens_seen": 3484184,
      "step": 5280
    },
    {
      "epoch": 3.1161556603773586,
      "grad_norm": 1.2537726163864136,
      "learning_rate": 9.905525871010412e-06,
      "loss": 0.4324,
      "num_input_tokens_seen": 3486744,
      "step": 5285
    },
    {
      "epoch": 3.119103773584906,
      "grad_norm": 4.12359094619751,
      "learning_rate": 9.905027465381363e-06,
      "loss": 0.4332,
      "num_input_tokens_seen": 3489816,
      "step": 5290
    },
    {
      "epoch": 3.1220518867924527,
      "grad_norm": 3.366502285003662,
      "learning_rate": 9.904527761125155e-06,
      "loss": 0.516,
      "num_input_tokens_seen": 3492568,
      "step": 5295
    },
    {
      "epoch": 3.125,
      "grad_norm": 1.3723126649856567,
      "learning_rate": 9.904026758374083e-06,
      "loss": 0.4546,
      "num_input_tokens_seen": 3495544,
      "step": 5300
    },
    {
      "epoch": 3.1279481132075473,
      "grad_norm": 1.7108049392700195,
      "learning_rate": 9.903524457260794e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 3499032,
      "step": 5305
    },
    {
      "epoch": 3.1308962264150946,
      "grad_norm": 2.319870710372925,
      "learning_rate": 9.90302085791827e-06,
      "loss": 0.4447,
      "num_input_tokens_seen": 3502200,
      "step": 5310
    },
    {
      "epoch": 3.1338443396226414,
      "grad_norm": 2.7945799827575684,
      "learning_rate": 9.902515960479844e-06,
      "loss": 0.5396,
      "num_input_tokens_seen": 3505720,
      "step": 5315
    },
    {
      "epoch": 3.1367924528301887,
      "grad_norm": 0.995719313621521,
      "learning_rate": 9.902009765079188e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 3509240,
      "step": 5320
    },
    {
      "epoch": 3.139740566037736,
      "grad_norm": 2.637019157409668,
      "learning_rate": 9.90150227185032e-06,
      "loss": 0.4725,
      "num_input_tokens_seen": 3511640,
      "step": 5325
    },
    {
      "epoch": 3.142688679245283,
      "grad_norm": 1.8110188245773315,
      "learning_rate": 9.900993480927603e-06,
      "loss": 0.4772,
      "num_input_tokens_seen": 3514904,
      "step": 5330
    },
    {
      "epoch": 3.14563679245283,
      "grad_norm": 1.7690271139144897,
      "learning_rate": 9.90048339244574e-06,
      "loss": 0.2791,
      "num_input_tokens_seen": 3520536,
      "step": 5335
    },
    {
      "epoch": 3.1485849056603774,
      "grad_norm": 1.153158187866211,
      "learning_rate": 9.899972006539776e-06,
      "loss": 0.5024,
      "num_input_tokens_seen": 3524120,
      "step": 5340
    },
    {
      "epoch": 3.1515330188679247,
      "grad_norm": 2.4987497329711914,
      "learning_rate": 9.899459323345106e-06,
      "loss": 0.3304,
      "num_input_tokens_seen": 3527480,
      "step": 5345
    },
    {
      "epoch": 3.1544811320754715,
      "grad_norm": 1.5526299476623535,
      "learning_rate": 9.898945342997467e-06,
      "loss": 0.5043,
      "num_input_tokens_seen": 3532216,
      "step": 5350
    },
    {
      "epoch": 3.157429245283019,
      "grad_norm": 2.2458126544952393,
      "learning_rate": 9.898430065632933e-06,
      "loss": 0.4677,
      "num_input_tokens_seen": 3534680,
      "step": 5355
    },
    {
      "epoch": 3.160377358490566,
      "grad_norm": 3.355865955352783,
      "learning_rate": 9.897913491387929e-06,
      "loss": 0.4903,
      "num_input_tokens_seen": 3537592,
      "step": 5360
    },
    {
      "epoch": 3.1633254716981134,
      "grad_norm": 1.2506219148635864,
      "learning_rate": 9.897395620399219e-06,
      "loss": 0.3865,
      "num_input_tokens_seen": 3540248,
      "step": 5365
    },
    {
      "epoch": 3.1662735849056602,
      "grad_norm": 1.8115522861480713,
      "learning_rate": 9.896876452803913e-06,
      "loss": 0.4733,
      "num_input_tokens_seen": 3543512,
      "step": 5370
    },
    {
      "epoch": 3.1692216981132075,
      "grad_norm": 2.092369318008423,
      "learning_rate": 9.896355988739461e-06,
      "loss": 0.3976,
      "num_input_tokens_seen": 3546488,
      "step": 5375
    },
    {
      "epoch": 3.172169811320755,
      "grad_norm": 1.4437311887741089,
      "learning_rate": 9.895834228343658e-06,
      "loss": 0.4173,
      "num_input_tokens_seen": 3548888,
      "step": 5380
    },
    {
      "epoch": 3.175117924528302,
      "grad_norm": 4.9079203605651855,
      "learning_rate": 9.895311171754644e-06,
      "loss": 0.552,
      "num_input_tokens_seen": 3551800,
      "step": 5385
    },
    {
      "epoch": 3.178066037735849,
      "grad_norm": 1.0746113061904907,
      "learning_rate": 9.8947868191109e-06,
      "loss": 0.5465,
      "num_input_tokens_seen": 3554904,
      "step": 5390
    },
    {
      "epoch": 3.1810141509433962,
      "grad_norm": 4.185611724853516,
      "learning_rate": 9.894261170551249e-06,
      "loss": 0.457,
      "num_input_tokens_seen": 3559160,
      "step": 5395
    },
    {
      "epoch": 3.1839622641509435,
      "grad_norm": 2.3797383308410645,
      "learning_rate": 9.893734226214861e-06,
      "loss": 0.4234,
      "num_input_tokens_seen": 3562712,
      "step": 5400
    },
    {
      "epoch": 3.1869103773584904,
      "grad_norm": 1.9186604022979736,
      "learning_rate": 9.893205986241246e-06,
      "loss": 0.5334,
      "num_input_tokens_seen": 3566040,
      "step": 5405
    },
    {
      "epoch": 3.1898584905660377,
      "grad_norm": 1.6732419729232788,
      "learning_rate": 9.892676450770257e-06,
      "loss": 0.464,
      "num_input_tokens_seen": 3569976,
      "step": 5410
    },
    {
      "epoch": 3.192806603773585,
      "grad_norm": 1.509825348854065,
      "learning_rate": 9.892145619942092e-06,
      "loss": 0.5705,
      "num_input_tokens_seen": 3572984,
      "step": 5415
    },
    {
      "epoch": 3.1957547169811322,
      "grad_norm": 2.3248484134674072,
      "learning_rate": 9.891613493897289e-06,
      "loss": 0.4542,
      "num_input_tokens_seen": 3576408,
      "step": 5420
    },
    {
      "epoch": 3.198702830188679,
      "grad_norm": 2.5423998832702637,
      "learning_rate": 9.891080072776733e-06,
      "loss": 0.4925,
      "num_input_tokens_seen": 3579608,
      "step": 5425
    },
    {
      "epoch": 3.2016509433962264,
      "grad_norm": 2.1048996448516846,
      "learning_rate": 9.890545356721649e-06,
      "loss": 0.4849,
      "num_input_tokens_seen": 3582552,
      "step": 5430
    },
    {
      "epoch": 3.2045990566037736,
      "grad_norm": 1.957683801651001,
      "learning_rate": 9.890009345873603e-06,
      "loss": 0.4255,
      "num_input_tokens_seen": 3585528,
      "step": 5435
    },
    {
      "epoch": 3.207547169811321,
      "grad_norm": 1.3933781385421753,
      "learning_rate": 9.889472040374509e-06,
      "loss": 0.4862,
      "num_input_tokens_seen": 3589624,
      "step": 5440
    },
    {
      "epoch": 3.2104952830188678,
      "grad_norm": 3.683094024658203,
      "learning_rate": 9.88893344036662e-06,
      "loss": 0.4965,
      "num_input_tokens_seen": 3593496,
      "step": 5445
    },
    {
      "epoch": 3.213443396226415,
      "grad_norm": 1.9547406435012817,
      "learning_rate": 9.888393545992531e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 3596760,
      "step": 5450
    },
    {
      "epoch": 3.2163915094339623,
      "grad_norm": 2.2385106086730957,
      "learning_rate": 9.887852357395184e-06,
      "loss": 0.7132,
      "num_input_tokens_seen": 3600664,
      "step": 5455
    },
    {
      "epoch": 3.2193396226415096,
      "grad_norm": 0.8917670845985413,
      "learning_rate": 9.88730987471786e-06,
      "loss": 0.4846,
      "num_input_tokens_seen": 3604312,
      "step": 5460
    },
    {
      "epoch": 3.2222877358490565,
      "grad_norm": 1.4458024501800537,
      "learning_rate": 9.886766098104183e-06,
      "loss": 0.3827,
      "num_input_tokens_seen": 3608088,
      "step": 5465
    },
    {
      "epoch": 3.2252358490566038,
      "grad_norm": 2.2719368934631348,
      "learning_rate": 9.886221027698122e-06,
      "loss": 0.6941,
      "num_input_tokens_seen": 3610808,
      "step": 5470
    },
    {
      "epoch": 3.228183962264151,
      "grad_norm": 1.3857052326202393,
      "learning_rate": 9.885674663643983e-06,
      "loss": 0.4603,
      "num_input_tokens_seen": 3614680,
      "step": 5475
    },
    {
      "epoch": 3.231132075471698,
      "grad_norm": 3.32948637008667,
      "learning_rate": 9.885127006086423e-06,
      "loss": 0.6486,
      "num_input_tokens_seen": 3617912,
      "step": 5480
    },
    {
      "epoch": 3.234080188679245,
      "grad_norm": 2.0632119178771973,
      "learning_rate": 9.884578055170434e-06,
      "loss": 0.4348,
      "num_input_tokens_seen": 3623800,
      "step": 5485
    },
    {
      "epoch": 3.2370283018867925,
      "grad_norm": 3.6889400482177734,
      "learning_rate": 9.884027811041353e-06,
      "loss": 0.4832,
      "num_input_tokens_seen": 3626744,
      "step": 5490
    },
    {
      "epoch": 3.2399764150943398,
      "grad_norm": 1.8935861587524414,
      "learning_rate": 9.883476273844861e-06,
      "loss": 0.3492,
      "num_input_tokens_seen": 3629880,
      "step": 5495
    },
    {
      "epoch": 3.2429245283018866,
      "grad_norm": 1.874763011932373,
      "learning_rate": 9.882923443726977e-06,
      "loss": 0.4926,
      "num_input_tokens_seen": 3632792,
      "step": 5500
    },
    {
      "epoch": 3.245872641509434,
      "grad_norm": 3.037036895751953,
      "learning_rate": 9.882369320834068e-06,
      "loss": 0.5218,
      "num_input_tokens_seen": 3638232,
      "step": 5505
    },
    {
      "epoch": 3.248820754716981,
      "grad_norm": 2.1290903091430664,
      "learning_rate": 9.88181390531284e-06,
      "loss": 0.6092,
      "num_input_tokens_seen": 3640920,
      "step": 5510
    },
    {
      "epoch": 3.2517688679245285,
      "grad_norm": 2.4716434478759766,
      "learning_rate": 9.88125719731034e-06,
      "loss": 0.3874,
      "num_input_tokens_seen": 3644344,
      "step": 5515
    },
    {
      "epoch": 3.2547169811320753,
      "grad_norm": 2.45539927482605,
      "learning_rate": 9.880699196973962e-06,
      "loss": 0.4581,
      "num_input_tokens_seen": 3647160,
      "step": 5520
    },
    {
      "epoch": 3.2576650943396226,
      "grad_norm": 2.4092705249786377,
      "learning_rate": 9.880139904451436e-06,
      "loss": 0.6643,
      "num_input_tokens_seen": 3650040,
      "step": 5525
    },
    {
      "epoch": 3.26061320754717,
      "grad_norm": 1.7549151182174683,
      "learning_rate": 9.879579319890838e-06,
      "loss": 0.4169,
      "num_input_tokens_seen": 3652824,
      "step": 5530
    },
    {
      "epoch": 3.263561320754717,
      "grad_norm": 1.3772081136703491,
      "learning_rate": 9.879017443440584e-06,
      "loss": 0.5452,
      "num_input_tokens_seen": 3657080,
      "step": 5535
    },
    {
      "epoch": 3.266509433962264,
      "grad_norm": 1.7863273620605469,
      "learning_rate": 9.878454275249436e-06,
      "loss": 0.348,
      "num_input_tokens_seen": 3660248,
      "step": 5540
    },
    {
      "epoch": 3.2694575471698113,
      "grad_norm": 3.0786070823669434,
      "learning_rate": 9.877889815466493e-06,
      "loss": 0.6093,
      "num_input_tokens_seen": 3663352,
      "step": 5545
    },
    {
      "epoch": 3.2724056603773586,
      "grad_norm": 1.9007587432861328,
      "learning_rate": 9.877324064241198e-06,
      "loss": 0.4911,
      "num_input_tokens_seen": 3666616,
      "step": 5550
    },
    {
      "epoch": 3.2753537735849054,
      "grad_norm": 1.8963282108306885,
      "learning_rate": 9.876757021723338e-06,
      "loss": 0.5634,
      "num_input_tokens_seen": 3669560,
      "step": 5555
    },
    {
      "epoch": 3.2783018867924527,
      "grad_norm": 1.261326789855957,
      "learning_rate": 9.876188688063038e-06,
      "loss": 0.4424,
      "num_input_tokens_seen": 3672536,
      "step": 5560
    },
    {
      "epoch": 3.28125,
      "grad_norm": 3.8353753089904785,
      "learning_rate": 9.875619063410768e-06,
      "loss": 0.5248,
      "num_input_tokens_seen": 3675960,
      "step": 5565
    },
    {
      "epoch": 3.2841981132075473,
      "grad_norm": 2.169651746749878,
      "learning_rate": 9.875048147917339e-06,
      "loss": 0.3852,
      "num_input_tokens_seen": 3679384,
      "step": 5570
    },
    {
      "epoch": 3.2871462264150946,
      "grad_norm": 2.7301642894744873,
      "learning_rate": 9.874475941733902e-06,
      "loss": 0.4924,
      "num_input_tokens_seen": 3681976,
      "step": 5575
    },
    {
      "epoch": 3.2900943396226414,
      "grad_norm": 1.126723289489746,
      "learning_rate": 9.873902445011952e-06,
      "loss": 0.4218,
      "num_input_tokens_seen": 3686488,
      "step": 5580
    },
    {
      "epoch": 3.2930424528301887,
      "grad_norm": 2.651505470275879,
      "learning_rate": 9.873327657903324e-06,
      "loss": 0.4233,
      "num_input_tokens_seen": 3689848,
      "step": 5585
    },
    {
      "epoch": 3.295990566037736,
      "grad_norm": 1.7688757181167603,
      "learning_rate": 9.872751580560194e-06,
      "loss": 0.4953,
      "num_input_tokens_seen": 3693368,
      "step": 5590
    },
    {
      "epoch": 3.298938679245283,
      "grad_norm": 1.7077308893203735,
      "learning_rate": 9.872174213135084e-06,
      "loss": 0.4239,
      "num_input_tokens_seen": 3696664,
      "step": 5595
    },
    {
      "epoch": 3.30188679245283,
      "grad_norm": 2.16625714302063,
      "learning_rate": 9.871595555780855e-06,
      "loss": 0.4299,
      "num_input_tokens_seen": 3699480,
      "step": 5600
    },
    {
      "epoch": 3.3048349056603774,
      "grad_norm": 1.9272631406784058,
      "learning_rate": 9.871015608650705e-06,
      "loss": 0.375,
      "num_input_tokens_seen": 3702328,
      "step": 5605
    },
    {
      "epoch": 3.3077830188679247,
      "grad_norm": 1.9644043445587158,
      "learning_rate": 9.870434371898182e-06,
      "loss": 0.4242,
      "num_input_tokens_seen": 3706712,
      "step": 5610
    },
    {
      "epoch": 3.3107311320754715,
      "grad_norm": 2.914085626602173,
      "learning_rate": 9.869851845677165e-06,
      "loss": 0.4119,
      "num_input_tokens_seen": 3709464,
      "step": 5615
    },
    {
      "epoch": 3.313679245283019,
      "grad_norm": 2.0667543411254883,
      "learning_rate": 9.869268030141886e-06,
      "loss": 0.5093,
      "num_input_tokens_seen": 3712088,
      "step": 5620
    },
    {
      "epoch": 3.316627358490566,
      "grad_norm": 2.5321643352508545,
      "learning_rate": 9.86868292544691e-06,
      "loss": 0.5983,
      "num_input_tokens_seen": 3714744,
      "step": 5625
    },
    {
      "epoch": 3.3195754716981134,
      "grad_norm": 1.5767210721969604,
      "learning_rate": 9.868096531747149e-06,
      "loss": 0.3779,
      "num_input_tokens_seen": 3717432,
      "step": 5630
    },
    {
      "epoch": 3.3225235849056602,
      "grad_norm": 1.7506160736083984,
      "learning_rate": 9.867508849197848e-06,
      "loss": 0.6017,
      "num_input_tokens_seen": 3721336,
      "step": 5635
    },
    {
      "epoch": 3.3254716981132075,
      "grad_norm": 2.446323871612549,
      "learning_rate": 9.866919877954602e-06,
      "loss": 0.3692,
      "num_input_tokens_seen": 3723960,
      "step": 5640
    },
    {
      "epoch": 3.328419811320755,
      "grad_norm": 2.0707485675811768,
      "learning_rate": 9.866329618173344e-06,
      "loss": 0.5357,
      "num_input_tokens_seen": 3727736,
      "step": 5645
    },
    {
      "epoch": 3.331367924528302,
      "grad_norm": 1.815556526184082,
      "learning_rate": 9.865738070010346e-06,
      "loss": 0.4942,
      "num_input_tokens_seen": 3731256,
      "step": 5650
    },
    {
      "epoch": 3.334316037735849,
      "grad_norm": 3.716630697250366,
      "learning_rate": 9.865145233622223e-06,
      "loss": 0.4271,
      "num_input_tokens_seen": 3734616,
      "step": 5655
    },
    {
      "epoch": 3.3372641509433962,
      "grad_norm": 2.213042736053467,
      "learning_rate": 9.864551109165935e-06,
      "loss": 0.5426,
      "num_input_tokens_seen": 3738072,
      "step": 5660
    },
    {
      "epoch": 3.3402122641509435,
      "grad_norm": 2.382355213165283,
      "learning_rate": 9.863955696798773e-06,
      "loss": 0.416,
      "num_input_tokens_seen": 3740536,
      "step": 5665
    },
    {
      "epoch": 3.3431603773584904,
      "grad_norm": 3.0343971252441406,
      "learning_rate": 9.863358996678378e-06,
      "loss": 0.487,
      "num_input_tokens_seen": 3742712,
      "step": 5670
    },
    {
      "epoch": 3.3461084905660377,
      "grad_norm": 2.006971597671509,
      "learning_rate": 9.86276100896273e-06,
      "loss": 0.5276,
      "num_input_tokens_seen": 3745560,
      "step": 5675
    },
    {
      "epoch": 3.349056603773585,
      "grad_norm": 2.0556857585906982,
      "learning_rate": 9.862161733810147e-06,
      "loss": 0.415,
      "num_input_tokens_seen": 3749880,
      "step": 5680
    },
    {
      "epoch": 3.3520047169811322,
      "grad_norm": 2.368375062942505,
      "learning_rate": 9.86156117137929e-06,
      "loss": 0.4606,
      "num_input_tokens_seen": 3753400,
      "step": 5685
    },
    {
      "epoch": 3.354952830188679,
      "grad_norm": 2.968602180480957,
      "learning_rate": 9.860959321829159e-06,
      "loss": 0.3834,
      "num_input_tokens_seen": 3757272,
      "step": 5690
    },
    {
      "epoch": 3.3579009433962264,
      "grad_norm": 1.8093825578689575,
      "learning_rate": 9.860356185319102e-06,
      "loss": 0.3978,
      "num_input_tokens_seen": 3761848,
      "step": 5695
    },
    {
      "epoch": 3.3608490566037736,
      "grad_norm": 1.3766168355941772,
      "learning_rate": 9.859751762008796e-06,
      "loss": 0.3167,
      "num_input_tokens_seen": 3765208,
      "step": 5700
    },
    {
      "epoch": 3.363797169811321,
      "grad_norm": 1.4748882055282593,
      "learning_rate": 9.859146052058266e-06,
      "loss": 0.2923,
      "num_input_tokens_seen": 3768056,
      "step": 5705
    },
    {
      "epoch": 3.3667452830188678,
      "grad_norm": 2.081285238265991,
      "learning_rate": 9.858539055627876e-06,
      "loss": 0.5204,
      "num_input_tokens_seen": 3771736,
      "step": 5710
    },
    {
      "epoch": 3.369693396226415,
      "grad_norm": 1.9172828197479248,
      "learning_rate": 9.857930772878333e-06,
      "loss": 0.455,
      "num_input_tokens_seen": 3774680,
      "step": 5715
    },
    {
      "epoch": 3.3726415094339623,
      "grad_norm": 3.287672519683838,
      "learning_rate": 9.857321203970682e-06,
      "loss": 0.4225,
      "num_input_tokens_seen": 3778168,
      "step": 5720
    },
    {
      "epoch": 3.3755896226415096,
      "grad_norm": 2.085043430328369,
      "learning_rate": 9.856710349066307e-06,
      "loss": 0.392,
      "num_input_tokens_seen": 3780856,
      "step": 5725
    },
    {
      "epoch": 3.3785377358490565,
      "grad_norm": 1.3676239252090454,
      "learning_rate": 9.856098208326937e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 3783960,
      "step": 5730
    },
    {
      "epoch": 3.3814858490566038,
      "grad_norm": 2.3357512950897217,
      "learning_rate": 9.855484781914639e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 3786936,
      "step": 5735
    },
    {
      "epoch": 3.384433962264151,
      "grad_norm": 1.9417906999588013,
      "learning_rate": 9.854870069991817e-06,
      "loss": 0.509,
      "num_input_tokens_seen": 3790712,
      "step": 5740
    },
    {
      "epoch": 3.387382075471698,
      "grad_norm": 2.8476386070251465,
      "learning_rate": 9.854254072721222e-06,
      "loss": 0.475,
      "num_input_tokens_seen": 3795864,
      "step": 5745
    },
    {
      "epoch": 3.390330188679245,
      "grad_norm": 2.6767914295196533,
      "learning_rate": 9.853636790265938e-06,
      "loss": 0.4216,
      "num_input_tokens_seen": 3799448,
      "step": 5750
    },
    {
      "epoch": 3.3932783018867925,
      "grad_norm": 2.035623788833618,
      "learning_rate": 9.853018222789397e-06,
      "loss": 0.4959,
      "num_input_tokens_seen": 3802552,
      "step": 5755
    },
    {
      "epoch": 3.3962264150943398,
      "grad_norm": 2.688955545425415,
      "learning_rate": 9.852398370455367e-06,
      "loss": 0.4481,
      "num_input_tokens_seen": 3807544,
      "step": 5760
    },
    {
      "epoch": 3.3991745283018866,
      "grad_norm": 1.4752839803695679,
      "learning_rate": 9.851777233427955e-06,
      "loss": 0.448,
      "num_input_tokens_seen": 3810840,
      "step": 5765
    },
    {
      "epoch": 3.402122641509434,
      "grad_norm": 2.4519567489624023,
      "learning_rate": 9.85115481187161e-06,
      "loss": 0.4949,
      "num_input_tokens_seen": 3813656,
      "step": 5770
    },
    {
      "epoch": 3.405070754716981,
      "grad_norm": 3.470754623413086,
      "learning_rate": 9.850531105951123e-06,
      "loss": 0.5965,
      "num_input_tokens_seen": 3816632,
      "step": 5775
    },
    {
      "epoch": 3.4080188679245285,
      "grad_norm": 1.803781509399414,
      "learning_rate": 9.84990611583162e-06,
      "loss": 0.4988,
      "num_input_tokens_seen": 3820024,
      "step": 5780
    },
    {
      "epoch": 3.4109669811320753,
      "grad_norm": 1.7882633209228516,
      "learning_rate": 9.849279841678572e-06,
      "loss": 0.4874,
      "num_input_tokens_seen": 3823160,
      "step": 5785
    },
    {
      "epoch": 3.4139150943396226,
      "grad_norm": 1.106061577796936,
      "learning_rate": 9.848652283657785e-06,
      "loss": 0.5397,
      "num_input_tokens_seen": 3826904,
      "step": 5790
    },
    {
      "epoch": 3.41686320754717,
      "grad_norm": 2.476064920425415,
      "learning_rate": 9.848023441935411e-06,
      "loss": 0.5086,
      "num_input_tokens_seen": 3830552,
      "step": 5795
    },
    {
      "epoch": 3.419811320754717,
      "grad_norm": 2.5693178176879883,
      "learning_rate": 9.847393316677935e-06,
      "loss": 0.5073,
      "num_input_tokens_seen": 3833944,
      "step": 5800
    },
    {
      "epoch": 3.422759433962264,
      "grad_norm": 3.5297389030456543,
      "learning_rate": 9.846761908052188e-06,
      "loss": 0.6502,
      "num_input_tokens_seen": 3837176,
      "step": 5805
    },
    {
      "epoch": 3.4257075471698113,
      "grad_norm": 0.9588062763214111,
      "learning_rate": 9.846129216225338e-06,
      "loss": 0.3878,
      "num_input_tokens_seen": 3841528,
      "step": 5810
    },
    {
      "epoch": 3.4286556603773586,
      "grad_norm": 1.0326237678527832,
      "learning_rate": 9.845495241364892e-06,
      "loss": 0.4118,
      "num_input_tokens_seen": 3845304,
      "step": 5815
    },
    {
      "epoch": 3.4316037735849054,
      "grad_norm": 4.396010398864746,
      "learning_rate": 9.844859983638696e-06,
      "loss": 0.6545,
      "num_input_tokens_seen": 3847992,
      "step": 5820
    },
    {
      "epoch": 3.4345518867924527,
      "grad_norm": 2.924666404724121,
      "learning_rate": 9.844223443214942e-06,
      "loss": 0.5724,
      "num_input_tokens_seen": 3851896,
      "step": 5825
    },
    {
      "epoch": 3.4375,
      "grad_norm": 2.143361806869507,
      "learning_rate": 9.843585620262153e-06,
      "loss": 0.531,
      "num_input_tokens_seen": 3855448,
      "step": 5830
    },
    {
      "epoch": 3.4404481132075473,
      "grad_norm": 2.109581470489502,
      "learning_rate": 9.842946514949197e-06,
      "loss": 0.4917,
      "num_input_tokens_seen": 3858040,
      "step": 5835
    },
    {
      "epoch": 3.4433962264150946,
      "grad_norm": 2.917799234390259,
      "learning_rate": 9.842306127445279e-06,
      "loss": 0.3642,
      "num_input_tokens_seen": 3860536,
      "step": 5840
    },
    {
      "epoch": 3.4463443396226414,
      "grad_norm": 1.9203991889953613,
      "learning_rate": 9.841664457919944e-06,
      "loss": 0.4526,
      "num_input_tokens_seen": 3863416,
      "step": 5845
    },
    {
      "epoch": 3.4492924528301887,
      "grad_norm": 1.8055039644241333,
      "learning_rate": 9.841021506543079e-06,
      "loss": 0.468,
      "num_input_tokens_seen": 3866296,
      "step": 5850
    },
    {
      "epoch": 3.452240566037736,
      "grad_norm": 1.7548140287399292,
      "learning_rate": 9.840377273484904e-06,
      "loss": 0.5662,
      "num_input_tokens_seen": 3869400,
      "step": 5855
    },
    {
      "epoch": 3.455188679245283,
      "grad_norm": 2.8091752529144287,
      "learning_rate": 9.839731758915986e-06,
      "loss": 0.5525,
      "num_input_tokens_seen": 3872632,
      "step": 5860
    },
    {
      "epoch": 3.45813679245283,
      "grad_norm": 2.408369779586792,
      "learning_rate": 9.839084963007226e-06,
      "loss": 0.3634,
      "num_input_tokens_seen": 3875800,
      "step": 5865
    },
    {
      "epoch": 3.4610849056603774,
      "grad_norm": 2.7587149143218994,
      "learning_rate": 9.838436885929868e-06,
      "loss": 0.4765,
      "num_input_tokens_seen": 3878520,
      "step": 5870
    },
    {
      "epoch": 3.4640330188679247,
      "grad_norm": 5.209230899810791,
      "learning_rate": 9.837787527855492e-06,
      "loss": 0.5723,
      "num_input_tokens_seen": 3882424,
      "step": 5875
    },
    {
      "epoch": 3.4669811320754715,
      "grad_norm": 1.8847943544387817,
      "learning_rate": 9.837136888956017e-06,
      "loss": 0.4035,
      "num_input_tokens_seen": 3885464,
      "step": 5880
    },
    {
      "epoch": 3.469929245283019,
      "grad_norm": 2.144526243209839,
      "learning_rate": 9.836484969403705e-06,
      "loss": 0.5837,
      "num_input_tokens_seen": 3888408,
      "step": 5885
    },
    {
      "epoch": 3.472877358490566,
      "grad_norm": 1.9771209955215454,
      "learning_rate": 9.835831769371152e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 3891960,
      "step": 5890
    },
    {
      "epoch": 3.4758254716981134,
      "grad_norm": 1.3798195123672485,
      "learning_rate": 9.835177289031298e-06,
      "loss": 0.5154,
      "num_input_tokens_seen": 3896344,
      "step": 5895
    },
    {
      "epoch": 3.4787735849056602,
      "grad_norm": 5.641699314117432,
      "learning_rate": 9.834521528557419e-06,
      "loss": 0.5181,
      "num_input_tokens_seen": 3899096,
      "step": 5900
    },
    {
      "epoch": 3.4817216981132075,
      "grad_norm": 2.4781124591827393,
      "learning_rate": 9.833864488123128e-06,
      "loss": 0.4399,
      "num_input_tokens_seen": 3901976,
      "step": 5905
    },
    {
      "epoch": 3.484669811320755,
      "grad_norm": 1.5114061832427979,
      "learning_rate": 9.83320616790238e-06,
      "loss": 0.4321,
      "num_input_tokens_seen": 3905848,
      "step": 5910
    },
    {
      "epoch": 3.487617924528302,
      "grad_norm": 2.3745806217193604,
      "learning_rate": 9.832546568069472e-06,
      "loss": 0.4312,
      "num_input_tokens_seen": 3909720,
      "step": 5915
    },
    {
      "epoch": 3.490566037735849,
      "grad_norm": 5.111372470855713,
      "learning_rate": 9.831885688799031e-06,
      "loss": 0.5204,
      "num_input_tokens_seen": 3912664,
      "step": 5920
    },
    {
      "epoch": 3.4935141509433962,
      "grad_norm": 1.9060786962509155,
      "learning_rate": 9.83122353026603e-06,
      "loss": 0.4592,
      "num_input_tokens_seen": 3919608,
      "step": 5925
    },
    {
      "epoch": 3.4964622641509435,
      "grad_norm": 3.092183828353882,
      "learning_rate": 9.830560092645778e-06,
      "loss": 0.3905,
      "num_input_tokens_seen": 3922136,
      "step": 5930
    },
    {
      "epoch": 3.4994103773584904,
      "grad_norm": 1.3127161264419556,
      "learning_rate": 9.829895376113923e-06,
      "loss": 0.4778,
      "num_input_tokens_seen": 3924536,
      "step": 5935
    },
    {
      "epoch": 3.5023584905660377,
      "grad_norm": 1.9892606735229492,
      "learning_rate": 9.829229380846452e-06,
      "loss": 0.5404,
      "num_input_tokens_seen": 3927224,
      "step": 5940
    },
    {
      "epoch": 3.505306603773585,
      "grad_norm": 1.5708200931549072,
      "learning_rate": 9.82856210701969e-06,
      "loss": 0.577,
      "num_input_tokens_seen": 3930776,
      "step": 5945
    },
    {
      "epoch": 3.5082547169811322,
      "grad_norm": 1.9528100490570068,
      "learning_rate": 9.827893554810298e-06,
      "loss": 0.5908,
      "num_input_tokens_seen": 3932952,
      "step": 5950
    },
    {
      "epoch": 3.5112028301886795,
      "grad_norm": 1.5022257566452026,
      "learning_rate": 9.827223724395281e-06,
      "loss": 0.4346,
      "num_input_tokens_seen": 3936504,
      "step": 5955
    },
    {
      "epoch": 3.5141509433962264,
      "grad_norm": 1.5970160961151123,
      "learning_rate": 9.82655261595198e-06,
      "loss": 0.5501,
      "num_input_tokens_seen": 3939384,
      "step": 5960
    },
    {
      "epoch": 3.5170990566037736,
      "grad_norm": 2.4326248168945312,
      "learning_rate": 9.825880229658073e-06,
      "loss": 0.5385,
      "num_input_tokens_seen": 3942136,
      "step": 5965
    },
    {
      "epoch": 3.5200471698113205,
      "grad_norm": 1.2024980783462524,
      "learning_rate": 9.825206565691576e-06,
      "loss": 0.515,
      "num_input_tokens_seen": 3945464,
      "step": 5970
    },
    {
      "epoch": 3.5229952830188678,
      "grad_norm": 1.285416841506958,
      "learning_rate": 9.824531624230844e-06,
      "loss": 0.5044,
      "num_input_tokens_seen": 3948248,
      "step": 5975
    },
    {
      "epoch": 3.525943396226415,
      "grad_norm": 1.4315446615219116,
      "learning_rate": 9.823855405454573e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 3951096,
      "step": 5980
    },
    {
      "epoch": 3.5288915094339623,
      "grad_norm": 3.358830451965332,
      "learning_rate": 9.823177909541795e-06,
      "loss": 0.6346,
      "num_input_tokens_seen": 3954456,
      "step": 5985
    },
    {
      "epoch": 3.5318396226415096,
      "grad_norm": 3.425950527191162,
      "learning_rate": 9.822499136671877e-06,
      "loss": 0.5115,
      "num_input_tokens_seen": 3958424,
      "step": 5990
    },
    {
      "epoch": 3.5347877358490565,
      "grad_norm": 1.5141605138778687,
      "learning_rate": 9.82181908702453e-06,
      "loss": 0.4226,
      "num_input_tokens_seen": 3961720,
      "step": 5995
    },
    {
      "epoch": 3.5377358490566038,
      "grad_norm": 1.7143033742904663,
      "learning_rate": 9.821137760779797e-06,
      "loss": 0.4087,
      "num_input_tokens_seen": 3965304,
      "step": 6000
    },
    {
      "epoch": 3.540683962264151,
      "grad_norm": 2.606755018234253,
      "learning_rate": 9.820455158118065e-06,
      "loss": 0.4586,
      "num_input_tokens_seen": 3968632,
      "step": 6005
    },
    {
      "epoch": 3.543632075471698,
      "grad_norm": 2.1948137283325195,
      "learning_rate": 9.819771279220053e-06,
      "loss": 0.4888,
      "num_input_tokens_seen": 3972600,
      "step": 6010
    },
    {
      "epoch": 3.546580188679245,
      "grad_norm": 1.398087739944458,
      "learning_rate": 9.819086124266825e-06,
      "loss": 0.4844,
      "num_input_tokens_seen": 3975704,
      "step": 6015
    },
    {
      "epoch": 3.5495283018867925,
      "grad_norm": 1.2942590713500977,
      "learning_rate": 9.818399693439778e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 3978232,
      "step": 6020
    },
    {
      "epoch": 3.5524764150943398,
      "grad_norm": 1.5676225423812866,
      "learning_rate": 9.817711986920644e-06,
      "loss": 0.495,
      "num_input_tokens_seen": 3981848,
      "step": 6025
    },
    {
      "epoch": 3.555424528301887,
      "grad_norm": 2.328906536102295,
      "learning_rate": 9.817023004891497e-06,
      "loss": 0.6072,
      "num_input_tokens_seen": 3985784,
      "step": 6030
    },
    {
      "epoch": 3.558372641509434,
      "grad_norm": 1.676172137260437,
      "learning_rate": 9.816332747534752e-06,
      "loss": 0.4338,
      "num_input_tokens_seen": 3989016,
      "step": 6035
    },
    {
      "epoch": 3.561320754716981,
      "grad_norm": 4.3357462882995605,
      "learning_rate": 9.815641215033153e-06,
      "loss": 0.6269,
      "num_input_tokens_seen": 3992920,
      "step": 6040
    },
    {
      "epoch": 3.5642688679245285,
      "grad_norm": 3.7040438652038574,
      "learning_rate": 9.814948407569789e-06,
      "loss": 0.4931,
      "num_input_tokens_seen": 3996216,
      "step": 6045
    },
    {
      "epoch": 3.5672169811320753,
      "grad_norm": 2.3188209533691406,
      "learning_rate": 9.814254325328082e-06,
      "loss": 0.4217,
      "num_input_tokens_seen": 4000472,
      "step": 6050
    },
    {
      "epoch": 3.5701650943396226,
      "grad_norm": 1.8227853775024414,
      "learning_rate": 9.813558968491794e-06,
      "loss": 0.4347,
      "num_input_tokens_seen": 4003384,
      "step": 6055
    },
    {
      "epoch": 3.57311320754717,
      "grad_norm": 2.281665086746216,
      "learning_rate": 9.812862337245024e-06,
      "loss": 0.4712,
      "num_input_tokens_seen": 4006488,
      "step": 6060
    },
    {
      "epoch": 3.576061320754717,
      "grad_norm": 1.4039090871810913,
      "learning_rate": 9.812164431772208e-06,
      "loss": 0.5346,
      "num_input_tokens_seen": 4009784,
      "step": 6065
    },
    {
      "epoch": 3.579009433962264,
      "grad_norm": 1.427514672279358,
      "learning_rate": 9.81146525225812e-06,
      "loss": 0.5006,
      "num_input_tokens_seen": 4014008,
      "step": 6070
    },
    {
      "epoch": 3.5819575471698113,
      "grad_norm": 2.4527087211608887,
      "learning_rate": 9.810764798887868e-06,
      "loss": 0.6148,
      "num_input_tokens_seen": 4016920,
      "step": 6075
    },
    {
      "epoch": 3.5849056603773586,
      "grad_norm": 2.1068873405456543,
      "learning_rate": 9.810063071846905e-06,
      "loss": 0.5133,
      "num_input_tokens_seen": 4020312,
      "step": 6080
    },
    {
      "epoch": 3.5878537735849054,
      "grad_norm": 4.161848068237305,
      "learning_rate": 9.809360071321013e-06,
      "loss": 0.4152,
      "num_input_tokens_seen": 4024024,
      "step": 6085
    },
    {
      "epoch": 3.5908018867924527,
      "grad_norm": 1.9931098222732544,
      "learning_rate": 9.808655797496314e-06,
      "loss": 0.4438,
      "num_input_tokens_seen": 4027064,
      "step": 6090
    },
    {
      "epoch": 3.59375,
      "grad_norm": 1.577402114868164,
      "learning_rate": 9.807950250559268e-06,
      "loss": 0.4349,
      "num_input_tokens_seen": 4030168,
      "step": 6095
    },
    {
      "epoch": 3.5966981132075473,
      "grad_norm": 1.9280726909637451,
      "learning_rate": 9.807243430696673e-06,
      "loss": 0.4176,
      "num_input_tokens_seen": 4033400,
      "step": 6100
    },
    {
      "epoch": 3.5996462264150946,
      "grad_norm": 1.887597680091858,
      "learning_rate": 9.806535338095661e-06,
      "loss": 0.5057,
      "num_input_tokens_seen": 4037368,
      "step": 6105
    },
    {
      "epoch": 3.6025943396226414,
      "grad_norm": 1.0374664068222046,
      "learning_rate": 9.805825972943706e-06,
      "loss": 0.4706,
      "num_input_tokens_seen": 4040248,
      "step": 6110
    },
    {
      "epoch": 3.6055424528301887,
      "grad_norm": 1.4565978050231934,
      "learning_rate": 9.80511533542861e-06,
      "loss": 0.3238,
      "num_input_tokens_seen": 4043224,
      "step": 6115
    },
    {
      "epoch": 3.608490566037736,
      "grad_norm": 0.4926975965499878,
      "learning_rate": 9.80440342573852e-06,
      "loss": 0.5076,
      "num_input_tokens_seen": 4048440,
      "step": 6120
    },
    {
      "epoch": 3.611438679245283,
      "grad_norm": 1.221228837966919,
      "learning_rate": 9.803690244061919e-06,
      "loss": 0.5652,
      "num_input_tokens_seen": 4052312,
      "step": 6125
    },
    {
      "epoch": 3.61438679245283,
      "grad_norm": 1.49860417842865,
      "learning_rate": 9.802975790587621e-06,
      "loss": 0.4541,
      "num_input_tokens_seen": 4055128,
      "step": 6130
    },
    {
      "epoch": 3.6173349056603774,
      "grad_norm": 2.765662431716919,
      "learning_rate": 9.802260065504783e-06,
      "loss": 0.4612,
      "num_input_tokens_seen": 4057560,
      "step": 6135
    },
    {
      "epoch": 3.6202830188679247,
      "grad_norm": 1.2077996730804443,
      "learning_rate": 9.801543069002897e-06,
      "loss": 0.504,
      "num_input_tokens_seen": 4060408,
      "step": 6140
    },
    {
      "epoch": 3.6232311320754715,
      "grad_norm": 3.16392183303833,
      "learning_rate": 9.80082480127179e-06,
      "loss": 0.4168,
      "num_input_tokens_seen": 4063064,
      "step": 6145
    },
    {
      "epoch": 3.626179245283019,
      "grad_norm": 4.2979230880737305,
      "learning_rate": 9.800105262501628e-06,
      "loss": 0.566,
      "num_input_tokens_seen": 4066040,
      "step": 6150
    },
    {
      "epoch": 3.629127358490566,
      "grad_norm": 1.7249126434326172,
      "learning_rate": 9.799384452882907e-06,
      "loss": 0.4525,
      "num_input_tokens_seen": 4068344,
      "step": 6155
    },
    {
      "epoch": 3.632075471698113,
      "grad_norm": 2.607527494430542,
      "learning_rate": 9.798662372606469e-06,
      "loss": 0.6746,
      "num_input_tokens_seen": 4072152,
      "step": 6160
    },
    {
      "epoch": 3.6350235849056602,
      "grad_norm": 1.9324184656143188,
      "learning_rate": 9.797939021863487e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 4074904,
      "step": 6165
    },
    {
      "epoch": 3.6379716981132075,
      "grad_norm": 1.8010355234146118,
      "learning_rate": 9.797214400845472e-06,
      "loss": 0.5732,
      "num_input_tokens_seen": 4077912,
      "step": 6170
    },
    {
      "epoch": 3.640919811320755,
      "grad_norm": 3.7704131603240967,
      "learning_rate": 9.796488509744269e-06,
      "loss": 0.4852,
      "num_input_tokens_seen": 4081624,
      "step": 6175
    },
    {
      "epoch": 3.643867924528302,
      "grad_norm": 1.8000379800796509,
      "learning_rate": 9.79576134875206e-06,
      "loss": 0.5188,
      "num_input_tokens_seen": 4084600,
      "step": 6180
    },
    {
      "epoch": 3.646816037735849,
      "grad_norm": 2.464503526687622,
      "learning_rate": 9.795032918061367e-06,
      "loss": 0.4701,
      "num_input_tokens_seen": 4087384,
      "step": 6185
    },
    {
      "epoch": 3.6497641509433962,
      "grad_norm": 2.4721627235412598,
      "learning_rate": 9.794303217865041e-06,
      "loss": 0.3887,
      "num_input_tokens_seen": 4090584,
      "step": 6190
    },
    {
      "epoch": 3.6527122641509435,
      "grad_norm": 1.407765507698059,
      "learning_rate": 9.79357224835628e-06,
      "loss": 0.4249,
      "num_input_tokens_seen": 4093304,
      "step": 6195
    },
    {
      "epoch": 3.6556603773584904,
      "grad_norm": 1.6793102025985718,
      "learning_rate": 9.792840009728605e-06,
      "loss": 0.391,
      "num_input_tokens_seen": 4096376,
      "step": 6200
    },
    {
      "epoch": 3.6586084905660377,
      "grad_norm": 2.0625193119049072,
      "learning_rate": 9.79210650217588e-06,
      "loss": 0.3752,
      "num_input_tokens_seen": 4102168,
      "step": 6205
    },
    {
      "epoch": 3.661556603773585,
      "grad_norm": 1.7321648597717285,
      "learning_rate": 9.791371725892307e-06,
      "loss": 0.4356,
      "num_input_tokens_seen": 4105944,
      "step": 6210
    },
    {
      "epoch": 3.6645047169811322,
      "grad_norm": 1.3256168365478516,
      "learning_rate": 9.79063568107242e-06,
      "loss": 0.5673,
      "num_input_tokens_seen": 4109368,
      "step": 6215
    },
    {
      "epoch": 3.6674528301886795,
      "grad_norm": 5.215675354003906,
      "learning_rate": 9.78989836791109e-06,
      "loss": 0.4157,
      "num_input_tokens_seen": 4112856,
      "step": 6220
    },
    {
      "epoch": 3.6704009433962264,
      "grad_norm": 1.4015494585037231,
      "learning_rate": 9.789159786603524e-06,
      "loss": 0.4811,
      "num_input_tokens_seen": 4115384,
      "step": 6225
    },
    {
      "epoch": 3.6733490566037736,
      "grad_norm": 2.039726734161377,
      "learning_rate": 9.788419937345263e-06,
      "loss": 0.4105,
      "num_input_tokens_seen": 4119128,
      "step": 6230
    },
    {
      "epoch": 3.6762971698113205,
      "grad_norm": 1.7354063987731934,
      "learning_rate": 9.787678820332188e-06,
      "loss": 0.4343,
      "num_input_tokens_seen": 4121816,
      "step": 6235
    },
    {
      "epoch": 3.6792452830188678,
      "grad_norm": 1.4166511297225952,
      "learning_rate": 9.78693643576051e-06,
      "loss": 0.3914,
      "num_input_tokens_seen": 4125272,
      "step": 6240
    },
    {
      "epoch": 3.682193396226415,
      "grad_norm": 1.1471704244613647,
      "learning_rate": 9.786192783826782e-06,
      "loss": 0.5531,
      "num_input_tokens_seen": 4128312,
      "step": 6245
    },
    {
      "epoch": 3.6851415094339623,
      "grad_norm": 1.9735784530639648,
      "learning_rate": 9.785447864727887e-06,
      "loss": 0.5043,
      "num_input_tokens_seen": 4131480,
      "step": 6250
    },
    {
      "epoch": 3.6880896226415096,
      "grad_norm": 2.915173053741455,
      "learning_rate": 9.784701678661045e-06,
      "loss": 0.4759,
      "num_input_tokens_seen": 4133880,
      "step": 6255
    },
    {
      "epoch": 3.6910377358490565,
      "grad_norm": 4.3081254959106445,
      "learning_rate": 9.783954225823813e-06,
      "loss": 0.428,
      "num_input_tokens_seen": 4137816,
      "step": 6260
    },
    {
      "epoch": 3.6939858490566038,
      "grad_norm": 3.823376417160034,
      "learning_rate": 9.783205506414082e-06,
      "loss": 0.5055,
      "num_input_tokens_seen": 4140632,
      "step": 6265
    },
    {
      "epoch": 3.696933962264151,
      "grad_norm": 3.010692834854126,
      "learning_rate": 9.782455520630079e-06,
      "loss": 0.5313,
      "num_input_tokens_seen": 4144184,
      "step": 6270
    },
    {
      "epoch": 3.699882075471698,
      "grad_norm": 2.216198444366455,
      "learning_rate": 9.781704268670364e-06,
      "loss": 0.411,
      "num_input_tokens_seen": 4146808,
      "step": 6275
    },
    {
      "epoch": 3.702830188679245,
      "grad_norm": 1.842231035232544,
      "learning_rate": 9.780951750733837e-06,
      "loss": 0.413,
      "num_input_tokens_seen": 4150840,
      "step": 6280
    },
    {
      "epoch": 3.7057783018867925,
      "grad_norm": 2.6239125728607178,
      "learning_rate": 9.780197967019728e-06,
      "loss": 0.3545,
      "num_input_tokens_seen": 4152920,
      "step": 6285
    },
    {
      "epoch": 3.7087264150943398,
      "grad_norm": 1.4602774381637573,
      "learning_rate": 9.779442917727608e-06,
      "loss": 0.4635,
      "num_input_tokens_seen": 4155704,
      "step": 6290
    },
    {
      "epoch": 3.711674528301887,
      "grad_norm": 2.101463794708252,
      "learning_rate": 9.778686603057377e-06,
      "loss": 0.4484,
      "num_input_tokens_seen": 4159416,
      "step": 6295
    },
    {
      "epoch": 3.714622641509434,
      "grad_norm": 2.164400577545166,
      "learning_rate": 9.777929023209271e-06,
      "loss": 0.6352,
      "num_input_tokens_seen": 4163096,
      "step": 6300
    },
    {
      "epoch": 3.717570754716981,
      "grad_norm": 3.560542106628418,
      "learning_rate": 9.777170178383866e-06,
      "loss": 0.5245,
      "num_input_tokens_seen": 4167576,
      "step": 6305
    },
    {
      "epoch": 3.7205188679245285,
      "grad_norm": 2.1344165802001953,
      "learning_rate": 9.776410068782068e-06,
      "loss": 0.5607,
      "num_input_tokens_seen": 4171192,
      "step": 6310
    },
    {
      "epoch": 3.7234669811320753,
      "grad_norm": 1.8679028749465942,
      "learning_rate": 9.775648694605118e-06,
      "loss": 0.5789,
      "num_input_tokens_seen": 4175128,
      "step": 6315
    },
    {
      "epoch": 3.7264150943396226,
      "grad_norm": 1.8186396360397339,
      "learning_rate": 9.774886056054593e-06,
      "loss": 0.3881,
      "num_input_tokens_seen": 4178584,
      "step": 6320
    },
    {
      "epoch": 3.72936320754717,
      "grad_norm": 2.299381971359253,
      "learning_rate": 9.774122153332408e-06,
      "loss": 0.5847,
      "num_input_tokens_seen": 4181688,
      "step": 6325
    },
    {
      "epoch": 3.732311320754717,
      "grad_norm": 3.302819013595581,
      "learning_rate": 9.773356986640807e-06,
      "loss": 0.5736,
      "num_input_tokens_seen": 4184152,
      "step": 6330
    },
    {
      "epoch": 3.735259433962264,
      "grad_norm": 3.4716250896453857,
      "learning_rate": 9.772590556182373e-06,
      "loss": 0.5008,
      "num_input_tokens_seen": 4186808,
      "step": 6335
    },
    {
      "epoch": 3.7382075471698113,
      "grad_norm": 1.9623467922210693,
      "learning_rate": 9.77182286216002e-06,
      "loss": 0.4284,
      "num_input_tokens_seen": 4189784,
      "step": 6340
    },
    {
      "epoch": 3.7411556603773586,
      "grad_norm": 2.107465982437134,
      "learning_rate": 9.771053904776998e-06,
      "loss": 0.4858,
      "num_input_tokens_seen": 4195512,
      "step": 6345
    },
    {
      "epoch": 3.7441037735849054,
      "grad_norm": 1.527693271636963,
      "learning_rate": 9.770283684236891e-06,
      "loss": 0.4829,
      "num_input_tokens_seen": 4199096,
      "step": 6350
    },
    {
      "epoch": 3.7470518867924527,
      "grad_norm": 1.0778632164001465,
      "learning_rate": 9.769512200743623e-06,
      "loss": 0.5858,
      "num_input_tokens_seen": 4202072,
      "step": 6355
    },
    {
      "epoch": 3.75,
      "grad_norm": 2.0923354625701904,
      "learning_rate": 9.768739454501444e-06,
      "loss": 0.4654,
      "num_input_tokens_seen": 4205208,
      "step": 6360
    },
    {
      "epoch": 3.7529481132075473,
      "grad_norm": 3.430405616760254,
      "learning_rate": 9.76796544571494e-06,
      "loss": 0.4521,
      "num_input_tokens_seen": 4208088,
      "step": 6365
    },
    {
      "epoch": 3.7558962264150946,
      "grad_norm": 2.1894729137420654,
      "learning_rate": 9.767190174589036e-06,
      "loss": 0.3361,
      "num_input_tokens_seen": 4210616,
      "step": 6370
    },
    {
      "epoch": 3.7588443396226414,
      "grad_norm": 2.2627360820770264,
      "learning_rate": 9.76641364132899e-06,
      "loss": 0.5059,
      "num_input_tokens_seen": 4213176,
      "step": 6375
    },
    {
      "epoch": 3.7617924528301887,
      "grad_norm": 2.377472162246704,
      "learning_rate": 9.765635846140389e-06,
      "loss": 0.62,
      "num_input_tokens_seen": 4219960,
      "step": 6380
    },
    {
      "epoch": 3.764740566037736,
      "grad_norm": 1.3392698764801025,
      "learning_rate": 9.764856789229157e-06,
      "loss": 0.4568,
      "num_input_tokens_seen": 4223352,
      "step": 6385
    },
    {
      "epoch": 3.767688679245283,
      "grad_norm": 1.3347755670547485,
      "learning_rate": 9.764076470801557e-06,
      "loss": 0.5276,
      "num_input_tokens_seen": 4227160,
      "step": 6390
    },
    {
      "epoch": 3.77063679245283,
      "grad_norm": 2.5613937377929688,
      "learning_rate": 9.763294891064182e-06,
      "loss": 0.5215,
      "num_input_tokens_seen": 4230008,
      "step": 6395
    },
    {
      "epoch": 3.7735849056603774,
      "grad_norm": 1.5592621564865112,
      "learning_rate": 9.762512050223951e-06,
      "loss": 0.3512,
      "num_input_tokens_seen": 4232760,
      "step": 6400
    },
    {
      "epoch": 3.7765330188679247,
      "grad_norm": 2.1239981651306152,
      "learning_rate": 9.761727948488132e-06,
      "loss": 0.4612,
      "num_input_tokens_seen": 4236056,
      "step": 6405
    },
    {
      "epoch": 3.7794811320754715,
      "grad_norm": 1.4761896133422852,
      "learning_rate": 9.760942586064315e-06,
      "loss": 0.3381,
      "num_input_tokens_seen": 4239768,
      "step": 6410
    },
    {
      "epoch": 3.782429245283019,
      "grad_norm": 1.3835372924804688,
      "learning_rate": 9.760155963160431e-06,
      "loss": 0.4596,
      "num_input_tokens_seen": 4242552,
      "step": 6415
    },
    {
      "epoch": 3.785377358490566,
      "grad_norm": 2.1916744709014893,
      "learning_rate": 9.759368079984741e-06,
      "loss": 0.8526,
      "num_input_tokens_seen": 4245912,
      "step": 6420
    },
    {
      "epoch": 3.788325471698113,
      "grad_norm": 1.7321279048919678,
      "learning_rate": 9.758578936745839e-06,
      "loss": 0.4198,
      "num_input_tokens_seen": 4248376,
      "step": 6425
    },
    {
      "epoch": 3.7912735849056602,
      "grad_norm": 1.8776565790176392,
      "learning_rate": 9.757788533652656e-06,
      "loss": 0.4698,
      "num_input_tokens_seen": 4251224,
      "step": 6430
    },
    {
      "epoch": 3.7942216981132075,
      "grad_norm": 1.9327930212020874,
      "learning_rate": 9.756996870914454e-06,
      "loss": 0.4677,
      "num_input_tokens_seen": 4254648,
      "step": 6435
    },
    {
      "epoch": 3.797169811320755,
      "grad_norm": 2.4032862186431885,
      "learning_rate": 9.756203948740828e-06,
      "loss": 0.6443,
      "num_input_tokens_seen": 4257304,
      "step": 6440
    },
    {
      "epoch": 3.800117924528302,
      "grad_norm": 1.9451631307601929,
      "learning_rate": 9.755409767341709e-06,
      "loss": 0.4746,
      "num_input_tokens_seen": 4260216,
      "step": 6445
    },
    {
      "epoch": 3.803066037735849,
      "grad_norm": 1.1113866567611694,
      "learning_rate": 9.75461432692736e-06,
      "loss": 0.3998,
      "num_input_tokens_seen": 4264568,
      "step": 6450
    },
    {
      "epoch": 3.8060141509433962,
      "grad_norm": 3.064146041870117,
      "learning_rate": 9.753817627708375e-06,
      "loss": 0.4958,
      "num_input_tokens_seen": 4268696,
      "step": 6455
    },
    {
      "epoch": 3.8089622641509435,
      "grad_norm": 1.8140944242477417,
      "learning_rate": 9.753019669895686e-06,
      "loss": 0.417,
      "num_input_tokens_seen": 4271800,
      "step": 6460
    },
    {
      "epoch": 3.8119103773584904,
      "grad_norm": 2.0475385189056396,
      "learning_rate": 9.752220453700556e-06,
      "loss": 0.4465,
      "num_input_tokens_seen": 4275256,
      "step": 6465
    },
    {
      "epoch": 3.8148584905660377,
      "grad_norm": 1.5813344717025757,
      "learning_rate": 9.75141997933458e-06,
      "loss": 0.4025,
      "num_input_tokens_seen": 4278040,
      "step": 6470
    },
    {
      "epoch": 3.817806603773585,
      "grad_norm": 1.3078409433364868,
      "learning_rate": 9.750618247009685e-06,
      "loss": 0.4501,
      "num_input_tokens_seen": 4281592,
      "step": 6475
    },
    {
      "epoch": 3.8207547169811322,
      "grad_norm": 0.9804493188858032,
      "learning_rate": 9.749815256938138e-06,
      "loss": 0.362,
      "num_input_tokens_seen": 4284920,
      "step": 6480
    },
    {
      "epoch": 3.8237028301886795,
      "grad_norm": 1.5708634853363037,
      "learning_rate": 9.749011009332529e-06,
      "loss": 0.542,
      "num_input_tokens_seen": 4287960,
      "step": 6485
    },
    {
      "epoch": 3.8266509433962264,
      "grad_norm": 1.6159676313400269,
      "learning_rate": 9.748205504405787e-06,
      "loss": 0.3796,
      "num_input_tokens_seen": 4290872,
      "step": 6490
    },
    {
      "epoch": 3.8295990566037736,
      "grad_norm": 1.8157576322555542,
      "learning_rate": 9.747398742371177e-06,
      "loss": 0.5179,
      "num_input_tokens_seen": 4294008,
      "step": 6495
    },
    {
      "epoch": 3.8325471698113205,
      "grad_norm": 2.140465021133423,
      "learning_rate": 9.746590723442289e-06,
      "loss": 0.517,
      "num_input_tokens_seen": 4297688,
      "step": 6500
    },
    {
      "epoch": 3.8354952830188678,
      "grad_norm": 1.1031979322433472,
      "learning_rate": 9.745781447833049e-06,
      "loss": 0.5107,
      "num_input_tokens_seen": 4301176,
      "step": 6505
    },
    {
      "epoch": 3.838443396226415,
      "grad_norm": 1.9653302431106567,
      "learning_rate": 9.74497091575772e-06,
      "loss": 0.3731,
      "num_input_tokens_seen": 4304152,
      "step": 6510
    },
    {
      "epoch": 3.8413915094339623,
      "grad_norm": 1.3907338380813599,
      "learning_rate": 9.744159127430888e-06,
      "loss": 0.6074,
      "num_input_tokens_seen": 4307320,
      "step": 6515
    },
    {
      "epoch": 3.8443396226415096,
      "grad_norm": 1.3061860799789429,
      "learning_rate": 9.743346083067482e-06,
      "loss": 0.407,
      "num_input_tokens_seen": 4310232,
      "step": 6520
    },
    {
      "epoch": 3.8472877358490565,
      "grad_norm": 1.886415719985962,
      "learning_rate": 9.742531782882758e-06,
      "loss": 0.586,
      "num_input_tokens_seen": 4313560,
      "step": 6525
    },
    {
      "epoch": 3.8502358490566038,
      "grad_norm": 2.0596909523010254,
      "learning_rate": 9.741716227092305e-06,
      "loss": 0.4633,
      "num_input_tokens_seen": 4316568,
      "step": 6530
    },
    {
      "epoch": 3.853183962264151,
      "grad_norm": 1.2584502696990967,
      "learning_rate": 9.740899415912048e-06,
      "loss": 0.3569,
      "num_input_tokens_seen": 4319480,
      "step": 6535
    },
    {
      "epoch": 3.856132075471698,
      "grad_norm": 1.4106732606887817,
      "learning_rate": 9.740081349558236e-06,
      "loss": 0.5087,
      "num_input_tokens_seen": 4322744,
      "step": 6540
    },
    {
      "epoch": 3.859080188679245,
      "grad_norm": 1.825339436531067,
      "learning_rate": 9.739262028247459e-06,
      "loss": 0.3377,
      "num_input_tokens_seen": 4325592,
      "step": 6545
    },
    {
      "epoch": 3.8620283018867925,
      "grad_norm": 1.4263280630111694,
      "learning_rate": 9.738441452196633e-06,
      "loss": 0.5049,
      "num_input_tokens_seen": 4328984,
      "step": 6550
    },
    {
      "epoch": 3.8649764150943398,
      "grad_norm": 1.5841104984283447,
      "learning_rate": 9.737619621623013e-06,
      "loss": 0.4847,
      "num_input_tokens_seen": 4332376,
      "step": 6555
    },
    {
      "epoch": 3.867924528301887,
      "grad_norm": 1.3872623443603516,
      "learning_rate": 9.73679653674418e-06,
      "loss": 0.4775,
      "num_input_tokens_seen": 4337112,
      "step": 6560
    },
    {
      "epoch": 3.870872641509434,
      "grad_norm": 1.5526912212371826,
      "learning_rate": 9.735972197778047e-06,
      "loss": 0.4618,
      "num_input_tokens_seen": 4339768,
      "step": 6565
    },
    {
      "epoch": 3.873820754716981,
      "grad_norm": 2.41532564163208,
      "learning_rate": 9.735146604942867e-06,
      "loss": 0.475,
      "num_input_tokens_seen": 4343224,
      "step": 6570
    },
    {
      "epoch": 3.8767688679245285,
      "grad_norm": 2.304701805114746,
      "learning_rate": 9.734319758457214e-06,
      "loss": 0.4637,
      "num_input_tokens_seen": 4347224,
      "step": 6575
    },
    {
      "epoch": 3.8797169811320753,
      "grad_norm": 1.5605089664459229,
      "learning_rate": 9.733491658540001e-06,
      "loss": 0.5076,
      "num_input_tokens_seen": 4350456,
      "step": 6580
    },
    {
      "epoch": 3.8826650943396226,
      "grad_norm": 2.9778549671173096,
      "learning_rate": 9.732662305410474e-06,
      "loss": 0.5303,
      "num_input_tokens_seen": 4353368,
      "step": 6585
    },
    {
      "epoch": 3.88561320754717,
      "grad_norm": 2.6693007946014404,
      "learning_rate": 9.731831699288203e-06,
      "loss": 0.5727,
      "num_input_tokens_seen": 4356120,
      "step": 6590
    },
    {
      "epoch": 3.888561320754717,
      "grad_norm": 1.6506192684173584,
      "learning_rate": 9.730999840393096e-06,
      "loss": 0.5971,
      "num_input_tokens_seen": 4359288,
      "step": 6595
    },
    {
      "epoch": 3.891509433962264,
      "grad_norm": 2.3919756412506104,
      "learning_rate": 9.730166728945391e-06,
      "loss": 0.5912,
      "num_input_tokens_seen": 4362648,
      "step": 6600
    },
    {
      "epoch": 3.8944575471698113,
      "grad_norm": 3.2053802013397217,
      "learning_rate": 9.72933236516566e-06,
      "loss": 0.4529,
      "num_input_tokens_seen": 4365176,
      "step": 6605
    },
    {
      "epoch": 3.8974056603773586,
      "grad_norm": 9.990158081054688,
      "learning_rate": 9.728496749274806e-06,
      "loss": 0.5033,
      "num_input_tokens_seen": 4368408,
      "step": 6610
    },
    {
      "epoch": 3.9003537735849054,
      "grad_norm": 5.007667064666748,
      "learning_rate": 9.727659881494054e-06,
      "loss": 0.5807,
      "num_input_tokens_seen": 4370520,
      "step": 6615
    },
    {
      "epoch": 3.9033018867924527,
      "grad_norm": 1.6647617816925049,
      "learning_rate": 9.726821762044975e-06,
      "loss": 0.4864,
      "num_input_tokens_seen": 4373400,
      "step": 6620
    },
    {
      "epoch": 3.90625,
      "grad_norm": 2.3470141887664795,
      "learning_rate": 9.725982391149465e-06,
      "loss": 0.6862,
      "num_input_tokens_seen": 4376600,
      "step": 6625
    },
    {
      "epoch": 3.9091981132075473,
      "grad_norm": 1.8078652620315552,
      "learning_rate": 9.725141769029747e-06,
      "loss": 0.3106,
      "num_input_tokens_seen": 4381592,
      "step": 6630
    },
    {
      "epoch": 3.9121462264150946,
      "grad_norm": 1.9628053903579712,
      "learning_rate": 9.72429989590838e-06,
      "loss": 0.5517,
      "num_input_tokens_seen": 4384600,
      "step": 6635
    },
    {
      "epoch": 3.9150943396226414,
      "grad_norm": 1.2512035369873047,
      "learning_rate": 9.723456772008257e-06,
      "loss": 0.5346,
      "num_input_tokens_seen": 4387864,
      "step": 6640
    },
    {
      "epoch": 3.9180424528301887,
      "grad_norm": 2.335808038711548,
      "learning_rate": 9.722612397552598e-06,
      "loss": 0.5225,
      "num_input_tokens_seen": 4390808,
      "step": 6645
    },
    {
      "epoch": 3.920990566037736,
      "grad_norm": 1.37651789188385,
      "learning_rate": 9.72176677276495e-06,
      "loss": 0.6735,
      "num_input_tokens_seen": 4394296,
      "step": 6650
    },
    {
      "epoch": 3.923938679245283,
      "grad_norm": 1.8377697467803955,
      "learning_rate": 9.7209198978692e-06,
      "loss": 0.634,
      "num_input_tokens_seen": 4396760,
      "step": 6655
    },
    {
      "epoch": 3.92688679245283,
      "grad_norm": 1.1158671379089355,
      "learning_rate": 9.720071773089564e-06,
      "loss": 0.7765,
      "num_input_tokens_seen": 4399128,
      "step": 6660
    },
    {
      "epoch": 3.9298349056603774,
      "grad_norm": 2.3328628540039062,
      "learning_rate": 9.71922239865058e-06,
      "loss": 0.4751,
      "num_input_tokens_seen": 4402456,
      "step": 6665
    },
    {
      "epoch": 3.9327830188679247,
      "grad_norm": 1.4910801649093628,
      "learning_rate": 9.718371774777131e-06,
      "loss": 0.4265,
      "num_input_tokens_seen": 4405400,
      "step": 6670
    },
    {
      "epoch": 3.9357311320754715,
      "grad_norm": 2.029214382171631,
      "learning_rate": 9.717519901694416e-06,
      "loss": 0.5091,
      "num_input_tokens_seen": 4407992,
      "step": 6675
    },
    {
      "epoch": 3.938679245283019,
      "grad_norm": 2.571441650390625,
      "learning_rate": 9.716666779627978e-06,
      "loss": 0.4487,
      "num_input_tokens_seen": 4411512,
      "step": 6680
    },
    {
      "epoch": 3.941627358490566,
      "grad_norm": 2.849327325820923,
      "learning_rate": 9.715812408803681e-06,
      "loss": 0.5084,
      "num_input_tokens_seen": 4414296,
      "step": 6685
    },
    {
      "epoch": 3.944575471698113,
      "grad_norm": 2.0784857273101807,
      "learning_rate": 9.714956789447726e-06,
      "loss": 0.3872,
      "num_input_tokens_seen": 4416856,
      "step": 6690
    },
    {
      "epoch": 3.9475235849056602,
      "grad_norm": 2.9747719764709473,
      "learning_rate": 9.71409992178664e-06,
      "loss": 0.5722,
      "num_input_tokens_seen": 4419928,
      "step": 6695
    },
    {
      "epoch": 3.9504716981132075,
      "grad_norm": 2.2301552295684814,
      "learning_rate": 9.713241806047282e-06,
      "loss": 0.5143,
      "num_input_tokens_seen": 4422328,
      "step": 6700
    },
    {
      "epoch": 3.953419811320755,
      "grad_norm": 2.2421693801879883,
      "learning_rate": 9.712382442456845e-06,
      "loss": 0.4196,
      "num_input_tokens_seen": 4426168,
      "step": 6705
    },
    {
      "epoch": 3.956367924528302,
      "grad_norm": 1.8638815879821777,
      "learning_rate": 9.711521831242846e-06,
      "loss": 0.5061,
      "num_input_tokens_seen": 4429112,
      "step": 6710
    },
    {
      "epoch": 3.959316037735849,
      "grad_norm": 2.725217580795288,
      "learning_rate": 9.710659972633137e-06,
      "loss": 0.5524,
      "num_input_tokens_seen": 4431896,
      "step": 6715
    },
    {
      "epoch": 3.9622641509433962,
      "grad_norm": 1.0478849411010742,
      "learning_rate": 9.709796866855899e-06,
      "loss": 0.5136,
      "num_input_tokens_seen": 4434808,
      "step": 6720
    },
    {
      "epoch": 3.9652122641509435,
      "grad_norm": 4.103155136108398,
      "learning_rate": 9.70893251413964e-06,
      "loss": 0.4646,
      "num_input_tokens_seen": 4437400,
      "step": 6725
    },
    {
      "epoch": 3.9681603773584904,
      "grad_norm": 1.4258785247802734,
      "learning_rate": 9.708066914713205e-06,
      "loss": 0.4121,
      "num_input_tokens_seen": 4440824,
      "step": 6730
    },
    {
      "epoch": 3.9711084905660377,
      "grad_norm": 2.5916833877563477,
      "learning_rate": 9.707200068805764e-06,
      "loss": 0.4789,
      "num_input_tokens_seen": 4445304,
      "step": 6735
    },
    {
      "epoch": 3.974056603773585,
      "grad_norm": 1.5275408029556274,
      "learning_rate": 9.706331976646817e-06,
      "loss": 0.2685,
      "num_input_tokens_seen": 4448056,
      "step": 6740
    },
    {
      "epoch": 3.9770047169811322,
      "grad_norm": 1.8412368297576904,
      "learning_rate": 9.705462638466197e-06,
      "loss": 0.4223,
      "num_input_tokens_seen": 4450872,
      "step": 6745
    },
    {
      "epoch": 3.9799528301886795,
      "grad_norm": 1.7090975046157837,
      "learning_rate": 9.704592054494065e-06,
      "loss": 0.4387,
      "num_input_tokens_seen": 4454072,
      "step": 6750
    },
    {
      "epoch": 3.9829009433962264,
      "grad_norm": 2.5332393646240234,
      "learning_rate": 9.703720224960909e-06,
      "loss": 0.5674,
      "num_input_tokens_seen": 4456952,
      "step": 6755
    },
    {
      "epoch": 3.9858490566037736,
      "grad_norm": 1.3531749248504639,
      "learning_rate": 9.702847150097552e-06,
      "loss": 0.3907,
      "num_input_tokens_seen": 4460888,
      "step": 6760
    },
    {
      "epoch": 3.9887971698113205,
      "grad_norm": 1.3580771684646606,
      "learning_rate": 9.701972830135143e-06,
      "loss": 0.4623,
      "num_input_tokens_seen": 4464152,
      "step": 6765
    },
    {
      "epoch": 3.9917452830188678,
      "grad_norm": 1.3621846437454224,
      "learning_rate": 9.701097265305164e-06,
      "loss": 0.4955,
      "num_input_tokens_seen": 4467704,
      "step": 6770
    },
    {
      "epoch": 3.994693396226415,
      "grad_norm": 1.3718831539154053,
      "learning_rate": 9.700220455839422e-06,
      "loss": 0.3201,
      "num_input_tokens_seen": 4471736,
      "step": 6775
    },
    {
      "epoch": 3.9976415094339623,
      "grad_norm": 3.7190420627593994,
      "learning_rate": 9.69934240197006e-06,
      "loss": 0.4011,
      "num_input_tokens_seen": 4475128,
      "step": 6780
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.4945891797542572,
      "eval_runtime": 18.3669,
      "eval_samples_per_second": 92.34,
      "eval_steps_per_second": 23.085,
      "num_input_tokens_seen": 4476776,
      "step": 6784
    },
    {
      "epoch": 4.00058962264151,
      "grad_norm": 2.452103614807129,
      "learning_rate": 9.698463103929542e-06,
      "loss": 0.3097,
      "num_input_tokens_seen": 4477256,
      "step": 6785
    },
    {
      "epoch": 4.003537735849057,
      "grad_norm": 2.1431336402893066,
      "learning_rate": 9.697582561950669e-06,
      "loss": 0.3667,
      "num_input_tokens_seen": 4481000,
      "step": 6790
    },
    {
      "epoch": 4.006485849056604,
      "grad_norm": 1.7345376014709473,
      "learning_rate": 9.696700776266568e-06,
      "loss": 0.4009,
      "num_input_tokens_seen": 4485000,
      "step": 6795
    },
    {
      "epoch": 4.009433962264151,
      "grad_norm": 1.703779697418213,
      "learning_rate": 9.695817747110694e-06,
      "loss": 0.3624,
      "num_input_tokens_seen": 4489352,
      "step": 6800
    },
    {
      "epoch": 4.012382075471698,
      "grad_norm": 1.9731892347335815,
      "learning_rate": 9.694933474716831e-06,
      "loss": 0.3813,
      "num_input_tokens_seen": 4492776,
      "step": 6805
    },
    {
      "epoch": 4.015330188679245,
      "grad_norm": 1.960410475730896,
      "learning_rate": 9.6940479593191e-06,
      "loss": 0.5304,
      "num_input_tokens_seen": 4495432,
      "step": 6810
    },
    {
      "epoch": 4.0182783018867925,
      "grad_norm": 0.9555906057357788,
      "learning_rate": 9.693161201151942e-06,
      "loss": 0.4689,
      "num_input_tokens_seen": 4499304,
      "step": 6815
    },
    {
      "epoch": 4.02122641509434,
      "grad_norm": 1.5389059782028198,
      "learning_rate": 9.692273200450128e-06,
      "loss": 0.43,
      "num_input_tokens_seen": 4502120,
      "step": 6820
    },
    {
      "epoch": 4.024174528301887,
      "grad_norm": 1.1791049242019653,
      "learning_rate": 9.69138395744876e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 4505800,
      "step": 6825
    },
    {
      "epoch": 4.027122641509434,
      "grad_norm": 4.463519096374512,
      "learning_rate": 9.690493472383274e-06,
      "loss": 0.5893,
      "num_input_tokens_seen": 4508776,
      "step": 6830
    },
    {
      "epoch": 4.030070754716981,
      "grad_norm": 1.9662586450576782,
      "learning_rate": 9.689601745489423e-06,
      "loss": 0.4979,
      "num_input_tokens_seen": 4511976,
      "step": 6835
    },
    {
      "epoch": 4.033018867924528,
      "grad_norm": 2.808878183364868,
      "learning_rate": 9.6887087770033e-06,
      "loss": 0.3465,
      "num_input_tokens_seen": 4515496,
      "step": 6840
    },
    {
      "epoch": 4.035966981132075,
      "grad_norm": 2.7323219776153564,
      "learning_rate": 9.687814567161322e-06,
      "loss": 0.4798,
      "num_input_tokens_seen": 4519752,
      "step": 6845
    },
    {
      "epoch": 4.038915094339623,
      "grad_norm": 2.265474557876587,
      "learning_rate": 9.686919116200232e-06,
      "loss": 0.5665,
      "num_input_tokens_seen": 4522824,
      "step": 6850
    },
    {
      "epoch": 4.04186320754717,
      "grad_norm": 1.9944087266921997,
      "learning_rate": 9.686022424357108e-06,
      "loss": 0.4818,
      "num_input_tokens_seen": 4525704,
      "step": 6855
    },
    {
      "epoch": 4.044811320754717,
      "grad_norm": 1.6783710718154907,
      "learning_rate": 9.685124491869353e-06,
      "loss": 0.4249,
      "num_input_tokens_seen": 4528616,
      "step": 6860
    },
    {
      "epoch": 4.0477594339622645,
      "grad_norm": 2.2345335483551025,
      "learning_rate": 9.684225318974696e-06,
      "loss": 0.6815,
      "num_input_tokens_seen": 4532136,
      "step": 6865
    },
    {
      "epoch": 4.050707547169812,
      "grad_norm": 1.5228191614151,
      "learning_rate": 9.683324905911197e-06,
      "loss": 0.4211,
      "num_input_tokens_seen": 4535048,
      "step": 6870
    },
    {
      "epoch": 4.053655660377358,
      "grad_norm": 2.54984712600708,
      "learning_rate": 9.682423252917245e-06,
      "loss": 0.5969,
      "num_input_tokens_seen": 4539272,
      "step": 6875
    },
    {
      "epoch": 4.056603773584905,
      "grad_norm": 1.7036157846450806,
      "learning_rate": 9.681520360231557e-06,
      "loss": 0.404,
      "num_input_tokens_seen": 4543432,
      "step": 6880
    },
    {
      "epoch": 4.059551886792453,
      "grad_norm": 2.4191641807556152,
      "learning_rate": 9.680616228093178e-06,
      "loss": 0.4648,
      "num_input_tokens_seen": 4546408,
      "step": 6885
    },
    {
      "epoch": 4.0625,
      "grad_norm": 2.5568714141845703,
      "learning_rate": 9.67971085674148e-06,
      "loss": 0.429,
      "num_input_tokens_seen": 4549512,
      "step": 6890
    },
    {
      "epoch": 4.065448113207547,
      "grad_norm": 1.8029133081436157,
      "learning_rate": 9.678804246416164e-06,
      "loss": 0.3032,
      "num_input_tokens_seen": 4552648,
      "step": 6895
    },
    {
      "epoch": 4.068396226415095,
      "grad_norm": 1.533697247505188,
      "learning_rate": 9.677896397357259e-06,
      "loss": 0.4,
      "num_input_tokens_seen": 4556744,
      "step": 6900
    },
    {
      "epoch": 4.071344339622642,
      "grad_norm": 1.9169901609420776,
      "learning_rate": 9.676987309805121e-06,
      "loss": 0.5477,
      "num_input_tokens_seen": 4559688,
      "step": 6905
    },
    {
      "epoch": 4.074292452830188,
      "grad_norm": 1.6373481750488281,
      "learning_rate": 9.67607698400044e-06,
      "loss": 0.3645,
      "num_input_tokens_seen": 4562376,
      "step": 6910
    },
    {
      "epoch": 4.0772405660377355,
      "grad_norm": 1.8949525356292725,
      "learning_rate": 9.67516542018422e-06,
      "loss": 0.5246,
      "num_input_tokens_seen": 4565032,
      "step": 6915
    },
    {
      "epoch": 4.080188679245283,
      "grad_norm": 1.9811887741088867,
      "learning_rate": 9.67425261859781e-06,
      "loss": 0.5031,
      "num_input_tokens_seen": 4567880,
      "step": 6920
    },
    {
      "epoch": 4.08313679245283,
      "grad_norm": 2.5787534713745117,
      "learning_rate": 9.673338579482871e-06,
      "loss": 0.439,
      "num_input_tokens_seen": 4571976,
      "step": 6925
    },
    {
      "epoch": 4.086084905660377,
      "grad_norm": 2.2286829948425293,
      "learning_rate": 9.672423303081404e-06,
      "loss": 0.2951,
      "num_input_tokens_seen": 4575720,
      "step": 6930
    },
    {
      "epoch": 4.089033018867925,
      "grad_norm": 3.015331983566284,
      "learning_rate": 9.67150678963573e-06,
      "loss": 0.3466,
      "num_input_tokens_seen": 4578664,
      "step": 6935
    },
    {
      "epoch": 4.091981132075472,
      "grad_norm": 2.1125972270965576,
      "learning_rate": 9.670589039388501e-06,
      "loss": 0.4631,
      "num_input_tokens_seen": 4581544,
      "step": 6940
    },
    {
      "epoch": 4.094929245283019,
      "grad_norm": 1.6815558671951294,
      "learning_rate": 9.669670052582695e-06,
      "loss": 0.3037,
      "num_input_tokens_seen": 4584072,
      "step": 6945
    },
    {
      "epoch": 4.097877358490566,
      "grad_norm": 3.152862548828125,
      "learning_rate": 9.668749829461617e-06,
      "loss": 0.4392,
      "num_input_tokens_seen": 4587432,
      "step": 6950
    },
    {
      "epoch": 4.100825471698113,
      "grad_norm": 2.743082284927368,
      "learning_rate": 9.667828370268898e-06,
      "loss": 0.4542,
      "num_input_tokens_seen": 4590088,
      "step": 6955
    },
    {
      "epoch": 4.10377358490566,
      "grad_norm": 2.8095293045043945,
      "learning_rate": 9.666905675248505e-06,
      "loss": 0.4443,
      "num_input_tokens_seen": 4592872,
      "step": 6960
    },
    {
      "epoch": 4.1067216981132075,
      "grad_norm": 1.5299906730651855,
      "learning_rate": 9.66598174464472e-06,
      "loss": 0.469,
      "num_input_tokens_seen": 4595848,
      "step": 6965
    },
    {
      "epoch": 4.109669811320755,
      "grad_norm": 1.445300579071045,
      "learning_rate": 9.665056578702157e-06,
      "loss": 0.507,
      "num_input_tokens_seen": 4598632,
      "step": 6970
    },
    {
      "epoch": 4.112617924528302,
      "grad_norm": 1.7791574001312256,
      "learning_rate": 9.66413017766576e-06,
      "loss": 0.4464,
      "num_input_tokens_seen": 4602056,
      "step": 6975
    },
    {
      "epoch": 4.115566037735849,
      "grad_norm": 1.0910542011260986,
      "learning_rate": 9.663202541780799e-06,
      "loss": 0.3441,
      "num_input_tokens_seen": 4605160,
      "step": 6980
    },
    {
      "epoch": 4.118514150943396,
      "grad_norm": 1.1069021224975586,
      "learning_rate": 9.662273671292866e-06,
      "loss": 0.3655,
      "num_input_tokens_seen": 4608840,
      "step": 6985
    },
    {
      "epoch": 4.121462264150943,
      "grad_norm": 0.9430096745491028,
      "learning_rate": 9.661343566447886e-06,
      "loss": 0.4636,
      "num_input_tokens_seen": 4612904,
      "step": 6990
    },
    {
      "epoch": 4.12441037735849,
      "grad_norm": 1.7672797441482544,
      "learning_rate": 9.660412227492107e-06,
      "loss": 0.4826,
      "num_input_tokens_seen": 4616072,
      "step": 6995
    },
    {
      "epoch": 4.127358490566038,
      "grad_norm": 2.905426502227783,
      "learning_rate": 9.659479654672106e-06,
      "loss": 0.5116,
      "num_input_tokens_seen": 4619144,
      "step": 7000
    },
    {
      "epoch": 4.130306603773585,
      "grad_norm": 1.5285683870315552,
      "learning_rate": 9.658545848234784e-06,
      "loss": 0.3966,
      "num_input_tokens_seen": 4621928,
      "step": 7005
    },
    {
      "epoch": 4.133254716981132,
      "grad_norm": 6.73098611831665,
      "learning_rate": 9.657610808427372e-06,
      "loss": 0.6536,
      "num_input_tokens_seen": 4625192,
      "step": 7010
    },
    {
      "epoch": 4.1362028301886795,
      "grad_norm": 0.9842777848243713,
      "learning_rate": 9.656674535497425e-06,
      "loss": 0.392,
      "num_input_tokens_seen": 4628488,
      "step": 7015
    },
    {
      "epoch": 4.139150943396227,
      "grad_norm": 2.8900506496429443,
      "learning_rate": 9.655737029692827e-06,
      "loss": 0.6573,
      "num_input_tokens_seen": 4631176,
      "step": 7020
    },
    {
      "epoch": 4.142099056603773,
      "grad_norm": 2.735351800918579,
      "learning_rate": 9.654798291261785e-06,
      "loss": 0.5352,
      "num_input_tokens_seen": 4634184,
      "step": 7025
    },
    {
      "epoch": 4.1450471698113205,
      "grad_norm": 2.742966890335083,
      "learning_rate": 9.653858320452833e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 4639944,
      "step": 7030
    },
    {
      "epoch": 4.147995283018868,
      "grad_norm": 2.418301582336426,
      "learning_rate": 9.652917117514836e-06,
      "loss": 0.4741,
      "num_input_tokens_seen": 4642920,
      "step": 7035
    },
    {
      "epoch": 4.150943396226415,
      "grad_norm": 1.4289723634719849,
      "learning_rate": 9.651974682696975e-06,
      "loss": 0.3218,
      "num_input_tokens_seen": 4646152,
      "step": 7040
    },
    {
      "epoch": 4.153891509433962,
      "grad_norm": 1.3688620328903198,
      "learning_rate": 9.651031016248773e-06,
      "loss": 0.3787,
      "num_input_tokens_seen": 4649288,
      "step": 7045
    },
    {
      "epoch": 4.15683962264151,
      "grad_norm": 2.7717173099517822,
      "learning_rate": 9.65008611842006e-06,
      "loss": 0.3528,
      "num_input_tokens_seen": 4652744,
      "step": 7050
    },
    {
      "epoch": 4.159787735849057,
      "grad_norm": 1.1341568231582642,
      "learning_rate": 9.64913998946101e-06,
      "loss": 0.4679,
      "num_input_tokens_seen": 4656200,
      "step": 7055
    },
    {
      "epoch": 4.162735849056604,
      "grad_norm": 2.4984230995178223,
      "learning_rate": 9.648192629622109e-06,
      "loss": 0.3994,
      "num_input_tokens_seen": 4658600,
      "step": 7060
    },
    {
      "epoch": 4.165683962264151,
      "grad_norm": 2.8273439407348633,
      "learning_rate": 9.647244039154178e-06,
      "loss": 0.6261,
      "num_input_tokens_seen": 4661352,
      "step": 7065
    },
    {
      "epoch": 4.168632075471698,
      "grad_norm": 1.3854660987854004,
      "learning_rate": 9.64629421830836e-06,
      "loss": 0.4016,
      "num_input_tokens_seen": 4664872,
      "step": 7070
    },
    {
      "epoch": 4.171580188679245,
      "grad_norm": 2.1194279193878174,
      "learning_rate": 9.64534316733612e-06,
      "loss": 0.3856,
      "num_input_tokens_seen": 4668008,
      "step": 7075
    },
    {
      "epoch": 4.1745283018867925,
      "grad_norm": 2.3048760890960693,
      "learning_rate": 9.644390886489258e-06,
      "loss": 0.4582,
      "num_input_tokens_seen": 4670504,
      "step": 7080
    },
    {
      "epoch": 4.17747641509434,
      "grad_norm": 2.0482678413391113,
      "learning_rate": 9.643437376019893e-06,
      "loss": 0.5076,
      "num_input_tokens_seen": 4674344,
      "step": 7085
    },
    {
      "epoch": 4.180424528301887,
      "grad_norm": 2.983699321746826,
      "learning_rate": 9.64248263618047e-06,
      "loss": 0.5465,
      "num_input_tokens_seen": 4676968,
      "step": 7090
    },
    {
      "epoch": 4.183372641509434,
      "grad_norm": 2.8827106952667236,
      "learning_rate": 9.64152666722376e-06,
      "loss": 0.355,
      "num_input_tokens_seen": 4679912,
      "step": 7095
    },
    {
      "epoch": 4.186320754716981,
      "grad_norm": 1.9402662515640259,
      "learning_rate": 9.640569469402863e-06,
      "loss": 0.4788,
      "num_input_tokens_seen": 4682856,
      "step": 7100
    },
    {
      "epoch": 4.189268867924528,
      "grad_norm": 1.410333514213562,
      "learning_rate": 9.639611042971198e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 4686216,
      "step": 7105
    },
    {
      "epoch": 4.192216981132075,
      "grad_norm": 1.7129716873168945,
      "learning_rate": 9.638651388182514e-06,
      "loss": 0.461,
      "num_input_tokens_seen": 4689192,
      "step": 7110
    },
    {
      "epoch": 4.195165094339623,
      "grad_norm": 1.7509336471557617,
      "learning_rate": 9.637690505290884e-06,
      "loss": 0.316,
      "num_input_tokens_seen": 4692552,
      "step": 7115
    },
    {
      "epoch": 4.19811320754717,
      "grad_norm": 1.1145939826965332,
      "learning_rate": 9.636728394550705e-06,
      "loss": 0.4007,
      "num_input_tokens_seen": 4695272,
      "step": 7120
    },
    {
      "epoch": 4.201061320754717,
      "grad_norm": 2.1704487800598145,
      "learning_rate": 9.6357650562167e-06,
      "loss": 0.498,
      "num_input_tokens_seen": 4698152,
      "step": 7125
    },
    {
      "epoch": 4.2040094339622645,
      "grad_norm": 2.428316831588745,
      "learning_rate": 9.634800490543918e-06,
      "loss": 0.6924,
      "num_input_tokens_seen": 4702280,
      "step": 7130
    },
    {
      "epoch": 4.206957547169812,
      "grad_norm": 2.2808945178985596,
      "learning_rate": 9.633834697787731e-06,
      "loss": 0.3773,
      "num_input_tokens_seen": 4705064,
      "step": 7135
    },
    {
      "epoch": 4.209905660377358,
      "grad_norm": 1.225632667541504,
      "learning_rate": 9.632867678203836e-06,
      "loss": 0.4358,
      "num_input_tokens_seen": 4707336,
      "step": 7140
    },
    {
      "epoch": 4.212853773584905,
      "grad_norm": 1.5542349815368652,
      "learning_rate": 9.631899432048258e-06,
      "loss": 0.4577,
      "num_input_tokens_seen": 4711176,
      "step": 7145
    },
    {
      "epoch": 4.215801886792453,
      "grad_norm": 1.7153784036636353,
      "learning_rate": 9.630929959577343e-06,
      "loss": 0.3796,
      "num_input_tokens_seen": 4714280,
      "step": 7150
    },
    {
      "epoch": 4.21875,
      "grad_norm": 1.7429203987121582,
      "learning_rate": 9.629959261047764e-06,
      "loss": 0.5988,
      "num_input_tokens_seen": 4717256,
      "step": 7155
    },
    {
      "epoch": 4.221698113207547,
      "grad_norm": 1.3366847038269043,
      "learning_rate": 9.628987336716513e-06,
      "loss": 0.5037,
      "num_input_tokens_seen": 4720520,
      "step": 7160
    },
    {
      "epoch": 4.224646226415095,
      "grad_norm": 1.1860090494155884,
      "learning_rate": 9.628014186840918e-06,
      "loss": 0.3138,
      "num_input_tokens_seen": 4726760,
      "step": 7165
    },
    {
      "epoch": 4.227594339622642,
      "grad_norm": 1.7789610624313354,
      "learning_rate": 9.62703981167862e-06,
      "loss": 0.4806,
      "num_input_tokens_seen": 4730632,
      "step": 7170
    },
    {
      "epoch": 4.230542452830188,
      "grad_norm": 2.8640425205230713,
      "learning_rate": 9.626064211487592e-06,
      "loss": 0.5059,
      "num_input_tokens_seen": 4733224,
      "step": 7175
    },
    {
      "epoch": 4.2334905660377355,
      "grad_norm": 1.5356751680374146,
      "learning_rate": 9.625087386526125e-06,
      "loss": 0.4392,
      "num_input_tokens_seen": 4736904,
      "step": 7180
    },
    {
      "epoch": 4.236438679245283,
      "grad_norm": 1.5610798597335815,
      "learning_rate": 9.624109337052839e-06,
      "loss": 0.4395,
      "num_input_tokens_seen": 4739336,
      "step": 7185
    },
    {
      "epoch": 4.23938679245283,
      "grad_norm": 2.061108112335205,
      "learning_rate": 9.623130063326678e-06,
      "loss": 0.5551,
      "num_input_tokens_seen": 4742376,
      "step": 7190
    },
    {
      "epoch": 4.242334905660377,
      "grad_norm": 1.3884388208389282,
      "learning_rate": 9.622149565606909e-06,
      "loss": 0.3712,
      "num_input_tokens_seen": 4745800,
      "step": 7195
    },
    {
      "epoch": 4.245283018867925,
      "grad_norm": 1.3294041156768799,
      "learning_rate": 9.621167844153122e-06,
      "loss": 0.4374,
      "num_input_tokens_seen": 4749224,
      "step": 7200
    },
    {
      "epoch": 4.248231132075472,
      "grad_norm": 3.0809078216552734,
      "learning_rate": 9.620184899225231e-06,
      "loss": 0.4197,
      "num_input_tokens_seen": 4751688,
      "step": 7205
    },
    {
      "epoch": 4.251179245283019,
      "grad_norm": 2.629441261291504,
      "learning_rate": 9.619200731083477e-06,
      "loss": 0.4389,
      "num_input_tokens_seen": 4755176,
      "step": 7210
    },
    {
      "epoch": 4.254127358490566,
      "grad_norm": 2.2905685901641846,
      "learning_rate": 9.618215339988422e-06,
      "loss": 0.4302,
      "num_input_tokens_seen": 4759528,
      "step": 7215
    },
    {
      "epoch": 4.257075471698113,
      "grad_norm": 2.0900614261627197,
      "learning_rate": 9.617228726200951e-06,
      "loss": 0.3862,
      "num_input_tokens_seen": 4762312,
      "step": 7220
    },
    {
      "epoch": 4.26002358490566,
      "grad_norm": 1.0832273960113525,
      "learning_rate": 9.616240889982277e-06,
      "loss": 0.3715,
      "num_input_tokens_seen": 4765512,
      "step": 7225
    },
    {
      "epoch": 4.2629716981132075,
      "grad_norm": 1.2484678030014038,
      "learning_rate": 9.61525183159393e-06,
      "loss": 0.428,
      "num_input_tokens_seen": 4767880,
      "step": 7230
    },
    {
      "epoch": 4.265919811320755,
      "grad_norm": 1.9305827617645264,
      "learning_rate": 9.614261551297774e-06,
      "loss": 0.5832,
      "num_input_tokens_seen": 4771752,
      "step": 7235
    },
    {
      "epoch": 4.268867924528302,
      "grad_norm": 5.226742267608643,
      "learning_rate": 9.613270049355983e-06,
      "loss": 0.5245,
      "num_input_tokens_seen": 4775464,
      "step": 7240
    },
    {
      "epoch": 4.271816037735849,
      "grad_norm": 2.46671199798584,
      "learning_rate": 9.612277326031065e-06,
      "loss": 0.4159,
      "num_input_tokens_seen": 4778568,
      "step": 7245
    },
    {
      "epoch": 4.274764150943396,
      "grad_norm": 1.8648021221160889,
      "learning_rate": 9.611283381585848e-06,
      "loss": 0.5231,
      "num_input_tokens_seen": 4781544,
      "step": 7250
    },
    {
      "epoch": 4.277712264150943,
      "grad_norm": 1.6778558492660522,
      "learning_rate": 9.61028821628348e-06,
      "loss": 0.432,
      "num_input_tokens_seen": 4784456,
      "step": 7255
    },
    {
      "epoch": 4.28066037735849,
      "grad_norm": 2.40561580657959,
      "learning_rate": 9.609291830387439e-06,
      "loss": 0.4013,
      "num_input_tokens_seen": 4786984,
      "step": 7260
    },
    {
      "epoch": 4.283608490566038,
      "grad_norm": 4.078433513641357,
      "learning_rate": 9.608294224161523e-06,
      "loss": 0.3934,
      "num_input_tokens_seen": 4790312,
      "step": 7265
    },
    {
      "epoch": 4.286556603773585,
      "grad_norm": 1.5456923246383667,
      "learning_rate": 9.607295397869847e-06,
      "loss": 0.4572,
      "num_input_tokens_seen": 4793224,
      "step": 7270
    },
    {
      "epoch": 4.289504716981132,
      "grad_norm": 1.7030737400054932,
      "learning_rate": 9.60629535177686e-06,
      "loss": 0.3633,
      "num_input_tokens_seen": 4795752,
      "step": 7275
    },
    {
      "epoch": 4.2924528301886795,
      "grad_norm": 1.3352009057998657,
      "learning_rate": 9.605294086147325e-06,
      "loss": 0.5116,
      "num_input_tokens_seen": 4798696,
      "step": 7280
    },
    {
      "epoch": 4.295400943396227,
      "grad_norm": 1.7045186758041382,
      "learning_rate": 9.604291601246333e-06,
      "loss": 0.4853,
      "num_input_tokens_seen": 4801768,
      "step": 7285
    },
    {
      "epoch": 4.298349056603773,
      "grad_norm": 3.1141064167022705,
      "learning_rate": 9.603287897339299e-06,
      "loss": 0.46,
      "num_input_tokens_seen": 4805768,
      "step": 7290
    },
    {
      "epoch": 4.3012971698113205,
      "grad_norm": 1.3982442617416382,
      "learning_rate": 9.602282974691953e-06,
      "loss": 0.5398,
      "num_input_tokens_seen": 4808552,
      "step": 7295
    },
    {
      "epoch": 4.304245283018868,
      "grad_norm": 1.3132338523864746,
      "learning_rate": 9.601276833570355e-06,
      "loss": 0.6175,
      "num_input_tokens_seen": 4811432,
      "step": 7300
    },
    {
      "epoch": 4.307193396226415,
      "grad_norm": 0.9858609437942505,
      "learning_rate": 9.600269474240885e-06,
      "loss": 0.4417,
      "num_input_tokens_seen": 4815464,
      "step": 7305
    },
    {
      "epoch": 4.310141509433962,
      "grad_norm": 2.790390729904175,
      "learning_rate": 9.599260896970246e-06,
      "loss": 0.5406,
      "num_input_tokens_seen": 4818344,
      "step": 7310
    },
    {
      "epoch": 4.31308962264151,
      "grad_norm": 1.2277859449386597,
      "learning_rate": 9.598251102025463e-06,
      "loss": 0.5049,
      "num_input_tokens_seen": 4822280,
      "step": 7315
    },
    {
      "epoch": 4.316037735849057,
      "grad_norm": 2.562476634979248,
      "learning_rate": 9.597240089673882e-06,
      "loss": 0.4784,
      "num_input_tokens_seen": 4825192,
      "step": 7320
    },
    {
      "epoch": 4.318985849056604,
      "grad_norm": 1.5467262268066406,
      "learning_rate": 9.596227860183175e-06,
      "loss": 0.4279,
      "num_input_tokens_seen": 4829128,
      "step": 7325
    },
    {
      "epoch": 4.321933962264151,
      "grad_norm": 1.0308811664581299,
      "learning_rate": 9.595214413821334e-06,
      "loss": 0.3524,
      "num_input_tokens_seen": 4832104,
      "step": 7330
    },
    {
      "epoch": 4.324882075471698,
      "grad_norm": 1.867716670036316,
      "learning_rate": 9.59419975085667e-06,
      "loss": 0.442,
      "num_input_tokens_seen": 4835016,
      "step": 7335
    },
    {
      "epoch": 4.327830188679245,
      "grad_norm": 2.123149871826172,
      "learning_rate": 9.593183871557826e-06,
      "loss": 0.3936,
      "num_input_tokens_seen": 4838216,
      "step": 7340
    },
    {
      "epoch": 4.3307783018867925,
      "grad_norm": 2.5819644927978516,
      "learning_rate": 9.592166776193754e-06,
      "loss": 0.4035,
      "num_input_tokens_seen": 4841096,
      "step": 7345
    },
    {
      "epoch": 4.33372641509434,
      "grad_norm": 1.7340352535247803,
      "learning_rate": 9.591148465033738e-06,
      "loss": 0.5548,
      "num_input_tokens_seen": 4845032,
      "step": 7350
    },
    {
      "epoch": 4.336674528301887,
      "grad_norm": 1.6399118900299072,
      "learning_rate": 9.590128938347378e-06,
      "loss": 0.4652,
      "num_input_tokens_seen": 4847784,
      "step": 7355
    },
    {
      "epoch": 4.339622641509434,
      "grad_norm": 1.9362975358963013,
      "learning_rate": 9.589108196404599e-06,
      "loss": 0.4244,
      "num_input_tokens_seen": 4850952,
      "step": 7360
    },
    {
      "epoch": 4.342570754716981,
      "grad_norm": 2.6924099922180176,
      "learning_rate": 9.588086239475649e-06,
      "loss": 0.4343,
      "num_input_tokens_seen": 4853800,
      "step": 7365
    },
    {
      "epoch": 4.345518867924528,
      "grad_norm": 13.95986557006836,
      "learning_rate": 9.587063067831092e-06,
      "loss": 0.6125,
      "num_input_tokens_seen": 4857000,
      "step": 7370
    },
    {
      "epoch": 4.348466981132075,
      "grad_norm": 2.478508949279785,
      "learning_rate": 9.586038681741818e-06,
      "loss": 0.4878,
      "num_input_tokens_seen": 4860520,
      "step": 7375
    },
    {
      "epoch": 4.351415094339623,
      "grad_norm": 2.3113229274749756,
      "learning_rate": 9.58501308147904e-06,
      "loss": 0.2957,
      "num_input_tokens_seen": 4863432,
      "step": 7380
    },
    {
      "epoch": 4.35436320754717,
      "grad_norm": 1.453588604927063,
      "learning_rate": 9.583986267314288e-06,
      "loss": 0.489,
      "num_input_tokens_seen": 4867112,
      "step": 7385
    },
    {
      "epoch": 4.357311320754717,
      "grad_norm": 2.3686020374298096,
      "learning_rate": 9.582958239519416e-06,
      "loss": 0.4858,
      "num_input_tokens_seen": 4870536,
      "step": 7390
    },
    {
      "epoch": 4.3602594339622645,
      "grad_norm": 4.589690208435059,
      "learning_rate": 9.581928998366597e-06,
      "loss": 0.4414,
      "num_input_tokens_seen": 4872840,
      "step": 7395
    },
    {
      "epoch": 4.363207547169811,
      "grad_norm": 1.829455852508545,
      "learning_rate": 9.58089854412833e-06,
      "loss": 0.625,
      "num_input_tokens_seen": 4875912,
      "step": 7400
    },
    {
      "epoch": 4.366155660377358,
      "grad_norm": 3.0968165397644043,
      "learning_rate": 9.579866877077431e-06,
      "loss": 0.4767,
      "num_input_tokens_seen": 4879816,
      "step": 7405
    },
    {
      "epoch": 4.369103773584905,
      "grad_norm": 2.5272092819213867,
      "learning_rate": 9.578833997487038e-06,
      "loss": 0.6627,
      "num_input_tokens_seen": 4883240,
      "step": 7410
    },
    {
      "epoch": 4.372051886792453,
      "grad_norm": 1.8744592666625977,
      "learning_rate": 9.57779990563061e-06,
      "loss": 0.595,
      "num_input_tokens_seen": 4887560,
      "step": 7415
    },
    {
      "epoch": 4.375,
      "grad_norm": 1.793652057647705,
      "learning_rate": 9.576764601781928e-06,
      "loss": 0.5417,
      "num_input_tokens_seen": 4891560,
      "step": 7420
    },
    {
      "epoch": 4.377948113207547,
      "grad_norm": 1.4144402742385864,
      "learning_rate": 9.575728086215093e-06,
      "loss": 0.5166,
      "num_input_tokens_seen": 4895016,
      "step": 7425
    },
    {
      "epoch": 4.380896226415095,
      "grad_norm": 2.0562267303466797,
      "learning_rate": 9.574690359204527e-06,
      "loss": 0.5226,
      "num_input_tokens_seen": 4898376,
      "step": 7430
    },
    {
      "epoch": 4.383844339622642,
      "grad_norm": 1.309874176979065,
      "learning_rate": 9.573651421024972e-06,
      "loss": 0.4887,
      "num_input_tokens_seen": 4901352,
      "step": 7435
    },
    {
      "epoch": 4.386792452830189,
      "grad_norm": 2.644780158996582,
      "learning_rate": 9.572611271951494e-06,
      "loss": 0.4941,
      "num_input_tokens_seen": 4904616,
      "step": 7440
    },
    {
      "epoch": 4.3897405660377355,
      "grad_norm": 2.1164112091064453,
      "learning_rate": 9.571569912259473e-06,
      "loss": 0.3854,
      "num_input_tokens_seen": 4907208,
      "step": 7445
    },
    {
      "epoch": 4.392688679245283,
      "grad_norm": 2.1132497787475586,
      "learning_rate": 9.570527342224614e-06,
      "loss": 0.5374,
      "num_input_tokens_seen": 4910536,
      "step": 7450
    },
    {
      "epoch": 4.39563679245283,
      "grad_norm": 4.449840545654297,
      "learning_rate": 9.569483562122945e-06,
      "loss": 0.7987,
      "num_input_tokens_seen": 4914536,
      "step": 7455
    },
    {
      "epoch": 4.398584905660377,
      "grad_norm": 1.6682238578796387,
      "learning_rate": 9.568438572230811e-06,
      "loss": 0.4393,
      "num_input_tokens_seen": 4918280,
      "step": 7460
    },
    {
      "epoch": 4.401533018867925,
      "grad_norm": 1.7804105281829834,
      "learning_rate": 9.567392372824873e-06,
      "loss": 0.5315,
      "num_input_tokens_seen": 4921384,
      "step": 7465
    },
    {
      "epoch": 4.404481132075472,
      "grad_norm": 1.4560558795928955,
      "learning_rate": 9.566344964182123e-06,
      "loss": 0.492,
      "num_input_tokens_seen": 4924040,
      "step": 7470
    },
    {
      "epoch": 4.407429245283019,
      "grad_norm": 5.318580150604248,
      "learning_rate": 9.565296346579862e-06,
      "loss": 0.5681,
      "num_input_tokens_seen": 4926856,
      "step": 7475
    },
    {
      "epoch": 4.410377358490566,
      "grad_norm": 1.8727731704711914,
      "learning_rate": 9.564246520295719e-06,
      "loss": 0.4125,
      "num_input_tokens_seen": 4929928,
      "step": 7480
    },
    {
      "epoch": 4.413325471698113,
      "grad_norm": 1.6377122402191162,
      "learning_rate": 9.563195485607638e-06,
      "loss": 0.4978,
      "num_input_tokens_seen": 4932776,
      "step": 7485
    },
    {
      "epoch": 4.41627358490566,
      "grad_norm": 3.8691999912261963,
      "learning_rate": 9.562143242793885e-06,
      "loss": 0.4401,
      "num_input_tokens_seen": 4935432,
      "step": 7490
    },
    {
      "epoch": 4.4192216981132075,
      "grad_norm": 2.8287158012390137,
      "learning_rate": 9.561089792133048e-06,
      "loss": 0.3923,
      "num_input_tokens_seen": 4939240,
      "step": 7495
    },
    {
      "epoch": 4.422169811320755,
      "grad_norm": 2.3411812782287598,
      "learning_rate": 9.560035133904031e-06,
      "loss": 0.4536,
      "num_input_tokens_seen": 4941960,
      "step": 7500
    },
    {
      "epoch": 4.425117924528302,
      "grad_norm": 1.7029826641082764,
      "learning_rate": 9.55897926838606e-06,
      "loss": 0.5492,
      "num_input_tokens_seen": 4945192,
      "step": 7505
    },
    {
      "epoch": 4.428066037735849,
      "grad_norm": 1.7048293352127075,
      "learning_rate": 9.55792219585868e-06,
      "loss": 0.3717,
      "num_input_tokens_seen": 4949192,
      "step": 7510
    },
    {
      "epoch": 4.431014150943396,
      "grad_norm": 1.7667900323867798,
      "learning_rate": 9.556863916601754e-06,
      "loss": 0.3789,
      "num_input_tokens_seen": 4953288,
      "step": 7515
    },
    {
      "epoch": 4.433962264150943,
      "grad_norm": 2.440966844558716,
      "learning_rate": 9.555804430895467e-06,
      "loss": 0.6161,
      "num_input_tokens_seen": 4956136,
      "step": 7520
    },
    {
      "epoch": 4.43691037735849,
      "grad_norm": 1.2371697425842285,
      "learning_rate": 9.554743739020325e-06,
      "loss": 0.4528,
      "num_input_tokens_seen": 4960008,
      "step": 7525
    },
    {
      "epoch": 4.439858490566038,
      "grad_norm": 1.6167486906051636,
      "learning_rate": 9.553681841257146e-06,
      "loss": 0.3395,
      "num_input_tokens_seen": 4962984,
      "step": 7530
    },
    {
      "epoch": 4.442806603773585,
      "grad_norm": 1.7759188413619995,
      "learning_rate": 9.552618737887073e-06,
      "loss": 0.5006,
      "num_input_tokens_seen": 4966216,
      "step": 7535
    },
    {
      "epoch": 4.445754716981132,
      "grad_norm": 3.332770824432373,
      "learning_rate": 9.55155442919157e-06,
      "loss": 0.5901,
      "num_input_tokens_seen": 4968712,
      "step": 7540
    },
    {
      "epoch": 4.4487028301886795,
      "grad_norm": 1.3297756910324097,
      "learning_rate": 9.550488915452416e-06,
      "loss": 0.349,
      "num_input_tokens_seen": 4971560,
      "step": 7545
    },
    {
      "epoch": 4.451650943396227,
      "grad_norm": 1.4916118383407593,
      "learning_rate": 9.54942219695171e-06,
      "loss": 0.5026,
      "num_input_tokens_seen": 4974568,
      "step": 7550
    },
    {
      "epoch": 4.454599056603773,
      "grad_norm": 1.3539879322052002,
      "learning_rate": 9.54835427397187e-06,
      "loss": 0.4905,
      "num_input_tokens_seen": 4977320,
      "step": 7555
    },
    {
      "epoch": 4.4575471698113205,
      "grad_norm": 2.441842794418335,
      "learning_rate": 9.547285146795634e-06,
      "loss": 0.453,
      "num_input_tokens_seen": 4979720,
      "step": 7560
    },
    {
      "epoch": 4.460495283018868,
      "grad_norm": 4.029641151428223,
      "learning_rate": 9.546214815706059e-06,
      "loss": 0.4132,
      "num_input_tokens_seen": 4982824,
      "step": 7565
    },
    {
      "epoch": 4.463443396226415,
      "grad_norm": 2.6732938289642334,
      "learning_rate": 9.545143280986518e-06,
      "loss": 0.5406,
      "num_input_tokens_seen": 4985928,
      "step": 7570
    },
    {
      "epoch": 4.466391509433962,
      "grad_norm": 1.574789047241211,
      "learning_rate": 9.544070542920703e-06,
      "loss": 0.4692,
      "num_input_tokens_seen": 4989160,
      "step": 7575
    },
    {
      "epoch": 4.46933962264151,
      "grad_norm": 3.0948591232299805,
      "learning_rate": 9.542996601792629e-06,
      "loss": 0.3725,
      "num_input_tokens_seen": 4991496,
      "step": 7580
    },
    {
      "epoch": 4.472287735849057,
      "grad_norm": 1.047343373298645,
      "learning_rate": 9.541921457886624e-06,
      "loss": 0.441,
      "num_input_tokens_seen": 4994280,
      "step": 7585
    },
    {
      "epoch": 4.475235849056604,
      "grad_norm": 1.2321743965148926,
      "learning_rate": 9.54084511148734e-06,
      "loss": 0.4247,
      "num_input_tokens_seen": 4997832,
      "step": 7590
    },
    {
      "epoch": 4.478183962264151,
      "grad_norm": 1.8912714719772339,
      "learning_rate": 9.539767562879742e-06,
      "loss": 0.5385,
      "num_input_tokens_seen": 5000424,
      "step": 7595
    },
    {
      "epoch": 4.481132075471698,
      "grad_norm": 1.6969109773635864,
      "learning_rate": 9.538688812349118e-06,
      "loss": 0.3145,
      "num_input_tokens_seen": 5003048,
      "step": 7600
    },
    {
      "epoch": 4.484080188679245,
      "grad_norm": 2.1449785232543945,
      "learning_rate": 9.537608860181069e-06,
      "loss": 0.4413,
      "num_input_tokens_seen": 5005832,
      "step": 7605
    },
    {
      "epoch": 4.4870283018867925,
      "grad_norm": 2.361207962036133,
      "learning_rate": 9.536527706661519e-06,
      "loss": 0.5411,
      "num_input_tokens_seen": 5008552,
      "step": 7610
    },
    {
      "epoch": 4.48997641509434,
      "grad_norm": 2.0780117511749268,
      "learning_rate": 9.535445352076707e-06,
      "loss": 0.5087,
      "num_input_tokens_seen": 5011016,
      "step": 7615
    },
    {
      "epoch": 4.492924528301887,
      "grad_norm": 1.5228490829467773,
      "learning_rate": 9.534361796713191e-06,
      "loss": 0.3792,
      "num_input_tokens_seen": 5013736,
      "step": 7620
    },
    {
      "epoch": 4.495872641509434,
      "grad_norm": 3.7025198936462402,
      "learning_rate": 9.533277040857847e-06,
      "loss": 0.5086,
      "num_input_tokens_seen": 5016968,
      "step": 7625
    },
    {
      "epoch": 4.498820754716981,
      "grad_norm": 1.3475953340530396,
      "learning_rate": 9.53219108479787e-06,
      "loss": 0.6187,
      "num_input_tokens_seen": 5019464,
      "step": 7630
    },
    {
      "epoch": 4.501768867924528,
      "grad_norm": 1.2320364713668823,
      "learning_rate": 9.53110392882077e-06,
      "loss": 0.4901,
      "num_input_tokens_seen": 5022376,
      "step": 7635
    },
    {
      "epoch": 4.504716981132075,
      "grad_norm": 1.8912127017974854,
      "learning_rate": 9.530015573214378e-06,
      "loss": 0.6376,
      "num_input_tokens_seen": 5026440,
      "step": 7640
    },
    {
      "epoch": 4.507665094339623,
      "grad_norm": 2.6435546875,
      "learning_rate": 9.528926018266837e-06,
      "loss": 0.5844,
      "num_input_tokens_seen": 5029896,
      "step": 7645
    },
    {
      "epoch": 4.51061320754717,
      "grad_norm": 3.006228446960449,
      "learning_rate": 9.527835264266617e-06,
      "loss": 0.4099,
      "num_input_tokens_seen": 5032808,
      "step": 7650
    },
    {
      "epoch": 4.513561320754717,
      "grad_norm": 1.984240174293518,
      "learning_rate": 9.526743311502496e-06,
      "loss": 0.4673,
      "num_input_tokens_seen": 5035784,
      "step": 7655
    },
    {
      "epoch": 4.5165094339622645,
      "grad_norm": 6.773003101348877,
      "learning_rate": 9.525650160263573e-06,
      "loss": 0.523,
      "num_input_tokens_seen": 5039240,
      "step": 7660
    },
    {
      "epoch": 4.519457547169811,
      "grad_norm": 4.032406806945801,
      "learning_rate": 9.524555810839267e-06,
      "loss": 0.4746,
      "num_input_tokens_seen": 5042664,
      "step": 7665
    },
    {
      "epoch": 4.522405660377358,
      "grad_norm": 1.7200754880905151,
      "learning_rate": 9.523460263519309e-06,
      "loss": 0.4488,
      "num_input_tokens_seen": 5045672,
      "step": 7670
    },
    {
      "epoch": 4.525353773584905,
      "grad_norm": 1.5583415031433105,
      "learning_rate": 9.522363518593753e-06,
      "loss": 0.5539,
      "num_input_tokens_seen": 5049160,
      "step": 7675
    },
    {
      "epoch": 4.528301886792453,
      "grad_norm": 1.6208946704864502,
      "learning_rate": 9.521265576352963e-06,
      "loss": 0.4399,
      "num_input_tokens_seen": 5052040,
      "step": 7680
    },
    {
      "epoch": 4.53125,
      "grad_norm": 1.607393741607666,
      "learning_rate": 9.520166437087628e-06,
      "loss": 0.4534,
      "num_input_tokens_seen": 5055272,
      "step": 7685
    },
    {
      "epoch": 4.534198113207547,
      "grad_norm": 2.237539291381836,
      "learning_rate": 9.519066101088748e-06,
      "loss": 0.3648,
      "num_input_tokens_seen": 5058248,
      "step": 7690
    },
    {
      "epoch": 4.537146226415095,
      "grad_norm": 2.3736941814422607,
      "learning_rate": 9.51796456864764e-06,
      "loss": 0.35,
      "num_input_tokens_seen": 5061192,
      "step": 7695
    },
    {
      "epoch": 4.540094339622642,
      "grad_norm": 1.0082405805587769,
      "learning_rate": 9.516861840055942e-06,
      "loss": 0.3693,
      "num_input_tokens_seen": 5064392,
      "step": 7700
    },
    {
      "epoch": 4.543042452830189,
      "grad_norm": 2.147836923599243,
      "learning_rate": 9.515757915605604e-06,
      "loss": 0.4428,
      "num_input_tokens_seen": 5067720,
      "step": 7705
    },
    {
      "epoch": 4.5459905660377355,
      "grad_norm": 2.818782329559326,
      "learning_rate": 9.514652795588899e-06,
      "loss": 0.4683,
      "num_input_tokens_seen": 5070696,
      "step": 7710
    },
    {
      "epoch": 4.548938679245283,
      "grad_norm": 1.2179821729660034,
      "learning_rate": 9.513546480298405e-06,
      "loss": 0.3422,
      "num_input_tokens_seen": 5074888,
      "step": 7715
    },
    {
      "epoch": 4.55188679245283,
      "grad_norm": 2.835498332977295,
      "learning_rate": 9.512438970027032e-06,
      "loss": 0.5659,
      "num_input_tokens_seen": 5077800,
      "step": 7720
    },
    {
      "epoch": 4.554834905660377,
      "grad_norm": 1.4974870681762695,
      "learning_rate": 9.511330265067992e-06,
      "loss": 0.4758,
      "num_input_tokens_seen": 5081608,
      "step": 7725
    },
    {
      "epoch": 4.557783018867925,
      "grad_norm": 2.3473994731903076,
      "learning_rate": 9.510220365714822e-06,
      "loss": 0.617,
      "num_input_tokens_seen": 5086312,
      "step": 7730
    },
    {
      "epoch": 4.560731132075472,
      "grad_norm": 2.641878843307495,
      "learning_rate": 9.509109272261373e-06,
      "loss": 0.5285,
      "num_input_tokens_seen": 5090024,
      "step": 7735
    },
    {
      "epoch": 4.563679245283019,
      "grad_norm": 2.608853816986084,
      "learning_rate": 9.50799698500181e-06,
      "loss": 0.4554,
      "num_input_tokens_seen": 5093288,
      "step": 7740
    },
    {
      "epoch": 4.566627358490566,
      "grad_norm": 2.565425395965576,
      "learning_rate": 9.506883504230618e-06,
      "loss": 0.5056,
      "num_input_tokens_seen": 5096744,
      "step": 7745
    },
    {
      "epoch": 4.569575471698113,
      "grad_norm": 1.8440033197402954,
      "learning_rate": 9.505768830242593e-06,
      "loss": 0.4201,
      "num_input_tokens_seen": 5100008,
      "step": 7750
    },
    {
      "epoch": 4.57252358490566,
      "grad_norm": 2.572190761566162,
      "learning_rate": 9.504652963332852e-06,
      "loss": 0.5464,
      "num_input_tokens_seen": 5103592,
      "step": 7755
    },
    {
      "epoch": 4.5754716981132075,
      "grad_norm": 1.485986590385437,
      "learning_rate": 9.503535903796825e-06,
      "loss": 0.4714,
      "num_input_tokens_seen": 5105832,
      "step": 7760
    },
    {
      "epoch": 4.578419811320755,
      "grad_norm": 1.8105634450912476,
      "learning_rate": 9.50241765193026e-06,
      "loss": 0.3874,
      "num_input_tokens_seen": 5110280,
      "step": 7765
    },
    {
      "epoch": 4.581367924528302,
      "grad_norm": 2.5712666511535645,
      "learning_rate": 9.501298208029214e-06,
      "loss": 0.6047,
      "num_input_tokens_seen": 5113224,
      "step": 7770
    },
    {
      "epoch": 4.584316037735849,
      "grad_norm": 2.2851805686950684,
      "learning_rate": 9.500177572390071e-06,
      "loss": 0.585,
      "num_input_tokens_seen": 5116456,
      "step": 7775
    },
    {
      "epoch": 4.587264150943396,
      "grad_norm": 4.001415729522705,
      "learning_rate": 9.49905574530952e-06,
      "loss": 0.5695,
      "num_input_tokens_seen": 5120776,
      "step": 7780
    },
    {
      "epoch": 4.590212264150943,
      "grad_norm": 1.2369303703308105,
      "learning_rate": 9.497932727084571e-06,
      "loss": 0.5144,
      "num_input_tokens_seen": 5124360,
      "step": 7785
    },
    {
      "epoch": 4.59316037735849,
      "grad_norm": 1.3818989992141724,
      "learning_rate": 9.496808518012545e-06,
      "loss": 0.4494,
      "num_input_tokens_seen": 5127816,
      "step": 7790
    },
    {
      "epoch": 4.596108490566038,
      "grad_norm": 2.795192003250122,
      "learning_rate": 9.495683118391087e-06,
      "loss": 0.538,
      "num_input_tokens_seen": 5130856,
      "step": 7795
    },
    {
      "epoch": 4.599056603773585,
      "grad_norm": 2.72634220123291,
      "learning_rate": 9.494556528518146e-06,
      "loss": 0.4488,
      "num_input_tokens_seen": 5133768,
      "step": 7800
    },
    {
      "epoch": 4.602004716981132,
      "grad_norm": 1.8947908878326416,
      "learning_rate": 9.493428748691995e-06,
      "loss": 0.5698,
      "num_input_tokens_seen": 5140584,
      "step": 7805
    },
    {
      "epoch": 4.6049528301886795,
      "grad_norm": 1.4524437189102173,
      "learning_rate": 9.492299779211215e-06,
      "loss": 0.307,
      "num_input_tokens_seen": 5144616,
      "step": 7810
    },
    {
      "epoch": 4.607900943396227,
      "grad_norm": 1.3324542045593262,
      "learning_rate": 9.49116962037471e-06,
      "loss": 0.45,
      "num_input_tokens_seen": 5148392,
      "step": 7815
    },
    {
      "epoch": 4.610849056603773,
      "grad_norm": 1.6141685247421265,
      "learning_rate": 9.490038272481691e-06,
      "loss": 0.4428,
      "num_input_tokens_seen": 5150952,
      "step": 7820
    },
    {
      "epoch": 4.6137971698113205,
      "grad_norm": 1.8393303155899048,
      "learning_rate": 9.488905735831689e-06,
      "loss": 0.4114,
      "num_input_tokens_seen": 5153928,
      "step": 7825
    },
    {
      "epoch": 4.616745283018868,
      "grad_norm": 1.5755314826965332,
      "learning_rate": 9.487772010724548e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 5157192,
      "step": 7830
    },
    {
      "epoch": 4.619693396226415,
      "grad_norm": 2.386803150177002,
      "learning_rate": 9.486637097460425e-06,
      "loss": 0.428,
      "num_input_tokens_seen": 5159720,
      "step": 7835
    },
    {
      "epoch": 4.622641509433962,
      "grad_norm": 2.1059467792510986,
      "learning_rate": 9.485500996339793e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 5163272,
      "step": 7840
    },
    {
      "epoch": 4.62558962264151,
      "grad_norm": 1.548727035522461,
      "learning_rate": 9.484363707663443e-06,
      "loss": 0.4253,
      "num_input_tokens_seen": 5166568,
      "step": 7845
    },
    {
      "epoch": 4.628537735849057,
      "grad_norm": 1.5566030740737915,
      "learning_rate": 9.483225231732474e-06,
      "loss": 0.3894,
      "num_input_tokens_seen": 5172104,
      "step": 7850
    },
    {
      "epoch": 4.631485849056604,
      "grad_norm": 3.064798593521118,
      "learning_rate": 9.482085568848302e-06,
      "loss": 0.4598,
      "num_input_tokens_seen": 5174376,
      "step": 7855
    },
    {
      "epoch": 4.634433962264151,
      "grad_norm": 1.6378132104873657,
      "learning_rate": 9.480944719312659e-06,
      "loss": 0.6963,
      "num_input_tokens_seen": 5177544,
      "step": 7860
    },
    {
      "epoch": 4.637382075471698,
      "grad_norm": 1.6780871152877808,
      "learning_rate": 9.47980268342759e-06,
      "loss": 0.4846,
      "num_input_tokens_seen": 5180680,
      "step": 7865
    },
    {
      "epoch": 4.640330188679245,
      "grad_norm": 1.864379644393921,
      "learning_rate": 9.478659461495456e-06,
      "loss": 0.3767,
      "num_input_tokens_seen": 5183336,
      "step": 7870
    },
    {
      "epoch": 4.6432783018867925,
      "grad_norm": 1.987644910812378,
      "learning_rate": 9.477515053818926e-06,
      "loss": 0.464,
      "num_input_tokens_seen": 5186568,
      "step": 7875
    },
    {
      "epoch": 4.64622641509434,
      "grad_norm": 2.5901570320129395,
      "learning_rate": 9.476369460700988e-06,
      "loss": 0.5222,
      "num_input_tokens_seen": 5189576,
      "step": 7880
    },
    {
      "epoch": 4.649174528301887,
      "grad_norm": 3.9750468730926514,
      "learning_rate": 9.475222682444944e-06,
      "loss": 0.5426,
      "num_input_tokens_seen": 5192360,
      "step": 7885
    },
    {
      "epoch": 4.652122641509434,
      "grad_norm": 1.163432240486145,
      "learning_rate": 9.474074719354406e-06,
      "loss": 0.4254,
      "num_input_tokens_seen": 5195048,
      "step": 7890
    },
    {
      "epoch": 4.655070754716981,
      "grad_norm": 1.509907841682434,
      "learning_rate": 9.472925571733306e-06,
      "loss": 0.4954,
      "num_input_tokens_seen": 5198568,
      "step": 7895
    },
    {
      "epoch": 4.658018867924528,
      "grad_norm": 1.3482563495635986,
      "learning_rate": 9.471775239885883e-06,
      "loss": 0.492,
      "num_input_tokens_seen": 5203112,
      "step": 7900
    },
    {
      "epoch": 4.660966981132075,
      "grad_norm": 1.8188621997833252,
      "learning_rate": 9.470623724116693e-06,
      "loss": 0.4329,
      "num_input_tokens_seen": 5206920,
      "step": 7905
    },
    {
      "epoch": 4.663915094339623,
      "grad_norm": 2.1917402744293213,
      "learning_rate": 9.469471024730606e-06,
      "loss": 0.437,
      "num_input_tokens_seen": 5210056,
      "step": 7910
    },
    {
      "epoch": 4.66686320754717,
      "grad_norm": 1.6490241289138794,
      "learning_rate": 9.4683171420328e-06,
      "loss": 0.5536,
      "num_input_tokens_seen": 5213256,
      "step": 7915
    },
    {
      "epoch": 4.669811320754717,
      "grad_norm": 2.6375937461853027,
      "learning_rate": 9.467162076328776e-06,
      "loss": 0.393,
      "num_input_tokens_seen": 5216168,
      "step": 7920
    },
    {
      "epoch": 4.6727594339622645,
      "grad_norm": 3.0387628078460693,
      "learning_rate": 9.466005827924337e-06,
      "loss": 0.4699,
      "num_input_tokens_seen": 5219368,
      "step": 7925
    },
    {
      "epoch": 4.675707547169811,
      "grad_norm": 2.4526593685150146,
      "learning_rate": 9.46484839712561e-06,
      "loss": 0.3703,
      "num_input_tokens_seen": 5222344,
      "step": 7930
    },
    {
      "epoch": 4.678655660377358,
      "grad_norm": 1.4520732164382935,
      "learning_rate": 9.463689784239026e-06,
      "loss": 0.5319,
      "num_input_tokens_seen": 5225576,
      "step": 7935
    },
    {
      "epoch": 4.681603773584905,
      "grad_norm": 4.552534580230713,
      "learning_rate": 9.462529989571334e-06,
      "loss": 0.3136,
      "num_input_tokens_seen": 5227816,
      "step": 7940
    },
    {
      "epoch": 4.684551886792453,
      "grad_norm": 2.1334056854248047,
      "learning_rate": 9.461369013429595e-06,
      "loss": 0.4923,
      "num_input_tokens_seen": 5230888,
      "step": 7945
    },
    {
      "epoch": 4.6875,
      "grad_norm": 2.1324379444122314,
      "learning_rate": 9.460206856121183e-06,
      "loss": 0.4404,
      "num_input_tokens_seen": 5234280,
      "step": 7950
    },
    {
      "epoch": 4.690448113207547,
      "grad_norm": 4.344475746154785,
      "learning_rate": 9.459043517953786e-06,
      "loss": 0.4538,
      "num_input_tokens_seen": 5237768,
      "step": 7955
    },
    {
      "epoch": 4.693396226415095,
      "grad_norm": 2.550166368484497,
      "learning_rate": 9.457878999235396e-06,
      "loss": 0.5962,
      "num_input_tokens_seen": 5240648,
      "step": 7960
    },
    {
      "epoch": 4.696344339622642,
      "grad_norm": 1.71035635471344,
      "learning_rate": 9.45671330027433e-06,
      "loss": 0.417,
      "num_input_tokens_seen": 5243496,
      "step": 7965
    },
    {
      "epoch": 4.699292452830189,
      "grad_norm": 2.453404426574707,
      "learning_rate": 9.45554642137921e-06,
      "loss": 0.5585,
      "num_input_tokens_seen": 5245832,
      "step": 7970
    },
    {
      "epoch": 4.7022405660377355,
      "grad_norm": 1.9945875406265259,
      "learning_rate": 9.454378362858974e-06,
      "loss": 0.4596,
      "num_input_tokens_seen": 5249608,
      "step": 7975
    },
    {
      "epoch": 4.705188679245283,
      "grad_norm": 4.432562828063965,
      "learning_rate": 9.453209125022867e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 5252840,
      "step": 7980
    },
    {
      "epoch": 4.70813679245283,
      "grad_norm": 2.3953752517700195,
      "learning_rate": 9.452038708180453e-06,
      "loss": 0.5127,
      "num_input_tokens_seen": 5256584,
      "step": 7985
    },
    {
      "epoch": 4.711084905660377,
      "grad_norm": 1.3791389465332031,
      "learning_rate": 9.450867112641603e-06,
      "loss": 0.5534,
      "num_input_tokens_seen": 5259816,
      "step": 7990
    },
    {
      "epoch": 4.714033018867925,
      "grad_norm": 1.6101717948913574,
      "learning_rate": 9.449694338716506e-06,
      "loss": 0.4088,
      "num_input_tokens_seen": 5263496,
      "step": 7995
    },
    {
      "epoch": 4.716981132075472,
      "grad_norm": 2.2771811485290527,
      "learning_rate": 9.448520386715653e-06,
      "loss": 0.4728,
      "num_input_tokens_seen": 5266184,
      "step": 8000
    },
    {
      "epoch": 4.719929245283019,
      "grad_norm": 2.3576252460479736,
      "learning_rate": 9.447345256949855e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 5268584,
      "step": 8005
    },
    {
      "epoch": 4.722877358490566,
      "grad_norm": 2.4585437774658203,
      "learning_rate": 9.446168949730234e-06,
      "loss": 0.4171,
      "num_input_tokens_seen": 5271912,
      "step": 8010
    },
    {
      "epoch": 4.725825471698113,
      "grad_norm": 0.8210746645927429,
      "learning_rate": 9.444991465368223e-06,
      "loss": 0.6409,
      "num_input_tokens_seen": 5274760,
      "step": 8015
    },
    {
      "epoch": 4.72877358490566,
      "grad_norm": 1.2236535549163818,
      "learning_rate": 9.443812804175562e-06,
      "loss": 0.3694,
      "num_input_tokens_seen": 5278376,
      "step": 8020
    },
    {
      "epoch": 4.7317216981132075,
      "grad_norm": 1.9746432304382324,
      "learning_rate": 9.44263296646431e-06,
      "loss": 0.3494,
      "num_input_tokens_seen": 5281320,
      "step": 8025
    },
    {
      "epoch": 4.734669811320755,
      "grad_norm": 3.665078639984131,
      "learning_rate": 9.441451952546835e-06,
      "loss": 0.563,
      "num_input_tokens_seen": 5284808,
      "step": 8030
    },
    {
      "epoch": 4.737617924528302,
      "grad_norm": 1.5734410285949707,
      "learning_rate": 9.440269762735814e-06,
      "loss": 0.5945,
      "num_input_tokens_seen": 5287496,
      "step": 8035
    },
    {
      "epoch": 4.740566037735849,
      "grad_norm": 2.1040127277374268,
      "learning_rate": 9.439086397344236e-06,
      "loss": 0.5003,
      "num_input_tokens_seen": 5290472,
      "step": 8040
    },
    {
      "epoch": 4.743514150943396,
      "grad_norm": 2.779982328414917,
      "learning_rate": 9.437901856685404e-06,
      "loss": 0.5254,
      "num_input_tokens_seen": 5293800,
      "step": 8045
    },
    {
      "epoch": 4.746462264150943,
      "grad_norm": 1.8347662687301636,
      "learning_rate": 9.436716141072925e-06,
      "loss": 0.544,
      "num_input_tokens_seen": 5296520,
      "step": 8050
    },
    {
      "epoch": 4.74941037735849,
      "grad_norm": 1.5035545825958252,
      "learning_rate": 9.435529250820732e-06,
      "loss": 0.4185,
      "num_input_tokens_seen": 5299496,
      "step": 8055
    },
    {
      "epoch": 4.752358490566038,
      "grad_norm": 1.4474151134490967,
      "learning_rate": 9.43434118624305e-06,
      "loss": 0.5485,
      "num_input_tokens_seen": 5303432,
      "step": 8060
    },
    {
      "epoch": 4.755306603773585,
      "grad_norm": 1.9702975749969482,
      "learning_rate": 9.433151947654428e-06,
      "loss": 0.5121,
      "num_input_tokens_seen": 5307720,
      "step": 8065
    },
    {
      "epoch": 4.758254716981132,
      "grad_norm": 1.8501344919204712,
      "learning_rate": 9.431961535369724e-06,
      "loss": 0.4827,
      "num_input_tokens_seen": 5311272,
      "step": 8070
    },
    {
      "epoch": 4.7612028301886795,
      "grad_norm": 2.4606504440307617,
      "learning_rate": 9.430769949704103e-06,
      "loss": 0.377,
      "num_input_tokens_seen": 5314568,
      "step": 8075
    },
    {
      "epoch": 4.764150943396227,
      "grad_norm": 2.351323127746582,
      "learning_rate": 9.42957719097304e-06,
      "loss": 0.5035,
      "num_input_tokens_seen": 5317864,
      "step": 8080
    },
    {
      "epoch": 4.767099056603773,
      "grad_norm": 1.163707971572876,
      "learning_rate": 9.42838325949233e-06,
      "loss": 0.3304,
      "num_input_tokens_seen": 5321096,
      "step": 8085
    },
    {
      "epoch": 4.7700471698113205,
      "grad_norm": 1.7007471323013306,
      "learning_rate": 9.427188155578062e-06,
      "loss": 0.4516,
      "num_input_tokens_seen": 5325000,
      "step": 8090
    },
    {
      "epoch": 4.772995283018868,
      "grad_norm": 1.8435297012329102,
      "learning_rate": 9.42599187954665e-06,
      "loss": 0.4767,
      "num_input_tokens_seen": 5327752,
      "step": 8095
    },
    {
      "epoch": 4.775943396226415,
      "grad_norm": 2.292861223220825,
      "learning_rate": 9.424794431714814e-06,
      "loss": 0.6225,
      "num_input_tokens_seen": 5331112,
      "step": 8100
    },
    {
      "epoch": 4.778891509433962,
      "grad_norm": 2.3473498821258545,
      "learning_rate": 9.423595812399581e-06,
      "loss": 0.4328,
      "num_input_tokens_seen": 5335336,
      "step": 8105
    },
    {
      "epoch": 4.78183962264151,
      "grad_norm": 1.8841150999069214,
      "learning_rate": 9.422396021918296e-06,
      "loss": 0.575,
      "num_input_tokens_seen": 5339240,
      "step": 8110
    },
    {
      "epoch": 4.784787735849057,
      "grad_norm": 1.7598662376403809,
      "learning_rate": 9.421195060588602e-06,
      "loss": 0.3858,
      "num_input_tokens_seen": 5343560,
      "step": 8115
    },
    {
      "epoch": 4.787735849056604,
      "grad_norm": 1.5389039516448975,
      "learning_rate": 9.419992928728461e-06,
      "loss": 0.5195,
      "num_input_tokens_seen": 5346920,
      "step": 8120
    },
    {
      "epoch": 4.790683962264151,
      "grad_norm": 1.6455289125442505,
      "learning_rate": 9.418789626656144e-06,
      "loss": 0.4021,
      "num_input_tokens_seen": 5350600,
      "step": 8125
    },
    {
      "epoch": 4.793632075471698,
      "grad_norm": 1.624757170677185,
      "learning_rate": 9.417585154690229e-06,
      "loss": 0.4669,
      "num_input_tokens_seen": 5353576,
      "step": 8130
    },
    {
      "epoch": 4.796580188679245,
      "grad_norm": 2.2532358169555664,
      "learning_rate": 9.416379513149605e-06,
      "loss": 0.4523,
      "num_input_tokens_seen": 5356968,
      "step": 8135
    },
    {
      "epoch": 4.7995283018867925,
      "grad_norm": 3.9590392112731934,
      "learning_rate": 9.415172702353471e-06,
      "loss": 0.5766,
      "num_input_tokens_seen": 5360744,
      "step": 8140
    },
    {
      "epoch": 4.80247641509434,
      "grad_norm": 1.8093452453613281,
      "learning_rate": 9.413964722621339e-06,
      "loss": 0.4836,
      "num_input_tokens_seen": 5364968,
      "step": 8145
    },
    {
      "epoch": 4.805424528301887,
      "grad_norm": 2.1799535751342773,
      "learning_rate": 9.41275557427302e-06,
      "loss": 0.4551,
      "num_input_tokens_seen": 5367688,
      "step": 8150
    },
    {
      "epoch": 4.808372641509434,
      "grad_norm": 1.5219428539276123,
      "learning_rate": 9.411545257628646e-06,
      "loss": 0.3686,
      "num_input_tokens_seen": 5370152,
      "step": 8155
    },
    {
      "epoch": 4.811320754716981,
      "grad_norm": 1.2784806489944458,
      "learning_rate": 9.410333773008653e-06,
      "loss": 0.5504,
      "num_input_tokens_seen": 5374056,
      "step": 8160
    },
    {
      "epoch": 4.814268867924528,
      "grad_norm": 3.306267499923706,
      "learning_rate": 9.409121120733784e-06,
      "loss": 0.4036,
      "num_input_tokens_seen": 5380904,
      "step": 8165
    },
    {
      "epoch": 4.817216981132075,
      "grad_norm": 2.0579781532287598,
      "learning_rate": 9.4079073011251e-06,
      "loss": 0.3967,
      "num_input_tokens_seen": 5383144,
      "step": 8170
    },
    {
      "epoch": 4.820165094339623,
      "grad_norm": 1.1466025114059448,
      "learning_rate": 9.406692314503956e-06,
      "loss": 0.486,
      "num_input_tokens_seen": 5388008,
      "step": 8175
    },
    {
      "epoch": 4.82311320754717,
      "grad_norm": 1.874875783920288,
      "learning_rate": 9.405476161192033e-06,
      "loss": 0.4633,
      "num_input_tokens_seen": 5390408,
      "step": 8180
    },
    {
      "epoch": 4.826061320754717,
      "grad_norm": 1.9117014408111572,
      "learning_rate": 9.40425884151131e-06,
      "loss": 0.4813,
      "num_input_tokens_seen": 5393736,
      "step": 8185
    },
    {
      "epoch": 4.8290094339622645,
      "grad_norm": 2.215020179748535,
      "learning_rate": 9.403040355784076e-06,
      "loss": 0.5507,
      "num_input_tokens_seen": 5397128,
      "step": 8190
    },
    {
      "epoch": 4.831957547169811,
      "grad_norm": 5.830589294433594,
      "learning_rate": 9.401820704332932e-06,
      "loss": 0.4842,
      "num_input_tokens_seen": 5400456,
      "step": 8195
    },
    {
      "epoch": 4.834905660377358,
      "grad_norm": 1.627112865447998,
      "learning_rate": 9.400599887480786e-06,
      "loss": 0.4822,
      "num_input_tokens_seen": 5403272,
      "step": 8200
    },
    {
      "epoch": 4.837853773584905,
      "grad_norm": 1.9509440660476685,
      "learning_rate": 9.399377905550854e-06,
      "loss": 0.5228,
      "num_input_tokens_seen": 5406952,
      "step": 8205
    },
    {
      "epoch": 4.840801886792453,
      "grad_norm": 1.1362595558166504,
      "learning_rate": 9.398154758866662e-06,
      "loss": 0.4535,
      "num_input_tokens_seen": 5411272,
      "step": 8210
    },
    {
      "epoch": 4.84375,
      "grad_norm": 1.1073936223983765,
      "learning_rate": 9.396930447752041e-06,
      "loss": 0.3678,
      "num_input_tokens_seen": 5414760,
      "step": 8215
    },
    {
      "epoch": 4.846698113207547,
      "grad_norm": 1.9867472648620605,
      "learning_rate": 9.395704972531137e-06,
      "loss": 0.4502,
      "num_input_tokens_seen": 5418248,
      "step": 8220
    },
    {
      "epoch": 4.849646226415095,
      "grad_norm": 2.039151191711426,
      "learning_rate": 9.394478333528396e-06,
      "loss": 0.4865,
      "num_input_tokens_seen": 5421096,
      "step": 8225
    },
    {
      "epoch": 4.852594339622642,
      "grad_norm": 2.574596881866455,
      "learning_rate": 9.393250531068576e-06,
      "loss": 0.5134,
      "num_input_tokens_seen": 5424648,
      "step": 8230
    },
    {
      "epoch": 4.855542452830189,
      "grad_norm": 2.8930532932281494,
      "learning_rate": 9.392021565476744e-06,
      "loss": 0.3596,
      "num_input_tokens_seen": 5428904,
      "step": 8235
    },
    {
      "epoch": 4.8584905660377355,
      "grad_norm": 1.8422049283981323,
      "learning_rate": 9.390791437078274e-06,
      "loss": 0.4312,
      "num_input_tokens_seen": 5431720,
      "step": 8240
    },
    {
      "epoch": 4.861438679245283,
      "grad_norm": 2.044743061065674,
      "learning_rate": 9.38956014619885e-06,
      "loss": 0.4761,
      "num_input_tokens_seen": 5434728,
      "step": 8245
    },
    {
      "epoch": 4.86438679245283,
      "grad_norm": 1.2801305055618286,
      "learning_rate": 9.388327693164456e-06,
      "loss": 0.4107,
      "num_input_tokens_seen": 5437288,
      "step": 8250
    },
    {
      "epoch": 4.867334905660377,
      "grad_norm": 2.955547332763672,
      "learning_rate": 9.387094078301395e-06,
      "loss": 0.5555,
      "num_input_tokens_seen": 5440392,
      "step": 8255
    },
    {
      "epoch": 4.870283018867925,
      "grad_norm": 4.032364845275879,
      "learning_rate": 9.385859301936269e-06,
      "loss": 0.5199,
      "num_input_tokens_seen": 5444392,
      "step": 8260
    },
    {
      "epoch": 4.873231132075472,
      "grad_norm": 1.256404161453247,
      "learning_rate": 9.38462336439599e-06,
      "loss": 0.3905,
      "num_input_tokens_seen": 5448840,
      "step": 8265
    },
    {
      "epoch": 4.876179245283019,
      "grad_norm": 2.508655309677124,
      "learning_rate": 9.383386266007779e-06,
      "loss": 0.402,
      "num_input_tokens_seen": 5452168,
      "step": 8270
    },
    {
      "epoch": 4.879127358490566,
      "grad_norm": 1.5617269277572632,
      "learning_rate": 9.382148007099164e-06,
      "loss": 0.5605,
      "num_input_tokens_seen": 5455784,
      "step": 8275
    },
    {
      "epoch": 4.882075471698113,
      "grad_norm": 2.1329777240753174,
      "learning_rate": 9.380908587997977e-06,
      "loss": 0.5192,
      "num_input_tokens_seen": 5459080,
      "step": 8280
    },
    {
      "epoch": 4.88502358490566,
      "grad_norm": 2.3926479816436768,
      "learning_rate": 9.37966800903236e-06,
      "loss": 0.3765,
      "num_input_tokens_seen": 5462920,
      "step": 8285
    },
    {
      "epoch": 4.8879716981132075,
      "grad_norm": 2.140406847000122,
      "learning_rate": 9.378426270530762e-06,
      "loss": 0.5797,
      "num_input_tokens_seen": 5465064,
      "step": 8290
    },
    {
      "epoch": 4.890919811320755,
      "grad_norm": 2.260577917098999,
      "learning_rate": 9.37718337282194e-06,
      "loss": 0.3695,
      "num_input_tokens_seen": 5467432,
      "step": 8295
    },
    {
      "epoch": 4.893867924528302,
      "grad_norm": 1.7586777210235596,
      "learning_rate": 9.375939316234956e-06,
      "loss": 0.4766,
      "num_input_tokens_seen": 5471624,
      "step": 8300
    },
    {
      "epoch": 4.896816037735849,
      "grad_norm": 2.0518639087677,
      "learning_rate": 9.374694101099178e-06,
      "loss": 0.4213,
      "num_input_tokens_seen": 5475144,
      "step": 8305
    },
    {
      "epoch": 4.899764150943396,
      "grad_norm": 1.2941980361938477,
      "learning_rate": 9.373447727744282e-06,
      "loss": 0.5072,
      "num_input_tokens_seen": 5478600,
      "step": 8310
    },
    {
      "epoch": 4.902712264150943,
      "grad_norm": 2.226104974746704,
      "learning_rate": 9.372200196500253e-06,
      "loss": 0.5028,
      "num_input_tokens_seen": 5483112,
      "step": 8315
    },
    {
      "epoch": 4.90566037735849,
      "grad_norm": 4.212259292602539,
      "learning_rate": 9.37095150769738e-06,
      "loss": 0.4719,
      "num_input_tokens_seen": 5486344,
      "step": 8320
    },
    {
      "epoch": 4.908608490566038,
      "grad_norm": 2.233198404312134,
      "learning_rate": 9.369701661666255e-06,
      "loss": 0.4273,
      "num_input_tokens_seen": 5489288,
      "step": 8325
    },
    {
      "epoch": 4.911556603773585,
      "grad_norm": 1.8000880479812622,
      "learning_rate": 9.368450658737782e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 5492360,
      "step": 8330
    },
    {
      "epoch": 4.914504716981132,
      "grad_norm": 4.031656265258789,
      "learning_rate": 9.367198499243173e-06,
      "loss": 0.5682,
      "num_input_tokens_seen": 5494952,
      "step": 8335
    },
    {
      "epoch": 4.9174528301886795,
      "grad_norm": 2.7462050914764404,
      "learning_rate": 9.365945183513938e-06,
      "loss": 0.4373,
      "num_input_tokens_seen": 5497736,
      "step": 8340
    },
    {
      "epoch": 4.920400943396227,
      "grad_norm": 1.3366950750350952,
      "learning_rate": 9.3646907118819e-06,
      "loss": 0.3846,
      "num_input_tokens_seen": 5501352,
      "step": 8345
    },
    {
      "epoch": 4.923349056603773,
      "grad_norm": 1.6242331266403198,
      "learning_rate": 9.363435084679185e-06,
      "loss": 0.345,
      "num_input_tokens_seen": 5504040,
      "step": 8350
    },
    {
      "epoch": 4.9262971698113205,
      "grad_norm": 1.4274379014968872,
      "learning_rate": 9.362178302238227e-06,
      "loss": 0.4816,
      "num_input_tokens_seen": 5507304,
      "step": 8355
    },
    {
      "epoch": 4.929245283018868,
      "grad_norm": 1.1250410079956055,
      "learning_rate": 9.360920364891762e-06,
      "loss": 0.4617,
      "num_input_tokens_seen": 5514216,
      "step": 8360
    },
    {
      "epoch": 4.932193396226415,
      "grad_norm": 2.693946361541748,
      "learning_rate": 9.359661272972836e-06,
      "loss": 0.4421,
      "num_input_tokens_seen": 5517544,
      "step": 8365
    },
    {
      "epoch": 4.935141509433962,
      "grad_norm": 3.258152484893799,
      "learning_rate": 9.3584010268148e-06,
      "loss": 0.4665,
      "num_input_tokens_seen": 5520584,
      "step": 8370
    },
    {
      "epoch": 4.93808962264151,
      "grad_norm": 1.2574036121368408,
      "learning_rate": 9.357139626751308e-06,
      "loss": 0.4224,
      "num_input_tokens_seen": 5523240,
      "step": 8375
    },
    {
      "epoch": 4.941037735849057,
      "grad_norm": 2.0517282485961914,
      "learning_rate": 9.355877073116321e-06,
      "loss": 0.494,
      "num_input_tokens_seen": 5526696,
      "step": 8380
    },
    {
      "epoch": 4.943985849056604,
      "grad_norm": 1.6362618207931519,
      "learning_rate": 9.354613366244108e-06,
      "loss": 0.5885,
      "num_input_tokens_seen": 5529320,
      "step": 8385
    },
    {
      "epoch": 4.946933962264151,
      "grad_norm": 1.6025888919830322,
      "learning_rate": 9.353348506469236e-06,
      "loss": 0.3566,
      "num_input_tokens_seen": 5532360,
      "step": 8390
    },
    {
      "epoch": 4.949882075471698,
      "grad_norm": 1.0173109769821167,
      "learning_rate": 9.352082494126586e-06,
      "loss": 0.4047,
      "num_input_tokens_seen": 5535752,
      "step": 8395
    },
    {
      "epoch": 4.952830188679245,
      "grad_norm": 2.847536325454712,
      "learning_rate": 9.350815329551341e-06,
      "loss": 0.4214,
      "num_input_tokens_seen": 5539208,
      "step": 8400
    },
    {
      "epoch": 4.9557783018867925,
      "grad_norm": 1.7493538856506348,
      "learning_rate": 9.349547013078986e-06,
      "loss": 0.5202,
      "num_input_tokens_seen": 5543240,
      "step": 8405
    },
    {
      "epoch": 4.95872641509434,
      "grad_norm": 6.680507659912109,
      "learning_rate": 9.348277545045312e-06,
      "loss": 0.6079,
      "num_input_tokens_seen": 5546120,
      "step": 8410
    },
    {
      "epoch": 4.961674528301887,
      "grad_norm": 1.404626488685608,
      "learning_rate": 9.347006925786418e-06,
      "loss": 0.5532,
      "num_input_tokens_seen": 5549736,
      "step": 8415
    },
    {
      "epoch": 4.964622641509434,
      "grad_norm": 2.169457197189331,
      "learning_rate": 9.34573515563871e-06,
      "loss": 0.4977,
      "num_input_tokens_seen": 5553224,
      "step": 8420
    },
    {
      "epoch": 4.967570754716981,
      "grad_norm": 2.713348627090454,
      "learning_rate": 9.344462234938885e-06,
      "loss": 0.3935,
      "num_input_tokens_seen": 5556168,
      "step": 8425
    },
    {
      "epoch": 4.970518867924528,
      "grad_norm": 1.51192045211792,
      "learning_rate": 9.343188164023962e-06,
      "loss": 0.8096,
      "num_input_tokens_seen": 5560072,
      "step": 8430
    },
    {
      "epoch": 4.973466981132075,
      "grad_norm": 2.6568405628204346,
      "learning_rate": 9.341912943231256e-06,
      "loss": 0.5861,
      "num_input_tokens_seen": 5562696,
      "step": 8435
    },
    {
      "epoch": 4.976415094339623,
      "grad_norm": 1.4927059412002563,
      "learning_rate": 9.340636572898383e-06,
      "loss": 0.5444,
      "num_input_tokens_seen": 5566984,
      "step": 8440
    },
    {
      "epoch": 4.97936320754717,
      "grad_norm": 7.546762466430664,
      "learning_rate": 9.339359053363272e-06,
      "loss": 0.4823,
      "num_input_tokens_seen": 5571112,
      "step": 8445
    },
    {
      "epoch": 4.982311320754717,
      "grad_norm": 2.077305793762207,
      "learning_rate": 9.338080384964148e-06,
      "loss": 0.4066,
      "num_input_tokens_seen": 5574024,
      "step": 8450
    },
    {
      "epoch": 4.9852594339622645,
      "grad_norm": 1.6700588464736938,
      "learning_rate": 9.336800568039548e-06,
      "loss": 0.4422,
      "num_input_tokens_seen": 5578600,
      "step": 8455
    },
    {
      "epoch": 4.988207547169811,
      "grad_norm": 1.4363809823989868,
      "learning_rate": 9.335519602928307e-06,
      "loss": 0.4496,
      "num_input_tokens_seen": 5581928,
      "step": 8460
    },
    {
      "epoch": 4.991155660377358,
      "grad_norm": 1.69771409034729,
      "learning_rate": 9.334237489969565e-06,
      "loss": 0.3551,
      "num_input_tokens_seen": 5584776,
      "step": 8465
    },
    {
      "epoch": 4.994103773584905,
      "grad_norm": 1.1503379344940186,
      "learning_rate": 9.332954229502768e-06,
      "loss": 0.5198,
      "num_input_tokens_seen": 5588040,
      "step": 8470
    },
    {
      "epoch": 4.997051886792453,
      "grad_norm": 2.725942373275757,
      "learning_rate": 9.331669821867665e-06,
      "loss": 0.3815,
      "num_input_tokens_seen": 5591208,
      "step": 8475
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.9226576089859009,
      "learning_rate": 9.33038426740431e-06,
      "loss": 0.458,
      "num_input_tokens_seen": 5593864,
      "step": 8480
    },
    {
      "epoch": 5.002948113207547,
      "grad_norm": 1.6762280464172363,
      "learning_rate": 9.329097566453055e-06,
      "loss": 0.3867,
      "num_input_tokens_seen": 5596936,
      "step": 8485
    },
    {
      "epoch": 5.005896226415095,
      "grad_norm": 2.7340221405029297,
      "learning_rate": 9.327809719354564e-06,
      "loss": 0.476,
      "num_input_tokens_seen": 5599752,
      "step": 8490
    },
    {
      "epoch": 5.008844339622642,
      "grad_norm": 2.4800002574920654,
      "learning_rate": 9.326520726449795e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 5602888,
      "step": 8495
    },
    {
      "epoch": 5.011792452830188,
      "grad_norm": 1.459059715270996,
      "learning_rate": 9.32523058808002e-06,
      "loss": 0.6877,
      "num_input_tokens_seen": 5606408,
      "step": 8500
    },
    {
      "epoch": 5.0147405660377355,
      "grad_norm": 1.9232797622680664,
      "learning_rate": 9.323939304586806e-06,
      "loss": 0.4303,
      "num_input_tokens_seen": 5609416,
      "step": 8505
    },
    {
      "epoch": 5.017688679245283,
      "grad_norm": 2.9160780906677246,
      "learning_rate": 9.322646876312025e-06,
      "loss": 0.3231,
      "num_input_tokens_seen": 5612200,
      "step": 8510
    },
    {
      "epoch": 5.02063679245283,
      "grad_norm": 1.8944355249404907,
      "learning_rate": 9.321353303597854e-06,
      "loss": 0.5221,
      "num_input_tokens_seen": 5618024,
      "step": 8515
    },
    {
      "epoch": 5.023584905660377,
      "grad_norm": 1.0977389812469482,
      "learning_rate": 9.320058586786771e-06,
      "loss": 0.4345,
      "num_input_tokens_seen": 5620648,
      "step": 8520
    },
    {
      "epoch": 5.026533018867925,
      "grad_norm": 1.4010158777236938,
      "learning_rate": 9.31876272622156e-06,
      "loss": 0.3552,
      "num_input_tokens_seen": 5623784,
      "step": 8525
    },
    {
      "epoch": 5.029481132075472,
      "grad_norm": 2.10634708404541,
      "learning_rate": 9.317465722245305e-06,
      "loss": 0.5272,
      "num_input_tokens_seen": 5626600,
      "step": 8530
    },
    {
      "epoch": 5.032429245283019,
      "grad_norm": 2.650879144668579,
      "learning_rate": 9.316167575201391e-06,
      "loss": 0.4404,
      "num_input_tokens_seen": 5630312,
      "step": 8535
    },
    {
      "epoch": 5.035377358490566,
      "grad_norm": 2.5495622158050537,
      "learning_rate": 9.31486828543351e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 5633288,
      "step": 8540
    },
    {
      "epoch": 5.038325471698113,
      "grad_norm": 1.474488377571106,
      "learning_rate": 9.313567853285656e-06,
      "loss": 0.3826,
      "num_input_tokens_seen": 5636616,
      "step": 8545
    },
    {
      "epoch": 5.04127358490566,
      "grad_norm": 1.7291898727416992,
      "learning_rate": 9.31226627910212e-06,
      "loss": 0.3861,
      "num_input_tokens_seen": 5641032,
      "step": 8550
    },
    {
      "epoch": 5.0442216981132075,
      "grad_norm": 2.323779821395874,
      "learning_rate": 9.310963563227504e-06,
      "loss": 0.4716,
      "num_input_tokens_seen": 5643912,
      "step": 8555
    },
    {
      "epoch": 5.047169811320755,
      "grad_norm": 1.9827604293823242,
      "learning_rate": 9.309659706006704e-06,
      "loss": 0.4181,
      "num_input_tokens_seen": 5648008,
      "step": 8560
    },
    {
      "epoch": 5.050117924528302,
      "grad_norm": 2.050431251525879,
      "learning_rate": 9.308354707784925e-06,
      "loss": 0.3722,
      "num_input_tokens_seen": 5652136,
      "step": 8565
    },
    {
      "epoch": 5.053066037735849,
      "grad_norm": 1.1452800035476685,
      "learning_rate": 9.307048568907669e-06,
      "loss": 0.3766,
      "num_input_tokens_seen": 5656264,
      "step": 8570
    },
    {
      "epoch": 5.056014150943396,
      "grad_norm": 3.99790358543396,
      "learning_rate": 9.30574128972074e-06,
      "loss": 0.4087,
      "num_input_tokens_seen": 5658408,
      "step": 8575
    },
    {
      "epoch": 5.058962264150943,
      "grad_norm": 1.7324029207229614,
      "learning_rate": 9.304432870570247e-06,
      "loss": 0.3831,
      "num_input_tokens_seen": 5661672,
      "step": 8580
    },
    {
      "epoch": 5.06191037735849,
      "grad_norm": 1.7526251077651978,
      "learning_rate": 9.303123311802605e-06,
      "loss": 0.4019,
      "num_input_tokens_seen": 5664328,
      "step": 8585
    },
    {
      "epoch": 5.064858490566038,
      "grad_norm": 2.0135161876678467,
      "learning_rate": 9.301812613764516e-06,
      "loss": 0.3171,
      "num_input_tokens_seen": 5667816,
      "step": 8590
    },
    {
      "epoch": 5.067806603773585,
      "grad_norm": 1.5631194114685059,
      "learning_rate": 9.300500776803001e-06,
      "loss": 0.4603,
      "num_input_tokens_seen": 5671528,
      "step": 8595
    },
    {
      "epoch": 5.070754716981132,
      "grad_norm": 2.033721923828125,
      "learning_rate": 9.29918780126537e-06,
      "loss": 0.3857,
      "num_input_tokens_seen": 5674248,
      "step": 8600
    },
    {
      "epoch": 5.0737028301886795,
      "grad_norm": 1.2417939901351929,
      "learning_rate": 9.297873687499239e-06,
      "loss": 0.4444,
      "num_input_tokens_seen": 5678760,
      "step": 8605
    },
    {
      "epoch": 5.076650943396227,
      "grad_norm": 1.8914475440979004,
      "learning_rate": 9.296558435852528e-06,
      "loss": 0.5118,
      "num_input_tokens_seen": 5681128,
      "step": 8610
    },
    {
      "epoch": 5.079599056603773,
      "grad_norm": 1.7732535600662231,
      "learning_rate": 9.295242046673454e-06,
      "loss": 0.3769,
      "num_input_tokens_seen": 5684008,
      "step": 8615
    },
    {
      "epoch": 5.0825471698113205,
      "grad_norm": 5.014444351196289,
      "learning_rate": 9.293924520310535e-06,
      "loss": 0.6049,
      "num_input_tokens_seen": 5686344,
      "step": 8620
    },
    {
      "epoch": 5.085495283018868,
      "grad_norm": 2.099168062210083,
      "learning_rate": 9.292605857112595e-06,
      "loss": 0.4212,
      "num_input_tokens_seen": 5689384,
      "step": 8625
    },
    {
      "epoch": 5.088443396226415,
      "grad_norm": 1.82022225856781,
      "learning_rate": 9.291286057428755e-06,
      "loss": 0.4699,
      "num_input_tokens_seen": 5692168,
      "step": 8630
    },
    {
      "epoch": 5.091391509433962,
      "grad_norm": 2.5631892681121826,
      "learning_rate": 9.289965121608436e-06,
      "loss": 0.4806,
      "num_input_tokens_seen": 5694952,
      "step": 8635
    },
    {
      "epoch": 5.09433962264151,
      "grad_norm": 2.134756088256836,
      "learning_rate": 9.288643050001362e-06,
      "loss": 0.5384,
      "num_input_tokens_seen": 5697832,
      "step": 8640
    },
    {
      "epoch": 5.097287735849057,
      "grad_norm": 2.1347601413726807,
      "learning_rate": 9.287319842957557e-06,
      "loss": 0.4735,
      "num_input_tokens_seen": 5700808,
      "step": 8645
    },
    {
      "epoch": 5.100235849056604,
      "grad_norm": 2.0051255226135254,
      "learning_rate": 9.285995500827348e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 5703624,
      "step": 8650
    },
    {
      "epoch": 5.103183962264151,
      "grad_norm": 3.748734951019287,
      "learning_rate": 9.284670023961355e-06,
      "loss": 0.4706,
      "num_input_tokens_seen": 5706056,
      "step": 8655
    },
    {
      "epoch": 5.106132075471698,
      "grad_norm": 4.612092971801758,
      "learning_rate": 9.28334341271051e-06,
      "loss": 0.8608,
      "num_input_tokens_seen": 5709512,
      "step": 8660
    },
    {
      "epoch": 5.109080188679245,
      "grad_norm": 2.895564079284668,
      "learning_rate": 9.282015667426036e-06,
      "loss": 0.3945,
      "num_input_tokens_seen": 5712872,
      "step": 8665
    },
    {
      "epoch": 5.1120283018867925,
      "grad_norm": 2.016547679901123,
      "learning_rate": 9.280686788459461e-06,
      "loss": 0.3572,
      "num_input_tokens_seen": 5715528,
      "step": 8670
    },
    {
      "epoch": 5.11497641509434,
      "grad_norm": 1.6915256977081299,
      "learning_rate": 9.279356776162606e-06,
      "loss": 0.3282,
      "num_input_tokens_seen": 5718824,
      "step": 8675
    },
    {
      "epoch": 5.117924528301887,
      "grad_norm": 1.5141220092773438,
      "learning_rate": 9.278025630887607e-06,
      "loss": 0.4042,
      "num_input_tokens_seen": 5721576,
      "step": 8680
    },
    {
      "epoch": 5.120872641509434,
      "grad_norm": 3.104158401489258,
      "learning_rate": 9.27669335298688e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 5724520,
      "step": 8685
    },
    {
      "epoch": 5.123820754716981,
      "grad_norm": 1.8404011726379395,
      "learning_rate": 9.275359942813158e-06,
      "loss": 0.4582,
      "num_input_tokens_seen": 5727272,
      "step": 8690
    },
    {
      "epoch": 5.126768867924528,
      "grad_norm": 1.777918815612793,
      "learning_rate": 9.274025400719466e-06,
      "loss": 0.5175,
      "num_input_tokens_seen": 5730088,
      "step": 8695
    },
    {
      "epoch": 5.129716981132075,
      "grad_norm": 2.0912628173828125,
      "learning_rate": 9.27268972705913e-06,
      "loss": 0.4695,
      "num_input_tokens_seen": 5733544,
      "step": 8700
    },
    {
      "epoch": 5.132665094339623,
      "grad_norm": 2.2114758491516113,
      "learning_rate": 9.271352922185772e-06,
      "loss": 0.4732,
      "num_input_tokens_seen": 5736552,
      "step": 8705
    },
    {
      "epoch": 5.13561320754717,
      "grad_norm": 1.940643548965454,
      "learning_rate": 9.270014986453321e-06,
      "loss": 0.4506,
      "num_input_tokens_seen": 5739848,
      "step": 8710
    },
    {
      "epoch": 5.138561320754717,
      "grad_norm": 1.5474761724472046,
      "learning_rate": 9.268675920215999e-06,
      "loss": 0.4228,
      "num_input_tokens_seen": 5742888,
      "step": 8715
    },
    {
      "epoch": 5.1415094339622645,
      "grad_norm": 1.8288825750350952,
      "learning_rate": 9.26733572382833e-06,
      "loss": 0.4177,
      "num_input_tokens_seen": 5746088,
      "step": 8720
    },
    {
      "epoch": 5.144457547169812,
      "grad_norm": 2.0953354835510254,
      "learning_rate": 9.265994397645137e-06,
      "loss": 0.4201,
      "num_input_tokens_seen": 5749000,
      "step": 8725
    },
    {
      "epoch": 5.147405660377358,
      "grad_norm": 2.005282163619995,
      "learning_rate": 9.264651942021543e-06,
      "loss": 0.3347,
      "num_input_tokens_seen": 5752168,
      "step": 8730
    },
    {
      "epoch": 5.150353773584905,
      "grad_norm": 2.44761061668396,
      "learning_rate": 9.263308357312966e-06,
      "loss": 0.6201,
      "num_input_tokens_seen": 5755208,
      "step": 8735
    },
    {
      "epoch": 5.153301886792453,
      "grad_norm": 2.6490228176116943,
      "learning_rate": 9.26196364387513e-06,
      "loss": 0.5187,
      "num_input_tokens_seen": 5758952,
      "step": 8740
    },
    {
      "epoch": 5.15625,
      "grad_norm": 2.070124387741089,
      "learning_rate": 9.26061780206405e-06,
      "loss": 0.4334,
      "num_input_tokens_seen": 5761928,
      "step": 8745
    },
    {
      "epoch": 5.159198113207547,
      "grad_norm": 2.475288152694702,
      "learning_rate": 9.259270832236043e-06,
      "loss": 0.3834,
      "num_input_tokens_seen": 5765672,
      "step": 8750
    },
    {
      "epoch": 5.162146226415095,
      "grad_norm": 1.2369657754898071,
      "learning_rate": 9.257922734747729e-06,
      "loss": 0.3384,
      "num_input_tokens_seen": 5768104,
      "step": 8755
    },
    {
      "epoch": 5.165094339622642,
      "grad_norm": 2.0992391109466553,
      "learning_rate": 9.256573509956018e-06,
      "loss": 0.5079,
      "num_input_tokens_seen": 5771624,
      "step": 8760
    },
    {
      "epoch": 5.168042452830188,
      "grad_norm": 3.781494379043579,
      "learning_rate": 9.255223158218127e-06,
      "loss": 0.4394,
      "num_input_tokens_seen": 5774632,
      "step": 8765
    },
    {
      "epoch": 5.1709905660377355,
      "grad_norm": 1.4892091751098633,
      "learning_rate": 9.253871679891566e-06,
      "loss": 0.5058,
      "num_input_tokens_seen": 5777608,
      "step": 8770
    },
    {
      "epoch": 5.173938679245283,
      "grad_norm": 2.5400776863098145,
      "learning_rate": 9.252519075334143e-06,
      "loss": 0.382,
      "num_input_tokens_seen": 5782056,
      "step": 8775
    },
    {
      "epoch": 5.17688679245283,
      "grad_norm": 2.238816022872925,
      "learning_rate": 9.251165344903969e-06,
      "loss": 0.6249,
      "num_input_tokens_seen": 5785608,
      "step": 8780
    },
    {
      "epoch": 5.179834905660377,
      "grad_norm": 1.4717497825622559,
      "learning_rate": 9.249810488959448e-06,
      "loss": 0.4679,
      "num_input_tokens_seen": 5788456,
      "step": 8785
    },
    {
      "epoch": 5.182783018867925,
      "grad_norm": 2.707275152206421,
      "learning_rate": 9.248454507859285e-06,
      "loss": 0.4688,
      "num_input_tokens_seen": 5791560,
      "step": 8790
    },
    {
      "epoch": 5.185731132075472,
      "grad_norm": 1.1439323425292969,
      "learning_rate": 9.247097401962482e-06,
      "loss": 0.3496,
      "num_input_tokens_seen": 5794952,
      "step": 8795
    },
    {
      "epoch": 5.188679245283019,
      "grad_norm": 1.4359208345413208,
      "learning_rate": 9.245739171628335e-06,
      "loss": 0.3448,
      "num_input_tokens_seen": 5798920,
      "step": 8800
    },
    {
      "epoch": 5.191627358490566,
      "grad_norm": 4.51532506942749,
      "learning_rate": 9.244379817216447e-06,
      "loss": 0.5942,
      "num_input_tokens_seen": 5801192,
      "step": 8805
    },
    {
      "epoch": 5.194575471698113,
      "grad_norm": 2.314096689224243,
      "learning_rate": 9.243019339086708e-06,
      "loss": 0.487,
      "num_input_tokens_seen": 5803752,
      "step": 8810
    },
    {
      "epoch": 5.19752358490566,
      "grad_norm": 1.6844842433929443,
      "learning_rate": 9.241657737599313e-06,
      "loss": 0.4448,
      "num_input_tokens_seen": 5807720,
      "step": 8815
    },
    {
      "epoch": 5.2004716981132075,
      "grad_norm": 3.2737040519714355,
      "learning_rate": 9.240295013114752e-06,
      "loss": 0.4242,
      "num_input_tokens_seen": 5811016,
      "step": 8820
    },
    {
      "epoch": 5.203419811320755,
      "grad_norm": 1.5740948915481567,
      "learning_rate": 9.238931165993811e-06,
      "loss": 0.3739,
      "num_input_tokens_seen": 5813608,
      "step": 8825
    },
    {
      "epoch": 5.206367924528302,
      "grad_norm": 3.027733325958252,
      "learning_rate": 9.237566196597577e-06,
      "loss": 0.3522,
      "num_input_tokens_seen": 5816648,
      "step": 8830
    },
    {
      "epoch": 5.209316037735849,
      "grad_norm": 1.8429605960845947,
      "learning_rate": 9.236200105287427e-06,
      "loss": 0.5021,
      "num_input_tokens_seen": 5819944,
      "step": 8835
    },
    {
      "epoch": 5.212264150943396,
      "grad_norm": 2.274684190750122,
      "learning_rate": 9.234832892425042e-06,
      "loss": 0.4048,
      "num_input_tokens_seen": 5823272,
      "step": 8840
    },
    {
      "epoch": 5.215212264150943,
      "grad_norm": 1.4623850584030151,
      "learning_rate": 9.2334645583724e-06,
      "loss": 0.3904,
      "num_input_tokens_seen": 5827208,
      "step": 8845
    },
    {
      "epoch": 5.21816037735849,
      "grad_norm": 1.6855360269546509,
      "learning_rate": 9.23209510349177e-06,
      "loss": 0.3229,
      "num_input_tokens_seen": 5829832,
      "step": 8850
    },
    {
      "epoch": 5.221108490566038,
      "grad_norm": 1.7666687965393066,
      "learning_rate": 9.230724528145722e-06,
      "loss": 0.5481,
      "num_input_tokens_seen": 5834376,
      "step": 8855
    },
    {
      "epoch": 5.224056603773585,
      "grad_norm": 2.4588568210601807,
      "learning_rate": 9.229352832697122e-06,
      "loss": 0.496,
      "num_input_tokens_seen": 5837960,
      "step": 8860
    },
    {
      "epoch": 5.227004716981132,
      "grad_norm": 2.1935038566589355,
      "learning_rate": 9.22798001750913e-06,
      "loss": 0.4828,
      "num_input_tokens_seen": 5841512,
      "step": 8865
    },
    {
      "epoch": 5.2299528301886795,
      "grad_norm": 1.2483973503112793,
      "learning_rate": 9.226606082945209e-06,
      "loss": 0.4136,
      "num_input_tokens_seen": 5845224,
      "step": 8870
    },
    {
      "epoch": 5.232900943396227,
      "grad_norm": 2.8223001956939697,
      "learning_rate": 9.225231029369112e-06,
      "loss": 0.5369,
      "num_input_tokens_seen": 5847656,
      "step": 8875
    },
    {
      "epoch": 5.235849056603773,
      "grad_norm": 1.7996333837509155,
      "learning_rate": 9.22385485714489e-06,
      "loss": 0.4265,
      "num_input_tokens_seen": 5850952,
      "step": 8880
    },
    {
      "epoch": 5.2387971698113205,
      "grad_norm": 1.8387082815170288,
      "learning_rate": 9.222477566636889e-06,
      "loss": 0.4105,
      "num_input_tokens_seen": 5853480,
      "step": 8885
    },
    {
      "epoch": 5.241745283018868,
      "grad_norm": 1.6025086641311646,
      "learning_rate": 9.221099158209757e-06,
      "loss": 0.3943,
      "num_input_tokens_seen": 5856872,
      "step": 8890
    },
    {
      "epoch": 5.244693396226415,
      "grad_norm": 1.3683639764785767,
      "learning_rate": 9.219719632228429e-06,
      "loss": 0.4919,
      "num_input_tokens_seen": 5860040,
      "step": 8895
    },
    {
      "epoch": 5.247641509433962,
      "grad_norm": 2.7074859142303467,
      "learning_rate": 9.218338989058141e-06,
      "loss": 0.402,
      "num_input_tokens_seen": 5862792,
      "step": 8900
    },
    {
      "epoch": 5.25058962264151,
      "grad_norm": 1.7999284267425537,
      "learning_rate": 9.21695722906443e-06,
      "loss": 0.4578,
      "num_input_tokens_seen": 5865640,
      "step": 8905
    },
    {
      "epoch": 5.253537735849057,
      "grad_norm": 2.915243625640869,
      "learning_rate": 9.215574352613115e-06,
      "loss": 0.3933,
      "num_input_tokens_seen": 5868968,
      "step": 8910
    },
    {
      "epoch": 5.256485849056604,
      "grad_norm": 2.6291558742523193,
      "learning_rate": 9.214190360070323e-06,
      "loss": 0.6287,
      "num_input_tokens_seen": 5872136,
      "step": 8915
    },
    {
      "epoch": 5.259433962264151,
      "grad_norm": 1.3465003967285156,
      "learning_rate": 9.212805251802471e-06,
      "loss": 0.4596,
      "num_input_tokens_seen": 5875752,
      "step": 8920
    },
    {
      "epoch": 5.262382075471698,
      "grad_norm": 2.785482883453369,
      "learning_rate": 9.211419028176273e-06,
      "loss": 0.4006,
      "num_input_tokens_seen": 5879784,
      "step": 8925
    },
    {
      "epoch": 5.265330188679245,
      "grad_norm": 3.058912754058838,
      "learning_rate": 9.210031689558738e-06,
      "loss": 0.4267,
      "num_input_tokens_seen": 5883560,
      "step": 8930
    },
    {
      "epoch": 5.2682783018867925,
      "grad_norm": 1.5160787105560303,
      "learning_rate": 9.208643236317166e-06,
      "loss": 0.38,
      "num_input_tokens_seen": 5886216,
      "step": 8935
    },
    {
      "epoch": 5.27122641509434,
      "grad_norm": 1.9245682954788208,
      "learning_rate": 9.20725366881916e-06,
      "loss": 0.3661,
      "num_input_tokens_seen": 5888968,
      "step": 8940
    },
    {
      "epoch": 5.274174528301887,
      "grad_norm": 2.2130980491638184,
      "learning_rate": 9.205862987432614e-06,
      "loss": 0.4707,
      "num_input_tokens_seen": 5892072,
      "step": 8945
    },
    {
      "epoch": 5.277122641509434,
      "grad_norm": 1.5467150211334229,
      "learning_rate": 9.204471192525715e-06,
      "loss": 0.3819,
      "num_input_tokens_seen": 5894632,
      "step": 8950
    },
    {
      "epoch": 5.280070754716981,
      "grad_norm": 2.203312873840332,
      "learning_rate": 9.203078284466949e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 5897544,
      "step": 8955
    },
    {
      "epoch": 5.283018867924528,
      "grad_norm": 2.081510305404663,
      "learning_rate": 9.201684263625091e-06,
      "loss": 0.5121,
      "num_input_tokens_seen": 5900712,
      "step": 8960
    },
    {
      "epoch": 5.285966981132075,
      "grad_norm": 7.534655570983887,
      "learning_rate": 9.200289130369218e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 5903848,
      "step": 8965
    },
    {
      "epoch": 5.288915094339623,
      "grad_norm": 1.9212706089019775,
      "learning_rate": 9.198892885068693e-06,
      "loss": 0.5105,
      "num_input_tokens_seen": 5907240,
      "step": 8970
    },
    {
      "epoch": 5.29186320754717,
      "grad_norm": 1.4610099792480469,
      "learning_rate": 9.197495528093182e-06,
      "loss": 0.3932,
      "num_input_tokens_seen": 5910664,
      "step": 8975
    },
    {
      "epoch": 5.294811320754717,
      "grad_norm": 1.3720444440841675,
      "learning_rate": 9.196097059812639e-06,
      "loss": 0.3516,
      "num_input_tokens_seen": 5914440,
      "step": 8980
    },
    {
      "epoch": 5.2977594339622645,
      "grad_norm": 1.2843266725540161,
      "learning_rate": 9.194697480597316e-06,
      "loss": 0.3633,
      "num_input_tokens_seen": 5918376,
      "step": 8985
    },
    {
      "epoch": 5.300707547169811,
      "grad_norm": 1.545376181602478,
      "learning_rate": 9.193296790817755e-06,
      "loss": 0.4814,
      "num_input_tokens_seen": 5922408,
      "step": 8990
    },
    {
      "epoch": 5.303655660377358,
      "grad_norm": 3.1219232082366943,
      "learning_rate": 9.1918949908448e-06,
      "loss": 0.5305,
      "num_input_tokens_seen": 5926824,
      "step": 8995
    },
    {
      "epoch": 5.306603773584905,
      "grad_norm": 3.152188539505005,
      "learning_rate": 9.190492081049578e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 5929352,
      "step": 9000
    },
    {
      "epoch": 5.309551886792453,
      "grad_norm": 2.4260165691375732,
      "learning_rate": 9.189088061803517e-06,
      "loss": 0.5947,
      "num_input_tokens_seen": 5932616,
      "step": 9005
    },
    {
      "epoch": 5.3125,
      "grad_norm": 2.9633803367614746,
      "learning_rate": 9.187682933478337e-06,
      "loss": 0.5621,
      "num_input_tokens_seen": 5935496,
      "step": 9010
    },
    {
      "epoch": 5.315448113207547,
      "grad_norm": 1.8959388732910156,
      "learning_rate": 9.186276696446054e-06,
      "loss": 0.4553,
      "num_input_tokens_seen": 5938600,
      "step": 9015
    },
    {
      "epoch": 5.318396226415095,
      "grad_norm": 2.037705421447754,
      "learning_rate": 9.184869351078974e-06,
      "loss": 0.4703,
      "num_input_tokens_seen": 5942408,
      "step": 9020
    },
    {
      "epoch": 5.321344339622642,
      "grad_norm": 1.8362700939178467,
      "learning_rate": 9.183460897749697e-06,
      "loss": 0.4289,
      "num_input_tokens_seen": 5944968,
      "step": 9025
    },
    {
      "epoch": 5.324292452830189,
      "grad_norm": 2.1846697330474854,
      "learning_rate": 9.182051336831117e-06,
      "loss": 0.4282,
      "num_input_tokens_seen": 5948104,
      "step": 9030
    },
    {
      "epoch": 5.3272405660377355,
      "grad_norm": 1.9703428745269775,
      "learning_rate": 9.180640668696424e-06,
      "loss": 0.5665,
      "num_input_tokens_seen": 5951432,
      "step": 9035
    },
    {
      "epoch": 5.330188679245283,
      "grad_norm": 1.4547951221466064,
      "learning_rate": 9.179228893719094e-06,
      "loss": 0.532,
      "num_input_tokens_seen": 5954696,
      "step": 9040
    },
    {
      "epoch": 5.33313679245283,
      "grad_norm": 5.3467888832092285,
      "learning_rate": 9.177816012272904e-06,
      "loss": 0.5023,
      "num_input_tokens_seen": 5957512,
      "step": 9045
    },
    {
      "epoch": 5.336084905660377,
      "grad_norm": 1.9283833503723145,
      "learning_rate": 9.17640202473192e-06,
      "loss": 0.4039,
      "num_input_tokens_seen": 5960616,
      "step": 9050
    },
    {
      "epoch": 5.339033018867925,
      "grad_norm": 1.6886876821517944,
      "learning_rate": 9.1749869314705e-06,
      "loss": 0.3783,
      "num_input_tokens_seen": 5963752,
      "step": 9055
    },
    {
      "epoch": 5.341981132075472,
      "grad_norm": 1.9627089500427246,
      "learning_rate": 9.173570732863295e-06,
      "loss": 0.4014,
      "num_input_tokens_seen": 5966792,
      "step": 9060
    },
    {
      "epoch": 5.344929245283019,
      "grad_norm": 1.4818049669265747,
      "learning_rate": 9.172153429285254e-06,
      "loss": 0.516,
      "num_input_tokens_seen": 5970088,
      "step": 9065
    },
    {
      "epoch": 5.347877358490566,
      "grad_norm": 2.214493989944458,
      "learning_rate": 9.17073502111161e-06,
      "loss": 0.3529,
      "num_input_tokens_seen": 5973960,
      "step": 9070
    },
    {
      "epoch": 5.350825471698113,
      "grad_norm": 1.902611255645752,
      "learning_rate": 9.169315508717895e-06,
      "loss": 0.4071,
      "num_input_tokens_seen": 5978440,
      "step": 9075
    },
    {
      "epoch": 5.35377358490566,
      "grad_norm": 3.8544881343841553,
      "learning_rate": 9.167894892479932e-06,
      "loss": 0.5922,
      "num_input_tokens_seen": 5980744,
      "step": 9080
    },
    {
      "epoch": 5.3567216981132075,
      "grad_norm": 1.9588565826416016,
      "learning_rate": 9.16647317277383e-06,
      "loss": 0.5414,
      "num_input_tokens_seen": 5983624,
      "step": 9085
    },
    {
      "epoch": 5.359669811320755,
      "grad_norm": 1.912695050239563,
      "learning_rate": 9.165050349976002e-06,
      "loss": 0.341,
      "num_input_tokens_seen": 5986824,
      "step": 9090
    },
    {
      "epoch": 5.362617924528302,
      "grad_norm": 1.8720180988311768,
      "learning_rate": 9.16362642446314e-06,
      "loss": 0.4061,
      "num_input_tokens_seen": 5990440,
      "step": 9095
    },
    {
      "epoch": 5.365566037735849,
      "grad_norm": 7.526110649108887,
      "learning_rate": 9.162201396612242e-06,
      "loss": 0.6646,
      "num_input_tokens_seen": 5993448,
      "step": 9100
    },
    {
      "epoch": 5.368514150943396,
      "grad_norm": 3.7018039226531982,
      "learning_rate": 9.160775266800583e-06,
      "loss": 0.4242,
      "num_input_tokens_seen": 5997096,
      "step": 9105
    },
    {
      "epoch": 5.371462264150943,
      "grad_norm": 1.390397548675537,
      "learning_rate": 9.159348035405742e-06,
      "loss": 0.4271,
      "num_input_tokens_seen": 6000232,
      "step": 9110
    },
    {
      "epoch": 5.37441037735849,
      "grad_norm": 2.355253219604492,
      "learning_rate": 9.157919702805582e-06,
      "loss": 0.4448,
      "num_input_tokens_seen": 6003720,
      "step": 9115
    },
    {
      "epoch": 5.377358490566038,
      "grad_norm": 2.1486170291900635,
      "learning_rate": 9.156490269378262e-06,
      "loss": 0.4384,
      "num_input_tokens_seen": 6007304,
      "step": 9120
    },
    {
      "epoch": 5.380306603773585,
      "grad_norm": 1.3157293796539307,
      "learning_rate": 9.15505973550223e-06,
      "loss": 0.3863,
      "num_input_tokens_seen": 6010440,
      "step": 9125
    },
    {
      "epoch": 5.383254716981132,
      "grad_norm": 1.2614086866378784,
      "learning_rate": 9.153628101556223e-06,
      "loss": 0.3743,
      "num_input_tokens_seen": 6013928,
      "step": 9130
    },
    {
      "epoch": 5.3862028301886795,
      "grad_norm": 1.7989447116851807,
      "learning_rate": 9.152195367919277e-06,
      "loss": 0.4262,
      "num_input_tokens_seen": 6017384,
      "step": 9135
    },
    {
      "epoch": 5.389150943396227,
      "grad_norm": 1.456613540649414,
      "learning_rate": 9.150761534970713e-06,
      "loss": 0.453,
      "num_input_tokens_seen": 6021544,
      "step": 9140
    },
    {
      "epoch": 5.392099056603773,
      "grad_norm": 1.8125258684158325,
      "learning_rate": 9.149326603090144e-06,
      "loss": 0.3703,
      "num_input_tokens_seen": 6025064,
      "step": 9145
    },
    {
      "epoch": 5.3950471698113205,
      "grad_norm": 1.4440068006515503,
      "learning_rate": 9.147890572657471e-06,
      "loss": 0.587,
      "num_input_tokens_seen": 6028040,
      "step": 9150
    },
    {
      "epoch": 5.397995283018868,
      "grad_norm": 2.4145240783691406,
      "learning_rate": 9.146453444052895e-06,
      "loss": 0.3036,
      "num_input_tokens_seen": 6031048,
      "step": 9155
    },
    {
      "epoch": 5.400943396226415,
      "grad_norm": 2.2258565425872803,
      "learning_rate": 9.145015217656899e-06,
      "loss": 0.471,
      "num_input_tokens_seen": 6034760,
      "step": 9160
    },
    {
      "epoch": 5.403891509433962,
      "grad_norm": 5.787347793579102,
      "learning_rate": 9.14357589385026e-06,
      "loss": 0.5143,
      "num_input_tokens_seen": 6037896,
      "step": 9165
    },
    {
      "epoch": 5.40683962264151,
      "grad_norm": 5.602572917938232,
      "learning_rate": 9.142135473014046e-06,
      "loss": 0.7033,
      "num_input_tokens_seen": 6040552,
      "step": 9170
    },
    {
      "epoch": 5.409787735849057,
      "grad_norm": 2.2250638008117676,
      "learning_rate": 9.140693955529614e-06,
      "loss": 0.4087,
      "num_input_tokens_seen": 6045032,
      "step": 9175
    },
    {
      "epoch": 5.412735849056604,
      "grad_norm": 2.185267925262451,
      "learning_rate": 9.13925134177861e-06,
      "loss": 0.4681,
      "num_input_tokens_seen": 6047400,
      "step": 9180
    },
    {
      "epoch": 5.415683962264151,
      "grad_norm": 2.297818899154663,
      "learning_rate": 9.137807632142977e-06,
      "loss": 0.4281,
      "num_input_tokens_seen": 6050344,
      "step": 9185
    },
    {
      "epoch": 5.418632075471698,
      "grad_norm": 2.409761428833008,
      "learning_rate": 9.136362827004937e-06,
      "loss": 0.436,
      "num_input_tokens_seen": 6053032,
      "step": 9190
    },
    {
      "epoch": 5.421580188679245,
      "grad_norm": 4.1479363441467285,
      "learning_rate": 9.134916926747015e-06,
      "loss": 0.5363,
      "num_input_tokens_seen": 6055272,
      "step": 9195
    },
    {
      "epoch": 5.4245283018867925,
      "grad_norm": 1.6529688835144043,
      "learning_rate": 9.133469931752016e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 6057800,
      "step": 9200
    },
    {
      "epoch": 5.42747641509434,
      "grad_norm": 2.3773839473724365,
      "learning_rate": 9.132021842403035e-06,
      "loss": 0.5956,
      "num_input_tokens_seen": 6060904,
      "step": 9205
    },
    {
      "epoch": 5.430424528301887,
      "grad_norm": 1.6071664094924927,
      "learning_rate": 9.130572659083465e-06,
      "loss": 0.3809,
      "num_input_tokens_seen": 6064296,
      "step": 9210
    },
    {
      "epoch": 5.433372641509434,
      "grad_norm": 2.262141704559326,
      "learning_rate": 9.129122382176982e-06,
      "loss": 0.5282,
      "num_input_tokens_seen": 6067944,
      "step": 9215
    },
    {
      "epoch": 5.436320754716981,
      "grad_norm": 1.4777801036834717,
      "learning_rate": 9.127671012067554e-06,
      "loss": 0.403,
      "num_input_tokens_seen": 6071080,
      "step": 9220
    },
    {
      "epoch": 5.439268867924528,
      "grad_norm": 1.084755778312683,
      "learning_rate": 9.126218549139434e-06,
      "loss": 0.3707,
      "num_input_tokens_seen": 6074408,
      "step": 9225
    },
    {
      "epoch": 5.442216981132075,
      "grad_norm": 1.329906940460205,
      "learning_rate": 9.124764993777171e-06,
      "loss": 0.3522,
      "num_input_tokens_seen": 6077960,
      "step": 9230
    },
    {
      "epoch": 5.445165094339623,
      "grad_norm": 2.036468982696533,
      "learning_rate": 9.1233103463656e-06,
      "loss": 0.412,
      "num_input_tokens_seen": 6081448,
      "step": 9235
    },
    {
      "epoch": 5.44811320754717,
      "grad_norm": 2.837557792663574,
      "learning_rate": 9.121854607289842e-06,
      "loss": 0.5152,
      "num_input_tokens_seen": 6084520,
      "step": 9240
    },
    {
      "epoch": 5.451061320754717,
      "grad_norm": 2.9350335597991943,
      "learning_rate": 9.120397776935314e-06,
      "loss": 0.3838,
      "num_input_tokens_seen": 6087336,
      "step": 9245
    },
    {
      "epoch": 5.4540094339622645,
      "grad_norm": 3.4278640747070312,
      "learning_rate": 9.118939855687717e-06,
      "loss": 0.3839,
      "num_input_tokens_seen": 6089576,
      "step": 9250
    },
    {
      "epoch": 5.456957547169811,
      "grad_norm": 1.5057499408721924,
      "learning_rate": 9.117480843933043e-06,
      "loss": 0.3567,
      "num_input_tokens_seen": 6092392,
      "step": 9255
    },
    {
      "epoch": 5.459905660377358,
      "grad_norm": 2.1233363151550293,
      "learning_rate": 9.116020742057567e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 6095336,
      "step": 9260
    },
    {
      "epoch": 5.462853773584905,
      "grad_norm": 3.783977746963501,
      "learning_rate": 9.114559550447863e-06,
      "loss": 0.6118,
      "num_input_tokens_seen": 6098056,
      "step": 9265
    },
    {
      "epoch": 5.465801886792453,
      "grad_norm": 2.538018226623535,
      "learning_rate": 9.113097269490784e-06,
      "loss": 0.4635,
      "num_input_tokens_seen": 6101160,
      "step": 9270
    },
    {
      "epoch": 5.46875,
      "grad_norm": 1.6230403184890747,
      "learning_rate": 9.111633899573476e-06,
      "loss": 0.5018,
      "num_input_tokens_seen": 6104744,
      "step": 9275
    },
    {
      "epoch": 5.471698113207547,
      "grad_norm": 2.0708417892456055,
      "learning_rate": 9.110169441083374e-06,
      "loss": 0.5002,
      "num_input_tokens_seen": 6108520,
      "step": 9280
    },
    {
      "epoch": 5.474646226415095,
      "grad_norm": 2.9829742908477783,
      "learning_rate": 9.108703894408198e-06,
      "loss": 0.421,
      "num_input_tokens_seen": 6111592,
      "step": 9285
    },
    {
      "epoch": 5.477594339622642,
      "grad_norm": 1.5134027004241943,
      "learning_rate": 9.107237259935959e-06,
      "loss": 0.6988,
      "num_input_tokens_seen": 6115432,
      "step": 9290
    },
    {
      "epoch": 5.480542452830189,
      "grad_norm": 1.4811124801635742,
      "learning_rate": 9.105769538054954e-06,
      "loss": 0.5272,
      "num_input_tokens_seen": 6119176,
      "step": 9295
    },
    {
      "epoch": 5.4834905660377355,
      "grad_norm": 1.3423231840133667,
      "learning_rate": 9.104300729153768e-06,
      "loss": 0.4106,
      "num_input_tokens_seen": 6122504,
      "step": 9300
    },
    {
      "epoch": 5.486438679245283,
      "grad_norm": 1.5402344465255737,
      "learning_rate": 9.102830833621277e-06,
      "loss": 0.5403,
      "num_input_tokens_seen": 6126088,
      "step": 9305
    },
    {
      "epoch": 5.48938679245283,
      "grad_norm": 1.8265899419784546,
      "learning_rate": 9.101359851846639e-06,
      "loss": 0.4298,
      "num_input_tokens_seen": 6128872,
      "step": 9310
    },
    {
      "epoch": 5.492334905660377,
      "grad_norm": 1.934721827507019,
      "learning_rate": 9.099887784219305e-06,
      "loss": 0.4689,
      "num_input_tokens_seen": 6132296,
      "step": 9315
    },
    {
      "epoch": 5.495283018867925,
      "grad_norm": 3.116152763366699,
      "learning_rate": 9.098414631129012e-06,
      "loss": 0.6609,
      "num_input_tokens_seen": 6134632,
      "step": 9320
    },
    {
      "epoch": 5.498231132075472,
      "grad_norm": 1.8180128335952759,
      "learning_rate": 9.09694039296578e-06,
      "loss": 0.5998,
      "num_input_tokens_seen": 6137800,
      "step": 9325
    },
    {
      "epoch": 5.501179245283019,
      "grad_norm": 1.4149829149246216,
      "learning_rate": 9.095465070119924e-06,
      "loss": 0.4197,
      "num_input_tokens_seen": 6140680,
      "step": 9330
    },
    {
      "epoch": 5.504127358490566,
      "grad_norm": 1.4578166007995605,
      "learning_rate": 9.09398866298204e-06,
      "loss": 0.5472,
      "num_input_tokens_seen": 6145096,
      "step": 9335
    },
    {
      "epoch": 5.507075471698113,
      "grad_norm": 1.4901857376098633,
      "learning_rate": 9.09251117194301e-06,
      "loss": 0.3219,
      "num_input_tokens_seen": 6148776,
      "step": 9340
    },
    {
      "epoch": 5.51002358490566,
      "grad_norm": 1.7039042711257935,
      "learning_rate": 9.091032597394012e-06,
      "loss": 0.4294,
      "num_input_tokens_seen": 6152040,
      "step": 9345
    },
    {
      "epoch": 5.5129716981132075,
      "grad_norm": 2.6316463947296143,
      "learning_rate": 9.089552939726503e-06,
      "loss": 0.4892,
      "num_input_tokens_seen": 6154856,
      "step": 9350
    },
    {
      "epoch": 5.515919811320755,
      "grad_norm": 2.805990219116211,
      "learning_rate": 9.088072199332227e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 6157736,
      "step": 9355
    },
    {
      "epoch": 5.518867924528302,
      "grad_norm": 2.2756097316741943,
      "learning_rate": 9.08659037660322e-06,
      "loss": 0.479,
      "num_input_tokens_seen": 6160392,
      "step": 9360
    },
    {
      "epoch": 5.521816037735849,
      "grad_norm": 1.918864369392395,
      "learning_rate": 9.085107471931797e-06,
      "loss": 0.3525,
      "num_input_tokens_seen": 6163624,
      "step": 9365
    },
    {
      "epoch": 5.524764150943396,
      "grad_norm": 1.4579954147338867,
      "learning_rate": 9.083623485710564e-06,
      "loss": 0.4945,
      "num_input_tokens_seen": 6167400,
      "step": 9370
    },
    {
      "epoch": 5.527712264150943,
      "grad_norm": 2.350581407546997,
      "learning_rate": 9.082138418332416e-06,
      "loss": 0.5094,
      "num_input_tokens_seen": 6171080,
      "step": 9375
    },
    {
      "epoch": 5.53066037735849,
      "grad_norm": 2.048426866531372,
      "learning_rate": 9.080652270190527e-06,
      "loss": 0.5392,
      "num_input_tokens_seen": 6173768,
      "step": 9380
    },
    {
      "epoch": 5.533608490566038,
      "grad_norm": 2.2405917644500732,
      "learning_rate": 9.079165041678363e-06,
      "loss": 0.5266,
      "num_input_tokens_seen": 6177000,
      "step": 9385
    },
    {
      "epoch": 5.536556603773585,
      "grad_norm": 1.9599385261535645,
      "learning_rate": 9.077676733189675e-06,
      "loss": 0.5089,
      "num_input_tokens_seen": 6180008,
      "step": 9390
    },
    {
      "epoch": 5.539504716981132,
      "grad_norm": 3.178906202316284,
      "learning_rate": 9.076187345118496e-06,
      "loss": 0.3749,
      "num_input_tokens_seen": 6182536,
      "step": 9395
    },
    {
      "epoch": 5.5424528301886795,
      "grad_norm": 2.6668336391448975,
      "learning_rate": 9.074696877859152e-06,
      "loss": 0.5415,
      "num_input_tokens_seen": 6185192,
      "step": 9400
    },
    {
      "epoch": 5.545400943396227,
      "grad_norm": 2.123929262161255,
      "learning_rate": 9.073205331806248e-06,
      "loss": 0.4007,
      "num_input_tokens_seen": 6188616,
      "step": 9405
    },
    {
      "epoch": 5.548349056603773,
      "grad_norm": 1.5857806205749512,
      "learning_rate": 9.071712707354676e-06,
      "loss": 0.5834,
      "num_input_tokens_seen": 6191656,
      "step": 9410
    },
    {
      "epoch": 5.5512971698113205,
      "grad_norm": 2.7303593158721924,
      "learning_rate": 9.070219004899618e-06,
      "loss": 0.3858,
      "num_input_tokens_seen": 6194792,
      "step": 9415
    },
    {
      "epoch": 5.554245283018868,
      "grad_norm": 2.4032747745513916,
      "learning_rate": 9.068724224836538e-06,
      "loss": 0.4335,
      "num_input_tokens_seen": 6199976,
      "step": 9420
    },
    {
      "epoch": 5.557193396226415,
      "grad_norm": 4.660754680633545,
      "learning_rate": 9.067228367561182e-06,
      "loss": 0.5576,
      "num_input_tokens_seen": 6202600,
      "step": 9425
    },
    {
      "epoch": 5.560141509433962,
      "grad_norm": 1.298729419708252,
      "learning_rate": 9.06573143346959e-06,
      "loss": 0.3729,
      "num_input_tokens_seen": 6205192,
      "step": 9430
    },
    {
      "epoch": 5.56308962264151,
      "grad_norm": 1.602677583694458,
      "learning_rate": 9.064233422958078e-06,
      "loss": 0.4612,
      "num_input_tokens_seen": 6208008,
      "step": 9435
    },
    {
      "epoch": 5.566037735849057,
      "grad_norm": 1.4287152290344238,
      "learning_rate": 9.062734336423248e-06,
      "loss": 0.4939,
      "num_input_tokens_seen": 6211688,
      "step": 9440
    },
    {
      "epoch": 5.568985849056604,
      "grad_norm": 2.2085180282592773,
      "learning_rate": 9.061234174261998e-06,
      "loss": 0.4234,
      "num_input_tokens_seen": 6214920,
      "step": 9445
    },
    {
      "epoch": 5.571933962264151,
      "grad_norm": 2.347740888595581,
      "learning_rate": 9.059732936871493e-06,
      "loss": 0.4944,
      "num_input_tokens_seen": 6218088,
      "step": 9450
    },
    {
      "epoch": 5.574882075471698,
      "grad_norm": 3.3669748306274414,
      "learning_rate": 9.058230624649198e-06,
      "loss": 0.4318,
      "num_input_tokens_seen": 6220808,
      "step": 9455
    },
    {
      "epoch": 5.577830188679245,
      "grad_norm": 2.2796716690063477,
      "learning_rate": 9.056727237992856e-06,
      "loss": 0.4201,
      "num_input_tokens_seen": 6225160,
      "step": 9460
    },
    {
      "epoch": 5.5807783018867925,
      "grad_norm": 2.045727252960205,
      "learning_rate": 9.055222777300493e-06,
      "loss": 0.437,
      "num_input_tokens_seen": 6228488,
      "step": 9465
    },
    {
      "epoch": 5.58372641509434,
      "grad_norm": 1.3146079778671265,
      "learning_rate": 9.053717242970423e-06,
      "loss": 0.3671,
      "num_input_tokens_seen": 6231848,
      "step": 9470
    },
    {
      "epoch": 5.586674528301887,
      "grad_norm": 2.1382973194122314,
      "learning_rate": 9.052210635401244e-06,
      "loss": 0.3815,
      "num_input_tokens_seen": 6235336,
      "step": 9475
    },
    {
      "epoch": 5.589622641509434,
      "grad_norm": 1.8254072666168213,
      "learning_rate": 9.050702954991833e-06,
      "loss": 0.4658,
      "num_input_tokens_seen": 6238696,
      "step": 9480
    },
    {
      "epoch": 5.592570754716981,
      "grad_norm": 2.0766518115997314,
      "learning_rate": 9.049194202141358e-06,
      "loss": 0.5575,
      "num_input_tokens_seen": 6241544,
      "step": 9485
    },
    {
      "epoch": 5.595518867924528,
      "grad_norm": 3.5932776927948,
      "learning_rate": 9.047684377249267e-06,
      "loss": 0.5996,
      "num_input_tokens_seen": 6244520,
      "step": 9490
    },
    {
      "epoch": 5.598466981132075,
      "grad_norm": 1.550055980682373,
      "learning_rate": 9.046173480715292e-06,
      "loss": 0.4497,
      "num_input_tokens_seen": 6248328,
      "step": 9495
    },
    {
      "epoch": 5.601415094339623,
      "grad_norm": 0.9342994689941406,
      "learning_rate": 9.044661512939451e-06,
      "loss": 0.5418,
      "num_input_tokens_seen": 6252392,
      "step": 9500
    },
    {
      "epoch": 5.60436320754717,
      "grad_norm": 2.579918622970581,
      "learning_rate": 9.043148474322043e-06,
      "loss": 0.397,
      "num_input_tokens_seen": 6255336,
      "step": 9505
    },
    {
      "epoch": 5.607311320754717,
      "grad_norm": 2.2248852252960205,
      "learning_rate": 9.04163436526365e-06,
      "loss": 0.374,
      "num_input_tokens_seen": 6258088,
      "step": 9510
    },
    {
      "epoch": 5.6102594339622645,
      "grad_norm": 2.27953839302063,
      "learning_rate": 9.040119186165142e-06,
      "loss": 0.4171,
      "num_input_tokens_seen": 6261288,
      "step": 9515
    },
    {
      "epoch": 5.613207547169811,
      "grad_norm": 2.300102710723877,
      "learning_rate": 9.038602937427665e-06,
      "loss": 0.5387,
      "num_input_tokens_seen": 6264520,
      "step": 9520
    },
    {
      "epoch": 5.616155660377358,
      "grad_norm": 2.4904747009277344,
      "learning_rate": 9.037085619452658e-06,
      "loss": 0.5055,
      "num_input_tokens_seen": 6267272,
      "step": 9525
    },
    {
      "epoch": 5.619103773584905,
      "grad_norm": 2.3131744861602783,
      "learning_rate": 9.035567232641833e-06,
      "loss": 0.4539,
      "num_input_tokens_seen": 6269832,
      "step": 9530
    },
    {
      "epoch": 5.622051886792453,
      "grad_norm": 2.0961525440216064,
      "learning_rate": 9.03404777739719e-06,
      "loss": 0.4009,
      "num_input_tokens_seen": 6272808,
      "step": 9535
    },
    {
      "epoch": 5.625,
      "grad_norm": 1.7636992931365967,
      "learning_rate": 9.032527254121013e-06,
      "loss": 0.343,
      "num_input_tokens_seen": 6276104,
      "step": 9540
    },
    {
      "epoch": 5.627948113207547,
      "grad_norm": 1.0846927165985107,
      "learning_rate": 9.031005663215867e-06,
      "loss": 0.437,
      "num_input_tokens_seen": 6279368,
      "step": 9545
    },
    {
      "epoch": 5.630896226415095,
      "grad_norm": 2.041990280151367,
      "learning_rate": 9.029483005084595e-06,
      "loss": 0.4777,
      "num_input_tokens_seen": 6283304,
      "step": 9550
    },
    {
      "epoch": 5.633844339622642,
      "grad_norm": 2.2044122219085693,
      "learning_rate": 9.027959280130337e-06,
      "loss": 0.5354,
      "num_input_tokens_seen": 6288328,
      "step": 9555
    },
    {
      "epoch": 5.636792452830189,
      "grad_norm": 2.1341066360473633,
      "learning_rate": 9.026434488756496e-06,
      "loss": 0.4689,
      "num_input_tokens_seen": 6291112,
      "step": 9560
    },
    {
      "epoch": 5.6397405660377355,
      "grad_norm": 1.046785831451416,
      "learning_rate": 9.024908631366774e-06,
      "loss": 0.5872,
      "num_input_tokens_seen": 6293864,
      "step": 9565
    },
    {
      "epoch": 5.642688679245283,
      "grad_norm": 2.6312742233276367,
      "learning_rate": 9.023381708365143e-06,
      "loss": 0.4576,
      "num_input_tokens_seen": 6297160,
      "step": 9570
    },
    {
      "epoch": 5.64563679245283,
      "grad_norm": 1.886027455329895,
      "learning_rate": 9.021853720155866e-06,
      "loss": 0.3747,
      "num_input_tokens_seen": 6299848,
      "step": 9575
    },
    {
      "epoch": 5.648584905660377,
      "grad_norm": 2.11794376373291,
      "learning_rate": 9.020324667143483e-06,
      "loss": 0.5638,
      "num_input_tokens_seen": 6302888,
      "step": 9580
    },
    {
      "epoch": 5.651533018867925,
      "grad_norm": 1.6454652547836304,
      "learning_rate": 9.018794549732819e-06,
      "loss": 0.4595,
      "num_input_tokens_seen": 6305960,
      "step": 9585
    },
    {
      "epoch": 5.654481132075472,
      "grad_norm": 1.8292570114135742,
      "learning_rate": 9.017263368328977e-06,
      "loss": 0.5312,
      "num_input_tokens_seen": 6309064,
      "step": 9590
    },
    {
      "epoch": 5.657429245283019,
      "grad_norm": 2.256633758544922,
      "learning_rate": 9.015731123337344e-06,
      "loss": 0.3853,
      "num_input_tokens_seen": 6311752,
      "step": 9595
    },
    {
      "epoch": 5.660377358490566,
      "grad_norm": 1.290195107460022,
      "learning_rate": 9.01419781516359e-06,
      "loss": 0.4249,
      "num_input_tokens_seen": 6314792,
      "step": 9600
    },
    {
      "epoch": 5.663325471698113,
      "grad_norm": 2.4256012439727783,
      "learning_rate": 9.012663444213664e-06,
      "loss": 0.5654,
      "num_input_tokens_seen": 6317640,
      "step": 9605
    },
    {
      "epoch": 5.66627358490566,
      "grad_norm": 2.887686014175415,
      "learning_rate": 9.011128010893797e-06,
      "loss": 0.6115,
      "num_input_tokens_seen": 6324680,
      "step": 9610
    },
    {
      "epoch": 5.6692216981132075,
      "grad_norm": 2.0892698764801025,
      "learning_rate": 9.009591515610503e-06,
      "loss": 0.4357,
      "num_input_tokens_seen": 6330152,
      "step": 9615
    },
    {
      "epoch": 5.672169811320755,
      "grad_norm": 1.2122193574905396,
      "learning_rate": 9.008053958770575e-06,
      "loss": 0.3721,
      "num_input_tokens_seen": 6333192,
      "step": 9620
    },
    {
      "epoch": 5.675117924528302,
      "grad_norm": 2.2959649562835693,
      "learning_rate": 9.006515340781087e-06,
      "loss": 0.4385,
      "num_input_tokens_seen": 6336520,
      "step": 9625
    },
    {
      "epoch": 5.678066037735849,
      "grad_norm": 1.3278429508209229,
      "learning_rate": 9.004975662049396e-06,
      "loss": 0.5003,
      "num_input_tokens_seen": 6340584,
      "step": 9630
    },
    {
      "epoch": 5.681014150943396,
      "grad_norm": 1.249680995941162,
      "learning_rate": 9.003434922983138e-06,
      "loss": 0.477,
      "num_input_tokens_seen": 6344936,
      "step": 9635
    },
    {
      "epoch": 5.683962264150943,
      "grad_norm": 1.5217032432556152,
      "learning_rate": 9.00189312399023e-06,
      "loss": 0.5499,
      "num_input_tokens_seen": 6348520,
      "step": 9640
    },
    {
      "epoch": 5.68691037735849,
      "grad_norm": 1.6650720834732056,
      "learning_rate": 9.00035026547887e-06,
      "loss": 0.5078,
      "num_input_tokens_seen": 6352296,
      "step": 9645
    },
    {
      "epoch": 5.689858490566038,
      "grad_norm": 2.293403387069702,
      "learning_rate": 8.998806347857537e-06,
      "loss": 0.5538,
      "num_input_tokens_seen": 6355464,
      "step": 9650
    },
    {
      "epoch": 5.692806603773585,
      "grad_norm": 1.5671502351760864,
      "learning_rate": 8.99726137153499e-06,
      "loss": 0.4835,
      "num_input_tokens_seen": 6358472,
      "step": 9655
    },
    {
      "epoch": 5.695754716981132,
      "grad_norm": 1.4876998662948608,
      "learning_rate": 8.995715336920266e-06,
      "loss": 0.5481,
      "num_input_tokens_seen": 6364680,
      "step": 9660
    },
    {
      "epoch": 5.6987028301886795,
      "grad_norm": 1.659095048904419,
      "learning_rate": 8.994168244422687e-06,
      "loss": 0.4963,
      "num_input_tokens_seen": 6369256,
      "step": 9665
    },
    {
      "epoch": 5.701650943396227,
      "grad_norm": 1.419141173362732,
      "learning_rate": 8.992620094451852e-06,
      "loss": 0.3547,
      "num_input_tokens_seen": 6372712,
      "step": 9670
    },
    {
      "epoch": 5.704599056603773,
      "grad_norm": 1.6844631433486938,
      "learning_rate": 8.991070887417639e-06,
      "loss": 0.3929,
      "num_input_tokens_seen": 6376616,
      "step": 9675
    },
    {
      "epoch": 5.7075471698113205,
      "grad_norm": 5.064459323883057,
      "learning_rate": 8.989520623730208e-06,
      "loss": 0.4064,
      "num_input_tokens_seen": 6379752,
      "step": 9680
    },
    {
      "epoch": 5.710495283018868,
      "grad_norm": 1.9799227714538574,
      "learning_rate": 8.987969303799998e-06,
      "loss": 0.4259,
      "num_input_tokens_seen": 6383304,
      "step": 9685
    },
    {
      "epoch": 5.713443396226415,
      "grad_norm": 2.029583692550659,
      "learning_rate": 8.986416928037728e-06,
      "loss": 0.4262,
      "num_input_tokens_seen": 6387336,
      "step": 9690
    },
    {
      "epoch": 5.716391509433962,
      "grad_norm": 1.2498348951339722,
      "learning_rate": 8.984863496854395e-06,
      "loss": 0.4522,
      "num_input_tokens_seen": 6390696,
      "step": 9695
    },
    {
      "epoch": 5.71933962264151,
      "grad_norm": 1.2268298864364624,
      "learning_rate": 8.983309010661279e-06,
      "loss": 0.4524,
      "num_input_tokens_seen": 6394504,
      "step": 9700
    },
    {
      "epoch": 5.722287735849057,
      "grad_norm": 2.671170949935913,
      "learning_rate": 8.981753469869934e-06,
      "loss": 0.4784,
      "num_input_tokens_seen": 6397512,
      "step": 9705
    },
    {
      "epoch": 5.725235849056604,
      "grad_norm": 0.9403424859046936,
      "learning_rate": 8.980196874892198e-06,
      "loss": 0.3025,
      "num_input_tokens_seen": 6401384,
      "step": 9710
    },
    {
      "epoch": 5.728183962264151,
      "grad_norm": 2.5676939487457275,
      "learning_rate": 8.978639226140184e-06,
      "loss": 0.4736,
      "num_input_tokens_seen": 6405384,
      "step": 9715
    },
    {
      "epoch": 5.731132075471698,
      "grad_norm": 1.5092235803604126,
      "learning_rate": 8.977080524026289e-06,
      "loss": 0.425,
      "num_input_tokens_seen": 6408488,
      "step": 9720
    },
    {
      "epoch": 5.734080188679245,
      "grad_norm": 3.4226386547088623,
      "learning_rate": 8.975520768963186e-06,
      "loss": 0.5438,
      "num_input_tokens_seen": 6411944,
      "step": 9725
    },
    {
      "epoch": 5.7370283018867925,
      "grad_norm": 1.895229697227478,
      "learning_rate": 8.973959961363825e-06,
      "loss": 0.4074,
      "num_input_tokens_seen": 6415048,
      "step": 9730
    },
    {
      "epoch": 5.73997641509434,
      "grad_norm": 1.9706082344055176,
      "learning_rate": 8.972398101641438e-06,
      "loss": 0.3532,
      "num_input_tokens_seen": 6417704,
      "step": 9735
    },
    {
      "epoch": 5.742924528301887,
      "grad_norm": 1.358763575553894,
      "learning_rate": 8.970835190209532e-06,
      "loss": 0.367,
      "num_input_tokens_seen": 6422344,
      "step": 9740
    },
    {
      "epoch": 5.745872641509434,
      "grad_norm": 2.2625620365142822,
      "learning_rate": 8.969271227481899e-06,
      "loss": 0.5322,
      "num_input_tokens_seen": 6425736,
      "step": 9745
    },
    {
      "epoch": 5.748820754716981,
      "grad_norm": 1.3481266498565674,
      "learning_rate": 8.967706213872599e-06,
      "loss": 0.472,
      "num_input_tokens_seen": 6428552,
      "step": 9750
    },
    {
      "epoch": 5.751768867924528,
      "grad_norm": 1.3100154399871826,
      "learning_rate": 8.966140149795981e-06,
      "loss": 0.4041,
      "num_input_tokens_seen": 6432392,
      "step": 9755
    },
    {
      "epoch": 5.754716981132075,
      "grad_norm": 1.7328647375106812,
      "learning_rate": 8.964573035666663e-06,
      "loss": 0.386,
      "num_input_tokens_seen": 6438248,
      "step": 9760
    },
    {
      "epoch": 5.757665094339623,
      "grad_norm": 2.2816624641418457,
      "learning_rate": 8.96300487189955e-06,
      "loss": 0.3704,
      "num_input_tokens_seen": 6441000,
      "step": 9765
    },
    {
      "epoch": 5.76061320754717,
      "grad_norm": 3.454463005065918,
      "learning_rate": 8.961435658909816e-06,
      "loss": 0.4554,
      "num_input_tokens_seen": 6443656,
      "step": 9770
    },
    {
      "epoch": 5.763561320754717,
      "grad_norm": 1.4408000707626343,
      "learning_rate": 8.959865397112918e-06,
      "loss": 0.4212,
      "num_input_tokens_seen": 6446056,
      "step": 9775
    },
    {
      "epoch": 5.7665094339622645,
      "grad_norm": 2.200247049331665,
      "learning_rate": 8.95829408692459e-06,
      "loss": 0.4737,
      "num_input_tokens_seen": 6449224,
      "step": 9780
    },
    {
      "epoch": 5.769457547169811,
      "grad_norm": 1.3835819959640503,
      "learning_rate": 8.956721728760845e-06,
      "loss": 0.4873,
      "num_input_tokens_seen": 6452552,
      "step": 9785
    },
    {
      "epoch": 5.772405660377358,
      "grad_norm": 2.1205122470855713,
      "learning_rate": 8.95514832303797e-06,
      "loss": 0.3992,
      "num_input_tokens_seen": 6455176,
      "step": 9790
    },
    {
      "epoch": 5.775353773584905,
      "grad_norm": 4.64229679107666,
      "learning_rate": 8.953573870172528e-06,
      "loss": 0.5534,
      "num_input_tokens_seen": 6458504,
      "step": 9795
    },
    {
      "epoch": 5.778301886792453,
      "grad_norm": 2.184582233428955,
      "learning_rate": 8.951998370581368e-06,
      "loss": 0.4223,
      "num_input_tokens_seen": 6461512,
      "step": 9800
    },
    {
      "epoch": 5.78125,
      "grad_norm": 1.722968339920044,
      "learning_rate": 8.950421824681605e-06,
      "loss": 0.3319,
      "num_input_tokens_seen": 6468552,
      "step": 9805
    },
    {
      "epoch": 5.784198113207547,
      "grad_norm": 1.565529465675354,
      "learning_rate": 8.948844232890638e-06,
      "loss": 0.4411,
      "num_input_tokens_seen": 6471688,
      "step": 9810
    },
    {
      "epoch": 5.787146226415095,
      "grad_norm": 1.9285593032836914,
      "learning_rate": 8.947265595626144e-06,
      "loss": 0.361,
      "num_input_tokens_seen": 6474856,
      "step": 9815
    },
    {
      "epoch": 5.790094339622642,
      "grad_norm": 0.9254405498504639,
      "learning_rate": 8.945685913306071e-06,
      "loss": 0.3831,
      "num_input_tokens_seen": 6477864,
      "step": 9820
    },
    {
      "epoch": 5.793042452830189,
      "grad_norm": 1.5471876859664917,
      "learning_rate": 8.944105186348646e-06,
      "loss": 0.4454,
      "num_input_tokens_seen": 6480488,
      "step": 9825
    },
    {
      "epoch": 5.7959905660377355,
      "grad_norm": 3.844320297241211,
      "learning_rate": 8.942523415172377e-06,
      "loss": 0.5029,
      "num_input_tokens_seen": 6483432,
      "step": 9830
    },
    {
      "epoch": 5.798938679245283,
      "grad_norm": 1.3756054639816284,
      "learning_rate": 8.94094060019604e-06,
      "loss": 0.3422,
      "num_input_tokens_seen": 6486696,
      "step": 9835
    },
    {
      "epoch": 5.80188679245283,
      "grad_norm": 3.35147762298584,
      "learning_rate": 8.939356741838696e-06,
      "loss": 0.6325,
      "num_input_tokens_seen": 6489352,
      "step": 9840
    },
    {
      "epoch": 5.804834905660377,
      "grad_norm": 1.912316083908081,
      "learning_rate": 8.937771840519677e-06,
      "loss": 0.2936,
      "num_input_tokens_seen": 6493192,
      "step": 9845
    },
    {
      "epoch": 5.807783018867925,
      "grad_norm": 2.5798542499542236,
      "learning_rate": 8.936185896658593e-06,
      "loss": 0.5337,
      "num_input_tokens_seen": 6496680,
      "step": 9850
    },
    {
      "epoch": 5.810731132075472,
      "grad_norm": 1.7604142427444458,
      "learning_rate": 8.934598910675329e-06,
      "loss": 0.366,
      "num_input_tokens_seen": 6500200,
      "step": 9855
    },
    {
      "epoch": 5.813679245283019,
      "grad_norm": 1.2059096097946167,
      "learning_rate": 8.933010882990044e-06,
      "loss": 0.3677,
      "num_input_tokens_seen": 6503464,
      "step": 9860
    },
    {
      "epoch": 5.816627358490566,
      "grad_norm": 1.7108402252197266,
      "learning_rate": 8.93142181402318e-06,
      "loss": 0.4401,
      "num_input_tokens_seen": 6507944,
      "step": 9865
    },
    {
      "epoch": 5.819575471698113,
      "grad_norm": 2.066962480545044,
      "learning_rate": 8.929831704195445e-06,
      "loss": 0.4738,
      "num_input_tokens_seen": 6510536,
      "step": 9870
    },
    {
      "epoch": 5.82252358490566,
      "grad_norm": 1.4766886234283447,
      "learning_rate": 8.928240553927831e-06,
      "loss": 0.4499,
      "num_input_tokens_seen": 6513928,
      "step": 9875
    },
    {
      "epoch": 5.8254716981132075,
      "grad_norm": 2.1442670822143555,
      "learning_rate": 8.926648363641602e-06,
      "loss": 0.4003,
      "num_input_tokens_seen": 6516648,
      "step": 9880
    },
    {
      "epoch": 5.828419811320755,
      "grad_norm": 2.33748197555542,
      "learning_rate": 8.925055133758294e-06,
      "loss": 0.4671,
      "num_input_tokens_seen": 6520488,
      "step": 9885
    },
    {
      "epoch": 5.831367924528302,
      "grad_norm": 2.069326639175415,
      "learning_rate": 8.923460864699723e-06,
      "loss": 0.448,
      "num_input_tokens_seen": 6524008,
      "step": 9890
    },
    {
      "epoch": 5.834316037735849,
      "grad_norm": 0.9562781453132629,
      "learning_rate": 8.921865556887979e-06,
      "loss": 0.3078,
      "num_input_tokens_seen": 6527496,
      "step": 9895
    },
    {
      "epoch": 5.837264150943396,
      "grad_norm": 1.9253634214401245,
      "learning_rate": 8.920269210745426e-06,
      "loss": 0.5449,
      "num_input_tokens_seen": 6531176,
      "step": 9900
    },
    {
      "epoch": 5.840212264150943,
      "grad_norm": 1.3615758419036865,
      "learning_rate": 8.918671826694704e-06,
      "loss": 0.3596,
      "num_input_tokens_seen": 6534504,
      "step": 9905
    },
    {
      "epoch": 5.84316037735849,
      "grad_norm": 3.1889381408691406,
      "learning_rate": 8.91707340515873e-06,
      "loss": 0.5187,
      "num_input_tokens_seen": 6538216,
      "step": 9910
    },
    {
      "epoch": 5.846108490566038,
      "grad_norm": 3.0599591732025146,
      "learning_rate": 8.915473946560688e-06,
      "loss": 0.4976,
      "num_input_tokens_seen": 6541288,
      "step": 9915
    },
    {
      "epoch": 5.849056603773585,
      "grad_norm": 2.2725131511688232,
      "learning_rate": 8.913873451324044e-06,
      "loss": 0.501,
      "num_input_tokens_seen": 6545224,
      "step": 9920
    },
    {
      "epoch": 5.852004716981132,
      "grad_norm": 4.73910665512085,
      "learning_rate": 8.912271919872538e-06,
      "loss": 0.4198,
      "num_input_tokens_seen": 6549896,
      "step": 9925
    },
    {
      "epoch": 5.8549528301886795,
      "grad_norm": 2.3406484127044678,
      "learning_rate": 8.910669352630176e-06,
      "loss": 0.4207,
      "num_input_tokens_seen": 6552872,
      "step": 9930
    },
    {
      "epoch": 5.857900943396227,
      "grad_norm": 1.6379930973052979,
      "learning_rate": 8.909065750021253e-06,
      "loss": 0.3752,
      "num_input_tokens_seen": 6555752,
      "step": 9935
    },
    {
      "epoch": 5.860849056603773,
      "grad_norm": 1.265820026397705,
      "learning_rate": 8.907461112470323e-06,
      "loss": 0.4117,
      "num_input_tokens_seen": 6559496,
      "step": 9940
    },
    {
      "epoch": 5.8637971698113205,
      "grad_norm": 1.661448359489441,
      "learning_rate": 8.905855440402225e-06,
      "loss": 0.3881,
      "num_input_tokens_seen": 6561768,
      "step": 9945
    },
    {
      "epoch": 5.866745283018868,
      "grad_norm": 1.7226784229278564,
      "learning_rate": 8.904248734242065e-06,
      "loss": 0.3801,
      "num_input_tokens_seen": 6565416,
      "step": 9950
    },
    {
      "epoch": 5.869693396226415,
      "grad_norm": 1.5241148471832275,
      "learning_rate": 8.902640994415226e-06,
      "loss": 0.4401,
      "num_input_tokens_seen": 6568584,
      "step": 9955
    },
    {
      "epoch": 5.872641509433962,
      "grad_norm": 1.7942253351211548,
      "learning_rate": 8.901032221347364e-06,
      "loss": 0.3571,
      "num_input_tokens_seen": 6572776,
      "step": 9960
    },
    {
      "epoch": 5.87558962264151,
      "grad_norm": 3.457502603530884,
      "learning_rate": 8.899422415464409e-06,
      "loss": 0.5938,
      "num_input_tokens_seen": 6575368,
      "step": 9965
    },
    {
      "epoch": 5.878537735849057,
      "grad_norm": 0.7188454866409302,
      "learning_rate": 8.897811577192565e-06,
      "loss": 0.3076,
      "num_input_tokens_seen": 6579688,
      "step": 9970
    },
    {
      "epoch": 5.881485849056604,
      "grad_norm": 1.4833704233169556,
      "learning_rate": 8.896199706958306e-06,
      "loss": 0.4899,
      "num_input_tokens_seen": 6583368,
      "step": 9975
    },
    {
      "epoch": 5.884433962264151,
      "grad_norm": 2.5317585468292236,
      "learning_rate": 8.894586805188384e-06,
      "loss": 0.4177,
      "num_input_tokens_seen": 6587336,
      "step": 9980
    },
    {
      "epoch": 5.887382075471698,
      "grad_norm": 3.1594760417938232,
      "learning_rate": 8.892972872309821e-06,
      "loss": 0.4969,
      "num_input_tokens_seen": 6591112,
      "step": 9985
    },
    {
      "epoch": 5.890330188679245,
      "grad_norm": 2.294553756713867,
      "learning_rate": 8.89135790874991e-06,
      "loss": 0.5387,
      "num_input_tokens_seen": 6594760,
      "step": 9990
    },
    {
      "epoch": 5.8932783018867925,
      "grad_norm": 2.795565605163574,
      "learning_rate": 8.889741914936224e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 6597576,
      "step": 9995
    },
    {
      "epoch": 5.89622641509434,
      "grad_norm": 2.4287197589874268,
      "learning_rate": 8.888124891296602e-06,
      "loss": 0.3631,
      "num_input_tokens_seen": 6600776,
      "step": 10000
    },
    {
      "epoch": 5.899174528301887,
      "grad_norm": 4.253491401672363,
      "learning_rate": 8.886506838259156e-06,
      "loss": 0.4111,
      "num_input_tokens_seen": 6603816,
      "step": 10005
    },
    {
      "epoch": 5.902122641509434,
      "grad_norm": 1.4540841579437256,
      "learning_rate": 8.884887756252279e-06,
      "loss": 0.3319,
      "num_input_tokens_seen": 6608040,
      "step": 10010
    },
    {
      "epoch": 5.905070754716981,
      "grad_norm": 1.4505618810653687,
      "learning_rate": 8.88326764570462e-06,
      "loss": 0.4116,
      "num_input_tokens_seen": 6611144,
      "step": 10015
    },
    {
      "epoch": 5.908018867924528,
      "grad_norm": 1.4506394863128662,
      "learning_rate": 8.88164650704512e-06,
      "loss": 0.5005,
      "num_input_tokens_seen": 6614408,
      "step": 10020
    },
    {
      "epoch": 5.910966981132075,
      "grad_norm": 3.147348642349243,
      "learning_rate": 8.880024340702978e-06,
      "loss": 0.4707,
      "num_input_tokens_seen": 6617864,
      "step": 10025
    },
    {
      "epoch": 5.913915094339623,
      "grad_norm": 1.645900011062622,
      "learning_rate": 8.878401147107667e-06,
      "loss": 0.3569,
      "num_input_tokens_seen": 6620200,
      "step": 10030
    },
    {
      "epoch": 5.91686320754717,
      "grad_norm": 2.159339427947998,
      "learning_rate": 8.87677692668894e-06,
      "loss": 0.2809,
      "num_input_tokens_seen": 6623240,
      "step": 10035
    },
    {
      "epoch": 5.919811320754717,
      "grad_norm": 1.8767893314361572,
      "learning_rate": 8.875151679876813e-06,
      "loss": 0.445,
      "num_input_tokens_seen": 6626760,
      "step": 10040
    },
    {
      "epoch": 5.9227594339622645,
      "grad_norm": 1.674679160118103,
      "learning_rate": 8.873525407101577e-06,
      "loss": 0.4078,
      "num_input_tokens_seen": 6629288,
      "step": 10045
    },
    {
      "epoch": 5.925707547169811,
      "grad_norm": 2.009742021560669,
      "learning_rate": 8.871898108793796e-06,
      "loss": 0.3865,
      "num_input_tokens_seen": 6633672,
      "step": 10050
    },
    {
      "epoch": 5.928655660377358,
      "grad_norm": 2.099512815475464,
      "learning_rate": 8.870269785384304e-06,
      "loss": 0.6145,
      "num_input_tokens_seen": 6636360,
      "step": 10055
    },
    {
      "epoch": 5.931603773584905,
      "grad_norm": 1.7563974857330322,
      "learning_rate": 8.868640437304206e-06,
      "loss": 0.4413,
      "num_input_tokens_seen": 6638920,
      "step": 10060
    },
    {
      "epoch": 5.934551886792453,
      "grad_norm": 0.9898931384086609,
      "learning_rate": 8.86701006498488e-06,
      "loss": 0.2583,
      "num_input_tokens_seen": 6642376,
      "step": 10065
    },
    {
      "epoch": 5.9375,
      "grad_norm": 2.325822591781616,
      "learning_rate": 8.865378668857972e-06,
      "loss": 0.4842,
      "num_input_tokens_seen": 6644424,
      "step": 10070
    },
    {
      "epoch": 5.940448113207547,
      "grad_norm": 3.539483070373535,
      "learning_rate": 8.863746249355404e-06,
      "loss": 0.515,
      "num_input_tokens_seen": 6647048,
      "step": 10075
    },
    {
      "epoch": 5.943396226415095,
      "grad_norm": 3.895735502243042,
      "learning_rate": 8.862112806909365e-06,
      "loss": 0.5031,
      "num_input_tokens_seen": 6649736,
      "step": 10080
    },
    {
      "epoch": 5.946344339622642,
      "grad_norm": 1.7629865407943726,
      "learning_rate": 8.860478341952314e-06,
      "loss": 0.4718,
      "num_input_tokens_seen": 6652424,
      "step": 10085
    },
    {
      "epoch": 5.949292452830189,
      "grad_norm": 1.4840354919433594,
      "learning_rate": 8.858842854916985e-06,
      "loss": 0.3692,
      "num_input_tokens_seen": 6655496,
      "step": 10090
    },
    {
      "epoch": 5.9522405660377355,
      "grad_norm": 2.2604644298553467,
      "learning_rate": 8.85720634623638e-06,
      "loss": 0.5314,
      "num_input_tokens_seen": 6658856,
      "step": 10095
    },
    {
      "epoch": 5.955188679245283,
      "grad_norm": 1.8268357515335083,
      "learning_rate": 8.855568816343769e-06,
      "loss": 0.2991,
      "num_input_tokens_seen": 6662376,
      "step": 10100
    },
    {
      "epoch": 5.95813679245283,
      "grad_norm": 1.5438791513442993,
      "learning_rate": 8.8539302656727e-06,
      "loss": 0.4345,
      "num_input_tokens_seen": 6665768,
      "step": 10105
    },
    {
      "epoch": 5.961084905660377,
      "grad_norm": 1.3336946964263916,
      "learning_rate": 8.852290694656983e-06,
      "loss": 0.4517,
      "num_input_tokens_seen": 6669832,
      "step": 10110
    },
    {
      "epoch": 5.964033018867925,
      "grad_norm": 4.659071445465088,
      "learning_rate": 8.8506501037307e-06,
      "loss": 0.4761,
      "num_input_tokens_seen": 6673608,
      "step": 10115
    },
    {
      "epoch": 5.966981132075472,
      "grad_norm": 2.671078681945801,
      "learning_rate": 8.849008493328209e-06,
      "loss": 0.6811,
      "num_input_tokens_seen": 6677544,
      "step": 10120
    },
    {
      "epoch": 5.969929245283019,
      "grad_norm": 1.8322943449020386,
      "learning_rate": 8.847365863884131e-06,
      "loss": 0.4753,
      "num_input_tokens_seen": 6680680,
      "step": 10125
    },
    {
      "epoch": 5.972877358490566,
      "grad_norm": 1.4643369913101196,
      "learning_rate": 8.845722215833359e-06,
      "loss": 0.405,
      "num_input_tokens_seen": 6683752,
      "step": 10130
    },
    {
      "epoch": 5.975825471698113,
      "grad_norm": 4.189075946807861,
      "learning_rate": 8.844077549611056e-06,
      "loss": 0.4319,
      "num_input_tokens_seen": 6686184,
      "step": 10135
    },
    {
      "epoch": 5.97877358490566,
      "grad_norm": 2.5693798065185547,
      "learning_rate": 8.842431865652654e-06,
      "loss": 0.4206,
      "num_input_tokens_seen": 6689928,
      "step": 10140
    },
    {
      "epoch": 5.9817216981132075,
      "grad_norm": 2.2869365215301514,
      "learning_rate": 8.840785164393858e-06,
      "loss": 0.5223,
      "num_input_tokens_seen": 6693032,
      "step": 10145
    },
    {
      "epoch": 5.984669811320755,
      "grad_norm": 3.2627298831939697,
      "learning_rate": 8.839137446270634e-06,
      "loss": 0.6188,
      "num_input_tokens_seen": 6695304,
      "step": 10150
    },
    {
      "epoch": 5.987617924528302,
      "grad_norm": 1.7268245220184326,
      "learning_rate": 8.837488711719226e-06,
      "loss": 0.5571,
      "num_input_tokens_seen": 6698280,
      "step": 10155
    },
    {
      "epoch": 5.990566037735849,
      "grad_norm": 1.7392959594726562,
      "learning_rate": 8.835838961176143e-06,
      "loss": 0.3638,
      "num_input_tokens_seen": 6701928,
      "step": 10160
    },
    {
      "epoch": 5.993514150943396,
      "grad_norm": 1.3603533506393433,
      "learning_rate": 8.834188195078164e-06,
      "loss": 0.4256,
      "num_input_tokens_seen": 6706088,
      "step": 10165
    },
    {
      "epoch": 5.996462264150943,
      "grad_norm": 2.205315113067627,
      "learning_rate": 8.832536413862337e-06,
      "loss": 0.3445,
      "num_input_tokens_seen": 6709128,
      "step": 10170
    },
    {
      "epoch": 5.99941037735849,
      "grad_norm": 1.365422010421753,
      "learning_rate": 8.830883617965976e-06,
      "loss": 0.6057,
      "num_input_tokens_seen": 6712264,
      "step": 10175
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.4917214512825012,
      "eval_runtime": 18.381,
      "eval_samples_per_second": 92.269,
      "eval_steps_per_second": 23.067,
      "num_input_tokens_seen": 6712376,
      "step": 10176
    },
    {
      "epoch": 6.002358490566038,
      "grad_norm": 1.287298321723938,
      "learning_rate": 8.829229807826665e-06,
      "loss": 0.3546,
      "num_input_tokens_seen": 6714904,
      "step": 10180
    },
    {
      "epoch": 6.005306603773585,
      "grad_norm": 1.8675764799118042,
      "learning_rate": 8.82757498388226e-06,
      "loss": 0.3122,
      "num_input_tokens_seen": 6718680,
      "step": 10185
    },
    {
      "epoch": 6.008254716981132,
      "grad_norm": 1.2993137836456299,
      "learning_rate": 8.825919146570884e-06,
      "loss": 0.2737,
      "num_input_tokens_seen": 6721688,
      "step": 10190
    },
    {
      "epoch": 6.0112028301886795,
      "grad_norm": 1.3435540199279785,
      "learning_rate": 8.824262296330925e-06,
      "loss": 0.4707,
      "num_input_tokens_seen": 6724888,
      "step": 10195
    },
    {
      "epoch": 6.014150943396227,
      "grad_norm": 1.4659404754638672,
      "learning_rate": 8.822604433601041e-06,
      "loss": 0.4338,
      "num_input_tokens_seen": 6727736,
      "step": 10200
    },
    {
      "epoch": 6.017099056603773,
      "grad_norm": 2.133643865585327,
      "learning_rate": 8.820945558820158e-06,
      "loss": 0.2976,
      "num_input_tokens_seen": 6731352,
      "step": 10205
    },
    {
      "epoch": 6.0200471698113205,
      "grad_norm": 4.136853218078613,
      "learning_rate": 8.81928567242747e-06,
      "loss": 0.5263,
      "num_input_tokens_seen": 6734136,
      "step": 10210
    },
    {
      "epoch": 6.022995283018868,
      "grad_norm": 2.613663673400879,
      "learning_rate": 8.817624774862443e-06,
      "loss": 0.3116,
      "num_input_tokens_seen": 6736952,
      "step": 10215
    },
    {
      "epoch": 6.025943396226415,
      "grad_norm": 1.6332001686096191,
      "learning_rate": 8.815962866564803e-06,
      "loss": 0.4562,
      "num_input_tokens_seen": 6740024,
      "step": 10220
    },
    {
      "epoch": 6.028891509433962,
      "grad_norm": 3.014108896255493,
      "learning_rate": 8.814299947974547e-06,
      "loss": 0.5364,
      "num_input_tokens_seen": 6742616,
      "step": 10225
    },
    {
      "epoch": 6.03183962264151,
      "grad_norm": 2.1580700874328613,
      "learning_rate": 8.812636019531942e-06,
      "loss": 0.4198,
      "num_input_tokens_seen": 6745336,
      "step": 10230
    },
    {
      "epoch": 6.034787735849057,
      "grad_norm": 4.626673221588135,
      "learning_rate": 8.810971081677517e-06,
      "loss": 0.4909,
      "num_input_tokens_seen": 6748856,
      "step": 10235
    },
    {
      "epoch": 6.037735849056604,
      "grad_norm": 3.094789505004883,
      "learning_rate": 8.809305134852076e-06,
      "loss": 0.4479,
      "num_input_tokens_seen": 6751768,
      "step": 10240
    },
    {
      "epoch": 6.040683962264151,
      "grad_norm": 1.487210750579834,
      "learning_rate": 8.807638179496684e-06,
      "loss": 0.3877,
      "num_input_tokens_seen": 6755160,
      "step": 10245
    },
    {
      "epoch": 6.043632075471698,
      "grad_norm": 2.23403000831604,
      "learning_rate": 8.805970216052673e-06,
      "loss": 0.3035,
      "num_input_tokens_seen": 6758040,
      "step": 10250
    },
    {
      "epoch": 6.046580188679245,
      "grad_norm": 1.715050220489502,
      "learning_rate": 8.804301244961645e-06,
      "loss": 0.422,
      "num_input_tokens_seen": 6760440,
      "step": 10255
    },
    {
      "epoch": 6.0495283018867925,
      "grad_norm": 3.074946641921997,
      "learning_rate": 8.802631266665465e-06,
      "loss": 0.4337,
      "num_input_tokens_seen": 6766232,
      "step": 10260
    },
    {
      "epoch": 6.05247641509434,
      "grad_norm": 2.7961418628692627,
      "learning_rate": 8.80096028160627e-06,
      "loss": 0.4287,
      "num_input_tokens_seen": 6770232,
      "step": 10265
    },
    {
      "epoch": 6.055424528301887,
      "grad_norm": 1.8929624557495117,
      "learning_rate": 8.799288290226457e-06,
      "loss": 0.3458,
      "num_input_tokens_seen": 6773240,
      "step": 10270
    },
    {
      "epoch": 6.058372641509434,
      "grad_norm": 2.5479612350463867,
      "learning_rate": 8.797615292968698e-06,
      "loss": 0.5066,
      "num_input_tokens_seen": 6776600,
      "step": 10275
    },
    {
      "epoch": 6.061320754716981,
      "grad_norm": 1.8109315633773804,
      "learning_rate": 8.795941290275923e-06,
      "loss": 0.5154,
      "num_input_tokens_seen": 6779896,
      "step": 10280
    },
    {
      "epoch": 6.064268867924528,
      "grad_norm": 1.8680700063705444,
      "learning_rate": 8.79426628259133e-06,
      "loss": 0.5529,
      "num_input_tokens_seen": 6782552,
      "step": 10285
    },
    {
      "epoch": 6.067216981132075,
      "grad_norm": 1.810267448425293,
      "learning_rate": 8.792590270358389e-06,
      "loss": 0.527,
      "num_input_tokens_seen": 6785624,
      "step": 10290
    },
    {
      "epoch": 6.070165094339623,
      "grad_norm": 2.100430727005005,
      "learning_rate": 8.790913254020827e-06,
      "loss": 0.4358,
      "num_input_tokens_seen": 6789112,
      "step": 10295
    },
    {
      "epoch": 6.07311320754717,
      "grad_norm": 1.7013344764709473,
      "learning_rate": 8.789235234022643e-06,
      "loss": 0.3886,
      "num_input_tokens_seen": 6793688,
      "step": 10300
    },
    {
      "epoch": 6.076061320754717,
      "grad_norm": 1.2420345544815063,
      "learning_rate": 8.787556210808101e-06,
      "loss": 0.4007,
      "num_input_tokens_seen": 6797624,
      "step": 10305
    },
    {
      "epoch": 6.0790094339622645,
      "grad_norm": 1.939249038696289,
      "learning_rate": 8.78587618482173e-06,
      "loss": 0.3217,
      "num_input_tokens_seen": 6799896,
      "step": 10310
    },
    {
      "epoch": 6.081957547169812,
      "grad_norm": 1.3290318250656128,
      "learning_rate": 8.78419515650832e-06,
      "loss": 0.4801,
      "num_input_tokens_seen": 6803416,
      "step": 10315
    },
    {
      "epoch": 6.084905660377358,
      "grad_norm": 6.7564778327941895,
      "learning_rate": 8.782513126312934e-06,
      "loss": 0.4218,
      "num_input_tokens_seen": 6807192,
      "step": 10320
    },
    {
      "epoch": 6.087853773584905,
      "grad_norm": 1.9599509239196777,
      "learning_rate": 8.780830094680897e-06,
      "loss": 0.467,
      "num_input_tokens_seen": 6810744,
      "step": 10325
    },
    {
      "epoch": 6.090801886792453,
      "grad_norm": 1.9176570177078247,
      "learning_rate": 8.779146062057797e-06,
      "loss": 0.4842,
      "num_input_tokens_seen": 6814200,
      "step": 10330
    },
    {
      "epoch": 6.09375,
      "grad_norm": 2.833617925643921,
      "learning_rate": 8.777461028889492e-06,
      "loss": 0.4476,
      "num_input_tokens_seen": 6816824,
      "step": 10335
    },
    {
      "epoch": 6.096698113207547,
      "grad_norm": 2.003995180130005,
      "learning_rate": 8.775774995622097e-06,
      "loss": 0.5366,
      "num_input_tokens_seen": 6819864,
      "step": 10340
    },
    {
      "epoch": 6.099646226415095,
      "grad_norm": 2.89768648147583,
      "learning_rate": 8.774087962702e-06,
      "loss": 0.4101,
      "num_input_tokens_seen": 6825688,
      "step": 10345
    },
    {
      "epoch": 6.102594339622642,
      "grad_norm": 1.8619805574417114,
      "learning_rate": 8.772399930575849e-06,
      "loss": 0.4961,
      "num_input_tokens_seen": 6828440,
      "step": 10350
    },
    {
      "epoch": 6.105542452830188,
      "grad_norm": 2.818035364151001,
      "learning_rate": 8.77071089969056e-06,
      "loss": 0.3896,
      "num_input_tokens_seen": 6831736,
      "step": 10355
    },
    {
      "epoch": 6.1084905660377355,
      "grad_norm": 1.5850906372070312,
      "learning_rate": 8.769020870493309e-06,
      "loss": 0.4842,
      "num_input_tokens_seen": 6835128,
      "step": 10360
    },
    {
      "epoch": 6.111438679245283,
      "grad_norm": 1.930755853652954,
      "learning_rate": 8.767329843431537e-06,
      "loss": 0.3925,
      "num_input_tokens_seen": 6838936,
      "step": 10365
    },
    {
      "epoch": 6.11438679245283,
      "grad_norm": 1.7107948064804077,
      "learning_rate": 8.765637818952954e-06,
      "loss": 0.4446,
      "num_input_tokens_seen": 6843288,
      "step": 10370
    },
    {
      "epoch": 6.117334905660377,
      "grad_norm": 2.314715623855591,
      "learning_rate": 8.76394479750553e-06,
      "loss": 0.3959,
      "num_input_tokens_seen": 6846392,
      "step": 10375
    },
    {
      "epoch": 6.120283018867925,
      "grad_norm": 4.676614761352539,
      "learning_rate": 8.762250779537499e-06,
      "loss": 0.5094,
      "num_input_tokens_seen": 6849784,
      "step": 10380
    },
    {
      "epoch": 6.123231132075472,
      "grad_norm": 1.1493842601776123,
      "learning_rate": 8.760555765497358e-06,
      "loss": 0.422,
      "num_input_tokens_seen": 6853048,
      "step": 10385
    },
    {
      "epoch": 6.126179245283019,
      "grad_norm": 1.218997836112976,
      "learning_rate": 8.758859755833873e-06,
      "loss": 0.5775,
      "num_input_tokens_seen": 6855832,
      "step": 10390
    },
    {
      "epoch": 6.129127358490566,
      "grad_norm": 3.813154935836792,
      "learning_rate": 8.757162750996066e-06,
      "loss": 0.377,
      "num_input_tokens_seen": 6859320,
      "step": 10395
    },
    {
      "epoch": 6.132075471698113,
      "grad_norm": 3.9837639331817627,
      "learning_rate": 8.755464751433229e-06,
      "loss": 0.4393,
      "num_input_tokens_seen": 6862264,
      "step": 10400
    },
    {
      "epoch": 6.13502358490566,
      "grad_norm": 2.651789665222168,
      "learning_rate": 8.753765757594915e-06,
      "loss": 0.4472,
      "num_input_tokens_seen": 6865112,
      "step": 10405
    },
    {
      "epoch": 6.1379716981132075,
      "grad_norm": 3.5008323192596436,
      "learning_rate": 8.752065769930938e-06,
      "loss": 0.472,
      "num_input_tokens_seen": 6867992,
      "step": 10410
    },
    {
      "epoch": 6.140919811320755,
      "grad_norm": 1.687278389930725,
      "learning_rate": 8.75036478889138e-06,
      "loss": 0.3743,
      "num_input_tokens_seen": 6872856,
      "step": 10415
    },
    {
      "epoch": 6.143867924528302,
      "grad_norm": 2.4227309226989746,
      "learning_rate": 8.748662814926576e-06,
      "loss": 0.5279,
      "num_input_tokens_seen": 6875832,
      "step": 10420
    },
    {
      "epoch": 6.146816037735849,
      "grad_norm": 1.9549332857131958,
      "learning_rate": 8.746959848487139e-06,
      "loss": 0.3651,
      "num_input_tokens_seen": 6878968,
      "step": 10425
    },
    {
      "epoch": 6.149764150943396,
      "grad_norm": 4.06995153427124,
      "learning_rate": 8.745255890023934e-06,
      "loss": 0.4598,
      "num_input_tokens_seen": 6881272,
      "step": 10430
    },
    {
      "epoch": 6.152712264150943,
      "grad_norm": 1.9422136545181274,
      "learning_rate": 8.74355093998809e-06,
      "loss": 0.3769,
      "num_input_tokens_seen": 6884568,
      "step": 10435
    },
    {
      "epoch": 6.15566037735849,
      "grad_norm": 1.605150818824768,
      "learning_rate": 8.741844998831001e-06,
      "loss": 0.4119,
      "num_input_tokens_seen": 6887384,
      "step": 10440
    },
    {
      "epoch": 6.158608490566038,
      "grad_norm": 1.3265048265457153,
      "learning_rate": 8.740138067004323e-06,
      "loss": 0.6006,
      "num_input_tokens_seen": 6890040,
      "step": 10445
    },
    {
      "epoch": 6.161556603773585,
      "grad_norm": 2.258838415145874,
      "learning_rate": 8.738430144959973e-06,
      "loss": 0.4364,
      "num_input_tokens_seen": 6892504,
      "step": 10450
    },
    {
      "epoch": 6.164504716981132,
      "grad_norm": 2.0912537574768066,
      "learning_rate": 8.73672123315013e-06,
      "loss": 0.4951,
      "num_input_tokens_seen": 6896312,
      "step": 10455
    },
    {
      "epoch": 6.1674528301886795,
      "grad_norm": 3.4911227226257324,
      "learning_rate": 8.735011332027234e-06,
      "loss": 0.4711,
      "num_input_tokens_seen": 6898968,
      "step": 10460
    },
    {
      "epoch": 6.170400943396227,
      "grad_norm": 1.6597505807876587,
      "learning_rate": 8.733300442043993e-06,
      "loss": 0.3913,
      "num_input_tokens_seen": 6903512,
      "step": 10465
    },
    {
      "epoch": 6.173349056603773,
      "grad_norm": 2.8113670349121094,
      "learning_rate": 8.73158856365337e-06,
      "loss": 0.4078,
      "num_input_tokens_seen": 6906552,
      "step": 10470
    },
    {
      "epoch": 6.1762971698113205,
      "grad_norm": 1.214758276939392,
      "learning_rate": 8.729875697308592e-06,
      "loss": 0.3746,
      "num_input_tokens_seen": 6908888,
      "step": 10475
    },
    {
      "epoch": 6.179245283018868,
      "grad_norm": 1.5151963233947754,
      "learning_rate": 8.728161843463148e-06,
      "loss": 0.3597,
      "num_input_tokens_seen": 6912056,
      "step": 10480
    },
    {
      "epoch": 6.182193396226415,
      "grad_norm": 2.2324090003967285,
      "learning_rate": 8.726447002570791e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 6915480,
      "step": 10485
    },
    {
      "epoch": 6.185141509433962,
      "grad_norm": 1.99545156955719,
      "learning_rate": 8.724731175085526e-06,
      "loss": 0.3809,
      "num_input_tokens_seen": 6918200,
      "step": 10490
    },
    {
      "epoch": 6.18808962264151,
      "grad_norm": 1.729260802268982,
      "learning_rate": 8.723014361461633e-06,
      "loss": 0.4802,
      "num_input_tokens_seen": 6922360,
      "step": 10495
    },
    {
      "epoch": 6.191037735849057,
      "grad_norm": 1.6530083417892456,
      "learning_rate": 8.72129656215364e-06,
      "loss": 0.5085,
      "num_input_tokens_seen": 6925976,
      "step": 10500
    },
    {
      "epoch": 6.193985849056604,
      "grad_norm": 1.8385859727859497,
      "learning_rate": 8.719577777616347e-06,
      "loss": 0.4434,
      "num_input_tokens_seen": 6928856,
      "step": 10505
    },
    {
      "epoch": 6.196933962264151,
      "grad_norm": 1.4353750944137573,
      "learning_rate": 8.717858008304804e-06,
      "loss": 0.3371,
      "num_input_tokens_seen": 6932728,
      "step": 10510
    },
    {
      "epoch": 6.199882075471698,
      "grad_norm": 3.4084675312042236,
      "learning_rate": 8.71613725467433e-06,
      "loss": 0.4266,
      "num_input_tokens_seen": 6935832,
      "step": 10515
    },
    {
      "epoch": 6.202830188679245,
      "grad_norm": 6.35112190246582,
      "learning_rate": 8.714415517180506e-06,
      "loss": 0.3919,
      "num_input_tokens_seen": 6939128,
      "step": 10520
    },
    {
      "epoch": 6.2057783018867925,
      "grad_norm": 2.6945888996124268,
      "learning_rate": 8.712692796279164e-06,
      "loss": 0.6216,
      "num_input_tokens_seen": 6947096,
      "step": 10525
    },
    {
      "epoch": 6.20872641509434,
      "grad_norm": 2.209061861038208,
      "learning_rate": 8.710969092426401e-06,
      "loss": 0.3592,
      "num_input_tokens_seen": 6950456,
      "step": 10530
    },
    {
      "epoch": 6.211674528301887,
      "grad_norm": 1.9699876308441162,
      "learning_rate": 8.70924440607858e-06,
      "loss": 0.5177,
      "num_input_tokens_seen": 6953752,
      "step": 10535
    },
    {
      "epoch": 6.214622641509434,
      "grad_norm": 1.688982367515564,
      "learning_rate": 8.707518737692315e-06,
      "loss": 0.3401,
      "num_input_tokens_seen": 6956248,
      "step": 10540
    },
    {
      "epoch": 6.217570754716981,
      "grad_norm": 2.171053409576416,
      "learning_rate": 8.705792087724485e-06,
      "loss": 0.4212,
      "num_input_tokens_seen": 6959224,
      "step": 10545
    },
    {
      "epoch": 6.220518867924528,
      "grad_norm": 3.1826138496398926,
      "learning_rate": 8.704064456632231e-06,
      "loss": 0.4576,
      "num_input_tokens_seen": 6961944,
      "step": 10550
    },
    {
      "epoch": 6.223466981132075,
      "grad_norm": 1.4436403512954712,
      "learning_rate": 8.702335844872946e-06,
      "loss": 0.3196,
      "num_input_tokens_seen": 6965464,
      "step": 10555
    },
    {
      "epoch": 6.226415094339623,
      "grad_norm": 1.4948680400848389,
      "learning_rate": 8.700606252904293e-06,
      "loss": 0.2915,
      "num_input_tokens_seen": 6968952,
      "step": 10560
    },
    {
      "epoch": 6.22936320754717,
      "grad_norm": 2.1730222702026367,
      "learning_rate": 8.698875681184183e-06,
      "loss": 0.4269,
      "num_input_tokens_seen": 6971960,
      "step": 10565
    },
    {
      "epoch": 6.232311320754717,
      "grad_norm": 2.602480173110962,
      "learning_rate": 8.697144130170797e-06,
      "loss": 0.3847,
      "num_input_tokens_seen": 6975992,
      "step": 10570
    },
    {
      "epoch": 6.2352594339622645,
      "grad_norm": 1.3325378894805908,
      "learning_rate": 8.695411600322568e-06,
      "loss": 0.3748,
      "num_input_tokens_seen": 6978488,
      "step": 10575
    },
    {
      "epoch": 6.238207547169812,
      "grad_norm": 1.5378551483154297,
      "learning_rate": 8.693678092098191e-06,
      "loss": 0.4509,
      "num_input_tokens_seen": 6982264,
      "step": 10580
    },
    {
      "epoch": 6.241155660377358,
      "grad_norm": 1.2544060945510864,
      "learning_rate": 8.691943605956621e-06,
      "loss": 0.4246,
      "num_input_tokens_seen": 6986008,
      "step": 10585
    },
    {
      "epoch": 6.244103773584905,
      "grad_norm": 3.179379463195801,
      "learning_rate": 8.690208142357069e-06,
      "loss": 0.4234,
      "num_input_tokens_seen": 6989080,
      "step": 10590
    },
    {
      "epoch": 6.247051886792453,
      "grad_norm": 1.4841461181640625,
      "learning_rate": 8.68847170175901e-06,
      "loss": 0.4418,
      "num_input_tokens_seen": 6992248,
      "step": 10595
    },
    {
      "epoch": 6.25,
      "grad_norm": 2.4652390480041504,
      "learning_rate": 8.686734284622168e-06,
      "loss": 0.5605,
      "num_input_tokens_seen": 6995256,
      "step": 10600
    },
    {
      "epoch": 6.252948113207547,
      "grad_norm": 1.4408543109893799,
      "learning_rate": 8.684995891406537e-06,
      "loss": 0.4648,
      "num_input_tokens_seen": 6999096,
      "step": 10605
    },
    {
      "epoch": 6.255896226415095,
      "grad_norm": 2.643850088119507,
      "learning_rate": 8.683256522572362e-06,
      "loss": 0.3969,
      "num_input_tokens_seen": 7001496,
      "step": 10610
    },
    {
      "epoch": 6.258844339622642,
      "grad_norm": 1.9333022832870483,
      "learning_rate": 8.68151617858015e-06,
      "loss": 0.5116,
      "num_input_tokens_seen": 7004984,
      "step": 10615
    },
    {
      "epoch": 6.261792452830189,
      "grad_norm": 1.465149164199829,
      "learning_rate": 8.67977485989066e-06,
      "loss": 0.3112,
      "num_input_tokens_seen": 7009368,
      "step": 10620
    },
    {
      "epoch": 6.2647405660377355,
      "grad_norm": 2.6632988452911377,
      "learning_rate": 8.67803256696492e-06,
      "loss": 0.5368,
      "num_input_tokens_seen": 7012952,
      "step": 10625
    },
    {
      "epoch": 6.267688679245283,
      "grad_norm": 2.314897298812866,
      "learning_rate": 8.676289300264205e-06,
      "loss": 0.4526,
      "num_input_tokens_seen": 7016792,
      "step": 10630
    },
    {
      "epoch": 6.27063679245283,
      "grad_norm": 4.414429664611816,
      "learning_rate": 8.674545060250054e-06,
      "loss": 0.377,
      "num_input_tokens_seen": 7020536,
      "step": 10635
    },
    {
      "epoch": 6.273584905660377,
      "grad_norm": 1.7064440250396729,
      "learning_rate": 8.672799847384263e-06,
      "loss": 0.4219,
      "num_input_tokens_seen": 7024344,
      "step": 10640
    },
    {
      "epoch": 6.276533018867925,
      "grad_norm": 2.383183479309082,
      "learning_rate": 8.671053662128883e-06,
      "loss": 0.4081,
      "num_input_tokens_seen": 7028856,
      "step": 10645
    },
    {
      "epoch": 6.279481132075472,
      "grad_norm": 2.0439820289611816,
      "learning_rate": 8.669306504946223e-06,
      "loss": 0.423,
      "num_input_tokens_seen": 7032312,
      "step": 10650
    },
    {
      "epoch": 6.282429245283019,
      "grad_norm": 2.133373975753784,
      "learning_rate": 8.667558376298854e-06,
      "loss": 0.4791,
      "num_input_tokens_seen": 7035096,
      "step": 10655
    },
    {
      "epoch": 6.285377358490566,
      "grad_norm": 2.815107822418213,
      "learning_rate": 8.665809276649597e-06,
      "loss": 0.4338,
      "num_input_tokens_seen": 7038200,
      "step": 10660
    },
    {
      "epoch": 6.288325471698113,
      "grad_norm": 2.659992218017578,
      "learning_rate": 8.664059206461537e-06,
      "loss": 0.3428,
      "num_input_tokens_seen": 7040888,
      "step": 10665
    },
    {
      "epoch": 6.29127358490566,
      "grad_norm": 4.182241916656494,
      "learning_rate": 8.662308166198009e-06,
      "loss": 0.5691,
      "num_input_tokens_seen": 7044088,
      "step": 10670
    },
    {
      "epoch": 6.2942216981132075,
      "grad_norm": 2.5040297508239746,
      "learning_rate": 8.660556156322611e-06,
      "loss": 0.4697,
      "num_input_tokens_seen": 7046584,
      "step": 10675
    },
    {
      "epoch": 6.297169811320755,
      "grad_norm": 0.3434968590736389,
      "learning_rate": 8.658803177299196e-06,
      "loss": 0.3717,
      "num_input_tokens_seen": 7053720,
      "step": 10680
    },
    {
      "epoch": 6.300117924528302,
      "grad_norm": 1.612398386001587,
      "learning_rate": 8.65704922959187e-06,
      "loss": 0.3932,
      "num_input_tokens_seen": 7056088,
      "step": 10685
    },
    {
      "epoch": 6.303066037735849,
      "grad_norm": 1.6986474990844727,
      "learning_rate": 8.655294313664998e-06,
      "loss": 0.4927,
      "num_input_tokens_seen": 7059992,
      "step": 10690
    },
    {
      "epoch": 6.306014150943396,
      "grad_norm": 2.445718765258789,
      "learning_rate": 8.653538429983204e-06,
      "loss": 0.5177,
      "num_input_tokens_seen": 7062392,
      "step": 10695
    },
    {
      "epoch": 6.308962264150943,
      "grad_norm": 1.9402421712875366,
      "learning_rate": 8.651781579011366e-06,
      "loss": 0.3208,
      "num_input_tokens_seen": 7065240,
      "step": 10700
    },
    {
      "epoch": 6.31191037735849,
      "grad_norm": 1.8633865118026733,
      "learning_rate": 8.650023761214615e-06,
      "loss": 0.4158,
      "num_input_tokens_seen": 7068472,
      "step": 10705
    },
    {
      "epoch": 6.314858490566038,
      "grad_norm": 2.3171324729919434,
      "learning_rate": 8.648264977058344e-06,
      "loss": 0.4388,
      "num_input_tokens_seen": 7073112,
      "step": 10710
    },
    {
      "epoch": 6.317806603773585,
      "grad_norm": 3.1951680183410645,
      "learning_rate": 8.646505227008197e-06,
      "loss": 0.489,
      "num_input_tokens_seen": 7075544,
      "step": 10715
    },
    {
      "epoch": 6.320754716981132,
      "grad_norm": 2.5260097980499268,
      "learning_rate": 8.644744511530074e-06,
      "loss": 0.489,
      "num_input_tokens_seen": 7078456,
      "step": 10720
    },
    {
      "epoch": 6.3237028301886795,
      "grad_norm": 1.9451972246170044,
      "learning_rate": 8.642982831090135e-06,
      "loss": 0.3252,
      "num_input_tokens_seen": 7082904,
      "step": 10725
    },
    {
      "epoch": 6.326650943396227,
      "grad_norm": 2.3145954608917236,
      "learning_rate": 8.64122018615479e-06,
      "loss": 0.4343,
      "num_input_tokens_seen": 7085976,
      "step": 10730
    },
    {
      "epoch": 6.329599056603773,
      "grad_norm": 2.222755193710327,
      "learning_rate": 8.63945657719071e-06,
      "loss": 0.4189,
      "num_input_tokens_seen": 7088888,
      "step": 10735
    },
    {
      "epoch": 6.3325471698113205,
      "grad_norm": 2.5750207901000977,
      "learning_rate": 8.637692004664816e-06,
      "loss": 0.4566,
      "num_input_tokens_seen": 7092120,
      "step": 10740
    },
    {
      "epoch": 6.335495283018868,
      "grad_norm": 2.286430835723877,
      "learning_rate": 8.635926469044284e-06,
      "loss": 0.4778,
      "num_input_tokens_seen": 7095672,
      "step": 10745
    },
    {
      "epoch": 6.338443396226415,
      "grad_norm": 2.610384941101074,
      "learning_rate": 8.63415997079655e-06,
      "loss": 0.3156,
      "num_input_tokens_seen": 7099032,
      "step": 10750
    },
    {
      "epoch": 6.341391509433962,
      "grad_norm": 1.4542592763900757,
      "learning_rate": 8.6323925103893e-06,
      "loss": 0.4353,
      "num_input_tokens_seen": 7102680,
      "step": 10755
    },
    {
      "epoch": 6.34433962264151,
      "grad_norm": 1.9442470073699951,
      "learning_rate": 8.63062408829048e-06,
      "loss": 0.4208,
      "num_input_tokens_seen": 7107192,
      "step": 10760
    },
    {
      "epoch": 6.347287735849057,
      "grad_norm": 1.7456800937652588,
      "learning_rate": 8.628854704968285e-06,
      "loss": 0.3723,
      "num_input_tokens_seen": 7109912,
      "step": 10765
    },
    {
      "epoch": 6.350235849056604,
      "grad_norm": 2.7073922157287598,
      "learning_rate": 8.62708436089117e-06,
      "loss": 0.4302,
      "num_input_tokens_seen": 7113176,
      "step": 10770
    },
    {
      "epoch": 6.353183962264151,
      "grad_norm": 2.8276631832122803,
      "learning_rate": 8.625313056527836e-06,
      "loss": 0.5979,
      "num_input_tokens_seen": 7116120,
      "step": 10775
    },
    {
      "epoch": 6.356132075471698,
      "grad_norm": 1.720881700515747,
      "learning_rate": 8.623540792347244e-06,
      "loss": 0.4104,
      "num_input_tokens_seen": 7119160,
      "step": 10780
    },
    {
      "epoch": 6.359080188679245,
      "grad_norm": 5.57639217376709,
      "learning_rate": 8.621767568818614e-06,
      "loss": 0.4433,
      "num_input_tokens_seen": 7122136,
      "step": 10785
    },
    {
      "epoch": 6.3620283018867925,
      "grad_norm": 1.9545085430145264,
      "learning_rate": 8.619993386411409e-06,
      "loss": 0.3679,
      "num_input_tokens_seen": 7125080,
      "step": 10790
    },
    {
      "epoch": 6.36497641509434,
      "grad_norm": 1.3263589143753052,
      "learning_rate": 8.618218245595356e-06,
      "loss": 0.5295,
      "num_input_tokens_seen": 7129400,
      "step": 10795
    },
    {
      "epoch": 6.367924528301887,
      "grad_norm": 2.5473361015319824,
      "learning_rate": 8.616442146840427e-06,
      "loss": 0.4853,
      "num_input_tokens_seen": 7132408,
      "step": 10800
    },
    {
      "epoch": 6.370872641509434,
      "grad_norm": 2.8116750717163086,
      "learning_rate": 8.614665090616854e-06,
      "loss": 0.5536,
      "num_input_tokens_seen": 7135160,
      "step": 10805
    },
    {
      "epoch": 6.373820754716981,
      "grad_norm": 1.719628930091858,
      "learning_rate": 8.61288707739512e-06,
      "loss": 0.5215,
      "num_input_tokens_seen": 7137752,
      "step": 10810
    },
    {
      "epoch": 6.376768867924528,
      "grad_norm": 2.0831449031829834,
      "learning_rate": 8.611108107645963e-06,
      "loss": 0.4412,
      "num_input_tokens_seen": 7141048,
      "step": 10815
    },
    {
      "epoch": 6.379716981132075,
      "grad_norm": 2.9031481742858887,
      "learning_rate": 8.609328181840368e-06,
      "loss": 0.5216,
      "num_input_tokens_seen": 7143864,
      "step": 10820
    },
    {
      "epoch": 6.382665094339623,
      "grad_norm": 2.0337836742401123,
      "learning_rate": 8.607547300449585e-06,
      "loss": 0.3668,
      "num_input_tokens_seen": 7146584,
      "step": 10825
    },
    {
      "epoch": 6.38561320754717,
      "grad_norm": 1.8151413202285767,
      "learning_rate": 8.605765463945105e-06,
      "loss": 0.3445,
      "num_input_tokens_seen": 7149272,
      "step": 10830
    },
    {
      "epoch": 6.388561320754717,
      "grad_norm": 1.7562928199768066,
      "learning_rate": 8.603982672798678e-06,
      "loss": 0.4165,
      "num_input_tokens_seen": 7152216,
      "step": 10835
    },
    {
      "epoch": 6.3915094339622645,
      "grad_norm": 1.335750937461853,
      "learning_rate": 8.602198927482309e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 7155800,
      "step": 10840
    },
    {
      "epoch": 6.394457547169811,
      "grad_norm": 3.26888370513916,
      "learning_rate": 8.600414228468245e-06,
      "loss": 0.3889,
      "num_input_tokens_seen": 7158456,
      "step": 10845
    },
    {
      "epoch": 6.397405660377358,
      "grad_norm": 3.800729990005493,
      "learning_rate": 8.598628576229e-06,
      "loss": 0.6064,
      "num_input_tokens_seen": 7160984,
      "step": 10850
    },
    {
      "epoch": 6.400353773584905,
      "grad_norm": 1.0547301769256592,
      "learning_rate": 8.596841971237328e-06,
      "loss": 0.3546,
      "num_input_tokens_seen": 7164568,
      "step": 10855
    },
    {
      "epoch": 6.403301886792453,
      "grad_norm": 3.0336806774139404,
      "learning_rate": 8.595054413966246e-06,
      "loss": 0.4047,
      "num_input_tokens_seen": 7167800,
      "step": 10860
    },
    {
      "epoch": 6.40625,
      "grad_norm": 1.3830182552337646,
      "learning_rate": 8.593265904889011e-06,
      "loss": 0.348,
      "num_input_tokens_seen": 7170776,
      "step": 10865
    },
    {
      "epoch": 6.409198113207547,
      "grad_norm": 2.137472629547119,
      "learning_rate": 8.591476444479141e-06,
      "loss": 0.4048,
      "num_input_tokens_seen": 7173496,
      "step": 10870
    },
    {
      "epoch": 6.412146226415095,
      "grad_norm": 1.7305073738098145,
      "learning_rate": 8.589686033210407e-06,
      "loss": 0.5386,
      "num_input_tokens_seen": 7178200,
      "step": 10875
    },
    {
      "epoch": 6.415094339622642,
      "grad_norm": 1.7456393241882324,
      "learning_rate": 8.587894671556823e-06,
      "loss": 0.3884,
      "num_input_tokens_seen": 7181240,
      "step": 10880
    },
    {
      "epoch": 6.418042452830189,
      "grad_norm": 2.073417901992798,
      "learning_rate": 8.586102359992663e-06,
      "loss": 0.4836,
      "num_input_tokens_seen": 7184472,
      "step": 10885
    },
    {
      "epoch": 6.4209905660377355,
      "grad_norm": 2.4187119007110596,
      "learning_rate": 8.584309098992447e-06,
      "loss": 0.4532,
      "num_input_tokens_seen": 7187320,
      "step": 10890
    },
    {
      "epoch": 6.423938679245283,
      "grad_norm": 1.805241346359253,
      "learning_rate": 8.58251488903095e-06,
      "loss": 0.3383,
      "num_input_tokens_seen": 7190424,
      "step": 10895
    },
    {
      "epoch": 6.42688679245283,
      "grad_norm": 1.2227383852005005,
      "learning_rate": 8.580719730583196e-06,
      "loss": 0.4092,
      "num_input_tokens_seen": 7193816,
      "step": 10900
    },
    {
      "epoch": 6.429834905660377,
      "grad_norm": 4.081076622009277,
      "learning_rate": 8.578923624124462e-06,
      "loss": 0.3982,
      "num_input_tokens_seen": 7197176,
      "step": 10905
    },
    {
      "epoch": 6.432783018867925,
      "grad_norm": 2.1511688232421875,
      "learning_rate": 8.577126570130273e-06,
      "loss": 0.3963,
      "num_input_tokens_seen": 7200120,
      "step": 10910
    },
    {
      "epoch": 6.435731132075472,
      "grad_norm": 1.4649438858032227,
      "learning_rate": 8.575328569076408e-06,
      "loss": 0.3412,
      "num_input_tokens_seen": 7203160,
      "step": 10915
    },
    {
      "epoch": 6.438679245283019,
      "grad_norm": 1.5180115699768066,
      "learning_rate": 8.573529621438896e-06,
      "loss": 0.522,
      "num_input_tokens_seen": 7206296,
      "step": 10920
    },
    {
      "epoch": 6.441627358490566,
      "grad_norm": 1.9561312198638916,
      "learning_rate": 8.571729727694015e-06,
      "loss": 0.4112,
      "num_input_tokens_seen": 7209560,
      "step": 10925
    },
    {
      "epoch": 6.444575471698113,
      "grad_norm": 2.0280184745788574,
      "learning_rate": 8.569928888318298e-06,
      "loss": 0.3771,
      "num_input_tokens_seen": 7212408,
      "step": 10930
    },
    {
      "epoch": 6.44752358490566,
      "grad_norm": 2.0849130153656006,
      "learning_rate": 8.56812710378852e-06,
      "loss": 0.6758,
      "num_input_tokens_seen": 7215992,
      "step": 10935
    },
    {
      "epoch": 6.4504716981132075,
      "grad_norm": 2.005202293395996,
      "learning_rate": 8.566324374581714e-06,
      "loss": 0.5332,
      "num_input_tokens_seen": 7221784,
      "step": 10940
    },
    {
      "epoch": 6.453419811320755,
      "grad_norm": 2.82804012298584,
      "learning_rate": 8.564520701175158e-06,
      "loss": 0.4538,
      "num_input_tokens_seen": 7224664,
      "step": 10945
    },
    {
      "epoch": 6.456367924528302,
      "grad_norm": 1.4794119596481323,
      "learning_rate": 8.562716084046387e-06,
      "loss": 0.4694,
      "num_input_tokens_seen": 7228536,
      "step": 10950
    },
    {
      "epoch": 6.459316037735849,
      "grad_norm": 1.6174602508544922,
      "learning_rate": 8.560910523673177e-06,
      "loss": 0.558,
      "num_input_tokens_seen": 7232248,
      "step": 10955
    },
    {
      "epoch": 6.462264150943396,
      "grad_norm": 0.30869436264038086,
      "learning_rate": 8.55910402053356e-06,
      "loss": 0.3462,
      "num_input_tokens_seen": 7237208,
      "step": 10960
    },
    {
      "epoch": 6.465212264150943,
      "grad_norm": 1.5695024728775024,
      "learning_rate": 8.557296575105814e-06,
      "loss": 0.3625,
      "num_input_tokens_seen": 7240216,
      "step": 10965
    },
    {
      "epoch": 6.46816037735849,
      "grad_norm": 1.6111716032028198,
      "learning_rate": 8.555488187868469e-06,
      "loss": 0.3941,
      "num_input_tokens_seen": 7244120,
      "step": 10970
    },
    {
      "epoch": 6.471108490566038,
      "grad_norm": 1.5303854942321777,
      "learning_rate": 8.5536788593003e-06,
      "loss": 0.334,
      "num_input_tokens_seen": 7247288,
      "step": 10975
    },
    {
      "epoch": 6.474056603773585,
      "grad_norm": 3.0635359287261963,
      "learning_rate": 8.55186858988034e-06,
      "loss": 0.4408,
      "num_input_tokens_seen": 7250904,
      "step": 10980
    },
    {
      "epoch": 6.477004716981132,
      "grad_norm": 2.443293809890747,
      "learning_rate": 8.550057380087863e-06,
      "loss": 0.3348,
      "num_input_tokens_seen": 7253912,
      "step": 10985
    },
    {
      "epoch": 6.4799528301886795,
      "grad_norm": 2.74592661857605,
      "learning_rate": 8.548245230402396e-06,
      "loss": 0.5838,
      "num_input_tokens_seen": 7256440,
      "step": 10990
    },
    {
      "epoch": 6.482900943396227,
      "grad_norm": 4.4192399978637695,
      "learning_rate": 8.546432141303711e-06,
      "loss": 0.5001,
      "num_input_tokens_seen": 7259736,
      "step": 10995
    },
    {
      "epoch": 6.485849056603773,
      "grad_norm": 3.6658389568328857,
      "learning_rate": 8.544618113271833e-06,
      "loss": 0.5421,
      "num_input_tokens_seen": 7262104,
      "step": 11000
    },
    {
      "epoch": 6.4887971698113205,
      "grad_norm": 1.7639294862747192,
      "learning_rate": 8.542803146787032e-06,
      "loss": 0.407,
      "num_input_tokens_seen": 7265144,
      "step": 11005
    },
    {
      "epoch": 6.491745283018868,
      "grad_norm": 2.627187490463257,
      "learning_rate": 8.54098724232983e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 7268088,
      "step": 11010
    },
    {
      "epoch": 6.494693396226415,
      "grad_norm": 3.009317398071289,
      "learning_rate": 8.539170400380994e-06,
      "loss": 0.494,
      "num_input_tokens_seen": 7270808,
      "step": 11015
    },
    {
      "epoch": 6.497641509433962,
      "grad_norm": 1.7488528490066528,
      "learning_rate": 8.537352621421542e-06,
      "loss": 0.4612,
      "num_input_tokens_seen": 7274616,
      "step": 11020
    },
    {
      "epoch": 6.50058962264151,
      "grad_norm": 1.8036093711853027,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.3647,
      "num_input_tokens_seen": 7277752,
      "step": 11025
    },
    {
      "epoch": 6.503537735849057,
      "grad_norm": 5.226635456085205,
      "learning_rate": 8.533714254396096e-06,
      "loss": 0.4609,
      "num_input_tokens_seen": 7280920,
      "step": 11030
    },
    {
      "epoch": 6.506485849056604,
      "grad_norm": 1.9431790113449097,
      "learning_rate": 8.531893667293375e-06,
      "loss": 0.3676,
      "num_input_tokens_seen": 7284952,
      "step": 11035
    },
    {
      "epoch": 6.509433962264151,
      "grad_norm": 2.1642608642578125,
      "learning_rate": 8.530072145106585e-06,
      "loss": 0.3634,
      "num_input_tokens_seen": 7287544,
      "step": 11040
    },
    {
      "epoch": 6.512382075471698,
      "grad_norm": 1.4208194017410278,
      "learning_rate": 8.528249688317978e-06,
      "loss": 0.4465,
      "num_input_tokens_seen": 7290232,
      "step": 11045
    },
    {
      "epoch": 6.515330188679245,
      "grad_norm": 2.2767086029052734,
      "learning_rate": 8.526426297410062e-06,
      "loss": 0.578,
      "num_input_tokens_seen": 7293176,
      "step": 11050
    },
    {
      "epoch": 6.5182783018867925,
      "grad_norm": 2.49550724029541,
      "learning_rate": 8.524601972865586e-06,
      "loss": 0.3039,
      "num_input_tokens_seen": 7296536,
      "step": 11055
    },
    {
      "epoch": 6.52122641509434,
      "grad_norm": 1.883424162864685,
      "learning_rate": 8.522776715167548e-06,
      "loss": 0.5273,
      "num_input_tokens_seen": 7299608,
      "step": 11060
    },
    {
      "epoch": 6.524174528301887,
      "grad_norm": 1.9240996837615967,
      "learning_rate": 8.520950524799192e-06,
      "loss": 0.4288,
      "num_input_tokens_seen": 7303160,
      "step": 11065
    },
    {
      "epoch": 6.527122641509434,
      "grad_norm": 1.7999619245529175,
      "learning_rate": 8.51912340224401e-06,
      "loss": 0.4529,
      "num_input_tokens_seen": 7305784,
      "step": 11070
    },
    {
      "epoch": 6.530070754716981,
      "grad_norm": 1.8529026508331299,
      "learning_rate": 8.51729534798574e-06,
      "loss": 0.387,
      "num_input_tokens_seen": 7309912,
      "step": 11075
    },
    {
      "epoch": 6.533018867924528,
      "grad_norm": 2.1299095153808594,
      "learning_rate": 8.515466362508369e-06,
      "loss": 0.3805,
      "num_input_tokens_seen": 7313496,
      "step": 11080
    },
    {
      "epoch": 6.535966981132075,
      "grad_norm": 2.700831890106201,
      "learning_rate": 8.513636446296125e-06,
      "loss": 0.5176,
      "num_input_tokens_seen": 7316920,
      "step": 11085
    },
    {
      "epoch": 6.538915094339623,
      "grad_norm": 1.1184563636779785,
      "learning_rate": 8.51180559983349e-06,
      "loss": 0.3954,
      "num_input_tokens_seen": 7319864,
      "step": 11090
    },
    {
      "epoch": 6.54186320754717,
      "grad_norm": 2.06292986869812,
      "learning_rate": 8.50997382360519e-06,
      "loss": 0.5401,
      "num_input_tokens_seen": 7324536,
      "step": 11095
    },
    {
      "epoch": 6.544811320754717,
      "grad_norm": 1.4979579448699951,
      "learning_rate": 8.508141118096191e-06,
      "loss": 0.3527,
      "num_input_tokens_seen": 7327768,
      "step": 11100
    },
    {
      "epoch": 6.5477594339622645,
      "grad_norm": 1.2651081085205078,
      "learning_rate": 8.506307483791712e-06,
      "loss": 0.4778,
      "num_input_tokens_seen": 7330872,
      "step": 11105
    },
    {
      "epoch": 6.550707547169811,
      "grad_norm": 2.1303884983062744,
      "learning_rate": 8.504472921177215e-06,
      "loss": 0.434,
      "num_input_tokens_seen": 7333368,
      "step": 11110
    },
    {
      "epoch": 6.553655660377358,
      "grad_norm": 24.833770751953125,
      "learning_rate": 8.502637430738409e-06,
      "loss": 0.4674,
      "num_input_tokens_seen": 7336952,
      "step": 11115
    },
    {
      "epoch": 6.556603773584905,
      "grad_norm": 1.7363064289093018,
      "learning_rate": 8.500801012961248e-06,
      "loss": 0.3752,
      "num_input_tokens_seen": 7340440,
      "step": 11120
    },
    {
      "epoch": 6.559551886792453,
      "grad_norm": 3.322340488433838,
      "learning_rate": 8.49896366833193e-06,
      "loss": 0.4025,
      "num_input_tokens_seen": 7342648,
      "step": 11125
    },
    {
      "epoch": 6.5625,
      "grad_norm": 1.27372407913208,
      "learning_rate": 8.497125397336903e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 7346200,
      "step": 11130
    },
    {
      "epoch": 6.565448113207547,
      "grad_norm": 1.8577194213867188,
      "learning_rate": 8.495286200462854e-06,
      "loss": 0.4314,
      "num_input_tokens_seen": 7350840,
      "step": 11135
    },
    {
      "epoch": 6.568396226415095,
      "grad_norm": 2.2137818336486816,
      "learning_rate": 8.49344607819672e-06,
      "loss": 0.4958,
      "num_input_tokens_seen": 7353432,
      "step": 11140
    },
    {
      "epoch": 6.571344339622642,
      "grad_norm": 4.457449913024902,
      "learning_rate": 8.49160503102568e-06,
      "loss": 0.4229,
      "num_input_tokens_seen": 7356216,
      "step": 11145
    },
    {
      "epoch": 6.574292452830189,
      "grad_norm": 4.778475284576416,
      "learning_rate": 8.489763059437161e-06,
      "loss": 0.4368,
      "num_input_tokens_seen": 7358904,
      "step": 11150
    },
    {
      "epoch": 6.5772405660377355,
      "grad_norm": 1.4943903684616089,
      "learning_rate": 8.487920163918833e-06,
      "loss": 0.3649,
      "num_input_tokens_seen": 7361944,
      "step": 11155
    },
    {
      "epoch": 6.580188679245283,
      "grad_norm": 2.475358009338379,
      "learning_rate": 8.486076344958607e-06,
      "loss": 0.6176,
      "num_input_tokens_seen": 7366136,
      "step": 11160
    },
    {
      "epoch": 6.58313679245283,
      "grad_norm": 1.657072901725769,
      "learning_rate": 8.484231603044647e-06,
      "loss": 0.3394,
      "num_input_tokens_seen": 7370424,
      "step": 11165
    },
    {
      "epoch": 6.586084905660377,
      "grad_norm": 2.5224525928497314,
      "learning_rate": 8.482385938665352e-06,
      "loss": 0.4776,
      "num_input_tokens_seen": 7373912,
      "step": 11170
    },
    {
      "epoch": 6.589033018867925,
      "grad_norm": 1.422773003578186,
      "learning_rate": 8.480539352309373e-06,
      "loss": 0.4372,
      "num_input_tokens_seen": 7377368,
      "step": 11175
    },
    {
      "epoch": 6.591981132075472,
      "grad_norm": 1.894020915031433,
      "learning_rate": 8.478691844465598e-06,
      "loss": 0.5315,
      "num_input_tokens_seen": 7380280,
      "step": 11180
    },
    {
      "epoch": 6.594929245283019,
      "grad_norm": 1.849149465560913,
      "learning_rate": 8.476843415623168e-06,
      "loss": 0.469,
      "num_input_tokens_seen": 7385016,
      "step": 11185
    },
    {
      "epoch": 6.597877358490566,
      "grad_norm": 2.407217025756836,
      "learning_rate": 8.474994066271458e-06,
      "loss": 0.4901,
      "num_input_tokens_seen": 7388312,
      "step": 11190
    },
    {
      "epoch": 6.600825471698113,
      "grad_norm": 2.296215295791626,
      "learning_rate": 8.473143796900089e-06,
      "loss": 0.2955,
      "num_input_tokens_seen": 7391288,
      "step": 11195
    },
    {
      "epoch": 6.60377358490566,
      "grad_norm": 2.221813917160034,
      "learning_rate": 8.471292607998936e-06,
      "loss": 0.4492,
      "num_input_tokens_seen": 7397304,
      "step": 11200
    },
    {
      "epoch": 6.6067216981132075,
      "grad_norm": 2.2313129901885986,
      "learning_rate": 8.469440500058104e-06,
      "loss": 0.4804,
      "num_input_tokens_seen": 7400056,
      "step": 11205
    },
    {
      "epoch": 6.609669811320755,
      "grad_norm": 2.6479995250701904,
      "learning_rate": 8.467587473567945e-06,
      "loss": 0.4484,
      "num_input_tokens_seen": 7402872,
      "step": 11210
    },
    {
      "epoch": 6.612617924528302,
      "grad_norm": 1.5647809505462646,
      "learning_rate": 8.46573352901906e-06,
      "loss": 0.4644,
      "num_input_tokens_seen": 7406552,
      "step": 11215
    },
    {
      "epoch": 6.615566037735849,
      "grad_norm": 3.20453143119812,
      "learning_rate": 8.463878666902286e-06,
      "loss": 0.4456,
      "num_input_tokens_seen": 7409400,
      "step": 11220
    },
    {
      "epoch": 6.618514150943396,
      "grad_norm": 2.3809099197387695,
      "learning_rate": 8.462022887708706e-06,
      "loss": 0.4297,
      "num_input_tokens_seen": 7412344,
      "step": 11225
    },
    {
      "epoch": 6.621462264150943,
      "grad_norm": 1.7990726232528687,
      "learning_rate": 8.460166191929646e-06,
      "loss": 0.393,
      "num_input_tokens_seen": 7415032,
      "step": 11230
    },
    {
      "epoch": 6.62441037735849,
      "grad_norm": 2.587315559387207,
      "learning_rate": 8.458308580056675e-06,
      "loss": 0.4511,
      "num_input_tokens_seen": 7418904,
      "step": 11235
    },
    {
      "epoch": 6.627358490566038,
      "grad_norm": 1.9690241813659668,
      "learning_rate": 8.456450052581602e-06,
      "loss": 0.336,
      "num_input_tokens_seen": 7421528,
      "step": 11240
    },
    {
      "epoch": 6.630306603773585,
      "grad_norm": 1.9113131761550903,
      "learning_rate": 8.45459060999648e-06,
      "loss": 0.3695,
      "num_input_tokens_seen": 7424248,
      "step": 11245
    },
    {
      "epoch": 6.633254716981132,
      "grad_norm": 2.598909616470337,
      "learning_rate": 8.452730252793608e-06,
      "loss": 0.3543,
      "num_input_tokens_seen": 7427384,
      "step": 11250
    },
    {
      "epoch": 6.6362028301886795,
      "grad_norm": 2.502568006515503,
      "learning_rate": 8.450868981465519e-06,
      "loss": 0.4579,
      "num_input_tokens_seen": 7429688,
      "step": 11255
    },
    {
      "epoch": 6.639150943396227,
      "grad_norm": 1.9466094970703125,
      "learning_rate": 8.449006796504997e-06,
      "loss": 0.4476,
      "num_input_tokens_seen": 7432888,
      "step": 11260
    },
    {
      "epoch": 6.642099056603773,
      "grad_norm": 2.793776273727417,
      "learning_rate": 8.44714369840506e-06,
      "loss": 0.4953,
      "num_input_tokens_seen": 7436184,
      "step": 11265
    },
    {
      "epoch": 6.6450471698113205,
      "grad_norm": 1.9508823156356812,
      "learning_rate": 8.445279687658973e-06,
      "loss": 0.4082,
      "num_input_tokens_seen": 7438712,
      "step": 11270
    },
    {
      "epoch": 6.647995283018868,
      "grad_norm": 3.2517547607421875,
      "learning_rate": 8.44341476476024e-06,
      "loss": 0.398,
      "num_input_tokens_seen": 7441496,
      "step": 11275
    },
    {
      "epoch": 6.650943396226415,
      "grad_norm": 2.2211153507232666,
      "learning_rate": 8.441548930202608e-06,
      "loss": 0.5777,
      "num_input_tokens_seen": 7444440,
      "step": 11280
    },
    {
      "epoch": 6.653891509433962,
      "grad_norm": 2.213388681411743,
      "learning_rate": 8.439682184480065e-06,
      "loss": 0.3995,
      "num_input_tokens_seen": 7448344,
      "step": 11285
    },
    {
      "epoch": 6.65683962264151,
      "grad_norm": 1.9117364883422852,
      "learning_rate": 8.437814528086837e-06,
      "loss": 0.4502,
      "num_input_tokens_seen": 7451576,
      "step": 11290
    },
    {
      "epoch": 6.659787735849057,
      "grad_norm": 2.763836622238159,
      "learning_rate": 8.435945961517398e-06,
      "loss": 0.6278,
      "num_input_tokens_seen": 7454840,
      "step": 11295
    },
    {
      "epoch": 6.662735849056604,
      "grad_norm": 2.619485378265381,
      "learning_rate": 8.434076485266458e-06,
      "loss": 0.379,
      "num_input_tokens_seen": 7458392,
      "step": 11300
    },
    {
      "epoch": 6.665683962264151,
      "grad_norm": 1.7167994976043701,
      "learning_rate": 8.432206099828969e-06,
      "loss": 0.387,
      "num_input_tokens_seen": 7462264,
      "step": 11305
    },
    {
      "epoch": 6.668632075471698,
      "grad_norm": 2.148216724395752,
      "learning_rate": 8.430334805700122e-06,
      "loss": 0.5668,
      "num_input_tokens_seen": 7465048,
      "step": 11310
    },
    {
      "epoch": 6.671580188679245,
      "grad_norm": 1.659861445426941,
      "learning_rate": 8.428462603375351e-06,
      "loss": 0.469,
      "num_input_tokens_seen": 7468376,
      "step": 11315
    },
    {
      "epoch": 6.6745283018867925,
      "grad_norm": 1.529465675354004,
      "learning_rate": 8.426589493350332e-06,
      "loss": 0.4044,
      "num_input_tokens_seen": 7471608,
      "step": 11320
    },
    {
      "epoch": 6.67747641509434,
      "grad_norm": 1.5699610710144043,
      "learning_rate": 8.424715476120976e-06,
      "loss": 0.4599,
      "num_input_tokens_seen": 7475576,
      "step": 11325
    },
    {
      "epoch": 6.680424528301887,
      "grad_norm": 1.6448436975479126,
      "learning_rate": 8.422840552183437e-06,
      "loss": 0.4661,
      "num_input_tokens_seen": 7478616,
      "step": 11330
    },
    {
      "epoch": 6.683372641509434,
      "grad_norm": 2.4215006828308105,
      "learning_rate": 8.420964722034111e-06,
      "loss": 0.42,
      "num_input_tokens_seen": 7482936,
      "step": 11335
    },
    {
      "epoch": 6.686320754716981,
      "grad_norm": 3.5841543674468994,
      "learning_rate": 8.41908798616963e-06,
      "loss": 0.3029,
      "num_input_tokens_seen": 7485272,
      "step": 11340
    },
    {
      "epoch": 6.689268867924528,
      "grad_norm": 2.160662889480591,
      "learning_rate": 8.41721034508687e-06,
      "loss": 0.5398,
      "num_input_tokens_seen": 7488344,
      "step": 11345
    },
    {
      "epoch": 6.692216981132075,
      "grad_norm": 2.1798691749572754,
      "learning_rate": 8.415331799282942e-06,
      "loss": 0.4474,
      "num_input_tokens_seen": 7492344,
      "step": 11350
    },
    {
      "epoch": 6.695165094339623,
      "grad_norm": 1.8749724626541138,
      "learning_rate": 8.413452349255205e-06,
      "loss": 0.3848,
      "num_input_tokens_seen": 7496184,
      "step": 11355
    },
    {
      "epoch": 6.69811320754717,
      "grad_norm": 1.8822273015975952,
      "learning_rate": 8.411571995501245e-06,
      "loss": 0.4236,
      "num_input_tokens_seen": 7499768,
      "step": 11360
    },
    {
      "epoch": 6.701061320754717,
      "grad_norm": 1.3169912099838257,
      "learning_rate": 8.409690738518895e-06,
      "loss": 0.4957,
      "num_input_tokens_seen": 7503192,
      "step": 11365
    },
    {
      "epoch": 6.7040094339622645,
      "grad_norm": 3.559767961502075,
      "learning_rate": 8.407808578806229e-06,
      "loss": 0.4984,
      "num_input_tokens_seen": 7506712,
      "step": 11370
    },
    {
      "epoch": 6.706957547169811,
      "grad_norm": 1.8243926763534546,
      "learning_rate": 8.405925516861555e-06,
      "loss": 0.5494,
      "num_input_tokens_seen": 7510680,
      "step": 11375
    },
    {
      "epoch": 6.709905660377358,
      "grad_norm": 2.3624627590179443,
      "learning_rate": 8.40404155318342e-06,
      "loss": 0.3637,
      "num_input_tokens_seen": 7513016,
      "step": 11380
    },
    {
      "epoch": 6.712853773584905,
      "grad_norm": 1.4853438138961792,
      "learning_rate": 8.402156688270613e-06,
      "loss": 0.3957,
      "num_input_tokens_seen": 7516696,
      "step": 11385
    },
    {
      "epoch": 6.715801886792453,
      "grad_norm": 2.080721616744995,
      "learning_rate": 8.400270922622162e-06,
      "loss": 0.4848,
      "num_input_tokens_seen": 7520344,
      "step": 11390
    },
    {
      "epoch": 6.71875,
      "grad_norm": 2.7527647018432617,
      "learning_rate": 8.398384256737328e-06,
      "loss": 0.3306,
      "num_input_tokens_seen": 7524024,
      "step": 11395
    },
    {
      "epoch": 6.721698113207547,
      "grad_norm": 3.116978168487549,
      "learning_rate": 8.396496691115619e-06,
      "loss": 0.4645,
      "num_input_tokens_seen": 7526616,
      "step": 11400
    },
    {
      "epoch": 6.724646226415095,
      "grad_norm": 2.231851816177368,
      "learning_rate": 8.39460822625677e-06,
      "loss": 0.4364,
      "num_input_tokens_seen": 7529400,
      "step": 11405
    },
    {
      "epoch": 6.727594339622642,
      "grad_norm": 3.740666627883911,
      "learning_rate": 8.392718862660765e-06,
      "loss": 0.573,
      "num_input_tokens_seen": 7531864,
      "step": 11410
    },
    {
      "epoch": 6.730542452830189,
      "grad_norm": 1.3972673416137695,
      "learning_rate": 8.390828600827818e-06,
      "loss": 0.4473,
      "num_input_tokens_seen": 7535224,
      "step": 11415
    },
    {
      "epoch": 6.7334905660377355,
      "grad_norm": 2.1909890174865723,
      "learning_rate": 8.388937441258385e-06,
      "loss": 0.425,
      "num_input_tokens_seen": 7539288,
      "step": 11420
    },
    {
      "epoch": 6.736438679245283,
      "grad_norm": 1.8012797832489014,
      "learning_rate": 8.387045384453162e-06,
      "loss": 0.441,
      "num_input_tokens_seen": 7542168,
      "step": 11425
    },
    {
      "epoch": 6.73938679245283,
      "grad_norm": 1.7865604162216187,
      "learning_rate": 8.385152430913073e-06,
      "loss": 0.554,
      "num_input_tokens_seen": 7545304,
      "step": 11430
    },
    {
      "epoch": 6.742334905660377,
      "grad_norm": 2.0888350009918213,
      "learning_rate": 8.383258581139288e-06,
      "loss": 0.4312,
      "num_input_tokens_seen": 7547928,
      "step": 11435
    },
    {
      "epoch": 6.745283018867925,
      "grad_norm": 1.9554308652877808,
      "learning_rate": 8.381363835633213e-06,
      "loss": 0.7171,
      "num_input_tokens_seen": 7551096,
      "step": 11440
    },
    {
      "epoch": 6.748231132075472,
      "grad_norm": 1.8480013608932495,
      "learning_rate": 8.379468194896492e-06,
      "loss": 0.3886,
      "num_input_tokens_seen": 7554712,
      "step": 11445
    },
    {
      "epoch": 6.751179245283019,
      "grad_norm": 1.4941104650497437,
      "learning_rate": 8.377571659431e-06,
      "loss": 0.392,
      "num_input_tokens_seen": 7558136,
      "step": 11450
    },
    {
      "epoch": 6.754127358490566,
      "grad_norm": 1.7909951210021973,
      "learning_rate": 8.375674229738855e-06,
      "loss": 0.5621,
      "num_input_tokens_seen": 7561944,
      "step": 11455
    },
    {
      "epoch": 6.757075471698113,
      "grad_norm": 1.705208420753479,
      "learning_rate": 8.37377590632241e-06,
      "loss": 0.5763,
      "num_input_tokens_seen": 7565528,
      "step": 11460
    },
    {
      "epoch": 6.76002358490566,
      "grad_norm": 1.9498188495635986,
      "learning_rate": 8.371876689684253e-06,
      "loss": 0.3349,
      "num_input_tokens_seen": 7568152,
      "step": 11465
    },
    {
      "epoch": 6.7629716981132075,
      "grad_norm": 4.07738733291626,
      "learning_rate": 8.369976580327211e-06,
      "loss": 0.4289,
      "num_input_tokens_seen": 7570680,
      "step": 11470
    },
    {
      "epoch": 6.765919811320755,
      "grad_norm": 2.308008909225464,
      "learning_rate": 8.368075578754345e-06,
      "loss": 0.4548,
      "num_input_tokens_seen": 7573848,
      "step": 11475
    },
    {
      "epoch": 6.768867924528302,
      "grad_norm": 2.5340194702148438,
      "learning_rate": 8.366173685468952e-06,
      "loss": 0.4193,
      "num_input_tokens_seen": 7576920,
      "step": 11480
    },
    {
      "epoch": 6.771816037735849,
      "grad_norm": 2.4999213218688965,
      "learning_rate": 8.364270900974572e-06,
      "loss": 0.4847,
      "num_input_tokens_seen": 7579768,
      "step": 11485
    },
    {
      "epoch": 6.774764150943396,
      "grad_norm": 2.529285192489624,
      "learning_rate": 8.362367225774968e-06,
      "loss": 0.523,
      "num_input_tokens_seen": 7582424,
      "step": 11490
    },
    {
      "epoch": 6.777712264150943,
      "grad_norm": 2.01870059967041,
      "learning_rate": 8.360462660374153e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 7585688,
      "step": 11495
    },
    {
      "epoch": 6.78066037735849,
      "grad_norm": 2.2806642055511475,
      "learning_rate": 8.358557205276365e-06,
      "loss": 0.4709,
      "num_input_tokens_seen": 7589144,
      "step": 11500
    },
    {
      "epoch": 6.783608490566038,
      "grad_norm": 2.040357828140259,
      "learning_rate": 8.356650860986083e-06,
      "loss": 0.4718,
      "num_input_tokens_seen": 7591704,
      "step": 11505
    },
    {
      "epoch": 6.786556603773585,
      "grad_norm": 2.2827541828155518,
      "learning_rate": 8.354743628008017e-06,
      "loss": 0.4928,
      "num_input_tokens_seen": 7594680,
      "step": 11510
    },
    {
      "epoch": 6.789504716981132,
      "grad_norm": 2.2496840953826904,
      "learning_rate": 8.35283550684712e-06,
      "loss": 0.4124,
      "num_input_tokens_seen": 7597208,
      "step": 11515
    },
    {
      "epoch": 6.7924528301886795,
      "grad_norm": 1.5128735303878784,
      "learning_rate": 8.350926498008572e-06,
      "loss": 0.3846,
      "num_input_tokens_seen": 7601880,
      "step": 11520
    },
    {
      "epoch": 6.795400943396227,
      "grad_norm": 1.6822419166564941,
      "learning_rate": 8.34901660199779e-06,
      "loss": 0.3437,
      "num_input_tokens_seen": 7605208,
      "step": 11525
    },
    {
      "epoch": 6.798349056603773,
      "grad_norm": 1.2152587175369263,
      "learning_rate": 8.347105819320432e-06,
      "loss": 0.3991,
      "num_input_tokens_seen": 7608184,
      "step": 11530
    },
    {
      "epoch": 6.8012971698113205,
      "grad_norm": 2.7006001472473145,
      "learning_rate": 8.34519415048238e-06,
      "loss": 0.4036,
      "num_input_tokens_seen": 7611640,
      "step": 11535
    },
    {
      "epoch": 6.804245283018868,
      "grad_norm": 1.9043618440628052,
      "learning_rate": 8.343281595989761e-06,
      "loss": 0.4668,
      "num_input_tokens_seen": 7614808,
      "step": 11540
    },
    {
      "epoch": 6.807193396226415,
      "grad_norm": 3.0847525596618652,
      "learning_rate": 8.341368156348933e-06,
      "loss": 0.436,
      "num_input_tokens_seen": 7618136,
      "step": 11545
    },
    {
      "epoch": 6.810141509433962,
      "grad_norm": 2.08365535736084,
      "learning_rate": 8.339453832066482e-06,
      "loss": 0.539,
      "num_input_tokens_seen": 7622264,
      "step": 11550
    },
    {
      "epoch": 6.81308962264151,
      "grad_norm": 2.911790132522583,
      "learning_rate": 8.337538623649237e-06,
      "loss": 0.4246,
      "num_input_tokens_seen": 7624600,
      "step": 11555
    },
    {
      "epoch": 6.816037735849057,
      "grad_norm": 2.4462971687316895,
      "learning_rate": 8.33562253160426e-06,
      "loss": 0.4031,
      "num_input_tokens_seen": 7628152,
      "step": 11560
    },
    {
      "epoch": 6.818985849056604,
      "grad_norm": 4.461142539978027,
      "learning_rate": 8.33370555643884e-06,
      "loss": 0.4248,
      "num_input_tokens_seen": 7631896,
      "step": 11565
    },
    {
      "epoch": 6.821933962264151,
      "grad_norm": 2.1225173473358154,
      "learning_rate": 8.331787698660507e-06,
      "loss": 0.3935,
      "num_input_tokens_seen": 7634616,
      "step": 11570
    },
    {
      "epoch": 6.824882075471698,
      "grad_norm": 2.2797019481658936,
      "learning_rate": 8.32986895877702e-06,
      "loss": 0.4515,
      "num_input_tokens_seen": 7637560,
      "step": 11575
    },
    {
      "epoch": 6.827830188679245,
      "grad_norm": 1.3433442115783691,
      "learning_rate": 8.327949337296378e-06,
      "loss": 0.3164,
      "num_input_tokens_seen": 7640696,
      "step": 11580
    },
    {
      "epoch": 6.8307783018867925,
      "grad_norm": 3.602182388305664,
      "learning_rate": 8.326028834726803e-06,
      "loss": 0.4385,
      "num_input_tokens_seen": 7643576,
      "step": 11585
    },
    {
      "epoch": 6.83372641509434,
      "grad_norm": 2.4206478595733643,
      "learning_rate": 8.324107451576762e-06,
      "loss": 0.4273,
      "num_input_tokens_seen": 7646840,
      "step": 11590
    },
    {
      "epoch": 6.836674528301887,
      "grad_norm": 2.5514912605285645,
      "learning_rate": 8.322185188354947e-06,
      "loss": 0.3965,
      "num_input_tokens_seen": 7649848,
      "step": 11595
    },
    {
      "epoch": 6.839622641509434,
      "grad_norm": 3.4101696014404297,
      "learning_rate": 8.320262045570284e-06,
      "loss": 0.4758,
      "num_input_tokens_seen": 7652344,
      "step": 11600
    },
    {
      "epoch": 6.842570754716981,
      "grad_norm": 3.1667909622192383,
      "learning_rate": 8.318338023731937e-06,
      "loss": 0.6088,
      "num_input_tokens_seen": 7654936,
      "step": 11605
    },
    {
      "epoch": 6.845518867924528,
      "grad_norm": 4.260312080383301,
      "learning_rate": 8.316413123349296e-06,
      "loss": 0.4064,
      "num_input_tokens_seen": 7657784,
      "step": 11610
    },
    {
      "epoch": 6.848466981132075,
      "grad_norm": 1.377808690071106,
      "learning_rate": 8.314487344931987e-06,
      "loss": 0.4841,
      "num_input_tokens_seen": 7660920,
      "step": 11615
    },
    {
      "epoch": 6.851415094339623,
      "grad_norm": 2.3571088314056396,
      "learning_rate": 8.31256068898987e-06,
      "loss": 0.4097,
      "num_input_tokens_seen": 7664120,
      "step": 11620
    },
    {
      "epoch": 6.85436320754717,
      "grad_norm": 1.4993841648101807,
      "learning_rate": 8.310633156033032e-06,
      "loss": 0.2865,
      "num_input_tokens_seen": 7667128,
      "step": 11625
    },
    {
      "epoch": 6.857311320754717,
      "grad_norm": 3.467444896697998,
      "learning_rate": 8.3087047465718e-06,
      "loss": 0.5364,
      "num_input_tokens_seen": 7669912,
      "step": 11630
    },
    {
      "epoch": 6.8602594339622645,
      "grad_norm": 2.6692237854003906,
      "learning_rate": 8.306775461116727e-06,
      "loss": 0.4674,
      "num_input_tokens_seen": 7673112,
      "step": 11635
    },
    {
      "epoch": 6.863207547169811,
      "grad_norm": 2.5979092121124268,
      "learning_rate": 8.304845300178597e-06,
      "loss": 0.4549,
      "num_input_tokens_seen": 7676088,
      "step": 11640
    },
    {
      "epoch": 6.866155660377358,
      "grad_norm": 1.4601154327392578,
      "learning_rate": 8.302914264268433e-06,
      "loss": 0.4324,
      "num_input_tokens_seen": 7678936,
      "step": 11645
    },
    {
      "epoch": 6.869103773584905,
      "grad_norm": 3.7053356170654297,
      "learning_rate": 8.300982353897482e-06,
      "loss": 0.6085,
      "num_input_tokens_seen": 7682360,
      "step": 11650
    },
    {
      "epoch": 6.872051886792453,
      "grad_norm": 1.4377856254577637,
      "learning_rate": 8.299049569577226e-06,
      "loss": 0.3583,
      "num_input_tokens_seen": 7685944,
      "step": 11655
    },
    {
      "epoch": 6.875,
      "grad_norm": 1.7450807094573975,
      "learning_rate": 8.297115911819379e-06,
      "loss": 0.4451,
      "num_input_tokens_seen": 7688824,
      "step": 11660
    },
    {
      "epoch": 6.877948113207547,
      "grad_norm": 1.7979644536972046,
      "learning_rate": 8.295181381135884e-06,
      "loss": 0.3842,
      "num_input_tokens_seen": 7692280,
      "step": 11665
    },
    {
      "epoch": 6.880896226415095,
      "grad_norm": 1.9835453033447266,
      "learning_rate": 8.293245978038917e-06,
      "loss": 0.422,
      "num_input_tokens_seen": 7695736,
      "step": 11670
    },
    {
      "epoch": 6.883844339622642,
      "grad_norm": 1.484939694404602,
      "learning_rate": 8.291309703040884e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 7699000,
      "step": 11675
    },
    {
      "epoch": 6.886792452830189,
      "grad_norm": 1.8347632884979248,
      "learning_rate": 8.289372556654422e-06,
      "loss": 0.4415,
      "num_input_tokens_seen": 7701944,
      "step": 11680
    },
    {
      "epoch": 6.8897405660377355,
      "grad_norm": 1.2075819969177246,
      "learning_rate": 8.287434539392401e-06,
      "loss": 0.4749,
      "num_input_tokens_seen": 7706360,
      "step": 11685
    },
    {
      "epoch": 6.892688679245283,
      "grad_norm": 4.052304267883301,
      "learning_rate": 8.285495651767916e-06,
      "loss": 0.3878,
      "num_input_tokens_seen": 7709112,
      "step": 11690
    },
    {
      "epoch": 6.89563679245283,
      "grad_norm": 1.9836914539337158,
      "learning_rate": 8.283555894294297e-06,
      "loss": 0.4794,
      "num_input_tokens_seen": 7715704,
      "step": 11695
    },
    {
      "epoch": 6.898584905660377,
      "grad_norm": 2.315230369567871,
      "learning_rate": 8.281615267485105e-06,
      "loss": 0.4183,
      "num_input_tokens_seen": 7719320,
      "step": 11700
    },
    {
      "epoch": 6.901533018867925,
      "grad_norm": 3.6921606063842773,
      "learning_rate": 8.279673771854127e-06,
      "loss": 0.53,
      "num_input_tokens_seen": 7722808,
      "step": 11705
    },
    {
      "epoch": 6.904481132075472,
      "grad_norm": 5.472204685211182,
      "learning_rate": 8.277731407915386e-06,
      "loss": 0.588,
      "num_input_tokens_seen": 7725848,
      "step": 11710
    },
    {
      "epoch": 6.907429245283019,
      "grad_norm": 2.171933889389038,
      "learning_rate": 8.275788176183126e-06,
      "loss": 0.5166,
      "num_input_tokens_seen": 7729848,
      "step": 11715
    },
    {
      "epoch": 6.910377358490566,
      "grad_norm": 1.3914026021957397,
      "learning_rate": 8.273844077171827e-06,
      "loss": 0.441,
      "num_input_tokens_seen": 7734136,
      "step": 11720
    },
    {
      "epoch": 6.913325471698113,
      "grad_norm": 1.2855972051620483,
      "learning_rate": 8.271899111396202e-06,
      "loss": 0.3997,
      "num_input_tokens_seen": 7737560,
      "step": 11725
    },
    {
      "epoch": 6.91627358490566,
      "grad_norm": 6.333295822143555,
      "learning_rate": 8.269953279371185e-06,
      "loss": 0.6019,
      "num_input_tokens_seen": 7740312,
      "step": 11730
    },
    {
      "epoch": 6.9192216981132075,
      "grad_norm": 3.2596304416656494,
      "learning_rate": 8.268006581611945e-06,
      "loss": 0.3484,
      "num_input_tokens_seen": 7742808,
      "step": 11735
    },
    {
      "epoch": 6.922169811320755,
      "grad_norm": 1.8003157377243042,
      "learning_rate": 8.266059018633878e-06,
      "loss": 0.4711,
      "num_input_tokens_seen": 7746168,
      "step": 11740
    },
    {
      "epoch": 6.925117924528302,
      "grad_norm": 1.733193278312683,
      "learning_rate": 8.264110590952609e-06,
      "loss": 0.3492,
      "num_input_tokens_seen": 7749880,
      "step": 11745
    },
    {
      "epoch": 6.928066037735849,
      "grad_norm": 2.6422455310821533,
      "learning_rate": 8.262161299083993e-06,
      "loss": 0.3994,
      "num_input_tokens_seen": 7752984,
      "step": 11750
    },
    {
      "epoch": 6.931014150943396,
      "grad_norm": 1.5457755327224731,
      "learning_rate": 8.260211143544117e-06,
      "loss": 0.5534,
      "num_input_tokens_seen": 7756312,
      "step": 11755
    },
    {
      "epoch": 6.933962264150943,
      "grad_norm": 1.6360622644424438,
      "learning_rate": 8.258260124849288e-06,
      "loss": 0.4149,
      "num_input_tokens_seen": 7758584,
      "step": 11760
    },
    {
      "epoch": 6.93691037735849,
      "grad_norm": 1.4063552618026733,
      "learning_rate": 8.256308243516048e-06,
      "loss": 0.4997,
      "num_input_tokens_seen": 7762680,
      "step": 11765
    },
    {
      "epoch": 6.939858490566038,
      "grad_norm": 3.1588690280914307,
      "learning_rate": 8.254355500061168e-06,
      "loss": 0.4599,
      "num_input_tokens_seen": 7766552,
      "step": 11770
    },
    {
      "epoch": 6.942806603773585,
      "grad_norm": 1.8129280805587769,
      "learning_rate": 8.252401895001643e-06,
      "loss": 0.3772,
      "num_input_tokens_seen": 7769752,
      "step": 11775
    },
    {
      "epoch": 6.945754716981132,
      "grad_norm": 2.067519426345825,
      "learning_rate": 8.2504474288547e-06,
      "loss": 0.393,
      "num_input_tokens_seen": 7773112,
      "step": 11780
    },
    {
      "epoch": 6.9487028301886795,
      "grad_norm": 1.6338342428207397,
      "learning_rate": 8.248492102137791e-06,
      "loss": 0.4728,
      "num_input_tokens_seen": 7775896,
      "step": 11785
    },
    {
      "epoch": 6.951650943396227,
      "grad_norm": 1.7647852897644043,
      "learning_rate": 8.246535915368596e-06,
      "loss": 0.2882,
      "num_input_tokens_seen": 7778264,
      "step": 11790
    },
    {
      "epoch": 6.954599056603773,
      "grad_norm": 2.8789138793945312,
      "learning_rate": 8.24457886906503e-06,
      "loss": 0.3778,
      "num_input_tokens_seen": 7781048,
      "step": 11795
    },
    {
      "epoch": 6.9575471698113205,
      "grad_norm": 3.2859253883361816,
      "learning_rate": 8.242620963745222e-06,
      "loss": 0.5072,
      "num_input_tokens_seen": 7784152,
      "step": 11800
    },
    {
      "epoch": 6.960495283018868,
      "grad_norm": 8.093803405761719,
      "learning_rate": 8.240662199927538e-06,
      "loss": 0.4466,
      "num_input_tokens_seen": 7787256,
      "step": 11805
    },
    {
      "epoch": 6.963443396226415,
      "grad_norm": 3.597944974899292,
      "learning_rate": 8.238702578130573e-06,
      "loss": 0.5972,
      "num_input_tokens_seen": 7789944,
      "step": 11810
    },
    {
      "epoch": 6.966391509433962,
      "grad_norm": 1.354906439781189,
      "learning_rate": 8.23674209887314e-06,
      "loss": 0.5488,
      "num_input_tokens_seen": 7793752,
      "step": 11815
    },
    {
      "epoch": 6.96933962264151,
      "grad_norm": 1.4744213819503784,
      "learning_rate": 8.234780762674288e-06,
      "loss": 0.3018,
      "num_input_tokens_seen": 7796664,
      "step": 11820
    },
    {
      "epoch": 6.972287735849057,
      "grad_norm": 2.7011330127716064,
      "learning_rate": 8.232818570053286e-06,
      "loss": 0.524,
      "num_input_tokens_seen": 7799352,
      "step": 11825
    },
    {
      "epoch": 6.975235849056604,
      "grad_norm": 2.558635950088501,
      "learning_rate": 8.230855521529637e-06,
      "loss": 0.5217,
      "num_input_tokens_seen": 7802360,
      "step": 11830
    },
    {
      "epoch": 6.978183962264151,
      "grad_norm": 3.0442981719970703,
      "learning_rate": 8.228891617623064e-06,
      "loss": 0.4067,
      "num_input_tokens_seen": 7804952,
      "step": 11835
    },
    {
      "epoch": 6.981132075471698,
      "grad_norm": 2.8908586502075195,
      "learning_rate": 8.22692685885352e-06,
      "loss": 0.4676,
      "num_input_tokens_seen": 7808312,
      "step": 11840
    },
    {
      "epoch": 6.984080188679245,
      "grad_norm": 4.1235737800598145,
      "learning_rate": 8.224961245741183e-06,
      "loss": 0.4633,
      "num_input_tokens_seen": 7811352,
      "step": 11845
    },
    {
      "epoch": 6.9870283018867925,
      "grad_norm": 7.284614086151123,
      "learning_rate": 8.222994778806457e-06,
      "loss": 0.3804,
      "num_input_tokens_seen": 7813976,
      "step": 11850
    },
    {
      "epoch": 6.98997641509434,
      "grad_norm": 2.3300626277923584,
      "learning_rate": 8.221027458569972e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 7816664,
      "step": 11855
    },
    {
      "epoch": 6.992924528301887,
      "grad_norm": 1.6573293209075928,
      "learning_rate": 8.219059285552586e-06,
      "loss": 0.318,
      "num_input_tokens_seen": 7819576,
      "step": 11860
    },
    {
      "epoch": 6.995872641509434,
      "grad_norm": 3.536315441131592,
      "learning_rate": 8.21709026027538e-06,
      "loss": 0.3466,
      "num_input_tokens_seen": 7822616,
      "step": 11865
    },
    {
      "epoch": 6.998820754716981,
      "grad_norm": 1.649599313735962,
      "learning_rate": 8.215120383259664e-06,
      "loss": 0.4409,
      "num_input_tokens_seen": 7825944,
      "step": 11870
    },
    {
      "epoch": 7.001768867924528,
      "grad_norm": 1.331912875175476,
      "learning_rate": 8.21314965502697e-06,
      "loss": 0.3555,
      "num_input_tokens_seen": 7828312,
      "step": 11875
    },
    {
      "epoch": 7.004716981132075,
      "grad_norm": 2.0116474628448486,
      "learning_rate": 8.211178076099056e-06,
      "loss": 0.2281,
      "num_input_tokens_seen": 7834872,
      "step": 11880
    },
    {
      "epoch": 7.007665094339623,
      "grad_norm": 1.8600317239761353,
      "learning_rate": 8.209205646997909e-06,
      "loss": 0.2771,
      "num_input_tokens_seen": 7837528,
      "step": 11885
    },
    {
      "epoch": 7.01061320754717,
      "grad_norm": 1.635477900505066,
      "learning_rate": 8.207232368245735e-06,
      "loss": 0.541,
      "num_input_tokens_seen": 7840088,
      "step": 11890
    },
    {
      "epoch": 7.013561320754717,
      "grad_norm": 2.261821746826172,
      "learning_rate": 8.205258240364968e-06,
      "loss": 0.4539,
      "num_input_tokens_seen": 7843032,
      "step": 11895
    },
    {
      "epoch": 7.0165094339622645,
      "grad_norm": 3.081247091293335,
      "learning_rate": 8.203283263878268e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 7846200,
      "step": 11900
    },
    {
      "epoch": 7.019457547169812,
      "grad_norm": 2.200901746749878,
      "learning_rate": 8.201307439308518e-06,
      "loss": 0.3884,
      "num_input_tokens_seen": 7849144,
      "step": 11905
    },
    {
      "epoch": 7.022405660377358,
      "grad_norm": 1.8020702600479126,
      "learning_rate": 8.199330767178828e-06,
      "loss": 0.4108,
      "num_input_tokens_seen": 7852536,
      "step": 11910
    },
    {
      "epoch": 7.025353773584905,
      "grad_norm": 4.368762016296387,
      "learning_rate": 8.19735324801253e-06,
      "loss": 0.2859,
      "num_input_tokens_seen": 7855480,
      "step": 11915
    },
    {
      "epoch": 7.028301886792453,
      "grad_norm": 1.4919480085372925,
      "learning_rate": 8.195374882333178e-06,
      "loss": 0.4175,
      "num_input_tokens_seen": 7858264,
      "step": 11920
    },
    {
      "epoch": 7.03125,
      "grad_norm": 2.0578794479370117,
      "learning_rate": 8.193395670664555e-06,
      "loss": 0.4256,
      "num_input_tokens_seen": 7860984,
      "step": 11925
    },
    {
      "epoch": 7.034198113207547,
      "grad_norm": 1.8628044128417969,
      "learning_rate": 8.191415613530667e-06,
      "loss": 0.3041,
      "num_input_tokens_seen": 7865400,
      "step": 11930
    },
    {
      "epoch": 7.037146226415095,
      "grad_norm": 1.6800198554992676,
      "learning_rate": 8.189434711455739e-06,
      "loss": 0.5428,
      "num_input_tokens_seen": 7868600,
      "step": 11935
    },
    {
      "epoch": 7.040094339622642,
      "grad_norm": 1.9686099290847778,
      "learning_rate": 8.187452964964226e-06,
      "loss": 0.2429,
      "num_input_tokens_seen": 7871352,
      "step": 11940
    },
    {
      "epoch": 7.043042452830188,
      "grad_norm": 1.6866626739501953,
      "learning_rate": 8.185470374580805e-06,
      "loss": 0.4978,
      "num_input_tokens_seen": 7874520,
      "step": 11945
    },
    {
      "epoch": 7.0459905660377355,
      "grad_norm": 1.575110912322998,
      "learning_rate": 8.183486940830371e-06,
      "loss": 0.5216,
      "num_input_tokens_seen": 7877944,
      "step": 11950
    },
    {
      "epoch": 7.048938679245283,
      "grad_norm": 3.247765064239502,
      "learning_rate": 8.18150266423805e-06,
      "loss": 0.3632,
      "num_input_tokens_seen": 7881112,
      "step": 11955
    },
    {
      "epoch": 7.05188679245283,
      "grad_norm": 3.237342357635498,
      "learning_rate": 8.179517545329188e-06,
      "loss": 0.3467,
      "num_input_tokens_seen": 7884920,
      "step": 11960
    },
    {
      "epoch": 7.054834905660377,
      "grad_norm": 2.1652863025665283,
      "learning_rate": 8.177531584629353e-06,
      "loss": 0.4874,
      "num_input_tokens_seen": 7891160,
      "step": 11965
    },
    {
      "epoch": 7.057783018867925,
      "grad_norm": 1.8061635494232178,
      "learning_rate": 8.175544782664335e-06,
      "loss": 0.3617,
      "num_input_tokens_seen": 7894776,
      "step": 11970
    },
    {
      "epoch": 7.060731132075472,
      "grad_norm": 2.4625208377838135,
      "learning_rate": 8.173557139960151e-06,
      "loss": 0.3546,
      "num_input_tokens_seen": 7897464,
      "step": 11975
    },
    {
      "epoch": 7.063679245283019,
      "grad_norm": 2.872295379638672,
      "learning_rate": 8.17156865704304e-06,
      "loss": 0.3924,
      "num_input_tokens_seen": 7899768,
      "step": 11980
    },
    {
      "epoch": 7.066627358490566,
      "grad_norm": 1.6909000873565674,
      "learning_rate": 8.169579334439453e-06,
      "loss": 0.5694,
      "num_input_tokens_seen": 7902968,
      "step": 11985
    },
    {
      "epoch": 7.069575471698113,
      "grad_norm": 2.326366901397705,
      "learning_rate": 8.16758917267608e-06,
      "loss": 0.3631,
      "num_input_tokens_seen": 7905304,
      "step": 11990
    },
    {
      "epoch": 7.07252358490566,
      "grad_norm": 2.382815361022949,
      "learning_rate": 8.165598172279822e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 7907704,
      "step": 11995
    },
    {
      "epoch": 7.0754716981132075,
      "grad_norm": 1.8243217468261719,
      "learning_rate": 8.163606333777804e-06,
      "loss": 0.4126,
      "num_input_tokens_seen": 7911000,
      "step": 12000
    },
    {
      "epoch": 7.078419811320755,
      "grad_norm": 2.1878278255462646,
      "learning_rate": 8.161613657697374e-06,
      "loss": 0.4333,
      "num_input_tokens_seen": 7913752,
      "step": 12005
    },
    {
      "epoch": 7.081367924528302,
      "grad_norm": 3.0243287086486816,
      "learning_rate": 8.159620144566103e-06,
      "loss": 0.3891,
      "num_input_tokens_seen": 7917240,
      "step": 12010
    },
    {
      "epoch": 7.084316037735849,
      "grad_norm": 2.7944583892822266,
      "learning_rate": 8.157625794911782e-06,
      "loss": 0.5114,
      "num_input_tokens_seen": 7920376,
      "step": 12015
    },
    {
      "epoch": 7.087264150943396,
      "grad_norm": 2.0660669803619385,
      "learning_rate": 8.155630609262424e-06,
      "loss": 0.4484,
      "num_input_tokens_seen": 7923608,
      "step": 12020
    },
    {
      "epoch": 7.090212264150943,
      "grad_norm": 3.4465794563293457,
      "learning_rate": 8.153634588146262e-06,
      "loss": 0.4673,
      "num_input_tokens_seen": 7926936,
      "step": 12025
    },
    {
      "epoch": 7.09316037735849,
      "grad_norm": 3.1670949459075928,
      "learning_rate": 8.15163773209175e-06,
      "loss": 0.3382,
      "num_input_tokens_seen": 7929880,
      "step": 12030
    },
    {
      "epoch": 7.096108490566038,
      "grad_norm": 5.855579376220703,
      "learning_rate": 8.149640041627566e-06,
      "loss": 0.6378,
      "num_input_tokens_seen": 7932824,
      "step": 12035
    },
    {
      "epoch": 7.099056603773585,
      "grad_norm": 2.709357976913452,
      "learning_rate": 8.147641517282608e-06,
      "loss": 0.3601,
      "num_input_tokens_seen": 7936088,
      "step": 12040
    },
    {
      "epoch": 7.102004716981132,
      "grad_norm": 2.8497965335845947,
      "learning_rate": 8.145642159585992e-06,
      "loss": 0.5363,
      "num_input_tokens_seen": 7939672,
      "step": 12045
    },
    {
      "epoch": 7.1049528301886795,
      "grad_norm": 1.5327235460281372,
      "learning_rate": 8.143641969067057e-06,
      "loss": 0.4101,
      "num_input_tokens_seen": 7942840,
      "step": 12050
    },
    {
      "epoch": 7.107900943396227,
      "grad_norm": 5.1803460121154785,
      "learning_rate": 8.141640946255362e-06,
      "loss": 0.4651,
      "num_input_tokens_seen": 7945720,
      "step": 12055
    },
    {
      "epoch": 7.110849056603773,
      "grad_norm": 1.544669270515442,
      "learning_rate": 8.139639091680687e-06,
      "loss": 0.381,
      "num_input_tokens_seen": 7948152,
      "step": 12060
    },
    {
      "epoch": 7.1137971698113205,
      "grad_norm": 4.220897674560547,
      "learning_rate": 8.137636405873031e-06,
      "loss": 0.5678,
      "num_input_tokens_seen": 7951704,
      "step": 12065
    },
    {
      "epoch": 7.116745283018868,
      "grad_norm": 3.654050350189209,
      "learning_rate": 8.135632889362614e-06,
      "loss": 0.4609,
      "num_input_tokens_seen": 7956600,
      "step": 12070
    },
    {
      "epoch": 7.119693396226415,
      "grad_norm": 2.6214230060577393,
      "learning_rate": 8.133628542679879e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 7964120,
      "step": 12075
    },
    {
      "epoch": 7.122641509433962,
      "grad_norm": 1.768797755241394,
      "learning_rate": 8.131623366355478e-06,
      "loss": 0.5668,
      "num_input_tokens_seen": 7967544,
      "step": 12080
    },
    {
      "epoch": 7.12558962264151,
      "grad_norm": 2.5396127700805664,
      "learning_rate": 8.129617360920297e-06,
      "loss": 0.4981,
      "num_input_tokens_seen": 7970264,
      "step": 12085
    },
    {
      "epoch": 7.128537735849057,
      "grad_norm": 2.810141086578369,
      "learning_rate": 8.12761052690543e-06,
      "loss": 0.4953,
      "num_input_tokens_seen": 7973496,
      "step": 12090
    },
    {
      "epoch": 7.131485849056604,
      "grad_norm": 1.9219634532928467,
      "learning_rate": 8.125602864842197e-06,
      "loss": 0.362,
      "num_input_tokens_seen": 7976952,
      "step": 12095
    },
    {
      "epoch": 7.134433962264151,
      "grad_norm": 2.030203104019165,
      "learning_rate": 8.123594375262135e-06,
      "loss": 0.3519,
      "num_input_tokens_seen": 7980088,
      "step": 12100
    },
    {
      "epoch": 7.137382075471698,
      "grad_norm": 2.768420457839966,
      "learning_rate": 8.121585058697e-06,
      "loss": 0.3875,
      "num_input_tokens_seen": 7982648,
      "step": 12105
    },
    {
      "epoch": 7.140330188679245,
      "grad_norm": 3.147123336791992,
      "learning_rate": 8.119574915678767e-06,
      "loss": 0.5038,
      "num_input_tokens_seen": 7985400,
      "step": 12110
    },
    {
      "epoch": 7.1432783018867925,
      "grad_norm": 2.289415121078491,
      "learning_rate": 8.117563946739632e-06,
      "loss": 0.5605,
      "num_input_tokens_seen": 7987928,
      "step": 12115
    },
    {
      "epoch": 7.14622641509434,
      "grad_norm": 2.4484426975250244,
      "learning_rate": 8.115552152412006e-06,
      "loss": 0.4284,
      "num_input_tokens_seen": 7991960,
      "step": 12120
    },
    {
      "epoch": 7.149174528301887,
      "grad_norm": 1.3474708795547485,
      "learning_rate": 8.11353953322852e-06,
      "loss": 0.4275,
      "num_input_tokens_seen": 7996696,
      "step": 12125
    },
    {
      "epoch": 7.152122641509434,
      "grad_norm": 2.6132190227508545,
      "learning_rate": 8.111526089722024e-06,
      "loss": 0.4642,
      "num_input_tokens_seen": 7999576,
      "step": 12130
    },
    {
      "epoch": 7.155070754716981,
      "grad_norm": 2.023289680480957,
      "learning_rate": 8.109511822425586e-06,
      "loss": 0.5544,
      "num_input_tokens_seen": 8004536,
      "step": 12135
    },
    {
      "epoch": 7.158018867924528,
      "grad_norm": 2.222384214401245,
      "learning_rate": 8.107496731872491e-06,
      "loss": 0.4123,
      "num_input_tokens_seen": 8007288,
      "step": 12140
    },
    {
      "epoch": 7.160966981132075,
      "grad_norm": 3.618490219116211,
      "learning_rate": 8.105480818596243e-06,
      "loss": 0.3759,
      "num_input_tokens_seen": 8010456,
      "step": 12145
    },
    {
      "epoch": 7.163915094339623,
      "grad_norm": 1.7266346216201782,
      "learning_rate": 8.103464083130566e-06,
      "loss": 0.287,
      "num_input_tokens_seen": 8013080,
      "step": 12150
    },
    {
      "epoch": 7.16686320754717,
      "grad_norm": 1.2044956684112549,
      "learning_rate": 8.101446526009397e-06,
      "loss": 0.5479,
      "num_input_tokens_seen": 8016056,
      "step": 12155
    },
    {
      "epoch": 7.169811320754717,
      "grad_norm": 3.3956425189971924,
      "learning_rate": 8.099428147766894e-06,
      "loss": 0.3829,
      "num_input_tokens_seen": 8018520,
      "step": 12160
    },
    {
      "epoch": 7.1727594339622645,
      "grad_norm": 1.5670268535614014,
      "learning_rate": 8.097408948937431e-06,
      "loss": 0.3065,
      "num_input_tokens_seen": 8022488,
      "step": 12165
    },
    {
      "epoch": 7.175707547169812,
      "grad_norm": 3.3333423137664795,
      "learning_rate": 8.095388930055599e-06,
      "loss": 0.4271,
      "num_input_tokens_seen": 8025720,
      "step": 12170
    },
    {
      "epoch": 7.178655660377358,
      "grad_norm": 1.9620200395584106,
      "learning_rate": 8.093368091656209e-06,
      "loss": 0.3369,
      "num_input_tokens_seen": 8028856,
      "step": 12175
    },
    {
      "epoch": 7.181603773584905,
      "grad_norm": 1.566028118133545,
      "learning_rate": 8.091346434274284e-06,
      "loss": 0.4204,
      "num_input_tokens_seen": 8032216,
      "step": 12180
    },
    {
      "epoch": 7.184551886792453,
      "grad_norm": 2.1722464561462402,
      "learning_rate": 8.089323958445068e-06,
      "loss": 0.4894,
      "num_input_tokens_seen": 8035608,
      "step": 12185
    },
    {
      "epoch": 7.1875,
      "grad_norm": 1.3422520160675049,
      "learning_rate": 8.08730066470402e-06,
      "loss": 0.4828,
      "num_input_tokens_seen": 8038808,
      "step": 12190
    },
    {
      "epoch": 7.190448113207547,
      "grad_norm": 2.483468532562256,
      "learning_rate": 8.085276553586814e-06,
      "loss": 0.368,
      "num_input_tokens_seen": 8042488,
      "step": 12195
    },
    {
      "epoch": 7.193396226415095,
      "grad_norm": 2.775367021560669,
      "learning_rate": 8.083251625629345e-06,
      "loss": 0.3617,
      "num_input_tokens_seen": 8045144,
      "step": 12200
    },
    {
      "epoch": 7.196344339622642,
      "grad_norm": 3.3825855255126953,
      "learning_rate": 8.08122588136772e-06,
      "loss": 0.4036,
      "num_input_tokens_seen": 8047864,
      "step": 12205
    },
    {
      "epoch": 7.199292452830188,
      "grad_norm": 2.4922938346862793,
      "learning_rate": 8.079199321338262e-06,
      "loss": 0.4021,
      "num_input_tokens_seen": 8050872,
      "step": 12210
    },
    {
      "epoch": 7.2022405660377355,
      "grad_norm": 2.1087450981140137,
      "learning_rate": 8.077171946077516e-06,
      "loss": 0.4112,
      "num_input_tokens_seen": 8053496,
      "step": 12215
    },
    {
      "epoch": 7.205188679245283,
      "grad_norm": 3.952951431274414,
      "learning_rate": 8.075143756122232e-06,
      "loss": 0.3453,
      "num_input_tokens_seen": 8056248,
      "step": 12220
    },
    {
      "epoch": 7.20813679245283,
      "grad_norm": 2.0332934856414795,
      "learning_rate": 8.073114752009388e-06,
      "loss": 0.3943,
      "num_input_tokens_seen": 8059032,
      "step": 12225
    },
    {
      "epoch": 7.211084905660377,
      "grad_norm": 2.531399965286255,
      "learning_rate": 8.071084934276168e-06,
      "loss": 0.3494,
      "num_input_tokens_seen": 8061816,
      "step": 12230
    },
    {
      "epoch": 7.214033018867925,
      "grad_norm": 2.1156930923461914,
      "learning_rate": 8.069054303459976e-06,
      "loss": 0.495,
      "num_input_tokens_seen": 8065176,
      "step": 12235
    },
    {
      "epoch": 7.216981132075472,
      "grad_norm": 2.871216297149658,
      "learning_rate": 8.06702286009843e-06,
      "loss": 0.4357,
      "num_input_tokens_seen": 8068120,
      "step": 12240
    },
    {
      "epoch": 7.219929245283019,
      "grad_norm": 2.4331352710723877,
      "learning_rate": 8.064990604729363e-06,
      "loss": 0.424,
      "num_input_tokens_seen": 8071896,
      "step": 12245
    },
    {
      "epoch": 7.222877358490566,
      "grad_norm": 1.656305193901062,
      "learning_rate": 8.062957537890827e-06,
      "loss": 0.4937,
      "num_input_tokens_seen": 8076504,
      "step": 12250
    },
    {
      "epoch": 7.225825471698113,
      "grad_norm": 2.1300554275512695,
      "learning_rate": 8.060923660121081e-06,
      "loss": 0.3749,
      "num_input_tokens_seen": 8078872,
      "step": 12255
    },
    {
      "epoch": 7.22877358490566,
      "grad_norm": 1.88933527469635,
      "learning_rate": 8.058888971958603e-06,
      "loss": 0.4491,
      "num_input_tokens_seen": 8081848,
      "step": 12260
    },
    {
      "epoch": 7.2317216981132075,
      "grad_norm": 3.227546215057373,
      "learning_rate": 8.056853473942085e-06,
      "loss": 0.3189,
      "num_input_tokens_seen": 8084344,
      "step": 12265
    },
    {
      "epoch": 7.234669811320755,
      "grad_norm": 2.3438096046447754,
      "learning_rate": 8.054817166610438e-06,
      "loss": 0.2589,
      "num_input_tokens_seen": 8086712,
      "step": 12270
    },
    {
      "epoch": 7.237617924528302,
      "grad_norm": 1.5949827432632446,
      "learning_rate": 8.052780050502781e-06,
      "loss": 0.3593,
      "num_input_tokens_seen": 8089848,
      "step": 12275
    },
    {
      "epoch": 7.240566037735849,
      "grad_norm": 2.1761293411254883,
      "learning_rate": 8.050742126158448e-06,
      "loss": 0.4967,
      "num_input_tokens_seen": 8093176,
      "step": 12280
    },
    {
      "epoch": 7.243514150943396,
      "grad_norm": 2.6541483402252197,
      "learning_rate": 8.04870339411699e-06,
      "loss": 0.2922,
      "num_input_tokens_seen": 8097080,
      "step": 12285
    },
    {
      "epoch": 7.246462264150943,
      "grad_norm": 4.628573417663574,
      "learning_rate": 8.046663854918166e-06,
      "loss": 0.6073,
      "num_input_tokens_seen": 8099832,
      "step": 12290
    },
    {
      "epoch": 7.24941037735849,
      "grad_norm": 1.2509291172027588,
      "learning_rate": 8.044623509101959e-06,
      "loss": 0.3052,
      "num_input_tokens_seen": 8102680,
      "step": 12295
    },
    {
      "epoch": 7.252358490566038,
      "grad_norm": 2.6653993129730225,
      "learning_rate": 8.042582357208557e-06,
      "loss": 0.557,
      "num_input_tokens_seen": 8106584,
      "step": 12300
    },
    {
      "epoch": 7.255306603773585,
      "grad_norm": 6.004422187805176,
      "learning_rate": 8.04054039977836e-06,
      "loss": 0.3727,
      "num_input_tokens_seen": 8109688,
      "step": 12305
    },
    {
      "epoch": 7.258254716981132,
      "grad_norm": 1.570355772972107,
      "learning_rate": 8.038497637351992e-06,
      "loss": 0.3327,
      "num_input_tokens_seen": 8112856,
      "step": 12310
    },
    {
      "epoch": 7.2612028301886795,
      "grad_norm": 2.0101826190948486,
      "learning_rate": 8.036454070470276e-06,
      "loss": 0.4205,
      "num_input_tokens_seen": 8115384,
      "step": 12315
    },
    {
      "epoch": 7.264150943396227,
      "grad_norm": 2.650496482849121,
      "learning_rate": 8.03440969967426e-06,
      "loss": 0.3896,
      "num_input_tokens_seen": 8118840,
      "step": 12320
    },
    {
      "epoch": 7.267099056603773,
      "grad_norm": 1.9886906147003174,
      "learning_rate": 8.032364525505198e-06,
      "loss": 0.3146,
      "num_input_tokens_seen": 8122808,
      "step": 12325
    },
    {
      "epoch": 7.2700471698113205,
      "grad_norm": 2.1889562606811523,
      "learning_rate": 8.030318548504561e-06,
      "loss": 0.3651,
      "num_input_tokens_seen": 8125912,
      "step": 12330
    },
    {
      "epoch": 7.272995283018868,
      "grad_norm": 2.1113924980163574,
      "learning_rate": 8.028271769214026e-06,
      "loss": 0.5903,
      "num_input_tokens_seen": 8130008,
      "step": 12335
    },
    {
      "epoch": 7.275943396226415,
      "grad_norm": 3.4248805046081543,
      "learning_rate": 8.02622418817549e-06,
      "loss": 0.3305,
      "num_input_tokens_seen": 8133112,
      "step": 12340
    },
    {
      "epoch": 7.278891509433962,
      "grad_norm": 1.7187933921813965,
      "learning_rate": 8.024175805931056e-06,
      "loss": 0.3108,
      "num_input_tokens_seen": 8137240,
      "step": 12345
    },
    {
      "epoch": 7.28183962264151,
      "grad_norm": 2.284886360168457,
      "learning_rate": 8.022126623023045e-06,
      "loss": 0.4009,
      "num_input_tokens_seen": 8140216,
      "step": 12350
    },
    {
      "epoch": 7.284787735849057,
      "grad_norm": 2.1870195865631104,
      "learning_rate": 8.020076639993987e-06,
      "loss": 0.3393,
      "num_input_tokens_seen": 8143256,
      "step": 12355
    },
    {
      "epoch": 7.287735849056604,
      "grad_norm": 2.631944179534912,
      "learning_rate": 8.018025857386622e-06,
      "loss": 0.3125,
      "num_input_tokens_seen": 8146648,
      "step": 12360
    },
    {
      "epoch": 7.290683962264151,
      "grad_norm": 2.1599678993225098,
      "learning_rate": 8.015974275743905e-06,
      "loss": 0.4056,
      "num_input_tokens_seen": 8149656,
      "step": 12365
    },
    {
      "epoch": 7.293632075471698,
      "grad_norm": 2.1827023029327393,
      "learning_rate": 8.013921895609e-06,
      "loss": 0.4872,
      "num_input_tokens_seen": 8152856,
      "step": 12370
    },
    {
      "epoch": 7.296580188679245,
      "grad_norm": 3.411473035812378,
      "learning_rate": 8.011868717525283e-06,
      "loss": 0.4503,
      "num_input_tokens_seen": 8155512,
      "step": 12375
    },
    {
      "epoch": 7.2995283018867925,
      "grad_norm": 1.2947078943252563,
      "learning_rate": 8.009814742036343e-06,
      "loss": 0.4159,
      "num_input_tokens_seen": 8158584,
      "step": 12380
    },
    {
      "epoch": 7.30247641509434,
      "grad_norm": 3.251697540283203,
      "learning_rate": 8.007759969685979e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 8161912,
      "step": 12385
    },
    {
      "epoch": 7.305424528301887,
      "grad_norm": 3.147472620010376,
      "learning_rate": 8.005704401018199e-06,
      "loss": 0.4315,
      "num_input_tokens_seen": 8165080,
      "step": 12390
    },
    {
      "epoch": 7.308372641509434,
      "grad_norm": 2.551408052444458,
      "learning_rate": 8.003648036577226e-06,
      "loss": 0.3556,
      "num_input_tokens_seen": 8169240,
      "step": 12395
    },
    {
      "epoch": 7.311320754716981,
      "grad_norm": 3.5299503803253174,
      "learning_rate": 8.00159087690749e-06,
      "loss": 0.3136,
      "num_input_tokens_seen": 8171544,
      "step": 12400
    },
    {
      "epoch": 7.314268867924528,
      "grad_norm": 2.5902328491210938,
      "learning_rate": 7.999532922553635e-06,
      "loss": 0.3956,
      "num_input_tokens_seen": 8174712,
      "step": 12405
    },
    {
      "epoch": 7.317216981132075,
      "grad_norm": 1.2297993898391724,
      "learning_rate": 7.997474174060508e-06,
      "loss": 0.3863,
      "num_input_tokens_seen": 8179192,
      "step": 12410
    },
    {
      "epoch": 7.320165094339623,
      "grad_norm": 2.5044496059417725,
      "learning_rate": 7.995414631973179e-06,
      "loss": 0.4067,
      "num_input_tokens_seen": 8183416,
      "step": 12415
    },
    {
      "epoch": 7.32311320754717,
      "grad_norm": 4.098498344421387,
      "learning_rate": 7.993354296836914e-06,
      "loss": 0.3012,
      "num_input_tokens_seen": 8186648,
      "step": 12420
    },
    {
      "epoch": 7.326061320754717,
      "grad_norm": 2.678738594055176,
      "learning_rate": 7.991293169197198e-06,
      "loss": 0.4547,
      "num_input_tokens_seen": 8189912,
      "step": 12425
    },
    {
      "epoch": 7.3290094339622645,
      "grad_norm": 3.098851203918457,
      "learning_rate": 7.989231249599725e-06,
      "loss": 0.4129,
      "num_input_tokens_seen": 8193496,
      "step": 12430
    },
    {
      "epoch": 7.331957547169811,
      "grad_norm": 1.5951907634735107,
      "learning_rate": 7.987168538590395e-06,
      "loss": 0.3314,
      "num_input_tokens_seen": 8196376,
      "step": 12435
    },
    {
      "epoch": 7.334905660377358,
      "grad_norm": 1.556486964225769,
      "learning_rate": 7.985105036715322e-06,
      "loss": 0.4108,
      "num_input_tokens_seen": 8200024,
      "step": 12440
    },
    {
      "epoch": 7.337853773584905,
      "grad_norm": 0.47100725769996643,
      "learning_rate": 7.983040744520823e-06,
      "loss": 0.4297,
      "num_input_tokens_seen": 8205080,
      "step": 12445
    },
    {
      "epoch": 7.340801886792453,
      "grad_norm": 1.6439805030822754,
      "learning_rate": 7.980975662553432e-06,
      "loss": 0.3954,
      "num_input_tokens_seen": 8208248,
      "step": 12450
    },
    {
      "epoch": 7.34375,
      "grad_norm": 1.765467643737793,
      "learning_rate": 7.978909791359888e-06,
      "loss": 0.3764,
      "num_input_tokens_seen": 8211896,
      "step": 12455
    },
    {
      "epoch": 7.346698113207547,
      "grad_norm": 2.6935784816741943,
      "learning_rate": 7.976843131487136e-06,
      "loss": 0.3915,
      "num_input_tokens_seen": 8214616,
      "step": 12460
    },
    {
      "epoch": 7.349646226415095,
      "grad_norm": 4.338754177093506,
      "learning_rate": 7.974775683482337e-06,
      "loss": 0.4368,
      "num_input_tokens_seen": 8217976,
      "step": 12465
    },
    {
      "epoch": 7.352594339622642,
      "grad_norm": 2.032939910888672,
      "learning_rate": 7.972707447892855e-06,
      "loss": 0.4141,
      "num_input_tokens_seen": 8221624,
      "step": 12470
    },
    {
      "epoch": 7.355542452830189,
      "grad_norm": 2.2449803352355957,
      "learning_rate": 7.970638425266264e-06,
      "loss": 0.2845,
      "num_input_tokens_seen": 8224632,
      "step": 12475
    },
    {
      "epoch": 7.3584905660377355,
      "grad_norm": 3.91483998298645,
      "learning_rate": 7.968568616150349e-06,
      "loss": 0.4526,
      "num_input_tokens_seen": 8228824,
      "step": 12480
    },
    {
      "epoch": 7.361438679245283,
      "grad_norm": 5.903486728668213,
      "learning_rate": 7.966498021093096e-06,
      "loss": 0.5487,
      "num_input_tokens_seen": 8231736,
      "step": 12485
    },
    {
      "epoch": 7.36438679245283,
      "grad_norm": 1.343704104423523,
      "learning_rate": 7.96442664064271e-06,
      "loss": 0.3745,
      "num_input_tokens_seen": 8235000,
      "step": 12490
    },
    {
      "epoch": 7.367334905660377,
      "grad_norm": 1.858373761177063,
      "learning_rate": 7.962354475347593e-06,
      "loss": 0.4795,
      "num_input_tokens_seen": 8238328,
      "step": 12495
    },
    {
      "epoch": 7.370283018867925,
      "grad_norm": 3.809966564178467,
      "learning_rate": 7.960281525756364e-06,
      "loss": 0.5202,
      "num_input_tokens_seen": 8242136,
      "step": 12500
    },
    {
      "epoch": 7.373231132075472,
      "grad_norm": 2.2396979331970215,
      "learning_rate": 7.95820779241784e-06,
      "loss": 0.3696,
      "num_input_tokens_seen": 8245496,
      "step": 12505
    },
    {
      "epoch": 7.376179245283019,
      "grad_norm": 1.9255937337875366,
      "learning_rate": 7.956133275881055e-06,
      "loss": 0.5257,
      "num_input_tokens_seen": 8249560,
      "step": 12510
    },
    {
      "epoch": 7.379127358490566,
      "grad_norm": 1.786914348602295,
      "learning_rate": 7.954057976695244e-06,
      "loss": 0.2659,
      "num_input_tokens_seen": 8252920,
      "step": 12515
    },
    {
      "epoch": 7.382075471698113,
      "grad_norm": 1.8643625974655151,
      "learning_rate": 7.951981895409854e-06,
      "loss": 0.418,
      "num_input_tokens_seen": 8256792,
      "step": 12520
    },
    {
      "epoch": 7.38502358490566,
      "grad_norm": 2.1549248695373535,
      "learning_rate": 7.949905032574534e-06,
      "loss": 0.3879,
      "num_input_tokens_seen": 8260088,
      "step": 12525
    },
    {
      "epoch": 7.3879716981132075,
      "grad_norm": 2.056917905807495,
      "learning_rate": 7.947827388739145e-06,
      "loss": 0.346,
      "num_input_tokens_seen": 8263096,
      "step": 12530
    },
    {
      "epoch": 7.390919811320755,
      "grad_norm": 2.9158382415771484,
      "learning_rate": 7.945748964453747e-06,
      "loss": 0.4257,
      "num_input_tokens_seen": 8267736,
      "step": 12535
    },
    {
      "epoch": 7.393867924528302,
      "grad_norm": 1.6099978685379028,
      "learning_rate": 7.943669760268618e-06,
      "loss": 0.3682,
      "num_input_tokens_seen": 8270296,
      "step": 12540
    },
    {
      "epoch": 7.396816037735849,
      "grad_norm": 2.1024599075317383,
      "learning_rate": 7.941589776734232e-06,
      "loss": 0.3684,
      "num_input_tokens_seen": 8274232,
      "step": 12545
    },
    {
      "epoch": 7.399764150943396,
      "grad_norm": 3.3841159343719482,
      "learning_rate": 7.939509014401277e-06,
      "loss": 0.4721,
      "num_input_tokens_seen": 8277240,
      "step": 12550
    },
    {
      "epoch": 7.402712264150943,
      "grad_norm": 1.8726433515548706,
      "learning_rate": 7.93742747382064e-06,
      "loss": 0.4506,
      "num_input_tokens_seen": 8280664,
      "step": 12555
    },
    {
      "epoch": 7.40566037735849,
      "grad_norm": 3.5017809867858887,
      "learning_rate": 7.93534515554342e-06,
      "loss": 0.4691,
      "num_input_tokens_seen": 8282968,
      "step": 12560
    },
    {
      "epoch": 7.408608490566038,
      "grad_norm": 2.4666459560394287,
      "learning_rate": 7.933262060120918e-06,
      "loss": 0.3627,
      "num_input_tokens_seen": 8285880,
      "step": 12565
    },
    {
      "epoch": 7.411556603773585,
      "grad_norm": 3.5416204929351807,
      "learning_rate": 7.931178188104646e-06,
      "loss": 0.5136,
      "num_input_tokens_seen": 8288440,
      "step": 12570
    },
    {
      "epoch": 7.414504716981132,
      "grad_norm": 2.024329662322998,
      "learning_rate": 7.929093540046317e-06,
      "loss": 0.3408,
      "num_input_tokens_seen": 8291576,
      "step": 12575
    },
    {
      "epoch": 7.4174528301886795,
      "grad_norm": 1.8266892433166504,
      "learning_rate": 7.927008116497848e-06,
      "loss": 0.3392,
      "num_input_tokens_seen": 8294552,
      "step": 12580
    },
    {
      "epoch": 7.420400943396227,
      "grad_norm": 2.4067749977111816,
      "learning_rate": 7.924921918011366e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 8297176,
      "step": 12585
    },
    {
      "epoch": 7.423349056603773,
      "grad_norm": 2.7393081188201904,
      "learning_rate": 7.9228349451392e-06,
      "loss": 0.618,
      "num_input_tokens_seen": 8301208,
      "step": 12590
    },
    {
      "epoch": 7.4262971698113205,
      "grad_norm": 1.9342297315597534,
      "learning_rate": 7.920747198433884e-06,
      "loss": 0.4359,
      "num_input_tokens_seen": 8304920,
      "step": 12595
    },
    {
      "epoch": 7.429245283018868,
      "grad_norm": 1.825609803199768,
      "learning_rate": 7.91865867844816e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 8307448,
      "step": 12600
    },
    {
      "epoch": 7.432193396226415,
      "grad_norm": 3.7735965251922607,
      "learning_rate": 7.916569385734976e-06,
      "loss": 0.3848,
      "num_input_tokens_seen": 8310840,
      "step": 12605
    },
    {
      "epoch": 7.435141509433962,
      "grad_norm": 1.7836674451828003,
      "learning_rate": 7.914479320847474e-06,
      "loss": 0.4,
      "num_input_tokens_seen": 8314136,
      "step": 12610
    },
    {
      "epoch": 7.43808962264151,
      "grad_norm": 1.4542292356491089,
      "learning_rate": 7.912388484339012e-06,
      "loss": 0.2877,
      "num_input_tokens_seen": 8318232,
      "step": 12615
    },
    {
      "epoch": 7.441037735849057,
      "grad_norm": 4.058559417724609,
      "learning_rate": 7.910296876763147e-06,
      "loss": 0.6555,
      "num_input_tokens_seen": 8321080,
      "step": 12620
    },
    {
      "epoch": 7.443985849056604,
      "grad_norm": 1.8675956726074219,
      "learning_rate": 7.90820449867364e-06,
      "loss": 0.4273,
      "num_input_tokens_seen": 8323928,
      "step": 12625
    },
    {
      "epoch": 7.446933962264151,
      "grad_norm": 1.363814353942871,
      "learning_rate": 7.90611135062446e-06,
      "loss": 0.5016,
      "num_input_tokens_seen": 8327640,
      "step": 12630
    },
    {
      "epoch": 7.449882075471698,
      "grad_norm": 2.4247448444366455,
      "learning_rate": 7.904017433169775e-06,
      "loss": 0.4359,
      "num_input_tokens_seen": 8332088,
      "step": 12635
    },
    {
      "epoch": 7.452830188679245,
      "grad_norm": 2.7405014038085938,
      "learning_rate": 7.901922746863957e-06,
      "loss": 0.4862,
      "num_input_tokens_seen": 8335192,
      "step": 12640
    },
    {
      "epoch": 7.4557783018867925,
      "grad_norm": 2.4757797718048096,
      "learning_rate": 7.899827292261589e-06,
      "loss": 0.4067,
      "num_input_tokens_seen": 8337944,
      "step": 12645
    },
    {
      "epoch": 7.45872641509434,
      "grad_norm": 2.463571548461914,
      "learning_rate": 7.897731069917444e-06,
      "loss": 0.3311,
      "num_input_tokens_seen": 8341208,
      "step": 12650
    },
    {
      "epoch": 7.461674528301887,
      "grad_norm": 2.851022720336914,
      "learning_rate": 7.895634080386512e-06,
      "loss": 0.3819,
      "num_input_tokens_seen": 8344376,
      "step": 12655
    },
    {
      "epoch": 7.464622641509434,
      "grad_norm": 1.6841213703155518,
      "learning_rate": 7.893536324223977e-06,
      "loss": 0.4465,
      "num_input_tokens_seen": 8347736,
      "step": 12660
    },
    {
      "epoch": 7.467570754716981,
      "grad_norm": 1.3044652938842773,
      "learning_rate": 7.89143780198523e-06,
      "loss": 0.3517,
      "num_input_tokens_seen": 8351992,
      "step": 12665
    },
    {
      "epoch": 7.470518867924528,
      "grad_norm": 2.006349802017212,
      "learning_rate": 7.889338514225862e-06,
      "loss": 0.4466,
      "num_input_tokens_seen": 8354968,
      "step": 12670
    },
    {
      "epoch": 7.473466981132075,
      "grad_norm": 2.305638313293457,
      "learning_rate": 7.887238461501671e-06,
      "loss": 0.4257,
      "num_input_tokens_seen": 8358360,
      "step": 12675
    },
    {
      "epoch": 7.476415094339623,
      "grad_norm": 1.2876256704330444,
      "learning_rate": 7.885137644368654e-06,
      "loss": 0.3245,
      "num_input_tokens_seen": 8361496,
      "step": 12680
    },
    {
      "epoch": 7.47936320754717,
      "grad_norm": 2.2189438343048096,
      "learning_rate": 7.883036063383012e-06,
      "loss": 0.3441,
      "num_input_tokens_seen": 8364344,
      "step": 12685
    },
    {
      "epoch": 7.482311320754717,
      "grad_norm": 1.9343363046646118,
      "learning_rate": 7.880933719101148e-06,
      "loss": 0.4532,
      "num_input_tokens_seen": 8368344,
      "step": 12690
    },
    {
      "epoch": 7.4852594339622645,
      "grad_norm": 1.899059534072876,
      "learning_rate": 7.878830612079664e-06,
      "loss": 0.3592,
      "num_input_tokens_seen": 8371224,
      "step": 12695
    },
    {
      "epoch": 7.488207547169811,
      "grad_norm": 1.8963152170181274,
      "learning_rate": 7.876726742875369e-06,
      "loss": 0.4829,
      "num_input_tokens_seen": 8375064,
      "step": 12700
    },
    {
      "epoch": 7.491155660377358,
      "grad_norm": 1.3428306579589844,
      "learning_rate": 7.874622112045269e-06,
      "loss": 0.3579,
      "num_input_tokens_seen": 8378840,
      "step": 12705
    },
    {
      "epoch": 7.494103773584905,
      "grad_norm": 1.5304847955703735,
      "learning_rate": 7.872516720146578e-06,
      "loss": 0.4134,
      "num_input_tokens_seen": 8382872,
      "step": 12710
    },
    {
      "epoch": 7.497051886792453,
      "grad_norm": 1.9032882452011108,
      "learning_rate": 7.870410567736705e-06,
      "loss": 0.3694,
      "num_input_tokens_seen": 8385464,
      "step": 12715
    },
    {
      "epoch": 7.5,
      "grad_norm": 1.5629794597625732,
      "learning_rate": 7.868303655373264e-06,
      "loss": 0.4036,
      "num_input_tokens_seen": 8388856,
      "step": 12720
    },
    {
      "epoch": 7.502948113207547,
      "grad_norm": 1.8683295249938965,
      "learning_rate": 7.866195983614066e-06,
      "loss": 0.4558,
      "num_input_tokens_seen": 8391832,
      "step": 12725
    },
    {
      "epoch": 7.505896226415095,
      "grad_norm": 1.6583585739135742,
      "learning_rate": 7.864087553017133e-06,
      "loss": 0.3876,
      "num_input_tokens_seen": 8395544,
      "step": 12730
    },
    {
      "epoch": 7.508844339622642,
      "grad_norm": 3.622056484222412,
      "learning_rate": 7.861978364140674e-06,
      "loss": 0.4424,
      "num_input_tokens_seen": 8398840,
      "step": 12735
    },
    {
      "epoch": 7.511792452830189,
      "grad_norm": 3.292670965194702,
      "learning_rate": 7.859868417543109e-06,
      "loss": 0.4938,
      "num_input_tokens_seen": 8401688,
      "step": 12740
    },
    {
      "epoch": 7.5147405660377355,
      "grad_norm": 1.4479866027832031,
      "learning_rate": 7.857757713783055e-06,
      "loss": 0.3881,
      "num_input_tokens_seen": 8406520,
      "step": 12745
    },
    {
      "epoch": 7.517688679245283,
      "grad_norm": 2.248171806335449,
      "learning_rate": 7.855646253419331e-06,
      "loss": 0.4287,
      "num_input_tokens_seen": 8410360,
      "step": 12750
    },
    {
      "epoch": 7.52063679245283,
      "grad_norm": 2.227574586868286,
      "learning_rate": 7.853534037010952e-06,
      "loss": 0.4079,
      "num_input_tokens_seen": 8413848,
      "step": 12755
    },
    {
      "epoch": 7.523584905660377,
      "grad_norm": 1.8479485511779785,
      "learning_rate": 7.851421065117142e-06,
      "loss": 0.478,
      "num_input_tokens_seen": 8418232,
      "step": 12760
    },
    {
      "epoch": 7.526533018867925,
      "grad_norm": 1.7565211057662964,
      "learning_rate": 7.849307338297314e-06,
      "loss": 0.4866,
      "num_input_tokens_seen": 8422360,
      "step": 12765
    },
    {
      "epoch": 7.529481132075472,
      "grad_norm": 4.349031925201416,
      "learning_rate": 7.847192857111087e-06,
      "loss": 0.3786,
      "num_input_tokens_seen": 8426040,
      "step": 12770
    },
    {
      "epoch": 7.532429245283019,
      "grad_norm": 1.8110990524291992,
      "learning_rate": 7.845077622118282e-06,
      "loss": 0.3595,
      "num_input_tokens_seen": 8428536,
      "step": 12775
    },
    {
      "epoch": 7.535377358490566,
      "grad_norm": 1.6380923986434937,
      "learning_rate": 7.842961633878916e-06,
      "loss": 0.4823,
      "num_input_tokens_seen": 8431736,
      "step": 12780
    },
    {
      "epoch": 7.538325471698113,
      "grad_norm": 1.9977571964263916,
      "learning_rate": 7.840844892953204e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 8435096,
      "step": 12785
    },
    {
      "epoch": 7.54127358490566,
      "grad_norm": 3.0037224292755127,
      "learning_rate": 7.838727399901562e-06,
      "loss": 0.3923,
      "num_input_tokens_seen": 8438072,
      "step": 12790
    },
    {
      "epoch": 7.5442216981132075,
      "grad_norm": 2.760622262954712,
      "learning_rate": 7.836609155284607e-06,
      "loss": 0.4602,
      "num_input_tokens_seen": 8441432,
      "step": 12795
    },
    {
      "epoch": 7.547169811320755,
      "grad_norm": 1.7210122346878052,
      "learning_rate": 7.834490159663154e-06,
      "loss": 0.3183,
      "num_input_tokens_seen": 8444472,
      "step": 12800
    },
    {
      "epoch": 7.550117924528302,
      "grad_norm": 1.8778008222579956,
      "learning_rate": 7.832370413598215e-06,
      "loss": 0.4815,
      "num_input_tokens_seen": 8448152,
      "step": 12805
    },
    {
      "epoch": 7.553066037735849,
      "grad_norm": 2.1387009620666504,
      "learning_rate": 7.830249917651003e-06,
      "loss": 0.3048,
      "num_input_tokens_seen": 8451288,
      "step": 12810
    },
    {
      "epoch": 7.556014150943396,
      "grad_norm": 1.8195284605026245,
      "learning_rate": 7.828128672382926e-06,
      "loss": 0.4801,
      "num_input_tokens_seen": 8453848,
      "step": 12815
    },
    {
      "epoch": 7.558962264150943,
      "grad_norm": 3.2908895015716553,
      "learning_rate": 7.826006678355596e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 8456536,
      "step": 12820
    },
    {
      "epoch": 7.56191037735849,
      "grad_norm": 2.637216329574585,
      "learning_rate": 7.823883936130817e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 8459032,
      "step": 12825
    },
    {
      "epoch": 7.564858490566038,
      "grad_norm": 1.9338364601135254,
      "learning_rate": 7.821760446270597e-06,
      "loss": 0.4196,
      "num_input_tokens_seen": 8462008,
      "step": 12830
    },
    {
      "epoch": 7.567806603773585,
      "grad_norm": 4.922972679138184,
      "learning_rate": 7.819636209337136e-06,
      "loss": 0.4354,
      "num_input_tokens_seen": 8465464,
      "step": 12835
    },
    {
      "epoch": 7.570754716981132,
      "grad_norm": 2.1567885875701904,
      "learning_rate": 7.817511225892838e-06,
      "loss": 0.3966,
      "num_input_tokens_seen": 8468856,
      "step": 12840
    },
    {
      "epoch": 7.5737028301886795,
      "grad_norm": 2.9766807556152344,
      "learning_rate": 7.8153854965003e-06,
      "loss": 0.3088,
      "num_input_tokens_seen": 8471576,
      "step": 12845
    },
    {
      "epoch": 7.576650943396227,
      "grad_norm": 3.5927374362945557,
      "learning_rate": 7.813259021722319e-06,
      "loss": 0.3737,
      "num_input_tokens_seen": 8474232,
      "step": 12850
    },
    {
      "epoch": 7.579599056603773,
      "grad_norm": 3.7578465938568115,
      "learning_rate": 7.811131802121885e-06,
      "loss": 0.3267,
      "num_input_tokens_seen": 8478744,
      "step": 12855
    },
    {
      "epoch": 7.5825471698113205,
      "grad_norm": 1.634164810180664,
      "learning_rate": 7.809003838262193e-06,
      "loss": 0.3616,
      "num_input_tokens_seen": 8481112,
      "step": 12860
    },
    {
      "epoch": 7.585495283018868,
      "grad_norm": 2.5660572052001953,
      "learning_rate": 7.806875130706628e-06,
      "loss": 0.4685,
      "num_input_tokens_seen": 8484472,
      "step": 12865
    },
    {
      "epoch": 7.588443396226415,
      "grad_norm": 1.7040224075317383,
      "learning_rate": 7.804745680018775e-06,
      "loss": 0.429,
      "num_input_tokens_seen": 8487800,
      "step": 12870
    },
    {
      "epoch": 7.591391509433962,
      "grad_norm": 1.7822401523590088,
      "learning_rate": 7.802615486762418e-06,
      "loss": 0.377,
      "num_input_tokens_seen": 8490680,
      "step": 12875
    },
    {
      "epoch": 7.59433962264151,
      "grad_norm": 2.3028523921966553,
      "learning_rate": 7.800484551501528e-06,
      "loss": 0.5189,
      "num_input_tokens_seen": 8493656,
      "step": 12880
    },
    {
      "epoch": 7.597287735849057,
      "grad_norm": 1.9862312078475952,
      "learning_rate": 7.798352874800285e-06,
      "loss": 0.431,
      "num_input_tokens_seen": 8497400,
      "step": 12885
    },
    {
      "epoch": 7.600235849056604,
      "grad_norm": 3.4986681938171387,
      "learning_rate": 7.79622045722306e-06,
      "loss": 0.3768,
      "num_input_tokens_seen": 8500504,
      "step": 12890
    },
    {
      "epoch": 7.603183962264151,
      "grad_norm": 3.4763734340667725,
      "learning_rate": 7.794087299334416e-06,
      "loss": 0.3749,
      "num_input_tokens_seen": 8503224,
      "step": 12895
    },
    {
      "epoch": 7.606132075471698,
      "grad_norm": 2.401961326599121,
      "learning_rate": 7.79195340169912e-06,
      "loss": 0.3836,
      "num_input_tokens_seen": 8506552,
      "step": 12900
    },
    {
      "epoch": 7.609080188679245,
      "grad_norm": 2.886000394821167,
      "learning_rate": 7.789818764882127e-06,
      "loss": 0.4344,
      "num_input_tokens_seen": 8509496,
      "step": 12905
    },
    {
      "epoch": 7.6120283018867925,
      "grad_norm": 2.1736953258514404,
      "learning_rate": 7.78768338944859e-06,
      "loss": 0.4151,
      "num_input_tokens_seen": 8512504,
      "step": 12910
    },
    {
      "epoch": 7.61497641509434,
      "grad_norm": 1.7958701848983765,
      "learning_rate": 7.785547275963865e-06,
      "loss": 0.4821,
      "num_input_tokens_seen": 8515224,
      "step": 12915
    },
    {
      "epoch": 7.617924528301887,
      "grad_norm": 1.9705222845077515,
      "learning_rate": 7.783410424993492e-06,
      "loss": 0.3697,
      "num_input_tokens_seen": 8518040,
      "step": 12920
    },
    {
      "epoch": 7.620872641509434,
      "grad_norm": 3.5560460090637207,
      "learning_rate": 7.781272837103213e-06,
      "loss": 0.4344,
      "num_input_tokens_seen": 8521176,
      "step": 12925
    },
    {
      "epoch": 7.623820754716981,
      "grad_norm": 2.9579155445098877,
      "learning_rate": 7.779134512858964e-06,
      "loss": 0.427,
      "num_input_tokens_seen": 8524312,
      "step": 12930
    },
    {
      "epoch": 7.626768867924528,
      "grad_norm": 2.3006367683410645,
      "learning_rate": 7.776995452826876e-06,
      "loss": 0.3135,
      "num_input_tokens_seen": 8527960,
      "step": 12935
    },
    {
      "epoch": 7.629716981132075,
      "grad_norm": 3.5295331478118896,
      "learning_rate": 7.774855657573274e-06,
      "loss": 0.5258,
      "num_input_tokens_seen": 8531064,
      "step": 12940
    },
    {
      "epoch": 7.632665094339623,
      "grad_norm": 3.229963779449463,
      "learning_rate": 7.772715127664676e-06,
      "loss": 0.5103,
      "num_input_tokens_seen": 8534488,
      "step": 12945
    },
    {
      "epoch": 7.63561320754717,
      "grad_norm": 4.332727909088135,
      "learning_rate": 7.7705738636678e-06,
      "loss": 0.408,
      "num_input_tokens_seen": 8537624,
      "step": 12950
    },
    {
      "epoch": 7.638561320754717,
      "grad_norm": 4.683002948760986,
      "learning_rate": 7.768431866149552e-06,
      "loss": 0.4022,
      "num_input_tokens_seen": 8540152,
      "step": 12955
    },
    {
      "epoch": 7.6415094339622645,
      "grad_norm": 3.2841179370880127,
      "learning_rate": 7.766289135677035e-06,
      "loss": 0.3094,
      "num_input_tokens_seen": 8542744,
      "step": 12960
    },
    {
      "epoch": 7.644457547169811,
      "grad_norm": 2.103149175643921,
      "learning_rate": 7.764145672817549e-06,
      "loss": 0.4244,
      "num_input_tokens_seen": 8546488,
      "step": 12965
    },
    {
      "epoch": 7.647405660377358,
      "grad_norm": 2.441077709197998,
      "learning_rate": 7.762001478138583e-06,
      "loss": 0.5385,
      "num_input_tokens_seen": 8551256,
      "step": 12970
    },
    {
      "epoch": 7.650353773584905,
      "grad_norm": 2.175731658935547,
      "learning_rate": 7.759856552207822e-06,
      "loss": 0.3986,
      "num_input_tokens_seen": 8554680,
      "step": 12975
    },
    {
      "epoch": 7.653301886792453,
      "grad_norm": 2.000058650970459,
      "learning_rate": 7.757710895593144e-06,
      "loss": 0.4563,
      "num_input_tokens_seen": 8558680,
      "step": 12980
    },
    {
      "epoch": 7.65625,
      "grad_norm": 2.986107349395752,
      "learning_rate": 7.755564508862623e-06,
      "loss": 0.5012,
      "num_input_tokens_seen": 8562328,
      "step": 12985
    },
    {
      "epoch": 7.659198113207547,
      "grad_norm": 1.8285999298095703,
      "learning_rate": 7.753417392584522e-06,
      "loss": 0.4466,
      "num_input_tokens_seen": 8565432,
      "step": 12990
    },
    {
      "epoch": 7.662146226415095,
      "grad_norm": 2.835665702819824,
      "learning_rate": 7.751269547327298e-06,
      "loss": 0.5145,
      "num_input_tokens_seen": 8568088,
      "step": 12995
    },
    {
      "epoch": 7.665094339622642,
      "grad_norm": 2.286983013153076,
      "learning_rate": 7.749120973659606e-06,
      "loss": 0.4619,
      "num_input_tokens_seen": 8570808,
      "step": 13000
    },
    {
      "epoch": 7.668042452830189,
      "grad_norm": 1.6988829374313354,
      "learning_rate": 7.746971672150286e-06,
      "loss": 0.3417,
      "num_input_tokens_seen": 8574232,
      "step": 13005
    },
    {
      "epoch": 7.6709905660377355,
      "grad_norm": 2.6970858573913574,
      "learning_rate": 7.74482164336838e-06,
      "loss": 0.4455,
      "num_input_tokens_seen": 8577464,
      "step": 13010
    },
    {
      "epoch": 7.673938679245283,
      "grad_norm": 2.123915910720825,
      "learning_rate": 7.742670887883111e-06,
      "loss": 0.4237,
      "num_input_tokens_seen": 8580600,
      "step": 13015
    },
    {
      "epoch": 7.67688679245283,
      "grad_norm": 3.5682930946350098,
      "learning_rate": 7.740519406263905e-06,
      "loss": 0.3882,
      "num_input_tokens_seen": 8584312,
      "step": 13020
    },
    {
      "epoch": 7.679834905660377,
      "grad_norm": 2.3823580741882324,
      "learning_rate": 7.738367199080376e-06,
      "loss": 0.355,
      "num_input_tokens_seen": 8587800,
      "step": 13025
    },
    {
      "epoch": 7.682783018867925,
      "grad_norm": 2.4183521270751953,
      "learning_rate": 7.73621426690233e-06,
      "loss": 0.6053,
      "num_input_tokens_seen": 8592344,
      "step": 13030
    },
    {
      "epoch": 7.685731132075472,
      "grad_norm": 5.163645267486572,
      "learning_rate": 7.734060610299764e-06,
      "loss": 0.3858,
      "num_input_tokens_seen": 8594840,
      "step": 13035
    },
    {
      "epoch": 7.688679245283019,
      "grad_norm": 1.4902913570404053,
      "learning_rate": 7.731906229842869e-06,
      "loss": 0.5705,
      "num_input_tokens_seen": 8598232,
      "step": 13040
    },
    {
      "epoch": 7.691627358490566,
      "grad_norm": 1.9202746152877808,
      "learning_rate": 7.729751126102023e-06,
      "loss": 0.409,
      "num_input_tokens_seen": 8602712,
      "step": 13045
    },
    {
      "epoch": 7.694575471698113,
      "grad_norm": 2.378814458847046,
      "learning_rate": 7.727595299647805e-06,
      "loss": 0.4245,
      "num_input_tokens_seen": 8606008,
      "step": 13050
    },
    {
      "epoch": 7.69752358490566,
      "grad_norm": 1.5978602170944214,
      "learning_rate": 7.725438751050973e-06,
      "loss": 0.4312,
      "num_input_tokens_seen": 8609080,
      "step": 13055
    },
    {
      "epoch": 7.7004716981132075,
      "grad_norm": 2.276920795440674,
      "learning_rate": 7.723281480882489e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 8611544,
      "step": 13060
    },
    {
      "epoch": 7.703419811320755,
      "grad_norm": 1.9206783771514893,
      "learning_rate": 7.721123489713494e-06,
      "loss": 0.3685,
      "num_input_tokens_seen": 8614424,
      "step": 13065
    },
    {
      "epoch": 7.706367924528302,
      "grad_norm": 2.032841920852661,
      "learning_rate": 7.718964778115328e-06,
      "loss": 0.2937,
      "num_input_tokens_seen": 8617688,
      "step": 13070
    },
    {
      "epoch": 7.709316037735849,
      "grad_norm": 1.76680326461792,
      "learning_rate": 7.716805346659519e-06,
      "loss": 0.3597,
      "num_input_tokens_seen": 8620408,
      "step": 13075
    },
    {
      "epoch": 7.712264150943396,
      "grad_norm": 2.39839506149292,
      "learning_rate": 7.714645195917788e-06,
      "loss": 0.3966,
      "num_input_tokens_seen": 8623768,
      "step": 13080
    },
    {
      "epoch": 7.715212264150943,
      "grad_norm": 3.303377866744995,
      "learning_rate": 7.712484326462038e-06,
      "loss": 0.3388,
      "num_input_tokens_seen": 8626712,
      "step": 13085
    },
    {
      "epoch": 7.71816037735849,
      "grad_norm": 2.1727609634399414,
      "learning_rate": 7.710322738864375e-06,
      "loss": 0.4173,
      "num_input_tokens_seen": 8629944,
      "step": 13090
    },
    {
      "epoch": 7.721108490566038,
      "grad_norm": 2.3533287048339844,
      "learning_rate": 7.708160433697085e-06,
      "loss": 0.4305,
      "num_input_tokens_seen": 8633080,
      "step": 13095
    },
    {
      "epoch": 7.724056603773585,
      "grad_norm": 1.6141928434371948,
      "learning_rate": 7.705997411532649e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 8637464,
      "step": 13100
    },
    {
      "epoch": 7.727004716981132,
      "grad_norm": 2.224712610244751,
      "learning_rate": 7.703833672943735e-06,
      "loss": 0.4051,
      "num_input_tokens_seen": 8641368,
      "step": 13105
    },
    {
      "epoch": 7.7299528301886795,
      "grad_norm": 2.14267635345459,
      "learning_rate": 7.701669218503206e-06,
      "loss": 0.3502,
      "num_input_tokens_seen": 8643960,
      "step": 13110
    },
    {
      "epoch": 7.732900943396227,
      "grad_norm": 2.4199633598327637,
      "learning_rate": 7.699504048784106e-06,
      "loss": 0.3713,
      "num_input_tokens_seen": 8646872,
      "step": 13115
    },
    {
      "epoch": 7.735849056603773,
      "grad_norm": 2.591615915298462,
      "learning_rate": 7.697338164359675e-06,
      "loss": 0.4108,
      "num_input_tokens_seen": 8649912,
      "step": 13120
    },
    {
      "epoch": 7.7387971698113205,
      "grad_norm": 2.1200666427612305,
      "learning_rate": 7.69517156580334e-06,
      "loss": 0.3408,
      "num_input_tokens_seen": 8654008,
      "step": 13125
    },
    {
      "epoch": 7.741745283018868,
      "grad_norm": 1.6616610288619995,
      "learning_rate": 7.693004253688716e-06,
      "loss": 0.5282,
      "num_input_tokens_seen": 8656888,
      "step": 13130
    },
    {
      "epoch": 7.744693396226415,
      "grad_norm": 1.8039122819900513,
      "learning_rate": 7.690836228589613e-06,
      "loss": 0.3486,
      "num_input_tokens_seen": 8659544,
      "step": 13135
    },
    {
      "epoch": 7.747641509433962,
      "grad_norm": 1.146605134010315,
      "learning_rate": 7.688667491080019e-06,
      "loss": 0.3006,
      "num_input_tokens_seen": 8664472,
      "step": 13140
    },
    {
      "epoch": 7.75058962264151,
      "grad_norm": 2.873058795928955,
      "learning_rate": 7.686498041734121e-06,
      "loss": 0.4796,
      "num_input_tokens_seen": 8667576,
      "step": 13145
    },
    {
      "epoch": 7.753537735849057,
      "grad_norm": 2.5881621837615967,
      "learning_rate": 7.684327881126285e-06,
      "loss": 0.6387,
      "num_input_tokens_seen": 8670936,
      "step": 13150
    },
    {
      "epoch": 7.756485849056604,
      "grad_norm": 3.444035530090332,
      "learning_rate": 7.682157009831078e-06,
      "loss": 0.4186,
      "num_input_tokens_seen": 8673752,
      "step": 13155
    },
    {
      "epoch": 7.759433962264151,
      "grad_norm": 1.9400792121887207,
      "learning_rate": 7.67998542842324e-06,
      "loss": 0.3413,
      "num_input_tokens_seen": 8677016,
      "step": 13160
    },
    {
      "epoch": 7.762382075471698,
      "grad_norm": 2.3622524738311768,
      "learning_rate": 7.677813137477711e-06,
      "loss": 0.4019,
      "num_input_tokens_seen": 8680536,
      "step": 13165
    },
    {
      "epoch": 7.765330188679245,
      "grad_norm": 3.377296209335327,
      "learning_rate": 7.675640137569614e-06,
      "loss": 0.49,
      "num_input_tokens_seen": 8683256,
      "step": 13170
    },
    {
      "epoch": 7.7682783018867925,
      "grad_norm": 3.6802563667297363,
      "learning_rate": 7.673466429274257e-06,
      "loss": 0.4392,
      "num_input_tokens_seen": 8685560,
      "step": 13175
    },
    {
      "epoch": 7.77122641509434,
      "grad_norm": 1.5118906497955322,
      "learning_rate": 7.671292013167143e-06,
      "loss": 0.5012,
      "num_input_tokens_seen": 8688312,
      "step": 13180
    },
    {
      "epoch": 7.774174528301887,
      "grad_norm": 3.00201678276062,
      "learning_rate": 7.669116889823955e-06,
      "loss": 0.5108,
      "num_input_tokens_seen": 8691032,
      "step": 13185
    },
    {
      "epoch": 7.777122641509434,
      "grad_norm": 1.3862141370773315,
      "learning_rate": 7.666941059820567e-06,
      "loss": 0.4321,
      "num_input_tokens_seen": 8694648,
      "step": 13190
    },
    {
      "epoch": 7.780070754716981,
      "grad_norm": 2.6622705459594727,
      "learning_rate": 7.66476452373304e-06,
      "loss": 0.5551,
      "num_input_tokens_seen": 8698552,
      "step": 13195
    },
    {
      "epoch": 7.783018867924528,
      "grad_norm": 3.244410753250122,
      "learning_rate": 7.66258728213762e-06,
      "loss": 0.4037,
      "num_input_tokens_seen": 8701976,
      "step": 13200
    },
    {
      "epoch": 7.785966981132075,
      "grad_norm": 3.7800047397613525,
      "learning_rate": 7.66040933561074e-06,
      "loss": 0.4969,
      "num_input_tokens_seen": 8704664,
      "step": 13205
    },
    {
      "epoch": 7.788915094339623,
      "grad_norm": 1.6505465507507324,
      "learning_rate": 7.658230684729027e-06,
      "loss": 0.5963,
      "num_input_tokens_seen": 8708312,
      "step": 13210
    },
    {
      "epoch": 7.79186320754717,
      "grad_norm": 1.9823917150497437,
      "learning_rate": 7.656051330069282e-06,
      "loss": 0.5418,
      "num_input_tokens_seen": 8712056,
      "step": 13215
    },
    {
      "epoch": 7.794811320754717,
      "grad_norm": 2.2153513431549072,
      "learning_rate": 7.6538712722085e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 8715352,
      "step": 13220
    },
    {
      "epoch": 7.7977594339622645,
      "grad_norm": 1.75736403465271,
      "learning_rate": 7.651690511723862e-06,
      "loss": 0.4084,
      "num_input_tokens_seen": 8718008,
      "step": 13225
    },
    {
      "epoch": 7.800707547169811,
      "grad_norm": 3.3481149673461914,
      "learning_rate": 7.64950904919273e-06,
      "loss": 0.5511,
      "num_input_tokens_seen": 8721048,
      "step": 13230
    },
    {
      "epoch": 7.803655660377358,
      "grad_norm": 1.9734175205230713,
      "learning_rate": 7.647326885192662e-06,
      "loss": 0.4106,
      "num_input_tokens_seen": 8723096,
      "step": 13235
    },
    {
      "epoch": 7.806603773584905,
      "grad_norm": 1.9829331636428833,
      "learning_rate": 7.645144020301392e-06,
      "loss": 0.3379,
      "num_input_tokens_seen": 8727032,
      "step": 13240
    },
    {
      "epoch": 7.809551886792453,
      "grad_norm": 4.338983058929443,
      "learning_rate": 7.64296045509684e-06,
      "loss": 0.3942,
      "num_input_tokens_seen": 8732568,
      "step": 13245
    },
    {
      "epoch": 7.8125,
      "grad_norm": 3.9344868659973145,
      "learning_rate": 7.64077619015712e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 8735832,
      "step": 13250
    },
    {
      "epoch": 7.815448113207547,
      "grad_norm": 2.494302988052368,
      "learning_rate": 7.638591226060519e-06,
      "loss": 0.4958,
      "num_input_tokens_seen": 8739192,
      "step": 13255
    },
    {
      "epoch": 7.818396226415095,
      "grad_norm": 5.048979759216309,
      "learning_rate": 7.636405563385522e-06,
      "loss": 0.3928,
      "num_input_tokens_seen": 8743352,
      "step": 13260
    },
    {
      "epoch": 7.821344339622642,
      "grad_norm": 3.2518227100372314,
      "learning_rate": 7.634219202710789e-06,
      "loss": 0.4979,
      "num_input_tokens_seen": 8745816,
      "step": 13265
    },
    {
      "epoch": 7.824292452830189,
      "grad_norm": 2.0998196601867676,
      "learning_rate": 7.632032144615168e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 8749016,
      "step": 13270
    },
    {
      "epoch": 7.8272405660377355,
      "grad_norm": 3.7662806510925293,
      "learning_rate": 7.629844389677695e-06,
      "loss": 0.4454,
      "num_input_tokens_seen": 8752120,
      "step": 13275
    },
    {
      "epoch": 7.830188679245283,
      "grad_norm": 2.3895509243011475,
      "learning_rate": 7.627655938477586e-06,
      "loss": 0.4533,
      "num_input_tokens_seen": 8755640,
      "step": 13280
    },
    {
      "epoch": 7.83313679245283,
      "grad_norm": 2.7380130290985107,
      "learning_rate": 7.6254667915942415e-06,
      "loss": 0.349,
      "num_input_tokens_seen": 8757976,
      "step": 13285
    },
    {
      "epoch": 7.836084905660377,
      "grad_norm": 1.819222092628479,
      "learning_rate": 7.62327694960725e-06,
      "loss": 0.4238,
      "num_input_tokens_seen": 8761112,
      "step": 13290
    },
    {
      "epoch": 7.839033018867925,
      "grad_norm": 1.5021588802337646,
      "learning_rate": 7.621086413096379e-06,
      "loss": 0.3783,
      "num_input_tokens_seen": 8764536,
      "step": 13295
    },
    {
      "epoch": 7.841981132075472,
      "grad_norm": 1.5905457735061646,
      "learning_rate": 7.618895182641584e-06,
      "loss": 0.5001,
      "num_input_tokens_seen": 8768152,
      "step": 13300
    },
    {
      "epoch": 7.844929245283019,
      "grad_norm": 2.514277696609497,
      "learning_rate": 7.6167032588230035e-06,
      "loss": 0.3837,
      "num_input_tokens_seen": 8770776,
      "step": 13305
    },
    {
      "epoch": 7.847877358490566,
      "grad_norm": 1.731438398361206,
      "learning_rate": 7.614510642220958e-06,
      "loss": 0.554,
      "num_input_tokens_seen": 8778328,
      "step": 13310
    },
    {
      "epoch": 7.850825471698113,
      "grad_norm": 2.2607998847961426,
      "learning_rate": 7.612317333415951e-06,
      "loss": 0.4062,
      "num_input_tokens_seen": 8781944,
      "step": 13315
    },
    {
      "epoch": 7.85377358490566,
      "grad_norm": 2.195303201675415,
      "learning_rate": 7.610123332988673e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 8785016,
      "step": 13320
    },
    {
      "epoch": 7.8567216981132075,
      "grad_norm": 1.3579221963882446,
      "learning_rate": 7.607928641519992e-06,
      "loss": 0.3732,
      "num_input_tokens_seen": 8788472,
      "step": 13325
    },
    {
      "epoch": 7.859669811320755,
      "grad_norm": 3.702352523803711,
      "learning_rate": 7.605733259590964e-06,
      "loss": 0.486,
      "num_input_tokens_seen": 8791096,
      "step": 13330
    },
    {
      "epoch": 7.862617924528302,
      "grad_norm": 1.9895697832107544,
      "learning_rate": 7.603537187782826e-06,
      "loss": 0.3774,
      "num_input_tokens_seen": 8795000,
      "step": 13335
    },
    {
      "epoch": 7.865566037735849,
      "grad_norm": 3.3092434406280518,
      "learning_rate": 7.601340426676996e-06,
      "loss": 0.5162,
      "num_input_tokens_seen": 8797880,
      "step": 13340
    },
    {
      "epoch": 7.868514150943396,
      "grad_norm": 4.736794471740723,
      "learning_rate": 7.599142976855077e-06,
      "loss": 0.4843,
      "num_input_tokens_seen": 8801240,
      "step": 13345
    },
    {
      "epoch": 7.871462264150943,
      "grad_norm": 2.170046329498291,
      "learning_rate": 7.596944838898854e-06,
      "loss": 0.3644,
      "num_input_tokens_seen": 8804216,
      "step": 13350
    },
    {
      "epoch": 7.87441037735849,
      "grad_norm": 1.5298583507537842,
      "learning_rate": 7.594746013390293e-06,
      "loss": 0.4207,
      "num_input_tokens_seen": 8807544,
      "step": 13355
    },
    {
      "epoch": 7.877358490566038,
      "grad_norm": 2.835103750228882,
      "learning_rate": 7.59254650091154e-06,
      "loss": 0.3643,
      "num_input_tokens_seen": 8813176,
      "step": 13360
    },
    {
      "epoch": 7.880306603773585,
      "grad_norm": 1.5254782438278198,
      "learning_rate": 7.59034630204493e-06,
      "loss": 0.414,
      "num_input_tokens_seen": 8815928,
      "step": 13365
    },
    {
      "epoch": 7.883254716981132,
      "grad_norm": 1.7785441875457764,
      "learning_rate": 7.588145417372972e-06,
      "loss": 0.4021,
      "num_input_tokens_seen": 8818776,
      "step": 13370
    },
    {
      "epoch": 7.8862028301886795,
      "grad_norm": 1.6062877178192139,
      "learning_rate": 7.585943847478361e-06,
      "loss": 0.4822,
      "num_input_tokens_seen": 8821688,
      "step": 13375
    },
    {
      "epoch": 7.889150943396227,
      "grad_norm": 2.6857597827911377,
      "learning_rate": 7.583741592943971e-06,
      "loss": 0.3712,
      "num_input_tokens_seen": 8824664,
      "step": 13380
    },
    {
      "epoch": 7.892099056603773,
      "grad_norm": 2.50944447517395,
      "learning_rate": 7.581538654352859e-06,
      "loss": 0.4835,
      "num_input_tokens_seen": 8827288,
      "step": 13385
    },
    {
      "epoch": 7.8950471698113205,
      "grad_norm": 2.0214028358459473,
      "learning_rate": 7.579335032288262e-06,
      "loss": 0.2788,
      "num_input_tokens_seen": 8829944,
      "step": 13390
    },
    {
      "epoch": 7.897995283018868,
      "grad_norm": 4.519385814666748,
      "learning_rate": 7.577130727333598e-06,
      "loss": 0.5815,
      "num_input_tokens_seen": 8832920,
      "step": 13395
    },
    {
      "epoch": 7.900943396226415,
      "grad_norm": 2.595771551132202,
      "learning_rate": 7.5749257400724695e-06,
      "loss": 0.4463,
      "num_input_tokens_seen": 8836120,
      "step": 13400
    },
    {
      "epoch": 7.903891509433962,
      "grad_norm": 2.945953607559204,
      "learning_rate": 7.572720071088653e-06,
      "loss": 0.5252,
      "num_input_tokens_seen": 8838648,
      "step": 13405
    },
    {
      "epoch": 7.90683962264151,
      "grad_norm": 1.875922441482544,
      "learning_rate": 7.570513720966108e-06,
      "loss": 0.3148,
      "num_input_tokens_seen": 8842136,
      "step": 13410
    },
    {
      "epoch": 7.909787735849057,
      "grad_norm": 3.3363001346588135,
      "learning_rate": 7.56830669028898e-06,
      "loss": 0.4732,
      "num_input_tokens_seen": 8845624,
      "step": 13415
    },
    {
      "epoch": 7.912735849056604,
      "grad_norm": 2.1710054874420166,
      "learning_rate": 7.566098979641588e-06,
      "loss": 0.5637,
      "num_input_tokens_seen": 8848216,
      "step": 13420
    },
    {
      "epoch": 7.915683962264151,
      "grad_norm": 2.622162342071533,
      "learning_rate": 7.563890589608427e-06,
      "loss": 0.3969,
      "num_input_tokens_seen": 8851928,
      "step": 13425
    },
    {
      "epoch": 7.918632075471698,
      "grad_norm": 2.3745248317718506,
      "learning_rate": 7.561681520774187e-06,
      "loss": 0.2593,
      "num_input_tokens_seen": 8854680,
      "step": 13430
    },
    {
      "epoch": 7.921580188679245,
      "grad_norm": 1.859104871749878,
      "learning_rate": 7.559471773723721e-06,
      "loss": 0.547,
      "num_input_tokens_seen": 8858200,
      "step": 13435
    },
    {
      "epoch": 7.9245283018867925,
      "grad_norm": 2.1873695850372314,
      "learning_rate": 7.557261349042073e-06,
      "loss": 0.5313,
      "num_input_tokens_seen": 8862680,
      "step": 13440
    },
    {
      "epoch": 7.92747641509434,
      "grad_norm": 1.6060349941253662,
      "learning_rate": 7.555050247314464e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 8866712,
      "step": 13445
    },
    {
      "epoch": 7.930424528301887,
      "grad_norm": 2.6133663654327393,
      "learning_rate": 7.552838469126289e-06,
      "loss": 0.478,
      "num_input_tokens_seen": 8870328,
      "step": 13450
    },
    {
      "epoch": 7.933372641509434,
      "grad_norm": 3.306678056716919,
      "learning_rate": 7.550626015063125e-06,
      "loss": 0.4273,
      "num_input_tokens_seen": 8873400,
      "step": 13455
    },
    {
      "epoch": 7.936320754716981,
      "grad_norm": 1.6566591262817383,
      "learning_rate": 7.548412885710734e-06,
      "loss": 0.4008,
      "num_input_tokens_seen": 8876568,
      "step": 13460
    },
    {
      "epoch": 7.939268867924528,
      "grad_norm": 1.3968334197998047,
      "learning_rate": 7.546199081655048e-06,
      "loss": 0.3885,
      "num_input_tokens_seen": 8880760,
      "step": 13465
    },
    {
      "epoch": 7.942216981132075,
      "grad_norm": 5.70344877243042,
      "learning_rate": 7.54398460348218e-06,
      "loss": 0.3752,
      "num_input_tokens_seen": 8883608,
      "step": 13470
    },
    {
      "epoch": 7.945165094339623,
      "grad_norm": 3.3598666191101074,
      "learning_rate": 7.541769451778425e-06,
      "loss": 0.4531,
      "num_input_tokens_seen": 8886904,
      "step": 13475
    },
    {
      "epoch": 7.94811320754717,
      "grad_norm": 2.371516704559326,
      "learning_rate": 7.5395536271302536e-06,
      "loss": 0.4012,
      "num_input_tokens_seen": 8889752,
      "step": 13480
    },
    {
      "epoch": 7.951061320754717,
      "grad_norm": 1.8155467510223389,
      "learning_rate": 7.5373371301243136e-06,
      "loss": 0.587,
      "num_input_tokens_seen": 8893048,
      "step": 13485
    },
    {
      "epoch": 7.9540094339622645,
      "grad_norm": 3.7426676750183105,
      "learning_rate": 7.535119961347433e-06,
      "loss": 0.3798,
      "num_input_tokens_seen": 8896024,
      "step": 13490
    },
    {
      "epoch": 7.956957547169811,
      "grad_norm": 2.2288289070129395,
      "learning_rate": 7.532902121386618e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 8898872,
      "step": 13495
    },
    {
      "epoch": 7.959905660377358,
      "grad_norm": 1.701873779296875,
      "learning_rate": 7.530683610829051e-06,
      "loss": 0.4316,
      "num_input_tokens_seen": 8901176,
      "step": 13500
    },
    {
      "epoch": 7.962853773584905,
      "grad_norm": 1.3052029609680176,
      "learning_rate": 7.5284644302620906e-06,
      "loss": 0.285,
      "num_input_tokens_seen": 8903832,
      "step": 13505
    },
    {
      "epoch": 7.965801886792453,
      "grad_norm": 2.572922468185425,
      "learning_rate": 7.526244580273274e-06,
      "loss": 0.322,
      "num_input_tokens_seen": 8906648,
      "step": 13510
    },
    {
      "epoch": 7.96875,
      "grad_norm": 3.9600603580474854,
      "learning_rate": 7.524024061450318e-06,
      "loss": 0.5009,
      "num_input_tokens_seen": 8910168,
      "step": 13515
    },
    {
      "epoch": 7.971698113207547,
      "grad_norm": 2.435194492340088,
      "learning_rate": 7.521802874381115e-06,
      "loss": 0.5671,
      "num_input_tokens_seen": 8913720,
      "step": 13520
    },
    {
      "epoch": 7.974646226415095,
      "grad_norm": 2.292811155319214,
      "learning_rate": 7.519581019653731e-06,
      "loss": 0.3402,
      "num_input_tokens_seen": 8916344,
      "step": 13525
    },
    {
      "epoch": 7.977594339622642,
      "grad_norm": 2.94150710105896,
      "learning_rate": 7.517358497856413e-06,
      "loss": 0.4036,
      "num_input_tokens_seen": 8919096,
      "step": 13530
    },
    {
      "epoch": 7.980542452830189,
      "grad_norm": 2.0033044815063477,
      "learning_rate": 7.515135309577584e-06,
      "loss": 0.378,
      "num_input_tokens_seen": 8921336,
      "step": 13535
    },
    {
      "epoch": 7.9834905660377355,
      "grad_norm": 1.4132635593414307,
      "learning_rate": 7.5129114554058425e-06,
      "loss": 0.5267,
      "num_input_tokens_seen": 8924696,
      "step": 13540
    },
    {
      "epoch": 7.986438679245283,
      "grad_norm": 3.0454537868499756,
      "learning_rate": 7.510686935929963e-06,
      "loss": 0.3761,
      "num_input_tokens_seen": 8928728,
      "step": 13545
    },
    {
      "epoch": 7.98938679245283,
      "grad_norm": 3.5960988998413086,
      "learning_rate": 7.5084617517388965e-06,
      "loss": 0.367,
      "num_input_tokens_seen": 8931640,
      "step": 13550
    },
    {
      "epoch": 7.992334905660377,
      "grad_norm": 1.6785526275634766,
      "learning_rate": 7.506235903421771e-06,
      "loss": 0.3144,
      "num_input_tokens_seen": 8934808,
      "step": 13555
    },
    {
      "epoch": 7.995283018867925,
      "grad_norm": 2.5356295108795166,
      "learning_rate": 7.504009391567889e-06,
      "loss": 0.4559,
      "num_input_tokens_seen": 8937624,
      "step": 13560
    },
    {
      "epoch": 7.998231132075472,
      "grad_norm": 1.6504689455032349,
      "learning_rate": 7.501782216766729e-06,
      "loss": 0.3139,
      "num_input_tokens_seen": 8940440,
      "step": 13565
    },
    {
      "epoch": 8.0,
      "eval_loss": 0.5016111135482788,
      "eval_runtime": 18.3078,
      "eval_samples_per_second": 92.638,
      "eval_steps_per_second": 23.159,
      "num_input_tokens_seen": 8942288,
      "step": 13568
    },
    {
      "epoch": 8.00117924528302,
      "grad_norm": 2.456636905670166,
      "learning_rate": 7.499554379607944e-06,
      "loss": 0.2684,
      "num_input_tokens_seen": 8943856,
      "step": 13570
    },
    {
      "epoch": 8.004127358490566,
      "grad_norm": 1.6309980154037476,
      "learning_rate": 7.497325880681365e-06,
      "loss": 0.352,
      "num_input_tokens_seen": 8946736,
      "step": 13575
    },
    {
      "epoch": 8.007075471698114,
      "grad_norm": 2.5087695121765137,
      "learning_rate": 7.495096720576994e-06,
      "loss": 0.3488,
      "num_input_tokens_seen": 8950256,
      "step": 13580
    },
    {
      "epoch": 8.01002358490566,
      "grad_norm": 2.5145702362060547,
      "learning_rate": 7.492866899885017e-06,
      "loss": 0.3393,
      "num_input_tokens_seen": 8952880,
      "step": 13585
    },
    {
      "epoch": 8.012971698113208,
      "grad_norm": 2.283844232559204,
      "learning_rate": 7.490636419195782e-06,
      "loss": 0.5546,
      "num_input_tokens_seen": 8956432,
      "step": 13590
    },
    {
      "epoch": 8.015919811320755,
      "grad_norm": 3.543396472930908,
      "learning_rate": 7.488405279099821e-06,
      "loss": 0.4678,
      "num_input_tokens_seen": 8959056,
      "step": 13595
    },
    {
      "epoch": 8.018867924528301,
      "grad_norm": 2.062455892562866,
      "learning_rate": 7.48617348018784e-06,
      "loss": 0.4259,
      "num_input_tokens_seen": 8962256,
      "step": 13600
    },
    {
      "epoch": 8.02181603773585,
      "grad_norm": 3.507220983505249,
      "learning_rate": 7.4839410230507134e-06,
      "loss": 0.4995,
      "num_input_tokens_seen": 8965008,
      "step": 13605
    },
    {
      "epoch": 8.024764150943396,
      "grad_norm": 3.072874069213867,
      "learning_rate": 7.481707908279496e-06,
      "loss": 0.3597,
      "num_input_tokens_seen": 8968496,
      "step": 13610
    },
    {
      "epoch": 8.027712264150944,
      "grad_norm": 1.6695091724395752,
      "learning_rate": 7.4794741364654144e-06,
      "loss": 0.3939,
      "num_input_tokens_seen": 8971792,
      "step": 13615
    },
    {
      "epoch": 8.03066037735849,
      "grad_norm": 2.226823329925537,
      "learning_rate": 7.477239708199871e-06,
      "loss": 0.4118,
      "num_input_tokens_seen": 8975728,
      "step": 13620
    },
    {
      "epoch": 8.033608490566039,
      "grad_norm": 2.2165298461914062,
      "learning_rate": 7.475004624074434e-06,
      "loss": 0.3325,
      "num_input_tokens_seen": 8978960,
      "step": 13625
    },
    {
      "epoch": 8.036556603773585,
      "grad_norm": 2.423633337020874,
      "learning_rate": 7.4727688846808595e-06,
      "loss": 0.3482,
      "num_input_tokens_seen": 8983408,
      "step": 13630
    },
    {
      "epoch": 8.039504716981131,
      "grad_norm": 4.094726085662842,
      "learning_rate": 7.4705324906110654e-06,
      "loss": 0.3403,
      "num_input_tokens_seen": 8986096,
      "step": 13635
    },
    {
      "epoch": 8.04245283018868,
      "grad_norm": 1.921600580215454,
      "learning_rate": 7.4682954424571466e-06,
      "loss": 0.3737,
      "num_input_tokens_seen": 8989136,
      "step": 13640
    },
    {
      "epoch": 8.045400943396226,
      "grad_norm": 5.216590881347656,
      "learning_rate": 7.466057740811372e-06,
      "loss": 0.4144,
      "num_input_tokens_seen": 8992240,
      "step": 13645
    },
    {
      "epoch": 8.048349056603774,
      "grad_norm": 2.242926836013794,
      "learning_rate": 7.463819386266182e-06,
      "loss": 0.3699,
      "num_input_tokens_seen": 8995472,
      "step": 13650
    },
    {
      "epoch": 8.05129716981132,
      "grad_norm": 2.8685178756713867,
      "learning_rate": 7.461580379414191e-06,
      "loss": 0.4968,
      "num_input_tokens_seen": 8998512,
      "step": 13655
    },
    {
      "epoch": 8.054245283018869,
      "grad_norm": 2.709294080734253,
      "learning_rate": 7.459340720848187e-06,
      "loss": 0.4761,
      "num_input_tokens_seen": 9001200,
      "step": 13660
    },
    {
      "epoch": 8.057193396226415,
      "grad_norm": 2.0148262977600098,
      "learning_rate": 7.457100411161128e-06,
      "loss": 0.3293,
      "num_input_tokens_seen": 9005008,
      "step": 13665
    },
    {
      "epoch": 8.060141509433961,
      "grad_norm": 2.417386531829834,
      "learning_rate": 7.454859450946144e-06,
      "loss": 0.4912,
      "num_input_tokens_seen": 9007728,
      "step": 13670
    },
    {
      "epoch": 8.06308962264151,
      "grad_norm": 2.5843067169189453,
      "learning_rate": 7.4526178407965396e-06,
      "loss": 0.3022,
      "num_input_tokens_seen": 9010864,
      "step": 13675
    },
    {
      "epoch": 8.066037735849056,
      "grad_norm": 3.808467388153076,
      "learning_rate": 7.450375581305794e-06,
      "loss": 0.365,
      "num_input_tokens_seen": 9013264,
      "step": 13680
    },
    {
      "epoch": 8.068985849056604,
      "grad_norm": 2.7796545028686523,
      "learning_rate": 7.448132673067552e-06,
      "loss": 0.4364,
      "num_input_tokens_seen": 9016112,
      "step": 13685
    },
    {
      "epoch": 8.07193396226415,
      "grad_norm": 2.7986409664154053,
      "learning_rate": 7.445889116675634e-06,
      "loss": 0.4565,
      "num_input_tokens_seen": 9019120,
      "step": 13690
    },
    {
      "epoch": 8.074882075471699,
      "grad_norm": 2.3738255500793457,
      "learning_rate": 7.443644912724031e-06,
      "loss": 0.3414,
      "num_input_tokens_seen": 9022096,
      "step": 13695
    },
    {
      "epoch": 8.077830188679245,
      "grad_norm": 2.517268657684326,
      "learning_rate": 7.441400061806907e-06,
      "loss": 0.316,
      "num_input_tokens_seen": 9024816,
      "step": 13700
    },
    {
      "epoch": 8.080778301886792,
      "grad_norm": 2.3621480464935303,
      "learning_rate": 7.439154564518592e-06,
      "loss": 0.4317,
      "num_input_tokens_seen": 9027472,
      "step": 13705
    },
    {
      "epoch": 8.08372641509434,
      "grad_norm": 1.9411017894744873,
      "learning_rate": 7.436908421453597e-06,
      "loss": 0.4304,
      "num_input_tokens_seen": 9031024,
      "step": 13710
    },
    {
      "epoch": 8.086674528301886,
      "grad_norm": 2.7997829914093018,
      "learning_rate": 7.434661633206593e-06,
      "loss": 0.387,
      "num_input_tokens_seen": 9034704,
      "step": 13715
    },
    {
      "epoch": 8.089622641509434,
      "grad_norm": 1.6025009155273438,
      "learning_rate": 7.4324142003724286e-06,
      "loss": 0.6207,
      "num_input_tokens_seen": 9039440,
      "step": 13720
    },
    {
      "epoch": 8.09257075471698,
      "grad_norm": 1.1632803678512573,
      "learning_rate": 7.430166123546122e-06,
      "loss": 0.4052,
      "num_input_tokens_seen": 9043184,
      "step": 13725
    },
    {
      "epoch": 8.095518867924529,
      "grad_norm": 1.8271898031234741,
      "learning_rate": 7.427917403322862e-06,
      "loss": 0.37,
      "num_input_tokens_seen": 9046544,
      "step": 13730
    },
    {
      "epoch": 8.098466981132075,
      "grad_norm": 2.0082435607910156,
      "learning_rate": 7.425668040298003e-06,
      "loss": 0.3334,
      "num_input_tokens_seen": 9050448,
      "step": 13735
    },
    {
      "epoch": 8.101415094339623,
      "grad_norm": 4.105738639831543,
      "learning_rate": 7.4234180350670785e-06,
      "loss": 0.4407,
      "num_input_tokens_seen": 9053776,
      "step": 13740
    },
    {
      "epoch": 8.10436320754717,
      "grad_norm": 2.1171960830688477,
      "learning_rate": 7.421167388225785e-06,
      "loss": 0.3709,
      "num_input_tokens_seen": 9056528,
      "step": 13745
    },
    {
      "epoch": 8.107311320754716,
      "grad_norm": 2.507201910018921,
      "learning_rate": 7.41891610036999e-06,
      "loss": 0.3862,
      "num_input_tokens_seen": 9060016,
      "step": 13750
    },
    {
      "epoch": 8.110259433962264,
      "grad_norm": 3.9508867263793945,
      "learning_rate": 7.416664172095732e-06,
      "loss": 0.4048,
      "num_input_tokens_seen": 9062768,
      "step": 13755
    },
    {
      "epoch": 8.11320754716981,
      "grad_norm": 2.4400722980499268,
      "learning_rate": 7.414411603999221e-06,
      "loss": 0.4656,
      "num_input_tokens_seen": 9066608,
      "step": 13760
    },
    {
      "epoch": 8.116155660377359,
      "grad_norm": 1.6895620822906494,
      "learning_rate": 7.4121583966768295e-06,
      "loss": 0.3562,
      "num_input_tokens_seen": 9069552,
      "step": 13765
    },
    {
      "epoch": 8.119103773584905,
      "grad_norm": 1.9156559705734253,
      "learning_rate": 7.409904550725109e-06,
      "loss": 0.389,
      "num_input_tokens_seen": 9072496,
      "step": 13770
    },
    {
      "epoch": 8.122051886792454,
      "grad_norm": 1.7932703495025635,
      "learning_rate": 7.407650066740771e-06,
      "loss": 0.3993,
      "num_input_tokens_seen": 9075408,
      "step": 13775
    },
    {
      "epoch": 8.125,
      "grad_norm": 3.3254969120025635,
      "learning_rate": 7.405394945320702e-06,
      "loss": 0.3386,
      "num_input_tokens_seen": 9077968,
      "step": 13780
    },
    {
      "epoch": 8.127948113207546,
      "grad_norm": 2.7022290229797363,
      "learning_rate": 7.403139187061955e-06,
      "loss": 0.3953,
      "num_input_tokens_seen": 9081040,
      "step": 13785
    },
    {
      "epoch": 8.130896226415095,
      "grad_norm": 1.7372500896453857,
      "learning_rate": 7.400882792561752e-06,
      "loss": 0.4323,
      "num_input_tokens_seen": 9084016,
      "step": 13790
    },
    {
      "epoch": 8.133844339622641,
      "grad_norm": 1.5180184841156006,
      "learning_rate": 7.39862576241748e-06,
      "loss": 0.41,
      "num_input_tokens_seen": 9087536,
      "step": 13795
    },
    {
      "epoch": 8.13679245283019,
      "grad_norm": 2.320495843887329,
      "learning_rate": 7.396368097226703e-06,
      "loss": 0.3829,
      "num_input_tokens_seen": 9090640,
      "step": 13800
    },
    {
      "epoch": 8.139740566037736,
      "grad_norm": 3.788991928100586,
      "learning_rate": 7.394109797587144e-06,
      "loss": 0.4993,
      "num_input_tokens_seen": 9093840,
      "step": 13805
    },
    {
      "epoch": 8.142688679245284,
      "grad_norm": 1.2949638366699219,
      "learning_rate": 7.3918508640966956e-06,
      "loss": 0.3093,
      "num_input_tokens_seen": 9097808,
      "step": 13810
    },
    {
      "epoch": 8.14563679245283,
      "grad_norm": 1.3919038772583008,
      "learning_rate": 7.389591297353424e-06,
      "loss": 0.4807,
      "num_input_tokens_seen": 9101040,
      "step": 13815
    },
    {
      "epoch": 8.148584905660377,
      "grad_norm": 2.102057933807373,
      "learning_rate": 7.3873310979555565e-06,
      "loss": 0.3468,
      "num_input_tokens_seen": 9103568,
      "step": 13820
    },
    {
      "epoch": 8.151533018867925,
      "grad_norm": 2.3809902667999268,
      "learning_rate": 7.385070266501495e-06,
      "loss": 0.403,
      "num_input_tokens_seen": 9106928,
      "step": 13825
    },
    {
      "epoch": 8.154481132075471,
      "grad_norm": 2.7066032886505127,
      "learning_rate": 7.382808803589798e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 9110000,
      "step": 13830
    },
    {
      "epoch": 8.15742924528302,
      "grad_norm": 2.228416919708252,
      "learning_rate": 7.380546709819204e-06,
      "loss": 0.3627,
      "num_input_tokens_seen": 9112080,
      "step": 13835
    },
    {
      "epoch": 8.160377358490566,
      "grad_norm": 1.5837563276290894,
      "learning_rate": 7.378283985788608e-06,
      "loss": 0.3426,
      "num_input_tokens_seen": 9115088,
      "step": 13840
    },
    {
      "epoch": 8.163325471698114,
      "grad_norm": 1.8645203113555908,
      "learning_rate": 7.376020632097076e-06,
      "loss": 0.3764,
      "num_input_tokens_seen": 9117968,
      "step": 13845
    },
    {
      "epoch": 8.16627358490566,
      "grad_norm": 1.7657654285430908,
      "learning_rate": 7.373756649343841e-06,
      "loss": 0.3365,
      "num_input_tokens_seen": 9121264,
      "step": 13850
    },
    {
      "epoch": 8.169221698113208,
      "grad_norm": 1.653975486755371,
      "learning_rate": 7.371492038128305e-06,
      "loss": 0.3409,
      "num_input_tokens_seen": 9125424,
      "step": 13855
    },
    {
      "epoch": 8.172169811320755,
      "grad_norm": 2.2432234287261963,
      "learning_rate": 7.36922679905003e-06,
      "loss": 0.4168,
      "num_input_tokens_seen": 9128976,
      "step": 13860
    },
    {
      "epoch": 8.175117924528301,
      "grad_norm": 2.2032556533813477,
      "learning_rate": 7.366960932708749e-06,
      "loss": 0.3741,
      "num_input_tokens_seen": 9131984,
      "step": 13865
    },
    {
      "epoch": 8.17806603773585,
      "grad_norm": 2.3998405933380127,
      "learning_rate": 7.364694439704361e-06,
      "loss": 0.5639,
      "num_input_tokens_seen": 9134928,
      "step": 13870
    },
    {
      "epoch": 8.181014150943396,
      "grad_norm": 2.380180835723877,
      "learning_rate": 7.3624273206369264e-06,
      "loss": 0.4021,
      "num_input_tokens_seen": 9138160,
      "step": 13875
    },
    {
      "epoch": 8.183962264150944,
      "grad_norm": 2.151339530944824,
      "learning_rate": 7.360159576106681e-06,
      "loss": 0.4155,
      "num_input_tokens_seen": 9142128,
      "step": 13880
    },
    {
      "epoch": 8.18691037735849,
      "grad_norm": 2.329141855239868,
      "learning_rate": 7.357891206714014e-06,
      "loss": 0.4404,
      "num_input_tokens_seen": 9145264,
      "step": 13885
    },
    {
      "epoch": 8.189858490566039,
      "grad_norm": 1.7748265266418457,
      "learning_rate": 7.355622213059487e-06,
      "loss": 0.3466,
      "num_input_tokens_seen": 9148112,
      "step": 13890
    },
    {
      "epoch": 8.192806603773585,
      "grad_norm": 2.362417221069336,
      "learning_rate": 7.353352595743829e-06,
      "loss": 0.3022,
      "num_input_tokens_seen": 9150704,
      "step": 13895
    },
    {
      "epoch": 8.195754716981131,
      "grad_norm": 3.4135189056396484,
      "learning_rate": 7.351082355367928e-06,
      "loss": 0.3419,
      "num_input_tokens_seen": 9153936,
      "step": 13900
    },
    {
      "epoch": 8.19870283018868,
      "grad_norm": 3.092414379119873,
      "learning_rate": 7.34881149253284e-06,
      "loss": 0.5638,
      "num_input_tokens_seen": 9157776,
      "step": 13905
    },
    {
      "epoch": 8.201650943396226,
      "grad_norm": 5.292367458343506,
      "learning_rate": 7.346540007839787e-06,
      "loss": 0.4448,
      "num_input_tokens_seen": 9161488,
      "step": 13910
    },
    {
      "epoch": 8.204599056603774,
      "grad_norm": 3.20351243019104,
      "learning_rate": 7.344267901890154e-06,
      "loss": 0.4064,
      "num_input_tokens_seen": 9164208,
      "step": 13915
    },
    {
      "epoch": 8.20754716981132,
      "grad_norm": 1.4838943481445312,
      "learning_rate": 7.341995175285491e-06,
      "loss": 0.3181,
      "num_input_tokens_seen": 9167344,
      "step": 13920
    },
    {
      "epoch": 8.210495283018869,
      "grad_norm": 1.9911689758300781,
      "learning_rate": 7.339721828627512e-06,
      "loss": 0.2954,
      "num_input_tokens_seen": 9171088,
      "step": 13925
    },
    {
      "epoch": 8.213443396226415,
      "grad_norm": 6.70450496673584,
      "learning_rate": 7.337447862518096e-06,
      "loss": 0.4682,
      "num_input_tokens_seen": 9174128,
      "step": 13930
    },
    {
      "epoch": 8.216391509433961,
      "grad_norm": 2.406334400177002,
      "learning_rate": 7.335173277559282e-06,
      "loss": 0.515,
      "num_input_tokens_seen": 9178576,
      "step": 13935
    },
    {
      "epoch": 8.21933962264151,
      "grad_norm": 2.0694847106933594,
      "learning_rate": 7.332898074353281e-06,
      "loss": 0.4989,
      "num_input_tokens_seen": 9181968,
      "step": 13940
    },
    {
      "epoch": 8.222287735849056,
      "grad_norm": 3.3669493198394775,
      "learning_rate": 7.330622253502461e-06,
      "loss": 0.5387,
      "num_input_tokens_seen": 9184912,
      "step": 13945
    },
    {
      "epoch": 8.225235849056604,
      "grad_norm": 2.382093667984009,
      "learning_rate": 7.3283458156093534e-06,
      "loss": 0.5097,
      "num_input_tokens_seen": 9188176,
      "step": 13950
    },
    {
      "epoch": 8.22818396226415,
      "grad_norm": 2.3606879711151123,
      "learning_rate": 7.326068761276657e-06,
      "loss": 0.4382,
      "num_input_tokens_seen": 9191376,
      "step": 13955
    },
    {
      "epoch": 8.231132075471699,
      "grad_norm": 1.553068995475769,
      "learning_rate": 7.323791091107231e-06,
      "loss": 0.4316,
      "num_input_tokens_seen": 9194800,
      "step": 13960
    },
    {
      "epoch": 8.234080188679245,
      "grad_norm": 1.6381582021713257,
      "learning_rate": 7.3215128057040986e-06,
      "loss": 0.356,
      "num_input_tokens_seen": 9197680,
      "step": 13965
    },
    {
      "epoch": 8.237028301886792,
      "grad_norm": 2.688473701477051,
      "learning_rate": 7.319233905670447e-06,
      "loss": 0.4317,
      "num_input_tokens_seen": 9200976,
      "step": 13970
    },
    {
      "epoch": 8.23997641509434,
      "grad_norm": 1.9909669160842896,
      "learning_rate": 7.316954391609622e-06,
      "loss": 0.4006,
      "num_input_tokens_seen": 9204944,
      "step": 13975
    },
    {
      "epoch": 8.242924528301886,
      "grad_norm": 2.326504945755005,
      "learning_rate": 7.314674264125137e-06,
      "loss": 0.3648,
      "num_input_tokens_seen": 9207536,
      "step": 13980
    },
    {
      "epoch": 8.245872641509434,
      "grad_norm": 2.967050313949585,
      "learning_rate": 7.312393523820665e-06,
      "loss": 0.4463,
      "num_input_tokens_seen": 9213136,
      "step": 13985
    },
    {
      "epoch": 8.24882075471698,
      "grad_norm": 1.58027184009552,
      "learning_rate": 7.310112171300041e-06,
      "loss": 0.3641,
      "num_input_tokens_seen": 9216752,
      "step": 13990
    },
    {
      "epoch": 8.251768867924529,
      "grad_norm": 3.6315321922302246,
      "learning_rate": 7.307830207167263e-06,
      "loss": 0.4339,
      "num_input_tokens_seen": 9220240,
      "step": 13995
    },
    {
      "epoch": 8.254716981132075,
      "grad_norm": 2.065805673599243,
      "learning_rate": 7.305547632026493e-06,
      "loss": 0.3773,
      "num_input_tokens_seen": 9222864,
      "step": 14000
    },
    {
      "epoch": 8.257665094339623,
      "grad_norm": 1.5632236003875732,
      "learning_rate": 7.3032644464820515e-06,
      "loss": 0.2655,
      "num_input_tokens_seen": 9225296,
      "step": 14005
    },
    {
      "epoch": 8.26061320754717,
      "grad_norm": 2.504754066467285,
      "learning_rate": 7.30098065113842e-06,
      "loss": 0.4156,
      "num_input_tokens_seen": 9227696,
      "step": 14010
    },
    {
      "epoch": 8.263561320754716,
      "grad_norm": 3.6041924953460693,
      "learning_rate": 7.298696246600244e-06,
      "loss": 0.51,
      "num_input_tokens_seen": 9230384,
      "step": 14015
    },
    {
      "epoch": 8.266509433962264,
      "grad_norm": 1.8761990070343018,
      "learning_rate": 7.2964112334723315e-06,
      "loss": 0.3572,
      "num_input_tokens_seen": 9234384,
      "step": 14020
    },
    {
      "epoch": 8.26945754716981,
      "grad_norm": 2.981865167617798,
      "learning_rate": 7.294125612359647e-06,
      "loss": 0.4624,
      "num_input_tokens_seen": 9239056,
      "step": 14025
    },
    {
      "epoch": 8.272405660377359,
      "grad_norm": 2.9544994831085205,
      "learning_rate": 7.291839383867318e-06,
      "loss": 0.3987,
      "num_input_tokens_seen": 9241616,
      "step": 14030
    },
    {
      "epoch": 8.275353773584905,
      "grad_norm": 1.814882516860962,
      "learning_rate": 7.289552548600638e-06,
      "loss": 0.4439,
      "num_input_tokens_seen": 9245200,
      "step": 14035
    },
    {
      "epoch": 8.278301886792454,
      "grad_norm": 1.5062743425369263,
      "learning_rate": 7.287265107165052e-06,
      "loss": 0.322,
      "num_input_tokens_seen": 9248304,
      "step": 14040
    },
    {
      "epoch": 8.28125,
      "grad_norm": 1.0594642162322998,
      "learning_rate": 7.284977060166171e-06,
      "loss": 0.3898,
      "num_input_tokens_seen": 9251664,
      "step": 14045
    },
    {
      "epoch": 8.284198113207546,
      "grad_norm": 2.4336118698120117,
      "learning_rate": 7.282688408209766e-06,
      "loss": 0.5222,
      "num_input_tokens_seen": 9254640,
      "step": 14050
    },
    {
      "epoch": 8.287146226415095,
      "grad_norm": 3.5021352767944336,
      "learning_rate": 7.2803991519017655e-06,
      "loss": 0.4618,
      "num_input_tokens_seen": 9257424,
      "step": 14055
    },
    {
      "epoch": 8.290094339622641,
      "grad_norm": 2.5127015113830566,
      "learning_rate": 7.2781092918482634e-06,
      "loss": 0.2689,
      "num_input_tokens_seen": 9261040,
      "step": 14060
    },
    {
      "epoch": 8.29304245283019,
      "grad_norm": 3.9050867557525635,
      "learning_rate": 7.275818828655508e-06,
      "loss": 0.2895,
      "num_input_tokens_seen": 9265744,
      "step": 14065
    },
    {
      "epoch": 8.295990566037736,
      "grad_norm": 1.436152458190918,
      "learning_rate": 7.27352776292991e-06,
      "loss": 0.4313,
      "num_input_tokens_seen": 9269904,
      "step": 14070
    },
    {
      "epoch": 8.298938679245284,
      "grad_norm": 3.703974723815918,
      "learning_rate": 7.271236095278036e-06,
      "loss": 0.3456,
      "num_input_tokens_seen": 9272976,
      "step": 14075
    },
    {
      "epoch": 8.30188679245283,
      "grad_norm": 2.3931262493133545,
      "learning_rate": 7.2689438263066195e-06,
      "loss": 0.5673,
      "num_input_tokens_seen": 9276016,
      "step": 14080
    },
    {
      "epoch": 8.304834905660377,
      "grad_norm": 7.234551429748535,
      "learning_rate": 7.266650956622546e-06,
      "loss": 0.5651,
      "num_input_tokens_seen": 9278640,
      "step": 14085
    },
    {
      "epoch": 8.307783018867925,
      "grad_norm": 2.5820810794830322,
      "learning_rate": 7.2643574868328625e-06,
      "loss": 0.4133,
      "num_input_tokens_seen": 9281744,
      "step": 14090
    },
    {
      "epoch": 8.310731132075471,
      "grad_norm": 2.4964523315429688,
      "learning_rate": 7.262063417544776e-06,
      "loss": 0.5088,
      "num_input_tokens_seen": 9285232,
      "step": 14095
    },
    {
      "epoch": 8.31367924528302,
      "grad_norm": 4.400596618652344,
      "learning_rate": 7.25976874936565e-06,
      "loss": 0.4382,
      "num_input_tokens_seen": 9288912,
      "step": 14100
    },
    {
      "epoch": 8.316627358490566,
      "grad_norm": 1.866166353225708,
      "learning_rate": 7.257473482903009e-06,
      "loss": 0.3888,
      "num_input_tokens_seen": 9292592,
      "step": 14105
    },
    {
      "epoch": 8.319575471698114,
      "grad_norm": 1.731223464012146,
      "learning_rate": 7.255177618764534e-06,
      "loss": 0.3185,
      "num_input_tokens_seen": 9297008,
      "step": 14110
    },
    {
      "epoch": 8.32252358490566,
      "grad_norm": 2.7531936168670654,
      "learning_rate": 7.252881157558065e-06,
      "loss": 0.4918,
      "num_input_tokens_seen": 9300464,
      "step": 14115
    },
    {
      "epoch": 8.325471698113208,
      "grad_norm": 2.119976043701172,
      "learning_rate": 7.250584099891602e-06,
      "loss": 0.349,
      "num_input_tokens_seen": 9303760,
      "step": 14120
    },
    {
      "epoch": 8.328419811320755,
      "grad_norm": 1.5405694246292114,
      "learning_rate": 7.248286446373296e-06,
      "loss": 0.315,
      "num_input_tokens_seen": 9307440,
      "step": 14125
    },
    {
      "epoch": 8.331367924528301,
      "grad_norm": 4.774034023284912,
      "learning_rate": 7.245988197611466e-06,
      "loss": 0.3633,
      "num_input_tokens_seen": 9310704,
      "step": 14130
    },
    {
      "epoch": 8.33431603773585,
      "grad_norm": 1.6064646244049072,
      "learning_rate": 7.2436893542145805e-06,
      "loss": 0.4213,
      "num_input_tokens_seen": 9313744,
      "step": 14135
    },
    {
      "epoch": 8.337264150943396,
      "grad_norm": 2.2275872230529785,
      "learning_rate": 7.241389916791269e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 9318224,
      "step": 14140
    },
    {
      "epoch": 8.340212264150944,
      "grad_norm": 2.707923173904419,
      "learning_rate": 7.239089885950317e-06,
      "loss": 0.4812,
      "num_input_tokens_seen": 9322064,
      "step": 14145
    },
    {
      "epoch": 8.34316037735849,
      "grad_norm": 1.983652949333191,
      "learning_rate": 7.236789262300667e-06,
      "loss": 0.3977,
      "num_input_tokens_seen": 9324912,
      "step": 14150
    },
    {
      "epoch": 8.346108490566039,
      "grad_norm": 1.7230956554412842,
      "learning_rate": 7.23448804645142e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 9327984,
      "step": 14155
    },
    {
      "epoch": 8.349056603773585,
      "grad_norm": 1.245060682296753,
      "learning_rate": 7.232186239011834e-06,
      "loss": 0.2551,
      "num_input_tokens_seen": 9331376,
      "step": 14160
    },
    {
      "epoch": 8.352004716981131,
      "grad_norm": 1.4531240463256836,
      "learning_rate": 7.2298838405913195e-06,
      "loss": 0.4941,
      "num_input_tokens_seen": 9335248,
      "step": 14165
    },
    {
      "epoch": 8.35495283018868,
      "grad_norm": 2.9049429893493652,
      "learning_rate": 7.227580851799448e-06,
      "loss": 0.4153,
      "num_input_tokens_seen": 9338064,
      "step": 14170
    },
    {
      "epoch": 8.357900943396226,
      "grad_norm": 3.1003968715667725,
      "learning_rate": 7.2252772732459455e-06,
      "loss": 0.4723,
      "num_input_tokens_seen": 9341232,
      "step": 14175
    },
    {
      "epoch": 8.360849056603774,
      "grad_norm": 2.5350661277770996,
      "learning_rate": 7.222973105540696e-06,
      "loss": 0.3703,
      "num_input_tokens_seen": 9343856,
      "step": 14180
    },
    {
      "epoch": 8.36379716981132,
      "grad_norm": 2.7308568954467773,
      "learning_rate": 7.2206683492937345e-06,
      "loss": 0.4622,
      "num_input_tokens_seen": 9346768,
      "step": 14185
    },
    {
      "epoch": 8.366745283018869,
      "grad_norm": 2.5367977619171143,
      "learning_rate": 7.218363005115259e-06,
      "loss": 0.4127,
      "num_input_tokens_seen": 9349264,
      "step": 14190
    },
    {
      "epoch": 8.369693396226415,
      "grad_norm": 1.7319698333740234,
      "learning_rate": 7.216057073615617e-06,
      "loss": 0.4411,
      "num_input_tokens_seen": 9352304,
      "step": 14195
    },
    {
      "epoch": 8.372641509433961,
      "grad_norm": 2.124692440032959,
      "learning_rate": 7.21375055540531e-06,
      "loss": 0.4545,
      "num_input_tokens_seen": 9356112,
      "step": 14200
    },
    {
      "epoch": 8.37558962264151,
      "grad_norm": 1.6472656726837158,
      "learning_rate": 7.211443451095007e-06,
      "loss": 0.3846,
      "num_input_tokens_seen": 9359216,
      "step": 14205
    },
    {
      "epoch": 8.378537735849056,
      "grad_norm": 2.7865869998931885,
      "learning_rate": 7.2091357612955185e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 9362160,
      "step": 14210
    },
    {
      "epoch": 8.381485849056604,
      "grad_norm": 1.3687759637832642,
      "learning_rate": 7.206827486617816e-06,
      "loss": 0.394,
      "num_input_tokens_seen": 9365616,
      "step": 14215
    },
    {
      "epoch": 8.38443396226415,
      "grad_norm": 2.360454797744751,
      "learning_rate": 7.204518627673026e-06,
      "loss": 0.4117,
      "num_input_tokens_seen": 9368880,
      "step": 14220
    },
    {
      "epoch": 8.387382075471699,
      "grad_norm": 1.9314924478530884,
      "learning_rate": 7.202209185072428e-06,
      "loss": 0.3784,
      "num_input_tokens_seen": 9371600,
      "step": 14225
    },
    {
      "epoch": 8.390330188679245,
      "grad_norm": 3.0449440479278564,
      "learning_rate": 7.199899159427457e-06,
      "loss": 0.4355,
      "num_input_tokens_seen": 9374992,
      "step": 14230
    },
    {
      "epoch": 8.393278301886792,
      "grad_norm": 2.6756412982940674,
      "learning_rate": 7.1975885513497035e-06,
      "loss": 0.4147,
      "num_input_tokens_seen": 9378096,
      "step": 14235
    },
    {
      "epoch": 8.39622641509434,
      "grad_norm": 1.9031940698623657,
      "learning_rate": 7.195277361450909e-06,
      "loss": 0.3048,
      "num_input_tokens_seen": 9381200,
      "step": 14240
    },
    {
      "epoch": 8.399174528301886,
      "grad_norm": 4.269694805145264,
      "learning_rate": 7.192965590342973e-06,
      "loss": 0.5134,
      "num_input_tokens_seen": 9383760,
      "step": 14245
    },
    {
      "epoch": 8.402122641509434,
      "grad_norm": 2.8767504692077637,
      "learning_rate": 7.190653238637945e-06,
      "loss": 0.516,
      "num_input_tokens_seen": 9386672,
      "step": 14250
    },
    {
      "epoch": 8.40507075471698,
      "grad_norm": 3.0686724185943604,
      "learning_rate": 7.18834030694803e-06,
      "loss": 0.4716,
      "num_input_tokens_seen": 9390128,
      "step": 14255
    },
    {
      "epoch": 8.408018867924529,
      "grad_norm": 2.6923599243164062,
      "learning_rate": 7.186026795885589e-06,
      "loss": 0.532,
      "num_input_tokens_seen": 9393328,
      "step": 14260
    },
    {
      "epoch": 8.410966981132075,
      "grad_norm": 3.0450279712677,
      "learning_rate": 7.183712706063133e-06,
      "loss": 0.6031,
      "num_input_tokens_seen": 9396336,
      "step": 14265
    },
    {
      "epoch": 8.413915094339623,
      "grad_norm": 2.337407350540161,
      "learning_rate": 7.1813980380933255e-06,
      "loss": 0.3649,
      "num_input_tokens_seen": 9399600,
      "step": 14270
    },
    {
      "epoch": 8.41686320754717,
      "grad_norm": 2.3568718433380127,
      "learning_rate": 7.179082792588986e-06,
      "loss": 0.3658,
      "num_input_tokens_seen": 9402864,
      "step": 14275
    },
    {
      "epoch": 8.419811320754716,
      "grad_norm": 2.629756450653076,
      "learning_rate": 7.176766970163087e-06,
      "loss": 0.3879,
      "num_input_tokens_seen": 9406256,
      "step": 14280
    },
    {
      "epoch": 8.422759433962264,
      "grad_norm": 3.610945224761963,
      "learning_rate": 7.17445057142875e-06,
      "loss": 0.4086,
      "num_input_tokens_seen": 9409872,
      "step": 14285
    },
    {
      "epoch": 8.42570754716981,
      "grad_norm": 3.0907557010650635,
      "learning_rate": 7.172133596999253e-06,
      "loss": 0.3461,
      "num_input_tokens_seen": 9413072,
      "step": 14290
    },
    {
      "epoch": 8.428655660377359,
      "grad_norm": 3.2312984466552734,
      "learning_rate": 7.1698160474880255e-06,
      "loss": 0.448,
      "num_input_tokens_seen": 9415536,
      "step": 14295
    },
    {
      "epoch": 8.431603773584905,
      "grad_norm": 3.0398788452148438,
      "learning_rate": 7.167497923508648e-06,
      "loss": 0.3945,
      "num_input_tokens_seen": 9418384,
      "step": 14300
    },
    {
      "epoch": 8.434551886792454,
      "grad_norm": 1.4276421070098877,
      "learning_rate": 7.165179225674854e-06,
      "loss": 0.4651,
      "num_input_tokens_seen": 9421360,
      "step": 14305
    },
    {
      "epoch": 8.4375,
      "grad_norm": 2.8382999897003174,
      "learning_rate": 7.1628599546005276e-06,
      "loss": 0.3265,
      "num_input_tokens_seen": 9424208,
      "step": 14310
    },
    {
      "epoch": 8.440448113207546,
      "grad_norm": 2.693293571472168,
      "learning_rate": 7.160540110899708e-06,
      "loss": 0.3163,
      "num_input_tokens_seen": 9427632,
      "step": 14315
    },
    {
      "epoch": 8.443396226415095,
      "grad_norm": 3.072364091873169,
      "learning_rate": 7.158219695186582e-06,
      "loss": 0.3772,
      "num_input_tokens_seen": 9431184,
      "step": 14320
    },
    {
      "epoch": 8.446344339622641,
      "grad_norm": 1.7499140501022339,
      "learning_rate": 7.1558987080754905e-06,
      "loss": 0.3829,
      "num_input_tokens_seen": 9434256,
      "step": 14325
    },
    {
      "epoch": 8.44929245283019,
      "grad_norm": 2.254178524017334,
      "learning_rate": 7.1535771501809245e-06,
      "loss": 0.5402,
      "num_input_tokens_seen": 9441680,
      "step": 14330
    },
    {
      "epoch": 8.452240566037736,
      "grad_norm": 1.8849350214004517,
      "learning_rate": 7.151255022117527e-06,
      "loss": 0.4155,
      "num_input_tokens_seen": 9444432,
      "step": 14335
    },
    {
      "epoch": 8.455188679245284,
      "grad_norm": 2.9151296615600586,
      "learning_rate": 7.148932324500091e-06,
      "loss": 0.509,
      "num_input_tokens_seen": 9447440,
      "step": 14340
    },
    {
      "epoch": 8.45813679245283,
      "grad_norm": 2.2213597297668457,
      "learning_rate": 7.146609057943559e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 9454768,
      "step": 14345
    },
    {
      "epoch": 8.461084905660377,
      "grad_norm": 2.137998342514038,
      "learning_rate": 7.14428522306303e-06,
      "loss": 0.5042,
      "num_input_tokens_seen": 9457872,
      "step": 14350
    },
    {
      "epoch": 8.464033018867925,
      "grad_norm": 3.301802635192871,
      "learning_rate": 7.141960820473745e-06,
      "loss": 0.5227,
      "num_input_tokens_seen": 9461008,
      "step": 14355
    },
    {
      "epoch": 8.466981132075471,
      "grad_norm": 2.393718957901001,
      "learning_rate": 7.139635850791102e-06,
      "loss": 0.4048,
      "num_input_tokens_seen": 9464080,
      "step": 14360
    },
    {
      "epoch": 8.46992924528302,
      "grad_norm": 3.436851739883423,
      "learning_rate": 7.137310314630647e-06,
      "loss": 0.3527,
      "num_input_tokens_seen": 9466672,
      "step": 14365
    },
    {
      "epoch": 8.472877358490566,
      "grad_norm": 2.316643476486206,
      "learning_rate": 7.134984212608074e-06,
      "loss": 0.3123,
      "num_input_tokens_seen": 9472528,
      "step": 14370
    },
    {
      "epoch": 8.475825471698114,
      "grad_norm": 2.429934024810791,
      "learning_rate": 7.13265754533923e-06,
      "loss": 0.5056,
      "num_input_tokens_seen": 9475792,
      "step": 14375
    },
    {
      "epoch": 8.47877358490566,
      "grad_norm": 1.8633633852005005,
      "learning_rate": 7.130330313440109e-06,
      "loss": 0.3099,
      "num_input_tokens_seen": 9479120,
      "step": 14380
    },
    {
      "epoch": 8.481721698113208,
      "grad_norm": 2.602985143661499,
      "learning_rate": 7.128002517526856e-06,
      "loss": 0.3164,
      "num_input_tokens_seen": 9482224,
      "step": 14385
    },
    {
      "epoch": 8.484669811320755,
      "grad_norm": 2.8924527168273926,
      "learning_rate": 7.1256741582157654e-06,
      "loss": 0.3905,
      "num_input_tokens_seen": 9484816,
      "step": 14390
    },
    {
      "epoch": 8.487617924528301,
      "grad_norm": 2.432419538497925,
      "learning_rate": 7.123345236123282e-06,
      "loss": 0.3638,
      "num_input_tokens_seen": 9487248,
      "step": 14395
    },
    {
      "epoch": 8.49056603773585,
      "grad_norm": 1.839847207069397,
      "learning_rate": 7.121015751865994e-06,
      "loss": 0.3615,
      "num_input_tokens_seen": 9490672,
      "step": 14400
    },
    {
      "epoch": 8.493514150943396,
      "grad_norm": 2.759432077407837,
      "learning_rate": 7.118685706060645e-06,
      "loss": 0.473,
      "num_input_tokens_seen": 9494288,
      "step": 14405
    },
    {
      "epoch": 8.496462264150944,
      "grad_norm": 2.478773593902588,
      "learning_rate": 7.116355099324126e-06,
      "loss": 0.2988,
      "num_input_tokens_seen": 9497264,
      "step": 14410
    },
    {
      "epoch": 8.49941037735849,
      "grad_norm": 7.984093189239502,
      "learning_rate": 7.114023932273471e-06,
      "loss": 0.36,
      "num_input_tokens_seen": 9499696,
      "step": 14415
    },
    {
      "epoch": 8.502358490566039,
      "grad_norm": 1.5038379430770874,
      "learning_rate": 7.111692205525871e-06,
      "loss": 0.2717,
      "num_input_tokens_seen": 9503408,
      "step": 14420
    },
    {
      "epoch": 8.505306603773585,
      "grad_norm": 3.8995625972747803,
      "learning_rate": 7.109359919698658e-06,
      "loss": 0.5106,
      "num_input_tokens_seen": 9507184,
      "step": 14425
    },
    {
      "epoch": 8.508254716981131,
      "grad_norm": 3.906346559524536,
      "learning_rate": 7.107027075409316e-06,
      "loss": 0.5292,
      "num_input_tokens_seen": 9510480,
      "step": 14430
    },
    {
      "epoch": 8.51120283018868,
      "grad_norm": 5.833251476287842,
      "learning_rate": 7.104693673275475e-06,
      "loss": 0.417,
      "num_input_tokens_seen": 9512912,
      "step": 14435
    },
    {
      "epoch": 8.514150943396226,
      "grad_norm": 2.2036306858062744,
      "learning_rate": 7.1023597139149115e-06,
      "loss": 0.393,
      "num_input_tokens_seen": 9516400,
      "step": 14440
    },
    {
      "epoch": 8.517099056603774,
      "grad_norm": 1.6834121942520142,
      "learning_rate": 7.100025197945555e-06,
      "loss": 0.2696,
      "num_input_tokens_seen": 9519536,
      "step": 14445
    },
    {
      "epoch": 8.52004716981132,
      "grad_norm": 2.2519686222076416,
      "learning_rate": 7.097690125985476e-06,
      "loss": 0.3735,
      "num_input_tokens_seen": 9523344,
      "step": 14450
    },
    {
      "epoch": 8.522995283018869,
      "grad_norm": 2.858194351196289,
      "learning_rate": 7.095354498652895e-06,
      "loss": 0.5123,
      "num_input_tokens_seen": 9526032,
      "step": 14455
    },
    {
      "epoch": 8.525943396226415,
      "grad_norm": 2.276395320892334,
      "learning_rate": 7.093018316566182e-06,
      "loss": 0.3797,
      "num_input_tokens_seen": 9529424,
      "step": 14460
    },
    {
      "epoch": 8.528891509433961,
      "grad_norm": 1.7488967180252075,
      "learning_rate": 7.0906815803438465e-06,
      "loss": 0.3898,
      "num_input_tokens_seen": 9532624,
      "step": 14465
    },
    {
      "epoch": 8.53183962264151,
      "grad_norm": 4.829418182373047,
      "learning_rate": 7.088344290604554e-06,
      "loss": 0.4562,
      "num_input_tokens_seen": 9535664,
      "step": 14470
    },
    {
      "epoch": 8.534787735849056,
      "grad_norm": 1.6662189960479736,
      "learning_rate": 7.086006447967111e-06,
      "loss": 0.2613,
      "num_input_tokens_seen": 9539056,
      "step": 14475
    },
    {
      "epoch": 8.537735849056604,
      "grad_norm": 2.655885696411133,
      "learning_rate": 7.08366805305047e-06,
      "loss": 0.3961,
      "num_input_tokens_seen": 9541968,
      "step": 14480
    },
    {
      "epoch": 8.54068396226415,
      "grad_norm": 2.8141062259674072,
      "learning_rate": 7.08132910647373e-06,
      "loss": 0.3961,
      "num_input_tokens_seen": 9545104,
      "step": 14485
    },
    {
      "epoch": 8.543632075471699,
      "grad_norm": 3.1151087284088135,
      "learning_rate": 7.078989608856142e-06,
      "loss": 0.4737,
      "num_input_tokens_seen": 9547696,
      "step": 14490
    },
    {
      "epoch": 8.546580188679245,
      "grad_norm": 2.5008981227874756,
      "learning_rate": 7.076649560817092e-06,
      "loss": 0.4596,
      "num_input_tokens_seen": 9550384,
      "step": 14495
    },
    {
      "epoch": 8.549528301886792,
      "grad_norm": 2.0268630981445312,
      "learning_rate": 7.0743089629761245e-06,
      "loss": 0.4897,
      "num_input_tokens_seen": 9553200,
      "step": 14500
    },
    {
      "epoch": 8.55247641509434,
      "grad_norm": 2.66355562210083,
      "learning_rate": 7.071967815952917e-06,
      "loss": 0.3657,
      "num_input_tokens_seen": 9558800,
      "step": 14505
    },
    {
      "epoch": 8.555424528301886,
      "grad_norm": 2.482041597366333,
      "learning_rate": 7.0696261203673e-06,
      "loss": 0.3844,
      "num_input_tokens_seen": 9562224,
      "step": 14510
    },
    {
      "epoch": 8.558372641509434,
      "grad_norm": 2.3528337478637695,
      "learning_rate": 7.067283876839249e-06,
      "loss": 0.4624,
      "num_input_tokens_seen": 9565200,
      "step": 14515
    },
    {
      "epoch": 8.56132075471698,
      "grad_norm": 2.5191736221313477,
      "learning_rate": 7.064941085988884e-06,
      "loss": 0.3077,
      "num_input_tokens_seen": 9568080,
      "step": 14520
    },
    {
      "epoch": 8.564268867924529,
      "grad_norm": 2.3859658241271973,
      "learning_rate": 7.062597748436464e-06,
      "loss": 0.45,
      "num_input_tokens_seen": 9571728,
      "step": 14525
    },
    {
      "epoch": 8.567216981132075,
      "grad_norm": 2.8242785930633545,
      "learning_rate": 7.060253864802402e-06,
      "loss": 0.4319,
      "num_input_tokens_seen": 9575760,
      "step": 14530
    },
    {
      "epoch": 8.570165094339622,
      "grad_norm": 4.674290657043457,
      "learning_rate": 7.05790943570725e-06,
      "loss": 0.3891,
      "num_input_tokens_seen": 9579184,
      "step": 14535
    },
    {
      "epoch": 8.57311320754717,
      "grad_norm": 2.893705368041992,
      "learning_rate": 7.055564461771704e-06,
      "loss": 0.4471,
      "num_input_tokens_seen": 9582160,
      "step": 14540
    },
    {
      "epoch": 8.576061320754716,
      "grad_norm": 2.4283933639526367,
      "learning_rate": 7.053218943616611e-06,
      "loss": 0.3732,
      "num_input_tokens_seen": 9584880,
      "step": 14545
    },
    {
      "epoch": 8.579009433962264,
      "grad_norm": 1.3208401203155518,
      "learning_rate": 7.050872881862952e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 9588176,
      "step": 14550
    },
    {
      "epoch": 8.58195754716981,
      "grad_norm": 1.6437225341796875,
      "learning_rate": 7.04852627713186e-06,
      "loss": 0.5314,
      "num_input_tokens_seen": 9591280,
      "step": 14555
    },
    {
      "epoch": 8.584905660377359,
      "grad_norm": 1.9384218454360962,
      "learning_rate": 7.0461791300446055e-06,
      "loss": 0.3788,
      "num_input_tokens_seen": 9594704,
      "step": 14560
    },
    {
      "epoch": 8.587853773584905,
      "grad_norm": 4.311281204223633,
      "learning_rate": 7.043831441222611e-06,
      "loss": 0.4005,
      "num_input_tokens_seen": 9596944,
      "step": 14565
    },
    {
      "epoch": 8.590801886792454,
      "grad_norm": 2.4299371242523193,
      "learning_rate": 7.04148321128743e-06,
      "loss": 0.4399,
      "num_input_tokens_seen": 9599792,
      "step": 14570
    },
    {
      "epoch": 8.59375,
      "grad_norm": 2.7164456844329834,
      "learning_rate": 7.039134440860773e-06,
      "loss": 0.4287,
      "num_input_tokens_seen": 9603760,
      "step": 14575
    },
    {
      "epoch": 8.596698113207546,
      "grad_norm": 1.4512977600097656,
      "learning_rate": 7.036785130564484e-06,
      "loss": 0.4367,
      "num_input_tokens_seen": 9606480,
      "step": 14580
    },
    {
      "epoch": 8.599646226415095,
      "grad_norm": 3.142260789871216,
      "learning_rate": 7.0344352810205544e-06,
      "loss": 0.3861,
      "num_input_tokens_seen": 9609264,
      "step": 14585
    },
    {
      "epoch": 8.602594339622641,
      "grad_norm": 3.457306385040283,
      "learning_rate": 7.032084892851115e-06,
      "loss": 0.2969,
      "num_input_tokens_seen": 9612528,
      "step": 14590
    },
    {
      "epoch": 8.60554245283019,
      "grad_norm": 4.14358377456665,
      "learning_rate": 7.0297339666784425e-06,
      "loss": 0.3879,
      "num_input_tokens_seen": 9615600,
      "step": 14595
    },
    {
      "epoch": 8.608490566037736,
      "grad_norm": 2.4456746578216553,
      "learning_rate": 7.0273825031249556e-06,
      "loss": 0.4943,
      "num_input_tokens_seen": 9618928,
      "step": 14600
    },
    {
      "epoch": 8.611438679245284,
      "grad_norm": 2.7797296047210693,
      "learning_rate": 7.025030502813213e-06,
      "loss": 0.5256,
      "num_input_tokens_seen": 9622512,
      "step": 14605
    },
    {
      "epoch": 8.61438679245283,
      "grad_norm": 2.4663758277893066,
      "learning_rate": 7.022677966365917e-06,
      "loss": 0.3294,
      "num_input_tokens_seen": 9625552,
      "step": 14610
    },
    {
      "epoch": 8.617334905660378,
      "grad_norm": 5.055086135864258,
      "learning_rate": 7.020324894405913e-06,
      "loss": 0.3854,
      "num_input_tokens_seen": 9627888,
      "step": 14615
    },
    {
      "epoch": 8.620283018867925,
      "grad_norm": 1.9466204643249512,
      "learning_rate": 7.017971287556185e-06,
      "loss": 0.4034,
      "num_input_tokens_seen": 9631440,
      "step": 14620
    },
    {
      "epoch": 8.623231132075471,
      "grad_norm": 2.4704909324645996,
      "learning_rate": 7.015617146439863e-06,
      "loss": 0.3278,
      "num_input_tokens_seen": 9636368,
      "step": 14625
    },
    {
      "epoch": 8.62617924528302,
      "grad_norm": 2.8759069442749023,
      "learning_rate": 7.0132624716802125e-06,
      "loss": 0.514,
      "num_input_tokens_seen": 9639344,
      "step": 14630
    },
    {
      "epoch": 8.629127358490566,
      "grad_norm": 2.3839504718780518,
      "learning_rate": 7.0109072639006474e-06,
      "loss": 0.3615,
      "num_input_tokens_seen": 9641808,
      "step": 14635
    },
    {
      "epoch": 8.632075471698114,
      "grad_norm": 2.29826283454895,
      "learning_rate": 7.008551523724717e-06,
      "loss": 0.3708,
      "num_input_tokens_seen": 9645104,
      "step": 14640
    },
    {
      "epoch": 8.63502358490566,
      "grad_norm": 1.2743711471557617,
      "learning_rate": 7.006195251776116e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 9648848,
      "step": 14645
    },
    {
      "epoch": 8.637971698113208,
      "grad_norm": 1.4084073305130005,
      "learning_rate": 7.003838448678674e-06,
      "loss": 0.31,
      "num_input_tokens_seen": 9652752,
      "step": 14650
    },
    {
      "epoch": 8.640919811320755,
      "grad_norm": 2.4081008434295654,
      "learning_rate": 7.0014811150563675e-06,
      "loss": 0.3395,
      "num_input_tokens_seen": 9656048,
      "step": 14655
    },
    {
      "epoch": 8.643867924528301,
      "grad_norm": 2.8717541694641113,
      "learning_rate": 6.999123251533311e-06,
      "loss": 0.3764,
      "num_input_tokens_seen": 9659472,
      "step": 14660
    },
    {
      "epoch": 8.64681603773585,
      "grad_norm": 4.054232120513916,
      "learning_rate": 6.996764858733756e-06,
      "loss": 0.4102,
      "num_input_tokens_seen": 9662544,
      "step": 14665
    },
    {
      "epoch": 8.649764150943396,
      "grad_norm": 2.5582754611968994,
      "learning_rate": 6.994405937282099e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 9665168,
      "step": 14670
    },
    {
      "epoch": 8.652712264150944,
      "grad_norm": 2.305922031402588,
      "learning_rate": 6.9920464878028745e-06,
      "loss": 0.5294,
      "num_input_tokens_seen": 9668336,
      "step": 14675
    },
    {
      "epoch": 8.65566037735849,
      "grad_norm": 2.139122486114502,
      "learning_rate": 6.989686510920758e-06,
      "loss": 0.4458,
      "num_input_tokens_seen": 9671344,
      "step": 14680
    },
    {
      "epoch": 8.658608490566039,
      "grad_norm": 2.0792086124420166,
      "learning_rate": 6.9873260072605634e-06,
      "loss": 0.4076,
      "num_input_tokens_seen": 9674480,
      "step": 14685
    },
    {
      "epoch": 8.661556603773585,
      "grad_norm": 2.4613237380981445,
      "learning_rate": 6.984964977447243e-06,
      "loss": 0.3586,
      "num_input_tokens_seen": 9678544,
      "step": 14690
    },
    {
      "epoch": 8.664504716981131,
      "grad_norm": 1.7798255681991577,
      "learning_rate": 6.982603422105889e-06,
      "loss": 0.3537,
      "num_input_tokens_seen": 9681040,
      "step": 14695
    },
    {
      "epoch": 8.66745283018868,
      "grad_norm": 1.7896438837051392,
      "learning_rate": 6.980241341861736e-06,
      "loss": 0.4171,
      "num_input_tokens_seen": 9684304,
      "step": 14700
    },
    {
      "epoch": 8.670400943396226,
      "grad_norm": 2.359628200531006,
      "learning_rate": 6.977878737340153e-06,
      "loss": 0.3896,
      "num_input_tokens_seen": 9687728,
      "step": 14705
    },
    {
      "epoch": 8.673349056603774,
      "grad_norm": 3.1049885749816895,
      "learning_rate": 6.97551560916665e-06,
      "loss": 0.3111,
      "num_input_tokens_seen": 9690512,
      "step": 14710
    },
    {
      "epoch": 8.67629716981132,
      "grad_norm": 3.458021640777588,
      "learning_rate": 6.973151957966875e-06,
      "loss": 0.4103,
      "num_input_tokens_seen": 9693776,
      "step": 14715
    },
    {
      "epoch": 8.679245283018869,
      "grad_norm": 1.738378882408142,
      "learning_rate": 6.970787784366616e-06,
      "loss": 0.3441,
      "num_input_tokens_seen": 9697584,
      "step": 14720
    },
    {
      "epoch": 8.682193396226415,
      "grad_norm": 1.0696730613708496,
      "learning_rate": 6.968423088991797e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 9701744,
      "step": 14725
    },
    {
      "epoch": 8.685141509433961,
      "grad_norm": 2.8522422313690186,
      "learning_rate": 6.966057872468481e-06,
      "loss": 0.3305,
      "num_input_tokens_seen": 9704880,
      "step": 14730
    },
    {
      "epoch": 8.68808962264151,
      "grad_norm": 1.6853421926498413,
      "learning_rate": 6.963692135422872e-06,
      "loss": 0.4792,
      "num_input_tokens_seen": 9708048,
      "step": 14735
    },
    {
      "epoch": 8.691037735849056,
      "grad_norm": 6.120649814605713,
      "learning_rate": 6.961325878481305e-06,
      "loss": 0.4285,
      "num_input_tokens_seen": 9712240,
      "step": 14740
    },
    {
      "epoch": 8.693985849056604,
      "grad_norm": 2.4640419483184814,
      "learning_rate": 6.958959102270259e-06,
      "loss": 0.4784,
      "num_input_tokens_seen": 9714960,
      "step": 14745
    },
    {
      "epoch": 8.69693396226415,
      "grad_norm": 1.4684659242630005,
      "learning_rate": 6.95659180741635e-06,
      "loss": 0.3839,
      "num_input_tokens_seen": 9717968,
      "step": 14750
    },
    {
      "epoch": 8.699882075471699,
      "grad_norm": 1.8378645181655884,
      "learning_rate": 6.954223994546326e-06,
      "loss": 0.5077,
      "num_input_tokens_seen": 9721424,
      "step": 14755
    },
    {
      "epoch": 8.702830188679245,
      "grad_norm": 2.9244353771209717,
      "learning_rate": 6.951855664287077e-06,
      "loss": 0.4426,
      "num_input_tokens_seen": 9724336,
      "step": 14760
    },
    {
      "epoch": 8.705778301886792,
      "grad_norm": 1.7795634269714355,
      "learning_rate": 6.9494868172656304e-06,
      "loss": 0.3786,
      "num_input_tokens_seen": 9727888,
      "step": 14765
    },
    {
      "epoch": 8.70872641509434,
      "grad_norm": 2.670854330062866,
      "learning_rate": 6.947117454109146e-06,
      "loss": 0.4551,
      "num_input_tokens_seen": 9730576,
      "step": 14770
    },
    {
      "epoch": 8.711674528301886,
      "grad_norm": 1.8359160423278809,
      "learning_rate": 6.944747575444924e-06,
      "loss": 0.3004,
      "num_input_tokens_seen": 9733520,
      "step": 14775
    },
    {
      "epoch": 8.714622641509434,
      "grad_norm": 2.8579297065734863,
      "learning_rate": 6.942377181900399e-06,
      "loss": 0.3492,
      "num_input_tokens_seen": 9736368,
      "step": 14780
    },
    {
      "epoch": 8.71757075471698,
      "grad_norm": 3.376680850982666,
      "learning_rate": 6.940006274103146e-06,
      "loss": 0.436,
      "num_input_tokens_seen": 9739152,
      "step": 14785
    },
    {
      "epoch": 8.720518867924529,
      "grad_norm": 3.4427871704101562,
      "learning_rate": 6.93763485268087e-06,
      "loss": 0.3965,
      "num_input_tokens_seen": 9741808,
      "step": 14790
    },
    {
      "epoch": 8.723466981132075,
      "grad_norm": 2.108930826187134,
      "learning_rate": 6.935262918261416e-06,
      "loss": 0.4341,
      "num_input_tokens_seen": 9745104,
      "step": 14795
    },
    {
      "epoch": 8.726415094339622,
      "grad_norm": 1.9158250093460083,
      "learning_rate": 6.932890471472764e-06,
      "loss": 0.3238,
      "num_input_tokens_seen": 9748368,
      "step": 14800
    },
    {
      "epoch": 8.72936320754717,
      "grad_norm": 2.067960500717163,
      "learning_rate": 6.930517512943029e-06,
      "loss": 0.3357,
      "num_input_tokens_seen": 9753648,
      "step": 14805
    },
    {
      "epoch": 8.732311320754716,
      "grad_norm": 2.813508987426758,
      "learning_rate": 6.928144043300463e-06,
      "loss": 0.3539,
      "num_input_tokens_seen": 9756368,
      "step": 14810
    },
    {
      "epoch": 8.735259433962264,
      "grad_norm": 1.6124988794326782,
      "learning_rate": 6.925770063173451e-06,
      "loss": 0.4372,
      "num_input_tokens_seen": 9761200,
      "step": 14815
    },
    {
      "epoch": 8.73820754716981,
      "grad_norm": 1.46916925907135,
      "learning_rate": 6.923395573190514e-06,
      "loss": 0.3592,
      "num_input_tokens_seen": 9764400,
      "step": 14820
    },
    {
      "epoch": 8.741155660377359,
      "grad_norm": 8.405824661254883,
      "learning_rate": 6.921020573980313e-06,
      "loss": 0.3501,
      "num_input_tokens_seen": 9766576,
      "step": 14825
    },
    {
      "epoch": 8.744103773584905,
      "grad_norm": 1.9271302223205566,
      "learning_rate": 6.918645066171634e-06,
      "loss": 0.4067,
      "num_input_tokens_seen": 9769584,
      "step": 14830
    },
    {
      "epoch": 8.747051886792454,
      "grad_norm": 2.30765962600708,
      "learning_rate": 6.916269050393404e-06,
      "loss": 0.3625,
      "num_input_tokens_seen": 9772912,
      "step": 14835
    },
    {
      "epoch": 8.75,
      "grad_norm": 3.688122510910034,
      "learning_rate": 6.913892527274686e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 9779248,
      "step": 14840
    },
    {
      "epoch": 8.752948113207546,
      "grad_norm": 2.7103614807128906,
      "learning_rate": 6.9115154974446716e-06,
      "loss": 0.4122,
      "num_input_tokens_seen": 9782384,
      "step": 14845
    },
    {
      "epoch": 8.755896226415095,
      "grad_norm": 2.5912909507751465,
      "learning_rate": 6.909137961532692e-06,
      "loss": 0.3307,
      "num_input_tokens_seen": 9786032,
      "step": 14850
    },
    {
      "epoch": 8.758844339622641,
      "grad_norm": 2.170214891433716,
      "learning_rate": 6.906759920168209e-06,
      "loss": 0.3788,
      "num_input_tokens_seen": 9789648,
      "step": 14855
    },
    {
      "epoch": 8.76179245283019,
      "grad_norm": 2.3454747200012207,
      "learning_rate": 6.90438137398082e-06,
      "loss": 0.5233,
      "num_input_tokens_seen": 9793264,
      "step": 14860
    },
    {
      "epoch": 8.764740566037736,
      "grad_norm": 1.946526288986206,
      "learning_rate": 6.902002323600252e-06,
      "loss": 0.3027,
      "num_input_tokens_seen": 9795856,
      "step": 14865
    },
    {
      "epoch": 8.767688679245284,
      "grad_norm": 2.0139667987823486,
      "learning_rate": 6.899622769656373e-06,
      "loss": 0.4261,
      "num_input_tokens_seen": 9799696,
      "step": 14870
    },
    {
      "epoch": 8.77063679245283,
      "grad_norm": 2.4342360496520996,
      "learning_rate": 6.897242712779179e-06,
      "loss": 0.3516,
      "num_input_tokens_seen": 9802576,
      "step": 14875
    },
    {
      "epoch": 8.773584905660378,
      "grad_norm": 2.1937906742095947,
      "learning_rate": 6.894862153598802e-06,
      "loss": 0.3853,
      "num_input_tokens_seen": 9806000,
      "step": 14880
    },
    {
      "epoch": 8.776533018867925,
      "grad_norm": 2.7592062950134277,
      "learning_rate": 6.892481092745502e-06,
      "loss": 0.5326,
      "num_input_tokens_seen": 9808720,
      "step": 14885
    },
    {
      "epoch": 8.779481132075471,
      "grad_norm": 4.450175762176514,
      "learning_rate": 6.890099530849677e-06,
      "loss": 0.4566,
      "num_input_tokens_seen": 9812176,
      "step": 14890
    },
    {
      "epoch": 8.78242924528302,
      "grad_norm": 2.820142984390259,
      "learning_rate": 6.887717468541855e-06,
      "loss": 0.4473,
      "num_input_tokens_seen": 9815568,
      "step": 14895
    },
    {
      "epoch": 8.785377358490566,
      "grad_norm": 2.4796173572540283,
      "learning_rate": 6.885334906452696e-06,
      "loss": 0.4307,
      "num_input_tokens_seen": 9817936,
      "step": 14900
    },
    {
      "epoch": 8.788325471698114,
      "grad_norm": 3.846301794052124,
      "learning_rate": 6.882951845212997e-06,
      "loss": 0.5848,
      "num_input_tokens_seen": 9823920,
      "step": 14905
    },
    {
      "epoch": 8.79127358490566,
      "grad_norm": 1.8591939210891724,
      "learning_rate": 6.880568285453682e-06,
      "loss": 0.4125,
      "num_input_tokens_seen": 9827952,
      "step": 14910
    },
    {
      "epoch": 8.794221698113208,
      "grad_norm": 2.9007062911987305,
      "learning_rate": 6.878184227805807e-06,
      "loss": 0.412,
      "num_input_tokens_seen": 9831760,
      "step": 14915
    },
    {
      "epoch": 8.797169811320755,
      "grad_norm": 2.7298314571380615,
      "learning_rate": 6.8757996729005645e-06,
      "loss": 0.3481,
      "num_input_tokens_seen": 9835024,
      "step": 14920
    },
    {
      "epoch": 8.800117924528301,
      "grad_norm": 2.524733543395996,
      "learning_rate": 6.8734146213692756e-06,
      "loss": 0.4268,
      "num_input_tokens_seen": 9838704,
      "step": 14925
    },
    {
      "epoch": 8.80306603773585,
      "grad_norm": 1.7292333841323853,
      "learning_rate": 6.87102907384339e-06,
      "loss": 0.4463,
      "num_input_tokens_seen": 9841392,
      "step": 14930
    },
    {
      "epoch": 8.806014150943396,
      "grad_norm": 2.7160139083862305,
      "learning_rate": 6.868643030954494e-06,
      "loss": 0.3902,
      "num_input_tokens_seen": 9844688,
      "step": 14935
    },
    {
      "epoch": 8.808962264150944,
      "grad_norm": 2.7027535438537598,
      "learning_rate": 6.866256493334302e-06,
      "loss": 0.4415,
      "num_input_tokens_seen": 9848272,
      "step": 14940
    },
    {
      "epoch": 8.81191037735849,
      "grad_norm": 2.26517915725708,
      "learning_rate": 6.863869461614659e-06,
      "loss": 0.378,
      "num_input_tokens_seen": 9851056,
      "step": 14945
    },
    {
      "epoch": 8.814858490566039,
      "grad_norm": 4.310174942016602,
      "learning_rate": 6.861481936427545e-06,
      "loss": 0.3488,
      "num_input_tokens_seen": 9853712,
      "step": 14950
    },
    {
      "epoch": 8.817806603773585,
      "grad_norm": 2.0135416984558105,
      "learning_rate": 6.859093918405067e-06,
      "loss": 0.4817,
      "num_input_tokens_seen": 9856272,
      "step": 14955
    },
    {
      "epoch": 8.820754716981131,
      "grad_norm": 2.9232869148254395,
      "learning_rate": 6.856705408179458e-06,
      "loss": 0.3902,
      "num_input_tokens_seen": 9858864,
      "step": 14960
    },
    {
      "epoch": 8.82370283018868,
      "grad_norm": 2.023966073989868,
      "learning_rate": 6.854316406383093e-06,
      "loss": 0.4003,
      "num_input_tokens_seen": 9862000,
      "step": 14965
    },
    {
      "epoch": 8.826650943396226,
      "grad_norm": 2.5712900161743164,
      "learning_rate": 6.8519269136484665e-06,
      "loss": 0.62,
      "num_input_tokens_seen": 9865232,
      "step": 14970
    },
    {
      "epoch": 8.829599056603774,
      "grad_norm": 2.82016658782959,
      "learning_rate": 6.849536930608208e-06,
      "loss": 0.3342,
      "num_input_tokens_seen": 9867792,
      "step": 14975
    },
    {
      "epoch": 8.83254716981132,
      "grad_norm": 3.0893096923828125,
      "learning_rate": 6.847146457895078e-06,
      "loss": 0.3836,
      "num_input_tokens_seen": 9871216,
      "step": 14980
    },
    {
      "epoch": 8.835495283018869,
      "grad_norm": 3.0179901123046875,
      "learning_rate": 6.8447554961419615e-06,
      "loss": 0.4956,
      "num_input_tokens_seen": 9874448,
      "step": 14985
    },
    {
      "epoch": 8.838443396226415,
      "grad_norm": 2.7465779781341553,
      "learning_rate": 6.842364045981876e-06,
      "loss": 0.4028,
      "num_input_tokens_seen": 9877968,
      "step": 14990
    },
    {
      "epoch": 8.841391509433961,
      "grad_norm": 3.120307445526123,
      "learning_rate": 6.83997210804797e-06,
      "loss": 0.4299,
      "num_input_tokens_seen": 9881968,
      "step": 14995
    },
    {
      "epoch": 8.84433962264151,
      "grad_norm": 2.6601014137268066,
      "learning_rate": 6.837579682973519e-06,
      "loss": 0.4149,
      "num_input_tokens_seen": 9885520,
      "step": 15000
    },
    {
      "epoch": 8.847287735849056,
      "grad_norm": 1.8869493007659912,
      "learning_rate": 6.835186771391926e-06,
      "loss": 0.4172,
      "num_input_tokens_seen": 9889840,
      "step": 15005
    },
    {
      "epoch": 8.850235849056604,
      "grad_norm": 1.5823521614074707,
      "learning_rate": 6.8327933739367266e-06,
      "loss": 0.3497,
      "num_input_tokens_seen": 9893168,
      "step": 15010
    },
    {
      "epoch": 8.85318396226415,
      "grad_norm": 2.4797677993774414,
      "learning_rate": 6.830399491241584e-06,
      "loss": 0.3034,
      "num_input_tokens_seen": 9896752,
      "step": 15015
    },
    {
      "epoch": 8.856132075471699,
      "grad_norm": 2.0257272720336914,
      "learning_rate": 6.828005123940287e-06,
      "loss": 0.4806,
      "num_input_tokens_seen": 9900496,
      "step": 15020
    },
    {
      "epoch": 8.859080188679245,
      "grad_norm": 3.3165676593780518,
      "learning_rate": 6.825610272666754e-06,
      "loss": 0.3455,
      "num_input_tokens_seen": 9903472,
      "step": 15025
    },
    {
      "epoch": 8.862028301886792,
      "grad_norm": 2.032927989959717,
      "learning_rate": 6.823214938055034e-06,
      "loss": 0.3066,
      "num_input_tokens_seen": 9907280,
      "step": 15030
    },
    {
      "epoch": 8.86497641509434,
      "grad_norm": 2.345848321914673,
      "learning_rate": 6.8208191207393e-06,
      "loss": 0.4943,
      "num_input_tokens_seen": 9910128,
      "step": 15035
    },
    {
      "epoch": 8.867924528301886,
      "grad_norm": 2.135824203491211,
      "learning_rate": 6.818422821353859e-06,
      "loss": 0.4882,
      "num_input_tokens_seen": 9914224,
      "step": 15040
    },
    {
      "epoch": 8.870872641509434,
      "grad_norm": 3.2091686725616455,
      "learning_rate": 6.816026040533139e-06,
      "loss": 0.3132,
      "num_input_tokens_seen": 9916912,
      "step": 15045
    },
    {
      "epoch": 8.87382075471698,
      "grad_norm": 2.6855640411376953,
      "learning_rate": 6.8136287789116966e-06,
      "loss": 0.3566,
      "num_input_tokens_seen": 9920464,
      "step": 15050
    },
    {
      "epoch": 8.876768867924529,
      "grad_norm": 5.611797332763672,
      "learning_rate": 6.81123103712422e-06,
      "loss": 0.4491,
      "num_input_tokens_seen": 9923376,
      "step": 15055
    },
    {
      "epoch": 8.879716981132075,
      "grad_norm": 5.340473651885986,
      "learning_rate": 6.808832815805518e-06,
      "loss": 0.5668,
      "num_input_tokens_seen": 9927184,
      "step": 15060
    },
    {
      "epoch": 8.882665094339622,
      "grad_norm": 1.306396722793579,
      "learning_rate": 6.806434115590534e-06,
      "loss": 0.3227,
      "num_input_tokens_seen": 9930384,
      "step": 15065
    },
    {
      "epoch": 8.88561320754717,
      "grad_norm": 2.1514837741851807,
      "learning_rate": 6.804034937114332e-06,
      "loss": 0.3758,
      "num_input_tokens_seen": 9933392,
      "step": 15070
    },
    {
      "epoch": 8.888561320754716,
      "grad_norm": 3.6378889083862305,
      "learning_rate": 6.8016352810121064e-06,
      "loss": 0.4805,
      "num_input_tokens_seen": 9936912,
      "step": 15075
    },
    {
      "epoch": 8.891509433962264,
      "grad_norm": 1.4703682661056519,
      "learning_rate": 6.799235147919176e-06,
      "loss": 0.475,
      "num_input_tokens_seen": 9939888,
      "step": 15080
    },
    {
      "epoch": 8.89445754716981,
      "grad_norm": 2.4102649688720703,
      "learning_rate": 6.796834538470985e-06,
      "loss": 0.4648,
      "num_input_tokens_seen": 9943216,
      "step": 15085
    },
    {
      "epoch": 8.897405660377359,
      "grad_norm": 2.5343873500823975,
      "learning_rate": 6.794433453303106e-06,
      "loss": 0.3022,
      "num_input_tokens_seen": 9946160,
      "step": 15090
    },
    {
      "epoch": 8.900353773584905,
      "grad_norm": 1.7328795194625854,
      "learning_rate": 6.792031893051238e-06,
      "loss": 0.4346,
      "num_input_tokens_seen": 9950352,
      "step": 15095
    },
    {
      "epoch": 8.903301886792454,
      "grad_norm": 1.9488881826400757,
      "learning_rate": 6.789629858351201e-06,
      "loss": 0.4233,
      "num_input_tokens_seen": 9954000,
      "step": 15100
    },
    {
      "epoch": 8.90625,
      "grad_norm": 4.485937118530273,
      "learning_rate": 6.787227349838946e-06,
      "loss": 0.434,
      "num_input_tokens_seen": 9957040,
      "step": 15105
    },
    {
      "epoch": 8.909198113207546,
      "grad_norm": 3.462059497833252,
      "learning_rate": 6.784824368150548e-06,
      "loss": 0.428,
      "num_input_tokens_seen": 9960112,
      "step": 15110
    },
    {
      "epoch": 8.912146226415095,
      "grad_norm": 1.8981168270111084,
      "learning_rate": 6.7824209139222076e-06,
      "loss": 0.4508,
      "num_input_tokens_seen": 9963600,
      "step": 15115
    },
    {
      "epoch": 8.915094339622641,
      "grad_norm": 2.1089026927948,
      "learning_rate": 6.780016987790248e-06,
      "loss": 0.3221,
      "num_input_tokens_seen": 9967120,
      "step": 15120
    },
    {
      "epoch": 8.91804245283019,
      "grad_norm": 3.4344449043273926,
      "learning_rate": 6.7776125903911194e-06,
      "loss": 0.4,
      "num_input_tokens_seen": 9969648,
      "step": 15125
    },
    {
      "epoch": 8.920990566037736,
      "grad_norm": 3.989816665649414,
      "learning_rate": 6.775207722361396e-06,
      "loss": 0.3599,
      "num_input_tokens_seen": 9972624,
      "step": 15130
    },
    {
      "epoch": 8.923938679245284,
      "grad_norm": 1.766324520111084,
      "learning_rate": 6.772802384337778e-06,
      "loss": 0.4664,
      "num_input_tokens_seen": 9975856,
      "step": 15135
    },
    {
      "epoch": 8.92688679245283,
      "grad_norm": 1.9986534118652344,
      "learning_rate": 6.770396576957088e-06,
      "loss": 0.3172,
      "num_input_tokens_seen": 9980336,
      "step": 15140
    },
    {
      "epoch": 8.929834905660378,
      "grad_norm": 2.5767836570739746,
      "learning_rate": 6.767990300856274e-06,
      "loss": 0.4792,
      "num_input_tokens_seen": 9982896,
      "step": 15145
    },
    {
      "epoch": 8.932783018867925,
      "grad_norm": 2.476541519165039,
      "learning_rate": 6.765583556672408e-06,
      "loss": 0.4385,
      "num_input_tokens_seen": 9985808,
      "step": 15150
    },
    {
      "epoch": 8.935731132075471,
      "grad_norm": 3.024822950363159,
      "learning_rate": 6.763176345042687e-06,
      "loss": 0.3542,
      "num_input_tokens_seen": 9988400,
      "step": 15155
    },
    {
      "epoch": 8.93867924528302,
      "grad_norm": 1.860805630683899,
      "learning_rate": 6.760768666604429e-06,
      "loss": 0.4296,
      "num_input_tokens_seen": 9991536,
      "step": 15160
    },
    {
      "epoch": 8.941627358490566,
      "grad_norm": 2.425370693206787,
      "learning_rate": 6.758360521995079e-06,
      "loss": 0.3158,
      "num_input_tokens_seen": 9994448,
      "step": 15165
    },
    {
      "epoch": 8.944575471698114,
      "grad_norm": 3.1805763244628906,
      "learning_rate": 6.755951911852202e-06,
      "loss": 0.43,
      "num_input_tokens_seen": 9997584,
      "step": 15170
    },
    {
      "epoch": 8.94752358490566,
      "grad_norm": 3.477212905883789,
      "learning_rate": 6.7535428368134885e-06,
      "loss": 0.332,
      "num_input_tokens_seen": 10000624,
      "step": 15175
    },
    {
      "epoch": 8.950471698113208,
      "grad_norm": 1.6188874244689941,
      "learning_rate": 6.751133297516752e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 10003696,
      "step": 15180
    },
    {
      "epoch": 8.953419811320755,
      "grad_norm": 2.6035051345825195,
      "learning_rate": 6.748723294599928e-06,
      "loss": 0.3971,
      "num_input_tokens_seen": 10006992,
      "step": 15185
    },
    {
      "epoch": 8.956367924528301,
      "grad_norm": 3.2990574836730957,
      "learning_rate": 6.746312828701075e-06,
      "loss": 0.5256,
      "num_input_tokens_seen": 10009456,
      "step": 15190
    },
    {
      "epoch": 8.95931603773585,
      "grad_norm": 1.3801720142364502,
      "learning_rate": 6.743901900458374e-06,
      "loss": 0.4394,
      "num_input_tokens_seen": 10012752,
      "step": 15195
    },
    {
      "epoch": 8.962264150943396,
      "grad_norm": 3.3572351932525635,
      "learning_rate": 6.741490510510129e-06,
      "loss": 0.3747,
      "num_input_tokens_seen": 10016016,
      "step": 15200
    },
    {
      "epoch": 8.965212264150944,
      "grad_norm": 2.1645312309265137,
      "learning_rate": 6.7390786594947665e-06,
      "loss": 0.4653,
      "num_input_tokens_seen": 10020176,
      "step": 15205
    },
    {
      "epoch": 8.96816037735849,
      "grad_norm": 3.0249624252319336,
      "learning_rate": 6.7366663480508335e-06,
      "loss": 0.574,
      "num_input_tokens_seen": 10022864,
      "step": 15210
    },
    {
      "epoch": 8.971108490566039,
      "grad_norm": 1.469975233078003,
      "learning_rate": 6.734253576817002e-06,
      "loss": 0.2998,
      "num_input_tokens_seen": 10025776,
      "step": 15215
    },
    {
      "epoch": 8.974056603773585,
      "grad_norm": 1.9074724912643433,
      "learning_rate": 6.731840346432061e-06,
      "loss": 0.3626,
      "num_input_tokens_seen": 10028560,
      "step": 15220
    },
    {
      "epoch": 8.977004716981131,
      "grad_norm": 2.755237102508545,
      "learning_rate": 6.729426657534922e-06,
      "loss": 0.3695,
      "num_input_tokens_seen": 10031408,
      "step": 15225
    },
    {
      "epoch": 8.97995283018868,
      "grad_norm": 1.7764102220535278,
      "learning_rate": 6.727012510764624e-06,
      "loss": 0.3048,
      "num_input_tokens_seen": 10034960,
      "step": 15230
    },
    {
      "epoch": 8.982900943396226,
      "grad_norm": 2.383282423019409,
      "learning_rate": 6.724597906760322e-06,
      "loss": 0.5018,
      "num_input_tokens_seen": 10038544,
      "step": 15235
    },
    {
      "epoch": 8.985849056603774,
      "grad_norm": 2.1160919666290283,
      "learning_rate": 6.722182846161289e-06,
      "loss": 0.376,
      "num_input_tokens_seen": 10041488,
      "step": 15240
    },
    {
      "epoch": 8.98879716981132,
      "grad_norm": 2.25359845161438,
      "learning_rate": 6.719767329606926e-06,
      "loss": 0.4869,
      "num_input_tokens_seen": 10045680,
      "step": 15245
    },
    {
      "epoch": 8.991745283018869,
      "grad_norm": 1.8892009258270264,
      "learning_rate": 6.717351357736751e-06,
      "loss": 0.312,
      "num_input_tokens_seen": 10048816,
      "step": 15250
    },
    {
      "epoch": 8.994693396226415,
      "grad_norm": 2.624884843826294,
      "learning_rate": 6.7149349311904025e-06,
      "loss": 0.4906,
      "num_input_tokens_seen": 10051632,
      "step": 15255
    },
    {
      "epoch": 8.997641509433961,
      "grad_norm": 1.9678272008895874,
      "learning_rate": 6.712518050607642e-06,
      "loss": 0.3336,
      "num_input_tokens_seen": 10055344,
      "step": 15260
    },
    {
      "epoch": 9.00058962264151,
      "grad_norm": 2.282907247543335,
      "learning_rate": 6.710100716628345e-06,
      "loss": 0.3089,
      "num_input_tokens_seen": 10058512,
      "step": 15265
    },
    {
      "epoch": 9.003537735849056,
      "grad_norm": 2.4234061241149902,
      "learning_rate": 6.707682929892513e-06,
      "loss": 0.3634,
      "num_input_tokens_seen": 10061808,
      "step": 15270
    },
    {
      "epoch": 9.006485849056604,
      "grad_norm": 1.3979854583740234,
      "learning_rate": 6.705264691040266e-06,
      "loss": 0.3887,
      "num_input_tokens_seen": 10064784,
      "step": 15275
    },
    {
      "epoch": 9.00943396226415,
      "grad_norm": 3.2424585819244385,
      "learning_rate": 6.7028460007118435e-06,
      "loss": 0.2358,
      "num_input_tokens_seen": 10068464,
      "step": 15280
    },
    {
      "epoch": 9.012382075471699,
      "grad_norm": 1.893096923828125,
      "learning_rate": 6.700426859547602e-06,
      "loss": 0.3526,
      "num_input_tokens_seen": 10073520,
      "step": 15285
    },
    {
      "epoch": 9.015330188679245,
      "grad_norm": 2.2241921424865723,
      "learning_rate": 6.6980072681880224e-06,
      "loss": 0.3503,
      "num_input_tokens_seen": 10076432,
      "step": 15290
    },
    {
      "epoch": 9.018278301886792,
      "grad_norm": 2.3221960067749023,
      "learning_rate": 6.695587227273699e-06,
      "loss": 0.4459,
      "num_input_tokens_seen": 10079408,
      "step": 15295
    },
    {
      "epoch": 9.02122641509434,
      "grad_norm": 1.2379316091537476,
      "learning_rate": 6.69316673744535e-06,
      "loss": 0.3703,
      "num_input_tokens_seen": 10082384,
      "step": 15300
    },
    {
      "epoch": 9.024174528301886,
      "grad_norm": 3.2070648670196533,
      "learning_rate": 6.6907457993438115e-06,
      "loss": 0.3479,
      "num_input_tokens_seen": 10085296,
      "step": 15305
    },
    {
      "epoch": 9.027122641509434,
      "grad_norm": 2.056140661239624,
      "learning_rate": 6.688324413610036e-06,
      "loss": 0.2981,
      "num_input_tokens_seen": 10088656,
      "step": 15310
    },
    {
      "epoch": 9.03007075471698,
      "grad_norm": 3.822244882583618,
      "learning_rate": 6.685902580885094e-06,
      "loss": 0.5301,
      "num_input_tokens_seen": 10092080,
      "step": 15315
    },
    {
      "epoch": 9.033018867924529,
      "grad_norm": 1.4569116830825806,
      "learning_rate": 6.6834803018101794e-06,
      "loss": 0.2865,
      "num_input_tokens_seen": 10096400,
      "step": 15320
    },
    {
      "epoch": 9.035966981132075,
      "grad_norm": 2.1244351863861084,
      "learning_rate": 6.681057577026599e-06,
      "loss": 0.2344,
      "num_input_tokens_seen": 10100688,
      "step": 15325
    },
    {
      "epoch": 9.038915094339623,
      "grad_norm": 1.4261304140090942,
      "learning_rate": 6.6786344071757795e-06,
      "loss": 0.3288,
      "num_input_tokens_seen": 10104208,
      "step": 15330
    },
    {
      "epoch": 9.04186320754717,
      "grad_norm": 1.599448800086975,
      "learning_rate": 6.676210792899267e-06,
      "loss": 0.3332,
      "num_input_tokens_seen": 10107504,
      "step": 15335
    },
    {
      "epoch": 9.044811320754716,
      "grad_norm": 3.7061350345611572,
      "learning_rate": 6.6737867348387235e-06,
      "loss": 0.3871,
      "num_input_tokens_seen": 10111952,
      "step": 15340
    },
    {
      "epoch": 9.047759433962264,
      "grad_norm": 2.3308610916137695,
      "learning_rate": 6.671362233635926e-06,
      "loss": 0.3008,
      "num_input_tokens_seen": 10115280,
      "step": 15345
    },
    {
      "epoch": 9.05070754716981,
      "grad_norm": 1.6614205837249756,
      "learning_rate": 6.668937289932775e-06,
      "loss": 0.3436,
      "num_input_tokens_seen": 10118224,
      "step": 15350
    },
    {
      "epoch": 9.053655660377359,
      "grad_norm": 3.224592447280884,
      "learning_rate": 6.666511904371285e-06,
      "loss": 0.3723,
      "num_input_tokens_seen": 10121648,
      "step": 15355
    },
    {
      "epoch": 9.056603773584905,
      "grad_norm": 3.1975808143615723,
      "learning_rate": 6.664086077593586e-06,
      "loss": 0.3351,
      "num_input_tokens_seen": 10124336,
      "step": 15360
    },
    {
      "epoch": 9.059551886792454,
      "grad_norm": 2.04424786567688,
      "learning_rate": 6.661659810241924e-06,
      "loss": 0.3092,
      "num_input_tokens_seen": 10127600,
      "step": 15365
    },
    {
      "epoch": 9.0625,
      "grad_norm": 3.5930447578430176,
      "learning_rate": 6.659233102958667e-06,
      "loss": 0.4212,
      "num_input_tokens_seen": 10130288,
      "step": 15370
    },
    {
      "epoch": 9.065448113207546,
      "grad_norm": 2.294623613357544,
      "learning_rate": 6.6568059563862965e-06,
      "loss": 0.4061,
      "num_input_tokens_seen": 10133808,
      "step": 15375
    },
    {
      "epoch": 9.068396226415095,
      "grad_norm": 3.0196073055267334,
      "learning_rate": 6.654378371167407e-06,
      "loss": 0.5556,
      "num_input_tokens_seen": 10136208,
      "step": 15380
    },
    {
      "epoch": 9.071344339622641,
      "grad_norm": 1.8528488874435425,
      "learning_rate": 6.651950347944715e-06,
      "loss": 0.3422,
      "num_input_tokens_seen": 10139216,
      "step": 15385
    },
    {
      "epoch": 9.07429245283019,
      "grad_norm": 2.864394187927246,
      "learning_rate": 6.649521887361048e-06,
      "loss": 0.4554,
      "num_input_tokens_seen": 10142256,
      "step": 15390
    },
    {
      "epoch": 9.077240566037736,
      "grad_norm": 1.8416887521743774,
      "learning_rate": 6.647092990059352e-06,
      "loss": 0.3383,
      "num_input_tokens_seen": 10145648,
      "step": 15395
    },
    {
      "epoch": 9.080188679245284,
      "grad_norm": 1.87056565284729,
      "learning_rate": 6.644663656682689e-06,
      "loss": 0.3647,
      "num_input_tokens_seen": 10148272,
      "step": 15400
    },
    {
      "epoch": 9.08313679245283,
      "grad_norm": 2.719341278076172,
      "learning_rate": 6.642233887874234e-06,
      "loss": 0.2996,
      "num_input_tokens_seen": 10151088,
      "step": 15405
    },
    {
      "epoch": 9.086084905660377,
      "grad_norm": 4.330259799957275,
      "learning_rate": 6.639803684277279e-06,
      "loss": 0.5501,
      "num_input_tokens_seen": 10154384,
      "step": 15410
    },
    {
      "epoch": 9.089033018867925,
      "grad_norm": 2.218951463699341,
      "learning_rate": 6.637373046535233e-06,
      "loss": 0.4086,
      "num_input_tokens_seen": 10157904,
      "step": 15415
    },
    {
      "epoch": 9.091981132075471,
      "grad_norm": 2.5910251140594482,
      "learning_rate": 6.634941975291617e-06,
      "loss": 0.5231,
      "num_input_tokens_seen": 10160784,
      "step": 15420
    },
    {
      "epoch": 9.09492924528302,
      "grad_norm": 1.758376955986023,
      "learning_rate": 6.632510471190065e-06,
      "loss": 0.3914,
      "num_input_tokens_seen": 10164400,
      "step": 15425
    },
    {
      "epoch": 9.097877358490566,
      "grad_norm": 2.729684829711914,
      "learning_rate": 6.630078534874332e-06,
      "loss": 0.3354,
      "num_input_tokens_seen": 10168080,
      "step": 15430
    },
    {
      "epoch": 9.100825471698114,
      "grad_norm": 1.6400256156921387,
      "learning_rate": 6.62764616698828e-06,
      "loss": 0.4626,
      "num_input_tokens_seen": 10171440,
      "step": 15435
    },
    {
      "epoch": 9.10377358490566,
      "grad_norm": 2.4344494342803955,
      "learning_rate": 6.625213368175889e-06,
      "loss": 0.3855,
      "num_input_tokens_seen": 10174736,
      "step": 15440
    },
    {
      "epoch": 9.106721698113208,
      "grad_norm": 1.8013712167739868,
      "learning_rate": 6.622780139081256e-06,
      "loss": 0.3896,
      "num_input_tokens_seen": 10177968,
      "step": 15445
    },
    {
      "epoch": 9.109669811320755,
      "grad_norm": 2.40651273727417,
      "learning_rate": 6.620346480348589e-06,
      "loss": 0.308,
      "num_input_tokens_seen": 10182096,
      "step": 15450
    },
    {
      "epoch": 9.112617924528301,
      "grad_norm": 3.144134759902954,
      "learning_rate": 6.617912392622206e-06,
      "loss": 0.4058,
      "num_input_tokens_seen": 10184976,
      "step": 15455
    },
    {
      "epoch": 9.11556603773585,
      "grad_norm": 3.9891562461853027,
      "learning_rate": 6.615477876546544e-06,
      "loss": 0.3297,
      "num_input_tokens_seen": 10189232,
      "step": 15460
    },
    {
      "epoch": 9.118514150943396,
      "grad_norm": 1.870292067527771,
      "learning_rate": 6.6130429327661535e-06,
      "loss": 0.3577,
      "num_input_tokens_seen": 10193872,
      "step": 15465
    },
    {
      "epoch": 9.121462264150944,
      "grad_norm": 4.294406414031982,
      "learning_rate": 6.610607561925694e-06,
      "loss": 0.5834,
      "num_input_tokens_seen": 10197360,
      "step": 15470
    },
    {
      "epoch": 9.12441037735849,
      "grad_norm": 2.5288782119750977,
      "learning_rate": 6.608171764669941e-06,
      "loss": 0.3124,
      "num_input_tokens_seen": 10200496,
      "step": 15475
    },
    {
      "epoch": 9.127358490566039,
      "grad_norm": 3.1440799236297607,
      "learning_rate": 6.605735541643783e-06,
      "loss": 0.4438,
      "num_input_tokens_seen": 10204784,
      "step": 15480
    },
    {
      "epoch": 9.130306603773585,
      "grad_norm": 2.8419699668884277,
      "learning_rate": 6.603298893492219e-06,
      "loss": 0.2676,
      "num_input_tokens_seen": 10207920,
      "step": 15485
    },
    {
      "epoch": 9.133254716981131,
      "grad_norm": 2.9024906158447266,
      "learning_rate": 6.6008618208603625e-06,
      "loss": 0.542,
      "num_input_tokens_seen": 10214032,
      "step": 15490
    },
    {
      "epoch": 9.13620283018868,
      "grad_norm": 2.1501247882843018,
      "learning_rate": 6.598424324393442e-06,
      "loss": 0.4325,
      "num_input_tokens_seen": 10217136,
      "step": 15495
    },
    {
      "epoch": 9.139150943396226,
      "grad_norm": 2.303682565689087,
      "learning_rate": 6.595986404736792e-06,
      "loss": 0.4116,
      "num_input_tokens_seen": 10219728,
      "step": 15500
    },
    {
      "epoch": 9.142099056603774,
      "grad_norm": 2.0934431552886963,
      "learning_rate": 6.5935480625358615e-06,
      "loss": 0.3537,
      "num_input_tokens_seen": 10222864,
      "step": 15505
    },
    {
      "epoch": 9.14504716981132,
      "grad_norm": 2.565025568008423,
      "learning_rate": 6.591109298436216e-06,
      "loss": 0.377,
      "num_input_tokens_seen": 10226160,
      "step": 15510
    },
    {
      "epoch": 9.147995283018869,
      "grad_norm": 2.4246695041656494,
      "learning_rate": 6.5886701130835255e-06,
      "loss": 0.3585,
      "num_input_tokens_seen": 10229296,
      "step": 15515
    },
    {
      "epoch": 9.150943396226415,
      "grad_norm": 2.3742833137512207,
      "learning_rate": 6.586230507123574e-06,
      "loss": 0.6145,
      "num_input_tokens_seen": 10232496,
      "step": 15520
    },
    {
      "epoch": 9.153891509433961,
      "grad_norm": 1.9642466306686401,
      "learning_rate": 6.583790481202261e-06,
      "loss": 0.5124,
      "num_input_tokens_seen": 10236176,
      "step": 15525
    },
    {
      "epoch": 9.15683962264151,
      "grad_norm": 2.1083314418792725,
      "learning_rate": 6.5813500359655925e-06,
      "loss": 0.3845,
      "num_input_tokens_seen": 10240976,
      "step": 15530
    },
    {
      "epoch": 9.159787735849056,
      "grad_norm": 5.172887325286865,
      "learning_rate": 6.578909172059687e-06,
      "loss": 0.4629,
      "num_input_tokens_seen": 10243760,
      "step": 15535
    },
    {
      "epoch": 9.162735849056604,
      "grad_norm": 1.6602531671524048,
      "learning_rate": 6.576467890130772e-06,
      "loss": 0.3208,
      "num_input_tokens_seen": 10249488,
      "step": 15540
    },
    {
      "epoch": 9.16568396226415,
      "grad_norm": 2.382962226867676,
      "learning_rate": 6.574026190825191e-06,
      "loss": 0.3414,
      "num_input_tokens_seen": 10253360,
      "step": 15545
    },
    {
      "epoch": 9.168632075471699,
      "grad_norm": 2.620737075805664,
      "learning_rate": 6.57158407478939e-06,
      "loss": 0.4506,
      "num_input_tokens_seen": 10256592,
      "step": 15550
    },
    {
      "epoch": 9.171580188679245,
      "grad_norm": 2.0388917922973633,
      "learning_rate": 6.569141542669935e-06,
      "loss": 0.3927,
      "num_input_tokens_seen": 10260272,
      "step": 15555
    },
    {
      "epoch": 9.174528301886792,
      "grad_norm": 2.737311363220215,
      "learning_rate": 6.566698595113492e-06,
      "loss": 0.4328,
      "num_input_tokens_seen": 10264464,
      "step": 15560
    },
    {
      "epoch": 9.17747641509434,
      "grad_norm": 1.7604910135269165,
      "learning_rate": 6.564255232766843e-06,
      "loss": 0.3105,
      "num_input_tokens_seen": 10267792,
      "step": 15565
    },
    {
      "epoch": 9.180424528301886,
      "grad_norm": 1.864619255065918,
      "learning_rate": 6.561811456276881e-06,
      "loss": 0.3905,
      "num_input_tokens_seen": 10270224,
      "step": 15570
    },
    {
      "epoch": 9.183372641509434,
      "grad_norm": 3.3131184577941895,
      "learning_rate": 6.559367266290605e-06,
      "loss": 0.4422,
      "num_input_tokens_seen": 10272688,
      "step": 15575
    },
    {
      "epoch": 9.18632075471698,
      "grad_norm": 2.789407968521118,
      "learning_rate": 6.556922663455123e-06,
      "loss": 0.3231,
      "num_input_tokens_seen": 10274928,
      "step": 15580
    },
    {
      "epoch": 9.189268867924529,
      "grad_norm": 3.648817777633667,
      "learning_rate": 6.554477648417657e-06,
      "loss": 0.512,
      "num_input_tokens_seen": 10278032,
      "step": 15585
    },
    {
      "epoch": 9.192216981132075,
      "grad_norm": 2.4818437099456787,
      "learning_rate": 6.552032221825535e-06,
      "loss": 0.4183,
      "num_input_tokens_seen": 10281392,
      "step": 15590
    },
    {
      "epoch": 9.195165094339623,
      "grad_norm": 2.6221120357513428,
      "learning_rate": 6.549586384326192e-06,
      "loss": 0.4397,
      "num_input_tokens_seen": 10285360,
      "step": 15595
    },
    {
      "epoch": 9.19811320754717,
      "grad_norm": 2.9214026927948,
      "learning_rate": 6.547140136567176e-06,
      "loss": 0.6144,
      "num_input_tokens_seen": 10288240,
      "step": 15600
    },
    {
      "epoch": 9.201061320754716,
      "grad_norm": 3.5428249835968018,
      "learning_rate": 6.544693479196142e-06,
      "loss": 0.4217,
      "num_input_tokens_seen": 10291920,
      "step": 15605
    },
    {
      "epoch": 9.204009433962264,
      "grad_norm": 1.8277359008789062,
      "learning_rate": 6.542246412860851e-06,
      "loss": 0.3631,
      "num_input_tokens_seen": 10295504,
      "step": 15610
    },
    {
      "epoch": 9.20695754716981,
      "grad_norm": 4.420856952667236,
      "learning_rate": 6.5397989382091754e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 10298544,
      "step": 15615
    },
    {
      "epoch": 9.209905660377359,
      "grad_norm": 2.4659056663513184,
      "learning_rate": 6.537351055889096e-06,
      "loss": 0.3868,
      "num_input_tokens_seen": 10302416,
      "step": 15620
    },
    {
      "epoch": 9.212853773584905,
      "grad_norm": 2.4215588569641113,
      "learning_rate": 6.534902766548698e-06,
      "loss": 0.2747,
      "num_input_tokens_seen": 10305872,
      "step": 15625
    },
    {
      "epoch": 9.215801886792454,
      "grad_norm": 4.575099945068359,
      "learning_rate": 6.532454070836176e-06,
      "loss": 0.3127,
      "num_input_tokens_seen": 10308784,
      "step": 15630
    },
    {
      "epoch": 9.21875,
      "grad_norm": 4.516048431396484,
      "learning_rate": 6.530004969399836e-06,
      "loss": 0.3404,
      "num_input_tokens_seen": 10311568,
      "step": 15635
    },
    {
      "epoch": 9.221698113207546,
      "grad_norm": 3.2812037467956543,
      "learning_rate": 6.5275554628880865e-06,
      "loss": 0.4196,
      "num_input_tokens_seen": 10314000,
      "step": 15640
    },
    {
      "epoch": 9.224646226415095,
      "grad_norm": 3.099055290222168,
      "learning_rate": 6.525105551949444e-06,
      "loss": 0.4763,
      "num_input_tokens_seen": 10316880,
      "step": 15645
    },
    {
      "epoch": 9.227594339622641,
      "grad_norm": 1.801507592201233,
      "learning_rate": 6.522655237232535e-06,
      "loss": 0.331,
      "num_input_tokens_seen": 10319856,
      "step": 15650
    },
    {
      "epoch": 9.23054245283019,
      "grad_norm": 1.1905454397201538,
      "learning_rate": 6.5202045193860885e-06,
      "loss": 0.2695,
      "num_input_tokens_seen": 10323184,
      "step": 15655
    },
    {
      "epoch": 9.233490566037736,
      "grad_norm": 3.2016761302948,
      "learning_rate": 6.517753399058944e-06,
      "loss": 0.3704,
      "num_input_tokens_seen": 10325744,
      "step": 15660
    },
    {
      "epoch": 9.236438679245284,
      "grad_norm": 1.6634570360183716,
      "learning_rate": 6.515301876900047e-06,
      "loss": 0.3358,
      "num_input_tokens_seen": 10329264,
      "step": 15665
    },
    {
      "epoch": 9.23938679245283,
      "grad_norm": 3.116687059402466,
      "learning_rate": 6.512849953558445e-06,
      "loss": 0.3235,
      "num_input_tokens_seen": 10332976,
      "step": 15670
    },
    {
      "epoch": 9.242334905660377,
      "grad_norm": 3.364919900894165,
      "learning_rate": 6.510397629683301e-06,
      "loss": 0.4397,
      "num_input_tokens_seen": 10335440,
      "step": 15675
    },
    {
      "epoch": 9.245283018867925,
      "grad_norm": 2.4848380088806152,
      "learning_rate": 6.507944905923872e-06,
      "loss": 0.3979,
      "num_input_tokens_seen": 10337840,
      "step": 15680
    },
    {
      "epoch": 9.248231132075471,
      "grad_norm": 2.57112717628479,
      "learning_rate": 6.505491782929531e-06,
      "loss": 0.3284,
      "num_input_tokens_seen": 10340464,
      "step": 15685
    },
    {
      "epoch": 9.25117924528302,
      "grad_norm": 3.4877071380615234,
      "learning_rate": 6.5030382613497535e-06,
      "loss": 0.3522,
      "num_input_tokens_seen": 10343568,
      "step": 15690
    },
    {
      "epoch": 9.254127358490566,
      "grad_norm": 3.749790906906128,
      "learning_rate": 6.500584341834119e-06,
      "loss": 0.3534,
      "num_input_tokens_seen": 10346480,
      "step": 15695
    },
    {
      "epoch": 9.257075471698114,
      "grad_norm": 1.7349170446395874,
      "learning_rate": 6.498130025032312e-06,
      "loss": 0.3925,
      "num_input_tokens_seen": 10349264,
      "step": 15700
    },
    {
      "epoch": 9.26002358490566,
      "grad_norm": 3.696713924407959,
      "learning_rate": 6.495675311594123e-06,
      "loss": 0.3727,
      "num_input_tokens_seen": 10352688,
      "step": 15705
    },
    {
      "epoch": 9.262971698113208,
      "grad_norm": 2.624979019165039,
      "learning_rate": 6.493220202169452e-06,
      "loss": 0.3559,
      "num_input_tokens_seen": 10356048,
      "step": 15710
    },
    {
      "epoch": 9.265919811320755,
      "grad_norm": 2.2379777431488037,
      "learning_rate": 6.490764697408295e-06,
      "loss": 0.3496,
      "num_input_tokens_seen": 10358800,
      "step": 15715
    },
    {
      "epoch": 9.268867924528301,
      "grad_norm": 2.226835012435913,
      "learning_rate": 6.48830879796076e-06,
      "loss": 0.4063,
      "num_input_tokens_seen": 10362096,
      "step": 15720
    },
    {
      "epoch": 9.27181603773585,
      "grad_norm": 2.1983163356781006,
      "learning_rate": 6.4858525044770546e-06,
      "loss": 0.3954,
      "num_input_tokens_seen": 10364688,
      "step": 15725
    },
    {
      "epoch": 9.274764150943396,
      "grad_norm": 2.3438870906829834,
      "learning_rate": 6.483395817607497e-06,
      "loss": 0.2945,
      "num_input_tokens_seen": 10368368,
      "step": 15730
    },
    {
      "epoch": 9.277712264150944,
      "grad_norm": 1.7183570861816406,
      "learning_rate": 6.480938738002504e-06,
      "loss": 0.4554,
      "num_input_tokens_seen": 10371568,
      "step": 15735
    },
    {
      "epoch": 9.28066037735849,
      "grad_norm": 2.629835605621338,
      "learning_rate": 6.478481266312597e-06,
      "loss": 0.4589,
      "num_input_tokens_seen": 10374608,
      "step": 15740
    },
    {
      "epoch": 9.283608490566039,
      "grad_norm": 2.281510591506958,
      "learning_rate": 6.476023403188403e-06,
      "loss": 0.3308,
      "num_input_tokens_seen": 10377360,
      "step": 15745
    },
    {
      "epoch": 9.286556603773585,
      "grad_norm": 3.8611767292022705,
      "learning_rate": 6.473565149280651e-06,
      "loss": 0.3199,
      "num_input_tokens_seen": 10380592,
      "step": 15750
    },
    {
      "epoch": 9.289504716981131,
      "grad_norm": 3.0594322681427,
      "learning_rate": 6.471106505240175e-06,
      "loss": 0.437,
      "num_input_tokens_seen": 10383920,
      "step": 15755
    },
    {
      "epoch": 9.29245283018868,
      "grad_norm": 1.9772053956985474,
      "learning_rate": 6.468647471717914e-06,
      "loss": 0.4269,
      "num_input_tokens_seen": 10386544,
      "step": 15760
    },
    {
      "epoch": 9.295400943396226,
      "grad_norm": 1.8059364557266235,
      "learning_rate": 6.466188049364902e-06,
      "loss": 0.4011,
      "num_input_tokens_seen": 10389808,
      "step": 15765
    },
    {
      "epoch": 9.298349056603774,
      "grad_norm": 1.816598892211914,
      "learning_rate": 6.463728238832288e-06,
      "loss": 0.3676,
      "num_input_tokens_seen": 10393456,
      "step": 15770
    },
    {
      "epoch": 9.30129716981132,
      "grad_norm": 2.0623247623443604,
      "learning_rate": 6.461268040771311e-06,
      "loss": 0.3108,
      "num_input_tokens_seen": 10396528,
      "step": 15775
    },
    {
      "epoch": 9.304245283018869,
      "grad_norm": 4.234267711639404,
      "learning_rate": 6.458807455833326e-06,
      "loss": 0.4034,
      "num_input_tokens_seen": 10399760,
      "step": 15780
    },
    {
      "epoch": 9.307193396226415,
      "grad_norm": 2.0294108390808105,
      "learning_rate": 6.456346484669778e-06,
      "loss": 0.3829,
      "num_input_tokens_seen": 10402768,
      "step": 15785
    },
    {
      "epoch": 9.310141509433961,
      "grad_norm": 3.2737181186676025,
      "learning_rate": 6.4538851279322225e-06,
      "loss": 0.5584,
      "num_input_tokens_seen": 10409360,
      "step": 15790
    },
    {
      "epoch": 9.31308962264151,
      "grad_norm": 2.7960312366485596,
      "learning_rate": 6.451423386272312e-06,
      "loss": 0.389,
      "num_input_tokens_seen": 10413680,
      "step": 15795
    },
    {
      "epoch": 9.316037735849056,
      "grad_norm": 4.625767707824707,
      "learning_rate": 6.448961260341806e-06,
      "loss": 0.3796,
      "num_input_tokens_seen": 10416208,
      "step": 15800
    },
    {
      "epoch": 9.318985849056604,
      "grad_norm": 8.183886528015137,
      "learning_rate": 6.446498750792563e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 10419888,
      "step": 15805
    },
    {
      "epoch": 9.32193396226415,
      "grad_norm": 2.6309826374053955,
      "learning_rate": 6.444035858276538e-06,
      "loss": 0.4398,
      "num_input_tokens_seen": 10423248,
      "step": 15810
    },
    {
      "epoch": 9.324882075471699,
      "grad_norm": 2.4407408237457275,
      "learning_rate": 6.441572583445799e-06,
      "loss": 0.3836,
      "num_input_tokens_seen": 10426224,
      "step": 15815
    },
    {
      "epoch": 9.327830188679245,
      "grad_norm": 1.789969563484192,
      "learning_rate": 6.439108926952504e-06,
      "loss": 0.3539,
      "num_input_tokens_seen": 10430032,
      "step": 15820
    },
    {
      "epoch": 9.330778301886792,
      "grad_norm": 1.9494305849075317,
      "learning_rate": 6.43664488944892e-06,
      "loss": 0.5264,
      "num_input_tokens_seen": 10433968,
      "step": 15825
    },
    {
      "epoch": 9.33372641509434,
      "grad_norm": 2.6973917484283447,
      "learning_rate": 6.434180471587409e-06,
      "loss": 0.3509,
      "num_input_tokens_seen": 10436720,
      "step": 15830
    },
    {
      "epoch": 9.336674528301886,
      "grad_norm": 2.756563663482666,
      "learning_rate": 6.431715674020438e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 10439920,
      "step": 15835
    },
    {
      "epoch": 9.339622641509434,
      "grad_norm": 4.871086120605469,
      "learning_rate": 6.429250497400571e-06,
      "loss": 0.3398,
      "num_input_tokens_seen": 10442800,
      "step": 15840
    },
    {
      "epoch": 9.34257075471698,
      "grad_norm": 2.063044548034668,
      "learning_rate": 6.426784942380475e-06,
      "loss": 0.4117,
      "num_input_tokens_seen": 10447216,
      "step": 15845
    },
    {
      "epoch": 9.345518867924529,
      "grad_norm": 2.764214277267456,
      "learning_rate": 6.424319009612917e-06,
      "loss": 0.3732,
      "num_input_tokens_seen": 10449904,
      "step": 15850
    },
    {
      "epoch": 9.348466981132075,
      "grad_norm": 2.880627393722534,
      "learning_rate": 6.421852699750763e-06,
      "loss": 0.463,
      "num_input_tokens_seen": 10452784,
      "step": 15855
    },
    {
      "epoch": 9.351415094339623,
      "grad_norm": 2.7520925998687744,
      "learning_rate": 6.4193860134469775e-06,
      "loss": 0.3649,
      "num_input_tokens_seen": 10455888,
      "step": 15860
    },
    {
      "epoch": 9.35436320754717,
      "grad_norm": 1.9379150867462158,
      "learning_rate": 6.416918951354629e-06,
      "loss": 0.3342,
      "num_input_tokens_seen": 10458992,
      "step": 15865
    },
    {
      "epoch": 9.357311320754716,
      "grad_norm": 2.8721039295196533,
      "learning_rate": 6.4144515141268816e-06,
      "loss": 0.3881,
      "num_input_tokens_seen": 10461712,
      "step": 15870
    },
    {
      "epoch": 9.360259433962264,
      "grad_norm": 4.322682857513428,
      "learning_rate": 6.411983702416999e-06,
      "loss": 0.3327,
      "num_input_tokens_seen": 10464496,
      "step": 15875
    },
    {
      "epoch": 9.36320754716981,
      "grad_norm": 2.0548036098480225,
      "learning_rate": 6.409515516878346e-06,
      "loss": 0.3557,
      "num_input_tokens_seen": 10467152,
      "step": 15880
    },
    {
      "epoch": 9.366155660377359,
      "grad_norm": 4.411008834838867,
      "learning_rate": 6.407046958164387e-06,
      "loss": 0.3156,
      "num_input_tokens_seen": 10470032,
      "step": 15885
    },
    {
      "epoch": 9.369103773584905,
      "grad_norm": 4.833184719085693,
      "learning_rate": 6.404578026928679e-06,
      "loss": 0.4149,
      "num_input_tokens_seen": 10472720,
      "step": 15890
    },
    {
      "epoch": 9.372051886792454,
      "grad_norm": 2.6710588932037354,
      "learning_rate": 6.402108723824887e-06,
      "loss": 0.3023,
      "num_input_tokens_seen": 10475952,
      "step": 15895
    },
    {
      "epoch": 9.375,
      "grad_norm": 3.339677095413208,
      "learning_rate": 6.399639049506767e-06,
      "loss": 0.3633,
      "num_input_tokens_seen": 10479376,
      "step": 15900
    },
    {
      "epoch": 9.377948113207546,
      "grad_norm": 3.013512134552002,
      "learning_rate": 6.397169004628177e-06,
      "loss": 0.43,
      "num_input_tokens_seen": 10483280,
      "step": 15905
    },
    {
      "epoch": 9.380896226415095,
      "grad_norm": 3.869028329849243,
      "learning_rate": 6.39469858984307e-06,
      "loss": 0.3532,
      "num_input_tokens_seen": 10486256,
      "step": 15910
    },
    {
      "epoch": 9.383844339622641,
      "grad_norm": 2.5533556938171387,
      "learning_rate": 6.3922278058055024e-06,
      "loss": 0.3761,
      "num_input_tokens_seen": 10489072,
      "step": 15915
    },
    {
      "epoch": 9.38679245283019,
      "grad_norm": 2.053295373916626,
      "learning_rate": 6.389756653169622e-06,
      "loss": 0.3796,
      "num_input_tokens_seen": 10491408,
      "step": 15920
    },
    {
      "epoch": 9.389740566037736,
      "grad_norm": 3.159531593322754,
      "learning_rate": 6.387285132589678e-06,
      "loss": 0.3763,
      "num_input_tokens_seen": 10494576,
      "step": 15925
    },
    {
      "epoch": 9.392688679245284,
      "grad_norm": 3.0816400051116943,
      "learning_rate": 6.3848132447200166e-06,
      "loss": 0.4519,
      "num_input_tokens_seen": 10497392,
      "step": 15930
    },
    {
      "epoch": 9.39563679245283,
      "grad_norm": 2.549391269683838,
      "learning_rate": 6.38234099021508e-06,
      "loss": 0.3953,
      "num_input_tokens_seen": 10500720,
      "step": 15935
    },
    {
      "epoch": 9.398584905660377,
      "grad_norm": 2.2547452449798584,
      "learning_rate": 6.379868369729409e-06,
      "loss": 0.5015,
      "num_input_tokens_seen": 10503856,
      "step": 15940
    },
    {
      "epoch": 9.401533018867925,
      "grad_norm": 5.8300909996032715,
      "learning_rate": 6.377395383917642e-06,
      "loss": 0.4907,
      "num_input_tokens_seen": 10507728,
      "step": 15945
    },
    {
      "epoch": 9.404481132075471,
      "grad_norm": 3.862147092819214,
      "learning_rate": 6.374922033434507e-06,
      "loss": 0.4059,
      "num_input_tokens_seen": 10510480,
      "step": 15950
    },
    {
      "epoch": 9.40742924528302,
      "grad_norm": 1.5028403997421265,
      "learning_rate": 6.372448318934842e-06,
      "loss": 0.4514,
      "num_input_tokens_seen": 10514128,
      "step": 15955
    },
    {
      "epoch": 9.410377358490566,
      "grad_norm": 2.6461141109466553,
      "learning_rate": 6.369974241073569e-06,
      "loss": 0.4656,
      "num_input_tokens_seen": 10518000,
      "step": 15960
    },
    {
      "epoch": 9.413325471698114,
      "grad_norm": 0.3764435052871704,
      "learning_rate": 6.367499800505709e-06,
      "loss": 0.3952,
      "num_input_tokens_seen": 10523728,
      "step": 15965
    },
    {
      "epoch": 9.41627358490566,
      "grad_norm": 2.31819486618042,
      "learning_rate": 6.365024997886384e-06,
      "loss": 0.447,
      "num_input_tokens_seen": 10527088,
      "step": 15970
    },
    {
      "epoch": 9.419221698113208,
      "grad_norm": 1.7464253902435303,
      "learning_rate": 6.362549833870808e-06,
      "loss": 0.3378,
      "num_input_tokens_seen": 10530768,
      "step": 15975
    },
    {
      "epoch": 9.422169811320755,
      "grad_norm": 1.6945459842681885,
      "learning_rate": 6.360074309114293e-06,
      "loss": 0.3106,
      "num_input_tokens_seen": 10533840,
      "step": 15980
    },
    {
      "epoch": 9.425117924528301,
      "grad_norm": 3.573486328125,
      "learning_rate": 6.357598424272241e-06,
      "loss": 0.3863,
      "num_input_tokens_seen": 10537264,
      "step": 15985
    },
    {
      "epoch": 9.42806603773585,
      "grad_norm": 2.953733444213867,
      "learning_rate": 6.355122180000156e-06,
      "loss": 0.4871,
      "num_input_tokens_seen": 10540624,
      "step": 15990
    },
    {
      "epoch": 9.431014150943396,
      "grad_norm": 3.915431261062622,
      "learning_rate": 6.352645576953635e-06,
      "loss": 0.3189,
      "num_input_tokens_seen": 10543280,
      "step": 15995
    },
    {
      "epoch": 9.433962264150944,
      "grad_norm": 4.202461242675781,
      "learning_rate": 6.350168615788366e-06,
      "loss": 0.412,
      "num_input_tokens_seen": 10546928,
      "step": 16000
    },
    {
      "epoch": 9.43691037735849,
      "grad_norm": 2.3377413749694824,
      "learning_rate": 6.347691297160137e-06,
      "loss": 0.3631,
      "num_input_tokens_seen": 10549424,
      "step": 16005
    },
    {
      "epoch": 9.439858490566039,
      "grad_norm": 3.7439870834350586,
      "learning_rate": 6.34521362172483e-06,
      "loss": 0.4464,
      "num_input_tokens_seen": 10553040,
      "step": 16010
    },
    {
      "epoch": 9.442806603773585,
      "grad_norm": 1.8553555011749268,
      "learning_rate": 6.342735590138417e-06,
      "loss": 0.3019,
      "num_input_tokens_seen": 10556240,
      "step": 16015
    },
    {
      "epoch": 9.445754716981131,
      "grad_norm": 3.1455471515655518,
      "learning_rate": 6.340257203056972e-06,
      "loss": 0.5718,
      "num_input_tokens_seen": 10559440,
      "step": 16020
    },
    {
      "epoch": 9.44870283018868,
      "grad_norm": 2.1478323936462402,
      "learning_rate": 6.3377784611366554e-06,
      "loss": 0.3251,
      "num_input_tokens_seen": 10563632,
      "step": 16025
    },
    {
      "epoch": 9.451650943396226,
      "grad_norm": 4.078900337219238,
      "learning_rate": 6.335299365033726e-06,
      "loss": 0.441,
      "num_input_tokens_seen": 10566256,
      "step": 16030
    },
    {
      "epoch": 9.454599056603774,
      "grad_norm": 3.6336936950683594,
      "learning_rate": 6.3328199154045346e-06,
      "loss": 0.4296,
      "num_input_tokens_seen": 10569232,
      "step": 16035
    },
    {
      "epoch": 9.45754716981132,
      "grad_norm": 1.7404676675796509,
      "learning_rate": 6.3303401129055265e-06,
      "loss": 0.3256,
      "num_input_tokens_seen": 10572368,
      "step": 16040
    },
    {
      "epoch": 9.460495283018869,
      "grad_norm": 8.509567260742188,
      "learning_rate": 6.32785995819324e-06,
      "loss": 0.39,
      "num_input_tokens_seen": 10575888,
      "step": 16045
    },
    {
      "epoch": 9.463443396226415,
      "grad_norm": 1.509987473487854,
      "learning_rate": 6.3253794519243075e-06,
      "loss": 0.3999,
      "num_input_tokens_seen": 10580592,
      "step": 16050
    },
    {
      "epoch": 9.466391509433961,
      "grad_norm": 3.6679422855377197,
      "learning_rate": 6.322898594755452e-06,
      "loss": 0.3657,
      "num_input_tokens_seen": 10583696,
      "step": 16055
    },
    {
      "epoch": 9.46933962264151,
      "grad_norm": 5.714085578918457,
      "learning_rate": 6.320417387343492e-06,
      "loss": 0.5301,
      "num_input_tokens_seen": 10587216,
      "step": 16060
    },
    {
      "epoch": 9.472287735849056,
      "grad_norm": 2.061235189437866,
      "learning_rate": 6.3179358303453386e-06,
      "loss": 0.3546,
      "num_input_tokens_seen": 10590320,
      "step": 16065
    },
    {
      "epoch": 9.475235849056604,
      "grad_norm": 2.6333069801330566,
      "learning_rate": 6.315453924417995e-06,
      "loss": 0.5066,
      "num_input_tokens_seen": 10593072,
      "step": 16070
    },
    {
      "epoch": 9.47818396226415,
      "grad_norm": 2.6872401237487793,
      "learning_rate": 6.312971670218554e-06,
      "loss": 0.4722,
      "num_input_tokens_seen": 10596208,
      "step": 16075
    },
    {
      "epoch": 9.481132075471699,
      "grad_norm": 1.8110181093215942,
      "learning_rate": 6.3104890684042055e-06,
      "loss": 0.3566,
      "num_input_tokens_seen": 10600880,
      "step": 16080
    },
    {
      "epoch": 9.484080188679245,
      "grad_norm": 2.6379189491271973,
      "learning_rate": 6.308006119632228e-06,
      "loss": 0.3187,
      "num_input_tokens_seen": 10604528,
      "step": 16085
    },
    {
      "epoch": 9.487028301886792,
      "grad_norm": 1.7770155668258667,
      "learning_rate": 6.305522824559993e-06,
      "loss": 0.4103,
      "num_input_tokens_seen": 10607856,
      "step": 16090
    },
    {
      "epoch": 9.48997641509434,
      "grad_norm": 3.1694509983062744,
      "learning_rate": 6.303039183844965e-06,
      "loss": 0.3644,
      "num_input_tokens_seen": 10611184,
      "step": 16095
    },
    {
      "epoch": 9.492924528301886,
      "grad_norm": 2.877859592437744,
      "learning_rate": 6.300555198144697e-06,
      "loss": 0.3904,
      "num_input_tokens_seen": 10614096,
      "step": 16100
    },
    {
      "epoch": 9.495872641509434,
      "grad_norm": 9.127762794494629,
      "learning_rate": 6.2980708681168335e-06,
      "loss": 0.3193,
      "num_input_tokens_seen": 10617392,
      "step": 16105
    },
    {
      "epoch": 9.49882075471698,
      "grad_norm": 2.739583730697632,
      "learning_rate": 6.2955861944191145e-06,
      "loss": 0.421,
      "num_input_tokens_seen": 10620304,
      "step": 16110
    },
    {
      "epoch": 9.501768867924529,
      "grad_norm": 1.7166441679000854,
      "learning_rate": 6.293101177709367e-06,
      "loss": 0.4589,
      "num_input_tokens_seen": 10624432,
      "step": 16115
    },
    {
      "epoch": 9.504716981132075,
      "grad_norm": 3.336759090423584,
      "learning_rate": 6.290615818645509e-06,
      "loss": 0.5776,
      "num_input_tokens_seen": 10627440,
      "step": 16120
    },
    {
      "epoch": 9.507665094339622,
      "grad_norm": 2.002686023712158,
      "learning_rate": 6.288130117885552e-06,
      "loss": 0.3821,
      "num_input_tokens_seen": 10631024,
      "step": 16125
    },
    {
      "epoch": 9.51061320754717,
      "grad_norm": 2.4905569553375244,
      "learning_rate": 6.285644076087594e-06,
      "loss": 0.3806,
      "num_input_tokens_seen": 10634896,
      "step": 16130
    },
    {
      "epoch": 9.513561320754716,
      "grad_norm": 1.9177695512771606,
      "learning_rate": 6.283157693909826e-06,
      "loss": 0.351,
      "num_input_tokens_seen": 10637232,
      "step": 16135
    },
    {
      "epoch": 9.516509433962264,
      "grad_norm": 3.274437665939331,
      "learning_rate": 6.280670972010528e-06,
      "loss": 0.4315,
      "num_input_tokens_seen": 10640144,
      "step": 16140
    },
    {
      "epoch": 9.51945754716981,
      "grad_norm": 2.238821029663086,
      "learning_rate": 6.278183911048072e-06,
      "loss": 0.3745,
      "num_input_tokens_seen": 10644464,
      "step": 16145
    },
    {
      "epoch": 9.522405660377359,
      "grad_norm": 1.9894393682479858,
      "learning_rate": 6.275696511680915e-06,
      "loss": 0.5696,
      "num_input_tokens_seen": 10647472,
      "step": 16150
    },
    {
      "epoch": 9.525353773584905,
      "grad_norm": 2.6201114654541016,
      "learning_rate": 6.27320877456761e-06,
      "loss": 0.3241,
      "num_input_tokens_seen": 10651184,
      "step": 16155
    },
    {
      "epoch": 9.528301886792454,
      "grad_norm": 2.0454821586608887,
      "learning_rate": 6.270720700366793e-06,
      "loss": 0.3144,
      "num_input_tokens_seen": 10654864,
      "step": 16160
    },
    {
      "epoch": 9.53125,
      "grad_norm": 2.591193675994873,
      "learning_rate": 6.2682322897371974e-06,
      "loss": 0.39,
      "num_input_tokens_seen": 10658064,
      "step": 16165
    },
    {
      "epoch": 9.534198113207546,
      "grad_norm": 3.7814273834228516,
      "learning_rate": 6.265743543337634e-06,
      "loss": 0.3637,
      "num_input_tokens_seen": 10660656,
      "step": 16170
    },
    {
      "epoch": 9.537146226415095,
      "grad_norm": 3.1395020484924316,
      "learning_rate": 6.2632544618270156e-06,
      "loss": 0.3809,
      "num_input_tokens_seen": 10663696,
      "step": 16175
    },
    {
      "epoch": 9.540094339622641,
      "grad_norm": 3.1324687004089355,
      "learning_rate": 6.260765045864332e-06,
      "loss": 0.3462,
      "num_input_tokens_seen": 10666992,
      "step": 16180
    },
    {
      "epoch": 9.54304245283019,
      "grad_norm": 3.10060977935791,
      "learning_rate": 6.258275296108669e-06,
      "loss": 0.3293,
      "num_input_tokens_seen": 10670160,
      "step": 16185
    },
    {
      "epoch": 9.545990566037736,
      "grad_norm": 1.7961713075637817,
      "learning_rate": 6.2557852132191985e-06,
      "loss": 0.468,
      "num_input_tokens_seen": 10673776,
      "step": 16190
    },
    {
      "epoch": 9.548938679245284,
      "grad_norm": 1.9511990547180176,
      "learning_rate": 6.253294797855182e-06,
      "loss": 0.392,
      "num_input_tokens_seen": 10676848,
      "step": 16195
    },
    {
      "epoch": 9.55188679245283,
      "grad_norm": 6.038342475891113,
      "learning_rate": 6.250804050675964e-06,
      "loss": 0.339,
      "num_input_tokens_seen": 10680048,
      "step": 16200
    },
    {
      "epoch": 9.554834905660378,
      "grad_norm": 2.755009651184082,
      "learning_rate": 6.248312972340984e-06,
      "loss": 0.3713,
      "num_input_tokens_seen": 10683056,
      "step": 16205
    },
    {
      "epoch": 9.557783018867925,
      "grad_norm": 1.8278007507324219,
      "learning_rate": 6.2458215635097656e-06,
      "loss": 0.4619,
      "num_input_tokens_seen": 10686096,
      "step": 16210
    },
    {
      "epoch": 9.560731132075471,
      "grad_norm": 1.8676174879074097,
      "learning_rate": 6.2433298248419175e-06,
      "loss": 0.3797,
      "num_input_tokens_seen": 10689104,
      "step": 16215
    },
    {
      "epoch": 9.56367924528302,
      "grad_norm": 1.7773725986480713,
      "learning_rate": 6.2408377569971405e-06,
      "loss": 0.4094,
      "num_input_tokens_seen": 10691952,
      "step": 16220
    },
    {
      "epoch": 9.566627358490566,
      "grad_norm": 2.3066773414611816,
      "learning_rate": 6.238345360635221e-06,
      "loss": 0.4849,
      "num_input_tokens_seen": 10695568,
      "step": 16225
    },
    {
      "epoch": 9.569575471698114,
      "grad_norm": 1.782120704650879,
      "learning_rate": 6.2358526364160274e-06,
      "loss": 0.4056,
      "num_input_tokens_seen": 10698704,
      "step": 16230
    },
    {
      "epoch": 9.57252358490566,
      "grad_norm": 2.2162013053894043,
      "learning_rate": 6.2333595849995245e-06,
      "loss": 0.4211,
      "num_input_tokens_seen": 10701840,
      "step": 16235
    },
    {
      "epoch": 9.575471698113208,
      "grad_norm": 2.9055964946746826,
      "learning_rate": 6.230866207045756e-06,
      "loss": 0.3481,
      "num_input_tokens_seen": 10706256,
      "step": 16240
    },
    {
      "epoch": 9.578419811320755,
      "grad_norm": 3.870896339416504,
      "learning_rate": 6.228372503214853e-06,
      "loss": 0.4383,
      "num_input_tokens_seen": 10708912,
      "step": 16245
    },
    {
      "epoch": 9.581367924528301,
      "grad_norm": 3.3745739459991455,
      "learning_rate": 6.225878474167035e-06,
      "loss": 0.2566,
      "num_input_tokens_seen": 10711504,
      "step": 16250
    },
    {
      "epoch": 9.58431603773585,
      "grad_norm": 2.271312713623047,
      "learning_rate": 6.22338412056261e-06,
      "loss": 0.4221,
      "num_input_tokens_seen": 10714192,
      "step": 16255
    },
    {
      "epoch": 9.587264150943396,
      "grad_norm": 3.155810594558716,
      "learning_rate": 6.220889443061966e-06,
      "loss": 0.4104,
      "num_input_tokens_seen": 10717520,
      "step": 16260
    },
    {
      "epoch": 9.590212264150944,
      "grad_norm": 2.1598165035247803,
      "learning_rate": 6.2183944423255796e-06,
      "loss": 0.3601,
      "num_input_tokens_seen": 10720880,
      "step": 16265
    },
    {
      "epoch": 9.59316037735849,
      "grad_norm": 2.481694459915161,
      "learning_rate": 6.215899119014015e-06,
      "loss": 0.421,
      "num_input_tokens_seen": 10723920,
      "step": 16270
    },
    {
      "epoch": 9.596108490566039,
      "grad_norm": 5.866611003875732,
      "learning_rate": 6.2134034737879175e-06,
      "loss": 0.5537,
      "num_input_tokens_seen": 10726672,
      "step": 16275
    },
    {
      "epoch": 9.599056603773585,
      "grad_norm": 2.6566009521484375,
      "learning_rate": 6.2109075073080205e-06,
      "loss": 0.4608,
      "num_input_tokens_seen": 10730128,
      "step": 16280
    },
    {
      "epoch": 9.602004716981131,
      "grad_norm": 2.552927017211914,
      "learning_rate": 6.2084112202351425e-06,
      "loss": 0.3784,
      "num_input_tokens_seen": 10732528,
      "step": 16285
    },
    {
      "epoch": 9.60495283018868,
      "grad_norm": 5.36952018737793,
      "learning_rate": 6.205914613230186e-06,
      "loss": 0.322,
      "num_input_tokens_seen": 10735568,
      "step": 16290
    },
    {
      "epoch": 9.607900943396226,
      "grad_norm": 1.4249122142791748,
      "learning_rate": 6.203417686954138e-06,
      "loss": 0.2567,
      "num_input_tokens_seen": 10738640,
      "step": 16295
    },
    {
      "epoch": 9.610849056603774,
      "grad_norm": 1.8614333868026733,
      "learning_rate": 6.2009204420680706e-06,
      "loss": 0.3694,
      "num_input_tokens_seen": 10741968,
      "step": 16300
    },
    {
      "epoch": 9.61379716981132,
      "grad_norm": 3.3930423259735107,
      "learning_rate": 6.198422879233141e-06,
      "loss": 0.4177,
      "num_input_tokens_seen": 10744784,
      "step": 16305
    },
    {
      "epoch": 9.616745283018869,
      "grad_norm": 2.899869680404663,
      "learning_rate": 6.1959249991105895e-06,
      "loss": 0.3804,
      "num_input_tokens_seen": 10748304,
      "step": 16310
    },
    {
      "epoch": 9.619693396226415,
      "grad_norm": 2.8398969173431396,
      "learning_rate": 6.19342680236174e-06,
      "loss": 0.518,
      "num_input_tokens_seen": 10752400,
      "step": 16315
    },
    {
      "epoch": 9.622641509433961,
      "grad_norm": 6.5490875244140625,
      "learning_rate": 6.190928289648003e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 10756848,
      "step": 16320
    },
    {
      "epoch": 9.62558962264151,
      "grad_norm": 1.9717347621917725,
      "learning_rate": 6.188429461630866e-06,
      "loss": 0.3584,
      "num_input_tokens_seen": 10761200,
      "step": 16325
    },
    {
      "epoch": 9.628537735849056,
      "grad_norm": 1.6516892910003662,
      "learning_rate": 6.185930318971909e-06,
      "loss": 0.3569,
      "num_input_tokens_seen": 10763632,
      "step": 16330
    },
    {
      "epoch": 9.631485849056604,
      "grad_norm": 2.141080141067505,
      "learning_rate": 6.1834308623327885e-06,
      "loss": 0.3204,
      "num_input_tokens_seen": 10768432,
      "step": 16335
    },
    {
      "epoch": 9.63443396226415,
      "grad_norm": 3.2092254161834717,
      "learning_rate": 6.180931092375247e-06,
      "loss": 0.3629,
      "num_input_tokens_seen": 10771696,
      "step": 16340
    },
    {
      "epoch": 9.637382075471699,
      "grad_norm": 2.1944491863250732,
      "learning_rate": 6.1784310097611075e-06,
      "loss": 0.4084,
      "num_input_tokens_seen": 10774512,
      "step": 16345
    },
    {
      "epoch": 9.640330188679245,
      "grad_norm": 3.2174742221832275,
      "learning_rate": 6.1759306151522815e-06,
      "loss": 0.4797,
      "num_input_tokens_seen": 10777200,
      "step": 16350
    },
    {
      "epoch": 9.643278301886792,
      "grad_norm": 1.3480253219604492,
      "learning_rate": 6.173429909210755e-06,
      "loss": 0.3854,
      "num_input_tokens_seen": 10780464,
      "step": 16355
    },
    {
      "epoch": 9.64622641509434,
      "grad_norm": 2.2446272373199463,
      "learning_rate": 6.170928892598606e-06,
      "loss": 0.3413,
      "num_input_tokens_seen": 10783920,
      "step": 16360
    },
    {
      "epoch": 9.649174528301886,
      "grad_norm": 1.5758671760559082,
      "learning_rate": 6.168427565977984e-06,
      "loss": 0.3881,
      "num_input_tokens_seen": 10787600,
      "step": 16365
    },
    {
      "epoch": 9.652122641509434,
      "grad_norm": 2.22428560256958,
      "learning_rate": 6.165925930011129e-06,
      "loss": 0.4713,
      "num_input_tokens_seen": 10790896,
      "step": 16370
    },
    {
      "epoch": 9.65507075471698,
      "grad_norm": 3.593498706817627,
      "learning_rate": 6.163423985360359e-06,
      "loss": 0.361,
      "num_input_tokens_seen": 10794256,
      "step": 16375
    },
    {
      "epoch": 9.658018867924529,
      "grad_norm": 2.109630584716797,
      "learning_rate": 6.160921732688076e-06,
      "loss": 0.375,
      "num_input_tokens_seen": 10797168,
      "step": 16380
    },
    {
      "epoch": 9.660966981132075,
      "grad_norm": 2.477705240249634,
      "learning_rate": 6.158419172656759e-06,
      "loss": 0.3736,
      "num_input_tokens_seen": 10799792,
      "step": 16385
    },
    {
      "epoch": 9.663915094339622,
      "grad_norm": 3.1172754764556885,
      "learning_rate": 6.155916305928974e-06,
      "loss": 0.3225,
      "num_input_tokens_seen": 10803088,
      "step": 16390
    },
    {
      "epoch": 9.66686320754717,
      "grad_norm": 1.466437578201294,
      "learning_rate": 6.153413133167366e-06,
      "loss": 0.3878,
      "num_input_tokens_seen": 10807728,
      "step": 16395
    },
    {
      "epoch": 9.669811320754716,
      "grad_norm": 3.7209537029266357,
      "learning_rate": 6.1509096550346596e-06,
      "loss": 0.3631,
      "num_input_tokens_seen": 10810960,
      "step": 16400
    },
    {
      "epoch": 9.672759433962264,
      "grad_norm": 1.5192146301269531,
      "learning_rate": 6.148405872193661e-06,
      "loss": 0.3592,
      "num_input_tokens_seen": 10813584,
      "step": 16405
    },
    {
      "epoch": 9.67570754716981,
      "grad_norm": 4.457874298095703,
      "learning_rate": 6.14590178530726e-06,
      "loss": 0.4353,
      "num_input_tokens_seen": 10816944,
      "step": 16410
    },
    {
      "epoch": 9.678655660377359,
      "grad_norm": 2.624408483505249,
      "learning_rate": 6.143397395038422e-06,
      "loss": 0.3223,
      "num_input_tokens_seen": 10820080,
      "step": 16415
    },
    {
      "epoch": 9.681603773584905,
      "grad_norm": 2.836275339126587,
      "learning_rate": 6.140892702050196e-06,
      "loss": 0.4101,
      "num_input_tokens_seen": 10822832,
      "step": 16420
    },
    {
      "epoch": 9.684551886792454,
      "grad_norm": 4.298067092895508,
      "learning_rate": 6.138387707005711e-06,
      "loss": 0.3621,
      "num_input_tokens_seen": 10825520,
      "step": 16425
    },
    {
      "epoch": 9.6875,
      "grad_norm": 2.77807879447937,
      "learning_rate": 6.135882410568172e-06,
      "loss": 0.402,
      "num_input_tokens_seen": 10828528,
      "step": 16430
    },
    {
      "epoch": 9.690448113207546,
      "grad_norm": 1.996529221534729,
      "learning_rate": 6.133376813400872e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 10832176,
      "step": 16435
    },
    {
      "epoch": 9.693396226415095,
      "grad_norm": 5.04343318939209,
      "learning_rate": 6.130870916167175e-06,
      "loss": 0.4009,
      "num_input_tokens_seen": 10835280,
      "step": 16440
    },
    {
      "epoch": 9.696344339622641,
      "grad_norm": 2.296175003051758,
      "learning_rate": 6.128364719530528e-06,
      "loss": 0.3993,
      "num_input_tokens_seen": 10838544,
      "step": 16445
    },
    {
      "epoch": 9.69929245283019,
      "grad_norm": 1.8875367641448975,
      "learning_rate": 6.125858224154459e-06,
      "loss": 0.4381,
      "num_input_tokens_seen": 10841936,
      "step": 16450
    },
    {
      "epoch": 9.702240566037736,
      "grad_norm": 2.871469736099243,
      "learning_rate": 6.123351430702576e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 10845168,
      "step": 16455
    },
    {
      "epoch": 9.705188679245284,
      "grad_norm": 3.7956702709198,
      "learning_rate": 6.1208443398385575e-06,
      "loss": 0.3487,
      "num_input_tokens_seen": 10848240,
      "step": 16460
    },
    {
      "epoch": 9.70813679245283,
      "grad_norm": 2.2703804969787598,
      "learning_rate": 6.118336952226169e-06,
      "loss": 0.59,
      "num_input_tokens_seen": 10850800,
      "step": 16465
    },
    {
      "epoch": 9.711084905660378,
      "grad_norm": 1.7670515775680542,
      "learning_rate": 6.115829268529254e-06,
      "loss": 0.3362,
      "num_input_tokens_seen": 10854544,
      "step": 16470
    },
    {
      "epoch": 9.714033018867925,
      "grad_norm": 4.724531650543213,
      "learning_rate": 6.1133212894117326e-06,
      "loss": 0.4121,
      "num_input_tokens_seen": 10858608,
      "step": 16475
    },
    {
      "epoch": 9.716981132075471,
      "grad_norm": 1.5524672269821167,
      "learning_rate": 6.1108130155375986e-06,
      "loss": 0.3382,
      "num_input_tokens_seen": 10861712,
      "step": 16480
    },
    {
      "epoch": 9.71992924528302,
      "grad_norm": 2.334449291229248,
      "learning_rate": 6.108304447570933e-06,
      "loss": 0.2772,
      "num_input_tokens_seen": 10864912,
      "step": 16485
    },
    {
      "epoch": 9.722877358490566,
      "grad_norm": 3.9100704193115234,
      "learning_rate": 6.105795586175888e-06,
      "loss": 0.2627,
      "num_input_tokens_seen": 10867344,
      "step": 16490
    },
    {
      "epoch": 9.725825471698114,
      "grad_norm": 1.9105088710784912,
      "learning_rate": 6.1032864320166954e-06,
      "loss": 0.3403,
      "num_input_tokens_seen": 10871056,
      "step": 16495
    },
    {
      "epoch": 9.72877358490566,
      "grad_norm": 4.489187717437744,
      "learning_rate": 6.100776985757666e-06,
      "loss": 0.3365,
      "num_input_tokens_seen": 10874512,
      "step": 16500
    },
    {
      "epoch": 9.731721698113208,
      "grad_norm": 1.5112676620483398,
      "learning_rate": 6.098267248063186e-06,
      "loss": 0.4998,
      "num_input_tokens_seen": 10877360,
      "step": 16505
    },
    {
      "epoch": 9.734669811320755,
      "grad_norm": 2.788297414779663,
      "learning_rate": 6.0957572195977165e-06,
      "loss": 0.3456,
      "num_input_tokens_seen": 10880272,
      "step": 16510
    },
    {
      "epoch": 9.737617924528301,
      "grad_norm": 3.2062230110168457,
      "learning_rate": 6.0932469010258025e-06,
      "loss": 0.3386,
      "num_input_tokens_seen": 10883568,
      "step": 16515
    },
    {
      "epoch": 9.74056603773585,
      "grad_norm": 2.00248384475708,
      "learning_rate": 6.0907362930120594e-06,
      "loss": 0.4295,
      "num_input_tokens_seen": 10887184,
      "step": 16520
    },
    {
      "epoch": 9.743514150943396,
      "grad_norm": 2.560000419616699,
      "learning_rate": 6.088225396221181e-06,
      "loss": 0.3841,
      "num_input_tokens_seen": 10890160,
      "step": 16525
    },
    {
      "epoch": 9.746462264150944,
      "grad_norm": 4.631329536437988,
      "learning_rate": 6.0857142113179415e-06,
      "loss": 0.4524,
      "num_input_tokens_seen": 10896944,
      "step": 16530
    },
    {
      "epoch": 9.74941037735849,
      "grad_norm": 5.107585906982422,
      "learning_rate": 6.083202738967182e-06,
      "loss": 0.417,
      "num_input_tokens_seen": 10899856,
      "step": 16535
    },
    {
      "epoch": 9.752358490566039,
      "grad_norm": 1.8961753845214844,
      "learning_rate": 6.0806909798338324e-06,
      "loss": 0.2901,
      "num_input_tokens_seen": 10902704,
      "step": 16540
    },
    {
      "epoch": 9.755306603773585,
      "grad_norm": 2.0145537853240967,
      "learning_rate": 6.0781789345828854e-06,
      "loss": 0.4776,
      "num_input_tokens_seen": 10905648,
      "step": 16545
    },
    {
      "epoch": 9.758254716981131,
      "grad_norm": 4.600338935852051,
      "learning_rate": 6.0756666038794195e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 10908752,
      "step": 16550
    },
    {
      "epoch": 9.76120283018868,
      "grad_norm": 2.3824007511138916,
      "learning_rate": 6.073153988388586e-06,
      "loss": 0.3116,
      "num_input_tokens_seen": 10911728,
      "step": 16555
    },
    {
      "epoch": 9.764150943396226,
      "grad_norm": 2.268585443496704,
      "learning_rate": 6.070641088775608e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 10914992,
      "step": 16560
    },
    {
      "epoch": 9.767099056603774,
      "grad_norm": 2.8551199436187744,
      "learning_rate": 6.068127905705787e-06,
      "loss": 0.3768,
      "num_input_tokens_seen": 10917424,
      "step": 16565
    },
    {
      "epoch": 9.77004716981132,
      "grad_norm": 1.9698865413665771,
      "learning_rate": 6.065614439844501e-06,
      "loss": 0.3254,
      "num_input_tokens_seen": 10921104,
      "step": 16570
    },
    {
      "epoch": 9.772995283018869,
      "grad_norm": 1.81951904296875,
      "learning_rate": 6.063100691857198e-06,
      "loss": 0.4148,
      "num_input_tokens_seen": 10924816,
      "step": 16575
    },
    {
      "epoch": 9.775943396226415,
      "grad_norm": 2.4315834045410156,
      "learning_rate": 6.060586662409407e-06,
      "loss": 0.3912,
      "num_input_tokens_seen": 10928784,
      "step": 16580
    },
    {
      "epoch": 9.778891509433961,
      "grad_norm": 3.3420653343200684,
      "learning_rate": 6.058072352166724e-06,
      "loss": 0.3441,
      "num_input_tokens_seen": 10932560,
      "step": 16585
    },
    {
      "epoch": 9.78183962264151,
      "grad_norm": 3.888482093811035,
      "learning_rate": 6.055557761794826e-06,
      "loss": 0.5294,
      "num_input_tokens_seen": 10935344,
      "step": 16590
    },
    {
      "epoch": 9.784787735849056,
      "grad_norm": 2.2944226264953613,
      "learning_rate": 6.053042891959462e-06,
      "loss": 0.4417,
      "num_input_tokens_seen": 10939408,
      "step": 16595
    },
    {
      "epoch": 9.787735849056604,
      "grad_norm": 3.5459773540496826,
      "learning_rate": 6.050527743326455e-06,
      "loss": 0.4129,
      "num_input_tokens_seen": 10942032,
      "step": 16600
    },
    {
      "epoch": 9.79068396226415,
      "grad_norm": 2.176516056060791,
      "learning_rate": 6.048012316561699e-06,
      "loss": 0.4349,
      "num_input_tokens_seen": 10944944,
      "step": 16605
    },
    {
      "epoch": 9.793632075471699,
      "grad_norm": 3.7128469944000244,
      "learning_rate": 6.045496612331166e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 10948272,
      "step": 16610
    },
    {
      "epoch": 9.796580188679245,
      "grad_norm": 2.540820360183716,
      "learning_rate": 6.0429806313009e-06,
      "loss": 0.4418,
      "num_input_tokens_seen": 10950992,
      "step": 16615
    },
    {
      "epoch": 9.799528301886792,
      "grad_norm": 2.756800889968872,
      "learning_rate": 6.040464374137015e-06,
      "loss": 0.2871,
      "num_input_tokens_seen": 10954640,
      "step": 16620
    },
    {
      "epoch": 9.80247641509434,
      "grad_norm": 1.8353554010391235,
      "learning_rate": 6.0379478415057045e-06,
      "loss": 0.2637,
      "num_input_tokens_seen": 10958032,
      "step": 16625
    },
    {
      "epoch": 9.805424528301886,
      "grad_norm": 3.081576347351074,
      "learning_rate": 6.035431034073228e-06,
      "loss": 0.3441,
      "num_input_tokens_seen": 10962224,
      "step": 16630
    },
    {
      "epoch": 9.808372641509434,
      "grad_norm": 4.996615886688232,
      "learning_rate": 6.0329139525059235e-06,
      "loss": 0.3613,
      "num_input_tokens_seen": 10964816,
      "step": 16635
    },
    {
      "epoch": 9.81132075471698,
      "grad_norm": 1.8016397953033447,
      "learning_rate": 6.030396597470198e-06,
      "loss": 0.361,
      "num_input_tokens_seen": 10968624,
      "step": 16640
    },
    {
      "epoch": 9.814268867924529,
      "grad_norm": 2.5757482051849365,
      "learning_rate": 6.027878969632534e-06,
      "loss": 0.3633,
      "num_input_tokens_seen": 10971792,
      "step": 16645
    },
    {
      "epoch": 9.817216981132075,
      "grad_norm": 2.217841863632202,
      "learning_rate": 6.025361069659482e-06,
      "loss": 0.2944,
      "num_input_tokens_seen": 10974768,
      "step": 16650
    },
    {
      "epoch": 9.820165094339622,
      "grad_norm": 3.674365758895874,
      "learning_rate": 6.022842898217668e-06,
      "loss": 0.4413,
      "num_input_tokens_seen": 10977456,
      "step": 16655
    },
    {
      "epoch": 9.82311320754717,
      "grad_norm": 2.1471965312957764,
      "learning_rate": 6.020324455973788e-06,
      "loss": 0.3004,
      "num_input_tokens_seen": 10981232,
      "step": 16660
    },
    {
      "epoch": 9.826061320754716,
      "grad_norm": 4.137482643127441,
      "learning_rate": 6.017805743594612e-06,
      "loss": 0.4032,
      "num_input_tokens_seen": 10983376,
      "step": 16665
    },
    {
      "epoch": 9.829009433962264,
      "grad_norm": 6.137082576751709,
      "learning_rate": 6.0152867617469776e-06,
      "loss": 0.3178,
      "num_input_tokens_seen": 10987824,
      "step": 16670
    },
    {
      "epoch": 9.83195754716981,
      "grad_norm": 3.176222562789917,
      "learning_rate": 6.012767511097799e-06,
      "loss": 0.4393,
      "num_input_tokens_seen": 10991408,
      "step": 16675
    },
    {
      "epoch": 9.834905660377359,
      "grad_norm": 3.8801279067993164,
      "learning_rate": 6.010247992314055e-06,
      "loss": 0.5009,
      "num_input_tokens_seen": 10994640,
      "step": 16680
    },
    {
      "epoch": 9.837853773584905,
      "grad_norm": 4.115037441253662,
      "learning_rate": 6.007728206062802e-06,
      "loss": 0.4586,
      "num_input_tokens_seen": 10998256,
      "step": 16685
    },
    {
      "epoch": 9.840801886792454,
      "grad_norm": 3.744828224182129,
      "learning_rate": 6.005208153011163e-06,
      "loss": 0.4408,
      "num_input_tokens_seen": 11001456,
      "step": 16690
    },
    {
      "epoch": 9.84375,
      "grad_norm": 3.0440428256988525,
      "learning_rate": 6.0026878338263335e-06,
      "loss": 0.3665,
      "num_input_tokens_seen": 11004880,
      "step": 16695
    },
    {
      "epoch": 9.846698113207546,
      "grad_norm": 2.0580384731292725,
      "learning_rate": 6.000167249175579e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 11008048,
      "step": 16700
    },
    {
      "epoch": 9.849646226415095,
      "grad_norm": 2.2548370361328125,
      "learning_rate": 5.997646399726236e-06,
      "loss": 0.4164,
      "num_input_tokens_seen": 11011504,
      "step": 16705
    },
    {
      "epoch": 9.852594339622641,
      "grad_norm": 2.140430212020874,
      "learning_rate": 5.995125286145707e-06,
      "loss": 0.2716,
      "num_input_tokens_seen": 11014128,
      "step": 16710
    },
    {
      "epoch": 9.85554245283019,
      "grad_norm": 0.5407708883285522,
      "learning_rate": 5.99260390910147e-06,
      "loss": 0.2733,
      "num_input_tokens_seen": 11019376,
      "step": 16715
    },
    {
      "epoch": 9.858490566037736,
      "grad_norm": 2.2536873817443848,
      "learning_rate": 5.990082269261071e-06,
      "loss": 0.3036,
      "num_input_tokens_seen": 11022128,
      "step": 16720
    },
    {
      "epoch": 9.861438679245284,
      "grad_norm": 1.9797874689102173,
      "learning_rate": 5.987560367292123e-06,
      "loss": 0.3551,
      "num_input_tokens_seen": 11024656,
      "step": 16725
    },
    {
      "epoch": 9.86438679245283,
      "grad_norm": 4.391763210296631,
      "learning_rate": 5.985038203862313e-06,
      "loss": 0.531,
      "num_input_tokens_seen": 11027280,
      "step": 16730
    },
    {
      "epoch": 9.867334905660378,
      "grad_norm": 2.214672327041626,
      "learning_rate": 5.982515779639393e-06,
      "loss": 0.3371,
      "num_input_tokens_seen": 11030672,
      "step": 16735
    },
    {
      "epoch": 9.870283018867925,
      "grad_norm": 4.609348297119141,
      "learning_rate": 5.979993095291186e-06,
      "loss": 0.3193,
      "num_input_tokens_seen": 11034544,
      "step": 16740
    },
    {
      "epoch": 9.873231132075471,
      "grad_norm": 3.162928819656372,
      "learning_rate": 5.977470151485582e-06,
      "loss": 0.4287,
      "num_input_tokens_seen": 11037072,
      "step": 16745
    },
    {
      "epoch": 9.87617924528302,
      "grad_norm": 2.5031254291534424,
      "learning_rate": 5.974946948890544e-06,
      "loss": 0.3153,
      "num_input_tokens_seen": 11040112,
      "step": 16750
    },
    {
      "epoch": 9.879127358490566,
      "grad_norm": 2.9616806507110596,
      "learning_rate": 5.9724234881740994e-06,
      "loss": 0.3546,
      "num_input_tokens_seen": 11045232,
      "step": 16755
    },
    {
      "epoch": 9.882075471698114,
      "grad_norm": 3.0209708213806152,
      "learning_rate": 5.9698997700043445e-06,
      "loss": 0.3767,
      "num_input_tokens_seen": 11047440,
      "step": 16760
    },
    {
      "epoch": 9.88502358490566,
      "grad_norm": 2.8031694889068604,
      "learning_rate": 5.9673757950494475e-06,
      "loss": 0.3728,
      "num_input_tokens_seen": 11051280,
      "step": 16765
    },
    {
      "epoch": 9.887971698113208,
      "grad_norm": 2.5094563961029053,
      "learning_rate": 5.964851563977639e-06,
      "loss": 0.3493,
      "num_input_tokens_seen": 11054608,
      "step": 16770
    },
    {
      "epoch": 9.890919811320755,
      "grad_norm": 2.2560312747955322,
      "learning_rate": 5.962327077457219e-06,
      "loss": 0.3556,
      "num_input_tokens_seen": 11057936,
      "step": 16775
    },
    {
      "epoch": 9.893867924528301,
      "grad_norm": 3.0720200538635254,
      "learning_rate": 5.959802336156558e-06,
      "loss": 0.2843,
      "num_input_tokens_seen": 11061488,
      "step": 16780
    },
    {
      "epoch": 9.89681603773585,
      "grad_norm": 2.8033299446105957,
      "learning_rate": 5.957277340744094e-06,
      "loss": 0.4644,
      "num_input_tokens_seen": 11064560,
      "step": 16785
    },
    {
      "epoch": 9.899764150943396,
      "grad_norm": 4.301828861236572,
      "learning_rate": 5.954752091888326e-06,
      "loss": 0.3493,
      "num_input_tokens_seen": 11067760,
      "step": 16790
    },
    {
      "epoch": 9.902712264150944,
      "grad_norm": 2.287024736404419,
      "learning_rate": 5.952226590257829e-06,
      "loss": 0.3598,
      "num_input_tokens_seen": 11070576,
      "step": 16795
    },
    {
      "epoch": 9.90566037735849,
      "grad_norm": 3.1033740043640137,
      "learning_rate": 5.949700836521239e-06,
      "loss": 0.3815,
      "num_input_tokens_seen": 11074416,
      "step": 16800
    },
    {
      "epoch": 9.908608490566039,
      "grad_norm": 5.4590959548950195,
      "learning_rate": 5.947174831347257e-06,
      "loss": 0.3763,
      "num_input_tokens_seen": 11078160,
      "step": 16805
    },
    {
      "epoch": 9.911556603773585,
      "grad_norm": 2.140302896499634,
      "learning_rate": 5.944648575404657e-06,
      "loss": 0.4018,
      "num_input_tokens_seen": 11080848,
      "step": 16810
    },
    {
      "epoch": 9.914504716981131,
      "grad_norm": 2.0905725955963135,
      "learning_rate": 5.942122069362276e-06,
      "loss": 0.3331,
      "num_input_tokens_seen": 11083760,
      "step": 16815
    },
    {
      "epoch": 9.91745283018868,
      "grad_norm": 3.884600877761841,
      "learning_rate": 5.939595313889016e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 11086256,
      "step": 16820
    },
    {
      "epoch": 9.920400943396226,
      "grad_norm": 2.157484531402588,
      "learning_rate": 5.937068309653848e-06,
      "loss": 0.4575,
      "num_input_tokens_seen": 11088848,
      "step": 16825
    },
    {
      "epoch": 9.923349056603774,
      "grad_norm": 3.806908130645752,
      "learning_rate": 5.934541057325807e-06,
      "loss": 0.5658,
      "num_input_tokens_seen": 11092464,
      "step": 16830
    },
    {
      "epoch": 9.92629716981132,
      "grad_norm": 2.377638101577759,
      "learning_rate": 5.932013557573992e-06,
      "loss": 0.4293,
      "num_input_tokens_seen": 11095824,
      "step": 16835
    },
    {
      "epoch": 9.929245283018869,
      "grad_norm": 2.282193422317505,
      "learning_rate": 5.929485811067572e-06,
      "loss": 0.283,
      "num_input_tokens_seen": 11099376,
      "step": 16840
    },
    {
      "epoch": 9.932193396226415,
      "grad_norm": 9.890433311462402,
      "learning_rate": 5.926957818475778e-06,
      "loss": 0.3157,
      "num_input_tokens_seen": 11104144,
      "step": 16845
    },
    {
      "epoch": 9.935141509433961,
      "grad_norm": 3.9738645553588867,
      "learning_rate": 5.924429580467905e-06,
      "loss": 0.4131,
      "num_input_tokens_seen": 11107856,
      "step": 16850
    },
    {
      "epoch": 9.93808962264151,
      "grad_norm": 4.6767096519470215,
      "learning_rate": 5.921901097713317e-06,
      "loss": 0.5004,
      "num_input_tokens_seen": 11110416,
      "step": 16855
    },
    {
      "epoch": 9.941037735849056,
      "grad_norm": 2.4528751373291016,
      "learning_rate": 5.919372370881442e-06,
      "loss": 0.3065,
      "num_input_tokens_seen": 11112848,
      "step": 16860
    },
    {
      "epoch": 9.943985849056604,
      "grad_norm": 3.2868075370788574,
      "learning_rate": 5.916843400641768e-06,
      "loss": 0.4216,
      "num_input_tokens_seen": 11115216,
      "step": 16865
    },
    {
      "epoch": 9.94693396226415,
      "grad_norm": 2.182129383087158,
      "learning_rate": 5.914314187663851e-06,
      "loss": 0.3907,
      "num_input_tokens_seen": 11118576,
      "step": 16870
    },
    {
      "epoch": 9.949882075471699,
      "grad_norm": 1.8008551597595215,
      "learning_rate": 5.911784732617314e-06,
      "loss": 0.3534,
      "num_input_tokens_seen": 11121456,
      "step": 16875
    },
    {
      "epoch": 9.952830188679245,
      "grad_norm": 5.291997909545898,
      "learning_rate": 5.90925503617184e-06,
      "loss": 0.324,
      "num_input_tokens_seen": 11124240,
      "step": 16880
    },
    {
      "epoch": 9.955778301886792,
      "grad_norm": 4.448596954345703,
      "learning_rate": 5.9067250989971745e-06,
      "loss": 0.3944,
      "num_input_tokens_seen": 11127888,
      "step": 16885
    },
    {
      "epoch": 9.95872641509434,
      "grad_norm": 2.5647881031036377,
      "learning_rate": 5.904194921763133e-06,
      "loss": 0.3444,
      "num_input_tokens_seen": 11131088,
      "step": 16890
    },
    {
      "epoch": 9.961674528301886,
      "grad_norm": 2.233388900756836,
      "learning_rate": 5.901664505139589e-06,
      "loss": 0.5355,
      "num_input_tokens_seen": 11135248,
      "step": 16895
    },
    {
      "epoch": 9.964622641509434,
      "grad_norm": 1.5022283792495728,
      "learning_rate": 5.8991338497964814e-06,
      "loss": 0.2978,
      "num_input_tokens_seen": 11138800,
      "step": 16900
    },
    {
      "epoch": 9.96757075471698,
      "grad_norm": 5.132346153259277,
      "learning_rate": 5.896602956403812e-06,
      "loss": 0.4102,
      "num_input_tokens_seen": 11142384,
      "step": 16905
    },
    {
      "epoch": 9.970518867924529,
      "grad_norm": 2.5933454036712646,
      "learning_rate": 5.894071825631645e-06,
      "loss": 0.3951,
      "num_input_tokens_seen": 11145680,
      "step": 16910
    },
    {
      "epoch": 9.973466981132075,
      "grad_norm": 2.4310953617095947,
      "learning_rate": 5.891540458150109e-06,
      "loss": 0.4012,
      "num_input_tokens_seen": 11149392,
      "step": 16915
    },
    {
      "epoch": 9.976415094339622,
      "grad_norm": 2.8937530517578125,
      "learning_rate": 5.889008854629395e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 11152240,
      "step": 16920
    },
    {
      "epoch": 9.97936320754717,
      "grad_norm": 2.086498260498047,
      "learning_rate": 5.886477015739754e-06,
      "loss": 0.3514,
      "num_input_tokens_seen": 11155376,
      "step": 16925
    },
    {
      "epoch": 9.982311320754716,
      "grad_norm": 2.0598578453063965,
      "learning_rate": 5.883944942151502e-06,
      "loss": 0.5083,
      "num_input_tokens_seen": 11158512,
      "step": 16930
    },
    {
      "epoch": 9.985259433962264,
      "grad_norm": 2.2231967449188232,
      "learning_rate": 5.88141263453502e-06,
      "loss": 0.4999,
      "num_input_tokens_seen": 11161264,
      "step": 16935
    },
    {
      "epoch": 9.98820754716981,
      "grad_norm": 1.8537381887435913,
      "learning_rate": 5.878880093560744e-06,
      "loss": 0.4499,
      "num_input_tokens_seen": 11164464,
      "step": 16940
    },
    {
      "epoch": 9.991155660377359,
      "grad_norm": 2.4095025062561035,
      "learning_rate": 5.876347319899173e-06,
      "loss": 0.3578,
      "num_input_tokens_seen": 11167728,
      "step": 16945
    },
    {
      "epoch": 9.994103773584905,
      "grad_norm": 5.645260334014893,
      "learning_rate": 5.873814314220874e-06,
      "loss": 0.3859,
      "num_input_tokens_seen": 11170256,
      "step": 16950
    },
    {
      "epoch": 9.997051886792454,
      "grad_norm": 4.420820236206055,
      "learning_rate": 5.871281077196469e-06,
      "loss": 0.4061,
      "num_input_tokens_seen": 11173200,
      "step": 16955
    },
    {
      "epoch": 10.0,
      "grad_norm": 5.013511657714844,
      "learning_rate": 5.868747609496643e-06,
      "loss": 0.4498,
      "num_input_tokens_seen": 11175920,
      "step": 16960
    },
    {
      "epoch": 10.0,
      "eval_loss": 0.517292857170105,
      "eval_runtime": 18.4753,
      "eval_samples_per_second": 91.798,
      "eval_steps_per_second": 22.95,
      "num_input_tokens_seen": 11175920,
      "step": 16960
    },
    {
      "epoch": 10.002948113207546,
      "grad_norm": 3.551954507827759,
      "learning_rate": 5.8662139117921435e-06,
      "loss": 0.4673,
      "num_input_tokens_seen": 11179632,
      "step": 16965
    },
    {
      "epoch": 10.005896226415095,
      "grad_norm": 2.372073173522949,
      "learning_rate": 5.8636799847537785e-06,
      "loss": 0.3146,
      "num_input_tokens_seen": 11182544,
      "step": 16970
    },
    {
      "epoch": 10.008844339622641,
      "grad_norm": 1.969895601272583,
      "learning_rate": 5.861145829052415e-06,
      "loss": 0.5098,
      "num_input_tokens_seen": 11186704,
      "step": 16975
    },
    {
      "epoch": 10.01179245283019,
      "grad_norm": 2.2855591773986816,
      "learning_rate": 5.858611445358982e-06,
      "loss": 0.3644,
      "num_input_tokens_seen": 11190032,
      "step": 16980
    },
    {
      "epoch": 10.014740566037736,
      "grad_norm": 2.4234156608581543,
      "learning_rate": 5.856076834344468e-06,
      "loss": 0.3688,
      "num_input_tokens_seen": 11193232,
      "step": 16985
    },
    {
      "epoch": 10.017688679245284,
      "grad_norm": 1.7573461532592773,
      "learning_rate": 5.853541996679924e-06,
      "loss": 0.3522,
      "num_input_tokens_seen": 11196688,
      "step": 16990
    },
    {
      "epoch": 10.02063679245283,
      "grad_norm": 3.1783456802368164,
      "learning_rate": 5.851006933036456e-06,
      "loss": 0.2796,
      "num_input_tokens_seen": 11199056,
      "step": 16995
    },
    {
      "epoch": 10.023584905660377,
      "grad_norm": 1.4724727869033813,
      "learning_rate": 5.848471644085236e-06,
      "loss": 0.2656,
      "num_input_tokens_seen": 11201488,
      "step": 17000
    },
    {
      "epoch": 10.026533018867925,
      "grad_norm": 2.105929136276245,
      "learning_rate": 5.84593613049749e-06,
      "loss": 0.3411,
      "num_input_tokens_seen": 11205264,
      "step": 17005
    },
    {
      "epoch": 10.029481132075471,
      "grad_norm": 2.2735376358032227,
      "learning_rate": 5.843400392944509e-06,
      "loss": 0.3283,
      "num_input_tokens_seen": 11208240,
      "step": 17010
    },
    {
      "epoch": 10.03242924528302,
      "grad_norm": 4.815243721008301,
      "learning_rate": 5.840864432097639e-06,
      "loss": 0.5012,
      "num_input_tokens_seen": 11211696,
      "step": 17015
    },
    {
      "epoch": 10.035377358490566,
      "grad_norm": 3.405665874481201,
      "learning_rate": 5.838328248628286e-06,
      "loss": 0.3471,
      "num_input_tokens_seen": 11213968,
      "step": 17020
    },
    {
      "epoch": 10.038325471698114,
      "grad_norm": 2.249446392059326,
      "learning_rate": 5.835791843207916e-06,
      "loss": 0.4902,
      "num_input_tokens_seen": 11217168,
      "step": 17025
    },
    {
      "epoch": 10.04127358490566,
      "grad_norm": 1.7196645736694336,
      "learning_rate": 5.833255216508056e-06,
      "loss": 0.2933,
      "num_input_tokens_seen": 11221040,
      "step": 17030
    },
    {
      "epoch": 10.044221698113208,
      "grad_norm": 3.2542104721069336,
      "learning_rate": 5.830718369200284e-06,
      "loss": 0.3766,
      "num_input_tokens_seen": 11224688,
      "step": 17035
    },
    {
      "epoch": 10.047169811320755,
      "grad_norm": 5.314554214477539,
      "learning_rate": 5.828181301956244e-06,
      "loss": 0.4673,
      "num_input_tokens_seen": 11227760,
      "step": 17040
    },
    {
      "epoch": 10.050117924528301,
      "grad_norm": 2.1805331707000732,
      "learning_rate": 5.825644015447636e-06,
      "loss": 0.4404,
      "num_input_tokens_seen": 11230288,
      "step": 17045
    },
    {
      "epoch": 10.05306603773585,
      "grad_norm": 3.858076572418213,
      "learning_rate": 5.823106510346216e-06,
      "loss": 0.3335,
      "num_input_tokens_seen": 11233520,
      "step": 17050
    },
    {
      "epoch": 10.056014150943396,
      "grad_norm": 2.7234296798706055,
      "learning_rate": 5.820568787323798e-06,
      "loss": 0.3012,
      "num_input_tokens_seen": 11236944,
      "step": 17055
    },
    {
      "epoch": 10.058962264150944,
      "grad_norm": 1.9072738885879517,
      "learning_rate": 5.818030847052258e-06,
      "loss": 0.344,
      "num_input_tokens_seen": 11240848,
      "step": 17060
    },
    {
      "epoch": 10.06191037735849,
      "grad_norm": 1.895639181137085,
      "learning_rate": 5.815492690203528e-06,
      "loss": 0.322,
      "num_input_tokens_seen": 11244528,
      "step": 17065
    },
    {
      "epoch": 10.064858490566039,
      "grad_norm": 2.5032248497009277,
      "learning_rate": 5.812954317449591e-06,
      "loss": 0.5287,
      "num_input_tokens_seen": 11247824,
      "step": 17070
    },
    {
      "epoch": 10.067806603773585,
      "grad_norm": 2.204399585723877,
      "learning_rate": 5.810415729462496e-06,
      "loss": 0.3295,
      "num_input_tokens_seen": 11251152,
      "step": 17075
    },
    {
      "epoch": 10.070754716981131,
      "grad_norm": 2.662820816040039,
      "learning_rate": 5.807876926914344e-06,
      "loss": 0.395,
      "num_input_tokens_seen": 11254064,
      "step": 17080
    },
    {
      "epoch": 10.07370283018868,
      "grad_norm": 5.665797710418701,
      "learning_rate": 5.8053379104772935e-06,
      "loss": 0.3323,
      "num_input_tokens_seen": 11256688,
      "step": 17085
    },
    {
      "epoch": 10.076650943396226,
      "grad_norm": 2.175389528274536,
      "learning_rate": 5.802798680823562e-06,
      "loss": 0.2621,
      "num_input_tokens_seen": 11259984,
      "step": 17090
    },
    {
      "epoch": 10.079599056603774,
      "grad_norm": 2.216608762741089,
      "learning_rate": 5.8002592386254186e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 11262384,
      "step": 17095
    },
    {
      "epoch": 10.08254716981132,
      "grad_norm": 2.7520909309387207,
      "learning_rate": 5.797719584555192e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 11266160,
      "step": 17100
    },
    {
      "epoch": 10.085495283018869,
      "grad_norm": 3.352846622467041,
      "learning_rate": 5.795179719285269e-06,
      "loss": 0.3396,
      "num_input_tokens_seen": 11269008,
      "step": 17105
    },
    {
      "epoch": 10.088443396226415,
      "grad_norm": 2.1978230476379395,
      "learning_rate": 5.792639643488086e-06,
      "loss": 0.377,
      "num_input_tokens_seen": 11272048,
      "step": 17110
    },
    {
      "epoch": 10.091391509433961,
      "grad_norm": 2.3395111560821533,
      "learning_rate": 5.7900993578361434e-06,
      "loss": 0.4302,
      "num_input_tokens_seen": 11275568,
      "step": 17115
    },
    {
      "epoch": 10.09433962264151,
      "grad_norm": 3.196056842803955,
      "learning_rate": 5.7875588630019895e-06,
      "loss": 0.3266,
      "num_input_tokens_seen": 11278128,
      "step": 17120
    },
    {
      "epoch": 10.097287735849056,
      "grad_norm": 2.456690549850464,
      "learning_rate": 5.7850181596582335e-06,
      "loss": 0.3205,
      "num_input_tokens_seen": 11281456,
      "step": 17125
    },
    {
      "epoch": 10.100235849056604,
      "grad_norm": 3.1088216304779053,
      "learning_rate": 5.782477248477535e-06,
      "loss": 0.252,
      "num_input_tokens_seen": 11284976,
      "step": 17130
    },
    {
      "epoch": 10.10318396226415,
      "grad_norm": 2.117962598800659,
      "learning_rate": 5.779936130132614e-06,
      "loss": 0.4114,
      "num_input_tokens_seen": 11288048,
      "step": 17135
    },
    {
      "epoch": 10.106132075471699,
      "grad_norm": 2.588182210922241,
      "learning_rate": 5.777394805296242e-06,
      "loss": 0.2595,
      "num_input_tokens_seen": 11291152,
      "step": 17140
    },
    {
      "epoch": 10.109080188679245,
      "grad_norm": 3.7348380088806152,
      "learning_rate": 5.774853274641243e-06,
      "loss": 0.3184,
      "num_input_tokens_seen": 11293936,
      "step": 17145
    },
    {
      "epoch": 10.112028301886792,
      "grad_norm": 3.588880777359009,
      "learning_rate": 5.772311538840501e-06,
      "loss": 0.3817,
      "num_input_tokens_seen": 11296912,
      "step": 17150
    },
    {
      "epoch": 10.11497641509434,
      "grad_norm": 3.00968599319458,
      "learning_rate": 5.76976959856695e-06,
      "loss": 0.414,
      "num_input_tokens_seen": 11300752,
      "step": 17155
    },
    {
      "epoch": 10.117924528301886,
      "grad_norm": 2.4523301124572754,
      "learning_rate": 5.767227454493582e-06,
      "loss": 0.3142,
      "num_input_tokens_seen": 11303152,
      "step": 17160
    },
    {
      "epoch": 10.120872641509434,
      "grad_norm": 2.74698543548584,
      "learning_rate": 5.764685107293436e-06,
      "loss": 0.3003,
      "num_input_tokens_seen": 11306064,
      "step": 17165
    },
    {
      "epoch": 10.12382075471698,
      "grad_norm": 2.0423147678375244,
      "learning_rate": 5.762142557639614e-06,
      "loss": 0.3024,
      "num_input_tokens_seen": 11309904,
      "step": 17170
    },
    {
      "epoch": 10.126768867924529,
      "grad_norm": 4.599206924438477,
      "learning_rate": 5.759599806205266e-06,
      "loss": 0.3384,
      "num_input_tokens_seen": 11313136,
      "step": 17175
    },
    {
      "epoch": 10.129716981132075,
      "grad_norm": 2.8574330806732178,
      "learning_rate": 5.757056853663594e-06,
      "loss": 0.2867,
      "num_input_tokens_seen": 11315568,
      "step": 17180
    },
    {
      "epoch": 10.132665094339623,
      "grad_norm": 4.185221195220947,
      "learning_rate": 5.754513700687858e-06,
      "loss": 0.4057,
      "num_input_tokens_seen": 11318960,
      "step": 17185
    },
    {
      "epoch": 10.13561320754717,
      "grad_norm": 3.3884873390197754,
      "learning_rate": 5.751970347951369e-06,
      "loss": 0.4012,
      "num_input_tokens_seen": 11322128,
      "step": 17190
    },
    {
      "epoch": 10.138561320754716,
      "grad_norm": 3.0383033752441406,
      "learning_rate": 5.7494267961274875e-06,
      "loss": 0.3865,
      "num_input_tokens_seen": 11325680,
      "step": 17195
    },
    {
      "epoch": 10.141509433962264,
      "grad_norm": 3.4862968921661377,
      "learning_rate": 5.746883045889633e-06,
      "loss": 0.4417,
      "num_input_tokens_seen": 11329200,
      "step": 17200
    },
    {
      "epoch": 10.14445754716981,
      "grad_norm": 2.5058434009552,
      "learning_rate": 5.744339097911272e-06,
      "loss": 0.4266,
      "num_input_tokens_seen": 11333648,
      "step": 17205
    },
    {
      "epoch": 10.147405660377359,
      "grad_norm": 3.6223361492156982,
      "learning_rate": 5.741794952865928e-06,
      "loss": 0.3007,
      "num_input_tokens_seen": 11337488,
      "step": 17210
    },
    {
      "epoch": 10.150353773584905,
      "grad_norm": 2.884761095046997,
      "learning_rate": 5.739250611427173e-06,
      "loss": 0.414,
      "num_input_tokens_seen": 11341136,
      "step": 17215
    },
    {
      "epoch": 10.153301886792454,
      "grad_norm": 2.070863962173462,
      "learning_rate": 5.736706074268633e-06,
      "loss": 0.3734,
      "num_input_tokens_seen": 11344400,
      "step": 17220
    },
    {
      "epoch": 10.15625,
      "grad_norm": 1.7586400508880615,
      "learning_rate": 5.734161342063984e-06,
      "loss": 0.3563,
      "num_input_tokens_seen": 11348272,
      "step": 17225
    },
    {
      "epoch": 10.159198113207546,
      "grad_norm": 3.5918009281158447,
      "learning_rate": 5.731616415486957e-06,
      "loss": 0.3563,
      "num_input_tokens_seen": 11351056,
      "step": 17230
    },
    {
      "epoch": 10.162146226415095,
      "grad_norm": 4.035353183746338,
      "learning_rate": 5.72907129521133e-06,
      "loss": 0.4593,
      "num_input_tokens_seen": 11353776,
      "step": 17235
    },
    {
      "epoch": 10.165094339622641,
      "grad_norm": 2.1585001945495605,
      "learning_rate": 5.726525981910935e-06,
      "loss": 0.3762,
      "num_input_tokens_seen": 11356944,
      "step": 17240
    },
    {
      "epoch": 10.16804245283019,
      "grad_norm": 2.705728769302368,
      "learning_rate": 5.723980476259658e-06,
      "loss": 0.32,
      "num_input_tokens_seen": 11359920,
      "step": 17245
    },
    {
      "epoch": 10.170990566037736,
      "grad_norm": 2.614358901977539,
      "learning_rate": 5.721434778931426e-06,
      "loss": 0.2306,
      "num_input_tokens_seen": 11363568,
      "step": 17250
    },
    {
      "epoch": 10.173938679245284,
      "grad_norm": 4.282238960266113,
      "learning_rate": 5.7188888906002284e-06,
      "loss": 0.3836,
      "num_input_tokens_seen": 11367728,
      "step": 17255
    },
    {
      "epoch": 10.17688679245283,
      "grad_norm": 2.7606329917907715,
      "learning_rate": 5.716342811940098e-06,
      "loss": 0.3498,
      "num_input_tokens_seen": 11371504,
      "step": 17260
    },
    {
      "epoch": 10.179834905660377,
      "grad_norm": 4.433975696563721,
      "learning_rate": 5.713796543625123e-06,
      "loss": 0.3458,
      "num_input_tokens_seen": 11373744,
      "step": 17265
    },
    {
      "epoch": 10.182783018867925,
      "grad_norm": 3.508326292037964,
      "learning_rate": 5.711250086329435e-06,
      "loss": 0.3732,
      "num_input_tokens_seen": 11376624,
      "step": 17270
    },
    {
      "epoch": 10.185731132075471,
      "grad_norm": 2.202234983444214,
      "learning_rate": 5.7087034407272225e-06,
      "loss": 0.3698,
      "num_input_tokens_seen": 11379984,
      "step": 17275
    },
    {
      "epoch": 10.18867924528302,
      "grad_norm": 3.7500624656677246,
      "learning_rate": 5.70615660749272e-06,
      "loss": 0.3569,
      "num_input_tokens_seen": 11382864,
      "step": 17280
    },
    {
      "epoch": 10.191627358490566,
      "grad_norm": 3.480886936187744,
      "learning_rate": 5.7036095873002106e-06,
      "loss": 0.3453,
      "num_input_tokens_seen": 11386320,
      "step": 17285
    },
    {
      "epoch": 10.194575471698114,
      "grad_norm": 2.2311577796936035,
      "learning_rate": 5.701062380824032e-06,
      "loss": 0.3316,
      "num_input_tokens_seen": 11390256,
      "step": 17290
    },
    {
      "epoch": 10.19752358490566,
      "grad_norm": 4.813438892364502,
      "learning_rate": 5.698514988738566e-06,
      "loss": 0.3709,
      "num_input_tokens_seen": 11393264,
      "step": 17295
    },
    {
      "epoch": 10.200471698113208,
      "grad_norm": 2.2433650493621826,
      "learning_rate": 5.6959674117182465e-06,
      "loss": 0.4348,
      "num_input_tokens_seen": 11397584,
      "step": 17300
    },
    {
      "epoch": 10.203419811320755,
      "grad_norm": 2.1734156608581543,
      "learning_rate": 5.693419650437554e-06,
      "loss": 0.602,
      "num_input_tokens_seen": 11400848,
      "step": 17305
    },
    {
      "epoch": 10.206367924528301,
      "grad_norm": 3.4959535598754883,
      "learning_rate": 5.690871705571022e-06,
      "loss": 0.328,
      "num_input_tokens_seen": 11403760,
      "step": 17310
    },
    {
      "epoch": 10.20931603773585,
      "grad_norm": 1.5809614658355713,
      "learning_rate": 5.688323577793229e-06,
      "loss": 0.4283,
      "num_input_tokens_seen": 11407056,
      "step": 17315
    },
    {
      "epoch": 10.212264150943396,
      "grad_norm": 3.7241950035095215,
      "learning_rate": 5.685775267778801e-06,
      "loss": 0.3968,
      "num_input_tokens_seen": 11409936,
      "step": 17320
    },
    {
      "epoch": 10.215212264150944,
      "grad_norm": 3.1269173622131348,
      "learning_rate": 5.683226776202416e-06,
      "loss": 0.3831,
      "num_input_tokens_seen": 11413680,
      "step": 17325
    },
    {
      "epoch": 10.21816037735849,
      "grad_norm": 3.8473024368286133,
      "learning_rate": 5.680678103738798e-06,
      "loss": 0.3491,
      "num_input_tokens_seen": 11416912,
      "step": 17330
    },
    {
      "epoch": 10.221108490566039,
      "grad_norm": 3.9015421867370605,
      "learning_rate": 5.678129251062717e-06,
      "loss": 0.3004,
      "num_input_tokens_seen": 11419792,
      "step": 17335
    },
    {
      "epoch": 10.224056603773585,
      "grad_norm": 2.681098699569702,
      "learning_rate": 5.675580218848995e-06,
      "loss": 0.4182,
      "num_input_tokens_seen": 11422992,
      "step": 17340
    },
    {
      "epoch": 10.227004716981131,
      "grad_norm": 2.9362740516662598,
      "learning_rate": 5.673031007772498e-06,
      "loss": 0.4616,
      "num_input_tokens_seen": 11425552,
      "step": 17345
    },
    {
      "epoch": 10.22995283018868,
      "grad_norm": 2.9138176441192627,
      "learning_rate": 5.670481618508141e-06,
      "loss": 0.3495,
      "num_input_tokens_seen": 11428592,
      "step": 17350
    },
    {
      "epoch": 10.232900943396226,
      "grad_norm": 2.7549450397491455,
      "learning_rate": 5.667932051730887e-06,
      "loss": 0.3195,
      "num_input_tokens_seen": 11433264,
      "step": 17355
    },
    {
      "epoch": 10.235849056603774,
      "grad_norm": 3.2874300479888916,
      "learning_rate": 5.6653823081157434e-06,
      "loss": 0.4186,
      "num_input_tokens_seen": 11436272,
      "step": 17360
    },
    {
      "epoch": 10.23879716981132,
      "grad_norm": 3.813516616821289,
      "learning_rate": 5.662832388337766e-06,
      "loss": 0.4221,
      "num_input_tokens_seen": 11439824,
      "step": 17365
    },
    {
      "epoch": 10.241745283018869,
      "grad_norm": 3.1911675930023193,
      "learning_rate": 5.660282293072057e-06,
      "loss": 0.259,
      "num_input_tokens_seen": 11443120,
      "step": 17370
    },
    {
      "epoch": 10.244693396226415,
      "grad_norm": 2.1353228092193604,
      "learning_rate": 5.657732022993765e-06,
      "loss": 0.4514,
      "num_input_tokens_seen": 11447376,
      "step": 17375
    },
    {
      "epoch": 10.247641509433961,
      "grad_norm": 4.3171820640563965,
      "learning_rate": 5.655181578778085e-06,
      "loss": 0.3614,
      "num_input_tokens_seen": 11450736,
      "step": 17380
    },
    {
      "epoch": 10.25058962264151,
      "grad_norm": 2.4539194107055664,
      "learning_rate": 5.65263096110026e-06,
      "loss": 0.4529,
      "num_input_tokens_seen": 11454576,
      "step": 17385
    },
    {
      "epoch": 10.253537735849056,
      "grad_norm": 2.5242199897766113,
      "learning_rate": 5.650080170635573e-06,
      "loss": 0.4595,
      "num_input_tokens_seen": 11458384,
      "step": 17390
    },
    {
      "epoch": 10.256485849056604,
      "grad_norm": 2.5216851234436035,
      "learning_rate": 5.647529208059359e-06,
      "loss": 0.4427,
      "num_input_tokens_seen": 11462768,
      "step": 17395
    },
    {
      "epoch": 10.25943396226415,
      "grad_norm": 4.360574722290039,
      "learning_rate": 5.6449780740469985e-06,
      "loss": 0.3722,
      "num_input_tokens_seen": 11466032,
      "step": 17400
    },
    {
      "epoch": 10.262382075471699,
      "grad_norm": 1.980837345123291,
      "learning_rate": 5.642426769273912e-06,
      "loss": 0.3084,
      "num_input_tokens_seen": 11469424,
      "step": 17405
    },
    {
      "epoch": 10.265330188679245,
      "grad_norm": 2.6903836727142334,
      "learning_rate": 5.63987529441557e-06,
      "loss": 0.3305,
      "num_input_tokens_seen": 11472720,
      "step": 17410
    },
    {
      "epoch": 10.268278301886792,
      "grad_norm": 2.267540454864502,
      "learning_rate": 5.637323650147487e-06,
      "loss": 0.2741,
      "num_input_tokens_seen": 11475280,
      "step": 17415
    },
    {
      "epoch": 10.27122641509434,
      "grad_norm": 3.054473876953125,
      "learning_rate": 5.63477183714522e-06,
      "loss": 0.4447,
      "num_input_tokens_seen": 11478608,
      "step": 17420
    },
    {
      "epoch": 10.274174528301886,
      "grad_norm": 3.633803606033325,
      "learning_rate": 5.632219856084373e-06,
      "loss": 0.3441,
      "num_input_tokens_seen": 11481840,
      "step": 17425
    },
    {
      "epoch": 10.277122641509434,
      "grad_norm": 3.495574951171875,
      "learning_rate": 5.6296677076405944e-06,
      "loss": 0.4164,
      "num_input_tokens_seen": 11484944,
      "step": 17430
    },
    {
      "epoch": 10.28007075471698,
      "grad_norm": 1.957209825515747,
      "learning_rate": 5.627115392489578e-06,
      "loss": 0.365,
      "num_input_tokens_seen": 11492016,
      "step": 17435
    },
    {
      "epoch": 10.283018867924529,
      "grad_norm": 2.851461887359619,
      "learning_rate": 5.624562911307058e-06,
      "loss": 0.3669,
      "num_input_tokens_seen": 11495536,
      "step": 17440
    },
    {
      "epoch": 10.285966981132075,
      "grad_norm": 3.406179428100586,
      "learning_rate": 5.622010264768813e-06,
      "loss": 0.3799,
      "num_input_tokens_seen": 11498832,
      "step": 17445
    },
    {
      "epoch": 10.288915094339623,
      "grad_norm": 2.5209360122680664,
      "learning_rate": 5.619457453550673e-06,
      "loss": 0.4193,
      "num_input_tokens_seen": 11501776,
      "step": 17450
    },
    {
      "epoch": 10.29186320754717,
      "grad_norm": 2.3313488960266113,
      "learning_rate": 5.616904478328502e-06,
      "loss": 0.3483,
      "num_input_tokens_seen": 11505712,
      "step": 17455
    },
    {
      "epoch": 10.294811320754716,
      "grad_norm": 2.2830545902252197,
      "learning_rate": 5.6143513397782104e-06,
      "loss": 0.5449,
      "num_input_tokens_seen": 11509232,
      "step": 17460
    },
    {
      "epoch": 10.297759433962264,
      "grad_norm": 2.9282026290893555,
      "learning_rate": 5.611798038575755e-06,
      "loss": 0.363,
      "num_input_tokens_seen": 11511824,
      "step": 17465
    },
    {
      "epoch": 10.30070754716981,
      "grad_norm": 3.6314353942871094,
      "learning_rate": 5.609244575397131e-06,
      "loss": 0.3302,
      "num_input_tokens_seen": 11514672,
      "step": 17470
    },
    {
      "epoch": 10.303655660377359,
      "grad_norm": 5.020689010620117,
      "learning_rate": 5.606690950918381e-06,
      "loss": 0.3756,
      "num_input_tokens_seen": 11517808,
      "step": 17475
    },
    {
      "epoch": 10.306603773584905,
      "grad_norm": 3.099705696105957,
      "learning_rate": 5.604137165815586e-06,
      "loss": 0.3818,
      "num_input_tokens_seen": 11520880,
      "step": 17480
    },
    {
      "epoch": 10.309551886792454,
      "grad_norm": 2.005061626434326,
      "learning_rate": 5.601583220764874e-06,
      "loss": 0.3999,
      "num_input_tokens_seen": 11523600,
      "step": 17485
    },
    {
      "epoch": 10.3125,
      "grad_norm": 2.8692615032196045,
      "learning_rate": 5.599029116442409e-06,
      "loss": 0.3503,
      "num_input_tokens_seen": 11526896,
      "step": 17490
    },
    {
      "epoch": 10.315448113207546,
      "grad_norm": 3.1540842056274414,
      "learning_rate": 5.596474853524406e-06,
      "loss": 0.2845,
      "num_input_tokens_seen": 11529584,
      "step": 17495
    },
    {
      "epoch": 10.318396226415095,
      "grad_norm": 4.052372932434082,
      "learning_rate": 5.593920432687115e-06,
      "loss": 0.4112,
      "num_input_tokens_seen": 11532240,
      "step": 17500
    },
    {
      "epoch": 10.321344339622641,
      "grad_norm": 2.187345266342163,
      "learning_rate": 5.591365854606829e-06,
      "loss": 0.4267,
      "num_input_tokens_seen": 11536560,
      "step": 17505
    },
    {
      "epoch": 10.32429245283019,
      "grad_norm": 3.6245505809783936,
      "learning_rate": 5.588811119959885e-06,
      "loss": 0.3476,
      "num_input_tokens_seen": 11539280,
      "step": 17510
    },
    {
      "epoch": 10.327240566037736,
      "grad_norm": 4.323597431182861,
      "learning_rate": 5.58625622942266e-06,
      "loss": 0.4788,
      "num_input_tokens_seen": 11541840,
      "step": 17515
    },
    {
      "epoch": 10.330188679245284,
      "grad_norm": 2.669999599456787,
      "learning_rate": 5.58370118367157e-06,
      "loss": 0.5361,
      "num_input_tokens_seen": 11545584,
      "step": 17520
    },
    {
      "epoch": 10.33313679245283,
      "grad_norm": 2.8092446327209473,
      "learning_rate": 5.581145983383077e-06,
      "loss": 0.5079,
      "num_input_tokens_seen": 11548208,
      "step": 17525
    },
    {
      "epoch": 10.336084905660377,
      "grad_norm": 3.180938243865967,
      "learning_rate": 5.57859062923368e-06,
      "loss": 0.347,
      "num_input_tokens_seen": 11551472,
      "step": 17530
    },
    {
      "epoch": 10.339033018867925,
      "grad_norm": 2.609952211380005,
      "learning_rate": 5.5760351218999194e-06,
      "loss": 0.4886,
      "num_input_tokens_seen": 11554128,
      "step": 17535
    },
    {
      "epoch": 10.341981132075471,
      "grad_norm": 3.578115940093994,
      "learning_rate": 5.573479462058379e-06,
      "loss": 0.2445,
      "num_input_tokens_seen": 11556912,
      "step": 17540
    },
    {
      "epoch": 10.34492924528302,
      "grad_norm": 4.661099910736084,
      "learning_rate": 5.570923650385679e-06,
      "loss": 0.3514,
      "num_input_tokens_seen": 11559312,
      "step": 17545
    },
    {
      "epoch": 10.347877358490566,
      "grad_norm": 2.2403664588928223,
      "learning_rate": 5.568367687558481e-06,
      "loss": 0.2918,
      "num_input_tokens_seen": 11562224,
      "step": 17550
    },
    {
      "epoch": 10.350825471698114,
      "grad_norm": 7.799225330352783,
      "learning_rate": 5.56581157425349e-06,
      "loss": 0.3667,
      "num_input_tokens_seen": 11564784,
      "step": 17555
    },
    {
      "epoch": 10.35377358490566,
      "grad_norm": 4.841939449310303,
      "learning_rate": 5.563255311147446e-06,
      "loss": 0.3993,
      "num_input_tokens_seen": 11568336,
      "step": 17560
    },
    {
      "epoch": 10.356721698113208,
      "grad_norm": 2.9595348834991455,
      "learning_rate": 5.560698898917129e-06,
      "loss": 0.2758,
      "num_input_tokens_seen": 11571568,
      "step": 17565
    },
    {
      "epoch": 10.359669811320755,
      "grad_norm": 2.2423505783081055,
      "learning_rate": 5.558142338239365e-06,
      "loss": 0.3208,
      "num_input_tokens_seen": 11576816,
      "step": 17570
    },
    {
      "epoch": 10.362617924528301,
      "grad_norm": 5.600538730621338,
      "learning_rate": 5.55558562979101e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 11579056,
      "step": 17575
    },
    {
      "epoch": 10.36556603773585,
      "grad_norm": 2.6588759422302246,
      "learning_rate": 5.553028774248964e-06,
      "loss": 0.4172,
      "num_input_tokens_seen": 11582480,
      "step": 17580
    },
    {
      "epoch": 10.368514150943396,
      "grad_norm": 3.6319053173065186,
      "learning_rate": 5.5504717722901665e-06,
      "loss": 0.4336,
      "num_input_tokens_seen": 11588304,
      "step": 17585
    },
    {
      "epoch": 10.371462264150944,
      "grad_norm": 3.4403176307678223,
      "learning_rate": 5.547914624591597e-06,
      "loss": 0.4142,
      "num_input_tokens_seen": 11592624,
      "step": 17590
    },
    {
      "epoch": 10.37441037735849,
      "grad_norm": 5.27024507522583,
      "learning_rate": 5.545357331830269e-06,
      "loss": 0.3614,
      "num_input_tokens_seen": 11595856,
      "step": 17595
    },
    {
      "epoch": 10.377358490566039,
      "grad_norm": 7.9467549324035645,
      "learning_rate": 5.542799894683235e-06,
      "loss": 0.3253,
      "num_input_tokens_seen": 11598000,
      "step": 17600
    },
    {
      "epoch": 10.380306603773585,
      "grad_norm": 1.6432124376296997,
      "learning_rate": 5.540242313827591e-06,
      "loss": 0.3058,
      "num_input_tokens_seen": 11602192,
      "step": 17605
    },
    {
      "epoch": 10.383254716981131,
      "grad_norm": 2.6428675651550293,
      "learning_rate": 5.537684589940466e-06,
      "loss": 0.2837,
      "num_input_tokens_seen": 11605712,
      "step": 17610
    },
    {
      "epoch": 10.38620283018868,
      "grad_norm": 1.6663713455200195,
      "learning_rate": 5.535126723699029e-06,
      "loss": 0.4152,
      "num_input_tokens_seen": 11609328,
      "step": 17615
    },
    {
      "epoch": 10.389150943396226,
      "grad_norm": 1.7129652500152588,
      "learning_rate": 5.532568715780485e-06,
      "loss": 0.3895,
      "num_input_tokens_seen": 11612880,
      "step": 17620
    },
    {
      "epoch": 10.392099056603774,
      "grad_norm": 1.973809003829956,
      "learning_rate": 5.530010566862077e-06,
      "loss": 0.4914,
      "num_input_tokens_seen": 11616976,
      "step": 17625
    },
    {
      "epoch": 10.39504716981132,
      "grad_norm": 2.405604600906372,
      "learning_rate": 5.527452277621089e-06,
      "loss": 0.3408,
      "num_input_tokens_seen": 11619696,
      "step": 17630
    },
    {
      "epoch": 10.397995283018869,
      "grad_norm": 4.192343711853027,
      "learning_rate": 5.524893848734837e-06,
      "loss": 0.2638,
      "num_input_tokens_seen": 11622384,
      "step": 17635
    },
    {
      "epoch": 10.400943396226415,
      "grad_norm": 2.72299861907959,
      "learning_rate": 5.522335280880676e-06,
      "loss": 0.3031,
      "num_input_tokens_seen": 11627024,
      "step": 17640
    },
    {
      "epoch": 10.403891509433961,
      "grad_norm": 2.3101727962493896,
      "learning_rate": 5.519776574735999e-06,
      "loss": 0.3545,
      "num_input_tokens_seen": 11629616,
      "step": 17645
    },
    {
      "epoch": 10.40683962264151,
      "grad_norm": 3.497653007507324,
      "learning_rate": 5.5172177309782325e-06,
      "loss": 0.3726,
      "num_input_tokens_seen": 11632272,
      "step": 17650
    },
    {
      "epoch": 10.409787735849056,
      "grad_norm": 2.794992208480835,
      "learning_rate": 5.514658750284844e-06,
      "loss": 0.4201,
      "num_input_tokens_seen": 11635632,
      "step": 17655
    },
    {
      "epoch": 10.412735849056604,
      "grad_norm": 4.138739585876465,
      "learning_rate": 5.512099633333332e-06,
      "loss": 0.5319,
      "num_input_tokens_seen": 11642224,
      "step": 17660
    },
    {
      "epoch": 10.41568396226415,
      "grad_norm": 2.6134753227233887,
      "learning_rate": 5.509540380801236e-06,
      "loss": 0.3848,
      "num_input_tokens_seen": 11644304,
      "step": 17665
    },
    {
      "epoch": 10.418632075471699,
      "grad_norm": 2.6758322715759277,
      "learning_rate": 5.506980993366129e-06,
      "loss": 0.2432,
      "num_input_tokens_seen": 11647536,
      "step": 17670
    },
    {
      "epoch": 10.421580188679245,
      "grad_norm": 3.0789496898651123,
      "learning_rate": 5.504421471705616e-06,
      "loss": 0.5112,
      "num_input_tokens_seen": 11650576,
      "step": 17675
    },
    {
      "epoch": 10.424528301886792,
      "grad_norm": 3.515486478805542,
      "learning_rate": 5.501861816497346e-06,
      "loss": 0.3255,
      "num_input_tokens_seen": 11653584,
      "step": 17680
    },
    {
      "epoch": 10.42747641509434,
      "grad_norm": 2.509601354598999,
      "learning_rate": 5.499302028418998e-06,
      "loss": 0.3423,
      "num_input_tokens_seen": 11657040,
      "step": 17685
    },
    {
      "epoch": 10.430424528301886,
      "grad_norm": 2.161174774169922,
      "learning_rate": 5.496742108148285e-06,
      "loss": 0.3767,
      "num_input_tokens_seen": 11660208,
      "step": 17690
    },
    {
      "epoch": 10.433372641509434,
      "grad_norm": 3.7319185733795166,
      "learning_rate": 5.494182056362959e-06,
      "loss": 0.3984,
      "num_input_tokens_seen": 11663248,
      "step": 17695
    },
    {
      "epoch": 10.43632075471698,
      "grad_norm": 6.046995639801025,
      "learning_rate": 5.491621873740804e-06,
      "loss": 0.3888,
      "num_input_tokens_seen": 11666224,
      "step": 17700
    },
    {
      "epoch": 10.439268867924529,
      "grad_norm": 2.422039031982422,
      "learning_rate": 5.4890615609596375e-06,
      "loss": 0.3159,
      "num_input_tokens_seen": 11669616,
      "step": 17705
    },
    {
      "epoch": 10.442216981132075,
      "grad_norm": 2.167799711227417,
      "learning_rate": 5.486501118697317e-06,
      "loss": 0.3419,
      "num_input_tokens_seen": 11673488,
      "step": 17710
    },
    {
      "epoch": 10.445165094339623,
      "grad_norm": 3.5396533012390137,
      "learning_rate": 5.483940547631727e-06,
      "loss": 0.3849,
      "num_input_tokens_seen": 11675824,
      "step": 17715
    },
    {
      "epoch": 10.44811320754717,
      "grad_norm": 3.40950870513916,
      "learning_rate": 5.481379848440792e-06,
      "loss": 0.4163,
      "num_input_tokens_seen": 11678960,
      "step": 17720
    },
    {
      "epoch": 10.451061320754716,
      "grad_norm": 2.119755268096924,
      "learning_rate": 5.478819021802468e-06,
      "loss": 0.3369,
      "num_input_tokens_seen": 11682480,
      "step": 17725
    },
    {
      "epoch": 10.454009433962264,
      "grad_norm": 3.072749614715576,
      "learning_rate": 5.476258068394743e-06,
      "loss": 0.5086,
      "num_input_tokens_seen": 11685328,
      "step": 17730
    },
    {
      "epoch": 10.45695754716981,
      "grad_norm": 2.2231833934783936,
      "learning_rate": 5.473696988895644e-06,
      "loss": 0.4602,
      "num_input_tokens_seen": 11688720,
      "step": 17735
    },
    {
      "epoch": 10.459905660377359,
      "grad_norm": 2.4915857315063477,
      "learning_rate": 5.471135783983224e-06,
      "loss": 0.3031,
      "num_input_tokens_seen": 11692592,
      "step": 17740
    },
    {
      "epoch": 10.462853773584905,
      "grad_norm": 3.2274670600891113,
      "learning_rate": 5.4685744543355745e-06,
      "loss": 0.4848,
      "num_input_tokens_seen": 11696400,
      "step": 17745
    },
    {
      "epoch": 10.465801886792454,
      "grad_norm": 3.1634624004364014,
      "learning_rate": 5.466013000630819e-06,
      "loss": 0.4125,
      "num_input_tokens_seen": 11699088,
      "step": 17750
    },
    {
      "epoch": 10.46875,
      "grad_norm": 2.430234670639038,
      "learning_rate": 5.463451423547114e-06,
      "loss": 0.4946,
      "num_input_tokens_seen": 11703472,
      "step": 17755
    },
    {
      "epoch": 10.471698113207546,
      "grad_norm": 2.0338642597198486,
      "learning_rate": 5.460889723762647e-06,
      "loss": 0.3876,
      "num_input_tokens_seen": 11706832,
      "step": 17760
    },
    {
      "epoch": 10.474646226415095,
      "grad_norm": 2.3984808921813965,
      "learning_rate": 5.458327901955639e-06,
      "loss": 0.4897,
      "num_input_tokens_seen": 11710416,
      "step": 17765
    },
    {
      "epoch": 10.477594339622641,
      "grad_norm": 6.889331817626953,
      "learning_rate": 5.455765958804344e-06,
      "loss": 0.3524,
      "num_input_tokens_seen": 11712816,
      "step": 17770
    },
    {
      "epoch": 10.48054245283019,
      "grad_norm": 2.5041205883026123,
      "learning_rate": 5.4532038949870455e-06,
      "loss": 0.438,
      "num_input_tokens_seen": 11715440,
      "step": 17775
    },
    {
      "epoch": 10.483490566037736,
      "grad_norm": 3.33964204788208,
      "learning_rate": 5.450641711182066e-06,
      "loss": 0.4489,
      "num_input_tokens_seen": 11718512,
      "step": 17780
    },
    {
      "epoch": 10.486438679245284,
      "grad_norm": 2.3872904777526855,
      "learning_rate": 5.448079408067748e-06,
      "loss": 0.381,
      "num_input_tokens_seen": 11722160,
      "step": 17785
    },
    {
      "epoch": 10.48938679245283,
      "grad_norm": 3.6524198055267334,
      "learning_rate": 5.4455169863224775e-06,
      "loss": 0.3153,
      "num_input_tokens_seen": 11725712,
      "step": 17790
    },
    {
      "epoch": 10.492334905660377,
      "grad_norm": 2.029592514038086,
      "learning_rate": 5.442954446624664e-06,
      "loss": 0.3751,
      "num_input_tokens_seen": 11729264,
      "step": 17795
    },
    {
      "epoch": 10.495283018867925,
      "grad_norm": 1.9283969402313232,
      "learning_rate": 5.440391789652752e-06,
      "loss": 0.4519,
      "num_input_tokens_seen": 11733520,
      "step": 17800
    },
    {
      "epoch": 10.498231132075471,
      "grad_norm": 7.783349990844727,
      "learning_rate": 5.437829016085216e-06,
      "loss": 0.2985,
      "num_input_tokens_seen": 11737904,
      "step": 17805
    },
    {
      "epoch": 10.50117924528302,
      "grad_norm": 2.7163524627685547,
      "learning_rate": 5.435266126600561e-06,
      "loss": 0.3346,
      "num_input_tokens_seen": 11741968,
      "step": 17810
    },
    {
      "epoch": 10.504127358490566,
      "grad_norm": 3.317854881286621,
      "learning_rate": 5.4327031218773215e-06,
      "loss": 0.3188,
      "num_input_tokens_seen": 11745840,
      "step": 17815
    },
    {
      "epoch": 10.507075471698114,
      "grad_norm": 2.2412831783294678,
      "learning_rate": 5.430140002594067e-06,
      "loss": 0.5295,
      "num_input_tokens_seen": 11749264,
      "step": 17820
    },
    {
      "epoch": 10.51002358490566,
      "grad_norm": 3.144230365753174,
      "learning_rate": 5.4275767694293934e-06,
      "loss": 0.3536,
      "num_input_tokens_seen": 11753520,
      "step": 17825
    },
    {
      "epoch": 10.512971698113208,
      "grad_norm": 2.9305832386016846,
      "learning_rate": 5.425013423061926e-06,
      "loss": 0.3533,
      "num_input_tokens_seen": 11756048,
      "step": 17830
    },
    {
      "epoch": 10.515919811320755,
      "grad_norm": 2.5863523483276367,
      "learning_rate": 5.422449964170324e-06,
      "loss": 0.4816,
      "num_input_tokens_seen": 11759344,
      "step": 17835
    },
    {
      "epoch": 10.518867924528301,
      "grad_norm": 2.41121506690979,
      "learning_rate": 5.419886393433275e-06,
      "loss": 0.4331,
      "num_input_tokens_seen": 11763440,
      "step": 17840
    },
    {
      "epoch": 10.52181603773585,
      "grad_norm": 4.899518966674805,
      "learning_rate": 5.417322711529491e-06,
      "loss": 0.3704,
      "num_input_tokens_seen": 11765904,
      "step": 17845
    },
    {
      "epoch": 10.524764150943396,
      "grad_norm": 4.60014533996582,
      "learning_rate": 5.4147589191377224e-06,
      "loss": 0.2673,
      "num_input_tokens_seen": 11768816,
      "step": 17850
    },
    {
      "epoch": 10.527712264150944,
      "grad_norm": 2.5996949672698975,
      "learning_rate": 5.412195016936742e-06,
      "loss": 0.2738,
      "num_input_tokens_seen": 11771536,
      "step": 17855
    },
    {
      "epoch": 10.53066037735849,
      "grad_norm": 2.091630697250366,
      "learning_rate": 5.409631005605354e-06,
      "loss": 0.3413,
      "num_input_tokens_seen": 11774832,
      "step": 17860
    },
    {
      "epoch": 10.533608490566039,
      "grad_norm": 2.4019992351531982,
      "learning_rate": 5.407066885822391e-06,
      "loss": 0.2595,
      "num_input_tokens_seen": 11777776,
      "step": 17865
    },
    {
      "epoch": 10.536556603773585,
      "grad_norm": 3.516969919204712,
      "learning_rate": 5.404502658266717e-06,
      "loss": 0.4319,
      "num_input_tokens_seen": 11780944,
      "step": 17870
    },
    {
      "epoch": 10.539504716981131,
      "grad_norm": 1.751245141029358,
      "learning_rate": 5.4019383236172195e-06,
      "loss": 0.316,
      "num_input_tokens_seen": 11784976,
      "step": 17875
    },
    {
      "epoch": 10.54245283018868,
      "grad_norm": 3.0780744552612305,
      "learning_rate": 5.39937388255282e-06,
      "loss": 0.4316,
      "num_input_tokens_seen": 11787760,
      "step": 17880
    },
    {
      "epoch": 10.545400943396226,
      "grad_norm": 1.6810095310211182,
      "learning_rate": 5.3968093357524645e-06,
      "loss": 0.2893,
      "num_input_tokens_seen": 11790352,
      "step": 17885
    },
    {
      "epoch": 10.548349056603774,
      "grad_norm": 2.38798451423645,
      "learning_rate": 5.3942446838951245e-06,
      "loss": 0.3111,
      "num_input_tokens_seen": 11793200,
      "step": 17890
    },
    {
      "epoch": 10.55129716981132,
      "grad_norm": 3.6871657371520996,
      "learning_rate": 5.3916799276598074e-06,
      "loss": 0.4033,
      "num_input_tokens_seen": 11796848,
      "step": 17895
    },
    {
      "epoch": 10.554245283018869,
      "grad_norm": 1.659034252166748,
      "learning_rate": 5.3891150677255425e-06,
      "loss": 0.3175,
      "num_input_tokens_seen": 11800176,
      "step": 17900
    },
    {
      "epoch": 10.557193396226415,
      "grad_norm": 3.098393440246582,
      "learning_rate": 5.386550104771384e-06,
      "loss": 0.4033,
      "num_input_tokens_seen": 11804368,
      "step": 17905
    },
    {
      "epoch": 10.560141509433961,
      "grad_norm": 3.3002068996429443,
      "learning_rate": 5.3839850394764205e-06,
      "loss": 0.3809,
      "num_input_tokens_seen": 11807760,
      "step": 17910
    },
    {
      "epoch": 10.56308962264151,
      "grad_norm": 4.567578315734863,
      "learning_rate": 5.381419872519763e-06,
      "loss": 0.3815,
      "num_input_tokens_seen": 11810384,
      "step": 17915
    },
    {
      "epoch": 10.566037735849056,
      "grad_norm": 3.6665830612182617,
      "learning_rate": 5.378854604580549e-06,
      "loss": 0.3471,
      "num_input_tokens_seen": 11813072,
      "step": 17920
    },
    {
      "epoch": 10.568985849056604,
      "grad_norm": 3.4409666061401367,
      "learning_rate": 5.376289236337946e-06,
      "loss": 0.42,
      "num_input_tokens_seen": 11815664,
      "step": 17925
    },
    {
      "epoch": 10.57193396226415,
      "grad_norm": 2.6604716777801514,
      "learning_rate": 5.373723768471147e-06,
      "loss": 0.3151,
      "num_input_tokens_seen": 11818544,
      "step": 17930
    },
    {
      "epoch": 10.574882075471699,
      "grad_norm": 2.385392189025879,
      "learning_rate": 5.37115820165937e-06,
      "loss": 0.304,
      "num_input_tokens_seen": 11821264,
      "step": 17935
    },
    {
      "epoch": 10.577830188679245,
      "grad_norm": 3.070085048675537,
      "learning_rate": 5.368592536581858e-06,
      "loss": 0.3656,
      "num_input_tokens_seen": 11824656,
      "step": 17940
    },
    {
      "epoch": 10.580778301886792,
      "grad_norm": 4.700594902038574,
      "learning_rate": 5.366026773917885e-06,
      "loss": 0.3716,
      "num_input_tokens_seen": 11827760,
      "step": 17945
    },
    {
      "epoch": 10.58372641509434,
      "grad_norm": 1.868026614189148,
      "learning_rate": 5.363460914346746e-06,
      "loss": 0.3941,
      "num_input_tokens_seen": 11830000,
      "step": 17950
    },
    {
      "epoch": 10.586674528301886,
      "grad_norm": 2.89733624458313,
      "learning_rate": 5.360894958547762e-06,
      "loss": 0.4464,
      "num_input_tokens_seen": 11833680,
      "step": 17955
    },
    {
      "epoch": 10.589622641509434,
      "grad_norm": 1.8699976205825806,
      "learning_rate": 5.358328907200284e-06,
      "loss": 0.352,
      "num_input_tokens_seen": 11837456,
      "step": 17960
    },
    {
      "epoch": 10.59257075471698,
      "grad_norm": 2.2661190032958984,
      "learning_rate": 5.355762760983682e-06,
      "loss": 0.3683,
      "num_input_tokens_seen": 11840784,
      "step": 17965
    },
    {
      "epoch": 10.595518867924529,
      "grad_norm": 2.1819849014282227,
      "learning_rate": 5.353196520577356e-06,
      "loss": 0.3316,
      "num_input_tokens_seen": 11843280,
      "step": 17970
    },
    {
      "epoch": 10.598466981132075,
      "grad_norm": 3.246781349182129,
      "learning_rate": 5.35063018666073e-06,
      "loss": 0.4579,
      "num_input_tokens_seen": 11846224,
      "step": 17975
    },
    {
      "epoch": 10.601415094339622,
      "grad_norm": 2.40185546875,
      "learning_rate": 5.3480637599132515e-06,
      "loss": 0.2483,
      "num_input_tokens_seen": 11848880,
      "step": 17980
    },
    {
      "epoch": 10.60436320754717,
      "grad_norm": 2.0615992546081543,
      "learning_rate": 5.34549724101439e-06,
      "loss": 0.419,
      "num_input_tokens_seen": 11852528,
      "step": 17985
    },
    {
      "epoch": 10.607311320754716,
      "grad_norm": 3.0229339599609375,
      "learning_rate": 5.342930630643646e-06,
      "loss": 0.4213,
      "num_input_tokens_seen": 11856560,
      "step": 17990
    },
    {
      "epoch": 10.610259433962264,
      "grad_norm": 1.9504142999649048,
      "learning_rate": 5.340363929480541e-06,
      "loss": 0.3263,
      "num_input_tokens_seen": 11860688,
      "step": 17995
    },
    {
      "epoch": 10.61320754716981,
      "grad_norm": 1.9965471029281616,
      "learning_rate": 5.3377971382046164e-06,
      "loss": 0.4248,
      "num_input_tokens_seen": 11866224,
      "step": 18000
    },
    {
      "epoch": 10.616155660377359,
      "grad_norm": 2.3091678619384766,
      "learning_rate": 5.335230257495446e-06,
      "loss": 0.3523,
      "num_input_tokens_seen": 11869008,
      "step": 18005
    },
    {
      "epoch": 10.619103773584905,
      "grad_norm": 2.558356761932373,
      "learning_rate": 5.3326632880326205e-06,
      "loss": 0.3124,
      "num_input_tokens_seen": 11873104,
      "step": 18010
    },
    {
      "epoch": 10.622051886792454,
      "grad_norm": 2.432101249694824,
      "learning_rate": 5.3300962304957515e-06,
      "loss": 0.4119,
      "num_input_tokens_seen": 11877680,
      "step": 18015
    },
    {
      "epoch": 10.625,
      "grad_norm": 3.1192338466644287,
      "learning_rate": 5.327529085564487e-06,
      "loss": 0.3613,
      "num_input_tokens_seen": 11880400,
      "step": 18020
    },
    {
      "epoch": 10.627948113207546,
      "grad_norm": 3.6285626888275146,
      "learning_rate": 5.324961853918485e-06,
      "loss": 0.4057,
      "num_input_tokens_seen": 11884112,
      "step": 18025
    },
    {
      "epoch": 10.630896226415095,
      "grad_norm": 3.5493407249450684,
      "learning_rate": 5.32239453623743e-06,
      "loss": 0.4549,
      "num_input_tokens_seen": 11886928,
      "step": 18030
    },
    {
      "epoch": 10.633844339622641,
      "grad_norm": 2.351642370223999,
      "learning_rate": 5.3198271332010335e-06,
      "loss": 0.3012,
      "num_input_tokens_seen": 11891280,
      "step": 18035
    },
    {
      "epoch": 10.63679245283019,
      "grad_norm": 2.466045618057251,
      "learning_rate": 5.317259645489024e-06,
      "loss": 0.3163,
      "num_input_tokens_seen": 11893840,
      "step": 18040
    },
    {
      "epoch": 10.639740566037736,
      "grad_norm": 2.035175085067749,
      "learning_rate": 5.314692073781157e-06,
      "loss": 0.448,
      "num_input_tokens_seen": 11897552,
      "step": 18045
    },
    {
      "epoch": 10.642688679245284,
      "grad_norm": 2.504878282546997,
      "learning_rate": 5.312124418757207e-06,
      "loss": 0.441,
      "num_input_tokens_seen": 11900688,
      "step": 18050
    },
    {
      "epoch": 10.64563679245283,
      "grad_norm": 2.51212477684021,
      "learning_rate": 5.309556681096972e-06,
      "loss": 0.5207,
      "num_input_tokens_seen": 11904592,
      "step": 18055
    },
    {
      "epoch": 10.648584905660378,
      "grad_norm": 3.7996890544891357,
      "learning_rate": 5.306988861480271e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 11908080,
      "step": 18060
    },
    {
      "epoch": 10.651533018867925,
      "grad_norm": 2.455049753189087,
      "learning_rate": 5.304420960586946e-06,
      "loss": 0.4562,
      "num_input_tokens_seen": 11910800,
      "step": 18065
    },
    {
      "epoch": 10.654481132075471,
      "grad_norm": 4.292541027069092,
      "learning_rate": 5.3018529790968606e-06,
      "loss": 0.4268,
      "num_input_tokens_seen": 11913936,
      "step": 18070
    },
    {
      "epoch": 10.65742924528302,
      "grad_norm": 1.9897202253341675,
      "learning_rate": 5.299284917689898e-06,
      "loss": 0.3427,
      "num_input_tokens_seen": 11917712,
      "step": 18075
    },
    {
      "epoch": 10.660377358490566,
      "grad_norm": 3.3097422122955322,
      "learning_rate": 5.296716777045962e-06,
      "loss": 0.3976,
      "num_input_tokens_seen": 11920752,
      "step": 18080
    },
    {
      "epoch": 10.663325471698114,
      "grad_norm": 4.698294639587402,
      "learning_rate": 5.294148557844983e-06,
      "loss": 0.3997,
      "num_input_tokens_seen": 11923280,
      "step": 18085
    },
    {
      "epoch": 10.66627358490566,
      "grad_norm": 2.4917378425598145,
      "learning_rate": 5.291580260766904e-06,
      "loss": 0.3161,
      "num_input_tokens_seen": 11926896,
      "step": 18090
    },
    {
      "epoch": 10.669221698113208,
      "grad_norm": 2.5687060356140137,
      "learning_rate": 5.289011886491694e-06,
      "loss": 0.3377,
      "num_input_tokens_seen": 11930896,
      "step": 18095
    },
    {
      "epoch": 10.672169811320755,
      "grad_norm": 2.1308069229125977,
      "learning_rate": 5.286443435699342e-06,
      "loss": 0.4421,
      "num_input_tokens_seen": 11935184,
      "step": 18100
    },
    {
      "epoch": 10.675117924528301,
      "grad_norm": 1.906985878944397,
      "learning_rate": 5.283874909069855e-06,
      "loss": 0.2512,
      "num_input_tokens_seen": 11938704,
      "step": 18105
    },
    {
      "epoch": 10.67806603773585,
      "grad_norm": 2.832040548324585,
      "learning_rate": 5.281306307283263e-06,
      "loss": 0.3036,
      "num_input_tokens_seen": 11941456,
      "step": 18110
    },
    {
      "epoch": 10.681014150943396,
      "grad_norm": 2.569648504257202,
      "learning_rate": 5.2787376310196145e-06,
      "loss": 0.4092,
      "num_input_tokens_seen": 11944240,
      "step": 18115
    },
    {
      "epoch": 10.683962264150944,
      "grad_norm": 1.8171806335449219,
      "learning_rate": 5.276168880958977e-06,
      "loss": 0.4398,
      "num_input_tokens_seen": 11947152,
      "step": 18120
    },
    {
      "epoch": 10.68691037735849,
      "grad_norm": 3.035888433456421,
      "learning_rate": 5.273600057781437e-06,
      "loss": 0.4022,
      "num_input_tokens_seen": 11950352,
      "step": 18125
    },
    {
      "epoch": 10.689858490566039,
      "grad_norm": 2.622309446334839,
      "learning_rate": 5.271031162167103e-06,
      "loss": 0.3938,
      "num_input_tokens_seen": 11953680,
      "step": 18130
    },
    {
      "epoch": 10.692806603773585,
      "grad_norm": 2.7524518966674805,
      "learning_rate": 5.268462194796101e-06,
      "loss": 0.4203,
      "num_input_tokens_seen": 11956464,
      "step": 18135
    },
    {
      "epoch": 10.695754716981131,
      "grad_norm": 4.866339683532715,
      "learning_rate": 5.265893156348576e-06,
      "loss": 0.2427,
      "num_input_tokens_seen": 11959312,
      "step": 18140
    },
    {
      "epoch": 10.69870283018868,
      "grad_norm": 2.714456558227539,
      "learning_rate": 5.2633240475046925e-06,
      "loss": 0.3927,
      "num_input_tokens_seen": 11961776,
      "step": 18145
    },
    {
      "epoch": 10.701650943396226,
      "grad_norm": 2.447784900665283,
      "learning_rate": 5.2607548689446305e-06,
      "loss": 0.3577,
      "num_input_tokens_seen": 11965328,
      "step": 18150
    },
    {
      "epoch": 10.704599056603774,
      "grad_norm": 3.8986616134643555,
      "learning_rate": 5.258185621348595e-06,
      "loss": 0.4081,
      "num_input_tokens_seen": 11968816,
      "step": 18155
    },
    {
      "epoch": 10.70754716981132,
      "grad_norm": 3.1010756492614746,
      "learning_rate": 5.255616305396801e-06,
      "loss": 0.3621,
      "num_input_tokens_seen": 11972144,
      "step": 18160
    },
    {
      "epoch": 10.710495283018869,
      "grad_norm": 6.094260215759277,
      "learning_rate": 5.253046921769491e-06,
      "loss": 0.4234,
      "num_input_tokens_seen": 11974896,
      "step": 18165
    },
    {
      "epoch": 10.713443396226415,
      "grad_norm": 3.9047651290893555,
      "learning_rate": 5.250477471146916e-06,
      "loss": 0.273,
      "num_input_tokens_seen": 11977904,
      "step": 18170
    },
    {
      "epoch": 10.716391509433961,
      "grad_norm": 1.6819186210632324,
      "learning_rate": 5.2479079542093535e-06,
      "loss": 0.2999,
      "num_input_tokens_seen": 11981136,
      "step": 18175
    },
    {
      "epoch": 10.71933962264151,
      "grad_norm": 1.8307020664215088,
      "learning_rate": 5.245338371637091e-06,
      "loss": 0.3352,
      "num_input_tokens_seen": 11984976,
      "step": 18180
    },
    {
      "epoch": 10.722287735849056,
      "grad_norm": 3.477308511734009,
      "learning_rate": 5.242768724110437e-06,
      "loss": 0.3329,
      "num_input_tokens_seen": 11987856,
      "step": 18185
    },
    {
      "epoch": 10.725235849056604,
      "grad_norm": 2.77197527885437,
      "learning_rate": 5.240199012309717e-06,
      "loss": 0.4604,
      "num_input_tokens_seen": 11990416,
      "step": 18190
    },
    {
      "epoch": 10.72818396226415,
      "grad_norm": 4.00085973739624,
      "learning_rate": 5.237629236915273e-06,
      "loss": 0.4422,
      "num_input_tokens_seen": 11993392,
      "step": 18195
    },
    {
      "epoch": 10.731132075471699,
      "grad_norm": 3.709183692932129,
      "learning_rate": 5.235059398607464e-06,
      "loss": 0.3743,
      "num_input_tokens_seen": 11996144,
      "step": 18200
    },
    {
      "epoch": 10.734080188679245,
      "grad_norm": 3.6735222339630127,
      "learning_rate": 5.232489498066665e-06,
      "loss": 0.4656,
      "num_input_tokens_seen": 11999312,
      "step": 18205
    },
    {
      "epoch": 10.737028301886792,
      "grad_norm": 3.2352261543273926,
      "learning_rate": 5.229919535973272e-06,
      "loss": 0.434,
      "num_input_tokens_seen": 12002128,
      "step": 18210
    },
    {
      "epoch": 10.73997641509434,
      "grad_norm": 2.7217581272125244,
      "learning_rate": 5.2273495130076905e-06,
      "loss": 0.2888,
      "num_input_tokens_seen": 12006000,
      "step": 18215
    },
    {
      "epoch": 10.742924528301886,
      "grad_norm": 2.013572931289673,
      "learning_rate": 5.224779429850344e-06,
      "loss": 0.3708,
      "num_input_tokens_seen": 12009072,
      "step": 18220
    },
    {
      "epoch": 10.745872641509434,
      "grad_norm": 5.004072666168213,
      "learning_rate": 5.222209287181677e-06,
      "loss": 0.5322,
      "num_input_tokens_seen": 12011856,
      "step": 18225
    },
    {
      "epoch": 10.74882075471698,
      "grad_norm": 2.6996967792510986,
      "learning_rate": 5.219639085682142e-06,
      "loss": 0.4388,
      "num_input_tokens_seen": 12014512,
      "step": 18230
    },
    {
      "epoch": 10.751768867924529,
      "grad_norm": 3.390348196029663,
      "learning_rate": 5.2170688260322124e-06,
      "loss": 0.3748,
      "num_input_tokens_seen": 12017520,
      "step": 18235
    },
    {
      "epoch": 10.754716981132075,
      "grad_norm": 2.863145112991333,
      "learning_rate": 5.214498508912376e-06,
      "loss": 0.3056,
      "num_input_tokens_seen": 12020336,
      "step": 18240
    },
    {
      "epoch": 10.757665094339622,
      "grad_norm": 3.240400552749634,
      "learning_rate": 5.211928135003135e-06,
      "loss": 0.3678,
      "num_input_tokens_seen": 12024240,
      "step": 18245
    },
    {
      "epoch": 10.76061320754717,
      "grad_norm": 3.3532252311706543,
      "learning_rate": 5.209357704985007e-06,
      "loss": 0.5466,
      "num_input_tokens_seen": 12027248,
      "step": 18250
    },
    {
      "epoch": 10.763561320754716,
      "grad_norm": 2.443021059036255,
      "learning_rate": 5.206787219538524e-06,
      "loss": 0.4075,
      "num_input_tokens_seen": 12030320,
      "step": 18255
    },
    {
      "epoch": 10.766509433962264,
      "grad_norm": 2.149935007095337,
      "learning_rate": 5.204216679344234e-06,
      "loss": 0.4164,
      "num_input_tokens_seen": 12034672,
      "step": 18260
    },
    {
      "epoch": 10.76945754716981,
      "grad_norm": 2.047245979309082,
      "learning_rate": 5.201646085082696e-06,
      "loss": 0.339,
      "num_input_tokens_seen": 12038032,
      "step": 18265
    },
    {
      "epoch": 10.772405660377359,
      "grad_norm": 1.989140510559082,
      "learning_rate": 5.199075437434491e-06,
      "loss": 0.4487,
      "num_input_tokens_seen": 12041552,
      "step": 18270
    },
    {
      "epoch": 10.775353773584905,
      "grad_norm": 3.122041940689087,
      "learning_rate": 5.1965047370802046e-06,
      "loss": 0.3815,
      "num_input_tokens_seen": 12044752,
      "step": 18275
    },
    {
      "epoch": 10.778301886792454,
      "grad_norm": 2.7195041179656982,
      "learning_rate": 5.193933984700441e-06,
      "loss": 0.3775,
      "num_input_tokens_seen": 12048944,
      "step": 18280
    },
    {
      "epoch": 10.78125,
      "grad_norm": 1.9588921070098877,
      "learning_rate": 5.1913631809758216e-06,
      "loss": 0.4895,
      "num_input_tokens_seen": 12052240,
      "step": 18285
    },
    {
      "epoch": 10.784198113207546,
      "grad_norm": 1.7437593936920166,
      "learning_rate": 5.188792326586973e-06,
      "loss": 0.4539,
      "num_input_tokens_seen": 12054736,
      "step": 18290
    },
    {
      "epoch": 10.787146226415095,
      "grad_norm": 3.1937031745910645,
      "learning_rate": 5.186221422214544e-06,
      "loss": 0.5651,
      "num_input_tokens_seen": 12058480,
      "step": 18295
    },
    {
      "epoch": 10.790094339622641,
      "grad_norm": 4.076644420623779,
      "learning_rate": 5.1836504685391885e-06,
      "loss": 0.3826,
      "num_input_tokens_seen": 12061168,
      "step": 18300
    },
    {
      "epoch": 10.79304245283019,
      "grad_norm": 3.150028705596924,
      "learning_rate": 5.181079466241582e-06,
      "loss": 0.2994,
      "num_input_tokens_seen": 12064048,
      "step": 18305
    },
    {
      "epoch": 10.795990566037736,
      "grad_norm": 3.371492862701416,
      "learning_rate": 5.178508416002406e-06,
      "loss": 0.3996,
      "num_input_tokens_seen": 12067536,
      "step": 18310
    },
    {
      "epoch": 10.798938679245284,
      "grad_norm": 2.1302404403686523,
      "learning_rate": 5.175937318502357e-06,
      "loss": 0.3335,
      "num_input_tokens_seen": 12070544,
      "step": 18315
    },
    {
      "epoch": 10.80188679245283,
      "grad_norm": 3.5495474338531494,
      "learning_rate": 5.173366174422147e-06,
      "loss": 0.3019,
      "num_input_tokens_seen": 12073168,
      "step": 18320
    },
    {
      "epoch": 10.804834905660378,
      "grad_norm": 3.0108745098114014,
      "learning_rate": 5.170794984442492e-06,
      "loss": 0.3791,
      "num_input_tokens_seen": 12077168,
      "step": 18325
    },
    {
      "epoch": 10.807783018867925,
      "grad_norm": 2.591184616088867,
      "learning_rate": 5.16822374924413e-06,
      "loss": 0.3573,
      "num_input_tokens_seen": 12080048,
      "step": 18330
    },
    {
      "epoch": 10.810731132075471,
      "grad_norm": 3.757197856903076,
      "learning_rate": 5.165652469507806e-06,
      "loss": 0.4951,
      "num_input_tokens_seen": 12082256,
      "step": 18335
    },
    {
      "epoch": 10.81367924528302,
      "grad_norm": 3.2827439308166504,
      "learning_rate": 5.163081145914276e-06,
      "loss": 0.2989,
      "num_input_tokens_seen": 12084336,
      "step": 18340
    },
    {
      "epoch": 10.816627358490566,
      "grad_norm": 3.1096506118774414,
      "learning_rate": 5.160509779144311e-06,
      "loss": 0.3409,
      "num_input_tokens_seen": 12087664,
      "step": 18345
    },
    {
      "epoch": 10.819575471698114,
      "grad_norm": 5.600069522857666,
      "learning_rate": 5.157938369878688e-06,
      "loss": 0.2602,
      "num_input_tokens_seen": 12090128,
      "step": 18350
    },
    {
      "epoch": 10.82252358490566,
      "grad_norm": 3.205955743789673,
      "learning_rate": 5.155366918798203e-06,
      "loss": 0.4003,
      "num_input_tokens_seen": 12093488,
      "step": 18355
    },
    {
      "epoch": 10.825471698113208,
      "grad_norm": 2.8583409786224365,
      "learning_rate": 5.152795426583654e-06,
      "loss": 0.279,
      "num_input_tokens_seen": 12097520,
      "step": 18360
    },
    {
      "epoch": 10.828419811320755,
      "grad_norm": 1.729613184928894,
      "learning_rate": 5.15022389391586e-06,
      "loss": 0.4818,
      "num_input_tokens_seen": 12100944,
      "step": 18365
    },
    {
      "epoch": 10.831367924528301,
      "grad_norm": 1.774673581123352,
      "learning_rate": 5.147652321475642e-06,
      "loss": 0.2197,
      "num_input_tokens_seen": 12103792,
      "step": 18370
    },
    {
      "epoch": 10.83431603773585,
      "grad_norm": 2.641188621520996,
      "learning_rate": 5.145080709943835e-06,
      "loss": 0.4992,
      "num_input_tokens_seen": 12106832,
      "step": 18375
    },
    {
      "epoch": 10.837264150943396,
      "grad_norm": 1.8763964176177979,
      "learning_rate": 5.142509060001285e-06,
      "loss": 0.3831,
      "num_input_tokens_seen": 12110064,
      "step": 18380
    },
    {
      "epoch": 10.840212264150944,
      "grad_norm": 1.714513897895813,
      "learning_rate": 5.139937372328847e-06,
      "loss": 0.4131,
      "num_input_tokens_seen": 12114032,
      "step": 18385
    },
    {
      "epoch": 10.84316037735849,
      "grad_norm": 2.4243905544281006,
      "learning_rate": 5.1373656476073876e-06,
      "loss": 0.3528,
      "num_input_tokens_seen": 12117328,
      "step": 18390
    },
    {
      "epoch": 10.846108490566039,
      "grad_norm": 4.313455581665039,
      "learning_rate": 5.134793886517779e-06,
      "loss": 0.5094,
      "num_input_tokens_seen": 12121040,
      "step": 18395
    },
    {
      "epoch": 10.849056603773585,
      "grad_norm": 2.502354383468628,
      "learning_rate": 5.1322220897409105e-06,
      "loss": 0.2512,
      "num_input_tokens_seen": 12123824,
      "step": 18400
    },
    {
      "epoch": 10.852004716981131,
      "grad_norm": 5.059865951538086,
      "learning_rate": 5.129650257957671e-06,
      "loss": 0.3785,
      "num_input_tokens_seen": 12126960,
      "step": 18405
    },
    {
      "epoch": 10.85495283018868,
      "grad_norm": 3.5637309551239014,
      "learning_rate": 5.12707839184897e-06,
      "loss": 0.3194,
      "num_input_tokens_seen": 12130096,
      "step": 18410
    },
    {
      "epoch": 10.857900943396226,
      "grad_norm": 3.733160972595215,
      "learning_rate": 5.124506492095716e-06,
      "loss": 0.3717,
      "num_input_tokens_seen": 12132848,
      "step": 18415
    },
    {
      "epoch": 10.860849056603774,
      "grad_norm": 1.545000672340393,
      "learning_rate": 5.121934559378831e-06,
      "loss": 0.2901,
      "num_input_tokens_seen": 12136304,
      "step": 18420
    },
    {
      "epoch": 10.86379716981132,
      "grad_norm": 1.9998152256011963,
      "learning_rate": 5.1193625943792456e-06,
      "loss": 0.3664,
      "num_input_tokens_seen": 12138960,
      "step": 18425
    },
    {
      "epoch": 10.866745283018869,
      "grad_norm": 3.435150146484375,
      "learning_rate": 5.116790597777901e-06,
      "loss": 0.3169,
      "num_input_tokens_seen": 12141808,
      "step": 18430
    },
    {
      "epoch": 10.869693396226415,
      "grad_norm": 1.9358536005020142,
      "learning_rate": 5.11421857025574e-06,
      "loss": 0.375,
      "num_input_tokens_seen": 12145264,
      "step": 18435
    },
    {
      "epoch": 10.872641509433961,
      "grad_norm": 3.2511627674102783,
      "learning_rate": 5.111646512493721e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 12148496,
      "step": 18440
    },
    {
      "epoch": 10.87558962264151,
      "grad_norm": 2.862417697906494,
      "learning_rate": 5.109074425172806e-06,
      "loss": 0.4905,
      "num_input_tokens_seen": 12151216,
      "step": 18445
    },
    {
      "epoch": 10.878537735849056,
      "grad_norm": 1.9851090908050537,
      "learning_rate": 5.106502308973967e-06,
      "loss": 0.4005,
      "num_input_tokens_seen": 12153840,
      "step": 18450
    },
    {
      "epoch": 10.881485849056604,
      "grad_norm": 3.1111514568328857,
      "learning_rate": 5.103930164578184e-06,
      "loss": 0.3567,
      "num_input_tokens_seen": 12158192,
      "step": 18455
    },
    {
      "epoch": 10.88443396226415,
      "grad_norm": 2.6240973472595215,
      "learning_rate": 5.101357992666441e-06,
      "loss": 0.2688,
      "num_input_tokens_seen": 12161328,
      "step": 18460
    },
    {
      "epoch": 10.887382075471699,
      "grad_norm": 2.4614510536193848,
      "learning_rate": 5.098785793919733e-06,
      "loss": 0.3491,
      "num_input_tokens_seen": 12164560,
      "step": 18465
    },
    {
      "epoch": 10.890330188679245,
      "grad_norm": 5.559538841247559,
      "learning_rate": 5.096213569019061e-06,
      "loss": 0.4664,
      "num_input_tokens_seen": 12166800,
      "step": 18470
    },
    {
      "epoch": 10.893278301886792,
      "grad_norm": 6.195026874542236,
      "learning_rate": 5.0936413186454315e-06,
      "loss": 0.358,
      "num_input_tokens_seen": 12169232,
      "step": 18475
    },
    {
      "epoch": 10.89622641509434,
      "grad_norm": 2.6421267986297607,
      "learning_rate": 5.0910690434798584e-06,
      "loss": 0.3205,
      "num_input_tokens_seen": 12172496,
      "step": 18480
    },
    {
      "epoch": 10.899174528301886,
      "grad_norm": 3.6238577365875244,
      "learning_rate": 5.088496744203364e-06,
      "loss": 0.4342,
      "num_input_tokens_seen": 12175216,
      "step": 18485
    },
    {
      "epoch": 10.902122641509434,
      "grad_norm": 2.561020612716675,
      "learning_rate": 5.085924421496976e-06,
      "loss": 0.354,
      "num_input_tokens_seen": 12181552,
      "step": 18490
    },
    {
      "epoch": 10.90507075471698,
      "grad_norm": 2.7466835975646973,
      "learning_rate": 5.083352076041725e-06,
      "loss": 0.3358,
      "num_input_tokens_seen": 12187632,
      "step": 18495
    },
    {
      "epoch": 10.908018867924529,
      "grad_norm": 3.319230556488037,
      "learning_rate": 5.080779708518654e-06,
      "loss": 0.3678,
      "num_input_tokens_seen": 12189904,
      "step": 18500
    },
    {
      "epoch": 10.910966981132075,
      "grad_norm": 2.543335437774658,
      "learning_rate": 5.078207319608807e-06,
      "loss": 0.4194,
      "num_input_tokens_seen": 12193904,
      "step": 18505
    },
    {
      "epoch": 10.913915094339622,
      "grad_norm": 2.4117026329040527,
      "learning_rate": 5.075634909993235e-06,
      "loss": 0.377,
      "num_input_tokens_seen": 12197264,
      "step": 18510
    },
    {
      "epoch": 10.91686320754717,
      "grad_norm": 3.913832902908325,
      "learning_rate": 5.073062480352995e-06,
      "loss": 0.3608,
      "num_input_tokens_seen": 12199856,
      "step": 18515
    },
    {
      "epoch": 10.919811320754716,
      "grad_norm": 2.4989800453186035,
      "learning_rate": 5.070490031369149e-06,
      "loss": 0.2692,
      "num_input_tokens_seen": 12204496,
      "step": 18520
    },
    {
      "epoch": 10.922759433962264,
      "grad_norm": 2.049950122833252,
      "learning_rate": 5.067917563722762e-06,
      "loss": 0.3354,
      "num_input_tokens_seen": 12207184,
      "step": 18525
    },
    {
      "epoch": 10.92570754716981,
      "grad_norm": 6.723957061767578,
      "learning_rate": 5.065345078094907e-06,
      "loss": 0.3568,
      "num_input_tokens_seen": 12210576,
      "step": 18530
    },
    {
      "epoch": 10.928655660377359,
      "grad_norm": 2.202289342880249,
      "learning_rate": 5.062772575166663e-06,
      "loss": 0.3528,
      "num_input_tokens_seen": 12213552,
      "step": 18535
    },
    {
      "epoch": 10.931603773584905,
      "grad_norm": 4.398038387298584,
      "learning_rate": 5.0602000556191075e-06,
      "loss": 0.4683,
      "num_input_tokens_seen": 12216208,
      "step": 18540
    },
    {
      "epoch": 10.934551886792454,
      "grad_norm": 3.060457944869995,
      "learning_rate": 5.0576275201333284e-06,
      "loss": 0.3367,
      "num_input_tokens_seen": 12220016,
      "step": 18545
    },
    {
      "epoch": 10.9375,
      "grad_norm": 2.016045331954956,
      "learning_rate": 5.055054969390415e-06,
      "loss": 0.3848,
      "num_input_tokens_seen": 12223408,
      "step": 18550
    },
    {
      "epoch": 10.940448113207546,
      "grad_norm": 7.037094593048096,
      "learning_rate": 5.052482404071461e-06,
      "loss": 0.3385,
      "num_input_tokens_seen": 12225712,
      "step": 18555
    },
    {
      "epoch": 10.943396226415095,
      "grad_norm": 2.3657450675964355,
      "learning_rate": 5.049909824857564e-06,
      "loss": 0.3845,
      "num_input_tokens_seen": 12229744,
      "step": 18560
    },
    {
      "epoch": 10.946344339622641,
      "grad_norm": 6.284314155578613,
      "learning_rate": 5.047337232429827e-06,
      "loss": 0.3439,
      "num_input_tokens_seen": 12232432,
      "step": 18565
    },
    {
      "epoch": 10.94929245283019,
      "grad_norm": 3.701519012451172,
      "learning_rate": 5.044764627469354e-06,
      "loss": 0.4077,
      "num_input_tokens_seen": 12235248,
      "step": 18570
    },
    {
      "epoch": 10.952240566037736,
      "grad_norm": 3.4963715076446533,
      "learning_rate": 5.042192010657251e-06,
      "loss": 0.3595,
      "num_input_tokens_seen": 12238160,
      "step": 18575
    },
    {
      "epoch": 10.955188679245284,
      "grad_norm": 3.388214111328125,
      "learning_rate": 5.039619382674632e-06,
      "loss": 0.2975,
      "num_input_tokens_seen": 12242000,
      "step": 18580
    },
    {
      "epoch": 10.95813679245283,
      "grad_norm": 4.11798095703125,
      "learning_rate": 5.0370467442026115e-06,
      "loss": 0.4315,
      "num_input_tokens_seen": 12244720,
      "step": 18585
    },
    {
      "epoch": 10.961084905660378,
      "grad_norm": 4.732944488525391,
      "learning_rate": 5.034474095922304e-06,
      "loss": 0.391,
      "num_input_tokens_seen": 12247344,
      "step": 18590
    },
    {
      "epoch": 10.964033018867925,
      "grad_norm": 2.123810052871704,
      "learning_rate": 5.031901438514832e-06,
      "loss": 0.3118,
      "num_input_tokens_seen": 12250928,
      "step": 18595
    },
    {
      "epoch": 10.966981132075471,
      "grad_norm": 2.176208972930908,
      "learning_rate": 5.0293287726613185e-06,
      "loss": 0.3918,
      "num_input_tokens_seen": 12253808,
      "step": 18600
    },
    {
      "epoch": 10.96992924528302,
      "grad_norm": 4.267393112182617,
      "learning_rate": 5.0267560990428836e-06,
      "loss": 0.2938,
      "num_input_tokens_seen": 12257296,
      "step": 18605
    },
    {
      "epoch": 10.972877358490566,
      "grad_norm": 1.7519735097885132,
      "learning_rate": 5.024183418340657e-06,
      "loss": 0.4132,
      "num_input_tokens_seen": 12260560,
      "step": 18610
    },
    {
      "epoch": 10.975825471698114,
      "grad_norm": 4.019475936889648,
      "learning_rate": 5.021610731235766e-06,
      "loss": 0.3085,
      "num_input_tokens_seen": 12262608,
      "step": 18615
    },
    {
      "epoch": 10.97877358490566,
      "grad_norm": 2.426175117492676,
      "learning_rate": 5.01903803840934e-06,
      "loss": 0.3083,
      "num_input_tokens_seen": 12265744,
      "step": 18620
    },
    {
      "epoch": 10.981721698113208,
      "grad_norm": 2.2213268280029297,
      "learning_rate": 5.016465340542514e-06,
      "loss": 0.3335,
      "num_input_tokens_seen": 12269424,
      "step": 18625
    },
    {
      "epoch": 10.984669811320755,
      "grad_norm": 3.617539167404175,
      "learning_rate": 5.013892638316417e-06,
      "loss": 0.5115,
      "num_input_tokens_seen": 12272016,
      "step": 18630
    },
    {
      "epoch": 10.987617924528301,
      "grad_norm": 2.3333542346954346,
      "learning_rate": 5.011319932412182e-06,
      "loss": 0.3576,
      "num_input_tokens_seen": 12275152,
      "step": 18635
    },
    {
      "epoch": 10.99056603773585,
      "grad_norm": 2.329491376876831,
      "learning_rate": 5.008747223510947e-06,
      "loss": 0.4238,
      "num_input_tokens_seen": 12278576,
      "step": 18640
    },
    {
      "epoch": 10.993514150943396,
      "grad_norm": 3.317375421524048,
      "learning_rate": 5.006174512293849e-06,
      "loss": 0.5779,
      "num_input_tokens_seen": 12282000,
      "step": 18645
    },
    {
      "epoch": 10.996462264150944,
      "grad_norm": 1.8646942377090454,
      "learning_rate": 5.003601799442019e-06,
      "loss": 0.5393,
      "num_input_tokens_seen": 12285712,
      "step": 18650
    },
    {
      "epoch": 10.99941037735849,
      "grad_norm": 2.3972671031951904,
      "learning_rate": 5.0010290856366e-06,
      "loss": 0.235,
      "num_input_tokens_seen": 12288784,
      "step": 18655
    },
    {
      "epoch": 11.002358490566039,
      "grad_norm": 3.3607842922210693,
      "learning_rate": 4.998456371558726e-06,
      "loss": 0.4223,
      "num_input_tokens_seen": 12292104,
      "step": 18660
    },
    {
      "epoch": 11.005306603773585,
      "grad_norm": 2.398604393005371,
      "learning_rate": 4.995883657889531e-06,
      "loss": 0.383,
      "num_input_tokens_seen": 12295528,
      "step": 18665
    },
    {
      "epoch": 11.008254716981131,
      "grad_norm": 3.345334053039551,
      "learning_rate": 4.993310945310158e-06,
      "loss": 0.325,
      "num_input_tokens_seen": 12298824,
      "step": 18670
    },
    {
      "epoch": 11.01120283018868,
      "grad_norm": 2.150770664215088,
      "learning_rate": 4.9907382345017416e-06,
      "loss": 0.3799,
      "num_input_tokens_seen": 12303240,
      "step": 18675
    },
    {
      "epoch": 11.014150943396226,
      "grad_norm": 1.824595332145691,
      "learning_rate": 4.988165526145416e-06,
      "loss": 0.362,
      "num_input_tokens_seen": 12308104,
      "step": 18680
    },
    {
      "epoch": 11.017099056603774,
      "grad_norm": 1.657428503036499,
      "learning_rate": 4.985592820922319e-06,
      "loss": 0.3713,
      "num_input_tokens_seen": 12311272,
      "step": 18685
    },
    {
      "epoch": 11.02004716981132,
      "grad_norm": 3.062992572784424,
      "learning_rate": 4.983020119513586e-06,
      "loss": 0.3629,
      "num_input_tokens_seen": 12314024,
      "step": 18690
    },
    {
      "epoch": 11.022995283018869,
      "grad_norm": 3.173238754272461,
      "learning_rate": 4.9804474226003465e-06,
      "loss": 0.4353,
      "num_input_tokens_seen": 12317128,
      "step": 18695
    },
    {
      "epoch": 11.025943396226415,
      "grad_norm": 3.8530690670013428,
      "learning_rate": 4.977874730863739e-06,
      "loss": 0.295,
      "num_input_tokens_seen": 12322600,
      "step": 18700
    },
    {
      "epoch": 11.028891509433961,
      "grad_norm": 2.374476671218872,
      "learning_rate": 4.975302044984889e-06,
      "loss": 0.3279,
      "num_input_tokens_seen": 12325416,
      "step": 18705
    },
    {
      "epoch": 11.03183962264151,
      "grad_norm": 2.0353593826293945,
      "learning_rate": 4.972729365644931e-06,
      "loss": 0.2791,
      "num_input_tokens_seen": 12328168,
      "step": 18710
    },
    {
      "epoch": 11.034787735849056,
      "grad_norm": 2.8604278564453125,
      "learning_rate": 4.97015669352499e-06,
      "loss": 0.4902,
      "num_input_tokens_seen": 12330760,
      "step": 18715
    },
    {
      "epoch": 11.037735849056604,
      "grad_norm": 5.7274580001831055,
      "learning_rate": 4.967584029306194e-06,
      "loss": 0.4364,
      "num_input_tokens_seen": 12333768,
      "step": 18720
    },
    {
      "epoch": 11.04068396226415,
      "grad_norm": 6.029636859893799,
      "learning_rate": 4.965011373669666e-06,
      "loss": 0.3335,
      "num_input_tokens_seen": 12336744,
      "step": 18725
    },
    {
      "epoch": 11.043632075471699,
      "grad_norm": 2.6295735836029053,
      "learning_rate": 4.962438727296527e-06,
      "loss": 0.3831,
      "num_input_tokens_seen": 12340872,
      "step": 18730
    },
    {
      "epoch": 11.046580188679245,
      "grad_norm": 1.9966658353805542,
      "learning_rate": 4.959866090867897e-06,
      "loss": 0.2469,
      "num_input_tokens_seen": 12346216,
      "step": 18735
    },
    {
      "epoch": 11.049528301886792,
      "grad_norm": 3.437544345855713,
      "learning_rate": 4.957293465064893e-06,
      "loss": 0.3019,
      "num_input_tokens_seen": 12348936,
      "step": 18740
    },
    {
      "epoch": 11.05247641509434,
      "grad_norm": 3.9508016109466553,
      "learning_rate": 4.954720850568627e-06,
      "loss": 0.3157,
      "num_input_tokens_seen": 12351816,
      "step": 18745
    },
    {
      "epoch": 11.055424528301886,
      "grad_norm": 5.375389575958252,
      "learning_rate": 4.952148248060212e-06,
      "loss": 0.3397,
      "num_input_tokens_seen": 12354952,
      "step": 18750
    },
    {
      "epoch": 11.058372641509434,
      "grad_norm": 4.616829872131348,
      "learning_rate": 4.949575658220755e-06,
      "loss": 0.4522,
      "num_input_tokens_seen": 12357704,
      "step": 18755
    },
    {
      "epoch": 11.06132075471698,
      "grad_norm": 2.8060903549194336,
      "learning_rate": 4.947003081731359e-06,
      "loss": 0.3697,
      "num_input_tokens_seen": 12361608,
      "step": 18760
    },
    {
      "epoch": 11.064268867924529,
      "grad_norm": 3.2946646213531494,
      "learning_rate": 4.944430519273126e-06,
      "loss": 0.4852,
      "num_input_tokens_seen": 12364744,
      "step": 18765
    },
    {
      "epoch": 11.067216981132075,
      "grad_norm": 1.5486385822296143,
      "learning_rate": 4.941857971527152e-06,
      "loss": 0.3393,
      "num_input_tokens_seen": 12368360,
      "step": 18770
    },
    {
      "epoch": 11.070165094339623,
      "grad_norm": 2.7326748371124268,
      "learning_rate": 4.93928543917453e-06,
      "loss": 0.3092,
      "num_input_tokens_seen": 12371720,
      "step": 18775
    },
    {
      "epoch": 11.07311320754717,
      "grad_norm": 1.8578720092773438,
      "learning_rate": 4.93671292289635e-06,
      "loss": 0.2634,
      "num_input_tokens_seen": 12375688,
      "step": 18780
    },
    {
      "epoch": 11.076061320754716,
      "grad_norm": 2.425044298171997,
      "learning_rate": 4.934140423373698e-06,
      "loss": 0.4472,
      "num_input_tokens_seen": 12378792,
      "step": 18785
    },
    {
      "epoch": 11.079009433962264,
      "grad_norm": 1.7066113948822021,
      "learning_rate": 4.931567941287651e-06,
      "loss": 0.4514,
      "num_input_tokens_seen": 12381736,
      "step": 18790
    },
    {
      "epoch": 11.08195754716981,
      "grad_norm": 4.265979290008545,
      "learning_rate": 4.9289954773192875e-06,
      "loss": 0.3914,
      "num_input_tokens_seen": 12384616,
      "step": 18795
    },
    {
      "epoch": 11.084905660377359,
      "grad_norm": 2.858393430709839,
      "learning_rate": 4.926423032149677e-06,
      "loss": 0.2951,
      "num_input_tokens_seen": 12388488,
      "step": 18800
    },
    {
      "epoch": 11.087853773584905,
      "grad_norm": 2.7882189750671387,
      "learning_rate": 4.923850606459883e-06,
      "loss": 0.3231,
      "num_input_tokens_seen": 12392520,
      "step": 18805
    },
    {
      "epoch": 11.090801886792454,
      "grad_norm": 3.360539197921753,
      "learning_rate": 4.921278200930972e-06,
      "loss": 0.3048,
      "num_input_tokens_seen": 12396360,
      "step": 18810
    },
    {
      "epoch": 11.09375,
      "grad_norm": 3.3079440593719482,
      "learning_rate": 4.918705816243996e-06,
      "loss": 0.4316,
      "num_input_tokens_seen": 12399336,
      "step": 18815
    },
    {
      "epoch": 11.096698113207546,
      "grad_norm": 2.616729259490967,
      "learning_rate": 4.916133453080007e-06,
      "loss": 0.306,
      "num_input_tokens_seen": 12402856,
      "step": 18820
    },
    {
      "epoch": 11.099646226415095,
      "grad_norm": 4.504759311676025,
      "learning_rate": 4.913561112120046e-06,
      "loss": 0.3136,
      "num_input_tokens_seen": 12407880,
      "step": 18825
    },
    {
      "epoch": 11.102594339622641,
      "grad_norm": 2.6495447158813477,
      "learning_rate": 4.910988794045154e-06,
      "loss": 0.3111,
      "num_input_tokens_seen": 12410280,
      "step": 18830
    },
    {
      "epoch": 11.10554245283019,
      "grad_norm": 3.3120932579040527,
      "learning_rate": 4.9084164995363626e-06,
      "loss": 0.351,
      "num_input_tokens_seen": 12413384,
      "step": 18835
    },
    {
      "epoch": 11.108490566037736,
      "grad_norm": 1.973710060119629,
      "learning_rate": 4.905844229274697e-06,
      "loss": 0.3218,
      "num_input_tokens_seen": 12417416,
      "step": 18840
    },
    {
      "epoch": 11.111438679245284,
      "grad_norm": 14.83027172088623,
      "learning_rate": 4.903271983941177e-06,
      "loss": 0.4754,
      "num_input_tokens_seen": 12419688,
      "step": 18845
    },
    {
      "epoch": 11.11438679245283,
      "grad_norm": 2.8816723823547363,
      "learning_rate": 4.900699764216818e-06,
      "loss": 0.3705,
      "num_input_tokens_seen": 12422920,
      "step": 18850
    },
    {
      "epoch": 11.117334905660377,
      "grad_norm": 5.430673599243164,
      "learning_rate": 4.898127570782622e-06,
      "loss": 0.2807,
      "num_input_tokens_seen": 12426408,
      "step": 18855
    },
    {
      "epoch": 11.120283018867925,
      "grad_norm": 2.7598323822021484,
      "learning_rate": 4.895555404319592e-06,
      "loss": 0.413,
      "num_input_tokens_seen": 12430472,
      "step": 18860
    },
    {
      "epoch": 11.123231132075471,
      "grad_norm": 3.7343430519104004,
      "learning_rate": 4.89298326550872e-06,
      "loss": 0.3242,
      "num_input_tokens_seen": 12433576,
      "step": 18865
    },
    {
      "epoch": 11.12617924528302,
      "grad_norm": 3.497344732284546,
      "learning_rate": 4.8904111550309876e-06,
      "loss": 0.4144,
      "num_input_tokens_seen": 12436552,
      "step": 18870
    },
    {
      "epoch": 11.129127358490566,
      "grad_norm": 1.7707535028457642,
      "learning_rate": 4.8878390735673755e-06,
      "loss": 0.2364,
      "num_input_tokens_seen": 12439752,
      "step": 18875
    },
    {
      "epoch": 11.132075471698114,
      "grad_norm": 2.1490697860717773,
      "learning_rate": 4.8852670217988505e-06,
      "loss": 0.3318,
      "num_input_tokens_seen": 12443048,
      "step": 18880
    },
    {
      "epoch": 11.13502358490566,
      "grad_norm": 3.946535348892212,
      "learning_rate": 4.882695000406377e-06,
      "loss": 0.4253,
      "num_input_tokens_seen": 12446376,
      "step": 18885
    },
    {
      "epoch": 11.137971698113208,
      "grad_norm": 2.5758707523345947,
      "learning_rate": 4.880123010070909e-06,
      "loss": 0.3583,
      "num_input_tokens_seen": 12449416,
      "step": 18890
    },
    {
      "epoch": 11.140919811320755,
      "grad_norm": 3.5226688385009766,
      "learning_rate": 4.877551051473388e-06,
      "loss": 0.3632,
      "num_input_tokens_seen": 12453000,
      "step": 18895
    },
    {
      "epoch": 11.143867924528301,
      "grad_norm": 2.022728204727173,
      "learning_rate": 4.874979125294755e-06,
      "loss": 0.319,
      "num_input_tokens_seen": 12456808,
      "step": 18900
    },
    {
      "epoch": 11.14681603773585,
      "grad_norm": 3.551072597503662,
      "learning_rate": 4.872407232215937e-06,
      "loss": 0.3529,
      "num_input_tokens_seen": 12459720,
      "step": 18905
    },
    {
      "epoch": 11.149764150943396,
      "grad_norm": 2.6980831623077393,
      "learning_rate": 4.8698353729178546e-06,
      "loss": 0.4736,
      "num_input_tokens_seen": 12462568,
      "step": 18910
    },
    {
      "epoch": 11.152712264150944,
      "grad_norm": 2.0211150646209717,
      "learning_rate": 4.867263548081418e-06,
      "loss": 0.3913,
      "num_input_tokens_seen": 12465672,
      "step": 18915
    },
    {
      "epoch": 11.15566037735849,
      "grad_norm": 2.5296103954315186,
      "learning_rate": 4.8646917583875304e-06,
      "loss": 0.4973,
      "num_input_tokens_seen": 12469288,
      "step": 18920
    },
    {
      "epoch": 11.158608490566039,
      "grad_norm": 2.053290605545044,
      "learning_rate": 4.862120004517082e-06,
      "loss": 0.4051,
      "num_input_tokens_seen": 12473032,
      "step": 18925
    },
    {
      "epoch": 11.161556603773585,
      "grad_norm": 3.2847726345062256,
      "learning_rate": 4.859548287150956e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 12475304,
      "step": 18930
    },
    {
      "epoch": 11.164504716981131,
      "grad_norm": 2.762706995010376,
      "learning_rate": 4.8569766069700275e-06,
      "loss": 0.2452,
      "num_input_tokens_seen": 12477992,
      "step": 18935
    },
    {
      "epoch": 11.16745283018868,
      "grad_norm": 3.87250018119812,
      "learning_rate": 4.854404964655158e-06,
      "loss": 0.2697,
      "num_input_tokens_seen": 12481384,
      "step": 18940
    },
    {
      "epoch": 11.170400943396226,
      "grad_norm": 3.9831371307373047,
      "learning_rate": 4.8518333608872015e-06,
      "loss": 0.4373,
      "num_input_tokens_seen": 12483848,
      "step": 18945
    },
    {
      "epoch": 11.173349056603774,
      "grad_norm": 2.6348490715026855,
      "learning_rate": 4.849261796347002e-06,
      "loss": 0.2967,
      "num_input_tokens_seen": 12487208,
      "step": 18950
    },
    {
      "epoch": 11.17629716981132,
      "grad_norm": 3.4063611030578613,
      "learning_rate": 4.846690271715391e-06,
      "loss": 0.3472,
      "num_input_tokens_seen": 12489928,
      "step": 18955
    },
    {
      "epoch": 11.179245283018869,
      "grad_norm": 3.0247020721435547,
      "learning_rate": 4.844118787673191e-06,
      "loss": 0.4645,
      "num_input_tokens_seen": 12493000,
      "step": 18960
    },
    {
      "epoch": 11.182193396226415,
      "grad_norm": 3.3598554134368896,
      "learning_rate": 4.841547344901214e-06,
      "loss": 0.5068,
      "num_input_tokens_seen": 12495816,
      "step": 18965
    },
    {
      "epoch": 11.185141509433961,
      "grad_norm": 2.107295513153076,
      "learning_rate": 4.838975944080261e-06,
      "loss": 0.3168,
      "num_input_tokens_seen": 12499656,
      "step": 18970
    },
    {
      "epoch": 11.18808962264151,
      "grad_norm": 2.6018173694610596,
      "learning_rate": 4.83640458589112e-06,
      "loss": 0.2596,
      "num_input_tokens_seen": 12503688,
      "step": 18975
    },
    {
      "epoch": 11.191037735849056,
      "grad_norm": 2.3236682415008545,
      "learning_rate": 4.833833271014571e-06,
      "loss": 0.4329,
      "num_input_tokens_seen": 12505992,
      "step": 18980
    },
    {
      "epoch": 11.193985849056604,
      "grad_norm": 2.6046409606933594,
      "learning_rate": 4.831262000131379e-06,
      "loss": 0.3256,
      "num_input_tokens_seen": 12509544,
      "step": 18985
    },
    {
      "epoch": 11.19693396226415,
      "grad_norm": 2.2731635570526123,
      "learning_rate": 4.828690773922299e-06,
      "loss": 0.3073,
      "num_input_tokens_seen": 12512744,
      "step": 18990
    },
    {
      "epoch": 11.199882075471699,
      "grad_norm": 3.0305395126342773,
      "learning_rate": 4.826119593068074e-06,
      "loss": 0.4185,
      "num_input_tokens_seen": 12518856,
      "step": 18995
    },
    {
      "epoch": 11.202830188679245,
      "grad_norm": 2.426881790161133,
      "learning_rate": 4.8235484582494375e-06,
      "loss": 0.3045,
      "num_input_tokens_seen": 12522120,
      "step": 19000
    },
    {
      "epoch": 11.205778301886792,
      "grad_norm": 1.8929892778396606,
      "learning_rate": 4.8209773701471076e-06,
      "loss": 0.3038,
      "num_input_tokens_seen": 12526056,
      "step": 19005
    },
    {
      "epoch": 11.20872641509434,
      "grad_norm": 2.8837413787841797,
      "learning_rate": 4.818406329441789e-06,
      "loss": 0.4454,
      "num_input_tokens_seen": 12528872,
      "step": 19010
    },
    {
      "epoch": 11.211674528301886,
      "grad_norm": 2.1331777572631836,
      "learning_rate": 4.815835336814179e-06,
      "loss": 0.3581,
      "num_input_tokens_seen": 12531816,
      "step": 19015
    },
    {
      "epoch": 11.214622641509434,
      "grad_norm": 3.154477119445801,
      "learning_rate": 4.813264392944957e-06,
      "loss": 0.3738,
      "num_input_tokens_seen": 12534280,
      "step": 19020
    },
    {
      "epoch": 11.21757075471698,
      "grad_norm": 1.9566569328308105,
      "learning_rate": 4.8106934985147905e-06,
      "loss": 0.3362,
      "num_input_tokens_seen": 12537672,
      "step": 19025
    },
    {
      "epoch": 11.220518867924529,
      "grad_norm": 2.236354112625122,
      "learning_rate": 4.808122654204338e-06,
      "loss": 0.3173,
      "num_input_tokens_seen": 12540936,
      "step": 19030
    },
    {
      "epoch": 11.223466981132075,
      "grad_norm": 3.9648280143737793,
      "learning_rate": 4.805551860694239e-06,
      "loss": 0.3173,
      "num_input_tokens_seen": 12544104,
      "step": 19035
    },
    {
      "epoch": 11.226415094339623,
      "grad_norm": 3.2587320804595947,
      "learning_rate": 4.802981118665122e-06,
      "loss": 0.3687,
      "num_input_tokens_seen": 12546344,
      "step": 19040
    },
    {
      "epoch": 11.22936320754717,
      "grad_norm": 2.6159462928771973,
      "learning_rate": 4.800410428797604e-06,
      "loss": 0.3838,
      "num_input_tokens_seen": 12549992,
      "step": 19045
    },
    {
      "epoch": 11.232311320754716,
      "grad_norm": 3.13700270652771,
      "learning_rate": 4.797839791772286e-06,
      "loss": 0.3666,
      "num_input_tokens_seen": 12553384,
      "step": 19050
    },
    {
      "epoch": 11.235259433962264,
      "grad_norm": 2.1830179691314697,
      "learning_rate": 4.795269208269752e-06,
      "loss": 0.2759,
      "num_input_tokens_seen": 12555976,
      "step": 19055
    },
    {
      "epoch": 11.23820754716981,
      "grad_norm": 2.6924431324005127,
      "learning_rate": 4.792698678970579e-06,
      "loss": 0.2561,
      "num_input_tokens_seen": 12558824,
      "step": 19060
    },
    {
      "epoch": 11.241155660377359,
      "grad_norm": 2.681445837020874,
      "learning_rate": 4.7901282045553245e-06,
      "loss": 0.4508,
      "num_input_tokens_seen": 12561800,
      "step": 19065
    },
    {
      "epoch": 11.244103773584905,
      "grad_norm": 2.2855594158172607,
      "learning_rate": 4.787557785704531e-06,
      "loss": 0.4146,
      "num_input_tokens_seen": 12564328,
      "step": 19070
    },
    {
      "epoch": 11.247051886792454,
      "grad_norm": 3.0382354259490967,
      "learning_rate": 4.784987423098731e-06,
      "loss": 0.4684,
      "num_input_tokens_seen": 12567400,
      "step": 19075
    },
    {
      "epoch": 11.25,
      "grad_norm": 2.422194242477417,
      "learning_rate": 4.7824171174184354e-06,
      "loss": 0.3197,
      "num_input_tokens_seen": 12570792,
      "step": 19080
    },
    {
      "epoch": 11.252948113207546,
      "grad_norm": 1.8935493230819702,
      "learning_rate": 4.779846869344146e-06,
      "loss": 0.2138,
      "num_input_tokens_seen": 12574184,
      "step": 19085
    },
    {
      "epoch": 11.255896226415095,
      "grad_norm": 2.4154887199401855,
      "learning_rate": 4.777276679556346e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 12578376,
      "step": 19090
    },
    {
      "epoch": 11.258844339622641,
      "grad_norm": 2.2390286922454834,
      "learning_rate": 4.774706548735507e-06,
      "loss": 0.3126,
      "num_input_tokens_seen": 12581256,
      "step": 19095
    },
    {
      "epoch": 11.26179245283019,
      "grad_norm": 5.7270002365112305,
      "learning_rate": 4.77213647756208e-06,
      "loss": 0.2921,
      "num_input_tokens_seen": 12584936,
      "step": 19100
    },
    {
      "epoch": 11.264740566037736,
      "grad_norm": 2.4644391536712646,
      "learning_rate": 4.769566466716501e-06,
      "loss": 0.2399,
      "num_input_tokens_seen": 12587944,
      "step": 19105
    },
    {
      "epoch": 11.267688679245284,
      "grad_norm": 2.7339460849761963,
      "learning_rate": 4.766996516879195e-06,
      "loss": 0.2623,
      "num_input_tokens_seen": 12591560,
      "step": 19110
    },
    {
      "epoch": 11.27063679245283,
      "grad_norm": 2.482053279876709,
      "learning_rate": 4.764426628730564e-06,
      "loss": 0.4082,
      "num_input_tokens_seen": 12596200,
      "step": 19115
    },
    {
      "epoch": 11.273584905660377,
      "grad_norm": 1.9033784866333008,
      "learning_rate": 4.761856802950999e-06,
      "loss": 0.4268,
      "num_input_tokens_seen": 12600168,
      "step": 19120
    },
    {
      "epoch": 11.276533018867925,
      "grad_norm": 2.3184351921081543,
      "learning_rate": 4.759287040220872e-06,
      "loss": 0.2966,
      "num_input_tokens_seen": 12603272,
      "step": 19125
    },
    {
      "epoch": 11.279481132075471,
      "grad_norm": 2.177517890930176,
      "learning_rate": 4.756717341220538e-06,
      "loss": 0.2558,
      "num_input_tokens_seen": 12605960,
      "step": 19130
    },
    {
      "epoch": 11.28242924528302,
      "grad_norm": 3.9570810794830322,
      "learning_rate": 4.7541477066303365e-06,
      "loss": 0.3553,
      "num_input_tokens_seen": 12608808,
      "step": 19135
    },
    {
      "epoch": 11.285377358490566,
      "grad_norm": 2.5323946475982666,
      "learning_rate": 4.75157813713059e-06,
      "loss": 0.3085,
      "num_input_tokens_seen": 12612392,
      "step": 19140
    },
    {
      "epoch": 11.288325471698114,
      "grad_norm": 4.13688850402832,
      "learning_rate": 4.7490086334016e-06,
      "loss": 0.3154,
      "num_input_tokens_seen": 12615304,
      "step": 19145
    },
    {
      "epoch": 11.29127358490566,
      "grad_norm": 3.863420248031616,
      "learning_rate": 4.746439196123659e-06,
      "loss": 0.4977,
      "num_input_tokens_seen": 12618536,
      "step": 19150
    },
    {
      "epoch": 11.294221698113208,
      "grad_norm": 3.4435009956359863,
      "learning_rate": 4.743869825977032e-06,
      "loss": 0.3529,
      "num_input_tokens_seen": 12622504,
      "step": 19155
    },
    {
      "epoch": 11.297169811320755,
      "grad_norm": 2.6902849674224854,
      "learning_rate": 4.741300523641972e-06,
      "loss": 0.4793,
      "num_input_tokens_seen": 12626216,
      "step": 19160
    },
    {
      "epoch": 11.300117924528301,
      "grad_norm": 3.484013795852661,
      "learning_rate": 4.738731289798715e-06,
      "loss": 0.3187,
      "num_input_tokens_seen": 12630440,
      "step": 19165
    },
    {
      "epoch": 11.30306603773585,
      "grad_norm": 5.49545955657959,
      "learning_rate": 4.7361621251274744e-06,
      "loss": 0.399,
      "num_input_tokens_seen": 12633384,
      "step": 19170
    },
    {
      "epoch": 11.306014150943396,
      "grad_norm": 2.8632378578186035,
      "learning_rate": 4.733593030308446e-06,
      "loss": 0.2544,
      "num_input_tokens_seen": 12636584,
      "step": 19175
    },
    {
      "epoch": 11.308962264150944,
      "grad_norm": 2.6174468994140625,
      "learning_rate": 4.731024006021814e-06,
      "loss": 0.3944,
      "num_input_tokens_seen": 12640168,
      "step": 19180
    },
    {
      "epoch": 11.31191037735849,
      "grad_norm": 1.7239999771118164,
      "learning_rate": 4.728455052947732e-06,
      "loss": 0.2802,
      "num_input_tokens_seen": 12643592,
      "step": 19185
    },
    {
      "epoch": 11.314858490566039,
      "grad_norm": 2.940037488937378,
      "learning_rate": 4.725886171766349e-06,
      "loss": 0.3276,
      "num_input_tokens_seen": 12647080,
      "step": 19190
    },
    {
      "epoch": 11.317806603773585,
      "grad_norm": 2.9013917446136475,
      "learning_rate": 4.723317363157781e-06,
      "loss": 0.5346,
      "num_input_tokens_seen": 12649800,
      "step": 19195
    },
    {
      "epoch": 11.320754716981131,
      "grad_norm": 2.906733989715576,
      "learning_rate": 4.720748627802135e-06,
      "loss": 0.4227,
      "num_input_tokens_seen": 12652648,
      "step": 19200
    },
    {
      "epoch": 11.32370283018868,
      "grad_norm": 3.986771583557129,
      "learning_rate": 4.718179966379492e-06,
      "loss": 0.3086,
      "num_input_tokens_seen": 12655176,
      "step": 19205
    },
    {
      "epoch": 11.326650943396226,
      "grad_norm": 2.1207046508789062,
      "learning_rate": 4.715611379569919e-06,
      "loss": 0.4626,
      "num_input_tokens_seen": 12658824,
      "step": 19210
    },
    {
      "epoch": 11.329599056603774,
      "grad_norm": 2.673875331878662,
      "learning_rate": 4.713042868053458e-06,
      "loss": 0.3588,
      "num_input_tokens_seen": 12661544,
      "step": 19215
    },
    {
      "epoch": 11.33254716981132,
      "grad_norm": 3.732043504714966,
      "learning_rate": 4.7104744325101345e-06,
      "loss": 0.3438,
      "num_input_tokens_seen": 12663944,
      "step": 19220
    },
    {
      "epoch": 11.335495283018869,
      "grad_norm": 2.594094753265381,
      "learning_rate": 4.7079060736199525e-06,
      "loss": 0.3697,
      "num_input_tokens_seen": 12666984,
      "step": 19225
    },
    {
      "epoch": 11.338443396226415,
      "grad_norm": 2.716944456100464,
      "learning_rate": 4.705337792062897e-06,
      "loss": 0.3259,
      "num_input_tokens_seen": 12669832,
      "step": 19230
    },
    {
      "epoch": 11.341391509433961,
      "grad_norm": 2.345000982284546,
      "learning_rate": 4.702769588518931e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 12672104,
      "step": 19235
    },
    {
      "epoch": 11.34433962264151,
      "grad_norm": 4.064178466796875,
      "learning_rate": 4.700201463667996e-06,
      "loss": 0.2529,
      "num_input_tokens_seen": 12674760,
      "step": 19240
    },
    {
      "epoch": 11.347287735849056,
      "grad_norm": 4.245500087738037,
      "learning_rate": 4.697633418190017e-06,
      "loss": 0.3019,
      "num_input_tokens_seen": 12677640,
      "step": 19245
    },
    {
      "epoch": 11.350235849056604,
      "grad_norm": 6.300871849060059,
      "learning_rate": 4.695065452764893e-06,
      "loss": 0.3517,
      "num_input_tokens_seen": 12680392,
      "step": 19250
    },
    {
      "epoch": 11.35318396226415,
      "grad_norm": 2.213015556335449,
      "learning_rate": 4.692497568072505e-06,
      "loss": 0.3473,
      "num_input_tokens_seen": 12684360,
      "step": 19255
    },
    {
      "epoch": 11.356132075471699,
      "grad_norm": 2.527740955352783,
      "learning_rate": 4.689929764792711e-06,
      "loss": 0.4447,
      "num_input_tokens_seen": 12687240,
      "step": 19260
    },
    {
      "epoch": 11.359080188679245,
      "grad_norm": 2.4260926246643066,
      "learning_rate": 4.687362043605349e-06,
      "loss": 0.3787,
      "num_input_tokens_seen": 12689928,
      "step": 19265
    },
    {
      "epoch": 11.362028301886792,
      "grad_norm": 2.87019419670105,
      "learning_rate": 4.684794405190231e-06,
      "loss": 0.381,
      "num_input_tokens_seen": 12693672,
      "step": 19270
    },
    {
      "epoch": 11.36497641509434,
      "grad_norm": 2.0644633769989014,
      "learning_rate": 4.682226850227155e-06,
      "loss": 0.3712,
      "num_input_tokens_seen": 12697640,
      "step": 19275
    },
    {
      "epoch": 11.367924528301886,
      "grad_norm": 3.186251640319824,
      "learning_rate": 4.6796593793958875e-06,
      "loss": 0.4843,
      "num_input_tokens_seen": 12701352,
      "step": 19280
    },
    {
      "epoch": 11.370872641509434,
      "grad_norm": 4.527063846588135,
      "learning_rate": 4.6770919933761815e-06,
      "loss": 0.3446,
      "num_input_tokens_seen": 12704360,
      "step": 19285
    },
    {
      "epoch": 11.37382075471698,
      "grad_norm": 4.775228977203369,
      "learning_rate": 4.6745246928477615e-06,
      "loss": 0.3604,
      "num_input_tokens_seen": 12706824,
      "step": 19290
    },
    {
      "epoch": 11.376768867924529,
      "grad_norm": 3.4020965099334717,
      "learning_rate": 4.671957478490332e-06,
      "loss": 0.3684,
      "num_input_tokens_seen": 12709832,
      "step": 19295
    },
    {
      "epoch": 11.379716981132075,
      "grad_norm": 4.501711845397949,
      "learning_rate": 4.669390350983574e-06,
      "loss": 0.365,
      "num_input_tokens_seen": 12713896,
      "step": 19300
    },
    {
      "epoch": 11.382665094339623,
      "grad_norm": 3.04833984375,
      "learning_rate": 4.666823311007145e-06,
      "loss": 0.3655,
      "num_input_tokens_seen": 12717640,
      "step": 19305
    },
    {
      "epoch": 11.38561320754717,
      "grad_norm": 4.221590995788574,
      "learning_rate": 4.664256359240681e-06,
      "loss": 0.438,
      "num_input_tokens_seen": 12721512,
      "step": 19310
    },
    {
      "epoch": 11.388561320754716,
      "grad_norm": 3.2570793628692627,
      "learning_rate": 4.661689496363793e-06,
      "loss": 0.3949,
      "num_input_tokens_seen": 12724136,
      "step": 19315
    },
    {
      "epoch": 11.391509433962264,
      "grad_norm": 2.3634183406829834,
      "learning_rate": 4.659122723056068e-06,
      "loss": 0.2777,
      "num_input_tokens_seen": 12726760,
      "step": 19320
    },
    {
      "epoch": 11.39445754716981,
      "grad_norm": 3.9188168048858643,
      "learning_rate": 4.656556039997072e-06,
      "loss": 0.4087,
      "num_input_tokens_seen": 12729192,
      "step": 19325
    },
    {
      "epoch": 11.397405660377359,
      "grad_norm": 6.984065532684326,
      "learning_rate": 4.653989447866345e-06,
      "loss": 0.5215,
      "num_input_tokens_seen": 12731592,
      "step": 19330
    },
    {
      "epoch": 11.400353773584905,
      "grad_norm": 18.07296371459961,
      "learning_rate": 4.651422947343401e-06,
      "loss": 0.5295,
      "num_input_tokens_seen": 12734344,
      "step": 19335
    },
    {
      "epoch": 11.403301886792454,
      "grad_norm": 2.5688369274139404,
      "learning_rate": 4.6488565391077355e-06,
      "loss": 0.4214,
      "num_input_tokens_seen": 12738856,
      "step": 19340
    },
    {
      "epoch": 11.40625,
      "grad_norm": 1.2210142612457275,
      "learning_rate": 4.646290223838815e-06,
      "loss": 0.307,
      "num_input_tokens_seen": 12742696,
      "step": 19345
    },
    {
      "epoch": 11.409198113207546,
      "grad_norm": 6.7526679039001465,
      "learning_rate": 4.64372400221608e-06,
      "loss": 0.3803,
      "num_input_tokens_seen": 12745736,
      "step": 19350
    },
    {
      "epoch": 11.412146226415095,
      "grad_norm": 4.742504596710205,
      "learning_rate": 4.641157874918953e-06,
      "loss": 0.4261,
      "num_input_tokens_seen": 12748680,
      "step": 19355
    },
    {
      "epoch": 11.415094339622641,
      "grad_norm": 2.710089683532715,
      "learning_rate": 4.6385918426268245e-06,
      "loss": 0.3316,
      "num_input_tokens_seen": 12752680,
      "step": 19360
    },
    {
      "epoch": 11.41804245283019,
      "grad_norm": 2.2463736534118652,
      "learning_rate": 4.636025906019062e-06,
      "loss": 0.3412,
      "num_input_tokens_seen": 12755560,
      "step": 19365
    },
    {
      "epoch": 11.420990566037736,
      "grad_norm": 1.9697859287261963,
      "learning_rate": 4.6334600657750115e-06,
      "loss": 0.2665,
      "num_input_tokens_seen": 12760136,
      "step": 19370
    },
    {
      "epoch": 11.423938679245284,
      "grad_norm": 6.640734672546387,
      "learning_rate": 4.6308943225739855e-06,
      "loss": 0.4247,
      "num_input_tokens_seen": 12763752,
      "step": 19375
    },
    {
      "epoch": 11.42688679245283,
      "grad_norm": 5.089629650115967,
      "learning_rate": 4.628328677095281e-06,
      "loss": 0.3022,
      "num_input_tokens_seen": 12766312,
      "step": 19380
    },
    {
      "epoch": 11.429834905660377,
      "grad_norm": 2.626344680786133,
      "learning_rate": 4.625763130018159e-06,
      "loss": 0.3922,
      "num_input_tokens_seen": 12769288,
      "step": 19385
    },
    {
      "epoch": 11.432783018867925,
      "grad_norm": 3.022249937057495,
      "learning_rate": 4.6231976820218635e-06,
      "loss": 0.4593,
      "num_input_tokens_seen": 12774696,
      "step": 19390
    },
    {
      "epoch": 11.435731132075471,
      "grad_norm": 4.034916400909424,
      "learning_rate": 4.620632333785605e-06,
      "loss": 0.3103,
      "num_input_tokens_seen": 12777192,
      "step": 19395
    },
    {
      "epoch": 11.43867924528302,
      "grad_norm": 4.855743408203125,
      "learning_rate": 4.618067085988569e-06,
      "loss": 0.3727,
      "num_input_tokens_seen": 12782408,
      "step": 19400
    },
    {
      "epoch": 11.441627358490566,
      "grad_norm": 2.2551424503326416,
      "learning_rate": 4.615501939309921e-06,
      "loss": 0.4308,
      "num_input_tokens_seen": 12786184,
      "step": 19405
    },
    {
      "epoch": 11.444575471698114,
      "grad_norm": 3.1894774436950684,
      "learning_rate": 4.612936894428791e-06,
      "loss": 0.3356,
      "num_input_tokens_seen": 12789224,
      "step": 19410
    },
    {
      "epoch": 11.44752358490566,
      "grad_norm": 3.9213485717773438,
      "learning_rate": 4.610371952024285e-06,
      "loss": 0.4528,
      "num_input_tokens_seen": 12792776,
      "step": 19415
    },
    {
      "epoch": 11.450471698113208,
      "grad_norm": 2.0451672077178955,
      "learning_rate": 4.607807112775485e-06,
      "loss": 0.4763,
      "num_input_tokens_seen": 12795816,
      "step": 19420
    },
    {
      "epoch": 11.453419811320755,
      "grad_norm": 2.856206178665161,
      "learning_rate": 4.605242377361441e-06,
      "loss": 0.2677,
      "num_input_tokens_seen": 12799464,
      "step": 19425
    },
    {
      "epoch": 11.456367924528301,
      "grad_norm": 3.437051296234131,
      "learning_rate": 4.60267774646118e-06,
      "loss": 0.3784,
      "num_input_tokens_seen": 12802024,
      "step": 19430
    },
    {
      "epoch": 11.45931603773585,
      "grad_norm": 5.403711795806885,
      "learning_rate": 4.600113220753698e-06,
      "loss": 0.5562,
      "num_input_tokens_seen": 12805000,
      "step": 19435
    },
    {
      "epoch": 11.462264150943396,
      "grad_norm": 3.1375086307525635,
      "learning_rate": 4.597548800917964e-06,
      "loss": 0.3627,
      "num_input_tokens_seen": 12807816,
      "step": 19440
    },
    {
      "epoch": 11.465212264150944,
      "grad_norm": 2.9291059970855713,
      "learning_rate": 4.594984487632919e-06,
      "loss": 0.3249,
      "num_input_tokens_seen": 12810632,
      "step": 19445
    },
    {
      "epoch": 11.46816037735849,
      "grad_norm": 2.4111692905426025,
      "learning_rate": 4.592420281577478e-06,
      "loss": 0.3067,
      "num_input_tokens_seen": 12813608,
      "step": 19450
    },
    {
      "epoch": 11.471108490566039,
      "grad_norm": 9.880294799804688,
      "learning_rate": 4.589856183430521e-06,
      "loss": 0.4389,
      "num_input_tokens_seen": 12817864,
      "step": 19455
    },
    {
      "epoch": 11.474056603773585,
      "grad_norm": 2.0387661457061768,
      "learning_rate": 4.58729219387091e-06,
      "loss": 0.322,
      "num_input_tokens_seen": 12821224,
      "step": 19460
    },
    {
      "epoch": 11.477004716981131,
      "grad_norm": 4.243068695068359,
      "learning_rate": 4.584728313577468e-06,
      "loss": 0.32,
      "num_input_tokens_seen": 12824392,
      "step": 19465
    },
    {
      "epoch": 11.47995283018868,
      "grad_norm": 2.6662120819091797,
      "learning_rate": 4.582164543228993e-06,
      "loss": 0.2951,
      "num_input_tokens_seen": 12827144,
      "step": 19470
    },
    {
      "epoch": 11.482900943396226,
      "grad_norm": 4.818130016326904,
      "learning_rate": 4.5796008835042574e-06,
      "loss": 0.4044,
      "num_input_tokens_seen": 12830344,
      "step": 19475
    },
    {
      "epoch": 11.485849056603774,
      "grad_norm": 3.073967218399048,
      "learning_rate": 4.577037335082e-06,
      "loss": 0.2758,
      "num_input_tokens_seen": 12833096,
      "step": 19480
    },
    {
      "epoch": 11.48879716981132,
      "grad_norm": 3.6895060539245605,
      "learning_rate": 4.574473898640933e-06,
      "loss": 0.4237,
      "num_input_tokens_seen": 12836680,
      "step": 19485
    },
    {
      "epoch": 11.491745283018869,
      "grad_norm": 2.773444652557373,
      "learning_rate": 4.571910574859732e-06,
      "loss": 0.345,
      "num_input_tokens_seen": 12839240,
      "step": 19490
    },
    {
      "epoch": 11.494693396226415,
      "grad_norm": 3.1892502307891846,
      "learning_rate": 4.5693473644170535e-06,
      "loss": 0.3723,
      "num_input_tokens_seen": 12841896,
      "step": 19495
    },
    {
      "epoch": 11.497641509433961,
      "grad_norm": 2.2859749794006348,
      "learning_rate": 4.566784267991516e-06,
      "loss": 0.2908,
      "num_input_tokens_seen": 12844904,
      "step": 19500
    },
    {
      "epoch": 11.50058962264151,
      "grad_norm": 2.377124071121216,
      "learning_rate": 4.564221286261709e-06,
      "loss": 0.3063,
      "num_input_tokens_seen": 12847848,
      "step": 19505
    },
    {
      "epoch": 11.503537735849056,
      "grad_norm": 1.5922348499298096,
      "learning_rate": 4.5616584199061964e-06,
      "loss": 0.3584,
      "num_input_tokens_seen": 12851656,
      "step": 19510
    },
    {
      "epoch": 11.506485849056604,
      "grad_norm": 5.7457685470581055,
      "learning_rate": 4.559095669603506e-06,
      "loss": 0.4252,
      "num_input_tokens_seen": 12855432,
      "step": 19515
    },
    {
      "epoch": 11.50943396226415,
      "grad_norm": 2.1401867866516113,
      "learning_rate": 4.556533036032136e-06,
      "loss": 0.269,
      "num_input_tokens_seen": 12858728,
      "step": 19520
    },
    {
      "epoch": 11.512382075471699,
      "grad_norm": 2.009727716445923,
      "learning_rate": 4.553970519870557e-06,
      "loss": 0.3452,
      "num_input_tokens_seen": 12862088,
      "step": 19525
    },
    {
      "epoch": 11.515330188679245,
      "grad_norm": 3.1603028774261475,
      "learning_rate": 4.551408121797205e-06,
      "loss": 0.4002,
      "num_input_tokens_seen": 12865160,
      "step": 19530
    },
    {
      "epoch": 11.518278301886792,
      "grad_norm": 9.410238265991211,
      "learning_rate": 4.548845842490486e-06,
      "loss": 0.4154,
      "num_input_tokens_seen": 12867976,
      "step": 19535
    },
    {
      "epoch": 11.52122641509434,
      "grad_norm": 4.4053120613098145,
      "learning_rate": 4.5462836826287745e-06,
      "loss": 0.3731,
      "num_input_tokens_seen": 12871432,
      "step": 19540
    },
    {
      "epoch": 11.524174528301886,
      "grad_norm": 2.6784112453460693,
      "learning_rate": 4.543721642890414e-06,
      "loss": 0.3937,
      "num_input_tokens_seen": 12874856,
      "step": 19545
    },
    {
      "epoch": 11.527122641509434,
      "grad_norm": 4.565755844116211,
      "learning_rate": 4.541159723953714e-06,
      "loss": 0.4041,
      "num_input_tokens_seen": 12878664,
      "step": 19550
    },
    {
      "epoch": 11.53007075471698,
      "grad_norm": 1.5392042398452759,
      "learning_rate": 4.538597926496955e-06,
      "loss": 0.325,
      "num_input_tokens_seen": 12882408,
      "step": 19555
    },
    {
      "epoch": 11.533018867924529,
      "grad_norm": 4.090847969055176,
      "learning_rate": 4.536036251198384e-06,
      "loss": 0.453,
      "num_input_tokens_seen": 12884648,
      "step": 19560
    },
    {
      "epoch": 11.535966981132075,
      "grad_norm": 3.9518730640411377,
      "learning_rate": 4.5334746987362124e-06,
      "loss": 0.354,
      "num_input_tokens_seen": 12887752,
      "step": 19565
    },
    {
      "epoch": 11.538915094339622,
      "grad_norm": 3.765857696533203,
      "learning_rate": 4.530913269788627e-06,
      "loss": 0.4474,
      "num_input_tokens_seen": 12890184,
      "step": 19570
    },
    {
      "epoch": 11.54186320754717,
      "grad_norm": 2.8429043292999268,
      "learning_rate": 4.528351965033775e-06,
      "loss": 0.2736,
      "num_input_tokens_seen": 12893576,
      "step": 19575
    },
    {
      "epoch": 11.544811320754716,
      "grad_norm": 3.5920979976654053,
      "learning_rate": 4.525790785149774e-06,
      "loss": 0.3647,
      "num_input_tokens_seen": 12896264,
      "step": 19580
    },
    {
      "epoch": 11.547759433962264,
      "grad_norm": 3.0469675064086914,
      "learning_rate": 4.523229730814705e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 12898888,
      "step": 19585
    },
    {
      "epoch": 11.55070754716981,
      "grad_norm": 3.8665781021118164,
      "learning_rate": 4.52066880270662e-06,
      "loss": 0.2521,
      "num_input_tokens_seen": 12901544,
      "step": 19590
    },
    {
      "epoch": 11.553655660377359,
      "grad_norm": 6.742138385772705,
      "learning_rate": 4.518108001503536e-06,
      "loss": 0.3089,
      "num_input_tokens_seen": 12904200,
      "step": 19595
    },
    {
      "epoch": 11.556603773584905,
      "grad_norm": 11.703137397766113,
      "learning_rate": 4.515547327883434e-06,
      "loss": 0.5799,
      "num_input_tokens_seen": 12907752,
      "step": 19600
    },
    {
      "epoch": 11.559551886792454,
      "grad_norm": 2.2839460372924805,
      "learning_rate": 4.512986782524266e-06,
      "loss": 0.2793,
      "num_input_tokens_seen": 12910760,
      "step": 19605
    },
    {
      "epoch": 11.5625,
      "grad_norm": 2.893782138824463,
      "learning_rate": 4.510426366103946e-06,
      "loss": 0.2867,
      "num_input_tokens_seen": 12913928,
      "step": 19610
    },
    {
      "epoch": 11.565448113207546,
      "grad_norm": 4.706821918487549,
      "learning_rate": 4.5078660793003544e-06,
      "loss": 0.337,
      "num_input_tokens_seen": 12917032,
      "step": 19615
    },
    {
      "epoch": 11.568396226415095,
      "grad_norm": 4.422600269317627,
      "learning_rate": 4.505305922791341e-06,
      "loss": 0.47,
      "num_input_tokens_seen": 12920584,
      "step": 19620
    },
    {
      "epoch": 11.571344339622641,
      "grad_norm": 4.011805534362793,
      "learning_rate": 4.502745897254716e-06,
      "loss": 0.3423,
      "num_input_tokens_seen": 12927112,
      "step": 19625
    },
    {
      "epoch": 11.57429245283019,
      "grad_norm": 4.956382751464844,
      "learning_rate": 4.500186003368257e-06,
      "loss": 0.3532,
      "num_input_tokens_seen": 12929576,
      "step": 19630
    },
    {
      "epoch": 11.577240566037736,
      "grad_norm": 3.1527163982391357,
      "learning_rate": 4.497626241809709e-06,
      "loss": 0.4085,
      "num_input_tokens_seen": 12932648,
      "step": 19635
    },
    {
      "epoch": 11.580188679245284,
      "grad_norm": 2.2625417709350586,
      "learning_rate": 4.4950666132567775e-06,
      "loss": 0.3159,
      "num_input_tokens_seen": 12937064,
      "step": 19640
    },
    {
      "epoch": 11.58313679245283,
      "grad_norm": 4.313152313232422,
      "learning_rate": 4.492507118387136e-06,
      "loss": 0.2678,
      "num_input_tokens_seen": 12939400,
      "step": 19645
    },
    {
      "epoch": 11.586084905660378,
      "grad_norm": 4.129263877868652,
      "learning_rate": 4.4899477578784235e-06,
      "loss": 0.2499,
      "num_input_tokens_seen": 12941736,
      "step": 19650
    },
    {
      "epoch": 11.589033018867925,
      "grad_norm": 1.8315268754959106,
      "learning_rate": 4.487388532408239e-06,
      "loss": 0.3875,
      "num_input_tokens_seen": 12945288,
      "step": 19655
    },
    {
      "epoch": 11.591981132075471,
      "grad_norm": 2.747342109680176,
      "learning_rate": 4.48482944265415e-06,
      "loss": 0.2571,
      "num_input_tokens_seen": 12948232,
      "step": 19660
    },
    {
      "epoch": 11.59492924528302,
      "grad_norm": 2.0491979122161865,
      "learning_rate": 4.482270489293685e-06,
      "loss": 0.4816,
      "num_input_tokens_seen": 12951624,
      "step": 19665
    },
    {
      "epoch": 11.597877358490566,
      "grad_norm": 4.273489952087402,
      "learning_rate": 4.479711673004341e-06,
      "loss": 0.3033,
      "num_input_tokens_seen": 12955144,
      "step": 19670
    },
    {
      "epoch": 11.600825471698114,
      "grad_norm": 3.587686538696289,
      "learning_rate": 4.477152994463575e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 12958856,
      "step": 19675
    },
    {
      "epoch": 11.60377358490566,
      "grad_norm": 3.102142095565796,
      "learning_rate": 4.474594454348805e-06,
      "loss": 0.3431,
      "num_input_tokens_seen": 12962248,
      "step": 19680
    },
    {
      "epoch": 11.606721698113208,
      "grad_norm": 2.2090024948120117,
      "learning_rate": 4.472036053337419e-06,
      "loss": 0.5126,
      "num_input_tokens_seen": 12966440,
      "step": 19685
    },
    {
      "epoch": 11.609669811320755,
      "grad_norm": 3.945950984954834,
      "learning_rate": 4.469477792106761e-06,
      "loss": 0.3579,
      "num_input_tokens_seen": 12969640,
      "step": 19690
    },
    {
      "epoch": 11.612617924528301,
      "grad_norm": 3.133949041366577,
      "learning_rate": 4.466919671334146e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 12973000,
      "step": 19695
    },
    {
      "epoch": 11.61556603773585,
      "grad_norm": 1.89557683467865,
      "learning_rate": 4.4643616916968456e-06,
      "loss": 0.404,
      "num_input_tokens_seen": 12976712,
      "step": 19700
    },
    {
      "epoch": 11.618514150943396,
      "grad_norm": 2.919226884841919,
      "learning_rate": 4.461803853872095e-06,
      "loss": 0.3176,
      "num_input_tokens_seen": 12980712,
      "step": 19705
    },
    {
      "epoch": 11.621462264150944,
      "grad_norm": 2.8891775608062744,
      "learning_rate": 4.459246158537094e-06,
      "loss": 0.5088,
      "num_input_tokens_seen": 12983912,
      "step": 19710
    },
    {
      "epoch": 11.62441037735849,
      "grad_norm": 3.460460662841797,
      "learning_rate": 4.456688606369003e-06,
      "loss": 0.4019,
      "num_input_tokens_seen": 12987464,
      "step": 19715
    },
    {
      "epoch": 11.627358490566039,
      "grad_norm": 4.202056407928467,
      "learning_rate": 4.4541311980449446e-06,
      "loss": 0.3761,
      "num_input_tokens_seen": 12990984,
      "step": 19720
    },
    {
      "epoch": 11.630306603773585,
      "grad_norm": 3.875872850418091,
      "learning_rate": 4.4515739342420045e-06,
      "loss": 0.3577,
      "num_input_tokens_seen": 12994504,
      "step": 19725
    },
    {
      "epoch": 11.633254716981131,
      "grad_norm": 2.230929136276245,
      "learning_rate": 4.4490168156372294e-06,
      "loss": 0.4644,
      "num_input_tokens_seen": 12997416,
      "step": 19730
    },
    {
      "epoch": 11.63620283018868,
      "grad_norm": 10.970770835876465,
      "learning_rate": 4.446459842907626e-06,
      "loss": 0.3106,
      "num_input_tokens_seen": 12999976,
      "step": 19735
    },
    {
      "epoch": 11.639150943396226,
      "grad_norm": 3.844879150390625,
      "learning_rate": 4.443903016730165e-06,
      "loss": 0.3702,
      "num_input_tokens_seen": 13003176,
      "step": 19740
    },
    {
      "epoch": 11.642099056603774,
      "grad_norm": 3.6625633239746094,
      "learning_rate": 4.4413463377817775e-06,
      "loss": 0.3244,
      "num_input_tokens_seen": 13006312,
      "step": 19745
    },
    {
      "epoch": 11.64504716981132,
      "grad_norm": 2.1409406661987305,
      "learning_rate": 4.438789806739353e-06,
      "loss": 0.3751,
      "num_input_tokens_seen": 13008776,
      "step": 19750
    },
    {
      "epoch": 11.647995283018869,
      "grad_norm": 2.406168222427368,
      "learning_rate": 4.436233424279746e-06,
      "loss": 0.4347,
      "num_input_tokens_seen": 13011944,
      "step": 19755
    },
    {
      "epoch": 11.650943396226415,
      "grad_norm": 3.264860153198242,
      "learning_rate": 4.433677191079771e-06,
      "loss": 0.3377,
      "num_input_tokens_seen": 13015464,
      "step": 19760
    },
    {
      "epoch": 11.653891509433961,
      "grad_norm": 3.4588074684143066,
      "learning_rate": 4.431121107816201e-06,
      "loss": 0.371,
      "num_input_tokens_seen": 13018728,
      "step": 19765
    },
    {
      "epoch": 11.65683962264151,
      "grad_norm": 3.2582268714904785,
      "learning_rate": 4.4285651751657676e-06,
      "loss": 0.3916,
      "num_input_tokens_seen": 13022024,
      "step": 19770
    },
    {
      "epoch": 11.659787735849056,
      "grad_norm": 3.544985055923462,
      "learning_rate": 4.4260093938051685e-06,
      "loss": 0.3605,
      "num_input_tokens_seen": 13024584,
      "step": 19775
    },
    {
      "epoch": 11.662735849056604,
      "grad_norm": 2.8161981105804443,
      "learning_rate": 4.423453764411056e-06,
      "loss": 0.3682,
      "num_input_tokens_seen": 13029896,
      "step": 19780
    },
    {
      "epoch": 11.66568396226415,
      "grad_norm": 2.3850317001342773,
      "learning_rate": 4.4208982876600425e-06,
      "loss": 0.3453,
      "num_input_tokens_seen": 13033480,
      "step": 19785
    },
    {
      "epoch": 11.668632075471699,
      "grad_norm": 1.811225414276123,
      "learning_rate": 4.418342964228705e-06,
      "loss": 0.4401,
      "num_input_tokens_seen": 13036616,
      "step": 19790
    },
    {
      "epoch": 11.671580188679245,
      "grad_norm": 6.750536918640137,
      "learning_rate": 4.415787794793574e-06,
      "loss": 0.3532,
      "num_input_tokens_seen": 13038856,
      "step": 19795
    },
    {
      "epoch": 11.674528301886792,
      "grad_norm": 1.8598994016647339,
      "learning_rate": 4.4132327800311414e-06,
      "loss": 0.3176,
      "num_input_tokens_seen": 13041416,
      "step": 19800
    },
    {
      "epoch": 11.67747641509434,
      "grad_norm": 4.803706645965576,
      "learning_rate": 4.41067792061786e-06,
      "loss": 0.3688,
      "num_input_tokens_seen": 13044008,
      "step": 19805
    },
    {
      "epoch": 11.680424528301886,
      "grad_norm": 2.575505256652832,
      "learning_rate": 4.408123217230139e-06,
      "loss": 0.3838,
      "num_input_tokens_seen": 13047560,
      "step": 19810
    },
    {
      "epoch": 11.683372641509434,
      "grad_norm": 6.593127727508545,
      "learning_rate": 4.4055686705443464e-06,
      "loss": 0.2903,
      "num_input_tokens_seen": 13050632,
      "step": 19815
    },
    {
      "epoch": 11.68632075471698,
      "grad_norm": 2.65437388420105,
      "learning_rate": 4.4030142812368106e-06,
      "loss": 0.3472,
      "num_input_tokens_seen": 13053576,
      "step": 19820
    },
    {
      "epoch": 11.689268867924529,
      "grad_norm": 4.872476100921631,
      "learning_rate": 4.400460049983817e-06,
      "loss": 0.2619,
      "num_input_tokens_seen": 13056584,
      "step": 19825
    },
    {
      "epoch": 11.692216981132075,
      "grad_norm": 3.2316503524780273,
      "learning_rate": 4.397905977461608e-06,
      "loss": 0.2889,
      "num_input_tokens_seen": 13061672,
      "step": 19830
    },
    {
      "epoch": 11.695165094339622,
      "grad_norm": 3.088627338409424,
      "learning_rate": 4.395352064346387e-06,
      "loss": 0.4467,
      "num_input_tokens_seen": 13064680,
      "step": 19835
    },
    {
      "epoch": 11.69811320754717,
      "grad_norm": 4.444119453430176,
      "learning_rate": 4.392798311314314e-06,
      "loss": 0.3919,
      "num_input_tokens_seen": 13068104,
      "step": 19840
    },
    {
      "epoch": 11.701061320754716,
      "grad_norm": 3.32989764213562,
      "learning_rate": 4.390244719041502e-06,
      "loss": 0.3168,
      "num_input_tokens_seen": 13071016,
      "step": 19845
    },
    {
      "epoch": 11.704009433962264,
      "grad_norm": 4.072218418121338,
      "learning_rate": 4.387691288204029e-06,
      "loss": 0.2882,
      "num_input_tokens_seen": 13073864,
      "step": 19850
    },
    {
      "epoch": 11.70695754716981,
      "grad_norm": 3.0464730262756348,
      "learning_rate": 4.385138019477928e-06,
      "loss": 0.2804,
      "num_input_tokens_seen": 13076264,
      "step": 19855
    },
    {
      "epoch": 11.709905660377359,
      "grad_norm": 4.558078289031982,
      "learning_rate": 4.3825849135391864e-06,
      "loss": 0.3444,
      "num_input_tokens_seen": 13078664,
      "step": 19860
    },
    {
      "epoch": 11.712853773584905,
      "grad_norm": 2.3492090702056885,
      "learning_rate": 4.380031971063748e-06,
      "loss": 0.3658,
      "num_input_tokens_seen": 13082216,
      "step": 19865
    },
    {
      "epoch": 11.715801886792454,
      "grad_norm": 3.908867359161377,
      "learning_rate": 4.37747919272752e-06,
      "loss": 0.4238,
      "num_input_tokens_seen": 13085064,
      "step": 19870
    },
    {
      "epoch": 11.71875,
      "grad_norm": 2.269629955291748,
      "learning_rate": 4.374926579206357e-06,
      "loss": 0.2985,
      "num_input_tokens_seen": 13087656,
      "step": 19875
    },
    {
      "epoch": 11.721698113207546,
      "grad_norm": 2.560128688812256,
      "learning_rate": 4.372374131176075e-06,
      "loss": 0.3818,
      "num_input_tokens_seen": 13091176,
      "step": 19880
    },
    {
      "epoch": 11.724646226415095,
      "grad_norm": 3.767742395401001,
      "learning_rate": 4.369821849312449e-06,
      "loss": 0.4284,
      "num_input_tokens_seen": 13094536,
      "step": 19885
    },
    {
      "epoch": 11.727594339622641,
      "grad_norm": 4.980728626251221,
      "learning_rate": 4.367269734291203e-06,
      "loss": 0.4888,
      "num_input_tokens_seen": 13097864,
      "step": 19890
    },
    {
      "epoch": 11.73054245283019,
      "grad_norm": 2.5639147758483887,
      "learning_rate": 4.364717786788022e-06,
      "loss": 0.3046,
      "num_input_tokens_seen": 13100840,
      "step": 19895
    },
    {
      "epoch": 11.733490566037736,
      "grad_norm": 3.242769956588745,
      "learning_rate": 4.362166007478545e-06,
      "loss": 0.3934,
      "num_input_tokens_seen": 13104136,
      "step": 19900
    },
    {
      "epoch": 11.736438679245284,
      "grad_norm": 2.016415596008301,
      "learning_rate": 4.3596143970383665e-06,
      "loss": 0.3322,
      "num_input_tokens_seen": 13107272,
      "step": 19905
    },
    {
      "epoch": 11.73938679245283,
      "grad_norm": 2.792510747909546,
      "learning_rate": 4.357062956143035e-06,
      "loss": 0.3601,
      "num_input_tokens_seen": 13110088,
      "step": 19910
    },
    {
      "epoch": 11.742334905660378,
      "grad_norm": 2.0355563163757324,
      "learning_rate": 4.354511685468059e-06,
      "loss": 0.3653,
      "num_input_tokens_seen": 13113512,
      "step": 19915
    },
    {
      "epoch": 11.745283018867925,
      "grad_norm": 2.9518566131591797,
      "learning_rate": 4.351960585688894e-06,
      "loss": 0.3774,
      "num_input_tokens_seen": 13116840,
      "step": 19920
    },
    {
      "epoch": 11.748231132075471,
      "grad_norm": 2.7240729331970215,
      "learning_rate": 4.349409657480959e-06,
      "loss": 0.3914,
      "num_input_tokens_seen": 13120072,
      "step": 19925
    },
    {
      "epoch": 11.75117924528302,
      "grad_norm": 4.241157054901123,
      "learning_rate": 4.34685890151962e-06,
      "loss": 0.3168,
      "num_input_tokens_seen": 13122728,
      "step": 19930
    },
    {
      "epoch": 11.754127358490566,
      "grad_norm": 1.631070613861084,
      "learning_rate": 4.344308318480201e-06,
      "loss": 0.3647,
      "num_input_tokens_seen": 13126472,
      "step": 19935
    },
    {
      "epoch": 11.757075471698114,
      "grad_norm": 5.32188606262207,
      "learning_rate": 4.341757909037981e-06,
      "loss": 0.3224,
      "num_input_tokens_seen": 13129800,
      "step": 19940
    },
    {
      "epoch": 11.76002358490566,
      "grad_norm": 3.3034322261810303,
      "learning_rate": 4.33920767386819e-06,
      "loss": 0.383,
      "num_input_tokens_seen": 13132744,
      "step": 19945
    },
    {
      "epoch": 11.762971698113208,
      "grad_norm": 6.38716983795166,
      "learning_rate": 4.336657613646017e-06,
      "loss": 0.3286,
      "num_input_tokens_seen": 13135528,
      "step": 19950
    },
    {
      "epoch": 11.765919811320755,
      "grad_norm": 2.799443244934082,
      "learning_rate": 4.3341077290466e-06,
      "loss": 0.2921,
      "num_input_tokens_seen": 13138344,
      "step": 19955
    },
    {
      "epoch": 11.768867924528301,
      "grad_norm": 2.130488157272339,
      "learning_rate": 4.331558020745031e-06,
      "loss": 0.4286,
      "num_input_tokens_seen": 13142728,
      "step": 19960
    },
    {
      "epoch": 11.77181603773585,
      "grad_norm": 3.6103005409240723,
      "learning_rate": 4.329008489416357e-06,
      "loss": 0.3877,
      "num_input_tokens_seen": 13147592,
      "step": 19965
    },
    {
      "epoch": 11.774764150943396,
      "grad_norm": 3.477743148803711,
      "learning_rate": 4.326459135735576e-06,
      "loss": 0.3075,
      "num_input_tokens_seen": 13150376,
      "step": 19970
    },
    {
      "epoch": 11.777712264150944,
      "grad_norm": 2.6173911094665527,
      "learning_rate": 4.323909960377644e-06,
      "loss": 0.3745,
      "num_input_tokens_seen": 13153960,
      "step": 19975
    },
    {
      "epoch": 11.78066037735849,
      "grad_norm": 2.8441643714904785,
      "learning_rate": 4.3213609640174625e-06,
      "loss": 0.2774,
      "num_input_tokens_seen": 13157128,
      "step": 19980
    },
    {
      "epoch": 11.783608490566039,
      "grad_norm": 4.3567962646484375,
      "learning_rate": 4.318812147329889e-06,
      "loss": 0.345,
      "num_input_tokens_seen": 13160968,
      "step": 19985
    },
    {
      "epoch": 11.786556603773585,
      "grad_norm": 2.931596040725708,
      "learning_rate": 4.316263510989737e-06,
      "loss": 0.3668,
      "num_input_tokens_seen": 13163304,
      "step": 19990
    },
    {
      "epoch": 11.789504716981131,
      "grad_norm": 4.722005844116211,
      "learning_rate": 4.313715055671768e-06,
      "loss": 0.2958,
      "num_input_tokens_seen": 13165992,
      "step": 19995
    },
    {
      "epoch": 11.79245283018868,
      "grad_norm": 4.139705657958984,
      "learning_rate": 4.311166782050694e-06,
      "loss": 0.3165,
      "num_input_tokens_seen": 13168744,
      "step": 20000
    },
    {
      "epoch": 11.795400943396226,
      "grad_norm": 3.164364814758301,
      "learning_rate": 4.308618690801184e-06,
      "loss": 0.3271,
      "num_input_tokens_seen": 13171624,
      "step": 20005
    },
    {
      "epoch": 11.798349056603774,
      "grad_norm": 1.8542406558990479,
      "learning_rate": 4.3060707825978564e-06,
      "loss": 0.3232,
      "num_input_tokens_seen": 13174408,
      "step": 20010
    },
    {
      "epoch": 11.80129716981132,
      "grad_norm": 3.6907529830932617,
      "learning_rate": 4.303523058115278e-06,
      "loss": 0.3995,
      "num_input_tokens_seen": 13177160,
      "step": 20015
    },
    {
      "epoch": 11.804245283018869,
      "grad_norm": 3.758117198944092,
      "learning_rate": 4.300975518027972e-06,
      "loss": 0.3439,
      "num_input_tokens_seen": 13179976,
      "step": 20020
    },
    {
      "epoch": 11.807193396226415,
      "grad_norm": 2.3794853687286377,
      "learning_rate": 4.298428163010411e-06,
      "loss": 0.4026,
      "num_input_tokens_seen": 13183272,
      "step": 20025
    },
    {
      "epoch": 11.810141509433961,
      "grad_norm": 1.74667227268219,
      "learning_rate": 4.295880993737016e-06,
      "loss": 0.3628,
      "num_input_tokens_seen": 13187528,
      "step": 20030
    },
    {
      "epoch": 11.81308962264151,
      "grad_norm": 2.457465887069702,
      "learning_rate": 4.293334010882164e-06,
      "loss": 0.3588,
      "num_input_tokens_seen": 13190344,
      "step": 20035
    },
    {
      "epoch": 11.816037735849056,
      "grad_norm": 3.6586437225341797,
      "learning_rate": 4.290787215120178e-06,
      "loss": 0.3792,
      "num_input_tokens_seen": 13193096,
      "step": 20040
    },
    {
      "epoch": 11.818985849056604,
      "grad_norm": 2.7739241123199463,
      "learning_rate": 4.2882406071253315e-06,
      "loss": 0.5469,
      "num_input_tokens_seen": 13196136,
      "step": 20045
    },
    {
      "epoch": 11.82193396226415,
      "grad_norm": 2.0254640579223633,
      "learning_rate": 4.285694187571852e-06,
      "loss": 0.3205,
      "num_input_tokens_seen": 13198952,
      "step": 20050
    },
    {
      "epoch": 11.824882075471699,
      "grad_norm": 9.308955192565918,
      "learning_rate": 4.283147957133917e-06,
      "loss": 0.3507,
      "num_input_tokens_seen": 13201768,
      "step": 20055
    },
    {
      "epoch": 11.827830188679245,
      "grad_norm": 6.894294738769531,
      "learning_rate": 4.280601916485648e-06,
      "loss": 0.4598,
      "num_input_tokens_seen": 13208552,
      "step": 20060
    },
    {
      "epoch": 11.830778301886792,
      "grad_norm": 2.520784378051758,
      "learning_rate": 4.278056066301123e-06,
      "loss": 0.4202,
      "num_input_tokens_seen": 13211464,
      "step": 20065
    },
    {
      "epoch": 11.83372641509434,
      "grad_norm": 9.282341003417969,
      "learning_rate": 4.275510407254366e-06,
      "loss": 0.4425,
      "num_input_tokens_seen": 13213896,
      "step": 20070
    },
    {
      "epoch": 11.836674528301886,
      "grad_norm": 2.9133453369140625,
      "learning_rate": 4.27296494001935e-06,
      "loss": 0.3867,
      "num_input_tokens_seen": 13216808,
      "step": 20075
    },
    {
      "epoch": 11.839622641509434,
      "grad_norm": 4.665931224822998,
      "learning_rate": 4.2704196652700005e-06,
      "loss": 0.339,
      "num_input_tokens_seen": 13219560,
      "step": 20080
    },
    {
      "epoch": 11.84257075471698,
      "grad_norm": 2.8308558464050293,
      "learning_rate": 4.2678745836801895e-06,
      "loss": 0.2976,
      "num_input_tokens_seen": 13222728,
      "step": 20085
    },
    {
      "epoch": 11.845518867924529,
      "grad_norm": 1.7903683185577393,
      "learning_rate": 4.265329695923736e-06,
      "loss": 0.4494,
      "num_input_tokens_seen": 13225992,
      "step": 20090
    },
    {
      "epoch": 11.848466981132075,
      "grad_norm": 6.438207626342773,
      "learning_rate": 4.262785002674412e-06,
      "loss": 0.3042,
      "num_input_tokens_seen": 13230312,
      "step": 20095
    },
    {
      "epoch": 11.851415094339622,
      "grad_norm": 2.6492247581481934,
      "learning_rate": 4.260240504605938e-06,
      "loss": 0.3685,
      "num_input_tokens_seen": 13233576,
      "step": 20100
    },
    {
      "epoch": 11.85436320754717,
      "grad_norm": 2.0827465057373047,
      "learning_rate": 4.2576962023919775e-06,
      "loss": 0.3861,
      "num_input_tokens_seen": 13237000,
      "step": 20105
    },
    {
      "epoch": 11.857311320754716,
      "grad_norm": 2.0183634757995605,
      "learning_rate": 4.255152096706145e-06,
      "loss": 0.3509,
      "num_input_tokens_seen": 13240360,
      "step": 20110
    },
    {
      "epoch": 11.860259433962264,
      "grad_norm": 4.896492958068848,
      "learning_rate": 4.252608188222007e-06,
      "loss": 0.4576,
      "num_input_tokens_seen": 13242952,
      "step": 20115
    },
    {
      "epoch": 11.86320754716981,
      "grad_norm": 1.7933363914489746,
      "learning_rate": 4.250064477613071e-06,
      "loss": 0.311,
      "num_input_tokens_seen": 13246248,
      "step": 20120
    },
    {
      "epoch": 11.866155660377359,
      "grad_norm": 3.3694632053375244,
      "learning_rate": 4.247520965552797e-06,
      "loss": 0.2458,
      "num_input_tokens_seen": 13249480,
      "step": 20125
    },
    {
      "epoch": 11.869103773584905,
      "grad_norm": 2.0234484672546387,
      "learning_rate": 4.24497765271459e-06,
      "loss": 0.4301,
      "num_input_tokens_seen": 13252680,
      "step": 20130
    },
    {
      "epoch": 11.872051886792454,
      "grad_norm": 2.807858943939209,
      "learning_rate": 4.242434539771804e-06,
      "loss": 0.3963,
      "num_input_tokens_seen": 13255784,
      "step": 20135
    },
    {
      "epoch": 11.875,
      "grad_norm": 6.929694175720215,
      "learning_rate": 4.239891627397737e-06,
      "loss": 0.3086,
      "num_input_tokens_seen": 13259656,
      "step": 20140
    },
    {
      "epoch": 11.877948113207546,
      "grad_norm": 1.6344819068908691,
      "learning_rate": 4.237348916265638e-06,
      "loss": 0.3578,
      "num_input_tokens_seen": 13263272,
      "step": 20145
    },
    {
      "epoch": 11.880896226415095,
      "grad_norm": 1.8872205018997192,
      "learning_rate": 4.234806407048702e-06,
      "loss": 0.3382,
      "num_input_tokens_seen": 13267720,
      "step": 20150
    },
    {
      "epoch": 11.883844339622641,
      "grad_norm": 2.773818254470825,
      "learning_rate": 4.232264100420066e-06,
      "loss": 0.3323,
      "num_input_tokens_seen": 13270952,
      "step": 20155
    },
    {
      "epoch": 11.88679245283019,
      "grad_norm": 3.8489224910736084,
      "learning_rate": 4.229721997052819e-06,
      "loss": 0.3585,
      "num_input_tokens_seen": 13274248,
      "step": 20160
    },
    {
      "epoch": 11.889740566037736,
      "grad_norm": 2.7346949577331543,
      "learning_rate": 4.227180097619993e-06,
      "loss": 0.3392,
      "num_input_tokens_seen": 13277288,
      "step": 20165
    },
    {
      "epoch": 11.892688679245284,
      "grad_norm": 3.7288436889648438,
      "learning_rate": 4.224638402794566e-06,
      "loss": 0.3371,
      "num_input_tokens_seen": 13279976,
      "step": 20170
    },
    {
      "epoch": 11.89563679245283,
      "grad_norm": 2.563570261001587,
      "learning_rate": 4.222096913249464e-06,
      "loss": 0.3385,
      "num_input_tokens_seen": 13283368,
      "step": 20175
    },
    {
      "epoch": 11.898584905660378,
      "grad_norm": 3.2197999954223633,
      "learning_rate": 4.219555629657558e-06,
      "loss": 0.2733,
      "num_input_tokens_seen": 13285960,
      "step": 20180
    },
    {
      "epoch": 11.901533018867925,
      "grad_norm": 3.3383734226226807,
      "learning_rate": 4.21701455269166e-06,
      "loss": 0.3836,
      "num_input_tokens_seen": 13289000,
      "step": 20185
    },
    {
      "epoch": 11.904481132075471,
      "grad_norm": 4.599055767059326,
      "learning_rate": 4.2144736830245356e-06,
      "loss": 0.299,
      "num_input_tokens_seen": 13291752,
      "step": 20190
    },
    {
      "epoch": 11.90742924528302,
      "grad_norm": 4.086666107177734,
      "learning_rate": 4.211933021328889e-06,
      "loss": 0.335,
      "num_input_tokens_seen": 13294696,
      "step": 20195
    },
    {
      "epoch": 11.910377358490566,
      "grad_norm": 2.9282591342926025,
      "learning_rate": 4.20939256827737e-06,
      "loss": 0.3452,
      "num_input_tokens_seen": 13298440,
      "step": 20200
    },
    {
      "epoch": 11.913325471698114,
      "grad_norm": 3.9769999980926514,
      "learning_rate": 4.206852324542578e-06,
      "loss": 0.3521,
      "num_input_tokens_seen": 13300808,
      "step": 20205
    },
    {
      "epoch": 11.91627358490566,
      "grad_norm": 3.25929594039917,
      "learning_rate": 4.2043122907970496e-06,
      "loss": 0.3458,
      "num_input_tokens_seen": 13305448,
      "step": 20210
    },
    {
      "epoch": 11.919221698113208,
      "grad_norm": 2.9742555618286133,
      "learning_rate": 4.2017724677132715e-06,
      "loss": 0.28,
      "num_input_tokens_seen": 13308968,
      "step": 20215
    },
    {
      "epoch": 11.922169811320755,
      "grad_norm": 4.91429328918457,
      "learning_rate": 4.1992328559636734e-06,
      "loss": 0.4215,
      "num_input_tokens_seen": 13311336,
      "step": 20220
    },
    {
      "epoch": 11.925117924528301,
      "grad_norm": 2.461606025695801,
      "learning_rate": 4.196693456220628e-06,
      "loss": 0.3134,
      "num_input_tokens_seen": 13313992,
      "step": 20225
    },
    {
      "epoch": 11.92806603773585,
      "grad_norm": 2.961885452270508,
      "learning_rate": 4.194154269156452e-06,
      "loss": 0.3342,
      "num_input_tokens_seen": 13318568,
      "step": 20230
    },
    {
      "epoch": 11.931014150943396,
      "grad_norm": 5.526954650878906,
      "learning_rate": 4.191615295443404e-06,
      "loss": 0.2755,
      "num_input_tokens_seen": 13322120,
      "step": 20235
    },
    {
      "epoch": 11.933962264150944,
      "grad_norm": 2.972928524017334,
      "learning_rate": 4.189076535753692e-06,
      "loss": 0.408,
      "num_input_tokens_seen": 13326024,
      "step": 20240
    },
    {
      "epoch": 11.93691037735849,
      "grad_norm": 2.980402946472168,
      "learning_rate": 4.186537990759464e-06,
      "loss": 0.353,
      "num_input_tokens_seen": 13329256,
      "step": 20245
    },
    {
      "epoch": 11.939858490566039,
      "grad_norm": 3.767648220062256,
      "learning_rate": 4.183999661132806e-06,
      "loss": 0.3288,
      "num_input_tokens_seen": 13332392,
      "step": 20250
    },
    {
      "epoch": 11.942806603773585,
      "grad_norm": 3.185074806213379,
      "learning_rate": 4.181461547545756e-06,
      "loss": 0.3394,
      "num_input_tokens_seen": 13335048,
      "step": 20255
    },
    {
      "epoch": 11.945754716981131,
      "grad_norm": 5.364509105682373,
      "learning_rate": 4.178923650670289e-06,
      "loss": 0.3645,
      "num_input_tokens_seen": 13337448,
      "step": 20260
    },
    {
      "epoch": 11.94870283018868,
      "grad_norm": 4.834733009338379,
      "learning_rate": 4.176385971178324e-06,
      "loss": 0.399,
      "num_input_tokens_seen": 13340360,
      "step": 20265
    },
    {
      "epoch": 11.951650943396226,
      "grad_norm": 2.5440657138824463,
      "learning_rate": 4.1738485097417225e-06,
      "loss": 0.3512,
      "num_input_tokens_seen": 13343624,
      "step": 20270
    },
    {
      "epoch": 11.954599056603774,
      "grad_norm": 3.327101230621338,
      "learning_rate": 4.1713112670322886e-06,
      "loss": 0.3277,
      "num_input_tokens_seen": 13346504,
      "step": 20275
    },
    {
      "epoch": 11.95754716981132,
      "grad_norm": 3.192842721939087,
      "learning_rate": 4.168774243721768e-06,
      "loss": 0.4813,
      "num_input_tokens_seen": 13349448,
      "step": 20280
    },
    {
      "epoch": 11.960495283018869,
      "grad_norm": 2.3087403774261475,
      "learning_rate": 4.166237440481849e-06,
      "loss": 0.4238,
      "num_input_tokens_seen": 13353352,
      "step": 20285
    },
    {
      "epoch": 11.963443396226415,
      "grad_norm": 2.307023763656616,
      "learning_rate": 4.163700857984162e-06,
      "loss": 0.3998,
      "num_input_tokens_seen": 13356168,
      "step": 20290
    },
    {
      "epoch": 11.966391509433961,
      "grad_norm": 2.5706050395965576,
      "learning_rate": 4.161164496900275e-06,
      "loss": 0.397,
      "num_input_tokens_seen": 13359240,
      "step": 20295
    },
    {
      "epoch": 11.96933962264151,
      "grad_norm": 3.9816160202026367,
      "learning_rate": 4.1586283579017036e-06,
      "loss": 0.3536,
      "num_input_tokens_seen": 13361608,
      "step": 20300
    },
    {
      "epoch": 11.972287735849056,
      "grad_norm": 2.7877213954925537,
      "learning_rate": 4.156092441659901e-06,
      "loss": 0.3228,
      "num_input_tokens_seen": 13364904,
      "step": 20305
    },
    {
      "epoch": 11.975235849056604,
      "grad_norm": 2.1165831089019775,
      "learning_rate": 4.153556748846261e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 13368456,
      "step": 20310
    },
    {
      "epoch": 11.97818396226415,
      "grad_norm": 2.5460431575775146,
      "learning_rate": 4.15102128013212e-06,
      "loss": 0.3552,
      "num_input_tokens_seen": 13372424,
      "step": 20315
    },
    {
      "epoch": 11.981132075471699,
      "grad_norm": 2.616804838180542,
      "learning_rate": 4.1484860361887544e-06,
      "loss": 0.415,
      "num_input_tokens_seen": 13376456,
      "step": 20320
    },
    {
      "epoch": 11.984080188679245,
      "grad_norm": 2.188920021057129,
      "learning_rate": 4.145951017687379e-06,
      "loss": 0.4113,
      "num_input_tokens_seen": 13379400,
      "step": 20325
    },
    {
      "epoch": 11.987028301886792,
      "grad_norm": 2.9708902835845947,
      "learning_rate": 4.1434162252991524e-06,
      "loss": 0.298,
      "num_input_tokens_seen": 13381768,
      "step": 20330
    },
    {
      "epoch": 11.98997641509434,
      "grad_norm": 2.6214592456817627,
      "learning_rate": 4.140881659695173e-06,
      "loss": 0.3688,
      "num_input_tokens_seen": 13389160,
      "step": 20335
    },
    {
      "epoch": 11.992924528301886,
      "grad_norm": 2.5089735984802246,
      "learning_rate": 4.138347321546477e-06,
      "loss": 0.3397,
      "num_input_tokens_seen": 13392104,
      "step": 20340
    },
    {
      "epoch": 11.995872641509434,
      "grad_norm": 2.2136952877044678,
      "learning_rate": 4.13581321152404e-06,
      "loss": 0.38,
      "num_input_tokens_seen": 13395240,
      "step": 20345
    },
    {
      "epoch": 11.99882075471698,
      "grad_norm": 5.251260280609131,
      "learning_rate": 4.133279330298781e-06,
      "loss": 0.4502,
      "num_input_tokens_seen": 13399304,
      "step": 20350
    },
    {
      "epoch": 12.0,
      "eval_loss": 0.5455325245857239,
      "eval_runtime": 18.4468,
      "eval_samples_per_second": 91.94,
      "eval_steps_per_second": 22.985,
      "num_input_tokens_seen": 13399864,
      "step": 20352
    },
    {
      "epoch": 12.001768867924529,
      "grad_norm": 2.724440813064575,
      "learning_rate": 4.130745678541555e-06,
      "loss": 0.3918,
      "num_input_tokens_seen": 13402104,
      "step": 20355
    },
    {
      "epoch": 12.004716981132075,
      "grad_norm": 2.755385160446167,
      "learning_rate": 4.128212256923155e-06,
      "loss": 0.2881,
      "num_input_tokens_seen": 13404728,
      "step": 20360
    },
    {
      "epoch": 12.007665094339623,
      "grad_norm": 3.3698441982269287,
      "learning_rate": 4.125679066114318e-06,
      "loss": 0.4624,
      "num_input_tokens_seen": 13408728,
      "step": 20365
    },
    {
      "epoch": 12.01061320754717,
      "grad_norm": 2.2034912109375,
      "learning_rate": 4.123146106785717e-06,
      "loss": 0.3383,
      "num_input_tokens_seen": 13411288,
      "step": 20370
    },
    {
      "epoch": 12.013561320754716,
      "grad_norm": 4.5424957275390625,
      "learning_rate": 4.1206133796079625e-06,
      "loss": 0.3917,
      "num_input_tokens_seen": 13415544,
      "step": 20375
    },
    {
      "epoch": 12.016509433962264,
      "grad_norm": 3.3364429473876953,
      "learning_rate": 4.1180808852516065e-06,
      "loss": 0.2445,
      "num_input_tokens_seen": 13418648,
      "step": 20380
    },
    {
      "epoch": 12.01945754716981,
      "grad_norm": 3.9468235969543457,
      "learning_rate": 4.115548624387136e-06,
      "loss": 0.415,
      "num_input_tokens_seen": 13422072,
      "step": 20385
    },
    {
      "epoch": 12.022405660377359,
      "grad_norm": 3.150576114654541,
      "learning_rate": 4.113016597684979e-06,
      "loss": 0.4026,
      "num_input_tokens_seen": 13424984,
      "step": 20390
    },
    {
      "epoch": 12.025353773584905,
      "grad_norm": 2.497067451477051,
      "learning_rate": 4.110484805815502e-06,
      "loss": 0.3644,
      "num_input_tokens_seen": 13427480,
      "step": 20395
    },
    {
      "epoch": 12.028301886792454,
      "grad_norm": 3.4397077560424805,
      "learning_rate": 4.107953249449005e-06,
      "loss": 0.3395,
      "num_input_tokens_seen": 13431192,
      "step": 20400
    },
    {
      "epoch": 12.03125,
      "grad_norm": 2.307830333709717,
      "learning_rate": 4.105421929255729e-06,
      "loss": 0.308,
      "num_input_tokens_seen": 13434936,
      "step": 20405
    },
    {
      "epoch": 12.034198113207546,
      "grad_norm": 3.0648703575134277,
      "learning_rate": 4.102890845905854e-06,
      "loss": 0.3309,
      "num_input_tokens_seen": 13437720,
      "step": 20410
    },
    {
      "epoch": 12.037146226415095,
      "grad_norm": 4.068867206573486,
      "learning_rate": 4.1003600000694935e-06,
      "loss": 0.4175,
      "num_input_tokens_seen": 13440440,
      "step": 20415
    },
    {
      "epoch": 12.040094339622641,
      "grad_norm": 2.6948561668395996,
      "learning_rate": 4.0978293924167e-06,
      "loss": 0.2912,
      "num_input_tokens_seen": 13445496,
      "step": 20420
    },
    {
      "epoch": 12.04304245283019,
      "grad_norm": 2.00643253326416,
      "learning_rate": 4.095299023617461e-06,
      "loss": 0.2865,
      "num_input_tokens_seen": 13448888,
      "step": 20425
    },
    {
      "epoch": 12.045990566037736,
      "grad_norm": 4.2179856300354,
      "learning_rate": 4.092768894341707e-06,
      "loss": 0.3225,
      "num_input_tokens_seen": 13451544,
      "step": 20430
    },
    {
      "epoch": 12.048938679245284,
      "grad_norm": 3.4564599990844727,
      "learning_rate": 4.090239005259298e-06,
      "loss": 0.4284,
      "num_input_tokens_seen": 13454648,
      "step": 20435
    },
    {
      "epoch": 12.05188679245283,
      "grad_norm": 7.077065467834473,
      "learning_rate": 4.087709357040033e-06,
      "loss": 0.3008,
      "num_input_tokens_seen": 13457464,
      "step": 20440
    },
    {
      "epoch": 12.054834905660377,
      "grad_norm": 2.4561595916748047,
      "learning_rate": 4.085179950353648e-06,
      "loss": 0.3506,
      "num_input_tokens_seen": 13460120,
      "step": 20445
    },
    {
      "epoch": 12.057783018867925,
      "grad_norm": 3.4686901569366455,
      "learning_rate": 4.0826507858698135e-06,
      "loss": 0.2364,
      "num_input_tokens_seen": 13463096,
      "step": 20450
    },
    {
      "epoch": 12.060731132075471,
      "grad_norm": 2.9564642906188965,
      "learning_rate": 4.080121864258136e-06,
      "loss": 0.3364,
      "num_input_tokens_seen": 13466136,
      "step": 20455
    },
    {
      "epoch": 12.06367924528302,
      "grad_norm": 3.7337093353271484,
      "learning_rate": 4.077593186188161e-06,
      "loss": 0.3335,
      "num_input_tokens_seen": 13469720,
      "step": 20460
    },
    {
      "epoch": 12.066627358490566,
      "grad_norm": 3.4187703132629395,
      "learning_rate": 4.075064752329364e-06,
      "loss": 0.417,
      "num_input_tokens_seen": 13472728,
      "step": 20465
    },
    {
      "epoch": 12.069575471698114,
      "grad_norm": 2.4854674339294434,
      "learning_rate": 4.0725365633511605e-06,
      "loss": 0.3535,
      "num_input_tokens_seen": 13476184,
      "step": 20470
    },
    {
      "epoch": 12.07252358490566,
      "grad_norm": 2.636674165725708,
      "learning_rate": 4.070008619922899e-06,
      "loss": 0.3399,
      "num_input_tokens_seen": 13479416,
      "step": 20475
    },
    {
      "epoch": 12.075471698113208,
      "grad_norm": 3.2502479553222656,
      "learning_rate": 4.067480922713864e-06,
      "loss": 0.4352,
      "num_input_tokens_seen": 13483704,
      "step": 20480
    },
    {
      "epoch": 12.078419811320755,
      "grad_norm": 6.644106864929199,
      "learning_rate": 4.064953472393273e-06,
      "loss": 0.3808,
      "num_input_tokens_seen": 13486040,
      "step": 20485
    },
    {
      "epoch": 12.081367924528301,
      "grad_norm": 5.161535739898682,
      "learning_rate": 4.06242626963028e-06,
      "loss": 0.3056,
      "num_input_tokens_seen": 13488856,
      "step": 20490
    },
    {
      "epoch": 12.08431603773585,
      "grad_norm": 4.811398983001709,
      "learning_rate": 4.059899315093972e-06,
      "loss": 0.2571,
      "num_input_tokens_seen": 13491288,
      "step": 20495
    },
    {
      "epoch": 12.087264150943396,
      "grad_norm": 3.977097749710083,
      "learning_rate": 4.057372609453374e-06,
      "loss": 0.5939,
      "num_input_tokens_seen": 13494296,
      "step": 20500
    },
    {
      "epoch": 12.090212264150944,
      "grad_norm": 1.9020870923995972,
      "learning_rate": 4.054846153377439e-06,
      "loss": 0.2981,
      "num_input_tokens_seen": 13497688,
      "step": 20505
    },
    {
      "epoch": 12.09316037735849,
      "grad_norm": 2.232905387878418,
      "learning_rate": 4.052319947535058e-06,
      "loss": 0.2914,
      "num_input_tokens_seen": 13500760,
      "step": 20510
    },
    {
      "epoch": 12.096108490566039,
      "grad_norm": 3.2454395294189453,
      "learning_rate": 4.049793992595056e-06,
      "loss": 0.5623,
      "num_input_tokens_seen": 13506936,
      "step": 20515
    },
    {
      "epoch": 12.099056603773585,
      "grad_norm": 3.8340814113616943,
      "learning_rate": 4.047268289226187e-06,
      "loss": 0.283,
      "num_input_tokens_seen": 13511096,
      "step": 20520
    },
    {
      "epoch": 12.102004716981131,
      "grad_norm": 2.5725975036621094,
      "learning_rate": 4.044742838097147e-06,
      "loss": 0.355,
      "num_input_tokens_seen": 13514168,
      "step": 20525
    },
    {
      "epoch": 12.10495283018868,
      "grad_norm": 3.5849738121032715,
      "learning_rate": 4.0422176398765564e-06,
      "loss": 0.2906,
      "num_input_tokens_seen": 13517656,
      "step": 20530
    },
    {
      "epoch": 12.107900943396226,
      "grad_norm": 3.4604034423828125,
      "learning_rate": 4.039692695232975e-06,
      "loss": 0.3926,
      "num_input_tokens_seen": 13521400,
      "step": 20535
    },
    {
      "epoch": 12.110849056603774,
      "grad_norm": 2.702260971069336,
      "learning_rate": 4.03716800483489e-06,
      "loss": 0.2696,
      "num_input_tokens_seen": 13525240,
      "step": 20540
    },
    {
      "epoch": 12.11379716981132,
      "grad_norm": 8.864147186279297,
      "learning_rate": 4.034643569350726e-06,
      "loss": 0.3555,
      "num_input_tokens_seen": 13528184,
      "step": 20545
    },
    {
      "epoch": 12.116745283018869,
      "grad_norm": 3.15940523147583,
      "learning_rate": 4.032119389448837e-06,
      "loss": 0.3308,
      "num_input_tokens_seen": 13531384,
      "step": 20550
    },
    {
      "epoch": 12.119693396226415,
      "grad_norm": 2.9877405166625977,
      "learning_rate": 4.0295954657975115e-06,
      "loss": 0.3624,
      "num_input_tokens_seen": 13535448,
      "step": 20555
    },
    {
      "epoch": 12.122641509433961,
      "grad_norm": 2.730006456375122,
      "learning_rate": 4.027071799064968e-06,
      "loss": 0.3538,
      "num_input_tokens_seen": 13538680,
      "step": 20560
    },
    {
      "epoch": 12.12558962264151,
      "grad_norm": 7.909416675567627,
      "learning_rate": 4.02454838991936e-06,
      "loss": 0.3277,
      "num_input_tokens_seen": 13541368,
      "step": 20565
    },
    {
      "epoch": 12.128537735849056,
      "grad_norm": 2.9686808586120605,
      "learning_rate": 4.022025239028768e-06,
      "loss": 0.2935,
      "num_input_tokens_seen": 13544472,
      "step": 20570
    },
    {
      "epoch": 12.131485849056604,
      "grad_norm": 3.3359463214874268,
      "learning_rate": 4.0195023470612095e-06,
      "loss": 0.2101,
      "num_input_tokens_seen": 13547480,
      "step": 20575
    },
    {
      "epoch": 12.13443396226415,
      "grad_norm": 3.4013454914093018,
      "learning_rate": 4.016979714684631e-06,
      "loss": 0.2765,
      "num_input_tokens_seen": 13550712,
      "step": 20580
    },
    {
      "epoch": 12.137382075471699,
      "grad_norm": 2.3605031967163086,
      "learning_rate": 4.014457342566909e-06,
      "loss": 0.2556,
      "num_input_tokens_seen": 13555800,
      "step": 20585
    },
    {
      "epoch": 12.140330188679245,
      "grad_norm": 8.674534797668457,
      "learning_rate": 4.011935231375853e-06,
      "loss": 0.3744,
      "num_input_tokens_seen": 13558296,
      "step": 20590
    },
    {
      "epoch": 12.143278301886792,
      "grad_norm": 7.818875312805176,
      "learning_rate": 4.009413381779203e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 13562584,
      "step": 20595
    },
    {
      "epoch": 12.14622641509434,
      "grad_norm": 2.79284405708313,
      "learning_rate": 4.00689179444463e-06,
      "loss": 0.2288,
      "num_input_tokens_seen": 13564632,
      "step": 20600
    },
    {
      "epoch": 12.149174528301886,
      "grad_norm": 4.616445541381836,
      "learning_rate": 4.004370470039733e-06,
      "loss": 0.3947,
      "num_input_tokens_seen": 13567160,
      "step": 20605
    },
    {
      "epoch": 12.152122641509434,
      "grad_norm": 2.695622205734253,
      "learning_rate": 4.001849409232046e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 13570808,
      "step": 20610
    },
    {
      "epoch": 12.15507075471698,
      "grad_norm": 3.521028995513916,
      "learning_rate": 3.9993286126890274e-06,
      "loss": 0.3488,
      "num_input_tokens_seen": 13574072,
      "step": 20615
    },
    {
      "epoch": 12.158018867924529,
      "grad_norm": 6.657188892364502,
      "learning_rate": 3.996808081078074e-06,
      "loss": 0.2963,
      "num_input_tokens_seen": 13578200,
      "step": 20620
    },
    {
      "epoch": 12.160966981132075,
      "grad_norm": 3.0878634452819824,
      "learning_rate": 3.9942878150665035e-06,
      "loss": 0.3547,
      "num_input_tokens_seen": 13581560,
      "step": 20625
    },
    {
      "epoch": 12.163915094339623,
      "grad_norm": 2.265911340713501,
      "learning_rate": 3.991767815321569e-06,
      "loss": 0.3796,
      "num_input_tokens_seen": 13584600,
      "step": 20630
    },
    {
      "epoch": 12.16686320754717,
      "grad_norm": 4.453630447387695,
      "learning_rate": 3.9892480825104504e-06,
      "loss": 0.2736,
      "num_input_tokens_seen": 13587800,
      "step": 20635
    },
    {
      "epoch": 12.169811320754716,
      "grad_norm": 3.7526326179504395,
      "learning_rate": 3.986728617300257e-06,
      "loss": 0.3343,
      "num_input_tokens_seen": 13590584,
      "step": 20640
    },
    {
      "epoch": 12.172759433962264,
      "grad_norm": 1.8279273509979248,
      "learning_rate": 3.984209420358031e-06,
      "loss": 0.2212,
      "num_input_tokens_seen": 13594040,
      "step": 20645
    },
    {
      "epoch": 12.17570754716981,
      "grad_norm": 2.1653671264648438,
      "learning_rate": 3.981690492350738e-06,
      "loss": 0.2849,
      "num_input_tokens_seen": 13596824,
      "step": 20650
    },
    {
      "epoch": 12.178655660377359,
      "grad_norm": 2.3601841926574707,
      "learning_rate": 3.979171833945276e-06,
      "loss": 0.3278,
      "num_input_tokens_seen": 13600248,
      "step": 20655
    },
    {
      "epoch": 12.181603773584905,
      "grad_norm": 2.318138360977173,
      "learning_rate": 3.97665344580847e-06,
      "loss": 0.4343,
      "num_input_tokens_seen": 13603544,
      "step": 20660
    },
    {
      "epoch": 12.184551886792454,
      "grad_norm": 3.1391801834106445,
      "learning_rate": 3.974135328607075e-06,
      "loss": 0.3486,
      "num_input_tokens_seen": 13607832,
      "step": 20665
    },
    {
      "epoch": 12.1875,
      "grad_norm": 4.183563709259033,
      "learning_rate": 3.971617483007773e-06,
      "loss": 0.4472,
      "num_input_tokens_seen": 13610968,
      "step": 20670
    },
    {
      "epoch": 12.190448113207546,
      "grad_norm": 2.3700196743011475,
      "learning_rate": 3.969099909677174e-06,
      "loss": 0.3344,
      "num_input_tokens_seen": 13613560,
      "step": 20675
    },
    {
      "epoch": 12.193396226415095,
      "grad_norm": 3.198573350906372,
      "learning_rate": 3.9665826092818165e-06,
      "loss": 0.2873,
      "num_input_tokens_seen": 13617176,
      "step": 20680
    },
    {
      "epoch": 12.196344339622641,
      "grad_norm": 3.715136766433716,
      "learning_rate": 3.964065582488168e-06,
      "loss": 0.4014,
      "num_input_tokens_seen": 13620408,
      "step": 20685
    },
    {
      "epoch": 12.19929245283019,
      "grad_norm": 3.3968021869659424,
      "learning_rate": 3.961548829962622e-06,
      "loss": 0.2578,
      "num_input_tokens_seen": 13623512,
      "step": 20690
    },
    {
      "epoch": 12.202240566037736,
      "grad_norm": 3.369189500808716,
      "learning_rate": 3.959032352371496e-06,
      "loss": 0.3586,
      "num_input_tokens_seen": 13626456,
      "step": 20695
    },
    {
      "epoch": 12.205188679245284,
      "grad_norm": 3.0138742923736572,
      "learning_rate": 3.956516150381043e-06,
      "loss": 0.379,
      "num_input_tokens_seen": 13629048,
      "step": 20700
    },
    {
      "epoch": 12.20813679245283,
      "grad_norm": 3.3599562644958496,
      "learning_rate": 3.954000224657436e-06,
      "loss": 0.353,
      "num_input_tokens_seen": 13631896,
      "step": 20705
    },
    {
      "epoch": 12.211084905660377,
      "grad_norm": 2.69208025932312,
      "learning_rate": 3.951484575866776e-06,
      "loss": 0.3479,
      "num_input_tokens_seen": 13634392,
      "step": 20710
    },
    {
      "epoch": 12.214033018867925,
      "grad_norm": 4.639314651489258,
      "learning_rate": 3.948969204675096e-06,
      "loss": 0.269,
      "num_input_tokens_seen": 13637240,
      "step": 20715
    },
    {
      "epoch": 12.216981132075471,
      "grad_norm": 2.4901840686798096,
      "learning_rate": 3.946454111748346e-06,
      "loss": 0.3036,
      "num_input_tokens_seen": 13639896,
      "step": 20720
    },
    {
      "epoch": 12.21992924528302,
      "grad_norm": 2.7769775390625,
      "learning_rate": 3.943939297752413e-06,
      "loss": 0.4327,
      "num_input_tokens_seen": 13642904,
      "step": 20725
    },
    {
      "epoch": 12.222877358490566,
      "grad_norm": 3.7667253017425537,
      "learning_rate": 3.9414247633531e-06,
      "loss": 0.2693,
      "num_input_tokens_seen": 13649240,
      "step": 20730
    },
    {
      "epoch": 12.225825471698114,
      "grad_norm": 3.5941853523254395,
      "learning_rate": 3.9389105092161454e-06,
      "loss": 0.3918,
      "num_input_tokens_seen": 13652568,
      "step": 20735
    },
    {
      "epoch": 12.22877358490566,
      "grad_norm": 4.662074565887451,
      "learning_rate": 3.936396536007205e-06,
      "loss": 0.2771,
      "num_input_tokens_seen": 13655768,
      "step": 20740
    },
    {
      "epoch": 12.231721698113208,
      "grad_norm": 3.9631381034851074,
      "learning_rate": 3.933882844391866e-06,
      "loss": 0.3925,
      "num_input_tokens_seen": 13658712,
      "step": 20745
    },
    {
      "epoch": 12.234669811320755,
      "grad_norm": 2.444359064102173,
      "learning_rate": 3.931369435035639e-06,
      "loss": 0.3587,
      "num_input_tokens_seen": 13661144,
      "step": 20750
    },
    {
      "epoch": 12.237617924528301,
      "grad_norm": 3.8669238090515137,
      "learning_rate": 3.92885630860396e-06,
      "loss": 0.3029,
      "num_input_tokens_seen": 13664472,
      "step": 20755
    },
    {
      "epoch": 12.24056603773585,
      "grad_norm": 4.58669376373291,
      "learning_rate": 3.926343465762189e-06,
      "loss": 0.381,
      "num_input_tokens_seen": 13667640,
      "step": 20760
    },
    {
      "epoch": 12.243514150943396,
      "grad_norm": 3.2335760593414307,
      "learning_rate": 3.923830907175613e-06,
      "loss": 0.4211,
      "num_input_tokens_seen": 13670488,
      "step": 20765
    },
    {
      "epoch": 12.246462264150944,
      "grad_norm": 4.269385814666748,
      "learning_rate": 3.921318633509442e-06,
      "loss": 0.474,
      "num_input_tokens_seen": 13672920,
      "step": 20770
    },
    {
      "epoch": 12.24941037735849,
      "grad_norm": 4.493213176727295,
      "learning_rate": 3.918806645428811e-06,
      "loss": 0.5038,
      "num_input_tokens_seen": 13676216,
      "step": 20775
    },
    {
      "epoch": 12.252358490566039,
      "grad_norm": 3.783764123916626,
      "learning_rate": 3.916294943598781e-06,
      "loss": 0.4584,
      "num_input_tokens_seen": 13679448,
      "step": 20780
    },
    {
      "epoch": 12.255306603773585,
      "grad_norm": 3.6430230140686035,
      "learning_rate": 3.913783528684336e-06,
      "loss": 0.3417,
      "num_input_tokens_seen": 13683096,
      "step": 20785
    },
    {
      "epoch": 12.258254716981131,
      "grad_norm": 1.7287626266479492,
      "learning_rate": 3.911272401350381e-06,
      "loss": 0.3367,
      "num_input_tokens_seen": 13686328,
      "step": 20790
    },
    {
      "epoch": 12.26120283018868,
      "grad_norm": 4.701297283172607,
      "learning_rate": 3.9087615622617525e-06,
      "loss": 0.3083,
      "num_input_tokens_seen": 13689496,
      "step": 20795
    },
    {
      "epoch": 12.264150943396226,
      "grad_norm": 2.6887338161468506,
      "learning_rate": 3.906251012083202e-06,
      "loss": 0.3437,
      "num_input_tokens_seen": 13692792,
      "step": 20800
    },
    {
      "epoch": 12.267099056603774,
      "grad_norm": 4.382042407989502,
      "learning_rate": 3.9037407514794085e-06,
      "loss": 0.5237,
      "num_input_tokens_seen": 13696056,
      "step": 20805
    },
    {
      "epoch": 12.27004716981132,
      "grad_norm": 3.841545820236206,
      "learning_rate": 3.901230781114976e-06,
      "loss": 0.3817,
      "num_input_tokens_seen": 13699608,
      "step": 20810
    },
    {
      "epoch": 12.272995283018869,
      "grad_norm": 5.231124401092529,
      "learning_rate": 3.898721101654431e-06,
      "loss": 0.4064,
      "num_input_tokens_seen": 13702232,
      "step": 20815
    },
    {
      "epoch": 12.275943396226415,
      "grad_norm": 4.505551815032959,
      "learning_rate": 3.896211713762221e-06,
      "loss": 0.526,
      "num_input_tokens_seen": 13705368,
      "step": 20820
    },
    {
      "epoch": 12.278891509433961,
      "grad_norm": 1.8062105178833008,
      "learning_rate": 3.893702618102715e-06,
      "loss": 0.2776,
      "num_input_tokens_seen": 13708696,
      "step": 20825
    },
    {
      "epoch": 12.28183962264151,
      "grad_norm": 2.5730252265930176,
      "learning_rate": 3.891193815340211e-06,
      "loss": 0.2624,
      "num_input_tokens_seen": 13712248,
      "step": 20830
    },
    {
      "epoch": 12.284787735849056,
      "grad_norm": 3.1488747596740723,
      "learning_rate": 3.888685306138922e-06,
      "loss": 0.3618,
      "num_input_tokens_seen": 13715000,
      "step": 20835
    },
    {
      "epoch": 12.287735849056604,
      "grad_norm": 4.67014217376709,
      "learning_rate": 3.886177091162987e-06,
      "loss": 0.4977,
      "num_input_tokens_seen": 13718520,
      "step": 20840
    },
    {
      "epoch": 12.29068396226415,
      "grad_norm": 2.9980976581573486,
      "learning_rate": 3.883669171076468e-06,
      "loss": 0.2921,
      "num_input_tokens_seen": 13721688,
      "step": 20845
    },
    {
      "epoch": 12.293632075471699,
      "grad_norm": 4.263103485107422,
      "learning_rate": 3.881161546543348e-06,
      "loss": 0.2632,
      "num_input_tokens_seen": 13724728,
      "step": 20850
    },
    {
      "epoch": 12.296580188679245,
      "grad_norm": 2.205002546310425,
      "learning_rate": 3.8786542182275295e-06,
      "loss": 0.3076,
      "num_input_tokens_seen": 13728888,
      "step": 20855
    },
    {
      "epoch": 12.299528301886792,
      "grad_norm": 3.6057634353637695,
      "learning_rate": 3.87614718679284e-06,
      "loss": 0.2329,
      "num_input_tokens_seen": 13732600,
      "step": 20860
    },
    {
      "epoch": 12.30247641509434,
      "grad_norm": 2.7665319442749023,
      "learning_rate": 3.8736404529030255e-06,
      "loss": 0.4568,
      "num_input_tokens_seen": 13737912,
      "step": 20865
    },
    {
      "epoch": 12.305424528301886,
      "grad_norm": 2.427044153213501,
      "learning_rate": 3.871134017221756e-06,
      "loss": 0.3721,
      "num_input_tokens_seen": 13742712,
      "step": 20870
    },
    {
      "epoch": 12.308372641509434,
      "grad_norm": 2.8023340702056885,
      "learning_rate": 3.868627880412621e-06,
      "loss": 0.2756,
      "num_input_tokens_seen": 13746168,
      "step": 20875
    },
    {
      "epoch": 12.31132075471698,
      "grad_norm": 2.2266931533813477,
      "learning_rate": 3.86612204313913e-06,
      "loss": 0.3901,
      "num_input_tokens_seen": 13749752,
      "step": 20880
    },
    {
      "epoch": 12.314268867924529,
      "grad_norm": 4.098459720611572,
      "learning_rate": 3.863616506064714e-06,
      "loss": 0.3804,
      "num_input_tokens_seen": 13752216,
      "step": 20885
    },
    {
      "epoch": 12.317216981132075,
      "grad_norm": 1.8928816318511963,
      "learning_rate": 3.861111269852727e-06,
      "loss": 0.3229,
      "num_input_tokens_seen": 13755480,
      "step": 20890
    },
    {
      "epoch": 12.320165094339623,
      "grad_norm": 4.0329461097717285,
      "learning_rate": 3.858606335166439e-06,
      "loss": 0.3774,
      "num_input_tokens_seen": 13759032,
      "step": 20895
    },
    {
      "epoch": 12.32311320754717,
      "grad_norm": 2.9302589893341064,
      "learning_rate": 3.8561017026690415e-06,
      "loss": 0.3099,
      "num_input_tokens_seen": 13761560,
      "step": 20900
    },
    {
      "epoch": 12.326061320754716,
      "grad_norm": 3.737579584121704,
      "learning_rate": 3.8535973730236495e-06,
      "loss": 0.342,
      "num_input_tokens_seen": 13764248,
      "step": 20905
    },
    {
      "epoch": 12.329009433962264,
      "grad_norm": 3.837052345275879,
      "learning_rate": 3.851093346893294e-06,
      "loss": 0.4026,
      "num_input_tokens_seen": 13767608,
      "step": 20910
    },
    {
      "epoch": 12.33195754716981,
      "grad_norm": 3.373927354812622,
      "learning_rate": 3.848589624940927e-06,
      "loss": 0.3981,
      "num_input_tokens_seen": 13771256,
      "step": 20915
    },
    {
      "epoch": 12.334905660377359,
      "grad_norm": 3.8167781829833984,
      "learning_rate": 3.846086207829417e-06,
      "loss": 0.3185,
      "num_input_tokens_seen": 13774648,
      "step": 20920
    },
    {
      "epoch": 12.337853773584905,
      "grad_norm": 2.6378438472747803,
      "learning_rate": 3.843583096221559e-06,
      "loss": 0.3138,
      "num_input_tokens_seen": 13778488,
      "step": 20925
    },
    {
      "epoch": 12.340801886792454,
      "grad_norm": 2.5384252071380615,
      "learning_rate": 3.8410802907800596e-06,
      "loss": 0.3687,
      "num_input_tokens_seen": 13781624,
      "step": 20930
    },
    {
      "epoch": 12.34375,
      "grad_norm": 2.972222328186035,
      "learning_rate": 3.838577792167546e-06,
      "loss": 0.5076,
      "num_input_tokens_seen": 13784568,
      "step": 20935
    },
    {
      "epoch": 12.346698113207546,
      "grad_norm": 1.8603423833847046,
      "learning_rate": 3.836075601046569e-06,
      "loss": 0.2932,
      "num_input_tokens_seen": 13788632,
      "step": 20940
    },
    {
      "epoch": 12.349646226415095,
      "grad_norm": 3.0411481857299805,
      "learning_rate": 3.833573718079594e-06,
      "loss": 0.374,
      "num_input_tokens_seen": 13791896,
      "step": 20945
    },
    {
      "epoch": 12.352594339622641,
      "grad_norm": 3.121147871017456,
      "learning_rate": 3.831072143929002e-06,
      "loss": 0.2401,
      "num_input_tokens_seen": 13795416,
      "step": 20950
    },
    {
      "epoch": 12.35554245283019,
      "grad_norm": 2.872634172439575,
      "learning_rate": 3.828570879257098e-06,
      "loss": 0.3482,
      "num_input_tokens_seen": 13797912,
      "step": 20955
    },
    {
      "epoch": 12.358490566037736,
      "grad_norm": 4.184272766113281,
      "learning_rate": 3.826069924726102e-06,
      "loss": 0.3678,
      "num_input_tokens_seen": 13801752,
      "step": 20960
    },
    {
      "epoch": 12.361438679245284,
      "grad_norm": 3.788057327270508,
      "learning_rate": 3.823569280998154e-06,
      "loss": 0.2997,
      "num_input_tokens_seen": 13809304,
      "step": 20965
    },
    {
      "epoch": 12.36438679245283,
      "grad_norm": 2.692173719406128,
      "learning_rate": 3.8210689487353095e-06,
      "loss": 0.3491,
      "num_input_tokens_seen": 13812280,
      "step": 20970
    },
    {
      "epoch": 12.367334905660377,
      "grad_norm": 2.4362316131591797,
      "learning_rate": 3.818568928599539e-06,
      "loss": 0.2966,
      "num_input_tokens_seen": 13817144,
      "step": 20975
    },
    {
      "epoch": 12.370283018867925,
      "grad_norm": 2.153895854949951,
      "learning_rate": 3.816069221252739e-06,
      "loss": 0.3286,
      "num_input_tokens_seen": 13821432,
      "step": 20980
    },
    {
      "epoch": 12.373231132075471,
      "grad_norm": 1.6125518083572388,
      "learning_rate": 3.813569827356715e-06,
      "loss": 0.3445,
      "num_input_tokens_seen": 13825016,
      "step": 20985
    },
    {
      "epoch": 12.37617924528302,
      "grad_norm": 3.53585147857666,
      "learning_rate": 3.811070747573191e-06,
      "loss": 0.3867,
      "num_input_tokens_seen": 13828312,
      "step": 20990
    },
    {
      "epoch": 12.379127358490566,
      "grad_norm": 3.7725484371185303,
      "learning_rate": 3.8085719825638098e-06,
      "loss": 0.453,
      "num_input_tokens_seen": 13832824,
      "step": 20995
    },
    {
      "epoch": 12.382075471698114,
      "grad_norm": 3.2140185832977295,
      "learning_rate": 3.806073532990132e-06,
      "loss": 0.3033,
      "num_input_tokens_seen": 13836120,
      "step": 21000
    },
    {
      "epoch": 12.38502358490566,
      "grad_norm": 2.7889668941497803,
      "learning_rate": 3.8035753995136333e-06,
      "loss": 0.3448,
      "num_input_tokens_seen": 13840216,
      "step": 21005
    },
    {
      "epoch": 12.387971698113208,
      "grad_norm": 4.629992961883545,
      "learning_rate": 3.801077582795702e-06,
      "loss": 0.4171,
      "num_input_tokens_seen": 13843288,
      "step": 21010
    },
    {
      "epoch": 12.390919811320755,
      "grad_norm": 3.1253862380981445,
      "learning_rate": 3.7985800834976494e-06,
      "loss": 0.4607,
      "num_input_tokens_seen": 13847256,
      "step": 21015
    },
    {
      "epoch": 12.393867924528301,
      "grad_norm": 3.134239673614502,
      "learning_rate": 3.7960829022806965e-06,
      "loss": 0.3428,
      "num_input_tokens_seen": 13850040,
      "step": 21020
    },
    {
      "epoch": 12.39681603773585,
      "grad_norm": 2.6707189083099365,
      "learning_rate": 3.793586039805984e-06,
      "loss": 0.3314,
      "num_input_tokens_seen": 13854104,
      "step": 21025
    },
    {
      "epoch": 12.399764150943396,
      "grad_norm": 3.0988571643829346,
      "learning_rate": 3.791089496734567e-06,
      "loss": 0.3496,
      "num_input_tokens_seen": 13856984,
      "step": 21030
    },
    {
      "epoch": 12.402712264150944,
      "grad_norm": 2.663912057876587,
      "learning_rate": 3.7885932737274163e-06,
      "loss": 0.2725,
      "num_input_tokens_seen": 13861144,
      "step": 21035
    },
    {
      "epoch": 12.40566037735849,
      "grad_norm": 4.721590995788574,
      "learning_rate": 3.7860973714454156e-06,
      "loss": 0.3929,
      "num_input_tokens_seen": 13864024,
      "step": 21040
    },
    {
      "epoch": 12.408608490566039,
      "grad_norm": 2.5282506942749023,
      "learning_rate": 3.7836017905493695e-06,
      "loss": 0.3058,
      "num_input_tokens_seen": 13867576,
      "step": 21045
    },
    {
      "epoch": 12.411556603773585,
      "grad_norm": 2.8556644916534424,
      "learning_rate": 3.7811065316999908e-06,
      "loss": 0.3554,
      "num_input_tokens_seen": 13871256,
      "step": 21050
    },
    {
      "epoch": 12.414504716981131,
      "grad_norm": 6.800180435180664,
      "learning_rate": 3.7786115955579105e-06,
      "loss": 0.3315,
      "num_input_tokens_seen": 13874904,
      "step": 21055
    },
    {
      "epoch": 12.41745283018868,
      "grad_norm": 4.650108337402344,
      "learning_rate": 3.7761169827836746e-06,
      "loss": 0.371,
      "num_input_tokens_seen": 13877624,
      "step": 21060
    },
    {
      "epoch": 12.420400943396226,
      "grad_norm": 5.165241241455078,
      "learning_rate": 3.773622694037743e-06,
      "loss": 0.3955,
      "num_input_tokens_seen": 13881560,
      "step": 21065
    },
    {
      "epoch": 12.423349056603774,
      "grad_norm": 3.8620200157165527,
      "learning_rate": 3.7711287299804865e-06,
      "loss": 0.3915,
      "num_input_tokens_seen": 13884760,
      "step": 21070
    },
    {
      "epoch": 12.42629716981132,
      "grad_norm": 2.9335944652557373,
      "learning_rate": 3.768635091272197e-06,
      "loss": 0.3324,
      "num_input_tokens_seen": 13887256,
      "step": 21075
    },
    {
      "epoch": 12.429245283018869,
      "grad_norm": 3.3839476108551025,
      "learning_rate": 3.7661417785730732e-06,
      "loss": 0.4246,
      "num_input_tokens_seen": 13890232,
      "step": 21080
    },
    {
      "epoch": 12.432193396226415,
      "grad_norm": 4.764308452606201,
      "learning_rate": 3.7636487925432304e-06,
      "loss": 0.3136,
      "num_input_tokens_seen": 13893400,
      "step": 21085
    },
    {
      "epoch": 12.435141509433961,
      "grad_norm": 1.8414232730865479,
      "learning_rate": 3.761156133842697e-06,
      "loss": 0.2962,
      "num_input_tokens_seen": 13896856,
      "step": 21090
    },
    {
      "epoch": 12.43808962264151,
      "grad_norm": 3.111112117767334,
      "learning_rate": 3.7586638031314182e-06,
      "loss": 0.3343,
      "num_input_tokens_seen": 13900568,
      "step": 21095
    },
    {
      "epoch": 12.441037735849056,
      "grad_norm": 2.8539090156555176,
      "learning_rate": 3.7561718010692477e-06,
      "loss": 0.5443,
      "num_input_tokens_seen": 13903576,
      "step": 21100
    },
    {
      "epoch": 12.443985849056604,
      "grad_norm": 4.2266154289245605,
      "learning_rate": 3.7536801283159523e-06,
      "loss": 0.3495,
      "num_input_tokens_seen": 13906296,
      "step": 21105
    },
    {
      "epoch": 12.44693396226415,
      "grad_norm": 3.6881165504455566,
      "learning_rate": 3.7511887855312155e-06,
      "loss": 0.4049,
      "num_input_tokens_seen": 13909752,
      "step": 21110
    },
    {
      "epoch": 12.449882075471699,
      "grad_norm": 3.0305981636047363,
      "learning_rate": 3.74869777337463e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 13912856,
      "step": 21115
    },
    {
      "epoch": 12.452830188679245,
      "grad_norm": 3.786229372024536,
      "learning_rate": 3.7462070925057004e-06,
      "loss": 0.3385,
      "num_input_tokens_seen": 13915640,
      "step": 21120
    },
    {
      "epoch": 12.455778301886792,
      "grad_norm": 3.096726417541504,
      "learning_rate": 3.7437167435838472e-06,
      "loss": 0.351,
      "num_input_tokens_seen": 13918904,
      "step": 21125
    },
    {
      "epoch": 12.45872641509434,
      "grad_norm": 4.561611175537109,
      "learning_rate": 3.741226727268401e-06,
      "loss": 0.3605,
      "num_input_tokens_seen": 13921624,
      "step": 21130
    },
    {
      "epoch": 12.461674528301886,
      "grad_norm": 2.395838737487793,
      "learning_rate": 3.7387370442186027e-06,
      "loss": 0.2984,
      "num_input_tokens_seen": 13924504,
      "step": 21135
    },
    {
      "epoch": 12.464622641509434,
      "grad_norm": 10.605175018310547,
      "learning_rate": 3.736247695093609e-06,
      "loss": 0.7629,
      "num_input_tokens_seen": 13928088,
      "step": 21140
    },
    {
      "epoch": 12.46757075471698,
      "grad_norm": 4.3378424644470215,
      "learning_rate": 3.7337586805524838e-06,
      "loss": 0.3161,
      "num_input_tokens_seen": 13930456,
      "step": 21145
    },
    {
      "epoch": 12.470518867924529,
      "grad_norm": 2.9740025997161865,
      "learning_rate": 3.731270001254205e-06,
      "loss": 0.291,
      "num_input_tokens_seen": 13934360,
      "step": 21150
    },
    {
      "epoch": 12.473466981132075,
      "grad_norm": 3.134476661682129,
      "learning_rate": 3.728781657857661e-06,
      "loss": 0.256,
      "num_input_tokens_seen": 13937752,
      "step": 21155
    },
    {
      "epoch": 12.476415094339623,
      "grad_norm": 2.7972285747528076,
      "learning_rate": 3.726293651021653e-06,
      "loss": 0.2928,
      "num_input_tokens_seen": 13941816,
      "step": 21160
    },
    {
      "epoch": 12.47936320754717,
      "grad_norm": 4.452069282531738,
      "learning_rate": 3.7238059814048888e-06,
      "loss": 0.3602,
      "num_input_tokens_seen": 13945400,
      "step": 21165
    },
    {
      "epoch": 12.482311320754716,
      "grad_norm": 2.927311658859253,
      "learning_rate": 3.7213186496659916e-06,
      "loss": 0.3461,
      "num_input_tokens_seen": 13948440,
      "step": 21170
    },
    {
      "epoch": 12.485259433962264,
      "grad_norm": 2.553675651550293,
      "learning_rate": 3.718831656463493e-06,
      "loss": 0.3034,
      "num_input_tokens_seen": 13952024,
      "step": 21175
    },
    {
      "epoch": 12.48820754716981,
      "grad_norm": 1.8201618194580078,
      "learning_rate": 3.7163450024558345e-06,
      "loss": 0.4156,
      "num_input_tokens_seen": 13955480,
      "step": 21180
    },
    {
      "epoch": 12.491155660377359,
      "grad_norm": 3.44057035446167,
      "learning_rate": 3.7138586883013694e-06,
      "loss": 0.3686,
      "num_input_tokens_seen": 13958456,
      "step": 21185
    },
    {
      "epoch": 12.494103773584905,
      "grad_norm": 4.204365253448486,
      "learning_rate": 3.7113727146583573e-06,
      "loss": 0.3435,
      "num_input_tokens_seen": 13961528,
      "step": 21190
    },
    {
      "epoch": 12.497051886792454,
      "grad_norm": 2.4280200004577637,
      "learning_rate": 3.7088870821849748e-06,
      "loss": 0.3313,
      "num_input_tokens_seen": 13965048,
      "step": 21195
    },
    {
      "epoch": 12.5,
      "grad_norm": 2.4901087284088135,
      "learning_rate": 3.706401791539301e-06,
      "loss": 0.4069,
      "num_input_tokens_seen": 13968888,
      "step": 21200
    },
    {
      "epoch": 12.502948113207546,
      "grad_norm": 2.792083501815796,
      "learning_rate": 3.703916843379328e-06,
      "loss": 0.5053,
      "num_input_tokens_seen": 13972056,
      "step": 21205
    },
    {
      "epoch": 12.505896226415095,
      "grad_norm": 6.957554340362549,
      "learning_rate": 3.7014322383629575e-06,
      "loss": 0.2706,
      "num_input_tokens_seen": 13974904,
      "step": 21210
    },
    {
      "epoch": 12.508844339622641,
      "grad_norm": 4.16129732131958,
      "learning_rate": 3.6989479771479976e-06,
      "loss": 0.3153,
      "num_input_tokens_seen": 13977752,
      "step": 21215
    },
    {
      "epoch": 12.51179245283019,
      "grad_norm": 2.5019354820251465,
      "learning_rate": 3.696464060392169e-06,
      "loss": 0.4234,
      "num_input_tokens_seen": 13980696,
      "step": 21220
    },
    {
      "epoch": 12.514740566037736,
      "grad_norm": 3.4191861152648926,
      "learning_rate": 3.6939804887530962e-06,
      "loss": 0.3982,
      "num_input_tokens_seen": 13985112,
      "step": 21225
    },
    {
      "epoch": 12.517688679245284,
      "grad_norm": 3.755082130432129,
      "learning_rate": 3.6914972628883196e-06,
      "loss": 0.3047,
      "num_input_tokens_seen": 13988696,
      "step": 21230
    },
    {
      "epoch": 12.52063679245283,
      "grad_norm": 3.841695785522461,
      "learning_rate": 3.6890143834552814e-06,
      "loss": 0.3805,
      "num_input_tokens_seen": 13993048,
      "step": 21235
    },
    {
      "epoch": 12.523584905660378,
      "grad_norm": 3.9239509105682373,
      "learning_rate": 3.6865318511113347e-06,
      "loss": 0.4184,
      "num_input_tokens_seen": 13996088,
      "step": 21240
    },
    {
      "epoch": 12.526533018867925,
      "grad_norm": 3.398409605026245,
      "learning_rate": 3.684049666513742e-06,
      "loss": 0.3745,
      "num_input_tokens_seen": 13999512,
      "step": 21245
    },
    {
      "epoch": 12.529481132075471,
      "grad_norm": 4.528263568878174,
      "learning_rate": 3.6815678303196715e-06,
      "loss": 0.3271,
      "num_input_tokens_seen": 14003512,
      "step": 21250
    },
    {
      "epoch": 12.53242924528302,
      "grad_norm": 2.8586881160736084,
      "learning_rate": 3.6790863431861988e-06,
      "loss": 0.388,
      "num_input_tokens_seen": 14007416,
      "step": 21255
    },
    {
      "epoch": 12.535377358490566,
      "grad_norm": 3.965280055999756,
      "learning_rate": 3.676605205770311e-06,
      "loss": 0.4144,
      "num_input_tokens_seen": 14010456,
      "step": 21260
    },
    {
      "epoch": 12.538325471698114,
      "grad_norm": 3.1216819286346436,
      "learning_rate": 3.674124418728898e-06,
      "loss": 0.2471,
      "num_input_tokens_seen": 14016792,
      "step": 21265
    },
    {
      "epoch": 12.54127358490566,
      "grad_norm": 2.9451146125793457,
      "learning_rate": 3.671643982718759e-06,
      "loss": 0.4034,
      "num_input_tokens_seen": 14019864,
      "step": 21270
    },
    {
      "epoch": 12.544221698113208,
      "grad_norm": 15.006597518920898,
      "learning_rate": 3.6691638983966017e-06,
      "loss": 0.4853,
      "num_input_tokens_seen": 14023320,
      "step": 21275
    },
    {
      "epoch": 12.547169811320755,
      "grad_norm": 2.6291091442108154,
      "learning_rate": 3.666684166419037e-06,
      "loss": 0.3578,
      "num_input_tokens_seen": 14026424,
      "step": 21280
    },
    {
      "epoch": 12.550117924528301,
      "grad_norm": 3.7135708332061768,
      "learning_rate": 3.6642047874425833e-06,
      "loss": 0.2375,
      "num_input_tokens_seen": 14029144,
      "step": 21285
    },
    {
      "epoch": 12.55306603773585,
      "grad_norm": 2.010483741760254,
      "learning_rate": 3.661725762123671e-06,
      "loss": 0.2837,
      "num_input_tokens_seen": 14031992,
      "step": 21290
    },
    {
      "epoch": 12.556014150943396,
      "grad_norm": 2.4940342903137207,
      "learning_rate": 3.659247091118631e-06,
      "loss": 0.2455,
      "num_input_tokens_seen": 14034744,
      "step": 21295
    },
    {
      "epoch": 12.558962264150944,
      "grad_norm": 2.024609327316284,
      "learning_rate": 3.6567687750837027e-06,
      "loss": 0.3363,
      "num_input_tokens_seen": 14037432,
      "step": 21300
    },
    {
      "epoch": 12.56191037735849,
      "grad_norm": 2.4837429523468018,
      "learning_rate": 3.6542908146750287e-06,
      "loss": 0.3465,
      "num_input_tokens_seen": 14040952,
      "step": 21305
    },
    {
      "epoch": 12.564858490566039,
      "grad_norm": 8.924711227416992,
      "learning_rate": 3.6518132105486624e-06,
      "loss": 0.367,
      "num_input_tokens_seen": 14043736,
      "step": 21310
    },
    {
      "epoch": 12.567806603773585,
      "grad_norm": 3.1269078254699707,
      "learning_rate": 3.649335963360559e-06,
      "loss": 0.2865,
      "num_input_tokens_seen": 14047256,
      "step": 21315
    },
    {
      "epoch": 12.570754716981131,
      "grad_norm": 2.704651117324829,
      "learning_rate": 3.6468590737665795e-06,
      "loss": 0.5007,
      "num_input_tokens_seen": 14049912,
      "step": 21320
    },
    {
      "epoch": 12.57370283018868,
      "grad_norm": 6.8434600830078125,
      "learning_rate": 3.6443825424224926e-06,
      "loss": 0.266,
      "num_input_tokens_seen": 14052504,
      "step": 21325
    },
    {
      "epoch": 12.576650943396226,
      "grad_norm": 6.765328884124756,
      "learning_rate": 3.64190636998397e-06,
      "loss": 0.3872,
      "num_input_tokens_seen": 14055064,
      "step": 21330
    },
    {
      "epoch": 12.579599056603774,
      "grad_norm": 4.76124382019043,
      "learning_rate": 3.639430557106588e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 14058008,
      "step": 21335
    },
    {
      "epoch": 12.58254716981132,
      "grad_norm": 3.0980629920959473,
      "learning_rate": 3.6369551044458314e-06,
      "loss": 0.3605,
      "num_input_tokens_seen": 14060856,
      "step": 21340
    },
    {
      "epoch": 12.585495283018869,
      "grad_norm": 2.23439359664917,
      "learning_rate": 3.6344800126570846e-06,
      "loss": 0.3589,
      "num_input_tokens_seen": 14064344,
      "step": 21345
    },
    {
      "epoch": 12.588443396226415,
      "grad_norm": 2.4385955333709717,
      "learning_rate": 3.6320052823956385e-06,
      "loss": 0.2627,
      "num_input_tokens_seen": 14067576,
      "step": 21350
    },
    {
      "epoch": 12.591391509433961,
      "grad_norm": 2.958765745162964,
      "learning_rate": 3.6295309143166906e-06,
      "loss": 0.3252,
      "num_input_tokens_seen": 14072024,
      "step": 21355
    },
    {
      "epoch": 12.59433962264151,
      "grad_norm": 1.8703536987304688,
      "learning_rate": 3.62705690907534e-06,
      "loss": 0.3495,
      "num_input_tokens_seen": 14075576,
      "step": 21360
    },
    {
      "epoch": 12.597287735849056,
      "grad_norm": 2.8979060649871826,
      "learning_rate": 3.624583267326588e-06,
      "loss": 0.2777,
      "num_input_tokens_seen": 14080312,
      "step": 21365
    },
    {
      "epoch": 12.600235849056604,
      "grad_norm": 3.162487030029297,
      "learning_rate": 3.6221099897253454e-06,
      "loss": 0.3433,
      "num_input_tokens_seen": 14083896,
      "step": 21370
    },
    {
      "epoch": 12.60318396226415,
      "grad_norm": 3.5749330520629883,
      "learning_rate": 3.619637076926421e-06,
      "loss": 0.2736,
      "num_input_tokens_seen": 14087960,
      "step": 21375
    },
    {
      "epoch": 12.606132075471699,
      "grad_norm": 2.8513612747192383,
      "learning_rate": 3.617164529584528e-06,
      "loss": 0.2839,
      "num_input_tokens_seen": 14090264,
      "step": 21380
    },
    {
      "epoch": 12.609080188679245,
      "grad_norm": 3.483586311340332,
      "learning_rate": 3.614692348354286e-06,
      "loss": 0.4146,
      "num_input_tokens_seen": 14093560,
      "step": 21385
    },
    {
      "epoch": 12.612028301886792,
      "grad_norm": 2.8989508152008057,
      "learning_rate": 3.612220533890216e-06,
      "loss": 0.3498,
      "num_input_tokens_seen": 14096472,
      "step": 21390
    },
    {
      "epoch": 12.61497641509434,
      "grad_norm": 2.1048102378845215,
      "learning_rate": 3.609749086846741e-06,
      "loss": 0.3698,
      "num_input_tokens_seen": 14100312,
      "step": 21395
    },
    {
      "epoch": 12.617924528301886,
      "grad_norm": 2.142822265625,
      "learning_rate": 3.607278007878186e-06,
      "loss": 0.2872,
      "num_input_tokens_seen": 14102616,
      "step": 21400
    },
    {
      "epoch": 12.620872641509434,
      "grad_norm": 2.0913431644439697,
      "learning_rate": 3.6048072976387817e-06,
      "loss": 0.314,
      "num_input_tokens_seen": 14105400,
      "step": 21405
    },
    {
      "epoch": 12.62382075471698,
      "grad_norm": 2.4466865062713623,
      "learning_rate": 3.6023369567826585e-06,
      "loss": 0.405,
      "num_input_tokens_seen": 14108504,
      "step": 21410
    },
    {
      "epoch": 12.626768867924529,
      "grad_norm": 3.1835923194885254,
      "learning_rate": 3.599866985963848e-06,
      "loss": 0.2865,
      "num_input_tokens_seen": 14111384,
      "step": 21415
    },
    {
      "epoch": 12.629716981132075,
      "grad_norm": 2.614841938018799,
      "learning_rate": 3.5973973858362885e-06,
      "loss": 0.283,
      "num_input_tokens_seen": 14114232,
      "step": 21420
    },
    {
      "epoch": 12.632665094339622,
      "grad_norm": 3.1119422912597656,
      "learning_rate": 3.594928157053816e-06,
      "loss": 0.2466,
      "num_input_tokens_seen": 14116888,
      "step": 21425
    },
    {
      "epoch": 12.63561320754717,
      "grad_norm": 2.9490344524383545,
      "learning_rate": 3.592459300270168e-06,
      "loss": 0.2654,
      "num_input_tokens_seen": 14119736,
      "step": 21430
    },
    {
      "epoch": 12.638561320754716,
      "grad_norm": 6.127936363220215,
      "learning_rate": 3.589990816138988e-06,
      "loss": 0.2769,
      "num_input_tokens_seen": 14122392,
      "step": 21435
    },
    {
      "epoch": 12.641509433962264,
      "grad_norm": 6.324710369110107,
      "learning_rate": 3.587522705313816e-06,
      "loss": 0.3971,
      "num_input_tokens_seen": 14125176,
      "step": 21440
    },
    {
      "epoch": 12.64445754716981,
      "grad_norm": 5.9515557289123535,
      "learning_rate": 3.585054968448094e-06,
      "loss": 0.3158,
      "num_input_tokens_seen": 14127384,
      "step": 21445
    },
    {
      "epoch": 12.647405660377359,
      "grad_norm": 3.0053892135620117,
      "learning_rate": 3.5825876061951686e-06,
      "loss": 0.3986,
      "num_input_tokens_seen": 14130840,
      "step": 21450
    },
    {
      "epoch": 12.650353773584905,
      "grad_norm": 4.570580959320068,
      "learning_rate": 3.5801206192082818e-06,
      "loss": 0.4326,
      "num_input_tokens_seen": 14133912,
      "step": 21455
    },
    {
      "epoch": 12.653301886792454,
      "grad_norm": 4.853326320648193,
      "learning_rate": 3.577654008140582e-06,
      "loss": 0.3284,
      "num_input_tokens_seen": 14137368,
      "step": 21460
    },
    {
      "epoch": 12.65625,
      "grad_norm": 2.7922213077545166,
      "learning_rate": 3.5751877736451123e-06,
      "loss": 0.2731,
      "num_input_tokens_seen": 14139896,
      "step": 21465
    },
    {
      "epoch": 12.659198113207546,
      "grad_norm": 3.358949899673462,
      "learning_rate": 3.5727219163748205e-06,
      "loss": 0.3394,
      "num_input_tokens_seen": 14142776,
      "step": 21470
    },
    {
      "epoch": 12.662146226415095,
      "grad_norm": 2.7493715286254883,
      "learning_rate": 3.570256436982552e-06,
      "loss": 0.3508,
      "num_input_tokens_seen": 14145720,
      "step": 21475
    },
    {
      "epoch": 12.665094339622641,
      "grad_norm": 3.3734424114227295,
      "learning_rate": 3.5677913361210536e-06,
      "loss": 0.2604,
      "num_input_tokens_seen": 14148632,
      "step": 21480
    },
    {
      "epoch": 12.66804245283019,
      "grad_norm": 5.397823333740234,
      "learning_rate": 3.565326614442972e-06,
      "loss": 0.2747,
      "num_input_tokens_seen": 14151320,
      "step": 21485
    },
    {
      "epoch": 12.670990566037736,
      "grad_norm": 2.5809326171875,
      "learning_rate": 3.5628622726008523e-06,
      "loss": 0.4355,
      "num_input_tokens_seen": 14153976,
      "step": 21490
    },
    {
      "epoch": 12.673938679245284,
      "grad_norm": 3.4694013595581055,
      "learning_rate": 3.56039831124714e-06,
      "loss": 0.3607,
      "num_input_tokens_seen": 14156728,
      "step": 21495
    },
    {
      "epoch": 12.67688679245283,
      "grad_norm": 3.826706647872925,
      "learning_rate": 3.557934731034179e-06,
      "loss": 0.3535,
      "num_input_tokens_seen": 14159640,
      "step": 21500
    },
    {
      "epoch": 12.679834905660378,
      "grad_norm": 3.2386560440063477,
      "learning_rate": 3.5554715326142126e-06,
      "loss": 0.2289,
      "num_input_tokens_seen": 14162968,
      "step": 21505
    },
    {
      "epoch": 12.682783018867925,
      "grad_norm": 4.162199974060059,
      "learning_rate": 3.553008716639384e-06,
      "loss": 0.5167,
      "num_input_tokens_seen": 14165784,
      "step": 21510
    },
    {
      "epoch": 12.685731132075471,
      "grad_norm": 3.149980306625366,
      "learning_rate": 3.5505462837617338e-06,
      "loss": 0.2372,
      "num_input_tokens_seen": 14169400,
      "step": 21515
    },
    {
      "epoch": 12.68867924528302,
      "grad_norm": 2.82551908493042,
      "learning_rate": 3.5480842346332013e-06,
      "loss": 0.2287,
      "num_input_tokens_seen": 14172792,
      "step": 21520
    },
    {
      "epoch": 12.691627358490566,
      "grad_norm": 3.4468936920166016,
      "learning_rate": 3.5456225699056256e-06,
      "loss": 0.3314,
      "num_input_tokens_seen": 14176920,
      "step": 21525
    },
    {
      "epoch": 12.694575471698114,
      "grad_norm": 3.3566982746124268,
      "learning_rate": 3.5431612902307426e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 14180120,
      "step": 21530
    },
    {
      "epoch": 12.69752358490566,
      "grad_norm": 4.8324809074401855,
      "learning_rate": 3.540700396260186e-06,
      "loss": 0.4488,
      "num_input_tokens_seen": 14183416,
      "step": 21535
    },
    {
      "epoch": 12.700471698113208,
      "grad_norm": 4.1219305992126465,
      "learning_rate": 3.538239888645489e-06,
      "loss": 0.4419,
      "num_input_tokens_seen": 14186392,
      "step": 21540
    },
    {
      "epoch": 12.703419811320755,
      "grad_norm": 4.278482913970947,
      "learning_rate": 3.535779768038082e-06,
      "loss": 0.459,
      "num_input_tokens_seen": 14189208,
      "step": 21545
    },
    {
      "epoch": 12.706367924528301,
      "grad_norm": 4.047165870666504,
      "learning_rate": 3.5333200350892905e-06,
      "loss": 0.4566,
      "num_input_tokens_seen": 14192984,
      "step": 21550
    },
    {
      "epoch": 12.70931603773585,
      "grad_norm": 2.213069200515747,
      "learning_rate": 3.530860690450342e-06,
      "loss": 0.3629,
      "num_input_tokens_seen": 14197080,
      "step": 21555
    },
    {
      "epoch": 12.712264150943396,
      "grad_norm": 3.375910520553589,
      "learning_rate": 3.528401734772357e-06,
      "loss": 0.3295,
      "num_input_tokens_seen": 14199256,
      "step": 21560
    },
    {
      "epoch": 12.715212264150944,
      "grad_norm": 2.663362979888916,
      "learning_rate": 3.5259431687063538e-06,
      "loss": 0.2552,
      "num_input_tokens_seen": 14201976,
      "step": 21565
    },
    {
      "epoch": 12.71816037735849,
      "grad_norm": 2.862776517868042,
      "learning_rate": 3.523484992903249e-06,
      "loss": 0.2791,
      "num_input_tokens_seen": 14205592,
      "step": 21570
    },
    {
      "epoch": 12.721108490566039,
      "grad_norm": 2.2752132415771484,
      "learning_rate": 3.5210272080138573e-06,
      "loss": 0.384,
      "num_input_tokens_seen": 14209464,
      "step": 21575
    },
    {
      "epoch": 12.724056603773585,
      "grad_norm": 1.9730361700057983,
      "learning_rate": 3.518569814688887e-06,
      "loss": 0.4105,
      "num_input_tokens_seen": 14212728,
      "step": 21580
    },
    {
      "epoch": 12.727004716981131,
      "grad_norm": 4.6072845458984375,
      "learning_rate": 3.5161128135789414e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 14216312,
      "step": 21585
    },
    {
      "epoch": 12.72995283018868,
      "grad_norm": 3.233093023300171,
      "learning_rate": 3.513656205334525e-06,
      "loss": 0.3197,
      "num_input_tokens_seen": 14218968,
      "step": 21590
    },
    {
      "epoch": 12.732900943396226,
      "grad_norm": 3.052339792251587,
      "learning_rate": 3.5111999906060336e-06,
      "loss": 0.2827,
      "num_input_tokens_seen": 14222328,
      "step": 21595
    },
    {
      "epoch": 12.735849056603774,
      "grad_norm": 3.6037604808807373,
      "learning_rate": 3.50874417004376e-06,
      "loss": 0.4675,
      "num_input_tokens_seen": 14225528,
      "step": 21600
    },
    {
      "epoch": 12.73879716981132,
      "grad_norm": 3.957268238067627,
      "learning_rate": 3.5062887442978956e-06,
      "loss": 0.3047,
      "num_input_tokens_seen": 14228408,
      "step": 21605
    },
    {
      "epoch": 12.741745283018869,
      "grad_norm": 2.296881675720215,
      "learning_rate": 3.503833714018524e-06,
      "loss": 0.297,
      "num_input_tokens_seen": 14232312,
      "step": 21610
    },
    {
      "epoch": 12.744693396226415,
      "grad_norm": 6.0509514808654785,
      "learning_rate": 3.5013790798556228e-06,
      "loss": 0.3918,
      "num_input_tokens_seen": 14235928,
      "step": 21615
    },
    {
      "epoch": 12.747641509433961,
      "grad_norm": 3.3210034370422363,
      "learning_rate": 3.4989248424590705e-06,
      "loss": 0.3303,
      "num_input_tokens_seen": 14238936,
      "step": 21620
    },
    {
      "epoch": 12.75058962264151,
      "grad_norm": 2.000312089920044,
      "learning_rate": 3.4964710024786354e-06,
      "loss": 0.2812,
      "num_input_tokens_seen": 14241656,
      "step": 21625
    },
    {
      "epoch": 12.753537735849056,
      "grad_norm": 3.5397191047668457,
      "learning_rate": 3.4940175605639813e-06,
      "loss": 0.2856,
      "num_input_tokens_seen": 14244248,
      "step": 21630
    },
    {
      "epoch": 12.756485849056604,
      "grad_norm": 10.677297592163086,
      "learning_rate": 3.4915645173646694e-06,
      "loss": 0.3195,
      "num_input_tokens_seen": 14247224,
      "step": 21635
    },
    {
      "epoch": 12.75943396226415,
      "grad_norm": 3.2434239387512207,
      "learning_rate": 3.489111873530153e-06,
      "loss": 0.315,
      "num_input_tokens_seen": 14250712,
      "step": 21640
    },
    {
      "epoch": 12.762382075471699,
      "grad_norm": 3.319376230239868,
      "learning_rate": 3.4866596297097776e-06,
      "loss": 0.3571,
      "num_input_tokens_seen": 14254648,
      "step": 21645
    },
    {
      "epoch": 12.765330188679245,
      "grad_norm": 2.9902639389038086,
      "learning_rate": 3.484207786552789e-06,
      "loss": 0.3032,
      "num_input_tokens_seen": 14258648,
      "step": 21650
    },
    {
      "epoch": 12.768278301886792,
      "grad_norm": 1.3031842708587646,
      "learning_rate": 3.4817563447083214e-06,
      "loss": 0.3669,
      "num_input_tokens_seen": 14262104,
      "step": 21655
    },
    {
      "epoch": 12.77122641509434,
      "grad_norm": 5.512714862823486,
      "learning_rate": 3.4793053048254044e-06,
      "loss": 0.3216,
      "num_input_tokens_seen": 14265336,
      "step": 21660
    },
    {
      "epoch": 12.774174528301886,
      "grad_norm": 5.991628646850586,
      "learning_rate": 3.47685466755296e-06,
      "loss": 0.3263,
      "num_input_tokens_seen": 14268120,
      "step": 21665
    },
    {
      "epoch": 12.777122641509434,
      "grad_norm": 2.2153561115264893,
      "learning_rate": 3.474404433539809e-06,
      "loss": 0.5095,
      "num_input_tokens_seen": 14271032,
      "step": 21670
    },
    {
      "epoch": 12.78007075471698,
      "grad_norm": 2.461944341659546,
      "learning_rate": 3.4719546034346598e-06,
      "loss": 0.3365,
      "num_input_tokens_seen": 14274584,
      "step": 21675
    },
    {
      "epoch": 12.783018867924529,
      "grad_norm": 3.683790445327759,
      "learning_rate": 3.4695051778861125e-06,
      "loss": 0.4179,
      "num_input_tokens_seen": 14277656,
      "step": 21680
    },
    {
      "epoch": 12.785966981132075,
      "grad_norm": 2.2377374172210693,
      "learning_rate": 3.4670561575426677e-06,
      "loss": 0.2655,
      "num_input_tokens_seen": 14282296,
      "step": 21685
    },
    {
      "epoch": 12.788915094339622,
      "grad_norm": 3.43609356880188,
      "learning_rate": 3.4646075430527115e-06,
      "loss": 0.3898,
      "num_input_tokens_seen": 14285624,
      "step": 21690
    },
    {
      "epoch": 12.79186320754717,
      "grad_norm": 4.267385959625244,
      "learning_rate": 3.4621593350645236e-06,
      "loss": 0.3078,
      "num_input_tokens_seen": 14288760,
      "step": 21695
    },
    {
      "epoch": 12.794811320754716,
      "grad_norm": 3.2280824184417725,
      "learning_rate": 3.4597115342262817e-06,
      "loss": 0.3469,
      "num_input_tokens_seen": 14292760,
      "step": 21700
    },
    {
      "epoch": 12.797759433962264,
      "grad_norm": 5.705853462219238,
      "learning_rate": 3.4572641411860484e-06,
      "loss": 0.3893,
      "num_input_tokens_seen": 14295672,
      "step": 21705
    },
    {
      "epoch": 12.80070754716981,
      "grad_norm": 3.2364864349365234,
      "learning_rate": 3.454817156591782e-06,
      "loss": 0.4166,
      "num_input_tokens_seen": 14298904,
      "step": 21710
    },
    {
      "epoch": 12.803655660377359,
      "grad_norm": 3.915583372116089,
      "learning_rate": 3.4523705810913344e-06,
      "loss": 0.2914,
      "num_input_tokens_seen": 14302136,
      "step": 21715
    },
    {
      "epoch": 12.806603773584905,
      "grad_norm": 3.8757832050323486,
      "learning_rate": 3.449924415332443e-06,
      "loss": 0.2723,
      "num_input_tokens_seen": 14305560,
      "step": 21720
    },
    {
      "epoch": 12.809551886792454,
      "grad_norm": 4.785258769989014,
      "learning_rate": 3.447478659962745e-06,
      "loss": 0.2802,
      "num_input_tokens_seen": 14308920,
      "step": 21725
    },
    {
      "epoch": 12.8125,
      "grad_norm": 4.441842555999756,
      "learning_rate": 3.4450333156297625e-06,
      "loss": 0.3263,
      "num_input_tokens_seen": 14311384,
      "step": 21730
    },
    {
      "epoch": 12.815448113207546,
      "grad_norm": 6.176799774169922,
      "learning_rate": 3.44258838298091e-06,
      "loss": 0.5398,
      "num_input_tokens_seen": 14314840,
      "step": 21735
    },
    {
      "epoch": 12.818396226415095,
      "grad_norm": 3.01104998588562,
      "learning_rate": 3.440143862663497e-06,
      "loss": 0.288,
      "num_input_tokens_seen": 14317784,
      "step": 21740
    },
    {
      "epoch": 12.821344339622641,
      "grad_norm": 4.722825527191162,
      "learning_rate": 3.4376997553247183e-06,
      "loss": 0.3975,
      "num_input_tokens_seen": 14320792,
      "step": 21745
    },
    {
      "epoch": 12.82429245283019,
      "grad_norm": 3.649303674697876,
      "learning_rate": 3.4352560616116617e-06,
      "loss": 0.3337,
      "num_input_tokens_seen": 14324280,
      "step": 21750
    },
    {
      "epoch": 12.827240566037736,
      "grad_norm": 3.7351903915405273,
      "learning_rate": 3.4328127821713077e-06,
      "loss": 0.3133,
      "num_input_tokens_seen": 14327448,
      "step": 21755
    },
    {
      "epoch": 12.830188679245284,
      "grad_norm": 2.845050096511841,
      "learning_rate": 3.430369917650521e-06,
      "loss": 0.3709,
      "num_input_tokens_seen": 14330648,
      "step": 21760
    },
    {
      "epoch": 12.83313679245283,
      "grad_norm": 3.7107129096984863,
      "learning_rate": 3.427927468696066e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 14333528,
      "step": 21765
    },
    {
      "epoch": 12.836084905660378,
      "grad_norm": 3.870365619659424,
      "learning_rate": 3.425485435954588e-06,
      "loss": 0.3723,
      "num_input_tokens_seen": 14337176,
      "step": 21770
    },
    {
      "epoch": 12.839033018867925,
      "grad_norm": 3.017430543899536,
      "learning_rate": 3.4230438200726274e-06,
      "loss": 0.3806,
      "num_input_tokens_seen": 14340408,
      "step": 21775
    },
    {
      "epoch": 12.841981132075471,
      "grad_norm": 5.459214210510254,
      "learning_rate": 3.4206026216966113e-06,
      "loss": 0.3038,
      "num_input_tokens_seen": 14343672,
      "step": 21780
    },
    {
      "epoch": 12.84492924528302,
      "grad_norm": 2.906524419784546,
      "learning_rate": 3.418161841472858e-06,
      "loss": 0.3872,
      "num_input_tokens_seen": 14347032,
      "step": 21785
    },
    {
      "epoch": 12.847877358490566,
      "grad_norm": 2.657097101211548,
      "learning_rate": 3.4157214800475746e-06,
      "loss": 0.3607,
      "num_input_tokens_seen": 14349976,
      "step": 21790
    },
    {
      "epoch": 12.850825471698114,
      "grad_norm": 1.8719127178192139,
      "learning_rate": 3.4132815380668577e-06,
      "loss": 0.4601,
      "num_input_tokens_seen": 14352920,
      "step": 21795
    },
    {
      "epoch": 12.85377358490566,
      "grad_norm": 4.124464511871338,
      "learning_rate": 3.410842016176691e-06,
      "loss": 0.2856,
      "num_input_tokens_seen": 14355064,
      "step": 21800
    },
    {
      "epoch": 12.856721698113208,
      "grad_norm": 3.8797495365142822,
      "learning_rate": 3.4084029150229503e-06,
      "loss": 0.2267,
      "num_input_tokens_seen": 14356920,
      "step": 21805
    },
    {
      "epoch": 12.859669811320755,
      "grad_norm": 3.9937987327575684,
      "learning_rate": 3.4059642352513965e-06,
      "loss": 0.3542,
      "num_input_tokens_seen": 14360632,
      "step": 21810
    },
    {
      "epoch": 12.862617924528301,
      "grad_norm": 2.6981263160705566,
      "learning_rate": 3.4035259775076813e-06,
      "loss": 0.3205,
      "num_input_tokens_seen": 14363736,
      "step": 21815
    },
    {
      "epoch": 12.86556603773585,
      "grad_norm": 4.719969272613525,
      "learning_rate": 3.401088142437344e-06,
      "loss": 0.3331,
      "num_input_tokens_seen": 14367768,
      "step": 21820
    },
    {
      "epoch": 12.868514150943396,
      "grad_norm": 5.7945709228515625,
      "learning_rate": 3.398650730685813e-06,
      "loss": 0.475,
      "num_input_tokens_seen": 14371704,
      "step": 21825
    },
    {
      "epoch": 12.871462264150944,
      "grad_norm": 2.0215790271759033,
      "learning_rate": 3.396213742898401e-06,
      "loss": 0.3183,
      "num_input_tokens_seen": 14375032,
      "step": 21830
    },
    {
      "epoch": 12.87441037735849,
      "grad_norm": 3.1485037803649902,
      "learning_rate": 3.3937771797203134e-06,
      "loss": 0.4404,
      "num_input_tokens_seen": 14377880,
      "step": 21835
    },
    {
      "epoch": 12.877358490566039,
      "grad_norm": 4.105297565460205,
      "learning_rate": 3.391341041796641e-06,
      "loss": 0.4054,
      "num_input_tokens_seen": 14380568,
      "step": 21840
    },
    {
      "epoch": 12.880306603773585,
      "grad_norm": 10.216995239257812,
      "learning_rate": 3.3889053297723585e-06,
      "loss": 0.378,
      "num_input_tokens_seen": 14384248,
      "step": 21845
    },
    {
      "epoch": 12.883254716981131,
      "grad_norm": 2.721057653427124,
      "learning_rate": 3.3864700442923342e-06,
      "loss": 0.361,
      "num_input_tokens_seen": 14387192,
      "step": 21850
    },
    {
      "epoch": 12.88620283018868,
      "grad_norm": 3.1528208255767822,
      "learning_rate": 3.384035186001318e-06,
      "loss": 0.3772,
      "num_input_tokens_seen": 14390296,
      "step": 21855
    },
    {
      "epoch": 12.889150943396226,
      "grad_norm": 2.0998780727386475,
      "learning_rate": 3.381600755543953e-06,
      "loss": 0.3419,
      "num_input_tokens_seen": 14392920,
      "step": 21860
    },
    {
      "epoch": 12.892099056603774,
      "grad_norm": 4.6158671379089355,
      "learning_rate": 3.3791667535647615e-06,
      "loss": 0.3246,
      "num_input_tokens_seen": 14396120,
      "step": 21865
    },
    {
      "epoch": 12.89504716981132,
      "grad_norm": 3.9936835765838623,
      "learning_rate": 3.3767331807081584e-06,
      "loss": 0.3179,
      "num_input_tokens_seen": 14399320,
      "step": 21870
    },
    {
      "epoch": 12.897995283018869,
      "grad_norm": 4.2193474769592285,
      "learning_rate": 3.374300037618442e-06,
      "loss": 0.3085,
      "num_input_tokens_seen": 14402488,
      "step": 21875
    },
    {
      "epoch": 12.900943396226415,
      "grad_norm": 3.6839141845703125,
      "learning_rate": 3.371867324939796e-06,
      "loss": 0.3202,
      "num_input_tokens_seen": 14406328,
      "step": 21880
    },
    {
      "epoch": 12.903891509433961,
      "grad_norm": 3.860703945159912,
      "learning_rate": 3.369435043316293e-06,
      "loss": 0.3929,
      "num_input_tokens_seen": 14409656,
      "step": 21885
    },
    {
      "epoch": 12.90683962264151,
      "grad_norm": 3.7890267372131348,
      "learning_rate": 3.36700319339189e-06,
      "loss": 0.4727,
      "num_input_tokens_seen": 14411640,
      "step": 21890
    },
    {
      "epoch": 12.909787735849056,
      "grad_norm": 2.4482390880584717,
      "learning_rate": 3.3645717758104286e-06,
      "loss": 0.4173,
      "num_input_tokens_seen": 14414200,
      "step": 21895
    },
    {
      "epoch": 12.912735849056604,
      "grad_norm": 2.8492074012756348,
      "learning_rate": 3.3621407912156383e-06,
      "loss": 0.3922,
      "num_input_tokens_seen": 14418104,
      "step": 21900
    },
    {
      "epoch": 12.91568396226415,
      "grad_norm": 2.8848514556884766,
      "learning_rate": 3.3597102402511326e-06,
      "loss": 0.3998,
      "num_input_tokens_seen": 14420984,
      "step": 21905
    },
    {
      "epoch": 12.918632075471699,
      "grad_norm": 11.969594955444336,
      "learning_rate": 3.3572801235604093e-06,
      "loss": 0.322,
      "num_input_tokens_seen": 14423896,
      "step": 21910
    },
    {
      "epoch": 12.921580188679245,
      "grad_norm": 2.6282787322998047,
      "learning_rate": 3.3548504417868538e-06,
      "loss": 0.321,
      "num_input_tokens_seen": 14428056,
      "step": 21915
    },
    {
      "epoch": 12.924528301886792,
      "grad_norm": 4.4809746742248535,
      "learning_rate": 3.352421195573734e-06,
      "loss": 0.3853,
      "num_input_tokens_seen": 14431576,
      "step": 21920
    },
    {
      "epoch": 12.92747641509434,
      "grad_norm": 2.8515939712524414,
      "learning_rate": 3.3499923855642026e-06,
      "loss": 0.4072,
      "num_input_tokens_seen": 14438936,
      "step": 21925
    },
    {
      "epoch": 12.930424528301886,
      "grad_norm": 1.88764488697052,
      "learning_rate": 3.3475640124012986e-06,
      "loss": 0.3307,
      "num_input_tokens_seen": 14442360,
      "step": 21930
    },
    {
      "epoch": 12.933372641509434,
      "grad_norm": 2.4072821140289307,
      "learning_rate": 3.345136076727945e-06,
      "loss": 0.3261,
      "num_input_tokens_seen": 14444952,
      "step": 21935
    },
    {
      "epoch": 12.93632075471698,
      "grad_norm": 3.044734001159668,
      "learning_rate": 3.3427085791869453e-06,
      "loss": 0.3067,
      "num_input_tokens_seen": 14449272,
      "step": 21940
    },
    {
      "epoch": 12.939268867924529,
      "grad_norm": 5.255236625671387,
      "learning_rate": 3.3402815204209926e-06,
      "loss": 0.3564,
      "num_input_tokens_seen": 14451992,
      "step": 21945
    },
    {
      "epoch": 12.942216981132075,
      "grad_norm": 3.854719400405884,
      "learning_rate": 3.337854901072659e-06,
      "loss": 0.3303,
      "num_input_tokens_seen": 14457336,
      "step": 21950
    },
    {
      "epoch": 12.945165094339622,
      "grad_norm": 3.046675205230713,
      "learning_rate": 3.3354287217844056e-06,
      "loss": 0.3419,
      "num_input_tokens_seen": 14461144,
      "step": 21955
    },
    {
      "epoch": 12.94811320754717,
      "grad_norm": 2.1688458919525146,
      "learning_rate": 3.3330029831985712e-06,
      "loss": 0.305,
      "num_input_tokens_seen": 14464088,
      "step": 21960
    },
    {
      "epoch": 12.951061320754716,
      "grad_norm": 5.974574565887451,
      "learning_rate": 3.330577685957382e-06,
      "loss": 0.3907,
      "num_input_tokens_seen": 14468856,
      "step": 21965
    },
    {
      "epoch": 12.954009433962264,
      "grad_norm": 3.1095848083496094,
      "learning_rate": 3.3281528307029454e-06,
      "loss": 0.4445,
      "num_input_tokens_seen": 14471960,
      "step": 21970
    },
    {
      "epoch": 12.95695754716981,
      "grad_norm": 3.5408709049224854,
      "learning_rate": 3.325728418077251e-06,
      "loss": 0.4279,
      "num_input_tokens_seen": 14474904,
      "step": 21975
    },
    {
      "epoch": 12.959905660377359,
      "grad_norm": 3.5984647274017334,
      "learning_rate": 3.3233044487221744e-06,
      "loss": 0.3593,
      "num_input_tokens_seen": 14477944,
      "step": 21980
    },
    {
      "epoch": 12.962853773584905,
      "grad_norm": 3.1663124561309814,
      "learning_rate": 3.3208809232794715e-06,
      "loss": 0.3274,
      "num_input_tokens_seen": 14482008,
      "step": 21985
    },
    {
      "epoch": 12.965801886792454,
      "grad_norm": 2.6490581035614014,
      "learning_rate": 3.3184578423907797e-06,
      "loss": 0.3005,
      "num_input_tokens_seen": 14485400,
      "step": 21990
    },
    {
      "epoch": 12.96875,
      "grad_norm": 3.1252307891845703,
      "learning_rate": 3.3160352066976224e-06,
      "loss": 0.3466,
      "num_input_tokens_seen": 14488408,
      "step": 21995
    },
    {
      "epoch": 12.971698113207546,
      "grad_norm": 2.2896461486816406,
      "learning_rate": 3.3136130168414003e-06,
      "loss": 0.3197,
      "num_input_tokens_seen": 14491416,
      "step": 22000
    },
    {
      "epoch": 12.974646226415095,
      "grad_norm": 4.066316604614258,
      "learning_rate": 3.311191273463401e-06,
      "loss": 0.3662,
      "num_input_tokens_seen": 14494424,
      "step": 22005
    },
    {
      "epoch": 12.977594339622641,
      "grad_norm": 5.540147304534912,
      "learning_rate": 3.3087699772047908e-06,
      "loss": 0.3907,
      "num_input_tokens_seen": 14497496,
      "step": 22010
    },
    {
      "epoch": 12.98054245283019,
      "grad_norm": 3.3207361698150635,
      "learning_rate": 3.3063491287066164e-06,
      "loss": 0.423,
      "num_input_tokens_seen": 14500568,
      "step": 22015
    },
    {
      "epoch": 12.983490566037736,
      "grad_norm": 3.8865349292755127,
      "learning_rate": 3.303928728609811e-06,
      "loss": 0.4309,
      "num_input_tokens_seen": 14502904,
      "step": 22020
    },
    {
      "epoch": 12.986438679245284,
      "grad_norm": 1.788732886314392,
      "learning_rate": 3.3015087775551835e-06,
      "loss": 0.2976,
      "num_input_tokens_seen": 14506328,
      "step": 22025
    },
    {
      "epoch": 12.98938679245283,
      "grad_norm": 2.829754114151001,
      "learning_rate": 3.299089276183427e-06,
      "loss": 0.2607,
      "num_input_tokens_seen": 14509304,
      "step": 22030
    },
    {
      "epoch": 12.992334905660378,
      "grad_norm": 3.5364861488342285,
      "learning_rate": 3.2966702251351157e-06,
      "loss": 0.3234,
      "num_input_tokens_seen": 14512344,
      "step": 22035
    },
    {
      "epoch": 12.995283018867925,
      "grad_norm": 4.186611652374268,
      "learning_rate": 3.2942516250507035e-06,
      "loss": 0.4604,
      "num_input_tokens_seen": 14515000,
      "step": 22040
    },
    {
      "epoch": 12.998231132075471,
      "grad_norm": 2.764451026916504,
      "learning_rate": 3.2918334765705227e-06,
      "loss": 0.3141,
      "num_input_tokens_seen": 14517976,
      "step": 22045
    },
    {
      "epoch": 13.00117924528302,
      "grad_norm": 3.013895034790039,
      "learning_rate": 3.289415780334792e-06,
      "loss": 0.3004,
      "num_input_tokens_seen": 14522088,
      "step": 22050
    },
    {
      "epoch": 13.004127358490566,
      "grad_norm": 3.030280590057373,
      "learning_rate": 3.2869985369836067e-06,
      "loss": 0.2574,
      "num_input_tokens_seen": 14525736,
      "step": 22055
    },
    {
      "epoch": 13.007075471698114,
      "grad_norm": 4.481865882873535,
      "learning_rate": 3.2845817471569406e-06,
      "loss": 0.3146,
      "num_input_tokens_seen": 14528712,
      "step": 22060
    },
    {
      "epoch": 13.01002358490566,
      "grad_norm": 3.335444927215576,
      "learning_rate": 3.28216541149465e-06,
      "loss": 0.291,
      "num_input_tokens_seen": 14531464,
      "step": 22065
    },
    {
      "epoch": 13.012971698113208,
      "grad_norm": 2.5526270866394043,
      "learning_rate": 3.2797495306364707e-06,
      "loss": 0.3103,
      "num_input_tokens_seen": 14534056,
      "step": 22070
    },
    {
      "epoch": 13.015919811320755,
      "grad_norm": 5.848972320556641,
      "learning_rate": 3.2773341052220174e-06,
      "loss": 0.3898,
      "num_input_tokens_seen": 14537480,
      "step": 22075
    },
    {
      "epoch": 13.018867924528301,
      "grad_norm": 5.087125778198242,
      "learning_rate": 3.274919135890783e-06,
      "loss": 0.2681,
      "num_input_tokens_seen": 14541864,
      "step": 22080
    },
    {
      "epoch": 13.02181603773585,
      "grad_norm": 2.660449743270874,
      "learning_rate": 3.2725046232821424e-06,
      "loss": 0.3226,
      "num_input_tokens_seen": 14547944,
      "step": 22085
    },
    {
      "epoch": 13.024764150943396,
      "grad_norm": 4.324846267700195,
      "learning_rate": 3.270090568035348e-06,
      "loss": 0.3688,
      "num_input_tokens_seen": 14550824,
      "step": 22090
    },
    {
      "epoch": 13.027712264150944,
      "grad_norm": 5.113674640655518,
      "learning_rate": 3.2676769707895306e-06,
      "loss": 0.388,
      "num_input_tokens_seen": 14554376,
      "step": 22095
    },
    {
      "epoch": 13.03066037735849,
      "grad_norm": 2.5404887199401855,
      "learning_rate": 3.2652638321837015e-06,
      "loss": 0.2622,
      "num_input_tokens_seen": 14556904,
      "step": 22100
    },
    {
      "epoch": 13.033608490566039,
      "grad_norm": 3.28525710105896,
      "learning_rate": 3.2628511528567497e-06,
      "loss": 0.3999,
      "num_input_tokens_seen": 14560488,
      "step": 22105
    },
    {
      "epoch": 13.036556603773585,
      "grad_norm": 3.513718366622925,
      "learning_rate": 3.2604389334474407e-06,
      "loss": 0.3344,
      "num_input_tokens_seen": 14563464,
      "step": 22110
    },
    {
      "epoch": 13.039504716981131,
      "grad_norm": 3.555864095687866,
      "learning_rate": 3.2580271745944224e-06,
      "loss": 0.2874,
      "num_input_tokens_seen": 14566120,
      "step": 22115
    },
    {
      "epoch": 13.04245283018868,
      "grad_norm": 4.514703273773193,
      "learning_rate": 3.255615876936217e-06,
      "loss": 0.4695,
      "num_input_tokens_seen": 14568936,
      "step": 22120
    },
    {
      "epoch": 13.045400943396226,
      "grad_norm": 2.2500171661376953,
      "learning_rate": 3.2532050411112248e-06,
      "loss": 0.3362,
      "num_input_tokens_seen": 14573736,
      "step": 22125
    },
    {
      "epoch": 13.048349056603774,
      "grad_norm": 4.0734171867370605,
      "learning_rate": 3.2507946677577274e-06,
      "loss": 0.4699,
      "num_input_tokens_seen": 14577512,
      "step": 22130
    },
    {
      "epoch": 13.05129716981132,
      "grad_norm": 3.0512630939483643,
      "learning_rate": 3.2483847575138807e-06,
      "loss": 0.2792,
      "num_input_tokens_seen": 14580584,
      "step": 22135
    },
    {
      "epoch": 13.054245283018869,
      "grad_norm": 3.8828585147857666,
      "learning_rate": 3.245975311017716e-06,
      "loss": 0.2794,
      "num_input_tokens_seen": 14583080,
      "step": 22140
    },
    {
      "epoch": 13.057193396226415,
      "grad_norm": 4.468921184539795,
      "learning_rate": 3.2435663289071486e-06,
      "loss": 0.4468,
      "num_input_tokens_seen": 14586056,
      "step": 22145
    },
    {
      "epoch": 13.060141509433961,
      "grad_norm": 3.2856369018554688,
      "learning_rate": 3.241157811819966e-06,
      "loss": 0.2751,
      "num_input_tokens_seen": 14589160,
      "step": 22150
    },
    {
      "epoch": 13.06308962264151,
      "grad_norm": 2.6212518215179443,
      "learning_rate": 3.2387497603938327e-06,
      "loss": 0.4967,
      "num_input_tokens_seen": 14592232,
      "step": 22155
    },
    {
      "epoch": 13.066037735849056,
      "grad_norm": 2.8212997913360596,
      "learning_rate": 3.2363421752662903e-06,
      "loss": 0.2827,
      "num_input_tokens_seen": 14596008,
      "step": 22160
    },
    {
      "epoch": 13.068985849056604,
      "grad_norm": 2.731687545776367,
      "learning_rate": 3.233935057074759e-06,
      "loss": 0.276,
      "num_input_tokens_seen": 14599144,
      "step": 22165
    },
    {
      "epoch": 13.07193396226415,
      "grad_norm": 2.946129083633423,
      "learning_rate": 3.2315284064565324e-06,
      "loss": 0.4082,
      "num_input_tokens_seen": 14602600,
      "step": 22170
    },
    {
      "epoch": 13.074882075471699,
      "grad_norm": 1.6761841773986816,
      "learning_rate": 3.2291222240487813e-06,
      "loss": 0.3568,
      "num_input_tokens_seen": 14606184,
      "step": 22175
    },
    {
      "epoch": 13.077830188679245,
      "grad_norm": 2.938220977783203,
      "learning_rate": 3.226716510488554e-06,
      "loss": 0.2755,
      "num_input_tokens_seen": 14609384,
      "step": 22180
    },
    {
      "epoch": 13.080778301886792,
      "grad_norm": 2.883180856704712,
      "learning_rate": 3.224311266412773e-06,
      "loss": 0.2669,
      "num_input_tokens_seen": 14612872,
      "step": 22185
    },
    {
      "epoch": 13.08372641509434,
      "grad_norm": 2.3305885791778564,
      "learning_rate": 3.2219064924582366e-06,
      "loss": 0.2443,
      "num_input_tokens_seen": 14616552,
      "step": 22190
    },
    {
      "epoch": 13.086674528301886,
      "grad_norm": 3.1506528854370117,
      "learning_rate": 3.2195021892616197e-06,
      "loss": 0.3891,
      "num_input_tokens_seen": 14619528,
      "step": 22195
    },
    {
      "epoch": 13.089622641509434,
      "grad_norm": 2.3986377716064453,
      "learning_rate": 3.217098357459472e-06,
      "loss": 0.3605,
      "num_input_tokens_seen": 14622312,
      "step": 22200
    },
    {
      "epoch": 13.09257075471698,
      "grad_norm": 4.7461652755737305,
      "learning_rate": 3.214694997688217e-06,
      "loss": 0.4457,
      "num_input_tokens_seen": 14625416,
      "step": 22205
    },
    {
      "epoch": 13.095518867924529,
      "grad_norm": 3.7387406826019287,
      "learning_rate": 3.2122921105841572e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 14628200,
      "step": 22210
    },
    {
      "epoch": 13.098466981132075,
      "grad_norm": 3.193002462387085,
      "learning_rate": 3.2098896967834647e-06,
      "loss": 0.3506,
      "num_input_tokens_seen": 14631464,
      "step": 22215
    },
    {
      "epoch": 13.101415094339623,
      "grad_norm": 3.852872133255005,
      "learning_rate": 3.2074877569221896e-06,
      "loss": 0.245,
      "num_input_tokens_seen": 14634472,
      "step": 22220
    },
    {
      "epoch": 13.10436320754717,
      "grad_norm": 3.483309745788574,
      "learning_rate": 3.205086291636257e-06,
      "loss": 0.2453,
      "num_input_tokens_seen": 14637384,
      "step": 22225
    },
    {
      "epoch": 13.107311320754716,
      "grad_norm": 7.046278476715088,
      "learning_rate": 3.202685301561463e-06,
      "loss": 0.2829,
      "num_input_tokens_seen": 14640776,
      "step": 22230
    },
    {
      "epoch": 13.110259433962264,
      "grad_norm": 3.166409969329834,
      "learning_rate": 3.200284787333482e-06,
      "loss": 0.346,
      "num_input_tokens_seen": 14643464,
      "step": 22235
    },
    {
      "epoch": 13.11320754716981,
      "grad_norm": 3.765007257461548,
      "learning_rate": 3.1978847495878595e-06,
      "loss": 0.3029,
      "num_input_tokens_seen": 14646504,
      "step": 22240
    },
    {
      "epoch": 13.116155660377359,
      "grad_norm": 5.930373191833496,
      "learning_rate": 3.1954851889600176e-06,
      "loss": 0.3179,
      "num_input_tokens_seen": 14649704,
      "step": 22245
    },
    {
      "epoch": 13.119103773584905,
      "grad_norm": 4.362338542938232,
      "learning_rate": 3.1930861060852485e-06,
      "loss": 0.3348,
      "num_input_tokens_seen": 14654216,
      "step": 22250
    },
    {
      "epoch": 13.122051886792454,
      "grad_norm": 3.536468267440796,
      "learning_rate": 3.1906875015987194e-06,
      "loss": 0.3633,
      "num_input_tokens_seen": 14657448,
      "step": 22255
    },
    {
      "epoch": 13.125,
      "grad_norm": 1.8814209699630737,
      "learning_rate": 3.188289376135473e-06,
      "loss": 0.4501,
      "num_input_tokens_seen": 14661032,
      "step": 22260
    },
    {
      "epoch": 13.127948113207546,
      "grad_norm": 2.3776233196258545,
      "learning_rate": 3.1858917303304213e-06,
      "loss": 0.2817,
      "num_input_tokens_seen": 14665320,
      "step": 22265
    },
    {
      "epoch": 13.130896226415095,
      "grad_norm": 3.592939615249634,
      "learning_rate": 3.1834945648183535e-06,
      "loss": 0.3238,
      "num_input_tokens_seen": 14669736,
      "step": 22270
    },
    {
      "epoch": 13.133844339622641,
      "grad_norm": 2.3296799659729004,
      "learning_rate": 3.1810978802339283e-06,
      "loss": 0.3776,
      "num_input_tokens_seen": 14673064,
      "step": 22275
    },
    {
      "epoch": 13.13679245283019,
      "grad_norm": 4.6650309562683105,
      "learning_rate": 3.1787016772116767e-06,
      "loss": 0.3201,
      "num_input_tokens_seen": 14676936,
      "step": 22280
    },
    {
      "epoch": 13.139740566037736,
      "grad_norm": 4.653472900390625,
      "learning_rate": 3.1763059563860073e-06,
      "loss": 0.3818,
      "num_input_tokens_seen": 14679720,
      "step": 22285
    },
    {
      "epoch": 13.142688679245284,
      "grad_norm": 2.6507644653320312,
      "learning_rate": 3.1739107183911953e-06,
      "loss": 0.3889,
      "num_input_tokens_seen": 14682472,
      "step": 22290
    },
    {
      "epoch": 13.14563679245283,
      "grad_norm": 3.855710506439209,
      "learning_rate": 3.1715159638613898e-06,
      "loss": 0.3778,
      "num_input_tokens_seen": 14685960,
      "step": 22295
    },
    {
      "epoch": 13.148584905660377,
      "grad_norm": 2.924415349960327,
      "learning_rate": 3.1691216934306134e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 14688904,
      "step": 22300
    },
    {
      "epoch": 13.151533018867925,
      "grad_norm": 4.696845531463623,
      "learning_rate": 3.16672790773276e-06,
      "loss": 0.3374,
      "num_input_tokens_seen": 14691720,
      "step": 22305
    },
    {
      "epoch": 13.154481132075471,
      "grad_norm": 3.5109925270080566,
      "learning_rate": 3.164334607401593e-06,
      "loss": 0.3709,
      "num_input_tokens_seen": 14694408,
      "step": 22310
    },
    {
      "epoch": 13.15742924528302,
      "grad_norm": 2.449213981628418,
      "learning_rate": 3.1619417930707506e-06,
      "loss": 0.318,
      "num_input_tokens_seen": 14697448,
      "step": 22315
    },
    {
      "epoch": 13.160377358490566,
      "grad_norm": 4.849468231201172,
      "learning_rate": 3.1595494653737408e-06,
      "loss": 0.3549,
      "num_input_tokens_seen": 14700584,
      "step": 22320
    },
    {
      "epoch": 13.163325471698114,
      "grad_norm": 3.480523109436035,
      "learning_rate": 3.1571576249439408e-06,
      "loss": 0.3209,
      "num_input_tokens_seen": 14703880,
      "step": 22325
    },
    {
      "epoch": 13.16627358490566,
      "grad_norm": 2.2303333282470703,
      "learning_rate": 3.1547662724146e-06,
      "loss": 0.3372,
      "num_input_tokens_seen": 14706920,
      "step": 22330
    },
    {
      "epoch": 13.169221698113208,
      "grad_norm": 5.1591572761535645,
      "learning_rate": 3.1523754084188436e-06,
      "loss": 0.3073,
      "num_input_tokens_seen": 14710248,
      "step": 22335
    },
    {
      "epoch": 13.172169811320755,
      "grad_norm": 3.2747061252593994,
      "learning_rate": 3.149985033589661e-06,
      "loss": 0.2152,
      "num_input_tokens_seen": 14715336,
      "step": 22340
    },
    {
      "epoch": 13.175117924528301,
      "grad_norm": 2.3089466094970703,
      "learning_rate": 3.147595148559912e-06,
      "loss": 0.3285,
      "num_input_tokens_seen": 14719304,
      "step": 22345
    },
    {
      "epoch": 13.17806603773585,
      "grad_norm": 5.251135349273682,
      "learning_rate": 3.1452057539623328e-06,
      "loss": 0.3234,
      "num_input_tokens_seen": 14722408,
      "step": 22350
    },
    {
      "epoch": 13.181014150943396,
      "grad_norm": 3.280151844024658,
      "learning_rate": 3.142816850429523e-06,
      "loss": 0.274,
      "num_input_tokens_seen": 14725032,
      "step": 22355
    },
    {
      "epoch": 13.183962264150944,
      "grad_norm": 4.869561672210693,
      "learning_rate": 3.1404284385939552e-06,
      "loss": 0.4382,
      "num_input_tokens_seen": 14728328,
      "step": 22360
    },
    {
      "epoch": 13.18691037735849,
      "grad_norm": 3.4073586463928223,
      "learning_rate": 3.138040519087975e-06,
      "loss": 0.3403,
      "num_input_tokens_seen": 14731464,
      "step": 22365
    },
    {
      "epoch": 13.189858490566039,
      "grad_norm": 2.498628854751587,
      "learning_rate": 3.13565309254379e-06,
      "loss": 0.2795,
      "num_input_tokens_seen": 14734344,
      "step": 22370
    },
    {
      "epoch": 13.192806603773585,
      "grad_norm": 3.698668956756592,
      "learning_rate": 3.1332661595934845e-06,
      "loss": 0.3897,
      "num_input_tokens_seen": 14738536,
      "step": 22375
    },
    {
      "epoch": 13.195754716981131,
      "grad_norm": 2.006817102432251,
      "learning_rate": 3.130879720869008e-06,
      "loss": 0.3192,
      "num_input_tokens_seen": 14741672,
      "step": 22380
    },
    {
      "epoch": 13.19870283018868,
      "grad_norm": 5.53879976272583,
      "learning_rate": 3.1284937770021815e-06,
      "loss": 0.2924,
      "num_input_tokens_seen": 14745288,
      "step": 22385
    },
    {
      "epoch": 13.201650943396226,
      "grad_norm": 2.8169736862182617,
      "learning_rate": 3.1261083286246916e-06,
      "loss": 0.2757,
      "num_input_tokens_seen": 14748360,
      "step": 22390
    },
    {
      "epoch": 13.204599056603774,
      "grad_norm": 3.0694119930267334,
      "learning_rate": 3.1237233763680997e-06,
      "loss": 0.317,
      "num_input_tokens_seen": 14751432,
      "step": 22395
    },
    {
      "epoch": 13.20754716981132,
      "grad_norm": 2.8201401233673096,
      "learning_rate": 3.1213389208638303e-06,
      "loss": 0.2729,
      "num_input_tokens_seen": 14753896,
      "step": 22400
    },
    {
      "epoch": 13.210495283018869,
      "grad_norm": 5.055774688720703,
      "learning_rate": 3.1189549627431757e-06,
      "loss": 0.2267,
      "num_input_tokens_seen": 14756840,
      "step": 22405
    },
    {
      "epoch": 13.213443396226415,
      "grad_norm": 4.8234076499938965,
      "learning_rate": 3.116571502637304e-06,
      "loss": 0.3488,
      "num_input_tokens_seen": 14759240,
      "step": 22410
    },
    {
      "epoch": 13.216391509433961,
      "grad_norm": 2.2664830684661865,
      "learning_rate": 3.1141885411772434e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 14764296,
      "step": 22415
    },
    {
      "epoch": 13.21933962264151,
      "grad_norm": 2.494596481323242,
      "learning_rate": 3.111806078993893e-06,
      "loss": 0.3271,
      "num_input_tokens_seen": 14767336,
      "step": 22420
    },
    {
      "epoch": 13.222287735849056,
      "grad_norm": 4.474058151245117,
      "learning_rate": 3.1094241167180223e-06,
      "loss": 0.5165,
      "num_input_tokens_seen": 14774344,
      "step": 22425
    },
    {
      "epoch": 13.225235849056604,
      "grad_norm": 3.5619685649871826,
      "learning_rate": 3.1070426549802623e-06,
      "loss": 0.2756,
      "num_input_tokens_seen": 14777448,
      "step": 22430
    },
    {
      "epoch": 13.22818396226415,
      "grad_norm": 2.2009530067443848,
      "learning_rate": 3.1046616944111196e-06,
      "loss": 0.3208,
      "num_input_tokens_seen": 14780296,
      "step": 22435
    },
    {
      "epoch": 13.231132075471699,
      "grad_norm": 2.543518543243408,
      "learning_rate": 3.1022812356409606e-06,
      "loss": 0.2922,
      "num_input_tokens_seen": 14783976,
      "step": 22440
    },
    {
      "epoch": 13.234080188679245,
      "grad_norm": 2.9902992248535156,
      "learning_rate": 3.0999012793000244e-06,
      "loss": 0.2837,
      "num_input_tokens_seen": 14787048,
      "step": 22445
    },
    {
      "epoch": 13.237028301886792,
      "grad_norm": 4.036132335662842,
      "learning_rate": 3.097521826018414e-06,
      "loss": 0.3257,
      "num_input_tokens_seen": 14790152,
      "step": 22450
    },
    {
      "epoch": 13.23997641509434,
      "grad_norm": 5.493289470672607,
      "learning_rate": 3.0951428764260973e-06,
      "loss": 0.4397,
      "num_input_tokens_seen": 14793480,
      "step": 22455
    },
    {
      "epoch": 13.242924528301886,
      "grad_norm": 6.205087184906006,
      "learning_rate": 3.092764431152915e-06,
      "loss": 0.3633,
      "num_input_tokens_seen": 14796200,
      "step": 22460
    },
    {
      "epoch": 13.245872641509434,
      "grad_norm": 3.0279223918914795,
      "learning_rate": 3.0903864908285693e-06,
      "loss": 0.2893,
      "num_input_tokens_seen": 14800808,
      "step": 22465
    },
    {
      "epoch": 13.24882075471698,
      "grad_norm": 3.8543894290924072,
      "learning_rate": 3.088009056082629e-06,
      "loss": 0.2858,
      "num_input_tokens_seen": 14804264,
      "step": 22470
    },
    {
      "epoch": 13.251768867924529,
      "grad_norm": 4.892855167388916,
      "learning_rate": 3.0856321275445324e-06,
      "loss": 0.2097,
      "num_input_tokens_seen": 14806920,
      "step": 22475
    },
    {
      "epoch": 13.254716981132075,
      "grad_norm": 3.1630098819732666,
      "learning_rate": 3.0832557058435808e-06,
      "loss": 0.2354,
      "num_input_tokens_seen": 14810696,
      "step": 22480
    },
    {
      "epoch": 13.257665094339623,
      "grad_norm": 5.491907596588135,
      "learning_rate": 3.0808797916089405e-06,
      "loss": 0.3885,
      "num_input_tokens_seen": 14813064,
      "step": 22485
    },
    {
      "epoch": 13.26061320754717,
      "grad_norm": 4.724816799163818,
      "learning_rate": 3.078504385469647e-06,
      "loss": 0.381,
      "num_input_tokens_seen": 14816360,
      "step": 22490
    },
    {
      "epoch": 13.263561320754716,
      "grad_norm": 3.9109933376312256,
      "learning_rate": 3.076129488054599e-06,
      "loss": 0.3041,
      "num_input_tokens_seen": 14820616,
      "step": 22495
    },
    {
      "epoch": 13.266509433962264,
      "grad_norm": 3.8833556175231934,
      "learning_rate": 3.0737550999925604e-06,
      "loss": 0.3681,
      "num_input_tokens_seen": 14823496,
      "step": 22500
    },
    {
      "epoch": 13.26945754716981,
      "grad_norm": 2.8164420127868652,
      "learning_rate": 3.0713812219121604e-06,
      "loss": 0.2802,
      "num_input_tokens_seen": 14826248,
      "step": 22505
    },
    {
      "epoch": 13.272405660377359,
      "grad_norm": 2.6226203441619873,
      "learning_rate": 3.0690078544418934e-06,
      "loss": 0.4064,
      "num_input_tokens_seen": 14829320,
      "step": 22510
    },
    {
      "epoch": 13.275353773584905,
      "grad_norm": 6.343196868896484,
      "learning_rate": 3.0666349982101198e-06,
      "loss": 0.399,
      "num_input_tokens_seen": 14832456,
      "step": 22515
    },
    {
      "epoch": 13.278301886792454,
      "grad_norm": 3.7419474124908447,
      "learning_rate": 3.0642626538450627e-06,
      "loss": 0.3385,
      "num_input_tokens_seen": 14836232,
      "step": 22520
    },
    {
      "epoch": 13.28125,
      "grad_norm": 3.0143239498138428,
      "learning_rate": 3.061890821974809e-06,
      "loss": 0.357,
      "num_input_tokens_seen": 14839208,
      "step": 22525
    },
    {
      "epoch": 13.284198113207546,
      "grad_norm": 6.119863510131836,
      "learning_rate": 3.059519503227313e-06,
      "loss": 0.2742,
      "num_input_tokens_seen": 14841832,
      "step": 22530
    },
    {
      "epoch": 13.287146226415095,
      "grad_norm": 3.7055110931396484,
      "learning_rate": 3.057148698230393e-06,
      "loss": 0.2822,
      "num_input_tokens_seen": 14846280,
      "step": 22535
    },
    {
      "epoch": 13.290094339622641,
      "grad_norm": 4.603560447692871,
      "learning_rate": 3.0547784076117294e-06,
      "loss": 0.3982,
      "num_input_tokens_seen": 14848584,
      "step": 22540
    },
    {
      "epoch": 13.29304245283019,
      "grad_norm": 4.087154388427734,
      "learning_rate": 3.0524086319988635e-06,
      "loss": 0.3064,
      "num_input_tokens_seen": 14851560,
      "step": 22545
    },
    {
      "epoch": 13.295990566037736,
      "grad_norm": 2.981139659881592,
      "learning_rate": 3.0500393720192074e-06,
      "loss": 0.3257,
      "num_input_tokens_seen": 14854888,
      "step": 22550
    },
    {
      "epoch": 13.298938679245284,
      "grad_norm": 2.686608076095581,
      "learning_rate": 3.047670628300031e-06,
      "loss": 0.3345,
      "num_input_tokens_seen": 14858120,
      "step": 22555
    },
    {
      "epoch": 13.30188679245283,
      "grad_norm": 2.4988136291503906,
      "learning_rate": 3.0453024014684694e-06,
      "loss": 0.2805,
      "num_input_tokens_seen": 14860616,
      "step": 22560
    },
    {
      "epoch": 13.304834905660377,
      "grad_norm": 2.9417054653167725,
      "learning_rate": 3.0429346921515225e-06,
      "loss": 0.3274,
      "num_input_tokens_seen": 14863944,
      "step": 22565
    },
    {
      "epoch": 13.307783018867925,
      "grad_norm": 5.696717739105225,
      "learning_rate": 3.04056750097605e-06,
      "loss": 0.2823,
      "num_input_tokens_seen": 14866568,
      "step": 22570
    },
    {
      "epoch": 13.310731132075471,
      "grad_norm": 3.2223105430603027,
      "learning_rate": 3.0382008285687754e-06,
      "loss": 0.2766,
      "num_input_tokens_seen": 14869896,
      "step": 22575
    },
    {
      "epoch": 13.31367924528302,
      "grad_norm": 7.409430027008057,
      "learning_rate": 3.035834675556287e-06,
      "loss": 0.347,
      "num_input_tokens_seen": 14873064,
      "step": 22580
    },
    {
      "epoch": 13.316627358490566,
      "grad_norm": 2.1268887519836426,
      "learning_rate": 3.0334690425650336e-06,
      "loss": 0.3775,
      "num_input_tokens_seen": 14876392,
      "step": 22585
    },
    {
      "epoch": 13.319575471698114,
      "grad_norm": 3.227909803390503,
      "learning_rate": 3.031103930221325e-06,
      "loss": 0.3082,
      "num_input_tokens_seen": 14878792,
      "step": 22590
    },
    {
      "epoch": 13.32252358490566,
      "grad_norm": 3.1433029174804688,
      "learning_rate": 3.028739339151338e-06,
      "loss": 0.2647,
      "num_input_tokens_seen": 14882120,
      "step": 22595
    },
    {
      "epoch": 13.325471698113208,
      "grad_norm": 3.6477267742156982,
      "learning_rate": 3.0263752699811067e-06,
      "loss": 0.3383,
      "num_input_tokens_seen": 14886120,
      "step": 22600
    },
    {
      "epoch": 13.328419811320755,
      "grad_norm": 2.749800205230713,
      "learning_rate": 3.0240117233365267e-06,
      "loss": 0.3003,
      "num_input_tokens_seen": 14889800,
      "step": 22605
    },
    {
      "epoch": 13.331367924528301,
      "grad_norm": 6.048524379730225,
      "learning_rate": 3.0216486998433604e-06,
      "loss": 0.418,
      "num_input_tokens_seen": 14892040,
      "step": 22610
    },
    {
      "epoch": 13.33431603773585,
      "grad_norm": 4.3562397956848145,
      "learning_rate": 3.0192862001272273e-06,
      "loss": 0.4085,
      "num_input_tokens_seen": 14895272,
      "step": 22615
    },
    {
      "epoch": 13.337264150943396,
      "grad_norm": 4.372014999389648,
      "learning_rate": 3.0169242248136066e-06,
      "loss": 0.4102,
      "num_input_tokens_seen": 14898216,
      "step": 22620
    },
    {
      "epoch": 13.340212264150944,
      "grad_norm": 5.77791166305542,
      "learning_rate": 3.0145627745278457e-06,
      "loss": 0.4281,
      "num_input_tokens_seen": 14902568,
      "step": 22625
    },
    {
      "epoch": 13.34316037735849,
      "grad_norm": 3.1691482067108154,
      "learning_rate": 3.0122018498951478e-06,
      "loss": 0.4025,
      "num_input_tokens_seen": 14905608,
      "step": 22630
    },
    {
      "epoch": 13.346108490566039,
      "grad_norm": 7.59244441986084,
      "learning_rate": 3.0098414515405765e-06,
      "loss": 0.3131,
      "num_input_tokens_seen": 14908616,
      "step": 22635
    },
    {
      "epoch": 13.349056603773585,
      "grad_norm": 3.3999767303466797,
      "learning_rate": 3.0074815800890576e-06,
      "loss": 0.3341,
      "num_input_tokens_seen": 14912168,
      "step": 22640
    },
    {
      "epoch": 13.352004716981131,
      "grad_norm": 4.161776542663574,
      "learning_rate": 3.005122236165378e-06,
      "loss": 0.3476,
      "num_input_tokens_seen": 14915016,
      "step": 22645
    },
    {
      "epoch": 13.35495283018868,
      "grad_norm": 4.206117153167725,
      "learning_rate": 3.0027634203941847e-06,
      "loss": 0.3392,
      "num_input_tokens_seen": 14917544,
      "step": 22650
    },
    {
      "epoch": 13.357900943396226,
      "grad_norm": 2.5467681884765625,
      "learning_rate": 3.0004051333999816e-06,
      "loss": 0.2329,
      "num_input_tokens_seen": 14923592,
      "step": 22655
    },
    {
      "epoch": 13.360849056603774,
      "grad_norm": 3.22763991355896,
      "learning_rate": 2.998047375807139e-06,
      "loss": 0.2849,
      "num_input_tokens_seen": 14928744,
      "step": 22660
    },
    {
      "epoch": 13.36379716981132,
      "grad_norm": 2.499744176864624,
      "learning_rate": 2.995690148239881e-06,
      "loss": 0.3877,
      "num_input_tokens_seen": 14932456,
      "step": 22665
    },
    {
      "epoch": 13.366745283018869,
      "grad_norm": 2.9635720252990723,
      "learning_rate": 2.993333451322293e-06,
      "loss": 0.332,
      "num_input_tokens_seen": 14935304,
      "step": 22670
    },
    {
      "epoch": 13.369693396226415,
      "grad_norm": 3.089754343032837,
      "learning_rate": 2.9909772856783242e-06,
      "loss": 0.4419,
      "num_input_tokens_seen": 14938600,
      "step": 22675
    },
    {
      "epoch": 13.372641509433961,
      "grad_norm": 4.198117256164551,
      "learning_rate": 2.988621651931777e-06,
      "loss": 0.3372,
      "num_input_tokens_seen": 14941736,
      "step": 22680
    },
    {
      "epoch": 13.37558962264151,
      "grad_norm": 2.7154273986816406,
      "learning_rate": 2.986266550706315e-06,
      "loss": 0.3654,
      "num_input_tokens_seen": 14944520,
      "step": 22685
    },
    {
      "epoch": 13.378537735849056,
      "grad_norm": 3.0125534534454346,
      "learning_rate": 2.9839119826254627e-06,
      "loss": 0.2798,
      "num_input_tokens_seen": 14947656,
      "step": 22690
    },
    {
      "epoch": 13.381485849056604,
      "grad_norm": 2.3469016551971436,
      "learning_rate": 2.981557948312602e-06,
      "loss": 0.4951,
      "num_input_tokens_seen": 14950568,
      "step": 22695
    },
    {
      "epoch": 13.38443396226415,
      "grad_norm": 3.1788690090179443,
      "learning_rate": 2.9792044483909733e-06,
      "loss": 0.3244,
      "num_input_tokens_seen": 14953416,
      "step": 22700
    },
    {
      "epoch": 13.387382075471699,
      "grad_norm": 6.825149059295654,
      "learning_rate": 2.9768514834836767e-06,
      "loss": 0.293,
      "num_input_tokens_seen": 14956104,
      "step": 22705
    },
    {
      "epoch": 13.390330188679245,
      "grad_norm": 5.152388572692871,
      "learning_rate": 2.9744990542136685e-06,
      "loss": 0.332,
      "num_input_tokens_seen": 14958760,
      "step": 22710
    },
    {
      "epoch": 13.393278301886792,
      "grad_norm": 4.954981327056885,
      "learning_rate": 2.9721471612037637e-06,
      "loss": 0.3326,
      "num_input_tokens_seen": 14962568,
      "step": 22715
    },
    {
      "epoch": 13.39622641509434,
      "grad_norm": 2.6211135387420654,
      "learning_rate": 2.9697958050766385e-06,
      "loss": 0.35,
      "num_input_tokens_seen": 14966440,
      "step": 22720
    },
    {
      "epoch": 13.399174528301886,
      "grad_norm": 2.557053804397583,
      "learning_rate": 2.967444986454825e-06,
      "loss": 0.3058,
      "num_input_tokens_seen": 14969704,
      "step": 22725
    },
    {
      "epoch": 13.402122641509434,
      "grad_norm": 3.7379140853881836,
      "learning_rate": 2.9650947059607106e-06,
      "loss": 0.352,
      "num_input_tokens_seen": 14973608,
      "step": 22730
    },
    {
      "epoch": 13.40507075471698,
      "grad_norm": 2.6138927936553955,
      "learning_rate": 2.962744964216542e-06,
      "loss": 0.3422,
      "num_input_tokens_seen": 14975976,
      "step": 22735
    },
    {
      "epoch": 13.408018867924529,
      "grad_norm": 5.103383541107178,
      "learning_rate": 2.960395761844425e-06,
      "loss": 0.3434,
      "num_input_tokens_seen": 14978792,
      "step": 22740
    },
    {
      "epoch": 13.410966981132075,
      "grad_norm": 3.2040605545043945,
      "learning_rate": 2.95804709946632e-06,
      "loss": 0.2488,
      "num_input_tokens_seen": 14981512,
      "step": 22745
    },
    {
      "epoch": 13.413915094339623,
      "grad_norm": 2.7998971939086914,
      "learning_rate": 2.9556989777040457e-06,
      "loss": 0.2805,
      "num_input_tokens_seen": 14984648,
      "step": 22750
    },
    {
      "epoch": 13.41686320754717,
      "grad_norm": 6.954358100891113,
      "learning_rate": 2.9533513971792776e-06,
      "loss": 0.423,
      "num_input_tokens_seen": 14988328,
      "step": 22755
    },
    {
      "epoch": 13.419811320754716,
      "grad_norm": 2.498765468597412,
      "learning_rate": 2.9510043585135473e-06,
      "loss": 0.3803,
      "num_input_tokens_seen": 14991336,
      "step": 22760
    },
    {
      "epoch": 13.422759433962264,
      "grad_norm": 3.579957962036133,
      "learning_rate": 2.948657862328244e-06,
      "loss": 0.3293,
      "num_input_tokens_seen": 14994504,
      "step": 22765
    },
    {
      "epoch": 13.42570754716981,
      "grad_norm": 3.0097336769104004,
      "learning_rate": 2.946311909244613e-06,
      "loss": 0.3032,
      "num_input_tokens_seen": 14997704,
      "step": 22770
    },
    {
      "epoch": 13.428655660377359,
      "grad_norm": 2.1497116088867188,
      "learning_rate": 2.9439664998837538e-06,
      "loss": 0.3291,
      "num_input_tokens_seen": 15000360,
      "step": 22775
    },
    {
      "epoch": 13.431603773584905,
      "grad_norm": 3.0195088386535645,
      "learning_rate": 2.941621634866626e-06,
      "loss": 0.343,
      "num_input_tokens_seen": 15003336,
      "step": 22780
    },
    {
      "epoch": 13.434551886792454,
      "grad_norm": 5.637275218963623,
      "learning_rate": 2.9392773148140406e-06,
      "loss": 0.3667,
      "num_input_tokens_seen": 15006248,
      "step": 22785
    },
    {
      "epoch": 13.4375,
      "grad_norm": 4.156398296356201,
      "learning_rate": 2.9369335403466676e-06,
      "loss": 0.3951,
      "num_input_tokens_seen": 15008968,
      "step": 22790
    },
    {
      "epoch": 13.440448113207546,
      "grad_norm": 2.431295156478882,
      "learning_rate": 2.9345903120850318e-06,
      "loss": 0.3026,
      "num_input_tokens_seen": 15012712,
      "step": 22795
    },
    {
      "epoch": 13.443396226415095,
      "grad_norm": 5.048667907714844,
      "learning_rate": 2.932247630649512e-06,
      "loss": 0.3053,
      "num_input_tokens_seen": 15015144,
      "step": 22800
    },
    {
      "epoch": 13.446344339622641,
      "grad_norm": 3.6480536460876465,
      "learning_rate": 2.9299054966603424e-06,
      "loss": 0.4331,
      "num_input_tokens_seen": 15019368,
      "step": 22805
    },
    {
      "epoch": 13.44929245283019,
      "grad_norm": 2.2546048164367676,
      "learning_rate": 2.927563910737613e-06,
      "loss": 0.3951,
      "num_input_tokens_seen": 15022888,
      "step": 22810
    },
    {
      "epoch": 13.452240566037736,
      "grad_norm": 3.278656244277954,
      "learning_rate": 2.9252228735012722e-06,
      "loss": 0.3408,
      "num_input_tokens_seen": 15026344,
      "step": 22815
    },
    {
      "epoch": 13.455188679245284,
      "grad_norm": 3.1558659076690674,
      "learning_rate": 2.9228823855711174e-06,
      "loss": 0.3159,
      "num_input_tokens_seen": 15030632,
      "step": 22820
    },
    {
      "epoch": 13.45813679245283,
      "grad_norm": 2.4551899433135986,
      "learning_rate": 2.920542447566802e-06,
      "loss": 0.3924,
      "num_input_tokens_seen": 15033640,
      "step": 22825
    },
    {
      "epoch": 13.461084905660377,
      "grad_norm": 3.8449480533599854,
      "learning_rate": 2.918203060107837e-06,
      "loss": 0.4323,
      "num_input_tokens_seen": 15037320,
      "step": 22830
    },
    {
      "epoch": 13.464033018867925,
      "grad_norm": 3.387789487838745,
      "learning_rate": 2.9158642238135813e-06,
      "loss": 0.3799,
      "num_input_tokens_seen": 15040904,
      "step": 22835
    },
    {
      "epoch": 13.466981132075471,
      "grad_norm": 2.7415077686309814,
      "learning_rate": 2.913525939303257e-06,
      "loss": 0.3112,
      "num_input_tokens_seen": 15044200,
      "step": 22840
    },
    {
      "epoch": 13.46992924528302,
      "grad_norm": 3.611816167831421,
      "learning_rate": 2.9111882071959317e-06,
      "loss": 0.3956,
      "num_input_tokens_seen": 15047144,
      "step": 22845
    },
    {
      "epoch": 13.472877358490566,
      "grad_norm": 4.127892971038818,
      "learning_rate": 2.908851028110532e-06,
      "loss": 0.3643,
      "num_input_tokens_seen": 15050248,
      "step": 22850
    },
    {
      "epoch": 13.475825471698114,
      "grad_norm": 3.7673816680908203,
      "learning_rate": 2.906514402665834e-06,
      "loss": 0.3488,
      "num_input_tokens_seen": 15054056,
      "step": 22855
    },
    {
      "epoch": 13.47877358490566,
      "grad_norm": 5.284366130828857,
      "learning_rate": 2.9041783314804705e-06,
      "loss": 0.3138,
      "num_input_tokens_seen": 15056936,
      "step": 22860
    },
    {
      "epoch": 13.481721698113208,
      "grad_norm": 3.0423853397369385,
      "learning_rate": 2.9018428151729238e-06,
      "loss": 0.2663,
      "num_input_tokens_seen": 15060488,
      "step": 22865
    },
    {
      "epoch": 13.484669811320755,
      "grad_norm": 6.758183002471924,
      "learning_rate": 2.899507854361537e-06,
      "loss": 0.2406,
      "num_input_tokens_seen": 15063144,
      "step": 22870
    },
    {
      "epoch": 13.487617924528301,
      "grad_norm": 3.880744218826294,
      "learning_rate": 2.8971734496644975e-06,
      "loss": 0.3982,
      "num_input_tokens_seen": 15065768,
      "step": 22875
    },
    {
      "epoch": 13.49056603773585,
      "grad_norm": 6.316175937652588,
      "learning_rate": 2.894839601699851e-06,
      "loss": 0.3479,
      "num_input_tokens_seen": 15068680,
      "step": 22880
    },
    {
      "epoch": 13.493514150943396,
      "grad_norm": 4.531544208526611,
      "learning_rate": 2.8925063110854923e-06,
      "loss": 0.3949,
      "num_input_tokens_seen": 15070984,
      "step": 22885
    },
    {
      "epoch": 13.496462264150944,
      "grad_norm": 5.2690815925598145,
      "learning_rate": 2.8901735784391683e-06,
      "loss": 0.4982,
      "num_input_tokens_seen": 15073640,
      "step": 22890
    },
    {
      "epoch": 13.49941037735849,
      "grad_norm": 2.9124860763549805,
      "learning_rate": 2.8878414043784844e-06,
      "loss": 0.356,
      "num_input_tokens_seen": 15077448,
      "step": 22895
    },
    {
      "epoch": 13.502358490566039,
      "grad_norm": 2.2917356491088867,
      "learning_rate": 2.885509789520891e-06,
      "loss": 0.2452,
      "num_input_tokens_seen": 15080744,
      "step": 22900
    },
    {
      "epoch": 13.505306603773585,
      "grad_norm": 6.619190692901611,
      "learning_rate": 2.8831787344836926e-06,
      "loss": 0.38,
      "num_input_tokens_seen": 15083208,
      "step": 22905
    },
    {
      "epoch": 13.508254716981131,
      "grad_norm": 4.88230037689209,
      "learning_rate": 2.880848239884049e-06,
      "loss": 0.3189,
      "num_input_tokens_seen": 15087048,
      "step": 22910
    },
    {
      "epoch": 13.51120283018868,
      "grad_norm": 6.183482646942139,
      "learning_rate": 2.8785183063389667e-06,
      "loss": 0.2381,
      "num_input_tokens_seen": 15090024,
      "step": 22915
    },
    {
      "epoch": 13.514150943396226,
      "grad_norm": 2.2965712547302246,
      "learning_rate": 2.876188934465306e-06,
      "loss": 0.3169,
      "num_input_tokens_seen": 15093352,
      "step": 22920
    },
    {
      "epoch": 13.517099056603774,
      "grad_norm": 4.357248783111572,
      "learning_rate": 2.8738601248797758e-06,
      "loss": 0.2277,
      "num_input_tokens_seen": 15096328,
      "step": 22925
    },
    {
      "epoch": 13.52004716981132,
      "grad_norm": 2.73578143119812,
      "learning_rate": 2.8715318781989432e-06,
      "loss": 0.3446,
      "num_input_tokens_seen": 15099144,
      "step": 22930
    },
    {
      "epoch": 13.522995283018869,
      "grad_norm": 4.9230828285217285,
      "learning_rate": 2.869204195039219e-06,
      "loss": 0.3539,
      "num_input_tokens_seen": 15102056,
      "step": 22935
    },
    {
      "epoch": 13.525943396226415,
      "grad_norm": 3.57020902633667,
      "learning_rate": 2.8668770760168673e-06,
      "loss": 0.3621,
      "num_input_tokens_seen": 15105864,
      "step": 22940
    },
    {
      "epoch": 13.528891509433961,
      "grad_norm": 3.817758560180664,
      "learning_rate": 2.864550521748003e-06,
      "loss": 0.3495,
      "num_input_tokens_seen": 15109960,
      "step": 22945
    },
    {
      "epoch": 13.53183962264151,
      "grad_norm": 1.837733268737793,
      "learning_rate": 2.862224532848591e-06,
      "loss": 0.1823,
      "num_input_tokens_seen": 15112488,
      "step": 22950
    },
    {
      "epoch": 13.534787735849056,
      "grad_norm": 3.627310276031494,
      "learning_rate": 2.8598991099344455e-06,
      "loss": 0.4086,
      "num_input_tokens_seen": 15115784,
      "step": 22955
    },
    {
      "epoch": 13.537735849056604,
      "grad_norm": 4.385852336883545,
      "learning_rate": 2.857574253621236e-06,
      "loss": 0.2942,
      "num_input_tokens_seen": 15119048,
      "step": 22960
    },
    {
      "epoch": 13.54068396226415,
      "grad_norm": 3.1501450538635254,
      "learning_rate": 2.855249964524476e-06,
      "loss": 0.3918,
      "num_input_tokens_seen": 15123208,
      "step": 22965
    },
    {
      "epoch": 13.543632075471699,
      "grad_norm": 3.249363660812378,
      "learning_rate": 2.852926243259531e-06,
      "loss": 0.3239,
      "num_input_tokens_seen": 15126024,
      "step": 22970
    },
    {
      "epoch": 13.546580188679245,
      "grad_norm": 3.5336320400238037,
      "learning_rate": 2.850603090441617e-06,
      "loss": 0.3887,
      "num_input_tokens_seen": 15129288,
      "step": 22975
    },
    {
      "epoch": 13.549528301886792,
      "grad_norm": 6.847203254699707,
      "learning_rate": 2.848280506685798e-06,
      "loss": 0.3817,
      "num_input_tokens_seen": 15132552,
      "step": 22980
    },
    {
      "epoch": 13.55247641509434,
      "grad_norm": 2.441509962081909,
      "learning_rate": 2.845958492606986e-06,
      "loss": 0.3326,
      "num_input_tokens_seen": 15135944,
      "step": 22985
    },
    {
      "epoch": 13.555424528301886,
      "grad_norm": 3.520421266555786,
      "learning_rate": 2.843637048819949e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 15139112,
      "step": 22990
    },
    {
      "epoch": 13.558372641509434,
      "grad_norm": 2.5772359371185303,
      "learning_rate": 2.8413161759392966e-06,
      "loss": 0.2403,
      "num_input_tokens_seen": 15141512,
      "step": 22995
    },
    {
      "epoch": 13.56132075471698,
      "grad_norm": 6.560758113861084,
      "learning_rate": 2.8389958745794878e-06,
      "loss": 0.3726,
      "num_input_tokens_seen": 15144808,
      "step": 23000
    },
    {
      "epoch": 13.564268867924529,
      "grad_norm": 4.806283473968506,
      "learning_rate": 2.8366761453548366e-06,
      "loss": 0.3846,
      "num_input_tokens_seen": 15147464,
      "step": 23005
    },
    {
      "epoch": 13.567216981132075,
      "grad_norm": 3.652193069458008,
      "learning_rate": 2.8343569888795e-06,
      "loss": 0.3504,
      "num_input_tokens_seen": 15150024,
      "step": 23010
    },
    {
      "epoch": 13.570165094339622,
      "grad_norm": 3.042025327682495,
      "learning_rate": 2.832038405767483e-06,
      "loss": 0.4037,
      "num_input_tokens_seen": 15154088,
      "step": 23015
    },
    {
      "epoch": 13.57311320754717,
      "grad_norm": 2.1061036586761475,
      "learning_rate": 2.8297203966326397e-06,
      "loss": 0.3224,
      "num_input_tokens_seen": 15156552,
      "step": 23020
    },
    {
      "epoch": 13.576061320754716,
      "grad_norm": 3.204195976257324,
      "learning_rate": 2.8274029620886773e-06,
      "loss": 0.4009,
      "num_input_tokens_seen": 15160936,
      "step": 23025
    },
    {
      "epoch": 13.579009433962264,
      "grad_norm": 2.185044288635254,
      "learning_rate": 2.825086102749144e-06,
      "loss": 0.3311,
      "num_input_tokens_seen": 15164264,
      "step": 23030
    },
    {
      "epoch": 13.58195754716981,
      "grad_norm": 7.227344036102295,
      "learning_rate": 2.822769819227438e-06,
      "loss": 0.3118,
      "num_input_tokens_seen": 15166696,
      "step": 23035
    },
    {
      "epoch": 13.584905660377359,
      "grad_norm": 13.260885238647461,
      "learning_rate": 2.8204541121368055e-06,
      "loss": 0.4028,
      "num_input_tokens_seen": 15169768,
      "step": 23040
    },
    {
      "epoch": 13.587853773584905,
      "grad_norm": 2.7827320098876953,
      "learning_rate": 2.8181389820903402e-06,
      "loss": 0.3091,
      "num_input_tokens_seen": 15173352,
      "step": 23045
    },
    {
      "epoch": 13.590801886792454,
      "grad_norm": 4.0434889793396,
      "learning_rate": 2.8158244297009814e-06,
      "loss": 0.3173,
      "num_input_tokens_seen": 15177448,
      "step": 23050
    },
    {
      "epoch": 13.59375,
      "grad_norm": 2.732492446899414,
      "learning_rate": 2.8135104555815196e-06,
      "loss": 0.2443,
      "num_input_tokens_seen": 15181256,
      "step": 23055
    },
    {
      "epoch": 13.596698113207546,
      "grad_norm": 3.107440948486328,
      "learning_rate": 2.811197060344588e-06,
      "loss": 0.2822,
      "num_input_tokens_seen": 15184808,
      "step": 23060
    },
    {
      "epoch": 13.599646226415095,
      "grad_norm": 1.4938898086547852,
      "learning_rate": 2.8088842446026677e-06,
      "loss": 0.3712,
      "num_input_tokens_seen": 15188328,
      "step": 23065
    },
    {
      "epoch": 13.602594339622641,
      "grad_norm": 5.091133117675781,
      "learning_rate": 2.806572008968087e-06,
      "loss": 0.2581,
      "num_input_tokens_seen": 15191272,
      "step": 23070
    },
    {
      "epoch": 13.60554245283019,
      "grad_norm": 4.324878692626953,
      "learning_rate": 2.80426035405302e-06,
      "loss": 0.312,
      "num_input_tokens_seen": 15194760,
      "step": 23075
    },
    {
      "epoch": 13.608490566037736,
      "grad_norm": 12.375216484069824,
      "learning_rate": 2.8019492804694852e-06,
      "loss": 0.5028,
      "num_input_tokens_seen": 15201224,
      "step": 23080
    },
    {
      "epoch": 13.611438679245284,
      "grad_norm": 5.323567867279053,
      "learning_rate": 2.799638788829354e-06,
      "loss": 0.2483,
      "num_input_tokens_seen": 15203944,
      "step": 23085
    },
    {
      "epoch": 13.61438679245283,
      "grad_norm": 4.221457481384277,
      "learning_rate": 2.7973288797443367e-06,
      "loss": 0.4172,
      "num_input_tokens_seen": 15206632,
      "step": 23090
    },
    {
      "epoch": 13.617334905660378,
      "grad_norm": 2.8959803581237793,
      "learning_rate": 2.7950195538259884e-06,
      "loss": 0.3293,
      "num_input_tokens_seen": 15209768,
      "step": 23095
    },
    {
      "epoch": 13.620283018867925,
      "grad_norm": 2.3050172328948975,
      "learning_rate": 2.792710811685719e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 15213032,
      "step": 23100
    },
    {
      "epoch": 13.623231132075471,
      "grad_norm": 4.192460060119629,
      "learning_rate": 2.7904026539347743e-06,
      "loss": 0.4736,
      "num_input_tokens_seen": 15216552,
      "step": 23105
    },
    {
      "epoch": 13.62617924528302,
      "grad_norm": 4.998299598693848,
      "learning_rate": 2.7880950811842507e-06,
      "loss": 0.3541,
      "num_input_tokens_seen": 15220648,
      "step": 23110
    },
    {
      "epoch": 13.629127358490566,
      "grad_norm": 3.7447524070739746,
      "learning_rate": 2.785788094045085e-06,
      "loss": 0.3123,
      "num_input_tokens_seen": 15223976,
      "step": 23115
    },
    {
      "epoch": 13.632075471698114,
      "grad_norm": 5.182453632354736,
      "learning_rate": 2.7834816931280655e-06,
      "loss": 0.3704,
      "num_input_tokens_seen": 15227208,
      "step": 23120
    },
    {
      "epoch": 13.63502358490566,
      "grad_norm": 3.093216896057129,
      "learning_rate": 2.781175879043821e-06,
      "loss": 0.4536,
      "num_input_tokens_seen": 15231176,
      "step": 23125
    },
    {
      "epoch": 13.637971698113208,
      "grad_norm": 4.654433250427246,
      "learning_rate": 2.778870652402825e-06,
      "loss": 0.3897,
      "num_input_tokens_seen": 15233896,
      "step": 23130
    },
    {
      "epoch": 13.640919811320755,
      "grad_norm": 4.292058944702148,
      "learning_rate": 2.776566013815396e-06,
      "loss": 0.2719,
      "num_input_tokens_seen": 15236776,
      "step": 23135
    },
    {
      "epoch": 13.643867924528301,
      "grad_norm": 4.768463134765625,
      "learning_rate": 2.774261963891698e-06,
      "loss": 0.3658,
      "num_input_tokens_seen": 15241064,
      "step": 23140
    },
    {
      "epoch": 13.64681603773585,
      "grad_norm": 3.0166449546813965,
      "learning_rate": 2.771958503241735e-06,
      "loss": 0.2484,
      "num_input_tokens_seen": 15244488,
      "step": 23145
    },
    {
      "epoch": 13.649764150943396,
      "grad_norm": 1.8331224918365479,
      "learning_rate": 2.769655632475362e-06,
      "loss": 0.2104,
      "num_input_tokens_seen": 15247816,
      "step": 23150
    },
    {
      "epoch": 13.652712264150944,
      "grad_norm": 4.186027526855469,
      "learning_rate": 2.7673533522022733e-06,
      "loss": 0.2938,
      "num_input_tokens_seen": 15251848,
      "step": 23155
    },
    {
      "epoch": 13.65566037735849,
      "grad_norm": 4.357036590576172,
      "learning_rate": 2.765051663032007e-06,
      "loss": 0.4161,
      "num_input_tokens_seen": 15254952,
      "step": 23160
    },
    {
      "epoch": 13.658608490566039,
      "grad_norm": 5.593072891235352,
      "learning_rate": 2.7627505655739446e-06,
      "loss": 0.3482,
      "num_input_tokens_seen": 15257448,
      "step": 23165
    },
    {
      "epoch": 13.661556603773585,
      "grad_norm": 1.9798201322555542,
      "learning_rate": 2.7604500604373097e-06,
      "loss": 0.3191,
      "num_input_tokens_seen": 15260680,
      "step": 23170
    },
    {
      "epoch": 13.664504716981131,
      "grad_norm": 2.251962661743164,
      "learning_rate": 2.7581501482311757e-06,
      "loss": 0.2575,
      "num_input_tokens_seen": 15263592,
      "step": 23175
    },
    {
      "epoch": 13.66745283018868,
      "grad_norm": 2.5115573406219482,
      "learning_rate": 2.7558508295644513e-06,
      "loss": 0.3436,
      "num_input_tokens_seen": 15267112,
      "step": 23180
    },
    {
      "epoch": 13.670400943396226,
      "grad_norm": 4.446223735809326,
      "learning_rate": 2.7535521050458922e-06,
      "loss": 0.4172,
      "num_input_tokens_seen": 15270536,
      "step": 23185
    },
    {
      "epoch": 13.673349056603774,
      "grad_norm": 3.995842695236206,
      "learning_rate": 2.7512539752840926e-06,
      "loss": 0.374,
      "num_input_tokens_seen": 15273064,
      "step": 23190
    },
    {
      "epoch": 13.67629716981132,
      "grad_norm": 3.3284246921539307,
      "learning_rate": 2.748956440887497e-06,
      "loss": 0.3433,
      "num_input_tokens_seen": 15275976,
      "step": 23195
    },
    {
      "epoch": 13.679245283018869,
      "grad_norm": 4.998994827270508,
      "learning_rate": 2.7466595024643843e-06,
      "loss": 0.3399,
      "num_input_tokens_seen": 15279592,
      "step": 23200
    },
    {
      "epoch": 13.682193396226415,
      "grad_norm": 5.062131881713867,
      "learning_rate": 2.744363160622878e-06,
      "loss": 0.312,
      "num_input_tokens_seen": 15282120,
      "step": 23205
    },
    {
      "epoch": 13.685141509433961,
      "grad_norm": 2.6361846923828125,
      "learning_rate": 2.742067415970948e-06,
      "loss": 0.2724,
      "num_input_tokens_seen": 15285320,
      "step": 23210
    },
    {
      "epoch": 13.68808962264151,
      "grad_norm": 3.6439900398254395,
      "learning_rate": 2.739772269116402e-06,
      "loss": 0.3669,
      "num_input_tokens_seen": 15288648,
      "step": 23215
    },
    {
      "epoch": 13.691037735849056,
      "grad_norm": 5.086939334869385,
      "learning_rate": 2.7374777206668874e-06,
      "loss": 0.3003,
      "num_input_tokens_seen": 15291688,
      "step": 23220
    },
    {
      "epoch": 13.693985849056604,
      "grad_norm": 1.6106923818588257,
      "learning_rate": 2.735183771229898e-06,
      "loss": 0.2798,
      "num_input_tokens_seen": 15295496,
      "step": 23225
    },
    {
      "epoch": 13.69693396226415,
      "grad_norm": 5.102258682250977,
      "learning_rate": 2.732890421412765e-06,
      "loss": 0.2315,
      "num_input_tokens_seen": 15298440,
      "step": 23230
    },
    {
      "epoch": 13.699882075471699,
      "grad_norm": 12.061673164367676,
      "learning_rate": 2.7305976718226624e-06,
      "loss": 0.5454,
      "num_input_tokens_seen": 15302632,
      "step": 23235
    },
    {
      "epoch": 13.702830188679245,
      "grad_norm": 2.1046104431152344,
      "learning_rate": 2.728305523066609e-06,
      "loss": 0.3489,
      "num_input_tokens_seen": 15306568,
      "step": 23240
    },
    {
      "epoch": 13.705778301886792,
      "grad_norm": 3.7645201683044434,
      "learning_rate": 2.726013975751458e-06,
      "loss": 0.3517,
      "num_input_tokens_seen": 15309992,
      "step": 23245
    },
    {
      "epoch": 13.70872641509434,
      "grad_norm": 5.3610358238220215,
      "learning_rate": 2.723723030483908e-06,
      "loss": 0.4006,
      "num_input_tokens_seen": 15312776,
      "step": 23250
    },
    {
      "epoch": 13.711674528301886,
      "grad_norm": 2.5524303913116455,
      "learning_rate": 2.7214326878704953e-06,
      "loss": 0.313,
      "num_input_tokens_seen": 15315752,
      "step": 23255
    },
    {
      "epoch": 13.714622641509434,
      "grad_norm": 3.1992132663726807,
      "learning_rate": 2.7191429485175993e-06,
      "loss": 0.1979,
      "num_input_tokens_seen": 15318856,
      "step": 23260
    },
    {
      "epoch": 13.71757075471698,
      "grad_norm": 2.304063558578491,
      "learning_rate": 2.716853813031435e-06,
      "loss": 0.3393,
      "num_input_tokens_seen": 15322376,
      "step": 23265
    },
    {
      "epoch": 13.720518867924529,
      "grad_norm": 3.885200023651123,
      "learning_rate": 2.714565282018066e-06,
      "loss": 0.3262,
      "num_input_tokens_seen": 15325512,
      "step": 23270
    },
    {
      "epoch": 13.723466981132075,
      "grad_norm": 7.148113250732422,
      "learning_rate": 2.7122773560833877e-06,
      "loss": 0.3941,
      "num_input_tokens_seen": 15329224,
      "step": 23275
    },
    {
      "epoch": 13.726415094339622,
      "grad_norm": 3.4381630420684814,
      "learning_rate": 2.709990035833139e-06,
      "loss": 0.263,
      "num_input_tokens_seen": 15331976,
      "step": 23280
    },
    {
      "epoch": 13.72936320754717,
      "grad_norm": 5.805787086486816,
      "learning_rate": 2.707703321872896e-06,
      "loss": 0.3499,
      "num_input_tokens_seen": 15336104,
      "step": 23285
    },
    {
      "epoch": 13.732311320754716,
      "grad_norm": 3.4407191276550293,
      "learning_rate": 2.705417214808079e-06,
      "loss": 0.3499,
      "num_input_tokens_seen": 15339432,
      "step": 23290
    },
    {
      "epoch": 13.735259433962264,
      "grad_norm": 3.6199920177459717,
      "learning_rate": 2.703131715243945e-06,
      "loss": 0.2893,
      "num_input_tokens_seen": 15342344,
      "step": 23295
    },
    {
      "epoch": 13.73820754716981,
      "grad_norm": 7.179245471954346,
      "learning_rate": 2.7008468237855855e-06,
      "loss": 0.2748,
      "num_input_tokens_seen": 15345224,
      "step": 23300
    },
    {
      "epoch": 13.741155660377359,
      "grad_norm": 5.248873233795166,
      "learning_rate": 2.69856254103794e-06,
      "loss": 0.2881,
      "num_input_tokens_seen": 15348296,
      "step": 23305
    },
    {
      "epoch": 13.744103773584905,
      "grad_norm": 5.091902732849121,
      "learning_rate": 2.6962788676057806e-06,
      "loss": 0.4243,
      "num_input_tokens_seen": 15351592,
      "step": 23310
    },
    {
      "epoch": 13.747051886792454,
      "grad_norm": 5.444810390472412,
      "learning_rate": 2.69399580409372e-06,
      "loss": 0.2624,
      "num_input_tokens_seen": 15354664,
      "step": 23315
    },
    {
      "epoch": 13.75,
      "grad_norm": 4.431718826293945,
      "learning_rate": 2.6917133511062076e-06,
      "loss": 0.3973,
      "num_input_tokens_seen": 15357000,
      "step": 23320
    },
    {
      "epoch": 13.752948113207546,
      "grad_norm": 6.028803825378418,
      "learning_rate": 2.6894315092475342e-06,
      "loss": 0.4247,
      "num_input_tokens_seen": 15360136,
      "step": 23325
    },
    {
      "epoch": 13.755896226415095,
      "grad_norm": 2.4867806434631348,
      "learning_rate": 2.6871502791218245e-06,
      "loss": 0.3191,
      "num_input_tokens_seen": 15364136,
      "step": 23330
    },
    {
      "epoch": 13.758844339622641,
      "grad_norm": 5.734257698059082,
      "learning_rate": 2.684869661333048e-06,
      "loss": 0.4235,
      "num_input_tokens_seen": 15367240,
      "step": 23335
    },
    {
      "epoch": 13.76179245283019,
      "grad_norm": 5.907078742980957,
      "learning_rate": 2.6825896564850074e-06,
      "loss": 0.2795,
      "num_input_tokens_seen": 15369864,
      "step": 23340
    },
    {
      "epoch": 13.764740566037736,
      "grad_norm": 5.276105880737305,
      "learning_rate": 2.6803102651813416e-06,
      "loss": 0.273,
      "num_input_tokens_seen": 15373224,
      "step": 23345
    },
    {
      "epoch": 13.767688679245284,
      "grad_norm": 3.898656129837036,
      "learning_rate": 2.6780314880255307e-06,
      "loss": 0.3606,
      "num_input_tokens_seen": 15376296,
      "step": 23350
    },
    {
      "epoch": 13.77063679245283,
      "grad_norm": 5.447664260864258,
      "learning_rate": 2.675753325620891e-06,
      "loss": 0.2954,
      "num_input_tokens_seen": 15379400,
      "step": 23355
    },
    {
      "epoch": 13.773584905660378,
      "grad_norm": 4.047502517700195,
      "learning_rate": 2.6734757785705727e-06,
      "loss": 0.2765,
      "num_input_tokens_seen": 15382184,
      "step": 23360
    },
    {
      "epoch": 13.776533018867925,
      "grad_norm": 5.873675346374512,
      "learning_rate": 2.6711988474775712e-06,
      "loss": 0.3166,
      "num_input_tokens_seen": 15385000,
      "step": 23365
    },
    {
      "epoch": 13.779481132075471,
      "grad_norm": 3.249819278717041,
      "learning_rate": 2.668922532944711e-06,
      "loss": 0.3679,
      "num_input_tokens_seen": 15388968,
      "step": 23370
    },
    {
      "epoch": 13.78242924528302,
      "grad_norm": 5.055783748626709,
      "learning_rate": 2.6666468355746566e-06,
      "loss": 0.3627,
      "num_input_tokens_seen": 15391912,
      "step": 23375
    },
    {
      "epoch": 13.785377358490566,
      "grad_norm": 3.954352378845215,
      "learning_rate": 2.6643717559699073e-06,
      "loss": 0.3694,
      "num_input_tokens_seen": 15396168,
      "step": 23380
    },
    {
      "epoch": 13.788325471698114,
      "grad_norm": 4.042622089385986,
      "learning_rate": 2.662097294732803e-06,
      "loss": 0.481,
      "num_input_tokens_seen": 15399208,
      "step": 23385
    },
    {
      "epoch": 13.79127358490566,
      "grad_norm": 1.8504462242126465,
      "learning_rate": 2.6598234524655165e-06,
      "loss": 0.285,
      "num_input_tokens_seen": 15402632,
      "step": 23390
    },
    {
      "epoch": 13.794221698113208,
      "grad_norm": 8.975600242614746,
      "learning_rate": 2.657550229770054e-06,
      "loss": 0.4652,
      "num_input_tokens_seen": 15406248,
      "step": 23395
    },
    {
      "epoch": 13.797169811320755,
      "grad_norm": 7.3131279945373535,
      "learning_rate": 2.655277627248265e-06,
      "loss": 0.4795,
      "num_input_tokens_seen": 15410728,
      "step": 23400
    },
    {
      "epoch": 13.800117924528301,
      "grad_norm": 4.0428643226623535,
      "learning_rate": 2.65300564550183e-06,
      "loss": 0.4031,
      "num_input_tokens_seen": 15414312,
      "step": 23405
    },
    {
      "epoch": 13.80306603773585,
      "grad_norm": 4.513278007507324,
      "learning_rate": 2.6507342851322647e-06,
      "loss": 0.4696,
      "num_input_tokens_seen": 15417576,
      "step": 23410
    },
    {
      "epoch": 13.806014150943396,
      "grad_norm": 4.72562837600708,
      "learning_rate": 2.6484635467409233e-06,
      "loss": 0.3221,
      "num_input_tokens_seen": 15420968,
      "step": 23415
    },
    {
      "epoch": 13.808962264150944,
      "grad_norm": 3.779479742050171,
      "learning_rate": 2.64619343092899e-06,
      "loss": 0.4141,
      "num_input_tokens_seen": 15424296,
      "step": 23420
    },
    {
      "epoch": 13.81191037735849,
      "grad_norm": 4.0299153327941895,
      "learning_rate": 2.643923938297492e-06,
      "loss": 0.306,
      "num_input_tokens_seen": 15426984,
      "step": 23425
    },
    {
      "epoch": 13.814858490566039,
      "grad_norm": 4.499722003936768,
      "learning_rate": 2.6416550694472855e-06,
      "loss": 0.4037,
      "num_input_tokens_seen": 15430152,
      "step": 23430
    },
    {
      "epoch": 13.817806603773585,
      "grad_norm": 3.184877872467041,
      "learning_rate": 2.639386824979063e-06,
      "loss": 0.4638,
      "num_input_tokens_seen": 15433000,
      "step": 23435
    },
    {
      "epoch": 13.820754716981131,
      "grad_norm": 5.199819087982178,
      "learning_rate": 2.6371192054933525e-06,
      "loss": 0.328,
      "num_input_tokens_seen": 15436584,
      "step": 23440
    },
    {
      "epoch": 13.82370283018868,
      "grad_norm": 9.575271606445312,
      "learning_rate": 2.634852211590516e-06,
      "loss": 0.3026,
      "num_input_tokens_seen": 15439272,
      "step": 23445
    },
    {
      "epoch": 13.826650943396226,
      "grad_norm": 5.869104385375977,
      "learning_rate": 2.6325858438707473e-06,
      "loss": 0.3002,
      "num_input_tokens_seen": 15441928,
      "step": 23450
    },
    {
      "epoch": 13.829599056603774,
      "grad_norm": 3.9495084285736084,
      "learning_rate": 2.630320102934082e-06,
      "loss": 0.362,
      "num_input_tokens_seen": 15444680,
      "step": 23455
    },
    {
      "epoch": 13.83254716981132,
      "grad_norm": 2.8376951217651367,
      "learning_rate": 2.628054989380382e-06,
      "loss": 0.3706,
      "num_input_tokens_seen": 15447304,
      "step": 23460
    },
    {
      "epoch": 13.835495283018869,
      "grad_norm": 2.8224995136260986,
      "learning_rate": 2.625790503809346e-06,
      "loss": 0.3384,
      "num_input_tokens_seen": 15451688,
      "step": 23465
    },
    {
      "epoch": 13.838443396226415,
      "grad_norm": 2.4192721843719482,
      "learning_rate": 2.6235266468205067e-06,
      "loss": 0.2223,
      "num_input_tokens_seen": 15454664,
      "step": 23470
    },
    {
      "epoch": 13.841391509433961,
      "grad_norm": 5.917155742645264,
      "learning_rate": 2.621263419013227e-06,
      "loss": 0.5469,
      "num_input_tokens_seen": 15458344,
      "step": 23475
    },
    {
      "epoch": 13.84433962264151,
      "grad_norm": 4.009472370147705,
      "learning_rate": 2.619000820986711e-06,
      "loss": 0.383,
      "num_input_tokens_seen": 15461128,
      "step": 23480
    },
    {
      "epoch": 13.847287735849056,
      "grad_norm": 3.641308069229126,
      "learning_rate": 2.616738853339988e-06,
      "loss": 0.2277,
      "num_input_tokens_seen": 15464648,
      "step": 23485
    },
    {
      "epoch": 13.850235849056604,
      "grad_norm": 2.9751739501953125,
      "learning_rate": 2.614477516671926e-06,
      "loss": 0.3104,
      "num_input_tokens_seen": 15468040,
      "step": 23490
    },
    {
      "epoch": 13.85318396226415,
      "grad_norm": 3.4363701343536377,
      "learning_rate": 2.612216811581223e-06,
      "loss": 0.4968,
      "num_input_tokens_seen": 15471528,
      "step": 23495
    },
    {
      "epoch": 13.856132075471699,
      "grad_norm": 4.130477428436279,
      "learning_rate": 2.6099567386664095e-06,
      "loss": 0.3312,
      "num_input_tokens_seen": 15474440,
      "step": 23500
    },
    {
      "epoch": 13.859080188679245,
      "grad_norm": 3.585278034210205,
      "learning_rate": 2.60769729852585e-06,
      "loss": 0.2659,
      "num_input_tokens_seen": 15477416,
      "step": 23505
    },
    {
      "epoch": 13.862028301886792,
      "grad_norm": 4.182701587677002,
      "learning_rate": 2.6054384917577413e-06,
      "loss": 0.3575,
      "num_input_tokens_seen": 15480392,
      "step": 23510
    },
    {
      "epoch": 13.86497641509434,
      "grad_norm": 2.517836570739746,
      "learning_rate": 2.60318031896011e-06,
      "loss": 0.3159,
      "num_input_tokens_seen": 15483592,
      "step": 23515
    },
    {
      "epoch": 13.867924528301886,
      "grad_norm": 9.471793174743652,
      "learning_rate": 2.60092278073082e-06,
      "loss": 0.3606,
      "num_input_tokens_seen": 15486504,
      "step": 23520
    },
    {
      "epoch": 13.870872641509434,
      "grad_norm": 3.2678987979888916,
      "learning_rate": 2.5986658776675644e-06,
      "loss": 0.5032,
      "num_input_tokens_seen": 15489736,
      "step": 23525
    },
    {
      "epoch": 13.87382075471698,
      "grad_norm": 3.7002339363098145,
      "learning_rate": 2.5964096103678666e-06,
      "loss": 0.2873,
      "num_input_tokens_seen": 15493096,
      "step": 23530
    },
    {
      "epoch": 13.876768867924529,
      "grad_norm": 4.512133598327637,
      "learning_rate": 2.5941539794290833e-06,
      "loss": 0.351,
      "num_input_tokens_seen": 15496616,
      "step": 23535
    },
    {
      "epoch": 13.879716981132075,
      "grad_norm": 2.608203172683716,
      "learning_rate": 2.5918989854484024e-06,
      "loss": 0.3289,
      "num_input_tokens_seen": 15500168,
      "step": 23540
    },
    {
      "epoch": 13.882665094339622,
      "grad_norm": 2.908334970474243,
      "learning_rate": 2.5896446290228417e-06,
      "loss": 0.2606,
      "num_input_tokens_seen": 15503464,
      "step": 23545
    },
    {
      "epoch": 13.88561320754717,
      "grad_norm": 2.772916078567505,
      "learning_rate": 2.5873909107492547e-06,
      "loss": 0.3382,
      "num_input_tokens_seen": 15506440,
      "step": 23550
    },
    {
      "epoch": 13.888561320754716,
      "grad_norm": 5.00309944152832,
      "learning_rate": 2.5851378312243224e-06,
      "loss": 0.3282,
      "num_input_tokens_seen": 15509032,
      "step": 23555
    },
    {
      "epoch": 13.891509433962264,
      "grad_norm": 3.8531744480133057,
      "learning_rate": 2.5828853910445572e-06,
      "loss": 0.2648,
      "num_input_tokens_seen": 15512264,
      "step": 23560
    },
    {
      "epoch": 13.89445754716981,
      "grad_norm": 3.0112502574920654,
      "learning_rate": 2.5806335908063012e-06,
      "loss": 0.4764,
      "num_input_tokens_seen": 15516072,
      "step": 23565
    },
    {
      "epoch": 13.897405660377359,
      "grad_norm": 3.2789182662963867,
      "learning_rate": 2.5783824311057293e-06,
      "loss": 0.2827,
      "num_input_tokens_seen": 15518952,
      "step": 23570
    },
    {
      "epoch": 13.900353773584905,
      "grad_norm": 3.5943098068237305,
      "learning_rate": 2.5761319125388433e-06,
      "loss": 0.2702,
      "num_input_tokens_seen": 15523272,
      "step": 23575
    },
    {
      "epoch": 13.903301886792454,
      "grad_norm": 3.1326611042022705,
      "learning_rate": 2.57388203570148e-06,
      "loss": 0.3749,
      "num_input_tokens_seen": 15526440,
      "step": 23580
    },
    {
      "epoch": 13.90625,
      "grad_norm": 3.0987515449523926,
      "learning_rate": 2.5716328011893055e-06,
      "loss": 0.2858,
      "num_input_tokens_seen": 15530152,
      "step": 23585
    },
    {
      "epoch": 13.909198113207546,
      "grad_norm": 2.587846279144287,
      "learning_rate": 2.5693842095978127e-06,
      "loss": 0.3613,
      "num_input_tokens_seen": 15534568,
      "step": 23590
    },
    {
      "epoch": 13.912146226415095,
      "grad_norm": 4.206803321838379,
      "learning_rate": 2.567136261522325e-06,
      "loss": 0.3121,
      "num_input_tokens_seen": 15537416,
      "step": 23595
    },
    {
      "epoch": 13.915094339622641,
      "grad_norm": 3.972228527069092,
      "learning_rate": 2.5648889575579985e-06,
      "loss": 0.292,
      "num_input_tokens_seen": 15541832,
      "step": 23600
    },
    {
      "epoch": 13.91804245283019,
      "grad_norm": 3.136620283126831,
      "learning_rate": 2.562642298299814e-06,
      "loss": 0.3659,
      "num_input_tokens_seen": 15544648,
      "step": 23605
    },
    {
      "epoch": 13.920990566037736,
      "grad_norm": 3.683452606201172,
      "learning_rate": 2.560396284342584e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 15547400,
      "step": 23610
    },
    {
      "epoch": 13.923938679245284,
      "grad_norm": 5.036207675933838,
      "learning_rate": 2.558150916280954e-06,
      "loss": 0.3518,
      "num_input_tokens_seen": 15549512,
      "step": 23615
    },
    {
      "epoch": 13.92688679245283,
      "grad_norm": 2.2549288272857666,
      "learning_rate": 2.555906194709392e-06,
      "loss": 0.329,
      "num_input_tokens_seen": 15553096,
      "step": 23620
    },
    {
      "epoch": 13.929834905660378,
      "grad_norm": 4.366921424865723,
      "learning_rate": 2.553662120222199e-06,
      "loss": 0.3133,
      "num_input_tokens_seen": 15556360,
      "step": 23625
    },
    {
      "epoch": 13.932783018867925,
      "grad_norm": 4.038727283477783,
      "learning_rate": 2.5514186934135026e-06,
      "loss": 0.2819,
      "num_input_tokens_seen": 15559272,
      "step": 23630
    },
    {
      "epoch": 13.935731132075471,
      "grad_norm": 2.267454147338867,
      "learning_rate": 2.54917591487726e-06,
      "loss": 0.414,
      "num_input_tokens_seen": 15563144,
      "step": 23635
    },
    {
      "epoch": 13.93867924528302,
      "grad_norm": 5.794092178344727,
      "learning_rate": 2.5469337852072547e-06,
      "loss": 0.2534,
      "num_input_tokens_seen": 15565768,
      "step": 23640
    },
    {
      "epoch": 13.941627358490566,
      "grad_norm": 1.7401381731033325,
      "learning_rate": 2.5446923049971035e-06,
      "loss": 0.3777,
      "num_input_tokens_seen": 15569192,
      "step": 23645
    },
    {
      "epoch": 13.944575471698114,
      "grad_norm": 3.4032106399536133,
      "learning_rate": 2.5424514748402463e-06,
      "loss": 0.2661,
      "num_input_tokens_seen": 15571400,
      "step": 23650
    },
    {
      "epoch": 13.94752358490566,
      "grad_norm": 4.20936393737793,
      "learning_rate": 2.540211295329953e-06,
      "loss": 0.3404,
      "num_input_tokens_seen": 15575432,
      "step": 23655
    },
    {
      "epoch": 13.950471698113208,
      "grad_norm": 3.217743158340454,
      "learning_rate": 2.5379717670593197e-06,
      "loss": 0.3464,
      "num_input_tokens_seen": 15579240,
      "step": 23660
    },
    {
      "epoch": 13.953419811320755,
      "grad_norm": 6.5911149978637695,
      "learning_rate": 2.53573289062127e-06,
      "loss": 0.5137,
      "num_input_tokens_seen": 15582120,
      "step": 23665
    },
    {
      "epoch": 13.956367924528301,
      "grad_norm": 3.5559744834899902,
      "learning_rate": 2.5334946666085605e-06,
      "loss": 0.4299,
      "num_input_tokens_seen": 15585288,
      "step": 23670
    },
    {
      "epoch": 13.95931603773585,
      "grad_norm": 3.7903659343719482,
      "learning_rate": 2.531257095613766e-06,
      "loss": 0.3311,
      "num_input_tokens_seen": 15588264,
      "step": 23675
    },
    {
      "epoch": 13.962264150943396,
      "grad_norm": 5.182392597198486,
      "learning_rate": 2.529020178229297e-06,
      "loss": 0.3197,
      "num_input_tokens_seen": 15591720,
      "step": 23680
    },
    {
      "epoch": 13.965212264150944,
      "grad_norm": 2.0627779960632324,
      "learning_rate": 2.5267839150473846e-06,
      "loss": 0.3382,
      "num_input_tokens_seen": 15595176,
      "step": 23685
    },
    {
      "epoch": 13.96816037735849,
      "grad_norm": 3.651944398880005,
      "learning_rate": 2.5245483066600896e-06,
      "loss": 0.4437,
      "num_input_tokens_seen": 15597736,
      "step": 23690
    },
    {
      "epoch": 13.971108490566039,
      "grad_norm": 4.575820446014404,
      "learning_rate": 2.5223133536592996e-06,
      "loss": 0.3949,
      "num_input_tokens_seen": 15601096,
      "step": 23695
    },
    {
      "epoch": 13.974056603773585,
      "grad_norm": 4.272735595703125,
      "learning_rate": 2.520079056636725e-06,
      "loss": 0.2182,
      "num_input_tokens_seen": 15604584,
      "step": 23700
    },
    {
      "epoch": 13.977004716981131,
      "grad_norm": 2.327298402786255,
      "learning_rate": 2.5178454161839106e-06,
      "loss": 0.2664,
      "num_input_tokens_seen": 15607560,
      "step": 23705
    },
    {
      "epoch": 13.97995283018868,
      "grad_norm": 2.7227251529693604,
      "learning_rate": 2.5156124328922195e-06,
      "loss": 0.3147,
      "num_input_tokens_seen": 15613512,
      "step": 23710
    },
    {
      "epoch": 13.982900943396226,
      "grad_norm": 2.5315701961517334,
      "learning_rate": 2.513380107352844e-06,
      "loss": 0.3868,
      "num_input_tokens_seen": 15617800,
      "step": 23715
    },
    {
      "epoch": 13.985849056603774,
      "grad_norm": 3.030085325241089,
      "learning_rate": 2.5111484401568014e-06,
      "loss": 0.3392,
      "num_input_tokens_seen": 15621224,
      "step": 23720
    },
    {
      "epoch": 13.98879716981132,
      "grad_norm": 3.772409200668335,
      "learning_rate": 2.508917431894936e-06,
      "loss": 0.2901,
      "num_input_tokens_seen": 15625032,
      "step": 23725
    },
    {
      "epoch": 13.991745283018869,
      "grad_norm": 3.4239954948425293,
      "learning_rate": 2.5066870831579144e-06,
      "loss": 0.2862,
      "num_input_tokens_seen": 15628008,
      "step": 23730
    },
    {
      "epoch": 13.994693396226415,
      "grad_norm": 3.3708486557006836,
      "learning_rate": 2.504457394536235e-06,
      "loss": 0.3552,
      "num_input_tokens_seen": 15632392,
      "step": 23735
    },
    {
      "epoch": 13.997641509433961,
      "grad_norm": 4.502853870391846,
      "learning_rate": 2.502228366620216e-06,
      "loss": 0.2905,
      "num_input_tokens_seen": 15635048,
      "step": 23740
    },
    {
      "epoch": 14.0,
      "eval_loss": 0.5701644420623779,
      "eval_runtime": 18.526,
      "eval_samples_per_second": 91.547,
      "eval_steps_per_second": 22.887,
      "num_input_tokens_seen": 15636912,
      "step": 23744
    },
    {
      "epoch": 14.00058962264151,
      "grad_norm": 2.56679368019104,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.4082,
      "num_input_tokens_seen": 15637744,
      "step": 23745
    },
    {
      "epoch": 14.003537735849056,
      "grad_norm": 2.838013172149658,
      "learning_rate": 2.497772295265561e-06,
      "loss": 0.2555,
      "num_input_tokens_seen": 15640912,
      "step": 23750
    },
    {
      "epoch": 14.006485849056604,
      "grad_norm": 3.334449529647827,
      "learning_rate": 2.4955452530066897e-06,
      "loss": 0.3281,
      "num_input_tokens_seen": 15643664,
      "step": 23755
    },
    {
      "epoch": 14.00943396226415,
      "grad_norm": 4.132632255554199,
      "learning_rate": 2.4933188738130043e-06,
      "loss": 0.3617,
      "num_input_tokens_seen": 15646640,
      "step": 23760
    },
    {
      "epoch": 14.012382075471699,
      "grad_norm": 3.010655641555786,
      "learning_rate": 2.49109315827395e-06,
      "loss": 0.3329,
      "num_input_tokens_seen": 15649648,
      "step": 23765
    },
    {
      "epoch": 14.015330188679245,
      "grad_norm": 3.435429811477661,
      "learning_rate": 2.4888681069787975e-06,
      "loss": 0.4057,
      "num_input_tokens_seen": 15652720,
      "step": 23770
    },
    {
      "epoch": 14.018278301886792,
      "grad_norm": 3.3699448108673096,
      "learning_rate": 2.4866437205166353e-06,
      "loss": 0.3807,
      "num_input_tokens_seen": 15655536,
      "step": 23775
    },
    {
      "epoch": 14.02122641509434,
      "grad_norm": 2.475311040878296,
      "learning_rate": 2.4844199994763803e-06,
      "loss": 0.2822,
      "num_input_tokens_seen": 15658640,
      "step": 23780
    },
    {
      "epoch": 14.024174528301886,
      "grad_norm": 2.091373920440674,
      "learning_rate": 2.482196944446772e-06,
      "loss": 0.3251,
      "num_input_tokens_seen": 15661360,
      "step": 23785
    },
    {
      "epoch": 14.027122641509434,
      "grad_norm": 3.9984920024871826,
      "learning_rate": 2.4799745560163736e-06,
      "loss": 0.3203,
      "num_input_tokens_seen": 15664368,
      "step": 23790
    },
    {
      "epoch": 14.03007075471698,
      "grad_norm": 3.2846572399139404,
      "learning_rate": 2.4777528347735707e-06,
      "loss": 0.3027,
      "num_input_tokens_seen": 15666704,
      "step": 23795
    },
    {
      "epoch": 14.033018867924529,
      "grad_norm": 3.927797555923462,
      "learning_rate": 2.4755317813065766e-06,
      "loss": 0.3001,
      "num_input_tokens_seen": 15669168,
      "step": 23800
    },
    {
      "epoch": 14.035966981132075,
      "grad_norm": 2.811427116394043,
      "learning_rate": 2.4733113962034234e-06,
      "loss": 0.2512,
      "num_input_tokens_seen": 15672880,
      "step": 23805
    },
    {
      "epoch": 14.038915094339623,
      "grad_norm": 8.852548599243164,
      "learning_rate": 2.4710916800519674e-06,
      "loss": 0.3952,
      "num_input_tokens_seen": 15674896,
      "step": 23810
    },
    {
      "epoch": 14.04186320754717,
      "grad_norm": 3.880732536315918,
      "learning_rate": 2.4688726334398883e-06,
      "loss": 0.3198,
      "num_input_tokens_seen": 15677296,
      "step": 23815
    },
    {
      "epoch": 14.044811320754716,
      "grad_norm": 3.7080793380737305,
      "learning_rate": 2.466654256954688e-06,
      "loss": 0.2681,
      "num_input_tokens_seen": 15680720,
      "step": 23820
    },
    {
      "epoch": 14.047759433962264,
      "grad_norm": 2.557708978652954,
      "learning_rate": 2.4644365511836895e-06,
      "loss": 0.2255,
      "num_input_tokens_seen": 15683696,
      "step": 23825
    },
    {
      "epoch": 14.05070754716981,
      "grad_norm": 4.113070011138916,
      "learning_rate": 2.4622195167140432e-06,
      "loss": 0.3898,
      "num_input_tokens_seen": 15686608,
      "step": 23830
    },
    {
      "epoch": 14.053655660377359,
      "grad_norm": 4.2034454345703125,
      "learning_rate": 2.4600031541327173e-06,
      "loss": 0.3289,
      "num_input_tokens_seen": 15690096,
      "step": 23835
    },
    {
      "epoch": 14.056603773584905,
      "grad_norm": 4.371036529541016,
      "learning_rate": 2.457787464026503e-06,
      "loss": 0.2721,
      "num_input_tokens_seen": 15692624,
      "step": 23840
    },
    {
      "epoch": 14.059551886792454,
      "grad_norm": 3.960744857788086,
      "learning_rate": 2.455572446982014e-06,
      "loss": 0.3008,
      "num_input_tokens_seen": 15697488,
      "step": 23845
    },
    {
      "epoch": 14.0625,
      "grad_norm": 2.991431474685669,
      "learning_rate": 2.453358103585686e-06,
      "loss": 0.3316,
      "num_input_tokens_seen": 15700432,
      "step": 23850
    },
    {
      "epoch": 14.065448113207546,
      "grad_norm": 3.1338746547698975,
      "learning_rate": 2.4511444344237733e-06,
      "loss": 0.3776,
      "num_input_tokens_seen": 15703696,
      "step": 23855
    },
    {
      "epoch": 14.068396226415095,
      "grad_norm": 3.21616792678833,
      "learning_rate": 2.4489314400823567e-06,
      "loss": 0.3077,
      "num_input_tokens_seen": 15706768,
      "step": 23860
    },
    {
      "epoch": 14.071344339622641,
      "grad_norm": 3.9419102668762207,
      "learning_rate": 2.446719121147337e-06,
      "loss": 0.2623,
      "num_input_tokens_seen": 15709552,
      "step": 23865
    },
    {
      "epoch": 14.07429245283019,
      "grad_norm": 7.354172229766846,
      "learning_rate": 2.4445074782044347e-06,
      "loss": 0.3027,
      "num_input_tokens_seen": 15712272,
      "step": 23870
    },
    {
      "epoch": 14.077240566037736,
      "grad_norm": 2.2231009006500244,
      "learning_rate": 2.442296511839191e-06,
      "loss": 0.2152,
      "num_input_tokens_seen": 15715728,
      "step": 23875
    },
    {
      "epoch": 14.080188679245284,
      "grad_norm": 4.845943450927734,
      "learning_rate": 2.4400862226369687e-06,
      "loss": 0.3359,
      "num_input_tokens_seen": 15718800,
      "step": 23880
    },
    {
      "epoch": 14.08313679245283,
      "grad_norm": 2.0881378650665283,
      "learning_rate": 2.4378766111829514e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 15721968,
      "step": 23885
    },
    {
      "epoch": 14.086084905660377,
      "grad_norm": 4.250504970550537,
      "learning_rate": 2.435667678062142e-06,
      "loss": 0.2865,
      "num_input_tokens_seen": 15725328,
      "step": 23890
    },
    {
      "epoch": 14.089033018867925,
      "grad_norm": 3.373232841491699,
      "learning_rate": 2.4334594238593682e-06,
      "loss": 0.3514,
      "num_input_tokens_seen": 15729232,
      "step": 23895
    },
    {
      "epoch": 14.091981132075471,
      "grad_norm": 2.396521806716919,
      "learning_rate": 2.4312518491592727e-06,
      "loss": 0.3263,
      "num_input_tokens_seen": 15732048,
      "step": 23900
    },
    {
      "epoch": 14.09492924528302,
      "grad_norm": 2.820758819580078,
      "learning_rate": 2.429044954546322e-06,
      "loss": 0.2597,
      "num_input_tokens_seen": 15735216,
      "step": 23905
    },
    {
      "epoch": 14.097877358490566,
      "grad_norm": 3.0249388217926025,
      "learning_rate": 2.426838740604799e-06,
      "loss": 0.3482,
      "num_input_tokens_seen": 15738352,
      "step": 23910
    },
    {
      "epoch": 14.100825471698114,
      "grad_norm": 2.601571559906006,
      "learning_rate": 2.4246332079188066e-06,
      "loss": 0.4198,
      "num_input_tokens_seen": 15742160,
      "step": 23915
    },
    {
      "epoch": 14.10377358490566,
      "grad_norm": 3.6393606662750244,
      "learning_rate": 2.4224283570722745e-06,
      "loss": 0.3097,
      "num_input_tokens_seen": 15746064,
      "step": 23920
    },
    {
      "epoch": 14.106721698113208,
      "grad_norm": 4.418652057647705,
      "learning_rate": 2.420224188648943e-06,
      "loss": 0.3429,
      "num_input_tokens_seen": 15749168,
      "step": 23925
    },
    {
      "epoch": 14.109669811320755,
      "grad_norm": 3.9300107955932617,
      "learning_rate": 2.418020703232376e-06,
      "loss": 0.32,
      "num_input_tokens_seen": 15751920,
      "step": 23930
    },
    {
      "epoch": 14.112617924528301,
      "grad_norm": 3.2761924266815186,
      "learning_rate": 2.4158179014059556e-06,
      "loss": 0.2826,
      "num_input_tokens_seen": 15754768,
      "step": 23935
    },
    {
      "epoch": 14.11556603773585,
      "grad_norm": 3.9057350158691406,
      "learning_rate": 2.413615783752883e-06,
      "loss": 0.2938,
      "num_input_tokens_seen": 15757808,
      "step": 23940
    },
    {
      "epoch": 14.118514150943396,
      "grad_norm": 2.243485450744629,
      "learning_rate": 2.4114143508561767e-06,
      "loss": 0.2979,
      "num_input_tokens_seen": 15761136,
      "step": 23945
    },
    {
      "epoch": 14.121462264150944,
      "grad_norm": 3.044534206390381,
      "learning_rate": 2.4092136032986783e-06,
      "loss": 0.3306,
      "num_input_tokens_seen": 15765968,
      "step": 23950
    },
    {
      "epoch": 14.12441037735849,
      "grad_norm": 4.630002498626709,
      "learning_rate": 2.407013541663043e-06,
      "loss": 0.3968,
      "num_input_tokens_seen": 15768784,
      "step": 23955
    },
    {
      "epoch": 14.127358490566039,
      "grad_norm": 4.7474188804626465,
      "learning_rate": 2.4048141665317493e-06,
      "loss": 0.3451,
      "num_input_tokens_seen": 15772624,
      "step": 23960
    },
    {
      "epoch": 14.130306603773585,
      "grad_norm": 3.465026617050171,
      "learning_rate": 2.40261547848709e-06,
      "loss": 0.3421,
      "num_input_tokens_seen": 15775984,
      "step": 23965
    },
    {
      "epoch": 14.133254716981131,
      "grad_norm": 2.938837766647339,
      "learning_rate": 2.400417478111176e-06,
      "loss": 0.4255,
      "num_input_tokens_seen": 15778416,
      "step": 23970
    },
    {
      "epoch": 14.13620283018868,
      "grad_norm": 3.9986002445220947,
      "learning_rate": 2.3982201659859387e-06,
      "loss": 0.3657,
      "num_input_tokens_seen": 15781424,
      "step": 23975
    },
    {
      "epoch": 14.139150943396226,
      "grad_norm": 5.5473175048828125,
      "learning_rate": 2.3960235426931237e-06,
      "loss": 0.2987,
      "num_input_tokens_seen": 15784240,
      "step": 23980
    },
    {
      "epoch": 14.142099056603774,
      "grad_norm": 5.178963661193848,
      "learning_rate": 2.3938276088143003e-06,
      "loss": 0.3372,
      "num_input_tokens_seen": 15787440,
      "step": 23985
    },
    {
      "epoch": 14.14504716981132,
      "grad_norm": 2.0741913318634033,
      "learning_rate": 2.391632364930849e-06,
      "loss": 0.3126,
      "num_input_tokens_seen": 15790768,
      "step": 23990
    },
    {
      "epoch": 14.147995283018869,
      "grad_norm": 2.0339224338531494,
      "learning_rate": 2.3894378116239706e-06,
      "loss": 0.3391,
      "num_input_tokens_seen": 15794128,
      "step": 23995
    },
    {
      "epoch": 14.150943396226415,
      "grad_norm": 3.825723171234131,
      "learning_rate": 2.387243949474683e-06,
      "loss": 0.3373,
      "num_input_tokens_seen": 15796880,
      "step": 24000
    },
    {
      "epoch": 14.153891509433961,
      "grad_norm": 2.4576375484466553,
      "learning_rate": 2.38505077906382e-06,
      "loss": 0.3723,
      "num_input_tokens_seen": 15801392,
      "step": 24005
    },
    {
      "epoch": 14.15683962264151,
      "grad_norm": 5.9865617752075195,
      "learning_rate": 2.382858300972031e-06,
      "loss": 0.3915,
      "num_input_tokens_seen": 15804592,
      "step": 24010
    },
    {
      "epoch": 14.159787735849056,
      "grad_norm": 4.289112567901611,
      "learning_rate": 2.380666515779788e-06,
      "loss": 0.2903,
      "num_input_tokens_seen": 15807408,
      "step": 24015
    },
    {
      "epoch": 14.162735849056604,
      "grad_norm": 3.9265055656433105,
      "learning_rate": 2.3784754240673734e-06,
      "loss": 0.3589,
      "num_input_tokens_seen": 15810384,
      "step": 24020
    },
    {
      "epoch": 14.16568396226415,
      "grad_norm": 5.746804714202881,
      "learning_rate": 2.3762850264148883e-06,
      "loss": 0.3619,
      "num_input_tokens_seen": 15813776,
      "step": 24025
    },
    {
      "epoch": 14.168632075471699,
      "grad_norm": 2.035407781600952,
      "learning_rate": 2.374095323402251e-06,
      "loss": 0.4063,
      "num_input_tokens_seen": 15820784,
      "step": 24030
    },
    {
      "epoch": 14.171580188679245,
      "grad_norm": 2.0965559482574463,
      "learning_rate": 2.371906315609193e-06,
      "loss": 0.1986,
      "num_input_tokens_seen": 15823824,
      "step": 24035
    },
    {
      "epoch": 14.174528301886792,
      "grad_norm": 4.300044059753418,
      "learning_rate": 2.369718003615263e-06,
      "loss": 0.4196,
      "num_input_tokens_seen": 15826608,
      "step": 24040
    },
    {
      "epoch": 14.17747641509434,
      "grad_norm": 4.150572776794434,
      "learning_rate": 2.3675303879998284e-06,
      "loss": 0.4197,
      "num_input_tokens_seen": 15829776,
      "step": 24045
    },
    {
      "epoch": 14.180424528301886,
      "grad_norm": 3.794004440307617,
      "learning_rate": 2.365343469342068e-06,
      "loss": 0.36,
      "num_input_tokens_seen": 15833424,
      "step": 24050
    },
    {
      "epoch": 14.183372641509434,
      "grad_norm": 2.7420709133148193,
      "learning_rate": 2.3631572482209803e-06,
      "loss": 0.2349,
      "num_input_tokens_seen": 15836656,
      "step": 24055
    },
    {
      "epoch": 14.18632075471698,
      "grad_norm": 5.262619972229004,
      "learning_rate": 2.3609717252153752e-06,
      "loss": 0.3839,
      "num_input_tokens_seen": 15839696,
      "step": 24060
    },
    {
      "epoch": 14.189268867924529,
      "grad_norm": 3.861194372177124,
      "learning_rate": 2.35878690090388e-06,
      "loss": 0.4087,
      "num_input_tokens_seen": 15843504,
      "step": 24065
    },
    {
      "epoch": 14.192216981132075,
      "grad_norm": 3.86714506149292,
      "learning_rate": 2.356602775864935e-06,
      "loss": 0.402,
      "num_input_tokens_seen": 15846800,
      "step": 24070
    },
    {
      "epoch": 14.195165094339623,
      "grad_norm": 4.423323631286621,
      "learning_rate": 2.354419350676796e-06,
      "loss": 0.3219,
      "num_input_tokens_seen": 15849936,
      "step": 24075
    },
    {
      "epoch": 14.19811320754717,
      "grad_norm": 3.148587942123413,
      "learning_rate": 2.3522366259175377e-06,
      "loss": 0.3113,
      "num_input_tokens_seen": 15852976,
      "step": 24080
    },
    {
      "epoch": 14.201061320754716,
      "grad_norm": 2.6742687225341797,
      "learning_rate": 2.350054602165044e-06,
      "loss": 0.2257,
      "num_input_tokens_seen": 15857008,
      "step": 24085
    },
    {
      "epoch": 14.204009433962264,
      "grad_norm": 7.279539108276367,
      "learning_rate": 2.3478732799970143e-06,
      "loss": 0.3784,
      "num_input_tokens_seen": 15859216,
      "step": 24090
    },
    {
      "epoch": 14.20695754716981,
      "grad_norm": 5.410972595214844,
      "learning_rate": 2.3456926599909646e-06,
      "loss": 0.3519,
      "num_input_tokens_seen": 15862512,
      "step": 24095
    },
    {
      "epoch": 14.209905660377359,
      "grad_norm": 2.055349111557007,
      "learning_rate": 2.343512742724222e-06,
      "loss": 0.274,
      "num_input_tokens_seen": 15865840,
      "step": 24100
    },
    {
      "epoch": 14.212853773584905,
      "grad_norm": 4.737220287322998,
      "learning_rate": 2.341333528773928e-06,
      "loss": 0.2337,
      "num_input_tokens_seen": 15869520,
      "step": 24105
    },
    {
      "epoch": 14.215801886792454,
      "grad_norm": 3.936187982559204,
      "learning_rate": 2.3391550187170427e-06,
      "loss": 0.4347,
      "num_input_tokens_seen": 15873072,
      "step": 24110
    },
    {
      "epoch": 14.21875,
      "grad_norm": 2.525886297225952,
      "learning_rate": 2.336977213130333e-06,
      "loss": 0.2922,
      "num_input_tokens_seen": 15876656,
      "step": 24115
    },
    {
      "epoch": 14.221698113207546,
      "grad_norm": 3.21205472946167,
      "learning_rate": 2.3348001125903837e-06,
      "loss": 0.2435,
      "num_input_tokens_seen": 15880368,
      "step": 24120
    },
    {
      "epoch": 14.224646226415095,
      "grad_norm": 3.671283006668091,
      "learning_rate": 2.3326237176735905e-06,
      "loss": 0.2685,
      "num_input_tokens_seen": 15884272,
      "step": 24125
    },
    {
      "epoch": 14.227594339622641,
      "grad_norm": 4.022294521331787,
      "learning_rate": 2.330448028956164e-06,
      "loss": 0.4165,
      "num_input_tokens_seen": 15887376,
      "step": 24130
    },
    {
      "epoch": 14.23054245283019,
      "grad_norm": 2.932244062423706,
      "learning_rate": 2.3282730470141255e-06,
      "loss": 0.3006,
      "num_input_tokens_seen": 15891632,
      "step": 24135
    },
    {
      "epoch": 14.233490566037736,
      "grad_norm": 2.539215326309204,
      "learning_rate": 2.3260987724233143e-06,
      "loss": 0.259,
      "num_input_tokens_seen": 15895408,
      "step": 24140
    },
    {
      "epoch": 14.236438679245284,
      "grad_norm": 3.1366467475891113,
      "learning_rate": 2.323925205759374e-06,
      "loss": 0.3564,
      "num_input_tokens_seen": 15898736,
      "step": 24145
    },
    {
      "epoch": 14.23938679245283,
      "grad_norm": 2.6431145668029785,
      "learning_rate": 2.3217523475977715e-06,
      "loss": 0.2949,
      "num_input_tokens_seen": 15901232,
      "step": 24150
    },
    {
      "epoch": 14.242334905660377,
      "grad_norm": 5.316417694091797,
      "learning_rate": 2.3195801985137773e-06,
      "loss": 0.3959,
      "num_input_tokens_seen": 15904912,
      "step": 24155
    },
    {
      "epoch": 14.245283018867925,
      "grad_norm": 4.18626070022583,
      "learning_rate": 2.317408759082478e-06,
      "loss": 0.2769,
      "num_input_tokens_seen": 15908176,
      "step": 24160
    },
    {
      "epoch": 14.248231132075471,
      "grad_norm": 7.394626140594482,
      "learning_rate": 2.31523802987877e-06,
      "loss": 0.3429,
      "num_input_tokens_seen": 15911600,
      "step": 24165
    },
    {
      "epoch": 14.25117924528302,
      "grad_norm": 5.118335247039795,
      "learning_rate": 2.3130680114773637e-06,
      "loss": 0.2988,
      "num_input_tokens_seen": 15914384,
      "step": 24170
    },
    {
      "epoch": 14.254127358490566,
      "grad_norm": 7.711819648742676,
      "learning_rate": 2.310898704452782e-06,
      "loss": 0.4612,
      "num_input_tokens_seen": 15917328,
      "step": 24175
    },
    {
      "epoch": 14.257075471698114,
      "grad_norm": 3.812613010406494,
      "learning_rate": 2.3087301093793584e-06,
      "loss": 0.2781,
      "num_input_tokens_seen": 15920528,
      "step": 24180
    },
    {
      "epoch": 14.26002358490566,
      "grad_norm": 2.316462516784668,
      "learning_rate": 2.306562226831237e-06,
      "loss": 0.268,
      "num_input_tokens_seen": 15924496,
      "step": 24185
    },
    {
      "epoch": 14.262971698113208,
      "grad_norm": 2.299791097640991,
      "learning_rate": 2.304395057382374e-06,
      "loss": 0.2596,
      "num_input_tokens_seen": 15928048,
      "step": 24190
    },
    {
      "epoch": 14.265919811320755,
      "grad_norm": 2.674621343612671,
      "learning_rate": 2.3022286016065354e-06,
      "loss": 0.3348,
      "num_input_tokens_seen": 15930320,
      "step": 24195
    },
    {
      "epoch": 14.268867924528301,
      "grad_norm": 3.6387057304382324,
      "learning_rate": 2.300062860077303e-06,
      "loss": 0.2971,
      "num_input_tokens_seen": 15933488,
      "step": 24200
    },
    {
      "epoch": 14.27181603773585,
      "grad_norm": 3.2060346603393555,
      "learning_rate": 2.297897833368064e-06,
      "loss": 0.3638,
      "num_input_tokens_seen": 15937456,
      "step": 24205
    },
    {
      "epoch": 14.274764150943396,
      "grad_norm": 6.082944393157959,
      "learning_rate": 2.2957335220520194e-06,
      "loss": 0.2844,
      "num_input_tokens_seen": 15940336,
      "step": 24210
    },
    {
      "epoch": 14.277712264150944,
      "grad_norm": 4.5357513427734375,
      "learning_rate": 2.293569926702179e-06,
      "loss": 0.3234,
      "num_input_tokens_seen": 15943664,
      "step": 24215
    },
    {
      "epoch": 14.28066037735849,
      "grad_norm": 7.216495037078857,
      "learning_rate": 2.291407047891366e-06,
      "loss": 0.2792,
      "num_input_tokens_seen": 15946160,
      "step": 24220
    },
    {
      "epoch": 14.283608490566039,
      "grad_norm": 3.5598535537719727,
      "learning_rate": 2.2892448861922075e-06,
      "loss": 0.3187,
      "num_input_tokens_seen": 15949552,
      "step": 24225
    },
    {
      "epoch": 14.286556603773585,
      "grad_norm": 1.691366195678711,
      "learning_rate": 2.2870834421771505e-06,
      "loss": 0.2214,
      "num_input_tokens_seen": 15953360,
      "step": 24230
    },
    {
      "epoch": 14.289504716981131,
      "grad_norm": 3.8162851333618164,
      "learning_rate": 2.2849227164184433e-06,
      "loss": 0.3072,
      "num_input_tokens_seen": 15955888,
      "step": 24235
    },
    {
      "epoch": 14.29245283018868,
      "grad_norm": 5.005375862121582,
      "learning_rate": 2.2827627094881473e-06,
      "loss": 0.3027,
      "num_input_tokens_seen": 15958480,
      "step": 24240
    },
    {
      "epoch": 14.295400943396226,
      "grad_norm": 3.608095169067383,
      "learning_rate": 2.2806034219581364e-06,
      "loss": 0.3307,
      "num_input_tokens_seen": 15960944,
      "step": 24245
    },
    {
      "epoch": 14.298349056603774,
      "grad_norm": 3.6757993698120117,
      "learning_rate": 2.278444854400089e-06,
      "loss": 0.2943,
      "num_input_tokens_seen": 15963952,
      "step": 24250
    },
    {
      "epoch": 14.30129716981132,
      "grad_norm": 2.5633411407470703,
      "learning_rate": 2.276287007385496e-06,
      "loss": 0.3075,
      "num_input_tokens_seen": 15967184,
      "step": 24255
    },
    {
      "epoch": 14.304245283018869,
      "grad_norm": 3.52093243598938,
      "learning_rate": 2.2741298814856542e-06,
      "loss": 0.2705,
      "num_input_tokens_seen": 15969776,
      "step": 24260
    },
    {
      "epoch": 14.307193396226415,
      "grad_norm": 3.359555959701538,
      "learning_rate": 2.2719734772716763e-06,
      "loss": 0.3575,
      "num_input_tokens_seen": 15973840,
      "step": 24265
    },
    {
      "epoch": 14.310141509433961,
      "grad_norm": 3.254502773284912,
      "learning_rate": 2.269817795314477e-06,
      "loss": 0.3722,
      "num_input_tokens_seen": 15981008,
      "step": 24270
    },
    {
      "epoch": 14.31308962264151,
      "grad_norm": 4.666511535644531,
      "learning_rate": 2.2676628361847834e-06,
      "loss": 0.357,
      "num_input_tokens_seen": 15983472,
      "step": 24275
    },
    {
      "epoch": 14.316037735849056,
      "grad_norm": 4.299685001373291,
      "learning_rate": 2.2655086004531296e-06,
      "loss": 0.4436,
      "num_input_tokens_seen": 15987216,
      "step": 24280
    },
    {
      "epoch": 14.318985849056604,
      "grad_norm": 1.2724264860153198,
      "learning_rate": 2.2633550886898583e-06,
      "loss": 0.2307,
      "num_input_tokens_seen": 15990352,
      "step": 24285
    },
    {
      "epoch": 14.32193396226415,
      "grad_norm": 2.9433021545410156,
      "learning_rate": 2.26120230146512e-06,
      "loss": 0.35,
      "num_input_tokens_seen": 15994064,
      "step": 24290
    },
    {
      "epoch": 14.324882075471699,
      "grad_norm": 4.659343242645264,
      "learning_rate": 2.2590502393488777e-06,
      "loss": 0.387,
      "num_input_tokens_seen": 15996912,
      "step": 24295
    },
    {
      "epoch": 14.327830188679245,
      "grad_norm": 3.0139153003692627,
      "learning_rate": 2.256898902910898e-06,
      "loss": 0.2961,
      "num_input_tokens_seen": 15999472,
      "step": 24300
    },
    {
      "epoch": 14.330778301886792,
      "grad_norm": 3.3156464099884033,
      "learning_rate": 2.2547482927207548e-06,
      "loss": 0.2827,
      "num_input_tokens_seen": 16003088,
      "step": 24305
    },
    {
      "epoch": 14.33372641509434,
      "grad_norm": 2.5410454273223877,
      "learning_rate": 2.252598409347833e-06,
      "loss": 0.3858,
      "num_input_tokens_seen": 16005808,
      "step": 24310
    },
    {
      "epoch": 14.336674528301886,
      "grad_norm": 2.9429991245269775,
      "learning_rate": 2.250449253361323e-06,
      "loss": 0.2975,
      "num_input_tokens_seen": 16012688,
      "step": 24315
    },
    {
      "epoch": 14.339622641509434,
      "grad_norm": 3.998054027557373,
      "learning_rate": 2.2483008253302214e-06,
      "loss": 0.3857,
      "num_input_tokens_seen": 16015504,
      "step": 24320
    },
    {
      "epoch": 14.34257075471698,
      "grad_norm": 4.876596450805664,
      "learning_rate": 2.246153125823337e-06,
      "loss": 0.233,
      "num_input_tokens_seen": 16019120,
      "step": 24325
    },
    {
      "epoch": 14.345518867924529,
      "grad_norm": 3.969491720199585,
      "learning_rate": 2.2440061554092813e-06,
      "loss": 0.2131,
      "num_input_tokens_seen": 16023184,
      "step": 24330
    },
    {
      "epoch": 14.348466981132075,
      "grad_norm": 3.9715402126312256,
      "learning_rate": 2.2418599146564714e-06,
      "loss": 0.37,
      "num_input_tokens_seen": 16025712,
      "step": 24335
    },
    {
      "epoch": 14.351415094339623,
      "grad_norm": 6.847485065460205,
      "learning_rate": 2.239714404133138e-06,
      "loss": 0.3985,
      "num_input_tokens_seen": 16028496,
      "step": 24340
    },
    {
      "epoch": 14.35436320754717,
      "grad_norm": 3.6991987228393555,
      "learning_rate": 2.2375696244073126e-06,
      "loss": 0.365,
      "num_input_tokens_seen": 16031216,
      "step": 24345
    },
    {
      "epoch": 14.357311320754716,
      "grad_norm": 3.6487772464752197,
      "learning_rate": 2.235425576046834e-06,
      "loss": 0.3398,
      "num_input_tokens_seen": 16035440,
      "step": 24350
    },
    {
      "epoch": 14.360259433962264,
      "grad_norm": 1.754635214805603,
      "learning_rate": 2.233282259619347e-06,
      "loss": 0.3048,
      "num_input_tokens_seen": 16040208,
      "step": 24355
    },
    {
      "epoch": 14.36320754716981,
      "grad_norm": 3.0837366580963135,
      "learning_rate": 2.231139675692308e-06,
      "loss": 0.305,
      "num_input_tokens_seen": 16043088,
      "step": 24360
    },
    {
      "epoch": 14.366155660377359,
      "grad_norm": 5.545005798339844,
      "learning_rate": 2.228997824832973e-06,
      "loss": 0.3536,
      "num_input_tokens_seen": 16046736,
      "step": 24365
    },
    {
      "epoch": 14.369103773584905,
      "grad_norm": 2.9053030014038086,
      "learning_rate": 2.226856707608406e-06,
      "loss": 0.309,
      "num_input_tokens_seen": 16049968,
      "step": 24370
    },
    {
      "epoch": 14.372051886792454,
      "grad_norm": 3.279773712158203,
      "learning_rate": 2.2247163245854768e-06,
      "loss": 0.3582,
      "num_input_tokens_seen": 16054224,
      "step": 24375
    },
    {
      "epoch": 14.375,
      "grad_norm": 2.570075035095215,
      "learning_rate": 2.222576676330862e-06,
      "loss": 0.319,
      "num_input_tokens_seen": 16057552,
      "step": 24380
    },
    {
      "epoch": 14.377948113207546,
      "grad_norm": 3.7926697731018066,
      "learning_rate": 2.2204377634110403e-06,
      "loss": 0.4006,
      "num_input_tokens_seen": 16061072,
      "step": 24385
    },
    {
      "epoch": 14.380896226415095,
      "grad_norm": 2.680518865585327,
      "learning_rate": 2.218299586392301e-06,
      "loss": 0.3404,
      "num_input_tokens_seen": 16064272,
      "step": 24390
    },
    {
      "epoch": 14.383844339622641,
      "grad_norm": 2.1203091144561768,
      "learning_rate": 2.2161621458407355e-06,
      "loss": 0.2965,
      "num_input_tokens_seen": 16068080,
      "step": 24395
    },
    {
      "epoch": 14.38679245283019,
      "grad_norm": 2.1112844944000244,
      "learning_rate": 2.2140254423222398e-06,
      "loss": 0.2487,
      "num_input_tokens_seen": 16071728,
      "step": 24400
    },
    {
      "epoch": 14.389740566037736,
      "grad_norm": 6.151156425476074,
      "learning_rate": 2.2118894764025146e-06,
      "loss": 0.409,
      "num_input_tokens_seen": 16074256,
      "step": 24405
    },
    {
      "epoch": 14.392688679245284,
      "grad_norm": 5.17071008682251,
      "learning_rate": 2.2097542486470667e-06,
      "loss": 0.317,
      "num_input_tokens_seen": 16076528,
      "step": 24410
    },
    {
      "epoch": 14.39563679245283,
      "grad_norm": 3.370053291320801,
      "learning_rate": 2.207619759621205e-06,
      "loss": 0.3905,
      "num_input_tokens_seen": 16079664,
      "step": 24415
    },
    {
      "epoch": 14.398584905660377,
      "grad_norm": 3.1228413581848145,
      "learning_rate": 2.205486009890049e-06,
      "loss": 0.2988,
      "num_input_tokens_seen": 16083696,
      "step": 24420
    },
    {
      "epoch": 14.401533018867925,
      "grad_norm": 3.219456911087036,
      "learning_rate": 2.2033530000185146e-06,
      "loss": 0.3919,
      "num_input_tokens_seen": 16086864,
      "step": 24425
    },
    {
      "epoch": 14.404481132075471,
      "grad_norm": 6.446666240692139,
      "learning_rate": 2.2012207305713244e-06,
      "loss": 0.2785,
      "num_input_tokens_seen": 16089392,
      "step": 24430
    },
    {
      "epoch": 14.40742924528302,
      "grad_norm": 3.6427319049835205,
      "learning_rate": 2.19908920211301e-06,
      "loss": 0.3006,
      "num_input_tokens_seen": 16092688,
      "step": 24435
    },
    {
      "epoch": 14.410377358490566,
      "grad_norm": 5.125553607940674,
      "learning_rate": 2.196958415207901e-06,
      "loss": 0.2934,
      "num_input_tokens_seen": 16096240,
      "step": 24440
    },
    {
      "epoch": 14.413325471698114,
      "grad_norm": 4.007106781005859,
      "learning_rate": 2.1948283704201312e-06,
      "loss": 0.3167,
      "num_input_tokens_seen": 16098928,
      "step": 24445
    },
    {
      "epoch": 14.41627358490566,
      "grad_norm": 3.9867866039276123,
      "learning_rate": 2.1926990683136383e-06,
      "loss": 0.2274,
      "num_input_tokens_seen": 16102480,
      "step": 24450
    },
    {
      "epoch": 14.419221698113208,
      "grad_norm": 2.3171842098236084,
      "learning_rate": 2.1905705094521685e-06,
      "loss": 0.2584,
      "num_input_tokens_seen": 16105424,
      "step": 24455
    },
    {
      "epoch": 14.422169811320755,
      "grad_norm": 3.9813897609710693,
      "learning_rate": 2.1884426943992635e-06,
      "loss": 0.3662,
      "num_input_tokens_seen": 16108880,
      "step": 24460
    },
    {
      "epoch": 14.425117924528301,
      "grad_norm": 2.9000203609466553,
      "learning_rate": 2.1863156237182727e-06,
      "loss": 0.2988,
      "num_input_tokens_seen": 16112880,
      "step": 24465
    },
    {
      "epoch": 14.42806603773585,
      "grad_norm": 2.292388677597046,
      "learning_rate": 2.1841892979723466e-06,
      "loss": 0.1975,
      "num_input_tokens_seen": 16115440,
      "step": 24470
    },
    {
      "epoch": 14.431014150943396,
      "grad_norm": 3.1306538581848145,
      "learning_rate": 2.1820637177244375e-06,
      "loss": 0.4245,
      "num_input_tokens_seen": 16118576,
      "step": 24475
    },
    {
      "epoch": 14.433962264150944,
      "grad_norm": 6.461433410644531,
      "learning_rate": 2.179938883537306e-06,
      "loss": 0.3043,
      "num_input_tokens_seen": 16121296,
      "step": 24480
    },
    {
      "epoch": 14.43691037735849,
      "grad_norm": 2.689602851867676,
      "learning_rate": 2.177814795973508e-06,
      "loss": 0.3252,
      "num_input_tokens_seen": 16125296,
      "step": 24485
    },
    {
      "epoch": 14.439858490566039,
      "grad_norm": 4.574707984924316,
      "learning_rate": 2.1756914555954064e-06,
      "loss": 0.2778,
      "num_input_tokens_seen": 16128720,
      "step": 24490
    },
    {
      "epoch": 14.442806603773585,
      "grad_norm": 2.7323415279388428,
      "learning_rate": 2.173568862965164e-06,
      "loss": 0.3819,
      "num_input_tokens_seen": 16132016,
      "step": 24495
    },
    {
      "epoch": 14.445754716981131,
      "grad_norm": 5.932122230529785,
      "learning_rate": 2.171447018644746e-06,
      "loss": 0.295,
      "num_input_tokens_seen": 16134544,
      "step": 24500
    },
    {
      "epoch": 14.44870283018868,
      "grad_norm": 3.772021770477295,
      "learning_rate": 2.1693259231959186e-06,
      "loss": 0.3824,
      "num_input_tokens_seen": 16137168,
      "step": 24505
    },
    {
      "epoch": 14.451650943396226,
      "grad_norm": 4.380198001861572,
      "learning_rate": 2.1672055771802545e-06,
      "loss": 0.3936,
      "num_input_tokens_seen": 16140784,
      "step": 24510
    },
    {
      "epoch": 14.454599056603774,
      "grad_norm": 3.6254918575286865,
      "learning_rate": 2.1650859811591224e-06,
      "loss": 0.321,
      "num_input_tokens_seen": 16144112,
      "step": 24515
    },
    {
      "epoch": 14.45754716981132,
      "grad_norm": 4.345432281494141,
      "learning_rate": 2.1629671356936943e-06,
      "loss": 0.2936,
      "num_input_tokens_seen": 16146800,
      "step": 24520
    },
    {
      "epoch": 14.460495283018869,
      "grad_norm": 2.94218111038208,
      "learning_rate": 2.1608490413449428e-06,
      "loss": 0.3489,
      "num_input_tokens_seen": 16150448,
      "step": 24525
    },
    {
      "epoch": 14.463443396226415,
      "grad_norm": 3.3080039024353027,
      "learning_rate": 2.158731698673645e-06,
      "loss": 0.3405,
      "num_input_tokens_seen": 16153968,
      "step": 24530
    },
    {
      "epoch": 14.466391509433961,
      "grad_norm": 3.667351245880127,
      "learning_rate": 2.1566151082403752e-06,
      "loss": 0.3724,
      "num_input_tokens_seen": 16158608,
      "step": 24535
    },
    {
      "epoch": 14.46933962264151,
      "grad_norm": 2.926290988922119,
      "learning_rate": 2.154499270605508e-06,
      "loss": 0.2717,
      "num_input_tokens_seen": 16161360,
      "step": 24540
    },
    {
      "epoch": 14.472287735849056,
      "grad_norm": 3.93237566947937,
      "learning_rate": 2.1523841863292243e-06,
      "loss": 0.3602,
      "num_input_tokens_seen": 16165168,
      "step": 24545
    },
    {
      "epoch": 14.475235849056604,
      "grad_norm": 5.611141204833984,
      "learning_rate": 2.1502698559714998e-06,
      "loss": 0.2977,
      "num_input_tokens_seen": 16168400,
      "step": 24550
    },
    {
      "epoch": 14.47818396226415,
      "grad_norm": 3.589850664138794,
      "learning_rate": 2.1481562800921125e-06,
      "loss": 0.3392,
      "num_input_tokens_seen": 16171600,
      "step": 24555
    },
    {
      "epoch": 14.481132075471699,
      "grad_norm": 2.8799080848693848,
      "learning_rate": 2.146043459250641e-06,
      "loss": 0.3625,
      "num_input_tokens_seen": 16175952,
      "step": 24560
    },
    {
      "epoch": 14.484080188679245,
      "grad_norm": 3.1596319675445557,
      "learning_rate": 2.1439313940064634e-06,
      "loss": 0.2612,
      "num_input_tokens_seen": 16179664,
      "step": 24565
    },
    {
      "epoch": 14.487028301886792,
      "grad_norm": 5.186420440673828,
      "learning_rate": 2.141820084918756e-06,
      "loss": 0.3384,
      "num_input_tokens_seen": 16182736,
      "step": 24570
    },
    {
      "epoch": 14.48997641509434,
      "grad_norm": 4.902623653411865,
      "learning_rate": 2.1397095325465013e-06,
      "loss": 0.3746,
      "num_input_tokens_seen": 16187216,
      "step": 24575
    },
    {
      "epoch": 14.492924528301886,
      "grad_norm": 2.915117025375366,
      "learning_rate": 2.1375997374484754e-06,
      "loss": 0.2135,
      "num_input_tokens_seen": 16190256,
      "step": 24580
    },
    {
      "epoch": 14.495872641509434,
      "grad_norm": 2.9557065963745117,
      "learning_rate": 2.1354907001832546e-06,
      "loss": 0.5057,
      "num_input_tokens_seen": 16193328,
      "step": 24585
    },
    {
      "epoch": 14.49882075471698,
      "grad_norm": 4.5990142822265625,
      "learning_rate": 2.133382421309217e-06,
      "loss": 0.3863,
      "num_input_tokens_seen": 16197200,
      "step": 24590
    },
    {
      "epoch": 14.501768867924529,
      "grad_norm": 4.9871745109558105,
      "learning_rate": 2.131274901384537e-06,
      "loss": 0.3252,
      "num_input_tokens_seen": 16200112,
      "step": 24595
    },
    {
      "epoch": 14.504716981132075,
      "grad_norm": 2.4814419746398926,
      "learning_rate": 2.1291681409671896e-06,
      "loss": 0.2839,
      "num_input_tokens_seen": 16203536,
      "step": 24600
    },
    {
      "epoch": 14.507665094339622,
      "grad_norm": 3.218010187149048,
      "learning_rate": 2.12706214061495e-06,
      "loss": 0.3152,
      "num_input_tokens_seen": 16206096,
      "step": 24605
    },
    {
      "epoch": 14.51061320754717,
      "grad_norm": 3.1929216384887695,
      "learning_rate": 2.124956900885391e-06,
      "loss": 0.2037,
      "num_input_tokens_seen": 16209520,
      "step": 24610
    },
    {
      "epoch": 14.513561320754716,
      "grad_norm": 4.169461727142334,
      "learning_rate": 2.1228524223358833e-06,
      "loss": 0.3195,
      "num_input_tokens_seen": 16213744,
      "step": 24615
    },
    {
      "epoch": 14.516509433962264,
      "grad_norm": 4.331352233886719,
      "learning_rate": 2.120748705523595e-06,
      "loss": 0.3604,
      "num_input_tokens_seen": 16217648,
      "step": 24620
    },
    {
      "epoch": 14.51945754716981,
      "grad_norm": 4.119107723236084,
      "learning_rate": 2.1186457510054976e-06,
      "loss": 0.3207,
      "num_input_tokens_seen": 16220592,
      "step": 24625
    },
    {
      "epoch": 14.522405660377359,
      "grad_norm": 4.031186103820801,
      "learning_rate": 2.116543559338355e-06,
      "loss": 0.3908,
      "num_input_tokens_seen": 16222832,
      "step": 24630
    },
    {
      "epoch": 14.525353773584905,
      "grad_norm": 2.5149168968200684,
      "learning_rate": 2.1144421310787305e-06,
      "loss": 0.2537,
      "num_input_tokens_seen": 16226800,
      "step": 24635
    },
    {
      "epoch": 14.528301886792454,
      "grad_norm": 4.278092861175537,
      "learning_rate": 2.11234146678299e-06,
      "loss": 0.3286,
      "num_input_tokens_seen": 16229424,
      "step": 24640
    },
    {
      "epoch": 14.53125,
      "grad_norm": 5.0662760734558105,
      "learning_rate": 2.1102415670072907e-06,
      "loss": 0.3161,
      "num_input_tokens_seen": 16233904,
      "step": 24645
    },
    {
      "epoch": 14.534198113207546,
      "grad_norm": 2.615816116333008,
      "learning_rate": 2.108142432307591e-06,
      "loss": 0.3441,
      "num_input_tokens_seen": 16237808,
      "step": 24650
    },
    {
      "epoch": 14.537146226415095,
      "grad_norm": 3.1167986392974854,
      "learning_rate": 2.1060440632396456e-06,
      "loss": 0.3469,
      "num_input_tokens_seen": 16241584,
      "step": 24655
    },
    {
      "epoch": 14.540094339622641,
      "grad_norm": 3.2220194339752197,
      "learning_rate": 2.103946460359007e-06,
      "loss": 0.3933,
      "num_input_tokens_seen": 16244432,
      "step": 24660
    },
    {
      "epoch": 14.54304245283019,
      "grad_norm": 3.06862211227417,
      "learning_rate": 2.101849624221022e-06,
      "loss": 0.2669,
      "num_input_tokens_seen": 16247344,
      "step": 24665
    },
    {
      "epoch": 14.545990566037736,
      "grad_norm": 3.794424295425415,
      "learning_rate": 2.0997535553808417e-06,
      "loss": 0.3155,
      "num_input_tokens_seen": 16250704,
      "step": 24670
    },
    {
      "epoch": 14.548938679245284,
      "grad_norm": 3.0622880458831787,
      "learning_rate": 2.0976582543934064e-06,
      "loss": 0.3706,
      "num_input_tokens_seen": 16253552,
      "step": 24675
    },
    {
      "epoch": 14.55188679245283,
      "grad_norm": 6.030378818511963,
      "learning_rate": 2.0955637218134573e-06,
      "loss": 0.384,
      "num_input_tokens_seen": 16256240,
      "step": 24680
    },
    {
      "epoch": 14.554834905660378,
      "grad_norm": 3.2519781589508057,
      "learning_rate": 2.09346995819553e-06,
      "loss": 0.3212,
      "num_input_tokens_seen": 16259760,
      "step": 24685
    },
    {
      "epoch": 14.557783018867925,
      "grad_norm": 6.006622314453125,
      "learning_rate": 2.0913769640939553e-06,
      "loss": 0.3434,
      "num_input_tokens_seen": 16265648,
      "step": 24690
    },
    {
      "epoch": 14.560731132075471,
      "grad_norm": 12.209541320800781,
      "learning_rate": 2.0892847400628674e-06,
      "loss": 0.4528,
      "num_input_tokens_seen": 16268144,
      "step": 24695
    },
    {
      "epoch": 14.56367924528302,
      "grad_norm": 3.378011703491211,
      "learning_rate": 2.0871932866561885e-06,
      "loss": 0.2354,
      "num_input_tokens_seen": 16271920,
      "step": 24700
    },
    {
      "epoch": 14.566627358490566,
      "grad_norm": 3.8420464992523193,
      "learning_rate": 2.0851026044276405e-06,
      "loss": 0.2196,
      "num_input_tokens_seen": 16275408,
      "step": 24705
    },
    {
      "epoch": 14.569575471698114,
      "grad_norm": 2.6815311908721924,
      "learning_rate": 2.083012693930741e-06,
      "loss": 0.2518,
      "num_input_tokens_seen": 16279376,
      "step": 24710
    },
    {
      "epoch": 14.57252358490566,
      "grad_norm": 4.135249614715576,
      "learning_rate": 2.0809235557188e-06,
      "loss": 0.4315,
      "num_input_tokens_seen": 16281776,
      "step": 24715
    },
    {
      "epoch": 14.575471698113208,
      "grad_norm": 3.9006848335266113,
      "learning_rate": 2.0788351903449307e-06,
      "loss": 0.4253,
      "num_input_tokens_seen": 16285264,
      "step": 24720
    },
    {
      "epoch": 14.578419811320755,
      "grad_norm": 2.861772060394287,
      "learning_rate": 2.0767475983620317e-06,
      "loss": 0.3887,
      "num_input_tokens_seen": 16289392,
      "step": 24725
    },
    {
      "epoch": 14.581367924528301,
      "grad_norm": 2.466074228286743,
      "learning_rate": 2.074660780322806e-06,
      "loss": 0.2601,
      "num_input_tokens_seen": 16292944,
      "step": 24730
    },
    {
      "epoch": 14.58431603773585,
      "grad_norm": 3.4227213859558105,
      "learning_rate": 2.0725747367797473e-06,
      "loss": 0.274,
      "num_input_tokens_seen": 16295984,
      "step": 24735
    },
    {
      "epoch": 14.587264150943396,
      "grad_norm": 2.081347703933716,
      "learning_rate": 2.070489468285143e-06,
      "loss": 0.301,
      "num_input_tokens_seen": 16298608,
      "step": 24740
    },
    {
      "epoch": 14.590212264150944,
      "grad_norm": 2.7359437942504883,
      "learning_rate": 2.068404975391077e-06,
      "loss": 0.2528,
      "num_input_tokens_seen": 16301712,
      "step": 24745
    },
    {
      "epoch": 14.59316037735849,
      "grad_norm": 3.7588119506835938,
      "learning_rate": 2.0663212586494293e-06,
      "loss": 0.4356,
      "num_input_tokens_seen": 16304272,
      "step": 24750
    },
    {
      "epoch": 14.596108490566039,
      "grad_norm": 3.614182472229004,
      "learning_rate": 2.064238318611869e-06,
      "loss": 0.3156,
      "num_input_tokens_seen": 16306800,
      "step": 24755
    },
    {
      "epoch": 14.599056603773585,
      "grad_norm": 4.32574987411499,
      "learning_rate": 2.0621561558298693e-06,
      "loss": 0.3129,
      "num_input_tokens_seen": 16310160,
      "step": 24760
    },
    {
      "epoch": 14.602004716981131,
      "grad_norm": 3.2861571311950684,
      "learning_rate": 2.0600747708546877e-06,
      "loss": 0.2342,
      "num_input_tokens_seen": 16312976,
      "step": 24765
    },
    {
      "epoch": 14.60495283018868,
      "grad_norm": 4.7337775230407715,
      "learning_rate": 2.0579941642373814e-06,
      "loss": 0.305,
      "num_input_tokens_seen": 16315984,
      "step": 24770
    },
    {
      "epoch": 14.607900943396226,
      "grad_norm": 2.9244415760040283,
      "learning_rate": 2.0559143365287993e-06,
      "loss": 0.2968,
      "num_input_tokens_seen": 16319216,
      "step": 24775
    },
    {
      "epoch": 14.610849056603774,
      "grad_norm": 3.3133490085601807,
      "learning_rate": 2.0538352882795846e-06,
      "loss": 0.4066,
      "num_input_tokens_seen": 16322256,
      "step": 24780
    },
    {
      "epoch": 14.61379716981132,
      "grad_norm": 4.905874252319336,
      "learning_rate": 2.051757020040173e-06,
      "loss": 0.2404,
      "num_input_tokens_seen": 16326512,
      "step": 24785
    },
    {
      "epoch": 14.616745283018869,
      "grad_norm": 4.198197841644287,
      "learning_rate": 2.0496795323607983e-06,
      "loss": 0.2921,
      "num_input_tokens_seen": 16330320,
      "step": 24790
    },
    {
      "epoch": 14.619693396226415,
      "grad_norm": 4.805810451507568,
      "learning_rate": 2.0476028257914825e-06,
      "loss": 0.3772,
      "num_input_tokens_seen": 16333488,
      "step": 24795
    },
    {
      "epoch": 14.622641509433961,
      "grad_norm": 6.784592151641846,
      "learning_rate": 2.0455269008820433e-06,
      "loss": 0.2256,
      "num_input_tokens_seen": 16336336,
      "step": 24800
    },
    {
      "epoch": 14.62558962264151,
      "grad_norm": 2.594897985458374,
      "learning_rate": 2.0434517581820893e-06,
      "loss": 0.3578,
      "num_input_tokens_seen": 16339760,
      "step": 24805
    },
    {
      "epoch": 14.628537735849056,
      "grad_norm": 5.944509029388428,
      "learning_rate": 2.041377398241025e-06,
      "loss": 0.2814,
      "num_input_tokens_seen": 16342032,
      "step": 24810
    },
    {
      "epoch": 14.631485849056604,
      "grad_norm": 3.9395346641540527,
      "learning_rate": 2.0393038216080433e-06,
      "loss": 0.3539,
      "num_input_tokens_seen": 16345200,
      "step": 24815
    },
    {
      "epoch": 14.63443396226415,
      "grad_norm": 3.1870226860046387,
      "learning_rate": 2.037231028832135e-06,
      "loss": 0.3397,
      "num_input_tokens_seen": 16348688,
      "step": 24820
    },
    {
      "epoch": 14.637382075471699,
      "grad_norm": 3.8634066581726074,
      "learning_rate": 2.0351590204620823e-06,
      "loss": 0.3364,
      "num_input_tokens_seen": 16351568,
      "step": 24825
    },
    {
      "epoch": 14.640330188679245,
      "grad_norm": 4.622304439544678,
      "learning_rate": 2.033087797046457e-06,
      "loss": 0.5132,
      "num_input_tokens_seen": 16354640,
      "step": 24830
    },
    {
      "epoch": 14.643278301886792,
      "grad_norm": 5.92756986618042,
      "learning_rate": 2.031017359133624e-06,
      "loss": 0.424,
      "num_input_tokens_seen": 16357872,
      "step": 24835
    },
    {
      "epoch": 14.64622641509434,
      "grad_norm": 3.390963554382324,
      "learning_rate": 2.0289477072717406e-06,
      "loss": 0.2375,
      "num_input_tokens_seen": 16363024,
      "step": 24840
    },
    {
      "epoch": 14.649174528301886,
      "grad_norm": 4.112465858459473,
      "learning_rate": 2.026878842008756e-06,
      "loss": 0.2398,
      "num_input_tokens_seen": 16365712,
      "step": 24845
    },
    {
      "epoch": 14.652122641509434,
      "grad_norm": 5.518312454223633,
      "learning_rate": 2.0248107638924105e-06,
      "loss": 0.2127,
      "num_input_tokens_seen": 16368528,
      "step": 24850
    },
    {
      "epoch": 14.65507075471698,
      "grad_norm": 5.552243232727051,
      "learning_rate": 2.0227434734702386e-06,
      "loss": 0.3572,
      "num_input_tokens_seen": 16371856,
      "step": 24855
    },
    {
      "epoch": 14.658018867924529,
      "grad_norm": 3.6714093685150146,
      "learning_rate": 2.020676971289563e-06,
      "loss": 0.3253,
      "num_input_tokens_seen": 16374192,
      "step": 24860
    },
    {
      "epoch": 14.660966981132075,
      "grad_norm": 2.8466460704803467,
      "learning_rate": 2.0186112578975005e-06,
      "loss": 0.3795,
      "num_input_tokens_seen": 16378160,
      "step": 24865
    },
    {
      "epoch": 14.663915094339622,
      "grad_norm": 4.940086841583252,
      "learning_rate": 2.016546333840956e-06,
      "loss": 0.2658,
      "num_input_tokens_seen": 16381808,
      "step": 24870
    },
    {
      "epoch": 14.66686320754717,
      "grad_norm": 3.7032687664031982,
      "learning_rate": 2.014482199666627e-06,
      "loss": 0.3011,
      "num_input_tokens_seen": 16385136,
      "step": 24875
    },
    {
      "epoch": 14.669811320754716,
      "grad_norm": 5.309580326080322,
      "learning_rate": 2.0124188559210017e-06,
      "loss": 0.4582,
      "num_input_tokens_seen": 16388176,
      "step": 24880
    },
    {
      "epoch": 14.672759433962264,
      "grad_norm": 4.7323994636535645,
      "learning_rate": 2.0103563031503613e-06,
      "loss": 0.3789,
      "num_input_tokens_seen": 16390864,
      "step": 24885
    },
    {
      "epoch": 14.67570754716981,
      "grad_norm": 3.843297243118286,
      "learning_rate": 2.0082945419007745e-06,
      "loss": 0.2624,
      "num_input_tokens_seen": 16394800,
      "step": 24890
    },
    {
      "epoch": 14.678655660377359,
      "grad_norm": 2.9527337551116943,
      "learning_rate": 2.0062335727181007e-06,
      "loss": 0.3858,
      "num_input_tokens_seen": 16398672,
      "step": 24895
    },
    {
      "epoch": 14.681603773584905,
      "grad_norm": 3.3890774250030518,
      "learning_rate": 2.004173396147992e-06,
      "loss": 0.2995,
      "num_input_tokens_seen": 16401776,
      "step": 24900
    },
    {
      "epoch": 14.684551886792454,
      "grad_norm": 5.482773780822754,
      "learning_rate": 2.0021140127358873e-06,
      "loss": 0.4345,
      "num_input_tokens_seen": 16404688,
      "step": 24905
    },
    {
      "epoch": 14.6875,
      "grad_norm": 4.808653831481934,
      "learning_rate": 2.0000554230270164e-06,
      "loss": 0.3103,
      "num_input_tokens_seen": 16407376,
      "step": 24910
    },
    {
      "epoch": 14.690448113207546,
      "grad_norm": 2.866006374359131,
      "learning_rate": 1.997997627566401e-06,
      "loss": 0.3559,
      "num_input_tokens_seen": 16411408,
      "step": 24915
    },
    {
      "epoch": 14.693396226415095,
      "grad_norm": 3.9763541221618652,
      "learning_rate": 1.9959406268988536e-06,
      "loss": 0.3286,
      "num_input_tokens_seen": 16414288,
      "step": 24920
    },
    {
      "epoch": 14.696344339622641,
      "grad_norm": 4.519501209259033,
      "learning_rate": 1.9938844215689717e-06,
      "loss": 0.4805,
      "num_input_tokens_seen": 16417616,
      "step": 24925
    },
    {
      "epoch": 14.69929245283019,
      "grad_norm": 3.3119068145751953,
      "learning_rate": 1.991829012121145e-06,
      "loss": 0.2547,
      "num_input_tokens_seen": 16422928,
      "step": 24930
    },
    {
      "epoch": 14.702240566037736,
      "grad_norm": 2.3337574005126953,
      "learning_rate": 1.989774399099552e-06,
      "loss": 0.295,
      "num_input_tokens_seen": 16426160,
      "step": 24935
    },
    {
      "epoch": 14.705188679245284,
      "grad_norm": 2.4422965049743652,
      "learning_rate": 1.98772058304816e-06,
      "loss": 0.2932,
      "num_input_tokens_seen": 16429104,
      "step": 24940
    },
    {
      "epoch": 14.70813679245283,
      "grad_norm": 3.090442180633545,
      "learning_rate": 1.9856675645107244e-06,
      "loss": 0.2979,
      "num_input_tokens_seen": 16432240,
      "step": 24945
    },
    {
      "epoch": 14.711084905660378,
      "grad_norm": 5.099664211273193,
      "learning_rate": 1.9836153440307936e-06,
      "loss": 0.3373,
      "num_input_tokens_seen": 16436016,
      "step": 24950
    },
    {
      "epoch": 14.714033018867925,
      "grad_norm": 3.402369737625122,
      "learning_rate": 1.9815639221517002e-06,
      "loss": 0.3179,
      "num_input_tokens_seen": 16439440,
      "step": 24955
    },
    {
      "epoch": 14.716981132075471,
      "grad_norm": 2.7976346015930176,
      "learning_rate": 1.9795132994165673e-06,
      "loss": 0.2812,
      "num_input_tokens_seen": 16442544,
      "step": 24960
    },
    {
      "epoch": 14.71992924528302,
      "grad_norm": 5.117912769317627,
      "learning_rate": 1.977463476368306e-06,
      "loss": 0.335,
      "num_input_tokens_seen": 16445296,
      "step": 24965
    },
    {
      "epoch": 14.722877358490566,
      "grad_norm": 1.5856884717941284,
      "learning_rate": 1.975414453549614e-06,
      "loss": 0.2729,
      "num_input_tokens_seen": 16449136,
      "step": 24970
    },
    {
      "epoch": 14.725825471698114,
      "grad_norm": 3.5651869773864746,
      "learning_rate": 1.9733662315029826e-06,
      "loss": 0.3417,
      "num_input_tokens_seen": 16452368,
      "step": 24975
    },
    {
      "epoch": 14.72877358490566,
      "grad_norm": 2.925691843032837,
      "learning_rate": 1.9713188107706856e-06,
      "loss": 0.3835,
      "num_input_tokens_seen": 16455952,
      "step": 24980
    },
    {
      "epoch": 14.731721698113208,
      "grad_norm": 5.0564446449279785,
      "learning_rate": 1.969272191894786e-06,
      "loss": 0.3134,
      "num_input_tokens_seen": 16459280,
      "step": 24985
    },
    {
      "epoch": 14.734669811320755,
      "grad_norm": 2.8612685203552246,
      "learning_rate": 1.967226375417135e-06,
      "loss": 0.3375,
      "num_input_tokens_seen": 16462288,
      "step": 24990
    },
    {
      "epoch": 14.737617924528301,
      "grad_norm": 5.829468727111816,
      "learning_rate": 1.965181361879372e-06,
      "loss": 0.5495,
      "num_input_tokens_seen": 16465040,
      "step": 24995
    },
    {
      "epoch": 14.74056603773585,
      "grad_norm": 8.896955490112305,
      "learning_rate": 1.9631371518229214e-06,
      "loss": 0.2847,
      "num_input_tokens_seen": 16468784,
      "step": 25000
    },
    {
      "epoch": 14.743514150943396,
      "grad_norm": 4.354038715362549,
      "learning_rate": 1.9610937457889975e-06,
      "loss": 0.4188,
      "num_input_tokens_seen": 16472752,
      "step": 25005
    },
    {
      "epoch": 14.746462264150944,
      "grad_norm": 8.365440368652344,
      "learning_rate": 1.9590511443186032e-06,
      "loss": 0.3339,
      "num_input_tokens_seen": 16476496,
      "step": 25010
    },
    {
      "epoch": 14.74941037735849,
      "grad_norm": 3.569338798522949,
      "learning_rate": 1.9570093479525243e-06,
      "loss": 0.4223,
      "num_input_tokens_seen": 16480560,
      "step": 25015
    },
    {
      "epoch": 14.752358490566039,
      "grad_norm": 4.173075199127197,
      "learning_rate": 1.954968357231335e-06,
      "loss": 0.3583,
      "num_input_tokens_seen": 16485136,
      "step": 25020
    },
    {
      "epoch": 14.755306603773585,
      "grad_norm": 3.5709125995635986,
      "learning_rate": 1.9529281726953964e-06,
      "loss": 0.3738,
      "num_input_tokens_seen": 16487952,
      "step": 25025
    },
    {
      "epoch": 14.758254716981131,
      "grad_norm": 2.663255214691162,
      "learning_rate": 1.9508887948848564e-06,
      "loss": 0.4055,
      "num_input_tokens_seen": 16491184,
      "step": 25030
    },
    {
      "epoch": 14.76120283018868,
      "grad_norm": 2.432349920272827,
      "learning_rate": 1.9488502243396475e-06,
      "loss": 0.4201,
      "num_input_tokens_seen": 16494672,
      "step": 25035
    },
    {
      "epoch": 14.764150943396226,
      "grad_norm": 1.709970474243164,
      "learning_rate": 1.946812461599492e-06,
      "loss": 0.2973,
      "num_input_tokens_seen": 16498928,
      "step": 25040
    },
    {
      "epoch": 14.767099056603774,
      "grad_norm": 4.615368366241455,
      "learning_rate": 1.944775507203897e-06,
      "loss": 0.31,
      "num_input_tokens_seen": 16501808,
      "step": 25045
    },
    {
      "epoch": 14.77004716981132,
      "grad_norm": 4.636023998260498,
      "learning_rate": 1.942739361692153e-06,
      "loss": 0.2633,
      "num_input_tokens_seen": 16505296,
      "step": 25050
    },
    {
      "epoch": 14.772995283018869,
      "grad_norm": 4.0775146484375,
      "learning_rate": 1.94070402560334e-06,
      "loss": 0.2471,
      "num_input_tokens_seen": 16508144,
      "step": 25055
    },
    {
      "epoch": 14.775943396226415,
      "grad_norm": 5.541779518127441,
      "learning_rate": 1.93866949947632e-06,
      "loss": 0.4129,
      "num_input_tokens_seen": 16510384,
      "step": 25060
    },
    {
      "epoch": 14.778891509433961,
      "grad_norm": 2.868513584136963,
      "learning_rate": 1.9366357838497423e-06,
      "loss": 0.2313,
      "num_input_tokens_seen": 16513968,
      "step": 25065
    },
    {
      "epoch": 14.78183962264151,
      "grad_norm": 4.9946136474609375,
      "learning_rate": 1.9346028792620454e-06,
      "loss": 0.3656,
      "num_input_tokens_seen": 16518128,
      "step": 25070
    },
    {
      "epoch": 14.784787735849056,
      "grad_norm": 4.270539283752441,
      "learning_rate": 1.9325707862514464e-06,
      "loss": 0.3232,
      "num_input_tokens_seen": 16521616,
      "step": 25075
    },
    {
      "epoch": 14.787735849056604,
      "grad_norm": 3.773979425430298,
      "learning_rate": 1.930539505355952e-06,
      "loss": 0.3344,
      "num_input_tokens_seen": 16525488,
      "step": 25080
    },
    {
      "epoch": 14.79068396226415,
      "grad_norm": 3.597813129425049,
      "learning_rate": 1.9285090371133524e-06,
      "loss": 0.3564,
      "num_input_tokens_seen": 16529072,
      "step": 25085
    },
    {
      "epoch": 14.793632075471699,
      "grad_norm": 3.757230043411255,
      "learning_rate": 1.9264793820612228e-06,
      "loss": 0.3292,
      "num_input_tokens_seen": 16531440,
      "step": 25090
    },
    {
      "epoch": 14.796580188679245,
      "grad_norm": 7.317485332489014,
      "learning_rate": 1.924450540736921e-06,
      "loss": 0.3884,
      "num_input_tokens_seen": 16533904,
      "step": 25095
    },
    {
      "epoch": 14.799528301886792,
      "grad_norm": 6.232478141784668,
      "learning_rate": 1.922422513677593e-06,
      "loss": 0.3897,
      "num_input_tokens_seen": 16536880,
      "step": 25100
    },
    {
      "epoch": 14.80247641509434,
      "grad_norm": 2.185513734817505,
      "learning_rate": 1.9203953014201703e-06,
      "loss": 0.3692,
      "num_input_tokens_seen": 16541424,
      "step": 25105
    },
    {
      "epoch": 14.805424528301886,
      "grad_norm": 3.519742012023926,
      "learning_rate": 1.918368904501364e-06,
      "loss": 0.2784,
      "num_input_tokens_seen": 16544592,
      "step": 25110
    },
    {
      "epoch": 14.808372641509434,
      "grad_norm": 4.64440393447876,
      "learning_rate": 1.9163433234576713e-06,
      "loss": 0.3215,
      "num_input_tokens_seen": 16548208,
      "step": 25115
    },
    {
      "epoch": 14.81132075471698,
      "grad_norm": 2.303187370300293,
      "learning_rate": 1.9143185588253733e-06,
      "loss": 0.2828,
      "num_input_tokens_seen": 16551632,
      "step": 25120
    },
    {
      "epoch": 14.814268867924529,
      "grad_norm": 2.5964179039001465,
      "learning_rate": 1.9122946111405354e-06,
      "loss": 0.3046,
      "num_input_tokens_seen": 16555824,
      "step": 25125
    },
    {
      "epoch": 14.817216981132075,
      "grad_norm": 5.444244384765625,
      "learning_rate": 1.910271480939005e-06,
      "loss": 0.2306,
      "num_input_tokens_seen": 16558512,
      "step": 25130
    },
    {
      "epoch": 14.820165094339622,
      "grad_norm": 4.2263312339782715,
      "learning_rate": 1.9082491687564176e-06,
      "loss": 0.3766,
      "num_input_tokens_seen": 16560784,
      "step": 25135
    },
    {
      "epoch": 14.82311320754717,
      "grad_norm": 4.615510940551758,
      "learning_rate": 1.9062276751281872e-06,
      "loss": 0.3122,
      "num_input_tokens_seen": 16564304,
      "step": 25140
    },
    {
      "epoch": 14.826061320754716,
      "grad_norm": 5.162353515625,
      "learning_rate": 1.9042070005895136e-06,
      "loss": 0.2729,
      "num_input_tokens_seen": 16566896,
      "step": 25145
    },
    {
      "epoch": 14.829009433962264,
      "grad_norm": 3.128815174102783,
      "learning_rate": 1.9021871456753788e-06,
      "loss": 0.4611,
      "num_input_tokens_seen": 16570960,
      "step": 25150
    },
    {
      "epoch": 14.83195754716981,
      "grad_norm": 4.614686965942383,
      "learning_rate": 1.9001681109205478e-06,
      "loss": 0.4518,
      "num_input_tokens_seen": 16573328,
      "step": 25155
    },
    {
      "epoch": 14.834905660377359,
      "grad_norm": 1.9690887928009033,
      "learning_rate": 1.898149896859567e-06,
      "loss": 0.297,
      "num_input_tokens_seen": 16576208,
      "step": 25160
    },
    {
      "epoch": 14.837853773584905,
      "grad_norm": 3.5053470134735107,
      "learning_rate": 1.8961325040267714e-06,
      "loss": 0.2816,
      "num_input_tokens_seen": 16579184,
      "step": 25165
    },
    {
      "epoch": 14.840801886792454,
      "grad_norm": 3.165558099746704,
      "learning_rate": 1.894115932956272e-06,
      "loss": 0.2444,
      "num_input_tokens_seen": 16585456,
      "step": 25170
    },
    {
      "epoch": 14.84375,
      "grad_norm": 5.427536964416504,
      "learning_rate": 1.8921001841819652e-06,
      "loss": 0.2764,
      "num_input_tokens_seen": 16589104,
      "step": 25175
    },
    {
      "epoch": 14.846698113207546,
      "grad_norm": 2.8021867275238037,
      "learning_rate": 1.8900852582375284e-06,
      "loss": 0.2748,
      "num_input_tokens_seen": 16591824,
      "step": 25180
    },
    {
      "epoch": 14.849646226415095,
      "grad_norm": 3.7297439575195312,
      "learning_rate": 1.8880711556564214e-06,
      "loss": 0.2306,
      "num_input_tokens_seen": 16594928,
      "step": 25185
    },
    {
      "epoch": 14.852594339622641,
      "grad_norm": 3.8010432720184326,
      "learning_rate": 1.8860578769718891e-06,
      "loss": 0.3496,
      "num_input_tokens_seen": 16597360,
      "step": 25190
    },
    {
      "epoch": 14.85554245283019,
      "grad_norm": 2.9346988201141357,
      "learning_rate": 1.8840454227169525e-06,
      "loss": 0.3245,
      "num_input_tokens_seen": 16600304,
      "step": 25195
    },
    {
      "epoch": 14.858490566037736,
      "grad_norm": 4.050928592681885,
      "learning_rate": 1.882033793424421e-06,
      "loss": 0.3751,
      "num_input_tokens_seen": 16603600,
      "step": 25200
    },
    {
      "epoch": 14.861438679245284,
      "grad_norm": 3.1907753944396973,
      "learning_rate": 1.88002298962688e-06,
      "loss": 0.2634,
      "num_input_tokens_seen": 16606608,
      "step": 25205
    },
    {
      "epoch": 14.86438679245283,
      "grad_norm": 2.7858662605285645,
      "learning_rate": 1.8780130118566996e-06,
      "loss": 0.2718,
      "num_input_tokens_seen": 16610800,
      "step": 25210
    },
    {
      "epoch": 14.867334905660378,
      "grad_norm": 3.440598964691162,
      "learning_rate": 1.876003860646029e-06,
      "loss": 0.307,
      "num_input_tokens_seen": 16614256,
      "step": 25215
    },
    {
      "epoch": 14.870283018867925,
      "grad_norm": 2.7228970527648926,
      "learning_rate": 1.8739955365267997e-06,
      "loss": 0.3207,
      "num_input_tokens_seen": 16617584,
      "step": 25220
    },
    {
      "epoch": 14.873231132075471,
      "grad_norm": 2.7097585201263428,
      "learning_rate": 1.8719880400307228e-06,
      "loss": 0.3272,
      "num_input_tokens_seen": 16621136,
      "step": 25225
    },
    {
      "epoch": 14.87617924528302,
      "grad_norm": 4.520415782928467,
      "learning_rate": 1.869981371689295e-06,
      "loss": 0.3853,
      "num_input_tokens_seen": 16624016,
      "step": 25230
    },
    {
      "epoch": 14.879127358490566,
      "grad_norm": 2.7989087104797363,
      "learning_rate": 1.867975532033789e-06,
      "loss": 0.3363,
      "num_input_tokens_seen": 16626704,
      "step": 25235
    },
    {
      "epoch": 14.882075471698114,
      "grad_norm": 2.982692241668701,
      "learning_rate": 1.8659705215952589e-06,
      "loss": 0.3152,
      "num_input_tokens_seen": 16629808,
      "step": 25240
    },
    {
      "epoch": 14.88502358490566,
      "grad_norm": 2.910404920578003,
      "learning_rate": 1.8639663409045405e-06,
      "loss": 0.2354,
      "num_input_tokens_seen": 16633296,
      "step": 25245
    },
    {
      "epoch": 14.887971698113208,
      "grad_norm": 2.9499402046203613,
      "learning_rate": 1.8619629904922466e-06,
      "loss": 0.5455,
      "num_input_tokens_seen": 16636240,
      "step": 25250
    },
    {
      "epoch": 14.890919811320755,
      "grad_norm": 6.518833637237549,
      "learning_rate": 1.859960470888777e-06,
      "loss": 0.2923,
      "num_input_tokens_seen": 16638928,
      "step": 25255
    },
    {
      "epoch": 14.893867924528301,
      "grad_norm": 3.675835132598877,
      "learning_rate": 1.857958782624306e-06,
      "loss": 0.2771,
      "num_input_tokens_seen": 16642448,
      "step": 25260
    },
    {
      "epoch": 14.89681603773585,
      "grad_norm": 3.448415517807007,
      "learning_rate": 1.8559579262287886e-06,
      "loss": 0.307,
      "num_input_tokens_seen": 16645616,
      "step": 25265
    },
    {
      "epoch": 14.899764150943396,
      "grad_norm": 3.6964330673217773,
      "learning_rate": 1.8539579022319599e-06,
      "loss": 0.3921,
      "num_input_tokens_seen": 16648880,
      "step": 25270
    },
    {
      "epoch": 14.902712264150944,
      "grad_norm": 3.248483657836914,
      "learning_rate": 1.8519587111633357e-06,
      "loss": 0.3236,
      "num_input_tokens_seen": 16651952,
      "step": 25275
    },
    {
      "epoch": 14.90566037735849,
      "grad_norm": 4.118393898010254,
      "learning_rate": 1.8499603535522082e-06,
      "loss": 0.3492,
      "num_input_tokens_seen": 16655312,
      "step": 25280
    },
    {
      "epoch": 14.908608490566039,
      "grad_norm": 3.7822751998901367,
      "learning_rate": 1.8479628299276543e-06,
      "loss": 0.2495,
      "num_input_tokens_seen": 16657552,
      "step": 25285
    },
    {
      "epoch": 14.911556603773585,
      "grad_norm": 4.050014972686768,
      "learning_rate": 1.8459661408185241e-06,
      "loss": 0.3478,
      "num_input_tokens_seen": 16660496,
      "step": 25290
    },
    {
      "epoch": 14.914504716981131,
      "grad_norm": 2.856618642807007,
      "learning_rate": 1.8439702867534536e-06,
      "loss": 0.5152,
      "num_input_tokens_seen": 16663504,
      "step": 25295
    },
    {
      "epoch": 14.91745283018868,
      "grad_norm": 4.287875175476074,
      "learning_rate": 1.841975268260851e-06,
      "loss": 0.3106,
      "num_input_tokens_seen": 16666000,
      "step": 25300
    },
    {
      "epoch": 14.920400943396226,
      "grad_norm": 3.5516889095306396,
      "learning_rate": 1.8399810858689066e-06,
      "loss": 0.3901,
      "num_input_tokens_seen": 16669584,
      "step": 25305
    },
    {
      "epoch": 14.923349056603774,
      "grad_norm": 2.945413589477539,
      "learning_rate": 1.8379877401055884e-06,
      "loss": 0.3902,
      "num_input_tokens_seen": 16673392,
      "step": 25310
    },
    {
      "epoch": 14.92629716981132,
      "grad_norm": 3.957106351852417,
      "learning_rate": 1.8359952314986418e-06,
      "loss": 0.2983,
      "num_input_tokens_seen": 16676592,
      "step": 25315
    },
    {
      "epoch": 14.929245283018869,
      "grad_norm": 10.18409538269043,
      "learning_rate": 1.8340035605755957e-06,
      "loss": 0.3539,
      "num_input_tokens_seen": 16679728,
      "step": 25320
    },
    {
      "epoch": 14.932193396226415,
      "grad_norm": 6.077990531921387,
      "learning_rate": 1.8320127278637518e-06,
      "loss": 0.3184,
      "num_input_tokens_seen": 16682896,
      "step": 25325
    },
    {
      "epoch": 14.935141509433961,
      "grad_norm": 4.562627792358398,
      "learning_rate": 1.830022733890191e-06,
      "loss": 0.3487,
      "num_input_tokens_seen": 16685456,
      "step": 25330
    },
    {
      "epoch": 14.93808962264151,
      "grad_norm": 4.362175464630127,
      "learning_rate": 1.8280335791817733e-06,
      "loss": 0.3706,
      "num_input_tokens_seen": 16688336,
      "step": 25335
    },
    {
      "epoch": 14.941037735849056,
      "grad_norm": 2.8425230979919434,
      "learning_rate": 1.826045264265136e-06,
      "loss": 0.5873,
      "num_input_tokens_seen": 16691248,
      "step": 25340
    },
    {
      "epoch": 14.943985849056604,
      "grad_norm": 2.407381057739258,
      "learning_rate": 1.8240577896666928e-06,
      "loss": 0.2915,
      "num_input_tokens_seen": 16694160,
      "step": 25345
    },
    {
      "epoch": 14.94693396226415,
      "grad_norm": 4.5882391929626465,
      "learning_rate": 1.8220711559126382e-06,
      "loss": 0.3426,
      "num_input_tokens_seen": 16697744,
      "step": 25350
    },
    {
      "epoch": 14.949882075471699,
      "grad_norm": 5.252303123474121,
      "learning_rate": 1.8200853635289417e-06,
      "loss": 0.303,
      "num_input_tokens_seen": 16700688,
      "step": 25355
    },
    {
      "epoch": 14.952830188679245,
      "grad_norm": 4.670369625091553,
      "learning_rate": 1.81810041304135e-06,
      "loss": 0.3338,
      "num_input_tokens_seen": 16702800,
      "step": 25360
    },
    {
      "epoch": 14.955778301886792,
      "grad_norm": 2.283191680908203,
      "learning_rate": 1.8161163049753865e-06,
      "loss": 0.3703,
      "num_input_tokens_seen": 16707056,
      "step": 25365
    },
    {
      "epoch": 14.95872641509434,
      "grad_norm": 2.3912887573242188,
      "learning_rate": 1.8141330398563533e-06,
      "loss": 0.2892,
      "num_input_tokens_seen": 16710416,
      "step": 25370
    },
    {
      "epoch": 14.961674528301886,
      "grad_norm": 2.7033655643463135,
      "learning_rate": 1.8121506182093268e-06,
      "loss": 0.2387,
      "num_input_tokens_seen": 16713552,
      "step": 25375
    },
    {
      "epoch": 14.964622641509434,
      "grad_norm": 5.875823020935059,
      "learning_rate": 1.8101690405591643e-06,
      "loss": 0.3566,
      "num_input_tokens_seen": 16716368,
      "step": 25380
    },
    {
      "epoch": 14.96757075471698,
      "grad_norm": 4.225627422332764,
      "learning_rate": 1.8081883074304945e-06,
      "loss": 0.2707,
      "num_input_tokens_seen": 16719536,
      "step": 25385
    },
    {
      "epoch": 14.970518867924529,
      "grad_norm": 3.0717480182647705,
      "learning_rate": 1.8062084193477275e-06,
      "loss": 0.4463,
      "num_input_tokens_seen": 16722736,
      "step": 25390
    },
    {
      "epoch": 14.973466981132075,
      "grad_norm": 2.8491227626800537,
      "learning_rate": 1.804229376835046e-06,
      "loss": 0.3108,
      "num_input_tokens_seen": 16725904,
      "step": 25395
    },
    {
      "epoch": 14.976415094339622,
      "grad_norm": 3.7304258346557617,
      "learning_rate": 1.8022511804164105e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 16728400,
      "step": 25400
    },
    {
      "epoch": 14.97936320754717,
      "grad_norm": 3.368119955062866,
      "learning_rate": 1.8002738306155559e-06,
      "loss": 0.3787,
      "num_input_tokens_seen": 16731568,
      "step": 25405
    },
    {
      "epoch": 14.982311320754716,
      "grad_norm": 3.8554372787475586,
      "learning_rate": 1.7982973279559935e-06,
      "loss": 0.3242,
      "num_input_tokens_seen": 16733872,
      "step": 25410
    },
    {
      "epoch": 14.985259433962264,
      "grad_norm": 7.260667324066162,
      "learning_rate": 1.7963216729610134e-06,
      "loss": 0.3405,
      "num_input_tokens_seen": 16737648,
      "step": 25415
    },
    {
      "epoch": 14.98820754716981,
      "grad_norm": 2.2868900299072266,
      "learning_rate": 1.7943468661536773e-06,
      "loss": 0.2376,
      "num_input_tokens_seen": 16740656,
      "step": 25420
    },
    {
      "epoch": 14.991155660377359,
      "grad_norm": 10.87716293334961,
      "learning_rate": 1.7923729080568242e-06,
      "loss": 0.4229,
      "num_input_tokens_seen": 16743280,
      "step": 25425
    },
    {
      "epoch": 14.994103773584905,
      "grad_norm": 4.239756107330322,
      "learning_rate": 1.7903997991930683e-06,
      "loss": 0.344,
      "num_input_tokens_seen": 16746384,
      "step": 25430
    },
    {
      "epoch": 14.997051886792454,
      "grad_norm": 3.87459659576416,
      "learning_rate": 1.7884275400847972e-06,
      "loss": 0.3275,
      "num_input_tokens_seen": 16749040,
      "step": 25435
    },
    {
      "epoch": 15.0,
      "grad_norm": 36.16889953613281,
      "learning_rate": 1.786456131254175e-06,
      "loss": 0.3927,
      "num_input_tokens_seen": 16751912,
      "step": 25440
    },
    {
      "epoch": 15.002948113207546,
      "grad_norm": 3.9149065017700195,
      "learning_rate": 1.784485573223143e-06,
      "loss": 0.3048,
      "num_input_tokens_seen": 16754600,
      "step": 25445
    },
    {
      "epoch": 15.005896226415095,
      "grad_norm": 4.309606075286865,
      "learning_rate": 1.782515866513414e-06,
      "loss": 0.3289,
      "num_input_tokens_seen": 16758024,
      "step": 25450
    },
    {
      "epoch": 15.008844339622641,
      "grad_norm": 4.3739118576049805,
      "learning_rate": 1.7805470116464758e-06,
      "loss": 0.4376,
      "num_input_tokens_seen": 16761160,
      "step": 25455
    },
    {
      "epoch": 15.01179245283019,
      "grad_norm": 6.3820600509643555,
      "learning_rate": 1.7785790091435911e-06,
      "loss": 0.3256,
      "num_input_tokens_seen": 16763784,
      "step": 25460
    },
    {
      "epoch": 15.014740566037736,
      "grad_norm": 3.553680658340454,
      "learning_rate": 1.776611859525796e-06,
      "loss": 0.249,
      "num_input_tokens_seen": 16767272,
      "step": 25465
    },
    {
      "epoch": 15.017688679245284,
      "grad_norm": 2.2173614501953125,
      "learning_rate": 1.7746455633139042e-06,
      "loss": 0.2414,
      "num_input_tokens_seen": 16770728,
      "step": 25470
    },
    {
      "epoch": 15.02063679245283,
      "grad_norm": 3.1242270469665527,
      "learning_rate": 1.7726801210285005e-06,
      "loss": 0.3562,
      "num_input_tokens_seen": 16773448,
      "step": 25475
    },
    {
      "epoch": 15.023584905660377,
      "grad_norm": 3.092578649520874,
      "learning_rate": 1.7707155331899418e-06,
      "loss": 0.3023,
      "num_input_tokens_seen": 16777320,
      "step": 25480
    },
    {
      "epoch": 15.026533018867925,
      "grad_norm": 3.3388478755950928,
      "learning_rate": 1.7687518003183645e-06,
      "loss": 0.3423,
      "num_input_tokens_seen": 16781608,
      "step": 25485
    },
    {
      "epoch": 15.029481132075471,
      "grad_norm": 6.992531776428223,
      "learning_rate": 1.766788922933675e-06,
      "loss": 0.2941,
      "num_input_tokens_seen": 16785000,
      "step": 25490
    },
    {
      "epoch": 15.03242924528302,
      "grad_norm": 4.1038384437561035,
      "learning_rate": 1.7648269015555514e-06,
      "loss": 0.5238,
      "num_input_tokens_seen": 16787720,
      "step": 25495
    },
    {
      "epoch": 15.035377358490566,
      "grad_norm": 3.50523042678833,
      "learning_rate": 1.7628657367034474e-06,
      "loss": 0.3335,
      "num_input_tokens_seen": 16790920,
      "step": 25500
    },
    {
      "epoch": 15.038325471698114,
      "grad_norm": 2.778996467590332,
      "learning_rate": 1.7609054288965922e-06,
      "loss": 0.256,
      "num_input_tokens_seen": 16795048,
      "step": 25505
    },
    {
      "epoch": 15.04127358490566,
      "grad_norm": 3.854374408721924,
      "learning_rate": 1.7589459786539847e-06,
      "loss": 0.4023,
      "num_input_tokens_seen": 16798344,
      "step": 25510
    },
    {
      "epoch": 15.044221698113208,
      "grad_norm": 2.645552396774292,
      "learning_rate": 1.7569873864943975e-06,
      "loss": 0.2624,
      "num_input_tokens_seen": 16800936,
      "step": 25515
    },
    {
      "epoch": 15.047169811320755,
      "grad_norm": 3.659799814224243,
      "learning_rate": 1.7550296529363764e-06,
      "loss": 0.3751,
      "num_input_tokens_seen": 16803848,
      "step": 25520
    },
    {
      "epoch": 15.050117924528301,
      "grad_norm": 4.540435314178467,
      "learning_rate": 1.7530727784982393e-06,
      "loss": 0.3708,
      "num_input_tokens_seen": 16807464,
      "step": 25525
    },
    {
      "epoch": 15.05306603773585,
      "grad_norm": 2.785172462463379,
      "learning_rate": 1.7511167636980765e-06,
      "loss": 0.3122,
      "num_input_tokens_seen": 16810632,
      "step": 25530
    },
    {
      "epoch": 15.056014150943396,
      "grad_norm": 3.3163857460021973,
      "learning_rate": 1.7491616090537539e-06,
      "loss": 0.3106,
      "num_input_tokens_seen": 16814792,
      "step": 25535
    },
    {
      "epoch": 15.058962264150944,
      "grad_norm": 4.399407863616943,
      "learning_rate": 1.7472073150829056e-06,
      "loss": 0.3038,
      "num_input_tokens_seen": 16817448,
      "step": 25540
    },
    {
      "epoch": 15.06191037735849,
      "grad_norm": 3.7950990200042725,
      "learning_rate": 1.745253882302939e-06,
      "loss": 0.3265,
      "num_input_tokens_seen": 16820040,
      "step": 25545
    },
    {
      "epoch": 15.064858490566039,
      "grad_norm": 4.250898838043213,
      "learning_rate": 1.743301311231035e-06,
      "loss": 0.3646,
      "num_input_tokens_seen": 16822312,
      "step": 25550
    },
    {
      "epoch": 15.067806603773585,
      "grad_norm": 2.4910881519317627,
      "learning_rate": 1.7413496023841437e-06,
      "loss": 0.4168,
      "num_input_tokens_seen": 16825928,
      "step": 25555
    },
    {
      "epoch": 15.070754716981131,
      "grad_norm": 2.7256107330322266,
      "learning_rate": 1.7393987562789876e-06,
      "loss": 0.3144,
      "num_input_tokens_seen": 16829416,
      "step": 25560
    },
    {
      "epoch": 15.07370283018868,
      "grad_norm": 2.2343404293060303,
      "learning_rate": 1.7374487734320655e-06,
      "loss": 0.2958,
      "num_input_tokens_seen": 16833256,
      "step": 25565
    },
    {
      "epoch": 15.076650943396226,
      "grad_norm": 3.00880765914917,
      "learning_rate": 1.7354996543596408e-06,
      "loss": 0.3411,
      "num_input_tokens_seen": 16836264,
      "step": 25570
    },
    {
      "epoch": 15.079599056603774,
      "grad_norm": 3.8639817237854004,
      "learning_rate": 1.7335513995777504e-06,
      "loss": 0.3851,
      "num_input_tokens_seen": 16839656,
      "step": 25575
    },
    {
      "epoch": 15.08254716981132,
      "grad_norm": 2.8675286769866943,
      "learning_rate": 1.7316040096022062e-06,
      "loss": 0.2808,
      "num_input_tokens_seen": 16842824,
      "step": 25580
    },
    {
      "epoch": 15.085495283018869,
      "grad_norm": 9.220807075500488,
      "learning_rate": 1.7296574849485863e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 16846088,
      "step": 25585
    },
    {
      "epoch": 15.088443396226415,
      "grad_norm": 3.7075583934783936,
      "learning_rate": 1.7277118261322423e-06,
      "loss": 0.4453,
      "num_input_tokens_seen": 16849896,
      "step": 25590
    },
    {
      "epoch": 15.091391509433961,
      "grad_norm": 7.526834487915039,
      "learning_rate": 1.7257670336682925e-06,
      "loss": 0.3374,
      "num_input_tokens_seen": 16852488,
      "step": 25595
    },
    {
      "epoch": 15.09433962264151,
      "grad_norm": 4.620419502258301,
      "learning_rate": 1.7238231080716339e-06,
      "loss": 0.2529,
      "num_input_tokens_seen": 16855400,
      "step": 25600
    },
    {
      "epoch": 15.097287735849056,
      "grad_norm": 3.780304431915283,
      "learning_rate": 1.721880049856927e-06,
      "loss": 0.3211,
      "num_input_tokens_seen": 16858536,
      "step": 25605
    },
    {
      "epoch": 15.100235849056604,
      "grad_norm": 9.810796737670898,
      "learning_rate": 1.7199378595386046e-06,
      "loss": 0.2696,
      "num_input_tokens_seen": 16861928,
      "step": 25610
    },
    {
      "epoch": 15.10318396226415,
      "grad_norm": 3.3232553005218506,
      "learning_rate": 1.7179965376308705e-06,
      "loss": 0.2418,
      "num_input_tokens_seen": 16865288,
      "step": 25615
    },
    {
      "epoch": 15.106132075471699,
      "grad_norm": 4.1386237144470215,
      "learning_rate": 1.7160560846476976e-06,
      "loss": 0.3228,
      "num_input_tokens_seen": 16868456,
      "step": 25620
    },
    {
      "epoch": 15.109080188679245,
      "grad_norm": 2.0626165866851807,
      "learning_rate": 1.7141165011028277e-06,
      "loss": 0.2527,
      "num_input_tokens_seen": 16872008,
      "step": 25625
    },
    {
      "epoch": 15.112028301886792,
      "grad_norm": 2.806382656097412,
      "learning_rate": 1.7121777875097767e-06,
      "loss": 0.3295,
      "num_input_tokens_seen": 16876136,
      "step": 25630
    },
    {
      "epoch": 15.11497641509434,
      "grad_norm": 4.6839728355407715,
      "learning_rate": 1.7102399443818268e-06,
      "loss": 0.269,
      "num_input_tokens_seen": 16878632,
      "step": 25635
    },
    {
      "epoch": 15.117924528301886,
      "grad_norm": 4.213815212249756,
      "learning_rate": 1.7083029722320294e-06,
      "loss": 0.3123,
      "num_input_tokens_seen": 16882216,
      "step": 25640
    },
    {
      "epoch": 15.120872641509434,
      "grad_norm": 3.3275089263916016,
      "learning_rate": 1.7063668715732063e-06,
      "loss": 0.2988,
      "num_input_tokens_seen": 16885288,
      "step": 25645
    },
    {
      "epoch": 15.12382075471698,
      "grad_norm": 5.396313667297363,
      "learning_rate": 1.7044316429179492e-06,
      "loss": 0.4179,
      "num_input_tokens_seen": 16887752,
      "step": 25650
    },
    {
      "epoch": 15.126768867924529,
      "grad_norm": 5.496801853179932,
      "learning_rate": 1.7024972867786155e-06,
      "loss": 0.2887,
      "num_input_tokens_seen": 16890856,
      "step": 25655
    },
    {
      "epoch": 15.129716981132075,
      "grad_norm": 2.286466360092163,
      "learning_rate": 1.7005638036673389e-06,
      "loss": 0.3167,
      "num_input_tokens_seen": 16895336,
      "step": 25660
    },
    {
      "epoch": 15.132665094339623,
      "grad_norm": 4.053893089294434,
      "learning_rate": 1.6986311940960148e-06,
      "loss": 0.2267,
      "num_input_tokens_seen": 16898312,
      "step": 25665
    },
    {
      "epoch": 15.13561320754717,
      "grad_norm": 4.303810119628906,
      "learning_rate": 1.696699458576308e-06,
      "loss": 0.3115,
      "num_input_tokens_seen": 16901128,
      "step": 25670
    },
    {
      "epoch": 15.138561320754716,
      "grad_norm": 3.048647165298462,
      "learning_rate": 1.6947685976196581e-06,
      "loss": 0.4206,
      "num_input_tokens_seen": 16903592,
      "step": 25675
    },
    {
      "epoch": 15.141509433962264,
      "grad_norm": 3.3775672912597656,
      "learning_rate": 1.692838611737267e-06,
      "loss": 0.2522,
      "num_input_tokens_seen": 16907208,
      "step": 25680
    },
    {
      "epoch": 15.14445754716981,
      "grad_norm": 6.1657233238220215,
      "learning_rate": 1.690909501440106e-06,
      "loss": 0.45,
      "num_input_tokens_seen": 16910024,
      "step": 25685
    },
    {
      "epoch": 15.147405660377359,
      "grad_norm": 3.0086843967437744,
      "learning_rate": 1.688981267238915e-06,
      "loss": 0.3583,
      "num_input_tokens_seen": 16912360,
      "step": 25690
    },
    {
      "epoch": 15.150353773584905,
      "grad_norm": 2.229591131210327,
      "learning_rate": 1.687053909644204e-06,
      "loss": 0.4727,
      "num_input_tokens_seen": 16916648,
      "step": 25695
    },
    {
      "epoch": 15.153301886792454,
      "grad_norm": 4.750855445861816,
      "learning_rate": 1.685127429166249e-06,
      "loss": 0.2969,
      "num_input_tokens_seen": 16919272,
      "step": 25700
    },
    {
      "epoch": 15.15625,
      "grad_norm": 3.6766469478607178,
      "learning_rate": 1.683201826315093e-06,
      "loss": 0.251,
      "num_input_tokens_seen": 16922600,
      "step": 25705
    },
    {
      "epoch": 15.159198113207546,
      "grad_norm": 3.9287946224212646,
      "learning_rate": 1.681277101600548e-06,
      "loss": 0.278,
      "num_input_tokens_seen": 16925928,
      "step": 25710
    },
    {
      "epoch": 15.162146226415095,
      "grad_norm": 2.788708448410034,
      "learning_rate": 1.6793532555321939e-06,
      "loss": 0.3313,
      "num_input_tokens_seen": 16929576,
      "step": 25715
    },
    {
      "epoch": 15.165094339622641,
      "grad_norm": 5.2446370124816895,
      "learning_rate": 1.6774302886193744e-06,
      "loss": 0.2601,
      "num_input_tokens_seen": 16933064,
      "step": 25720
    },
    {
      "epoch": 15.16804245283019,
      "grad_norm": 4.047692775726318,
      "learning_rate": 1.6755082013712076e-06,
      "loss": 0.2837,
      "num_input_tokens_seen": 16935784,
      "step": 25725
    },
    {
      "epoch": 15.170990566037736,
      "grad_norm": 4.766485214233398,
      "learning_rate": 1.6735869942965716e-06,
      "loss": 0.3122,
      "num_input_tokens_seen": 16938344,
      "step": 25730
    },
    {
      "epoch": 15.173938679245284,
      "grad_norm": 2.8420543670654297,
      "learning_rate": 1.6716666679041155e-06,
      "loss": 0.3814,
      "num_input_tokens_seen": 16941704,
      "step": 25735
    },
    {
      "epoch": 15.17688679245283,
      "grad_norm": 3.6234500408172607,
      "learning_rate": 1.6697472227022533e-06,
      "loss": 0.299,
      "num_input_tokens_seen": 16945320,
      "step": 25740
    },
    {
      "epoch": 15.179834905660377,
      "grad_norm": 3.8771653175354004,
      "learning_rate": 1.6678286591991644e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 16947528,
      "step": 25745
    },
    {
      "epoch": 15.182783018867925,
      "grad_norm": 6.672211170196533,
      "learning_rate": 1.665910977902801e-06,
      "loss": 0.3161,
      "num_input_tokens_seen": 16950344,
      "step": 25750
    },
    {
      "epoch": 15.185731132075471,
      "grad_norm": 3.972094774246216,
      "learning_rate": 1.6639941793208747e-06,
      "loss": 0.27,
      "num_input_tokens_seen": 16954088,
      "step": 25755
    },
    {
      "epoch": 15.18867924528302,
      "grad_norm": 7.512986183166504,
      "learning_rate": 1.6620782639608674e-06,
      "loss": 0.278,
      "num_input_tokens_seen": 16957192,
      "step": 25760
    },
    {
      "epoch": 15.191627358490566,
      "grad_norm": 2.645540714263916,
      "learning_rate": 1.6601632323300231e-06,
      "loss": 0.3524,
      "num_input_tokens_seen": 16961704,
      "step": 25765
    },
    {
      "epoch": 15.194575471698114,
      "grad_norm": 4.0152716636657715,
      "learning_rate": 1.6582490849353595e-06,
      "loss": 0.3723,
      "num_input_tokens_seen": 16964648,
      "step": 25770
    },
    {
      "epoch": 15.19752358490566,
      "grad_norm": 3.321730136871338,
      "learning_rate": 1.6563358222836523e-06,
      "loss": 0.3161,
      "num_input_tokens_seen": 16967720,
      "step": 25775
    },
    {
      "epoch": 15.200471698113208,
      "grad_norm": 4.080368518829346,
      "learning_rate": 1.654423444881445e-06,
      "loss": 0.3366,
      "num_input_tokens_seen": 16970600,
      "step": 25780
    },
    {
      "epoch": 15.203419811320755,
      "grad_norm": 6.4055938720703125,
      "learning_rate": 1.652511953235051e-06,
      "loss": 0.2729,
      "num_input_tokens_seen": 16974216,
      "step": 25785
    },
    {
      "epoch": 15.206367924528301,
      "grad_norm": 5.073522567749023,
      "learning_rate": 1.650601347850544e-06,
      "loss": 0.3119,
      "num_input_tokens_seen": 16978152,
      "step": 25790
    },
    {
      "epoch": 15.20931603773585,
      "grad_norm": 3.2803750038146973,
      "learning_rate": 1.6486916292337652e-06,
      "loss": 0.5124,
      "num_input_tokens_seen": 16981000,
      "step": 25795
    },
    {
      "epoch": 15.212264150943396,
      "grad_norm": 2.8930437564849854,
      "learning_rate": 1.6467827978903212e-06,
      "loss": 0.2796,
      "num_input_tokens_seen": 16983880,
      "step": 25800
    },
    {
      "epoch": 15.215212264150944,
      "grad_norm": 2.0512545108795166,
      "learning_rate": 1.6448748543255827e-06,
      "loss": 0.2777,
      "num_input_tokens_seen": 16986248,
      "step": 25805
    },
    {
      "epoch": 15.21816037735849,
      "grad_norm": 1.7940404415130615,
      "learning_rate": 1.6429677990446845e-06,
      "loss": 0.3486,
      "num_input_tokens_seen": 16990408,
      "step": 25810
    },
    {
      "epoch": 15.221108490566039,
      "grad_norm": 4.588990211486816,
      "learning_rate": 1.6410616325525319e-06,
      "loss": 0.3122,
      "num_input_tokens_seen": 16992872,
      "step": 25815
    },
    {
      "epoch": 15.224056603773585,
      "grad_norm": 2.580562114715576,
      "learning_rate": 1.6391563553537875e-06,
      "loss": 0.2662,
      "num_input_tokens_seen": 16995240,
      "step": 25820
    },
    {
      "epoch": 15.227004716981131,
      "grad_norm": 4.677535533905029,
      "learning_rate": 1.6372519679528832e-06,
      "loss": 0.3,
      "num_input_tokens_seen": 16997672,
      "step": 25825
    },
    {
      "epoch": 15.22995283018868,
      "grad_norm": 3.529975652694702,
      "learning_rate": 1.6353484708540124e-06,
      "loss": 0.3838,
      "num_input_tokens_seen": 17000776,
      "step": 25830
    },
    {
      "epoch": 15.232900943396226,
      "grad_norm": 4.612621784210205,
      "learning_rate": 1.633445864561135e-06,
      "loss": 0.2832,
      "num_input_tokens_seen": 17004488,
      "step": 25835
    },
    {
      "epoch": 15.235849056603774,
      "grad_norm": 2.439095973968506,
      "learning_rate": 1.6315441495779726e-06,
      "loss": 0.436,
      "num_input_tokens_seen": 17007624,
      "step": 25840
    },
    {
      "epoch": 15.23879716981132,
      "grad_norm": 2.1081342697143555,
      "learning_rate": 1.6296433264080152e-06,
      "loss": 0.294,
      "num_input_tokens_seen": 17010760,
      "step": 25845
    },
    {
      "epoch": 15.241745283018869,
      "grad_norm": 1.8995634317398071,
      "learning_rate": 1.627743395554513e-06,
      "loss": 0.4026,
      "num_input_tokens_seen": 17014120,
      "step": 25850
    },
    {
      "epoch": 15.244693396226415,
      "grad_norm": 2.016895294189453,
      "learning_rate": 1.6258443575204802e-06,
      "loss": 0.2662,
      "num_input_tokens_seen": 17017736,
      "step": 25855
    },
    {
      "epoch": 15.247641509433961,
      "grad_norm": 2.775164842605591,
      "learning_rate": 1.6239462128086936e-06,
      "loss": 0.3901,
      "num_input_tokens_seen": 17021384,
      "step": 25860
    },
    {
      "epoch": 15.25058962264151,
      "grad_norm": 2.688603639602661,
      "learning_rate": 1.6220489619216988e-06,
      "loss": 0.2683,
      "num_input_tokens_seen": 17025128,
      "step": 25865
    },
    {
      "epoch": 15.253537735849056,
      "grad_norm": 2.732625722885132,
      "learning_rate": 1.6201526053618e-06,
      "loss": 0.364,
      "num_input_tokens_seen": 17028296,
      "step": 25870
    },
    {
      "epoch": 15.256485849056604,
      "grad_norm": 3.8182291984558105,
      "learning_rate": 1.6182571436310634e-06,
      "loss": 0.2027,
      "num_input_tokens_seen": 17030504,
      "step": 25875
    },
    {
      "epoch": 15.25943396226415,
      "grad_norm": 2.6365573406219482,
      "learning_rate": 1.616362577231324e-06,
      "loss": 0.417,
      "num_input_tokens_seen": 17033448,
      "step": 25880
    },
    {
      "epoch": 15.262382075471699,
      "grad_norm": 4.69075870513916,
      "learning_rate": 1.614468906664175e-06,
      "loss": 0.2369,
      "num_input_tokens_seen": 17036808,
      "step": 25885
    },
    {
      "epoch": 15.265330188679245,
      "grad_norm": 3.2866363525390625,
      "learning_rate": 1.612576132430974e-06,
      "loss": 0.2515,
      "num_input_tokens_seen": 17039880,
      "step": 25890
    },
    {
      "epoch": 15.268278301886792,
      "grad_norm": 5.834691047668457,
      "learning_rate": 1.6106842550328406e-06,
      "loss": 0.2961,
      "num_input_tokens_seen": 17043048,
      "step": 25895
    },
    {
      "epoch": 15.27122641509434,
      "grad_norm": 5.453880786895752,
      "learning_rate": 1.6087932749706582e-06,
      "loss": 0.3847,
      "num_input_tokens_seen": 17045800,
      "step": 25900
    },
    {
      "epoch": 15.274174528301886,
      "grad_norm": 3.2745604515075684,
      "learning_rate": 1.6069031927450696e-06,
      "loss": 0.2803,
      "num_input_tokens_seen": 17049576,
      "step": 25905
    },
    {
      "epoch": 15.277122641509434,
      "grad_norm": 4.617825031280518,
      "learning_rate": 1.605014008856486e-06,
      "loss": 0.3702,
      "num_input_tokens_seen": 17052424,
      "step": 25910
    },
    {
      "epoch": 15.28007075471698,
      "grad_norm": 7.34128475189209,
      "learning_rate": 1.6031257238050745e-06,
      "loss": 0.3557,
      "num_input_tokens_seen": 17055272,
      "step": 25915
    },
    {
      "epoch": 15.283018867924529,
      "grad_norm": 2.979889154434204,
      "learning_rate": 1.601238338090768e-06,
      "loss": 0.2642,
      "num_input_tokens_seen": 17059080,
      "step": 25920
    },
    {
      "epoch": 15.285966981132075,
      "grad_norm": 3.274409770965576,
      "learning_rate": 1.5993518522132595e-06,
      "loss": 0.3431,
      "num_input_tokens_seen": 17061672,
      "step": 25925
    },
    {
      "epoch": 15.288915094339623,
      "grad_norm": 3.380492687225342,
      "learning_rate": 1.5974662666720037e-06,
      "loss": 0.2612,
      "num_input_tokens_seen": 17065032,
      "step": 25930
    },
    {
      "epoch": 15.29186320754717,
      "grad_norm": 4.199792385101318,
      "learning_rate": 1.5955815819662162e-06,
      "loss": 0.3122,
      "num_input_tokens_seen": 17068008,
      "step": 25935
    },
    {
      "epoch": 15.294811320754716,
      "grad_norm": 3.254772901535034,
      "learning_rate": 1.5936977985948788e-06,
      "loss": 0.3056,
      "num_input_tokens_seen": 17070984,
      "step": 25940
    },
    {
      "epoch": 15.297759433962264,
      "grad_norm": 9.852694511413574,
      "learning_rate": 1.5918149170567298e-06,
      "loss": 0.2519,
      "num_input_tokens_seen": 17073256,
      "step": 25945
    },
    {
      "epoch": 15.30070754716981,
      "grad_norm": 6.64829683303833,
      "learning_rate": 1.5899329378502698e-06,
      "loss": 0.3043,
      "num_input_tokens_seen": 17076712,
      "step": 25950
    },
    {
      "epoch": 15.303655660377359,
      "grad_norm": 3.8638787269592285,
      "learning_rate": 1.588051861473761e-06,
      "loss": 0.3923,
      "num_input_tokens_seen": 17079400,
      "step": 25955
    },
    {
      "epoch": 15.306603773584905,
      "grad_norm": 3.006399393081665,
      "learning_rate": 1.5861716884252253e-06,
      "loss": 0.2076,
      "num_input_tokens_seen": 17084936,
      "step": 25960
    },
    {
      "epoch": 15.309551886792454,
      "grad_norm": 3.8750052452087402,
      "learning_rate": 1.5842924192024489e-06,
      "loss": 0.3876,
      "num_input_tokens_seen": 17088168,
      "step": 25965
    },
    {
      "epoch": 15.3125,
      "grad_norm": 3.9259297847747803,
      "learning_rate": 1.5824140543029742e-06,
      "loss": 0.3964,
      "num_input_tokens_seen": 17090952,
      "step": 25970
    },
    {
      "epoch": 15.315448113207546,
      "grad_norm": 5.009559154510498,
      "learning_rate": 1.5805365942241092e-06,
      "loss": 0.3467,
      "num_input_tokens_seen": 17093288,
      "step": 25975
    },
    {
      "epoch": 15.318396226415095,
      "grad_norm": 5.040842533111572,
      "learning_rate": 1.5786600394629181e-06,
      "loss": 0.2501,
      "num_input_tokens_seen": 17096648,
      "step": 25980
    },
    {
      "epoch": 15.321344339622641,
      "grad_norm": 3.2107787132263184,
      "learning_rate": 1.5767843905162261e-06,
      "loss": 0.2978,
      "num_input_tokens_seen": 17099912,
      "step": 25985
    },
    {
      "epoch": 15.32429245283019,
      "grad_norm": 6.128623962402344,
      "learning_rate": 1.5749096478806209e-06,
      "loss": 0.2724,
      "num_input_tokens_seen": 17102664,
      "step": 25990
    },
    {
      "epoch": 15.327240566037736,
      "grad_norm": 4.198847770690918,
      "learning_rate": 1.5730358120524452e-06,
      "loss": 0.31,
      "num_input_tokens_seen": 17106184,
      "step": 25995
    },
    {
      "epoch": 15.330188679245284,
      "grad_norm": 3.625493288040161,
      "learning_rate": 1.5711628835278098e-06,
      "loss": 0.4387,
      "num_input_tokens_seen": 17109320,
      "step": 26000
    },
    {
      "epoch": 15.33313679245283,
      "grad_norm": 3.7062995433807373,
      "learning_rate": 1.5692908628025782e-06,
      "loss": 0.2836,
      "num_input_tokens_seen": 17112392,
      "step": 26005
    },
    {
      "epoch": 15.336084905660377,
      "grad_norm": 2.8011069297790527,
      "learning_rate": 1.5674197503723765e-06,
      "loss": 0.3777,
      "num_input_tokens_seen": 17116840,
      "step": 26010
    },
    {
      "epoch": 15.339033018867925,
      "grad_norm": 5.595163822174072,
      "learning_rate": 1.5655495467325893e-06,
      "loss": 0.3213,
      "num_input_tokens_seen": 17119912,
      "step": 26015
    },
    {
      "epoch": 15.341981132075471,
      "grad_norm": 3.773224115371704,
      "learning_rate": 1.5636802523783613e-06,
      "loss": 0.3411,
      "num_input_tokens_seen": 17123304,
      "step": 26020
    },
    {
      "epoch": 15.34492924528302,
      "grad_norm": 3.2918760776519775,
      "learning_rate": 1.5618118678045947e-06,
      "loss": 0.3106,
      "num_input_tokens_seen": 17127272,
      "step": 26025
    },
    {
      "epoch": 15.347877358490566,
      "grad_norm": 10.885247230529785,
      "learning_rate": 1.5599443935059549e-06,
      "loss": 0.2838,
      "num_input_tokens_seen": 17129544,
      "step": 26030
    },
    {
      "epoch": 15.350825471698114,
      "grad_norm": 3.2444005012512207,
      "learning_rate": 1.5580778299768635e-06,
      "loss": 0.2892,
      "num_input_tokens_seen": 17132872,
      "step": 26035
    },
    {
      "epoch": 15.35377358490566,
      "grad_norm": 4.475638389587402,
      "learning_rate": 1.5562121777114997e-06,
      "loss": 0.3581,
      "num_input_tokens_seen": 17135624,
      "step": 26040
    },
    {
      "epoch": 15.356721698113208,
      "grad_norm": 2.728590488433838,
      "learning_rate": 1.5543474372038043e-06,
      "loss": 0.2923,
      "num_input_tokens_seen": 17138824,
      "step": 26045
    },
    {
      "epoch": 15.359669811320755,
      "grad_norm": 4.945248126983643,
      "learning_rate": 1.5524836089474748e-06,
      "loss": 0.2701,
      "num_input_tokens_seen": 17142184,
      "step": 26050
    },
    {
      "epoch": 15.362617924528301,
      "grad_norm": 4.1867194175720215,
      "learning_rate": 1.5506206934359664e-06,
      "loss": 0.3542,
      "num_input_tokens_seen": 17144744,
      "step": 26055
    },
    {
      "epoch": 15.36556603773585,
      "grad_norm": 2.8861184120178223,
      "learning_rate": 1.5487586911624947e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 17148072,
      "step": 26060
    },
    {
      "epoch": 15.368514150943396,
      "grad_norm": 4.6002655029296875,
      "learning_rate": 1.5468976026200355e-06,
      "loss": 0.3564,
      "num_input_tokens_seen": 17151496,
      "step": 26065
    },
    {
      "epoch": 15.371462264150944,
      "grad_norm": 4.002133846282959,
      "learning_rate": 1.5450374283013187e-06,
      "loss": 0.3105,
      "num_input_tokens_seen": 17154344,
      "step": 26070
    },
    {
      "epoch": 15.37441037735849,
      "grad_norm": 5.580099105834961,
      "learning_rate": 1.5431781686988317e-06,
      "loss": 0.2704,
      "num_input_tokens_seen": 17156904,
      "step": 26075
    },
    {
      "epoch": 15.377358490566039,
      "grad_norm": 2.7816567420959473,
      "learning_rate": 1.5413198243048233e-06,
      "loss": 0.2789,
      "num_input_tokens_seen": 17159464,
      "step": 26080
    },
    {
      "epoch": 15.380306603773585,
      "grad_norm": 4.160646915435791,
      "learning_rate": 1.5394623956112974e-06,
      "loss": 0.3597,
      "num_input_tokens_seen": 17162504,
      "step": 26085
    },
    {
      "epoch": 15.383254716981131,
      "grad_norm": 3.303999185562134,
      "learning_rate": 1.537605883110015e-06,
      "loss": 0.2882,
      "num_input_tokens_seen": 17167112,
      "step": 26090
    },
    {
      "epoch": 15.38620283018868,
      "grad_norm": 12.190568923950195,
      "learning_rate": 1.5357502872924984e-06,
      "loss": 0.3652,
      "num_input_tokens_seen": 17169768,
      "step": 26095
    },
    {
      "epoch": 15.389150943396226,
      "grad_norm": 2.7069199085235596,
      "learning_rate": 1.5338956086500235e-06,
      "loss": 0.3263,
      "num_input_tokens_seen": 17173096,
      "step": 26100
    },
    {
      "epoch": 15.392099056603774,
      "grad_norm": 8.043289184570312,
      "learning_rate": 1.5320418476736237e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 17175848,
      "step": 26105
    },
    {
      "epoch": 15.39504716981132,
      "grad_norm": 8.297844886779785,
      "learning_rate": 1.5301890048540912e-06,
      "loss": 0.2568,
      "num_input_tokens_seen": 17179752,
      "step": 26110
    },
    {
      "epoch": 15.397995283018869,
      "grad_norm": 2.586132526397705,
      "learning_rate": 1.5283370806819743e-06,
      "loss": 0.2369,
      "num_input_tokens_seen": 17183304,
      "step": 26115
    },
    {
      "epoch": 15.400943396226415,
      "grad_norm": 3.1854960918426514,
      "learning_rate": 1.5264860756475752e-06,
      "loss": 0.3078,
      "num_input_tokens_seen": 17186504,
      "step": 26120
    },
    {
      "epoch": 15.403891509433961,
      "grad_norm": 3.2159981727600098,
      "learning_rate": 1.5246359902409592e-06,
      "loss": 0.2922,
      "num_input_tokens_seen": 17190376,
      "step": 26125
    },
    {
      "epoch": 15.40683962264151,
      "grad_norm": 4.208527088165283,
      "learning_rate": 1.5227868249519423e-06,
      "loss": 0.2848,
      "num_input_tokens_seen": 17192776,
      "step": 26130
    },
    {
      "epoch": 15.409787735849056,
      "grad_norm": 3.420417070388794,
      "learning_rate": 1.5209385802700999e-06,
      "loss": 0.2947,
      "num_input_tokens_seen": 17196872,
      "step": 26135
    },
    {
      "epoch": 15.412735849056604,
      "grad_norm": 3.3229494094848633,
      "learning_rate": 1.5190912566847626e-06,
      "loss": 0.3127,
      "num_input_tokens_seen": 17200520,
      "step": 26140
    },
    {
      "epoch": 15.41568396226415,
      "grad_norm": 4.77260160446167,
      "learning_rate": 1.5172448546850166e-06,
      "loss": 0.3653,
      "num_input_tokens_seen": 17203944,
      "step": 26145
    },
    {
      "epoch": 15.418632075471699,
      "grad_norm": 3.089524507522583,
      "learning_rate": 1.515399374759704e-06,
      "loss": 0.2561,
      "num_input_tokens_seen": 17206984,
      "step": 26150
    },
    {
      "epoch": 15.421580188679245,
      "grad_norm": 3.7699460983276367,
      "learning_rate": 1.513554817397424e-06,
      "loss": 0.3816,
      "num_input_tokens_seen": 17210376,
      "step": 26155
    },
    {
      "epoch": 15.424528301886792,
      "grad_norm": 7.022712707519531,
      "learning_rate": 1.5117111830865338e-06,
      "loss": 0.3376,
      "num_input_tokens_seen": 17213160,
      "step": 26160
    },
    {
      "epoch": 15.42747641509434,
      "grad_norm": 2.8840978145599365,
      "learning_rate": 1.509868472315142e-06,
      "loss": 0.3159,
      "num_input_tokens_seen": 17216776,
      "step": 26165
    },
    {
      "epoch": 15.430424528301886,
      "grad_norm": 3.6613941192626953,
      "learning_rate": 1.508026685571113e-06,
      "loss": 0.2942,
      "num_input_tokens_seen": 17220168,
      "step": 26170
    },
    {
      "epoch": 15.433372641509434,
      "grad_norm": 4.754935264587402,
      "learning_rate": 1.506185823342069e-06,
      "loss": 0.4062,
      "num_input_tokens_seen": 17223720,
      "step": 26175
    },
    {
      "epoch": 15.43632075471698,
      "grad_norm": 3.7031733989715576,
      "learning_rate": 1.504345886115386e-06,
      "loss": 0.2379,
      "num_input_tokens_seen": 17226984,
      "step": 26180
    },
    {
      "epoch": 15.439268867924529,
      "grad_norm": 4.104978084564209,
      "learning_rate": 1.502506874378193e-06,
      "loss": 0.3352,
      "num_input_tokens_seen": 17230984,
      "step": 26185
    },
    {
      "epoch": 15.442216981132075,
      "grad_norm": 3.214662551879883,
      "learning_rate": 1.5006687886173805e-06,
      "loss": 0.3597,
      "num_input_tokens_seen": 17234376,
      "step": 26190
    },
    {
      "epoch": 15.445165094339623,
      "grad_norm": 7.0019917488098145,
      "learning_rate": 1.498831629319587e-06,
      "loss": 0.2993,
      "num_input_tokens_seen": 17237064,
      "step": 26195
    },
    {
      "epoch": 15.44811320754717,
      "grad_norm": 6.645082473754883,
      "learning_rate": 1.4969953969712087e-06,
      "loss": 0.3496,
      "num_input_tokens_seen": 17239656,
      "step": 26200
    },
    {
      "epoch": 15.451061320754716,
      "grad_norm": 5.372827529907227,
      "learning_rate": 1.4951600920583963e-06,
      "loss": 0.3404,
      "num_input_tokens_seen": 17242184,
      "step": 26205
    },
    {
      "epoch": 15.454009433962264,
      "grad_norm": 7.471951007843018,
      "learning_rate": 1.493325715067055e-06,
      "loss": 0.3468,
      "num_input_tokens_seen": 17245768,
      "step": 26210
    },
    {
      "epoch": 15.45695754716981,
      "grad_norm": 2.635089635848999,
      "learning_rate": 1.4914922664828417e-06,
      "loss": 0.296,
      "num_input_tokens_seen": 17249384,
      "step": 26215
    },
    {
      "epoch": 15.459905660377359,
      "grad_norm": 3.4694414138793945,
      "learning_rate": 1.4896597467911732e-06,
      "loss": 0.2319,
      "num_input_tokens_seen": 17252520,
      "step": 26220
    },
    {
      "epoch": 15.462853773584905,
      "grad_norm": 3.1510226726531982,
      "learning_rate": 1.4878281564772156e-06,
      "loss": 0.248,
      "num_input_tokens_seen": 17256072,
      "step": 26225
    },
    {
      "epoch": 15.465801886792454,
      "grad_norm": 4.008911609649658,
      "learning_rate": 1.4859974960258898e-06,
      "loss": 0.3952,
      "num_input_tokens_seen": 17258728,
      "step": 26230
    },
    {
      "epoch": 15.46875,
      "grad_norm": 3.076681613922119,
      "learning_rate": 1.4841677659218723e-06,
      "loss": 0.4726,
      "num_input_tokens_seen": 17263304,
      "step": 26235
    },
    {
      "epoch": 15.471698113207546,
      "grad_norm": 5.7277913093566895,
      "learning_rate": 1.4823389666495886e-06,
      "loss": 0.2997,
      "num_input_tokens_seen": 17265832,
      "step": 26240
    },
    {
      "epoch": 15.474646226415095,
      "grad_norm": 2.672200918197632,
      "learning_rate": 1.4805110986932258e-06,
      "loss": 0.3383,
      "num_input_tokens_seen": 17272584,
      "step": 26245
    },
    {
      "epoch": 15.477594339622641,
      "grad_norm": 6.042201519012451,
      "learning_rate": 1.4786841625367166e-06,
      "loss": 0.2323,
      "num_input_tokens_seen": 17275944,
      "step": 26250
    },
    {
      "epoch": 15.48054245283019,
      "grad_norm": 4.966555118560791,
      "learning_rate": 1.476858158663752e-06,
      "loss": 0.2672,
      "num_input_tokens_seen": 17279528,
      "step": 26255
    },
    {
      "epoch": 15.483490566037736,
      "grad_norm": 3.3852689266204834,
      "learning_rate": 1.4750330875577745e-06,
      "loss": 0.2695,
      "num_input_tokens_seen": 17282152,
      "step": 26260
    },
    {
      "epoch": 15.486438679245284,
      "grad_norm": 2.3446621894836426,
      "learning_rate": 1.4732089497019787e-06,
      "loss": 0.2342,
      "num_input_tokens_seen": 17285448,
      "step": 26265
    },
    {
      "epoch": 15.48938679245283,
      "grad_norm": 3.9275126457214355,
      "learning_rate": 1.471385745579313e-06,
      "loss": 0.2585,
      "num_input_tokens_seen": 17288936,
      "step": 26270
    },
    {
      "epoch": 15.492334905660377,
      "grad_norm": 4.406417369842529,
      "learning_rate": 1.4695634756724775e-06,
      "loss": 0.3861,
      "num_input_tokens_seen": 17292488,
      "step": 26275
    },
    {
      "epoch": 15.495283018867925,
      "grad_norm": 3.357255220413208,
      "learning_rate": 1.4677421404639281e-06,
      "loss": 0.2907,
      "num_input_tokens_seen": 17295016,
      "step": 26280
    },
    {
      "epoch": 15.498231132075471,
      "grad_norm": 5.405197620391846,
      "learning_rate": 1.4659217404358706e-06,
      "loss": 0.3993,
      "num_input_tokens_seen": 17297960,
      "step": 26285
    },
    {
      "epoch": 15.50117924528302,
      "grad_norm": 3.444270372390747,
      "learning_rate": 1.4641022760702627e-06,
      "loss": 0.223,
      "num_input_tokens_seen": 17301160,
      "step": 26290
    },
    {
      "epoch": 15.504127358490566,
      "grad_norm": 2.879955530166626,
      "learning_rate": 1.4622837478488172e-06,
      "loss": 0.2293,
      "num_input_tokens_seen": 17307720,
      "step": 26295
    },
    {
      "epoch": 15.507075471698114,
      "grad_norm": 3.277707099914551,
      "learning_rate": 1.4604661562529953e-06,
      "loss": 0.2191,
      "num_input_tokens_seen": 17315656,
      "step": 26300
    },
    {
      "epoch": 15.51002358490566,
      "grad_norm": 5.907118797302246,
      "learning_rate": 1.4586495017640119e-06,
      "loss": 0.3279,
      "num_input_tokens_seen": 17318536,
      "step": 26305
    },
    {
      "epoch": 15.512971698113208,
      "grad_norm": 2.400219202041626,
      "learning_rate": 1.4568337848628366e-06,
      "loss": 0.2405,
      "num_input_tokens_seen": 17321256,
      "step": 26310
    },
    {
      "epoch": 15.515919811320755,
      "grad_norm": 12.104177474975586,
      "learning_rate": 1.4550190060301872e-06,
      "loss": 0.2085,
      "num_input_tokens_seen": 17324584,
      "step": 26315
    },
    {
      "epoch": 15.518867924528301,
      "grad_norm": 3.660285472869873,
      "learning_rate": 1.4532051657465335e-06,
      "loss": 0.4389,
      "num_input_tokens_seen": 17327848,
      "step": 26320
    },
    {
      "epoch": 15.52181603773585,
      "grad_norm": 5.437450408935547,
      "learning_rate": 1.4513922644920985e-06,
      "loss": 0.3576,
      "num_input_tokens_seen": 17330600,
      "step": 26325
    },
    {
      "epoch": 15.524764150943396,
      "grad_norm": 4.824192047119141,
      "learning_rate": 1.4495803027468552e-06,
      "loss": 0.337,
      "num_input_tokens_seen": 17334088,
      "step": 26330
    },
    {
      "epoch": 15.527712264150944,
      "grad_norm": 2.513124942779541,
      "learning_rate": 1.4477692809905263e-06,
      "loss": 0.3193,
      "num_input_tokens_seen": 17337288,
      "step": 26335
    },
    {
      "epoch": 15.53066037735849,
      "grad_norm": 3.0200414657592773,
      "learning_rate": 1.4459591997025896e-06,
      "loss": 0.2271,
      "num_input_tokens_seen": 17340808,
      "step": 26340
    },
    {
      "epoch": 15.533608490566039,
      "grad_norm": 5.431547164916992,
      "learning_rate": 1.4441500593622737e-06,
      "loss": 0.343,
      "num_input_tokens_seen": 17343848,
      "step": 26345
    },
    {
      "epoch": 15.536556603773585,
      "grad_norm": 4.616041660308838,
      "learning_rate": 1.4423418604485539e-06,
      "loss": 0.3277,
      "num_input_tokens_seen": 17347592,
      "step": 26350
    },
    {
      "epoch": 15.539504716981131,
      "grad_norm": 4.8668365478515625,
      "learning_rate": 1.4405346034401597e-06,
      "loss": 0.4064,
      "num_input_tokens_seen": 17350312,
      "step": 26355
    },
    {
      "epoch": 15.54245283018868,
      "grad_norm": 2.4935977458953857,
      "learning_rate": 1.4387282888155695e-06,
      "loss": 0.3379,
      "num_input_tokens_seen": 17353192,
      "step": 26360
    },
    {
      "epoch": 15.545400943396226,
      "grad_norm": 5.1062703132629395,
      "learning_rate": 1.436922917053013e-06,
      "loss": 0.2627,
      "num_input_tokens_seen": 17355752,
      "step": 26365
    },
    {
      "epoch": 15.548349056603774,
      "grad_norm": 3.314178705215454,
      "learning_rate": 1.4351184886304686e-06,
      "loss": 0.3735,
      "num_input_tokens_seen": 17359496,
      "step": 26370
    },
    {
      "epoch": 15.55129716981132,
      "grad_norm": 4.066352367401123,
      "learning_rate": 1.4333150040256699e-06,
      "loss": 0.3545,
      "num_input_tokens_seen": 17362760,
      "step": 26375
    },
    {
      "epoch": 15.554245283018869,
      "grad_norm": 3.3013906478881836,
      "learning_rate": 1.4315124637160954e-06,
      "loss": 0.335,
      "num_input_tokens_seen": 17365992,
      "step": 26380
    },
    {
      "epoch": 15.557193396226415,
      "grad_norm": 7.206908226013184,
      "learning_rate": 1.4297108681789752e-06,
      "loss": 0.3401,
      "num_input_tokens_seen": 17368936,
      "step": 26385
    },
    {
      "epoch": 15.560141509433961,
      "grad_norm": 3.3131630420684814,
      "learning_rate": 1.4279102178912902e-06,
      "loss": 0.3757,
      "num_input_tokens_seen": 17373000,
      "step": 26390
    },
    {
      "epoch": 15.56308962264151,
      "grad_norm": 3.2862048149108887,
      "learning_rate": 1.4261105133297693e-06,
      "loss": 0.5194,
      "num_input_tokens_seen": 17376200,
      "step": 26395
    },
    {
      "epoch": 15.566037735849056,
      "grad_norm": 3.9058074951171875,
      "learning_rate": 1.4243117549708913e-06,
      "loss": 0.3226,
      "num_input_tokens_seen": 17382728,
      "step": 26400
    },
    {
      "epoch": 15.568985849056604,
      "grad_norm": 3.995713233947754,
      "learning_rate": 1.422513943290888e-06,
      "loss": 0.2979,
      "num_input_tokens_seen": 17386376,
      "step": 26405
    },
    {
      "epoch": 15.57193396226415,
      "grad_norm": 2.8996832370758057,
      "learning_rate": 1.4207170787657365e-06,
      "loss": 0.2804,
      "num_input_tokens_seen": 17390216,
      "step": 26410
    },
    {
      "epoch": 15.574882075471699,
      "grad_norm": 4.785772800445557,
      "learning_rate": 1.4189211618711646e-06,
      "loss": 0.3759,
      "num_input_tokens_seen": 17393064,
      "step": 26415
    },
    {
      "epoch": 15.577830188679245,
      "grad_norm": 4.123541355133057,
      "learning_rate": 1.417126193082648e-06,
      "loss": 0.3598,
      "num_input_tokens_seen": 17396328,
      "step": 26420
    },
    {
      "epoch": 15.580778301886792,
      "grad_norm": 6.128563404083252,
      "learning_rate": 1.4153321728754133e-06,
      "loss": 0.3908,
      "num_input_tokens_seen": 17399656,
      "step": 26425
    },
    {
      "epoch": 15.58372641509434,
      "grad_norm": 2.3929264545440674,
      "learning_rate": 1.4135391017244338e-06,
      "loss": 0.2471,
      "num_input_tokens_seen": 17402664,
      "step": 26430
    },
    {
      "epoch": 15.586674528301886,
      "grad_norm": 2.484192371368408,
      "learning_rate": 1.4117469801044332e-06,
      "loss": 0.337,
      "num_input_tokens_seen": 17406504,
      "step": 26435
    },
    {
      "epoch": 15.589622641509434,
      "grad_norm": 4.599382400512695,
      "learning_rate": 1.4099558084898862e-06,
      "loss": 0.3602,
      "num_input_tokens_seen": 17409544,
      "step": 26440
    },
    {
      "epoch": 15.59257075471698,
      "grad_norm": 5.225503921508789,
      "learning_rate": 1.408165587355011e-06,
      "loss": 0.4229,
      "num_input_tokens_seen": 17412872,
      "step": 26445
    },
    {
      "epoch": 15.595518867924529,
      "grad_norm": 2.9405081272125244,
      "learning_rate": 1.4063763171737766e-06,
      "loss": 0.2604,
      "num_input_tokens_seen": 17416552,
      "step": 26450
    },
    {
      "epoch": 15.598466981132075,
      "grad_norm": 3.640458583831787,
      "learning_rate": 1.4045879984198996e-06,
      "loss": 0.3157,
      "num_input_tokens_seen": 17420360,
      "step": 26455
    },
    {
      "epoch": 15.601415094339622,
      "grad_norm": 3.270742893218994,
      "learning_rate": 1.4028006315668457e-06,
      "loss": 0.3493,
      "num_input_tokens_seen": 17424520,
      "step": 26460
    },
    {
      "epoch": 15.60436320754717,
      "grad_norm": 4.251747131347656,
      "learning_rate": 1.4010142170878261e-06,
      "loss": 0.3189,
      "num_input_tokens_seen": 17427656,
      "step": 26465
    },
    {
      "epoch": 15.607311320754716,
      "grad_norm": 2.49747633934021,
      "learning_rate": 1.3992287554558042e-06,
      "loss": 0.2774,
      "num_input_tokens_seen": 17431304,
      "step": 26470
    },
    {
      "epoch": 15.610259433962264,
      "grad_norm": 2.64349627494812,
      "learning_rate": 1.3974442471434885e-06,
      "loss": 0.2503,
      "num_input_tokens_seen": 17434824,
      "step": 26475
    },
    {
      "epoch": 15.61320754716981,
      "grad_norm": 5.284521102905273,
      "learning_rate": 1.395660692623334e-06,
      "loss": 0.3014,
      "num_input_tokens_seen": 17437928,
      "step": 26480
    },
    {
      "epoch": 15.616155660377359,
      "grad_norm": 3.2819457054138184,
      "learning_rate": 1.3938780923675454e-06,
      "loss": 0.2935,
      "num_input_tokens_seen": 17440616,
      "step": 26485
    },
    {
      "epoch": 15.619103773584905,
      "grad_norm": 4.64290714263916,
      "learning_rate": 1.3920964468480718e-06,
      "loss": 0.3573,
      "num_input_tokens_seen": 17443624,
      "step": 26490
    },
    {
      "epoch": 15.622051886792454,
      "grad_norm": 3.6371469497680664,
      "learning_rate": 1.3903157565366143e-06,
      "loss": 0.3904,
      "num_input_tokens_seen": 17447208,
      "step": 26495
    },
    {
      "epoch": 15.625,
      "grad_norm": 3.6835179328918457,
      "learning_rate": 1.3885360219046172e-06,
      "loss": 0.3225,
      "num_input_tokens_seen": 17450120,
      "step": 26500
    },
    {
      "epoch": 15.627948113207546,
      "grad_norm": 3.5799365043640137,
      "learning_rate": 1.386757243423273e-06,
      "loss": 0.357,
      "num_input_tokens_seen": 17453128,
      "step": 26505
    },
    {
      "epoch": 15.630896226415095,
      "grad_norm": 5.802070140838623,
      "learning_rate": 1.384979421563521e-06,
      "loss": 0.3195,
      "num_input_tokens_seen": 17456488,
      "step": 26510
    },
    {
      "epoch": 15.633844339622641,
      "grad_norm": 2.168452501296997,
      "learning_rate": 1.3832025567960465e-06,
      "loss": 0.2787,
      "num_input_tokens_seen": 17460424,
      "step": 26515
    },
    {
      "epoch": 15.63679245283019,
      "grad_norm": 2.3698618412017822,
      "learning_rate": 1.3814266495912815e-06,
      "loss": 0.2895,
      "num_input_tokens_seen": 17464104,
      "step": 26520
    },
    {
      "epoch": 15.639740566037736,
      "grad_norm": 2.293449878692627,
      "learning_rate": 1.3796517004194078e-06,
      "loss": 0.3036,
      "num_input_tokens_seen": 17467464,
      "step": 26525
    },
    {
      "epoch": 15.642688679245284,
      "grad_norm": 3.728227138519287,
      "learning_rate": 1.3778777097503476e-06,
      "loss": 0.2344,
      "num_input_tokens_seen": 17470184,
      "step": 26530
    },
    {
      "epoch": 15.64563679245283,
      "grad_norm": 1.977829098701477,
      "learning_rate": 1.3761046780537757e-06,
      "loss": 0.3325,
      "num_input_tokens_seen": 17473480,
      "step": 26535
    },
    {
      "epoch": 15.648584905660378,
      "grad_norm": 2.314697265625,
      "learning_rate": 1.3743326057991086e-06,
      "loss": 0.3329,
      "num_input_tokens_seen": 17476520,
      "step": 26540
    },
    {
      "epoch": 15.651533018867925,
      "grad_norm": 4.851804733276367,
      "learning_rate": 1.3725614934555093e-06,
      "loss": 0.4122,
      "num_input_tokens_seen": 17480232,
      "step": 26545
    },
    {
      "epoch": 15.654481132075471,
      "grad_norm": 3.292724609375,
      "learning_rate": 1.3707913414918882e-06,
      "loss": 0.3684,
      "num_input_tokens_seen": 17483208,
      "step": 26550
    },
    {
      "epoch": 15.65742924528302,
      "grad_norm": 3.5148065090179443,
      "learning_rate": 1.3690221503768996e-06,
      "loss": 0.5072,
      "num_input_tokens_seen": 17486888,
      "step": 26555
    },
    {
      "epoch": 15.660377358490566,
      "grad_norm": 3.2296030521392822,
      "learning_rate": 1.3672539205789465e-06,
      "loss": 0.2706,
      "num_input_tokens_seen": 17491048,
      "step": 26560
    },
    {
      "epoch": 15.663325471698114,
      "grad_norm": 3.8956077098846436,
      "learning_rate": 1.3654866525661737e-06,
      "loss": 0.3145,
      "num_input_tokens_seen": 17494120,
      "step": 26565
    },
    {
      "epoch": 15.66627358490566,
      "grad_norm": 4.456235408782959,
      "learning_rate": 1.3637203468064741e-06,
      "loss": 0.3022,
      "num_input_tokens_seen": 17496936,
      "step": 26570
    },
    {
      "epoch": 15.669221698113208,
      "grad_norm": 1.7804807424545288,
      "learning_rate": 1.3619550037674838e-06,
      "loss": 0.2437,
      "num_input_tokens_seen": 17501288,
      "step": 26575
    },
    {
      "epoch": 15.672169811320755,
      "grad_norm": 6.188649654388428,
      "learning_rate": 1.3601906239165857e-06,
      "loss": 0.3058,
      "num_input_tokens_seen": 17505064,
      "step": 26580
    },
    {
      "epoch": 15.675117924528301,
      "grad_norm": 3.3427727222442627,
      "learning_rate": 1.3584272077209048e-06,
      "loss": 0.281,
      "num_input_tokens_seen": 17508264,
      "step": 26585
    },
    {
      "epoch": 15.67806603773585,
      "grad_norm": 2.5544400215148926,
      "learning_rate": 1.3566647556473168e-06,
      "loss": 0.4189,
      "num_input_tokens_seen": 17511240,
      "step": 26590
    },
    {
      "epoch": 15.681014150943396,
      "grad_norm": 3.368595600128174,
      "learning_rate": 1.3549032681624363e-06,
      "loss": 0.2596,
      "num_input_tokens_seen": 17513512,
      "step": 26595
    },
    {
      "epoch": 15.683962264150944,
      "grad_norm": 2.758190870285034,
      "learning_rate": 1.3531427457326252e-06,
      "loss": 0.222,
      "num_input_tokens_seen": 17517320,
      "step": 26600
    },
    {
      "epoch": 15.68691037735849,
      "grad_norm": 3.7079899311065674,
      "learning_rate": 1.3513831888239893e-06,
      "loss": 0.4369,
      "num_input_tokens_seen": 17521160,
      "step": 26605
    },
    {
      "epoch": 15.689858490566039,
      "grad_norm": 4.926781177520752,
      "learning_rate": 1.3496245979023786e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 17523880,
      "step": 26610
    },
    {
      "epoch": 15.692806603773585,
      "grad_norm": 6.095827579498291,
      "learning_rate": 1.3478669734333865e-06,
      "loss": 0.3075,
      "num_input_tokens_seen": 17526920,
      "step": 26615
    },
    {
      "epoch": 15.695754716981131,
      "grad_norm": 2.9893884658813477,
      "learning_rate": 1.3461103158823546e-06,
      "loss": 0.3274,
      "num_input_tokens_seen": 17529992,
      "step": 26620
    },
    {
      "epoch": 15.69870283018868,
      "grad_norm": 3.381091356277466,
      "learning_rate": 1.3443546257143624e-06,
      "loss": 0.3276,
      "num_input_tokens_seen": 17532648,
      "step": 26625
    },
    {
      "epoch": 15.701650943396226,
      "grad_norm": 4.352731704711914,
      "learning_rate": 1.3425999033942395e-06,
      "loss": 0.4221,
      "num_input_tokens_seen": 17535784,
      "step": 26630
    },
    {
      "epoch": 15.704599056603774,
      "grad_norm": 4.921932220458984,
      "learning_rate": 1.3408461493865549e-06,
      "loss": 0.2913,
      "num_input_tokens_seen": 17538632,
      "step": 26635
    },
    {
      "epoch": 15.70754716981132,
      "grad_norm": 4.2075114250183105,
      "learning_rate": 1.339093364155622e-06,
      "loss": 0.4294,
      "num_input_tokens_seen": 17541384,
      "step": 26640
    },
    {
      "epoch": 15.710495283018869,
      "grad_norm": 4.702098846435547,
      "learning_rate": 1.3373415481654988e-06,
      "loss": 0.3212,
      "num_input_tokens_seen": 17544200,
      "step": 26645
    },
    {
      "epoch": 15.713443396226415,
      "grad_norm": 4.32122802734375,
      "learning_rate": 1.335590701879984e-06,
      "loss": 0.3172,
      "num_input_tokens_seen": 17547240,
      "step": 26650
    },
    {
      "epoch": 15.716391509433961,
      "grad_norm": 4.272386074066162,
      "learning_rate": 1.3338408257626257e-06,
      "loss": 0.2738,
      "num_input_tokens_seen": 17549992,
      "step": 26655
    },
    {
      "epoch": 15.71933962264151,
      "grad_norm": 3.1140694618225098,
      "learning_rate": 1.3320919202767086e-06,
      "loss": 0.2783,
      "num_input_tokens_seen": 17553384,
      "step": 26660
    },
    {
      "epoch": 15.722287735849056,
      "grad_norm": 4.05529260635376,
      "learning_rate": 1.3303439858852636e-06,
      "loss": 0.2985,
      "num_input_tokens_seen": 17556168,
      "step": 26665
    },
    {
      "epoch": 15.725235849056604,
      "grad_norm": 2.7551493644714355,
      "learning_rate": 1.3285970230510636e-06,
      "loss": 0.3536,
      "num_input_tokens_seen": 17560168,
      "step": 26670
    },
    {
      "epoch": 15.72818396226415,
      "grad_norm": 2.940675735473633,
      "learning_rate": 1.3268510322366246e-06,
      "loss": 0.3398,
      "num_input_tokens_seen": 17562984,
      "step": 26675
    },
    {
      "epoch": 15.731132075471699,
      "grad_norm": 3.09352445602417,
      "learning_rate": 1.3251060139042038e-06,
      "loss": 0.2983,
      "num_input_tokens_seen": 17565928,
      "step": 26680
    },
    {
      "epoch": 15.734080188679245,
      "grad_norm": 2.5389864444732666,
      "learning_rate": 1.3233619685158056e-06,
      "loss": 0.2653,
      "num_input_tokens_seen": 17569480,
      "step": 26685
    },
    {
      "epoch": 15.737028301886792,
      "grad_norm": 4.569450378417969,
      "learning_rate": 1.3216188965331712e-06,
      "loss": 0.3667,
      "num_input_tokens_seen": 17572424,
      "step": 26690
    },
    {
      "epoch": 15.73997641509434,
      "grad_norm": 5.618490219116211,
      "learning_rate": 1.3198767984177869e-06,
      "loss": 0.1821,
      "num_input_tokens_seen": 17575176,
      "step": 26695
    },
    {
      "epoch": 15.742924528301886,
      "grad_norm": 8.23602294921875,
      "learning_rate": 1.3181356746308805e-06,
      "loss": 0.2795,
      "num_input_tokens_seen": 17578024,
      "step": 26700
    },
    {
      "epoch": 15.745872641509434,
      "grad_norm": 3.066056728363037,
      "learning_rate": 1.3163955256334226e-06,
      "loss": 0.2513,
      "num_input_tokens_seen": 17581384,
      "step": 26705
    },
    {
      "epoch": 15.74882075471698,
      "grad_norm": 5.679169178009033,
      "learning_rate": 1.3146563518861227e-06,
      "loss": 0.3124,
      "num_input_tokens_seen": 17585128,
      "step": 26710
    },
    {
      "epoch": 15.751768867924529,
      "grad_norm": 5.253920555114746,
      "learning_rate": 1.3129181538494384e-06,
      "loss": 0.3575,
      "num_input_tokens_seen": 17588392,
      "step": 26715
    },
    {
      "epoch": 15.754716981132075,
      "grad_norm": 2.9442129135131836,
      "learning_rate": 1.3111809319835622e-06,
      "loss": 0.566,
      "num_input_tokens_seen": 17591016,
      "step": 26720
    },
    {
      "epoch": 15.757665094339622,
      "grad_norm": 2.8937923908233643,
      "learning_rate": 1.3094446867484335e-06,
      "loss": 0.2884,
      "num_input_tokens_seen": 17594536,
      "step": 26725
    },
    {
      "epoch": 15.76061320754717,
      "grad_norm": 8.077138900756836,
      "learning_rate": 1.3077094186037287e-06,
      "loss": 0.2849,
      "num_input_tokens_seen": 17597096,
      "step": 26730
    },
    {
      "epoch": 15.763561320754716,
      "grad_norm": 4.450479030609131,
      "learning_rate": 1.305975128008869e-06,
      "loss": 0.2928,
      "num_input_tokens_seen": 17599400,
      "step": 26735
    },
    {
      "epoch": 15.766509433962264,
      "grad_norm": 3.2897849082946777,
      "learning_rate": 1.304241815423014e-06,
      "loss": 0.2106,
      "num_input_tokens_seen": 17602472,
      "step": 26740
    },
    {
      "epoch": 15.76945754716981,
      "grad_norm": 5.190310955047607,
      "learning_rate": 1.3025094813050655e-06,
      "loss": 0.3341,
      "num_input_tokens_seen": 17605576,
      "step": 26745
    },
    {
      "epoch": 15.772405660377359,
      "grad_norm": 3.498211622238159,
      "learning_rate": 1.3007781261136675e-06,
      "loss": 0.3119,
      "num_input_tokens_seen": 17612648,
      "step": 26750
    },
    {
      "epoch": 15.775353773584905,
      "grad_norm": 11.196240425109863,
      "learning_rate": 1.299047750307204e-06,
      "loss": 0.337,
      "num_input_tokens_seen": 17615624,
      "step": 26755
    },
    {
      "epoch": 15.778301886792454,
      "grad_norm": 3.425107479095459,
      "learning_rate": 1.297318354343799e-06,
      "loss": 0.4219,
      "num_input_tokens_seen": 17618472,
      "step": 26760
    },
    {
      "epoch": 15.78125,
      "grad_norm": 9.0136137008667,
      "learning_rate": 1.295589938681317e-06,
      "loss": 0.3267,
      "num_input_tokens_seen": 17622248,
      "step": 26765
    },
    {
      "epoch": 15.784198113207546,
      "grad_norm": 3.8815271854400635,
      "learning_rate": 1.2938625037773628e-06,
      "loss": 0.3135,
      "num_input_tokens_seen": 17625576,
      "step": 26770
    },
    {
      "epoch": 15.787146226415095,
      "grad_norm": 3.5528552532196045,
      "learning_rate": 1.2921360500892843e-06,
      "loss": 0.425,
      "num_input_tokens_seen": 17628424,
      "step": 26775
    },
    {
      "epoch": 15.790094339622641,
      "grad_norm": 6.037795066833496,
      "learning_rate": 1.290410578074167e-06,
      "loss": 0.2519,
      "num_input_tokens_seen": 17634600,
      "step": 26780
    },
    {
      "epoch": 15.79304245283019,
      "grad_norm": 3.8415191173553467,
      "learning_rate": 1.2886860881888362e-06,
      "loss": 0.3263,
      "num_input_tokens_seen": 17637384,
      "step": 26785
    },
    {
      "epoch": 15.795990566037736,
      "grad_norm": 5.050204753875732,
      "learning_rate": 1.2869625808898584e-06,
      "loss": 0.3675,
      "num_input_tokens_seen": 17644712,
      "step": 26790
    },
    {
      "epoch": 15.798938679245284,
      "grad_norm": 4.383384704589844,
      "learning_rate": 1.2852400566335398e-06,
      "loss": 0.2799,
      "num_input_tokens_seen": 17647304,
      "step": 26795
    },
    {
      "epoch": 15.80188679245283,
      "grad_norm": 3.503438711166382,
      "learning_rate": 1.2835185158759244e-06,
      "loss": 0.2208,
      "num_input_tokens_seen": 17650536,
      "step": 26800
    },
    {
      "epoch": 15.804834905660378,
      "grad_norm": 4.631406307220459,
      "learning_rate": 1.2817979590728009e-06,
      "loss": 0.3264,
      "num_input_tokens_seen": 17653352,
      "step": 26805
    },
    {
      "epoch": 15.807783018867925,
      "grad_norm": 4.545910835266113,
      "learning_rate": 1.2800783866796918e-06,
      "loss": 0.3008,
      "num_input_tokens_seen": 17656808,
      "step": 26810
    },
    {
      "epoch": 15.810731132075471,
      "grad_norm": 5.085482597351074,
      "learning_rate": 1.2783597991518604e-06,
      "loss": 0.3224,
      "num_input_tokens_seen": 17660008,
      "step": 26815
    },
    {
      "epoch": 15.81367924528302,
      "grad_norm": 4.06857967376709,
      "learning_rate": 1.2766421969443131e-06,
      "loss": 0.3061,
      "num_input_tokens_seen": 17663752,
      "step": 26820
    },
    {
      "epoch": 15.816627358490566,
      "grad_norm": 2.8813469409942627,
      "learning_rate": 1.274925580511791e-06,
      "loss": 0.4666,
      "num_input_tokens_seen": 17667944,
      "step": 26825
    },
    {
      "epoch": 15.819575471698114,
      "grad_norm": 7.237903594970703,
      "learning_rate": 1.2732099503087757e-06,
      "loss": 0.2726,
      "num_input_tokens_seen": 17670600,
      "step": 26830
    },
    {
      "epoch": 15.82252358490566,
      "grad_norm": 2.8409063816070557,
      "learning_rate": 1.2714953067894859e-06,
      "loss": 0.2364,
      "num_input_tokens_seen": 17673384,
      "step": 26835
    },
    {
      "epoch": 15.825471698113208,
      "grad_norm": 3.5230512619018555,
      "learning_rate": 1.2697816504078847e-06,
      "loss": 0.3049,
      "num_input_tokens_seen": 17676232,
      "step": 26840
    },
    {
      "epoch": 15.828419811320755,
      "grad_norm": 6.421483993530273,
      "learning_rate": 1.2680689816176672e-06,
      "loss": 0.3372,
      "num_input_tokens_seen": 17679784,
      "step": 26845
    },
    {
      "epoch": 15.831367924528301,
      "grad_norm": 1.969249963760376,
      "learning_rate": 1.2663573008722707e-06,
      "loss": 0.2205,
      "num_input_tokens_seen": 17683944,
      "step": 26850
    },
    {
      "epoch": 15.83431603773585,
      "grad_norm": 3.0161070823669434,
      "learning_rate": 1.2646466086248698e-06,
      "loss": 0.4675,
      "num_input_tokens_seen": 17687176,
      "step": 26855
    },
    {
      "epoch": 15.837264150943396,
      "grad_norm": 6.529024124145508,
      "learning_rate": 1.2629369053283779e-06,
      "loss": 0.269,
      "num_input_tokens_seen": 17689864,
      "step": 26860
    },
    {
      "epoch": 15.840212264150944,
      "grad_norm": 2.9326086044311523,
      "learning_rate": 1.2612281914354452e-06,
      "loss": 0.3144,
      "num_input_tokens_seen": 17693736,
      "step": 26865
    },
    {
      "epoch": 15.84316037735849,
      "grad_norm": 2.8180153369903564,
      "learning_rate": 1.259520467398463e-06,
      "loss": 0.2838,
      "num_input_tokens_seen": 17696616,
      "step": 26870
    },
    {
      "epoch": 15.846108490566039,
      "grad_norm": 5.854559421539307,
      "learning_rate": 1.2578137336695573e-06,
      "loss": 0.2717,
      "num_input_tokens_seen": 17699272,
      "step": 26875
    },
    {
      "epoch": 15.849056603773585,
      "grad_norm": 4.689304351806641,
      "learning_rate": 1.256107990700594e-06,
      "loss": 0.354,
      "num_input_tokens_seen": 17703176,
      "step": 26880
    },
    {
      "epoch": 15.852004716981131,
      "grad_norm": 7.426267147064209,
      "learning_rate": 1.2544032389431753e-06,
      "loss": 0.2978,
      "num_input_tokens_seen": 17706024,
      "step": 26885
    },
    {
      "epoch": 15.85495283018868,
      "grad_norm": 3.2108702659606934,
      "learning_rate": 1.2526994788486418e-06,
      "loss": 0.3543,
      "num_input_tokens_seen": 17709800,
      "step": 26890
    },
    {
      "epoch": 15.857900943396226,
      "grad_norm": 4.127328395843506,
      "learning_rate": 1.2509967108680697e-06,
      "loss": 0.3739,
      "num_input_tokens_seen": 17713256,
      "step": 26895
    },
    {
      "epoch": 15.860849056603774,
      "grad_norm": 3.768780469894409,
      "learning_rate": 1.249294935452277e-06,
      "loss": 0.3074,
      "num_input_tokens_seen": 17716488,
      "step": 26900
    },
    {
      "epoch": 15.86379716981132,
      "grad_norm": 5.616596221923828,
      "learning_rate": 1.247594153051815e-06,
      "loss": 0.328,
      "num_input_tokens_seen": 17720744,
      "step": 26905
    },
    {
      "epoch": 15.866745283018869,
      "grad_norm": 3.9310624599456787,
      "learning_rate": 1.2458943641169718e-06,
      "loss": 0.2243,
      "num_input_tokens_seen": 17723912,
      "step": 26910
    },
    {
      "epoch": 15.869693396226415,
      "grad_norm": 2.319870710372925,
      "learning_rate": 1.2441955690977758e-06,
      "loss": 0.2874,
      "num_input_tokens_seen": 17728392,
      "step": 26915
    },
    {
      "epoch": 15.872641509433961,
      "grad_norm": 3.426382541656494,
      "learning_rate": 1.2424977684439898e-06,
      "loss": 0.2862,
      "num_input_tokens_seen": 17730952,
      "step": 26920
    },
    {
      "epoch": 15.87558962264151,
      "grad_norm": 8.35843563079834,
      "learning_rate": 1.2408009626051137e-06,
      "loss": 0.2427,
      "num_input_tokens_seen": 17733576,
      "step": 26925
    },
    {
      "epoch": 15.878537735849056,
      "grad_norm": 5.907161235809326,
      "learning_rate": 1.2391051520303826e-06,
      "loss": 0.3543,
      "num_input_tokens_seen": 17736008,
      "step": 26930
    },
    {
      "epoch": 15.881485849056604,
      "grad_norm": 4.72495698928833,
      "learning_rate": 1.2374103371687723e-06,
      "loss": 0.3884,
      "num_input_tokens_seen": 17739048,
      "step": 26935
    },
    {
      "epoch": 15.88443396226415,
      "grad_norm": 4.1627678871154785,
      "learning_rate": 1.2357165184689906e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 17742440,
      "step": 26940
    },
    {
      "epoch": 15.887382075471699,
      "grad_norm": 7.628767013549805,
      "learning_rate": 1.2340236963794845e-06,
      "loss": 0.378,
      "num_input_tokens_seen": 17744904,
      "step": 26945
    },
    {
      "epoch": 15.890330188679245,
      "grad_norm": 5.873554229736328,
      "learning_rate": 1.232331871348435e-06,
      "loss": 0.3055,
      "num_input_tokens_seen": 17747624,
      "step": 26950
    },
    {
      "epoch": 15.893278301886792,
      "grad_norm": 4.113348484039307,
      "learning_rate": 1.2306410438237603e-06,
      "loss": 0.3857,
      "num_input_tokens_seen": 17751560,
      "step": 26955
    },
    {
      "epoch": 15.89622641509434,
      "grad_norm": 4.993739128112793,
      "learning_rate": 1.228951214253113e-06,
      "loss": 0.572,
      "num_input_tokens_seen": 17754088,
      "step": 26960
    },
    {
      "epoch": 15.899174528301886,
      "grad_norm": 3.8513247966766357,
      "learning_rate": 1.2272623830838854e-06,
      "loss": 0.3336,
      "num_input_tokens_seen": 17757224,
      "step": 26965
    },
    {
      "epoch": 15.902122641509434,
      "grad_norm": 3.664928913116455,
      "learning_rate": 1.2255745507632016e-06,
      "loss": 0.3798,
      "num_input_tokens_seen": 17760520,
      "step": 26970
    },
    {
      "epoch": 15.90507075471698,
      "grad_norm": 2.723865509033203,
      "learning_rate": 1.223887717737922e-06,
      "loss": 0.3172,
      "num_input_tokens_seen": 17763368,
      "step": 26975
    },
    {
      "epoch": 15.908018867924529,
      "grad_norm": 3.438472032546997,
      "learning_rate": 1.2222018844546434e-06,
      "loss": 0.2383,
      "num_input_tokens_seen": 17766184,
      "step": 26980
    },
    {
      "epoch": 15.910966981132075,
      "grad_norm": 4.567757606506348,
      "learning_rate": 1.2205170513596975e-06,
      "loss": 0.3076,
      "num_input_tokens_seen": 17769256,
      "step": 26985
    },
    {
      "epoch": 15.913915094339622,
      "grad_norm": 3.751547336578369,
      "learning_rate": 1.2188332188991493e-06,
      "loss": 0.3074,
      "num_input_tokens_seen": 17772744,
      "step": 26990
    },
    {
      "epoch": 15.91686320754717,
      "grad_norm": 5.2963762283325195,
      "learning_rate": 1.217150387518804e-06,
      "loss": 0.3461,
      "num_input_tokens_seen": 17776264,
      "step": 26995
    },
    {
      "epoch": 15.919811320754716,
      "grad_norm": 4.231935977935791,
      "learning_rate": 1.2154685576641967e-06,
      "loss": 0.2916,
      "num_input_tokens_seen": 17779464,
      "step": 27000
    },
    {
      "epoch": 15.922759433962264,
      "grad_norm": 2.703111171722412,
      "learning_rate": 1.2137877297805972e-06,
      "loss": 0.2223,
      "num_input_tokens_seen": 17782664,
      "step": 27005
    },
    {
      "epoch": 15.92570754716981,
      "grad_norm": 2.7506067752838135,
      "learning_rate": 1.2121079043130162e-06,
      "loss": 0.2116,
      "num_input_tokens_seen": 17785320,
      "step": 27010
    },
    {
      "epoch": 15.928655660377359,
      "grad_norm": 10.583368301391602,
      "learning_rate": 1.210429081706192e-06,
      "loss": 0.3258,
      "num_input_tokens_seen": 17788648,
      "step": 27015
    },
    {
      "epoch": 15.931603773584905,
      "grad_norm": 3.904362678527832,
      "learning_rate": 1.2087512624046005e-06,
      "loss": 0.3654,
      "num_input_tokens_seen": 17792456,
      "step": 27020
    },
    {
      "epoch": 15.934551886792454,
      "grad_norm": 4.202816009521484,
      "learning_rate": 1.2070744468524503e-06,
      "loss": 0.2825,
      "num_input_tokens_seen": 17795400,
      "step": 27025
    },
    {
      "epoch": 15.9375,
      "grad_norm": 3.770707368850708,
      "learning_rate": 1.2053986354936887e-06,
      "loss": 0.4297,
      "num_input_tokens_seen": 17799176,
      "step": 27030
    },
    {
      "epoch": 15.940448113207546,
      "grad_norm": 2.7429051399230957,
      "learning_rate": 1.2037238287719916e-06,
      "loss": 0.3196,
      "num_input_tokens_seen": 17802664,
      "step": 27035
    },
    {
      "epoch": 15.943396226415095,
      "grad_norm": 4.248921871185303,
      "learning_rate": 1.2020500271307721e-06,
      "loss": 0.2412,
      "num_input_tokens_seen": 17805576,
      "step": 27040
    },
    {
      "epoch": 15.946344339622641,
      "grad_norm": 3.903898239135742,
      "learning_rate": 1.200377231013176e-06,
      "loss": 0.2979,
      "num_input_tokens_seen": 17808904,
      "step": 27045
    },
    {
      "epoch": 15.94929245283019,
      "grad_norm": 4.91021728515625,
      "learning_rate": 1.1987054408620825e-06,
      "loss": 0.2812,
      "num_input_tokens_seen": 17812008,
      "step": 27050
    },
    {
      "epoch": 15.952240566037736,
      "grad_norm": 3.711174488067627,
      "learning_rate": 1.197034657120107e-06,
      "loss": 0.2554,
      "num_input_tokens_seen": 17815688,
      "step": 27055
    },
    {
      "epoch": 15.955188679245284,
      "grad_norm": 5.967428207397461,
      "learning_rate": 1.1953648802295964e-06,
      "loss": 0.2937,
      "num_input_tokens_seen": 17819176,
      "step": 27060
    },
    {
      "epoch": 15.95813679245283,
      "grad_norm": 3.905522584915161,
      "learning_rate": 1.1936961106326307e-06,
      "loss": 0.369,
      "num_input_tokens_seen": 17822280,
      "step": 27065
    },
    {
      "epoch": 15.961084905660378,
      "grad_norm": 4.5226216316223145,
      "learning_rate": 1.1920283487710237e-06,
      "loss": 0.3169,
      "num_input_tokens_seen": 17825288,
      "step": 27070
    },
    {
      "epoch": 15.964033018867925,
      "grad_norm": 2.4400792121887207,
      "learning_rate": 1.1903615950863228e-06,
      "loss": 0.2662,
      "num_input_tokens_seen": 17827976,
      "step": 27075
    },
    {
      "epoch": 15.966981132075471,
      "grad_norm": 3.416958808898926,
      "learning_rate": 1.1886958500198076e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 17831016,
      "step": 27080
    },
    {
      "epoch": 15.96992924528302,
      "grad_norm": 4.157408237457275,
      "learning_rate": 1.1870311140124923e-06,
      "loss": 0.254,
      "num_input_tokens_seen": 17833896,
      "step": 27085
    },
    {
      "epoch": 15.972877358490566,
      "grad_norm": 4.612361907958984,
      "learning_rate": 1.185367387505123e-06,
      "loss": 0.2919,
      "num_input_tokens_seen": 17836616,
      "step": 27090
    },
    {
      "epoch": 15.975825471698114,
      "grad_norm": 4.490530490875244,
      "learning_rate": 1.1837046709381783e-06,
      "loss": 0.3475,
      "num_input_tokens_seen": 17839592,
      "step": 27095
    },
    {
      "epoch": 15.97877358490566,
      "grad_norm": 3.806129217147827,
      "learning_rate": 1.1820429647518678e-06,
      "loss": 0.407,
      "num_input_tokens_seen": 17842440,
      "step": 27100
    },
    {
      "epoch": 15.981721698113208,
      "grad_norm": 3.4101455211639404,
      "learning_rate": 1.1803822693861377e-06,
      "loss": 0.3028,
      "num_input_tokens_seen": 17846344,
      "step": 27105
    },
    {
      "epoch": 15.984669811320755,
      "grad_norm": 4.038768291473389,
      "learning_rate": 1.1787225852806639e-06,
      "loss": 0.4727,
      "num_input_tokens_seen": 17849768,
      "step": 27110
    },
    {
      "epoch": 15.987617924528301,
      "grad_norm": 3.708556890487671,
      "learning_rate": 1.177063912874853e-06,
      "loss": 0.2607,
      "num_input_tokens_seen": 17852872,
      "step": 27115
    },
    {
      "epoch": 15.99056603773585,
      "grad_norm": 4.013854503631592,
      "learning_rate": 1.1754062526078487e-06,
      "loss": 0.255,
      "num_input_tokens_seen": 17856072,
      "step": 27120
    },
    {
      "epoch": 15.993514150943396,
      "grad_norm": 3.5344231128692627,
      "learning_rate": 1.1737496049185215e-06,
      "loss": 0.376,
      "num_input_tokens_seen": 17859336,
      "step": 27125
    },
    {
      "epoch": 15.996462264150944,
      "grad_norm": 2.477252960205078,
      "learning_rate": 1.172093970245477e-06,
      "loss": 0.335,
      "num_input_tokens_seen": 17863720,
      "step": 27130
    },
    {
      "epoch": 15.99941037735849,
      "grad_norm": 3.5357959270477295,
      "learning_rate": 1.1704393490270516e-06,
      "loss": 0.2389,
      "num_input_tokens_seen": 17866952,
      "step": 27135
    },
    {
      "epoch": 16.0,
      "eval_loss": 0.5845065712928772,
      "eval_runtime": 18.7393,
      "eval_samples_per_second": 90.505,
      "eval_steps_per_second": 22.626,
      "num_input_tokens_seen": 17867032,
      "step": 27136
    },
    {
      "epoch": 16.00235849056604,
      "grad_norm": 3.7729625701904297,
      "learning_rate": 1.1687857417013126e-06,
      "loss": 0.3395,
      "num_input_tokens_seen": 17868728,
      "step": 27140
    },
    {
      "epoch": 16.005306603773583,
      "grad_norm": 3.126826763153076,
      "learning_rate": 1.1671331487060583e-06,
      "loss": 0.3546,
      "num_input_tokens_seen": 17872984,
      "step": 27145
    },
    {
      "epoch": 16.00825471698113,
      "grad_norm": 3.1340529918670654,
      "learning_rate": 1.1654815704788237e-06,
      "loss": 0.4194,
      "num_input_tokens_seen": 17876408,
      "step": 27150
    },
    {
      "epoch": 16.01120283018868,
      "grad_norm": 2.152880907058716,
      "learning_rate": 1.1638310074568687e-06,
      "loss": 0.2686,
      "num_input_tokens_seen": 17880152,
      "step": 27155
    },
    {
      "epoch": 16.014150943396228,
      "grad_norm": 3.9543418884277344,
      "learning_rate": 1.162181460077188e-06,
      "loss": 0.289,
      "num_input_tokens_seen": 17883384,
      "step": 27160
    },
    {
      "epoch": 16.017099056603772,
      "grad_norm": 6.424722194671631,
      "learning_rate": 1.1605329287765056e-06,
      "loss": 0.3603,
      "num_input_tokens_seen": 17886072,
      "step": 27165
    },
    {
      "epoch": 16.02004716981132,
      "grad_norm": 4.0223164558410645,
      "learning_rate": 1.1588854139912775e-06,
      "loss": 0.3244,
      "num_input_tokens_seen": 17888920,
      "step": 27170
    },
    {
      "epoch": 16.02299528301887,
      "grad_norm": 3.346309185028076,
      "learning_rate": 1.1572389161576886e-06,
      "loss": 0.2864,
      "num_input_tokens_seen": 17891768,
      "step": 27175
    },
    {
      "epoch": 16.025943396226417,
      "grad_norm": 3.7274253368377686,
      "learning_rate": 1.15559343571166e-06,
      "loss": 0.2625,
      "num_input_tokens_seen": 17894264,
      "step": 27180
    },
    {
      "epoch": 16.02889150943396,
      "grad_norm": 3.021092414855957,
      "learning_rate": 1.153948973088837e-06,
      "loss": 0.2994,
      "num_input_tokens_seen": 17897752,
      "step": 27185
    },
    {
      "epoch": 16.03183962264151,
      "grad_norm": 4.4075517654418945,
      "learning_rate": 1.1523055287245993e-06,
      "loss": 0.2649,
      "num_input_tokens_seen": 17900600,
      "step": 27190
    },
    {
      "epoch": 16.034787735849058,
      "grad_norm": 2.3155007362365723,
      "learning_rate": 1.150663103054056e-06,
      "loss": 0.4161,
      "num_input_tokens_seen": 17903512,
      "step": 27195
    },
    {
      "epoch": 16.037735849056602,
      "grad_norm": 2.7349088191986084,
      "learning_rate": 1.1490216965120438e-06,
      "loss": 0.2875,
      "num_input_tokens_seen": 17906776,
      "step": 27200
    },
    {
      "epoch": 16.04068396226415,
      "grad_norm": 2.879483699798584,
      "learning_rate": 1.147381309533136e-06,
      "loss": 0.3543,
      "num_input_tokens_seen": 17909656,
      "step": 27205
    },
    {
      "epoch": 16.0436320754717,
      "grad_norm": 2.98820424079895,
      "learning_rate": 1.1457419425516287e-06,
      "loss": 0.3684,
      "num_input_tokens_seen": 17912344,
      "step": 27210
    },
    {
      "epoch": 16.046580188679247,
      "grad_norm": 4.034277439117432,
      "learning_rate": 1.1441035960015544e-06,
      "loss": 0.4083,
      "num_input_tokens_seen": 17916024,
      "step": 27215
    },
    {
      "epoch": 16.04952830188679,
      "grad_norm": 5.40632963180542,
      "learning_rate": 1.1424662703166716e-06,
      "loss": 0.3038,
      "num_input_tokens_seen": 17919032,
      "step": 27220
    },
    {
      "epoch": 16.05247641509434,
      "grad_norm": 3.3359272480010986,
      "learning_rate": 1.1408299659304684e-06,
      "loss": 0.318,
      "num_input_tokens_seen": 17922648,
      "step": 27225
    },
    {
      "epoch": 16.055424528301888,
      "grad_norm": 3.5588250160217285,
      "learning_rate": 1.1391946832761642e-06,
      "loss": 0.3252,
      "num_input_tokens_seen": 17925720,
      "step": 27230
    },
    {
      "epoch": 16.058372641509433,
      "grad_norm": 3.4008805751800537,
      "learning_rate": 1.137560422786706e-06,
      "loss": 0.2592,
      "num_input_tokens_seen": 17928440,
      "step": 27235
    },
    {
      "epoch": 16.06132075471698,
      "grad_norm": 1.5024508237838745,
      "learning_rate": 1.1359271848947712e-06,
      "loss": 0.3138,
      "num_input_tokens_seen": 17932216,
      "step": 27240
    },
    {
      "epoch": 16.06426886792453,
      "grad_norm": 3.5096805095672607,
      "learning_rate": 1.1342949700327688e-06,
      "loss": 0.2315,
      "num_input_tokens_seen": 17935480,
      "step": 27245
    },
    {
      "epoch": 16.067216981132077,
      "grad_norm": 4.720620155334473,
      "learning_rate": 1.1326637786328332e-06,
      "loss": 0.2512,
      "num_input_tokens_seen": 17939096,
      "step": 27250
    },
    {
      "epoch": 16.07016509433962,
      "grad_norm": 4.8753662109375,
      "learning_rate": 1.1310336111268293e-06,
      "loss": 0.3076,
      "num_input_tokens_seen": 17941944,
      "step": 27255
    },
    {
      "epoch": 16.07311320754717,
      "grad_norm": 2.8330440521240234,
      "learning_rate": 1.1294044679463517e-06,
      "loss": 0.3169,
      "num_input_tokens_seen": 17945816,
      "step": 27260
    },
    {
      "epoch": 16.076061320754718,
      "grad_norm": 2.46581768989563,
      "learning_rate": 1.1277763495227207e-06,
      "loss": 0.3028,
      "num_input_tokens_seen": 17948984,
      "step": 27265
    },
    {
      "epoch": 16.079009433962263,
      "grad_norm": 4.411411762237549,
      "learning_rate": 1.1261492562869913e-06,
      "loss": 0.2857,
      "num_input_tokens_seen": 17951960,
      "step": 27270
    },
    {
      "epoch": 16.08195754716981,
      "grad_norm": 5.904330253601074,
      "learning_rate": 1.1245231886699415e-06,
      "loss": 0.2725,
      "num_input_tokens_seen": 17954712,
      "step": 27275
    },
    {
      "epoch": 16.08490566037736,
      "grad_norm": 3.7460391521453857,
      "learning_rate": 1.12289814710208e-06,
      "loss": 0.4063,
      "num_input_tokens_seen": 17958424,
      "step": 27280
    },
    {
      "epoch": 16.087853773584907,
      "grad_norm": 7.076168060302734,
      "learning_rate": 1.1212741320136433e-06,
      "loss": 0.4015,
      "num_input_tokens_seen": 17961880,
      "step": 27285
    },
    {
      "epoch": 16.090801886792452,
      "grad_norm": 3.2214839458465576,
      "learning_rate": 1.1196511438345963e-06,
      "loss": 0.4246,
      "num_input_tokens_seen": 17964856,
      "step": 27290
    },
    {
      "epoch": 16.09375,
      "grad_norm": 3.583343982696533,
      "learning_rate": 1.118029182994631e-06,
      "loss": 0.2875,
      "num_input_tokens_seen": 17967384,
      "step": 27295
    },
    {
      "epoch": 16.096698113207548,
      "grad_norm": 4.860898494720459,
      "learning_rate": 1.1164082499231704e-06,
      "loss": 0.5074,
      "num_input_tokens_seen": 17970136,
      "step": 27300
    },
    {
      "epoch": 16.099646226415093,
      "grad_norm": 4.807193756103516,
      "learning_rate": 1.114788345049364e-06,
      "loss": 0.2891,
      "num_input_tokens_seen": 17973784,
      "step": 27305
    },
    {
      "epoch": 16.10259433962264,
      "grad_norm": 4.213859558105469,
      "learning_rate": 1.1131694688020872e-06,
      "loss": 0.1839,
      "num_input_tokens_seen": 17976568,
      "step": 27310
    },
    {
      "epoch": 16.10554245283019,
      "grad_norm": 4.234954833984375,
      "learning_rate": 1.1115516216099453e-06,
      "loss": 0.1612,
      "num_input_tokens_seen": 17979224,
      "step": 27315
    },
    {
      "epoch": 16.108490566037737,
      "grad_norm": 2.4744150638580322,
      "learning_rate": 1.1099348039012698e-06,
      "loss": 0.2355,
      "num_input_tokens_seen": 17983800,
      "step": 27320
    },
    {
      "epoch": 16.111438679245282,
      "grad_norm": 4.440317153930664,
      "learning_rate": 1.1083190161041202e-06,
      "loss": 0.4102,
      "num_input_tokens_seen": 17987128,
      "step": 27325
    },
    {
      "epoch": 16.11438679245283,
      "grad_norm": 4.13766622543335,
      "learning_rate": 1.1067042586462822e-06,
      "loss": 0.2256,
      "num_input_tokens_seen": 17990776,
      "step": 27330
    },
    {
      "epoch": 16.11733490566038,
      "grad_norm": 3.6878440380096436,
      "learning_rate": 1.1050905319552718e-06,
      "loss": 0.36,
      "num_input_tokens_seen": 17993496,
      "step": 27335
    },
    {
      "epoch": 16.120283018867923,
      "grad_norm": 5.9211320877075195,
      "learning_rate": 1.1034778364583293e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 17996088,
      "step": 27340
    },
    {
      "epoch": 16.12323113207547,
      "grad_norm": 7.335329532623291,
      "learning_rate": 1.1018661725824231e-06,
      "loss": 0.3209,
      "num_input_tokens_seen": 17999096,
      "step": 27345
    },
    {
      "epoch": 16.12617924528302,
      "grad_norm": 3.0737504959106445,
      "learning_rate": 1.100255540754247e-06,
      "loss": 0.4202,
      "num_input_tokens_seen": 18002296,
      "step": 27350
    },
    {
      "epoch": 16.129127358490567,
      "grad_norm": 3.8166749477386475,
      "learning_rate": 1.0986459414002244e-06,
      "loss": 0.3024,
      "num_input_tokens_seen": 18006584,
      "step": 27355
    },
    {
      "epoch": 16.132075471698112,
      "grad_norm": 3.714792490005493,
      "learning_rate": 1.0970373749465008e-06,
      "loss": 0.3106,
      "num_input_tokens_seen": 18009304,
      "step": 27360
    },
    {
      "epoch": 16.13502358490566,
      "grad_norm": 4.089264869689941,
      "learning_rate": 1.095429841818954e-06,
      "loss": 0.5085,
      "num_input_tokens_seen": 18012440,
      "step": 27365
    },
    {
      "epoch": 16.13797169811321,
      "grad_norm": 3.9846575260162354,
      "learning_rate": 1.093823342443185e-06,
      "loss": 0.203,
      "num_input_tokens_seen": 18015352,
      "step": 27370
    },
    {
      "epoch": 16.140919811320753,
      "grad_norm": 3.2786312103271484,
      "learning_rate": 1.0922178772445203e-06,
      "loss": 0.3381,
      "num_input_tokens_seen": 18018776,
      "step": 27375
    },
    {
      "epoch": 16.1438679245283,
      "grad_norm": 5.269048690795898,
      "learning_rate": 1.0906134466480146e-06,
      "loss": 0.27,
      "num_input_tokens_seen": 18021976,
      "step": 27380
    },
    {
      "epoch": 16.14681603773585,
      "grad_norm": 4.228376865386963,
      "learning_rate": 1.0890100510784473e-06,
      "loss": 0.2429,
      "num_input_tokens_seen": 18025432,
      "step": 27385
    },
    {
      "epoch": 16.149764150943398,
      "grad_norm": 5.043367385864258,
      "learning_rate": 1.0874076909603227e-06,
      "loss": 0.3018,
      "num_input_tokens_seen": 18029432,
      "step": 27390
    },
    {
      "epoch": 16.152712264150942,
      "grad_norm": 3.6022918224334717,
      "learning_rate": 1.0858063667178747e-06,
      "loss": 0.2821,
      "num_input_tokens_seen": 18032696,
      "step": 27395
    },
    {
      "epoch": 16.15566037735849,
      "grad_norm": 2.8313167095184326,
      "learning_rate": 1.0842060787750614e-06,
      "loss": 0.3381,
      "num_input_tokens_seen": 18035800,
      "step": 27400
    },
    {
      "epoch": 16.15860849056604,
      "grad_norm": 2.796313762664795,
      "learning_rate": 1.0826068275555652e-06,
      "loss": 0.2692,
      "num_input_tokens_seen": 18038904,
      "step": 27405
    },
    {
      "epoch": 16.161556603773583,
      "grad_norm": 6.844527721405029,
      "learning_rate": 1.081008613482794e-06,
      "loss": 0.3032,
      "num_input_tokens_seen": 18042488,
      "step": 27410
    },
    {
      "epoch": 16.16450471698113,
      "grad_norm": 2.964731454849243,
      "learning_rate": 1.079411436979883e-06,
      "loss": 0.2949,
      "num_input_tokens_seen": 18046136,
      "step": 27415
    },
    {
      "epoch": 16.16745283018868,
      "grad_norm": 3.4290027618408203,
      "learning_rate": 1.0778152984696905e-06,
      "loss": 0.3068,
      "num_input_tokens_seen": 18049144,
      "step": 27420
    },
    {
      "epoch": 16.170400943396228,
      "grad_norm": 3.3122148513793945,
      "learning_rate": 1.0762201983747993e-06,
      "loss": 0.3206,
      "num_input_tokens_seen": 18052760,
      "step": 27425
    },
    {
      "epoch": 16.173349056603772,
      "grad_norm": 19.445199966430664,
      "learning_rate": 1.0746261371175238e-06,
      "loss": 0.2272,
      "num_input_tokens_seen": 18055576,
      "step": 27430
    },
    {
      "epoch": 16.17629716981132,
      "grad_norm": 6.077767848968506,
      "learning_rate": 1.0730331151198953e-06,
      "loss": 0.3295,
      "num_input_tokens_seen": 18058968,
      "step": 27435
    },
    {
      "epoch": 16.17924528301887,
      "grad_norm": 3.2576231956481934,
      "learning_rate": 1.0714411328036733e-06,
      "loss": 0.2993,
      "num_input_tokens_seen": 18061944,
      "step": 27440
    },
    {
      "epoch": 16.182193396226417,
      "grad_norm": 2.896879196166992,
      "learning_rate": 1.0698501905903435e-06,
      "loss": 0.2261,
      "num_input_tokens_seen": 18065752,
      "step": 27445
    },
    {
      "epoch": 16.18514150943396,
      "grad_norm": 4.278557777404785,
      "learning_rate": 1.0682602889011134e-06,
      "loss": 0.3262,
      "num_input_tokens_seen": 18068376,
      "step": 27450
    },
    {
      "epoch": 16.18808962264151,
      "grad_norm": 2.820657968521118,
      "learning_rate": 1.0666714281569152e-06,
      "loss": 0.1967,
      "num_input_tokens_seen": 18073624,
      "step": 27455
    },
    {
      "epoch": 16.191037735849058,
      "grad_norm": 4.710768222808838,
      "learning_rate": 1.0650836087784095e-06,
      "loss": 0.332,
      "num_input_tokens_seen": 18077560,
      "step": 27460
    },
    {
      "epoch": 16.193985849056602,
      "grad_norm": 3.5744786262512207,
      "learning_rate": 1.0634968311859768e-06,
      "loss": 0.3759,
      "num_input_tokens_seen": 18081400,
      "step": 27465
    },
    {
      "epoch": 16.19693396226415,
      "grad_norm": 3.625427007675171,
      "learning_rate": 1.0619110957997237e-06,
      "loss": 0.3733,
      "num_input_tokens_seen": 18084952,
      "step": 27470
    },
    {
      "epoch": 16.1998820754717,
      "grad_norm": 2.8784661293029785,
      "learning_rate": 1.06032640303948e-06,
      "loss": 0.2352,
      "num_input_tokens_seen": 18088696,
      "step": 27475
    },
    {
      "epoch": 16.202830188679247,
      "grad_norm": 5.323342323303223,
      "learning_rate": 1.0587427533248002e-06,
      "loss": 0.2857,
      "num_input_tokens_seen": 18092376,
      "step": 27480
    },
    {
      "epoch": 16.20577830188679,
      "grad_norm": 3.046125650405884,
      "learning_rate": 1.057160147074961e-06,
      "loss": 0.2259,
      "num_input_tokens_seen": 18095416,
      "step": 27485
    },
    {
      "epoch": 16.20872641509434,
      "grad_norm": 2.4502177238464355,
      "learning_rate": 1.0555785847089657e-06,
      "loss": 0.2687,
      "num_input_tokens_seen": 18099032,
      "step": 27490
    },
    {
      "epoch": 16.211674528301888,
      "grad_norm": 5.955513954162598,
      "learning_rate": 1.0539980666455407e-06,
      "loss": 0.2523,
      "num_input_tokens_seen": 18102104,
      "step": 27495
    },
    {
      "epoch": 16.214622641509433,
      "grad_norm": 3.6034934520721436,
      "learning_rate": 1.052418593303134e-06,
      "loss": 0.284,
      "num_input_tokens_seen": 18106712,
      "step": 27500
    },
    {
      "epoch": 16.21757075471698,
      "grad_norm": 4.636280536651611,
      "learning_rate": 1.0508401650999178e-06,
      "loss": 0.2114,
      "num_input_tokens_seen": 18110488,
      "step": 27505
    },
    {
      "epoch": 16.22051886792453,
      "grad_norm": 1.7119940519332886,
      "learning_rate": 1.0492627824537877e-06,
      "loss": 0.234,
      "num_input_tokens_seen": 18113560,
      "step": 27510
    },
    {
      "epoch": 16.223466981132077,
      "grad_norm": 5.418909549713135,
      "learning_rate": 1.0476864457823626e-06,
      "loss": 0.3532,
      "num_input_tokens_seen": 18117240,
      "step": 27515
    },
    {
      "epoch": 16.22641509433962,
      "grad_norm": 3.000133752822876,
      "learning_rate": 1.0461111555029836e-06,
      "loss": 0.2024,
      "num_input_tokens_seen": 18120408,
      "step": 27520
    },
    {
      "epoch": 16.22936320754717,
      "grad_norm": 3.707765579223633,
      "learning_rate": 1.0445369120327175e-06,
      "loss": 0.398,
      "num_input_tokens_seen": 18124824,
      "step": 27525
    },
    {
      "epoch": 16.232311320754718,
      "grad_norm": 2.66023850440979,
      "learning_rate": 1.0429637157883516e-06,
      "loss": 0.2983,
      "num_input_tokens_seen": 18127544,
      "step": 27530
    },
    {
      "epoch": 16.235259433962263,
      "grad_norm": 3.0265307426452637,
      "learning_rate": 1.041391567186395e-06,
      "loss": 0.2387,
      "num_input_tokens_seen": 18130936,
      "step": 27535
    },
    {
      "epoch": 16.23820754716981,
      "grad_norm": 5.656672477722168,
      "learning_rate": 1.0398204666430821e-06,
      "loss": 0.3117,
      "num_input_tokens_seen": 18134104,
      "step": 27540
    },
    {
      "epoch": 16.24115566037736,
      "grad_norm": 4.254795551300049,
      "learning_rate": 1.0382504145743667e-06,
      "loss": 0.3714,
      "num_input_tokens_seen": 18137016,
      "step": 27545
    },
    {
      "epoch": 16.244103773584907,
      "grad_norm": 3.9661431312561035,
      "learning_rate": 1.0366814113959294e-06,
      "loss": 0.361,
      "num_input_tokens_seen": 18140376,
      "step": 27550
    },
    {
      "epoch": 16.247051886792452,
      "grad_norm": 3.494716167449951,
      "learning_rate": 1.0351134575231697e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 18142968,
      "step": 27555
    },
    {
      "epoch": 16.25,
      "grad_norm": 3.895977258682251,
      "learning_rate": 1.0335465533712098e-06,
      "loss": 0.3326,
      "num_input_tokens_seen": 18146072,
      "step": 27560
    },
    {
      "epoch": 16.252948113207548,
      "grad_norm": 4.488941192626953,
      "learning_rate": 1.031980699354894e-06,
      "loss": 0.2529,
      "num_input_tokens_seen": 18149432,
      "step": 27565
    },
    {
      "epoch": 16.255896226415093,
      "grad_norm": 3.3819167613983154,
      "learning_rate": 1.03041589588879e-06,
      "loss": 0.3099,
      "num_input_tokens_seen": 18152600,
      "step": 27570
    },
    {
      "epoch": 16.25884433962264,
      "grad_norm": 2.952915906906128,
      "learning_rate": 1.0288521433871834e-06,
      "loss": 0.2692,
      "num_input_tokens_seen": 18156024,
      "step": 27575
    },
    {
      "epoch": 16.26179245283019,
      "grad_norm": 6.3623833656311035,
      "learning_rate": 1.0272894422640866e-06,
      "loss": 0.2361,
      "num_input_tokens_seen": 18158904,
      "step": 27580
    },
    {
      "epoch": 16.264740566037737,
      "grad_norm": 3.325237512588501,
      "learning_rate": 1.0257277929332332e-06,
      "loss": 0.4583,
      "num_input_tokens_seen": 18162136,
      "step": 27585
    },
    {
      "epoch": 16.267688679245282,
      "grad_norm": 4.5835394859313965,
      "learning_rate": 1.0241671958080745e-06,
      "loss": 0.2312,
      "num_input_tokens_seen": 18164888,
      "step": 27590
    },
    {
      "epoch": 16.27063679245283,
      "grad_norm": 3.3347957134246826,
      "learning_rate": 1.0226076513017858e-06,
      "loss": 0.3288,
      "num_input_tokens_seen": 18168440,
      "step": 27595
    },
    {
      "epoch": 16.27358490566038,
      "grad_norm": 5.125063896179199,
      "learning_rate": 1.0210491598272625e-06,
      "loss": 0.3015,
      "num_input_tokens_seen": 18172216,
      "step": 27600
    },
    {
      "epoch": 16.276533018867923,
      "grad_norm": 4.404242992401123,
      "learning_rate": 1.0194917217971229e-06,
      "loss": 0.3318,
      "num_input_tokens_seen": 18176344,
      "step": 27605
    },
    {
      "epoch": 16.27948113207547,
      "grad_norm": 2.3488943576812744,
      "learning_rate": 1.0179353376237038e-06,
      "loss": 0.2806,
      "num_input_tokens_seen": 18179704,
      "step": 27610
    },
    {
      "epoch": 16.28242924528302,
      "grad_norm": 8.276045799255371,
      "learning_rate": 1.0163800077190672e-06,
      "loss": 0.5186,
      "num_input_tokens_seen": 18182584,
      "step": 27615
    },
    {
      "epoch": 16.285377358490567,
      "grad_norm": 2.587984323501587,
      "learning_rate": 1.0148257324949916e-06,
      "loss": 0.3226,
      "num_input_tokens_seen": 18186712,
      "step": 27620
    },
    {
      "epoch": 16.288325471698112,
      "grad_norm": 5.912843227386475,
      "learning_rate": 1.0132725123629783e-06,
      "loss": 0.3296,
      "num_input_tokens_seen": 18189848,
      "step": 27625
    },
    {
      "epoch": 16.29127358490566,
      "grad_norm": 3.7596449851989746,
      "learning_rate": 1.0117203477342497e-06,
      "loss": 0.4389,
      "num_input_tokens_seen": 18192664,
      "step": 27630
    },
    {
      "epoch": 16.29422169811321,
      "grad_norm": 4.719682216644287,
      "learning_rate": 1.0101692390197477e-06,
      "loss": 0.2543,
      "num_input_tokens_seen": 18195832,
      "step": 27635
    },
    {
      "epoch": 16.297169811320753,
      "grad_norm": 2.9655518531799316,
      "learning_rate": 1.0086191866301331e-06,
      "loss": 0.3331,
      "num_input_tokens_seen": 18199032,
      "step": 27640
    },
    {
      "epoch": 16.3001179245283,
      "grad_norm": 2.4002327919006348,
      "learning_rate": 1.0070701909757918e-06,
      "loss": 0.2416,
      "num_input_tokens_seen": 18202456,
      "step": 27645
    },
    {
      "epoch": 16.30306603773585,
      "grad_norm": 5.256389617919922,
      "learning_rate": 1.0055222524668267e-06,
      "loss": 0.3351,
      "num_input_tokens_seen": 18206296,
      "step": 27650
    },
    {
      "epoch": 16.306014150943398,
      "grad_norm": 3.0611696243286133,
      "learning_rate": 1.00397537151306e-06,
      "loss": 0.3665,
      "num_input_tokens_seen": 18210168,
      "step": 27655
    },
    {
      "epoch": 16.308962264150942,
      "grad_norm": 3.6888086795806885,
      "learning_rate": 1.002429548524036e-06,
      "loss": 0.3292,
      "num_input_tokens_seen": 18213016,
      "step": 27660
    },
    {
      "epoch": 16.31191037735849,
      "grad_norm": 3.6992480754852295,
      "learning_rate": 1.0008847839090175e-06,
      "loss": 0.4002,
      "num_input_tokens_seen": 18215608,
      "step": 27665
    },
    {
      "epoch": 16.31485849056604,
      "grad_norm": 5.334972858428955,
      "learning_rate": 9.993410780769862e-07,
      "loss": 0.3133,
      "num_input_tokens_seen": 18218840,
      "step": 27670
    },
    {
      "epoch": 16.317806603773583,
      "grad_norm": 4.137313365936279,
      "learning_rate": 9.977984314366463e-07,
      "loss": 0.2918,
      "num_input_tokens_seen": 18221368,
      "step": 27675
    },
    {
      "epoch": 16.32075471698113,
      "grad_norm": 4.698786735534668,
      "learning_rate": 9.962568443964216e-07,
      "loss": 0.3002,
      "num_input_tokens_seen": 18224984,
      "step": 27680
    },
    {
      "epoch": 16.32370283018868,
      "grad_norm": 4.36729621887207,
      "learning_rate": 9.947163173644524e-07,
      "loss": 0.4216,
      "num_input_tokens_seen": 18228152,
      "step": 27685
    },
    {
      "epoch": 16.326650943396228,
      "grad_norm": 4.332581520080566,
      "learning_rate": 9.931768507486007e-07,
      "loss": 0.2827,
      "num_input_tokens_seen": 18231256,
      "step": 27690
    },
    {
      "epoch": 16.329599056603772,
      "grad_norm": 3.574216604232788,
      "learning_rate": 9.916384449564453e-07,
      "loss": 0.2464,
      "num_input_tokens_seen": 18234200,
      "step": 27695
    },
    {
      "epoch": 16.33254716981132,
      "grad_norm": 3.9117281436920166,
      "learning_rate": 9.90101100395287e-07,
      "loss": 0.3234,
      "num_input_tokens_seen": 18237368,
      "step": 27700
    },
    {
      "epoch": 16.33549528301887,
      "grad_norm": 2.3664724826812744,
      "learning_rate": 9.885648174721428e-07,
      "loss": 0.3815,
      "num_input_tokens_seen": 18241080,
      "step": 27705
    },
    {
      "epoch": 16.338443396226417,
      "grad_norm": 4.250330448150635,
      "learning_rate": 9.870295965937532e-07,
      "loss": 0.2977,
      "num_input_tokens_seen": 18243512,
      "step": 27710
    },
    {
      "epoch": 16.34139150943396,
      "grad_norm": 4.738950729370117,
      "learning_rate": 9.854954381665727e-07,
      "loss": 0.3667,
      "num_input_tokens_seen": 18247000,
      "step": 27715
    },
    {
      "epoch": 16.34433962264151,
      "grad_norm": 2.0135891437530518,
      "learning_rate": 9.83962342596776e-07,
      "loss": 0.2263,
      "num_input_tokens_seen": 18250840,
      "step": 27720
    },
    {
      "epoch": 16.347287735849058,
      "grad_norm": 4.174996852874756,
      "learning_rate": 9.824303102902576e-07,
      "loss": 0.3732,
      "num_input_tokens_seen": 18255000,
      "step": 27725
    },
    {
      "epoch": 16.350235849056602,
      "grad_norm": 4.054419040679932,
      "learning_rate": 9.808993416526292e-07,
      "loss": 0.2576,
      "num_input_tokens_seen": 18257528,
      "step": 27730
    },
    {
      "epoch": 16.35318396226415,
      "grad_norm": 3.4507157802581787,
      "learning_rate": 9.793694370892204e-07,
      "loss": 0.4106,
      "num_input_tokens_seen": 18261816,
      "step": 27735
    },
    {
      "epoch": 16.3561320754717,
      "grad_norm": 3.023228645324707,
      "learning_rate": 9.77840597005082e-07,
      "loss": 0.2587,
      "num_input_tokens_seen": 18265976,
      "step": 27740
    },
    {
      "epoch": 16.359080188679247,
      "grad_norm": 4.151533603668213,
      "learning_rate": 9.763128218049806e-07,
      "loss": 0.3032,
      "num_input_tokens_seen": 18268920,
      "step": 27745
    },
    {
      "epoch": 16.36202830188679,
      "grad_norm": 2.803666353225708,
      "learning_rate": 9.747861118934005e-07,
      "loss": 0.2935,
      "num_input_tokens_seen": 18271640,
      "step": 27750
    },
    {
      "epoch": 16.36497641509434,
      "grad_norm": 3.344740867614746,
      "learning_rate": 9.732604676745443e-07,
      "loss": 0.3236,
      "num_input_tokens_seen": 18274936,
      "step": 27755
    },
    {
      "epoch": 16.367924528301888,
      "grad_norm": 4.895694732666016,
      "learning_rate": 9.717358895523333e-07,
      "loss": 0.253,
      "num_input_tokens_seen": 18278232,
      "step": 27760
    },
    {
      "epoch": 16.370872641509433,
      "grad_norm": 3.3425962924957275,
      "learning_rate": 9.702123779304074e-07,
      "loss": 0.3291,
      "num_input_tokens_seen": 18281016,
      "step": 27765
    },
    {
      "epoch": 16.37382075471698,
      "grad_norm": 6.471497058868408,
      "learning_rate": 9.686899332121203e-07,
      "loss": 0.3134,
      "num_input_tokens_seen": 18283736,
      "step": 27770
    },
    {
      "epoch": 16.37676886792453,
      "grad_norm": 4.920362949371338,
      "learning_rate": 9.671685558005488e-07,
      "loss": 0.3354,
      "num_input_tokens_seen": 18286712,
      "step": 27775
    },
    {
      "epoch": 16.379716981132077,
      "grad_norm": 4.27953577041626,
      "learning_rate": 9.656482460984828e-07,
      "loss": 0.2761,
      "num_input_tokens_seen": 18291544,
      "step": 27780
    },
    {
      "epoch": 16.38266509433962,
      "grad_norm": 4.572440147399902,
      "learning_rate": 9.641290045084307e-07,
      "loss": 0.2704,
      "num_input_tokens_seen": 18296376,
      "step": 27785
    },
    {
      "epoch": 16.38561320754717,
      "grad_norm": 3.313114643096924,
      "learning_rate": 9.626108314326182e-07,
      "loss": 0.3555,
      "num_input_tokens_seen": 18299320,
      "step": 27790
    },
    {
      "epoch": 16.388561320754718,
      "grad_norm": 2.3730883598327637,
      "learning_rate": 9.610937272729881e-07,
      "loss": 0.2833,
      "num_input_tokens_seen": 18302616,
      "step": 27795
    },
    {
      "epoch": 16.391509433962263,
      "grad_norm": 5.782884120941162,
      "learning_rate": 9.595776924311996e-07,
      "loss": 0.3318,
      "num_input_tokens_seen": 18307448,
      "step": 27800
    },
    {
      "epoch": 16.39445754716981,
      "grad_norm": 7.0055437088012695,
      "learning_rate": 9.580627273086313e-07,
      "loss": 0.3685,
      "num_input_tokens_seen": 18309944,
      "step": 27805
    },
    {
      "epoch": 16.39740566037736,
      "grad_norm": 5.0226898193359375,
      "learning_rate": 9.565488323063754e-07,
      "loss": 0.2967,
      "num_input_tokens_seen": 18313240,
      "step": 27810
    },
    {
      "epoch": 16.400353773584907,
      "grad_norm": 3.5612664222717285,
      "learning_rate": 9.55036007825243e-07,
      "loss": 0.3081,
      "num_input_tokens_seen": 18317176,
      "step": 27815
    },
    {
      "epoch": 16.403301886792452,
      "grad_norm": 2.822277545928955,
      "learning_rate": 9.535242542657602e-07,
      "loss": 0.2591,
      "num_input_tokens_seen": 18320280,
      "step": 27820
    },
    {
      "epoch": 16.40625,
      "grad_norm": 2.7134501934051514,
      "learning_rate": 9.520135720281692e-07,
      "loss": 0.293,
      "num_input_tokens_seen": 18324152,
      "step": 27825
    },
    {
      "epoch": 16.409198113207548,
      "grad_norm": 2.1314103603363037,
      "learning_rate": 9.505039615124318e-07,
      "loss": 0.3441,
      "num_input_tokens_seen": 18326328,
      "step": 27830
    },
    {
      "epoch": 16.412146226415093,
      "grad_norm": 4.7439799308776855,
      "learning_rate": 9.489954231182235e-07,
      "loss": 0.2869,
      "num_input_tokens_seen": 18328856,
      "step": 27835
    },
    {
      "epoch": 16.41509433962264,
      "grad_norm": 2.0246989727020264,
      "learning_rate": 9.474879572449352e-07,
      "loss": 0.2635,
      "num_input_tokens_seen": 18332696,
      "step": 27840
    },
    {
      "epoch": 16.41804245283019,
      "grad_norm": 3.249680995941162,
      "learning_rate": 9.459815642916759e-07,
      "loss": 0.431,
      "num_input_tokens_seen": 18336440,
      "step": 27845
    },
    {
      "epoch": 16.420990566037737,
      "grad_norm": 3.570563554763794,
      "learning_rate": 9.444762446572692e-07,
      "loss": 0.2783,
      "num_input_tokens_seen": 18339736,
      "step": 27850
    },
    {
      "epoch": 16.423938679245282,
      "grad_norm": 1.9866399765014648,
      "learning_rate": 9.429719987402541e-07,
      "loss": 0.3791,
      "num_input_tokens_seen": 18343096,
      "step": 27855
    },
    {
      "epoch": 16.42688679245283,
      "grad_norm": 4.635288238525391,
      "learning_rate": 9.414688269388883e-07,
      "loss": 0.2765,
      "num_input_tokens_seen": 18346040,
      "step": 27860
    },
    {
      "epoch": 16.42983490566038,
      "grad_norm": 3.203303813934326,
      "learning_rate": 9.3996672965114e-07,
      "loss": 0.2686,
      "num_input_tokens_seen": 18348568,
      "step": 27865
    },
    {
      "epoch": 16.432783018867923,
      "grad_norm": 4.082492828369141,
      "learning_rate": 9.384657072747e-07,
      "loss": 0.2725,
      "num_input_tokens_seen": 18350872,
      "step": 27870
    },
    {
      "epoch": 16.43573113207547,
      "grad_norm": 3.7214388847351074,
      "learning_rate": 9.369657602069676e-07,
      "loss": 0.2648,
      "num_input_tokens_seen": 18353560,
      "step": 27875
    },
    {
      "epoch": 16.43867924528302,
      "grad_norm": 3.5706076622009277,
      "learning_rate": 9.354668888450608e-07,
      "loss": 0.2792,
      "num_input_tokens_seen": 18357016,
      "step": 27880
    },
    {
      "epoch": 16.441627358490567,
      "grad_norm": 4.7808709144592285,
      "learning_rate": 9.339690935858125e-07,
      "loss": 0.343,
      "num_input_tokens_seen": 18360376,
      "step": 27885
    },
    {
      "epoch": 16.444575471698112,
      "grad_norm": 4.132315158843994,
      "learning_rate": 9.324723748257697e-07,
      "loss": 0.2888,
      "num_input_tokens_seen": 18363480,
      "step": 27890
    },
    {
      "epoch": 16.44752358490566,
      "grad_norm": 5.297361373901367,
      "learning_rate": 9.309767329611963e-07,
      "loss": 0.29,
      "num_input_tokens_seen": 18367512,
      "step": 27895
    },
    {
      "epoch": 16.45047169811321,
      "grad_norm": 2.6479098796844482,
      "learning_rate": 9.294821683880695e-07,
      "loss": 0.3776,
      "num_input_tokens_seen": 18371064,
      "step": 27900
    },
    {
      "epoch": 16.453419811320753,
      "grad_norm": 6.533544540405273,
      "learning_rate": 9.279886815020816e-07,
      "loss": 0.2196,
      "num_input_tokens_seen": 18375352,
      "step": 27905
    },
    {
      "epoch": 16.4563679245283,
      "grad_norm": 4.349218845367432,
      "learning_rate": 9.264962726986393e-07,
      "loss": 0.371,
      "num_input_tokens_seen": 18379672,
      "step": 27910
    },
    {
      "epoch": 16.45931603773585,
      "grad_norm": 3.1020421981811523,
      "learning_rate": 9.250049423728652e-07,
      "loss": 0.3069,
      "num_input_tokens_seen": 18382552,
      "step": 27915
    },
    {
      "epoch": 16.462264150943398,
      "grad_norm": 5.283880233764648,
      "learning_rate": 9.235146909195936e-07,
      "loss": 0.2243,
      "num_input_tokens_seen": 18385624,
      "step": 27920
    },
    {
      "epoch": 16.465212264150942,
      "grad_norm": 3.5904481410980225,
      "learning_rate": 9.220255187333771e-07,
      "loss": 0.296,
      "num_input_tokens_seen": 18388152,
      "step": 27925
    },
    {
      "epoch": 16.46816037735849,
      "grad_norm": 3.7692980766296387,
      "learning_rate": 9.205374262084798e-07,
      "loss": 0.4211,
      "num_input_tokens_seen": 18391160,
      "step": 27930
    },
    {
      "epoch": 16.47110849056604,
      "grad_norm": 2.754218578338623,
      "learning_rate": 9.190504137388806e-07,
      "loss": 0.3012,
      "num_input_tokens_seen": 18394808,
      "step": 27935
    },
    {
      "epoch": 16.474056603773583,
      "grad_norm": 5.1711344718933105,
      "learning_rate": 9.175644817182722e-07,
      "loss": 0.29,
      "num_input_tokens_seen": 18397240,
      "step": 27940
    },
    {
      "epoch": 16.47700471698113,
      "grad_norm": 4.459451198577881,
      "learning_rate": 9.16079630540061e-07,
      "loss": 0.335,
      "num_input_tokens_seen": 18400184,
      "step": 27945
    },
    {
      "epoch": 16.47995283018868,
      "grad_norm": 5.53680419921875,
      "learning_rate": 9.145958605973676e-07,
      "loss": 0.2811,
      "num_input_tokens_seen": 18402808,
      "step": 27950
    },
    {
      "epoch": 16.482900943396228,
      "grad_norm": 4.761890888214111,
      "learning_rate": 9.131131722830289e-07,
      "loss": 0.3584,
      "num_input_tokens_seen": 18408632,
      "step": 27955
    },
    {
      "epoch": 16.485849056603772,
      "grad_norm": 3.2619709968566895,
      "learning_rate": 9.116315659895892e-07,
      "loss": 0.271,
      "num_input_tokens_seen": 18411128,
      "step": 27960
    },
    {
      "epoch": 16.48879716981132,
      "grad_norm": 3.634657859802246,
      "learning_rate": 9.10151042109314e-07,
      "loss": 0.3459,
      "num_input_tokens_seen": 18414104,
      "step": 27965
    },
    {
      "epoch": 16.49174528301887,
      "grad_norm": 2.844452381134033,
      "learning_rate": 9.086716010341767e-07,
      "loss": 0.2784,
      "num_input_tokens_seen": 18417112,
      "step": 27970
    },
    {
      "epoch": 16.494693396226417,
      "grad_norm": 6.325484752655029,
      "learning_rate": 9.071932431558655e-07,
      "loss": 0.4917,
      "num_input_tokens_seen": 18421240,
      "step": 27975
    },
    {
      "epoch": 16.49764150943396,
      "grad_norm": 4.112940311431885,
      "learning_rate": 9.057159688657824e-07,
      "loss": 0.3356,
      "num_input_tokens_seen": 18424472,
      "step": 27980
    },
    {
      "epoch": 16.50058962264151,
      "grad_norm": 5.0034942626953125,
      "learning_rate": 9.042397785550405e-07,
      "loss": 0.3031,
      "num_input_tokens_seen": 18427160,
      "step": 27985
    },
    {
      "epoch": 16.503537735849058,
      "grad_norm": 3.4083549976348877,
      "learning_rate": 9.027646726144707e-07,
      "loss": 0.2985,
      "num_input_tokens_seen": 18430040,
      "step": 27990
    },
    {
      "epoch": 16.506485849056602,
      "grad_norm": 4.041662216186523,
      "learning_rate": 9.012906514346115e-07,
      "loss": 0.2886,
      "num_input_tokens_seen": 18432920,
      "step": 27995
    },
    {
      "epoch": 16.50943396226415,
      "grad_norm": 4.778041839599609,
      "learning_rate": 8.99817715405717e-07,
      "loss": 0.2862,
      "num_input_tokens_seen": 18437112,
      "step": 28000
    },
    {
      "epoch": 16.5123820754717,
      "grad_norm": 5.506007671356201,
      "learning_rate": 8.983458649177529e-07,
      "loss": 0.3602,
      "num_input_tokens_seen": 18439480,
      "step": 28005
    },
    {
      "epoch": 16.515330188679247,
      "grad_norm": 2.2714619636535645,
      "learning_rate": 8.968751003603982e-07,
      "loss": 0.3461,
      "num_input_tokens_seen": 18442040,
      "step": 28010
    },
    {
      "epoch": 16.51827830188679,
      "grad_norm": 4.901690483093262,
      "learning_rate": 8.95405422123043e-07,
      "loss": 0.3541,
      "num_input_tokens_seen": 18444984,
      "step": 28015
    },
    {
      "epoch": 16.52122641509434,
      "grad_norm": 5.613073825836182,
      "learning_rate": 8.939368305947932e-07,
      "loss": 0.3928,
      "num_input_tokens_seen": 18448088,
      "step": 28020
    },
    {
      "epoch": 16.524174528301888,
      "grad_norm": 3.7055649757385254,
      "learning_rate": 8.92469326164464e-07,
      "loss": 0.3648,
      "num_input_tokens_seen": 18451864,
      "step": 28025
    },
    {
      "epoch": 16.527122641509433,
      "grad_norm": 4.2378082275390625,
      "learning_rate": 8.910029092205829e-07,
      "loss": 0.2061,
      "num_input_tokens_seen": 18455448,
      "step": 28030
    },
    {
      "epoch": 16.53007075471698,
      "grad_norm": 3.3359925746917725,
      "learning_rate": 8.895375801513906e-07,
      "loss": 0.292,
      "num_input_tokens_seen": 18458392,
      "step": 28035
    },
    {
      "epoch": 16.53301886792453,
      "grad_norm": 4.083994388580322,
      "learning_rate": 8.880733393448377e-07,
      "loss": 0.3379,
      "num_input_tokens_seen": 18462136,
      "step": 28040
    },
    {
      "epoch": 16.535966981132077,
      "grad_norm": 3.8306102752685547,
      "learning_rate": 8.866101871885907e-07,
      "loss": 0.3793,
      "num_input_tokens_seen": 18465208,
      "step": 28045
    },
    {
      "epoch": 16.53891509433962,
      "grad_norm": 5.632137775421143,
      "learning_rate": 8.851481240700249e-07,
      "loss": 0.2346,
      "num_input_tokens_seen": 18467544,
      "step": 28050
    },
    {
      "epoch": 16.54186320754717,
      "grad_norm": 4.578236103057861,
      "learning_rate": 8.836871503762257e-07,
      "loss": 0.2625,
      "num_input_tokens_seen": 18470712,
      "step": 28055
    },
    {
      "epoch": 16.544811320754718,
      "grad_norm": 3.0723836421966553,
      "learning_rate": 8.822272664939946e-07,
      "loss": 0.3232,
      "num_input_tokens_seen": 18474008,
      "step": 28060
    },
    {
      "epoch": 16.547759433962263,
      "grad_norm": 3.0204155445098877,
      "learning_rate": 8.80768472809842e-07,
      "loss": 0.3403,
      "num_input_tokens_seen": 18476824,
      "step": 28065
    },
    {
      "epoch": 16.55070754716981,
      "grad_norm": 3.5329294204711914,
      "learning_rate": 8.793107697099884e-07,
      "loss": 0.3451,
      "num_input_tokens_seen": 18479800,
      "step": 28070
    },
    {
      "epoch": 16.55365566037736,
      "grad_norm": 3.5446345806121826,
      "learning_rate": 8.778541575803673e-07,
      "loss": 0.2581,
      "num_input_tokens_seen": 18483832,
      "step": 28075
    },
    {
      "epoch": 16.556603773584907,
      "grad_norm": 4.6515398025512695,
      "learning_rate": 8.763986368066241e-07,
      "loss": 0.2903,
      "num_input_tokens_seen": 18486904,
      "step": 28080
    },
    {
      "epoch": 16.559551886792452,
      "grad_norm": 3.7316181659698486,
      "learning_rate": 8.749442077741138e-07,
      "loss": 0.4272,
      "num_input_tokens_seen": 18490776,
      "step": 28085
    },
    {
      "epoch": 16.5625,
      "grad_norm": 3.1215574741363525,
      "learning_rate": 8.734908708679024e-07,
      "loss": 0.2738,
      "num_input_tokens_seen": 18493304,
      "step": 28090
    },
    {
      "epoch": 16.565448113207548,
      "grad_norm": 4.6116533279418945,
      "learning_rate": 8.72038626472767e-07,
      "loss": 0.2339,
      "num_input_tokens_seen": 18496696,
      "step": 28095
    },
    {
      "epoch": 16.568396226415093,
      "grad_norm": 3.6758310794830322,
      "learning_rate": 8.705874749731962e-07,
      "loss": 0.3231,
      "num_input_tokens_seen": 18505016,
      "step": 28100
    },
    {
      "epoch": 16.57134433962264,
      "grad_norm": 2.9239470958709717,
      "learning_rate": 8.691374167533867e-07,
      "loss": 0.433,
      "num_input_tokens_seen": 18508120,
      "step": 28105
    },
    {
      "epoch": 16.57429245283019,
      "grad_norm": 4.45897912979126,
      "learning_rate": 8.6768845219725e-07,
      "loss": 0.315,
      "num_input_tokens_seen": 18511064,
      "step": 28110
    },
    {
      "epoch": 16.577240566037737,
      "grad_norm": 4.514261722564697,
      "learning_rate": 8.662405816884056e-07,
      "loss": 0.4038,
      "num_input_tokens_seen": 18514360,
      "step": 28115
    },
    {
      "epoch": 16.580188679245282,
      "grad_norm": 3.2292368412017822,
      "learning_rate": 8.647938056101824e-07,
      "loss": 0.1735,
      "num_input_tokens_seen": 18517336,
      "step": 28120
    },
    {
      "epoch": 16.58313679245283,
      "grad_norm": 3.93446946144104,
      "learning_rate": 8.63348124345621e-07,
      "loss": 0.2839,
      "num_input_tokens_seen": 18520152,
      "step": 28125
    },
    {
      "epoch": 16.58608490566038,
      "grad_norm": 2.9556124210357666,
      "learning_rate": 8.619035382774716e-07,
      "loss": 0.308,
      "num_input_tokens_seen": 18522680,
      "step": 28130
    },
    {
      "epoch": 16.589033018867923,
      "grad_norm": 5.431281566619873,
      "learning_rate": 8.60460047788193e-07,
      "loss": 0.4035,
      "num_input_tokens_seen": 18526040,
      "step": 28135
    },
    {
      "epoch": 16.59198113207547,
      "grad_norm": 3.8193869590759277,
      "learning_rate": 8.590176532599587e-07,
      "loss": 0.2745,
      "num_input_tokens_seen": 18529560,
      "step": 28140
    },
    {
      "epoch": 16.59492924528302,
      "grad_norm": 6.53911018371582,
      "learning_rate": 8.575763550746475e-07,
      "loss": 0.2488,
      "num_input_tokens_seen": 18532248,
      "step": 28145
    },
    {
      "epoch": 16.597877358490567,
      "grad_norm": 5.721733093261719,
      "learning_rate": 8.56136153613848e-07,
      "loss": 0.2905,
      "num_input_tokens_seen": 18534456,
      "step": 28150
    },
    {
      "epoch": 16.600825471698112,
      "grad_norm": 3.398972272872925,
      "learning_rate": 8.546970492588619e-07,
      "loss": 0.3022,
      "num_input_tokens_seen": 18537144,
      "step": 28155
    },
    {
      "epoch": 16.60377358490566,
      "grad_norm": 3.715719223022461,
      "learning_rate": 8.532590423906973e-07,
      "loss": 0.2192,
      "num_input_tokens_seen": 18540376,
      "step": 28160
    },
    {
      "epoch": 16.60672169811321,
      "grad_norm": 3.3727025985717773,
      "learning_rate": 8.518221333900728e-07,
      "loss": 0.2789,
      "num_input_tokens_seen": 18543352,
      "step": 28165
    },
    {
      "epoch": 16.609669811320753,
      "grad_norm": 4.447512626647949,
      "learning_rate": 8.503863226374148e-07,
      "loss": 0.2842,
      "num_input_tokens_seen": 18546264,
      "step": 28170
    },
    {
      "epoch": 16.6126179245283,
      "grad_norm": 5.846355438232422,
      "learning_rate": 8.489516105128632e-07,
      "loss": 0.2735,
      "num_input_tokens_seen": 18549944,
      "step": 28175
    },
    {
      "epoch": 16.61556603773585,
      "grad_norm": 3.2531940937042236,
      "learning_rate": 8.475179973962621e-07,
      "loss": 0.4499,
      "num_input_tokens_seen": 18553240,
      "step": 28180
    },
    {
      "epoch": 16.618514150943398,
      "grad_norm": 3.079339027404785,
      "learning_rate": 8.460854836671678e-07,
      "loss": 0.2656,
      "num_input_tokens_seen": 18555864,
      "step": 28185
    },
    {
      "epoch": 16.621462264150942,
      "grad_norm": 3.1238672733306885,
      "learning_rate": 8.446540697048445e-07,
      "loss": 0.4284,
      "num_input_tokens_seen": 18559128,
      "step": 28190
    },
    {
      "epoch": 16.62441037735849,
      "grad_norm": 4.4841437339782715,
      "learning_rate": 8.432237558882639e-07,
      "loss": 0.3038,
      "num_input_tokens_seen": 18562264,
      "step": 28195
    },
    {
      "epoch": 16.62735849056604,
      "grad_norm": 4.798705577850342,
      "learning_rate": 8.417945425961083e-07,
      "loss": 0.4736,
      "num_input_tokens_seen": 18565208,
      "step": 28200
    },
    {
      "epoch": 16.630306603773583,
      "grad_norm": 3.5159404277801514,
      "learning_rate": 8.403664302067688e-07,
      "loss": 0.2664,
      "num_input_tokens_seen": 18568696,
      "step": 28205
    },
    {
      "epoch": 16.63325471698113,
      "grad_norm": 4.486374855041504,
      "learning_rate": 8.389394190983446e-07,
      "loss": 0.2916,
      "num_input_tokens_seen": 18575672,
      "step": 28210
    },
    {
      "epoch": 16.63620283018868,
      "grad_norm": 2.79045033454895,
      "learning_rate": 8.37513509648642e-07,
      "loss": 0.195,
      "num_input_tokens_seen": 18578776,
      "step": 28215
    },
    {
      "epoch": 16.639150943396228,
      "grad_norm": 3.2014050483703613,
      "learning_rate": 8.360887022351771e-07,
      "loss": 0.2758,
      "num_input_tokens_seen": 18582296,
      "step": 28220
    },
    {
      "epoch": 16.642099056603772,
      "grad_norm": 4.7932515144348145,
      "learning_rate": 8.346649972351739e-07,
      "loss": 0.2916,
      "num_input_tokens_seen": 18585336,
      "step": 28225
    },
    {
      "epoch": 16.64504716981132,
      "grad_norm": 5.699954509735107,
      "learning_rate": 8.33242395025563e-07,
      "loss": 0.3221,
      "num_input_tokens_seen": 18588440,
      "step": 28230
    },
    {
      "epoch": 16.64799528301887,
      "grad_norm": 3.966028928756714,
      "learning_rate": 8.318208959829871e-07,
      "loss": 0.3019,
      "num_input_tokens_seen": 18591864,
      "step": 28235
    },
    {
      "epoch": 16.650943396226417,
      "grad_norm": 4.777929782867432,
      "learning_rate": 8.304005004837929e-07,
      "loss": 0.2339,
      "num_input_tokens_seen": 18594360,
      "step": 28240
    },
    {
      "epoch": 16.65389150943396,
      "grad_norm": 7.578763484954834,
      "learning_rate": 8.289812089040344e-07,
      "loss": 0.4869,
      "num_input_tokens_seen": 18597560,
      "step": 28245
    },
    {
      "epoch": 16.65683962264151,
      "grad_norm": 5.610607624053955,
      "learning_rate": 8.275630216194785e-07,
      "loss": 0.349,
      "num_input_tokens_seen": 18600472,
      "step": 28250
    },
    {
      "epoch": 16.659787735849058,
      "grad_norm": 3.8890607357025146,
      "learning_rate": 8.261459390055948e-07,
      "loss": 0.3496,
      "num_input_tokens_seen": 18603192,
      "step": 28255
    },
    {
      "epoch": 16.662735849056602,
      "grad_norm": 3.803394317626953,
      "learning_rate": 8.24729961437562e-07,
      "loss": 0.2788,
      "num_input_tokens_seen": 18606392,
      "step": 28260
    },
    {
      "epoch": 16.66568396226415,
      "grad_norm": 3.944870948791504,
      "learning_rate": 8.233150892902653e-07,
      "loss": 0.2656,
      "num_input_tokens_seen": 18609176,
      "step": 28265
    },
    {
      "epoch": 16.6686320754717,
      "grad_norm": 2.813323497772217,
      "learning_rate": 8.219013229383005e-07,
      "loss": 0.454,
      "num_input_tokens_seen": 18612280,
      "step": 28270
    },
    {
      "epoch": 16.671580188679247,
      "grad_norm": 3.7285170555114746,
      "learning_rate": 8.204886627559666e-07,
      "loss": 0.3329,
      "num_input_tokens_seen": 18614712,
      "step": 28275
    },
    {
      "epoch": 16.67452830188679,
      "grad_norm": 3.6183383464813232,
      "learning_rate": 8.190771091172722e-07,
      "loss": 0.2392,
      "num_input_tokens_seen": 18618136,
      "step": 28280
    },
    {
      "epoch": 16.67747641509434,
      "grad_norm": 3.396876096725464,
      "learning_rate": 8.176666623959323e-07,
      "loss": 0.3215,
      "num_input_tokens_seen": 18621816,
      "step": 28285
    },
    {
      "epoch": 16.680424528301888,
      "grad_norm": 3.9186625480651855,
      "learning_rate": 8.162573229653681e-07,
      "loss": 0.3087,
      "num_input_tokens_seen": 18625592,
      "step": 28290
    },
    {
      "epoch": 16.683372641509433,
      "grad_norm": 2.7781436443328857,
      "learning_rate": 8.148490911987073e-07,
      "loss": 0.2302,
      "num_input_tokens_seen": 18629592,
      "step": 28295
    },
    {
      "epoch": 16.68632075471698,
      "grad_norm": 4.508155345916748,
      "learning_rate": 8.134419674687876e-07,
      "loss": 0.3124,
      "num_input_tokens_seen": 18631928,
      "step": 28300
    },
    {
      "epoch": 16.68926886792453,
      "grad_norm": 4.450277805328369,
      "learning_rate": 8.120359521481502e-07,
      "loss": 0.3479,
      "num_input_tokens_seen": 18635000,
      "step": 28305
    },
    {
      "epoch": 16.692216981132077,
      "grad_norm": 4.002969741821289,
      "learning_rate": 8.106310456090438e-07,
      "loss": 0.2552,
      "num_input_tokens_seen": 18637784,
      "step": 28310
    },
    {
      "epoch": 16.69516509433962,
      "grad_norm": 2.7369446754455566,
      "learning_rate": 8.092272482234231e-07,
      "loss": 0.2903,
      "num_input_tokens_seen": 18641496,
      "step": 28315
    },
    {
      "epoch": 16.69811320754717,
      "grad_norm": 5.067294597625732,
      "learning_rate": 8.078245603629486e-07,
      "loss": 0.3211,
      "num_input_tokens_seen": 18644568,
      "step": 28320
    },
    {
      "epoch": 16.701061320754718,
      "grad_norm": 3.1355276107788086,
      "learning_rate": 8.0642298239899e-07,
      "loss": 0.3889,
      "num_input_tokens_seen": 18647928,
      "step": 28325
    },
    {
      "epoch": 16.704009433962263,
      "grad_norm": 8.701777458190918,
      "learning_rate": 8.050225147026202e-07,
      "loss": 0.2747,
      "num_input_tokens_seen": 18651096,
      "step": 28330
    },
    {
      "epoch": 16.70695754716981,
      "grad_norm": 5.2595624923706055,
      "learning_rate": 8.03623157644619e-07,
      "loss": 0.3594,
      "num_input_tokens_seen": 18654072,
      "step": 28335
    },
    {
      "epoch": 16.70990566037736,
      "grad_norm": 6.3293633460998535,
      "learning_rate": 8.022249115954728e-07,
      "loss": 0.4267,
      "num_input_tokens_seen": 18656920,
      "step": 28340
    },
    {
      "epoch": 16.712853773584907,
      "grad_norm": 3.0945725440979004,
      "learning_rate": 8.008277769253709e-07,
      "loss": 0.2899,
      "num_input_tokens_seen": 18659992,
      "step": 28345
    },
    {
      "epoch": 16.715801886792452,
      "grad_norm": 3.1182284355163574,
      "learning_rate": 7.994317540042135e-07,
      "loss": 0.2287,
      "num_input_tokens_seen": 18662520,
      "step": 28350
    },
    {
      "epoch": 16.71875,
      "grad_norm": 4.502434253692627,
      "learning_rate": 7.980368432016017e-07,
      "loss": 0.3209,
      "num_input_tokens_seen": 18665176,
      "step": 28355
    },
    {
      "epoch": 16.721698113207548,
      "grad_norm": 4.263815402984619,
      "learning_rate": 7.966430448868461e-07,
      "loss": 0.2147,
      "num_input_tokens_seen": 18667992,
      "step": 28360
    },
    {
      "epoch": 16.724646226415093,
      "grad_norm": 2.8520796298980713,
      "learning_rate": 7.952503594289601e-07,
      "loss": 0.237,
      "num_input_tokens_seen": 18671320,
      "step": 28365
    },
    {
      "epoch": 16.72759433962264,
      "grad_norm": 2.9391887187957764,
      "learning_rate": 7.93858787196663e-07,
      "loss": 0.3905,
      "num_input_tokens_seen": 18674456,
      "step": 28370
    },
    {
      "epoch": 16.73054245283019,
      "grad_norm": 3.9708733558654785,
      "learning_rate": 7.92468328558379e-07,
      "loss": 0.3396,
      "num_input_tokens_seen": 18677784,
      "step": 28375
    },
    {
      "epoch": 16.733490566037737,
      "grad_norm": 3.5253713130950928,
      "learning_rate": 7.910789838822386e-07,
      "loss": 0.3874,
      "num_input_tokens_seen": 18681464,
      "step": 28380
    },
    {
      "epoch": 16.736438679245282,
      "grad_norm": 3.594465970993042,
      "learning_rate": 7.89690753536076e-07,
      "loss": 0.3085,
      "num_input_tokens_seen": 18685784,
      "step": 28385
    },
    {
      "epoch": 16.73938679245283,
      "grad_norm": 4.965672492980957,
      "learning_rate": 7.883036378874326e-07,
      "loss": 0.3756,
      "num_input_tokens_seen": 18688056,
      "step": 28390
    },
    {
      "epoch": 16.74233490566038,
      "grad_norm": 3.0073394775390625,
      "learning_rate": 7.86917637303552e-07,
      "loss": 0.3254,
      "num_input_tokens_seen": 18690616,
      "step": 28395
    },
    {
      "epoch": 16.745283018867923,
      "grad_norm": 2.2653534412384033,
      "learning_rate": 7.855327521513851e-07,
      "loss": 0.2289,
      "num_input_tokens_seen": 18693080,
      "step": 28400
    },
    {
      "epoch": 16.74823113207547,
      "grad_norm": 2.9346888065338135,
      "learning_rate": 7.841489827975851e-07,
      "loss": 0.2533,
      "num_input_tokens_seen": 18695928,
      "step": 28405
    },
    {
      "epoch": 16.75117924528302,
      "grad_norm": 3.505211353302002,
      "learning_rate": 7.827663296085109e-07,
      "loss": 0.331,
      "num_input_tokens_seen": 18698808,
      "step": 28410
    },
    {
      "epoch": 16.754127358490567,
      "grad_norm": 3.482731342315674,
      "learning_rate": 7.813847929502255e-07,
      "loss": 0.2758,
      "num_input_tokens_seen": 18701880,
      "step": 28415
    },
    {
      "epoch": 16.757075471698112,
      "grad_norm": 6.1247687339782715,
      "learning_rate": 7.800043731884982e-07,
      "loss": 0.3215,
      "num_input_tokens_seen": 18705080,
      "step": 28420
    },
    {
      "epoch": 16.76002358490566,
      "grad_norm": 10.56716537475586,
      "learning_rate": 7.786250706888005e-07,
      "loss": 0.2227,
      "num_input_tokens_seen": 18708408,
      "step": 28425
    },
    {
      "epoch": 16.76297169811321,
      "grad_norm": 3.884247303009033,
      "learning_rate": 7.772468858163085e-07,
      "loss": 0.3437,
      "num_input_tokens_seen": 18712312,
      "step": 28430
    },
    {
      "epoch": 16.765919811320753,
      "grad_norm": 5.344446182250977,
      "learning_rate": 7.758698189359026e-07,
      "loss": 0.2489,
      "num_input_tokens_seen": 18715064,
      "step": 28435
    },
    {
      "epoch": 16.7688679245283,
      "grad_norm": 3.4517247676849365,
      "learning_rate": 7.744938704121658e-07,
      "loss": 0.4029,
      "num_input_tokens_seen": 18718552,
      "step": 28440
    },
    {
      "epoch": 16.77181603773585,
      "grad_norm": 2.808821439743042,
      "learning_rate": 7.731190406093892e-07,
      "loss": 0.2401,
      "num_input_tokens_seen": 18725048,
      "step": 28445
    },
    {
      "epoch": 16.774764150943398,
      "grad_norm": 4.214805603027344,
      "learning_rate": 7.717453298915617e-07,
      "loss": 0.365,
      "num_input_tokens_seen": 18727736,
      "step": 28450
    },
    {
      "epoch": 16.777712264150942,
      "grad_norm": 8.828495025634766,
      "learning_rate": 7.703727386223825e-07,
      "loss": 0.3972,
      "num_input_tokens_seen": 18732888,
      "step": 28455
    },
    {
      "epoch": 16.78066037735849,
      "grad_norm": 3.146965980529785,
      "learning_rate": 7.690012671652491e-07,
      "loss": 0.2851,
      "num_input_tokens_seen": 18735768,
      "step": 28460
    },
    {
      "epoch": 16.78360849056604,
      "grad_norm": 1.7784008979797363,
      "learning_rate": 7.676309158832651e-07,
      "loss": 0.3289,
      "num_input_tokens_seen": 18739192,
      "step": 28465
    },
    {
      "epoch": 16.786556603773583,
      "grad_norm": 4.752766132354736,
      "learning_rate": 7.662616851392362e-07,
      "loss": 0.3194,
      "num_input_tokens_seen": 18742136,
      "step": 28470
    },
    {
      "epoch": 16.78950471698113,
      "grad_norm": 3.949188232421875,
      "learning_rate": 7.648935752956732e-07,
      "loss": 0.3906,
      "num_input_tokens_seen": 18744344,
      "step": 28475
    },
    {
      "epoch": 16.79245283018868,
      "grad_norm": 4.544149875640869,
      "learning_rate": 7.635265867147867e-07,
      "loss": 0.2042,
      "num_input_tokens_seen": 18747576,
      "step": 28480
    },
    {
      "epoch": 16.795400943396228,
      "grad_norm": 2.9175615310668945,
      "learning_rate": 7.621607197584963e-07,
      "loss": 0.4581,
      "num_input_tokens_seen": 18751608,
      "step": 28485
    },
    {
      "epoch": 16.798349056603772,
      "grad_norm": 3.9110360145568848,
      "learning_rate": 7.607959747884186e-07,
      "loss": 0.2733,
      "num_input_tokens_seen": 18754296,
      "step": 28490
    },
    {
      "epoch": 16.80129716981132,
      "grad_norm": 6.293025493621826,
      "learning_rate": 7.594323521658769e-07,
      "loss": 0.2711,
      "num_input_tokens_seen": 18758200,
      "step": 28495
    },
    {
      "epoch": 16.80424528301887,
      "grad_norm": 7.017145156860352,
      "learning_rate": 7.580698522518958e-07,
      "loss": 0.3673,
      "num_input_tokens_seen": 18761432,
      "step": 28500
    },
    {
      "epoch": 16.807193396226417,
      "grad_norm": 4.1823577880859375,
      "learning_rate": 7.567084754072035e-07,
      "loss": 0.3825,
      "num_input_tokens_seen": 18764600,
      "step": 28505
    },
    {
      "epoch": 16.81014150943396,
      "grad_norm": 1.9230045080184937,
      "learning_rate": 7.553482219922282e-07,
      "loss": 0.3376,
      "num_input_tokens_seen": 18772664,
      "step": 28510
    },
    {
      "epoch": 16.81308962264151,
      "grad_norm": 3.234172821044922,
      "learning_rate": 7.539890923671061e-07,
      "loss": 0.3282,
      "num_input_tokens_seen": 18776792,
      "step": 28515
    },
    {
      "epoch": 16.816037735849058,
      "grad_norm": 3.998917579650879,
      "learning_rate": 7.526310868916708e-07,
      "loss": 0.3245,
      "num_input_tokens_seen": 18779768,
      "step": 28520
    },
    {
      "epoch": 16.818985849056602,
      "grad_norm": 5.14208459854126,
      "learning_rate": 7.512742059254602e-07,
      "loss": 0.2924,
      "num_input_tokens_seen": 18782680,
      "step": 28525
    },
    {
      "epoch": 16.82193396226415,
      "grad_norm": 4.620311737060547,
      "learning_rate": 7.499184498277151e-07,
      "loss": 0.3955,
      "num_input_tokens_seen": 18785176,
      "step": 28530
    },
    {
      "epoch": 16.8248820754717,
      "grad_norm": 1.9465115070343018,
      "learning_rate": 7.485638189573758e-07,
      "loss": 0.3055,
      "num_input_tokens_seen": 18788632,
      "step": 28535
    },
    {
      "epoch": 16.827830188679247,
      "grad_norm": 3.266594648361206,
      "learning_rate": 7.472103136730891e-07,
      "loss": 0.311,
      "num_input_tokens_seen": 18791928,
      "step": 28540
    },
    {
      "epoch": 16.83077830188679,
      "grad_norm": 4.758243560791016,
      "learning_rate": 7.458579343331996e-07,
      "loss": 0.3034,
      "num_input_tokens_seen": 18795640,
      "step": 28545
    },
    {
      "epoch": 16.83372641509434,
      "grad_norm": 3.150561571121216,
      "learning_rate": 7.445066812957569e-07,
      "loss": 0.261,
      "num_input_tokens_seen": 18799448,
      "step": 28550
    },
    {
      "epoch": 16.836674528301888,
      "grad_norm": 4.858627796173096,
      "learning_rate": 7.43156554918511e-07,
      "loss": 0.3331,
      "num_input_tokens_seen": 18804152,
      "step": 28555
    },
    {
      "epoch": 16.839622641509433,
      "grad_norm": 5.260900497436523,
      "learning_rate": 7.418075555589132e-07,
      "loss": 0.2929,
      "num_input_tokens_seen": 18807256,
      "step": 28560
    },
    {
      "epoch": 16.84257075471698,
      "grad_norm": 3.100715160369873,
      "learning_rate": 7.404596835741168e-07,
      "loss": 0.2974,
      "num_input_tokens_seen": 18811000,
      "step": 28565
    },
    {
      "epoch": 16.84551886792453,
      "grad_norm": 3.5762195587158203,
      "learning_rate": 7.391129393209751e-07,
      "loss": 0.2669,
      "num_input_tokens_seen": 18814712,
      "step": 28570
    },
    {
      "epoch": 16.848466981132077,
      "grad_norm": 3.930922508239746,
      "learning_rate": 7.377673231560478e-07,
      "loss": 0.3345,
      "num_input_tokens_seen": 18818840,
      "step": 28575
    },
    {
      "epoch": 16.85141509433962,
      "grad_norm": 3.7862467765808105,
      "learning_rate": 7.364228354355907e-07,
      "loss": 0.3647,
      "num_input_tokens_seen": 18822392,
      "step": 28580
    },
    {
      "epoch": 16.85436320754717,
      "grad_norm": 3.4929964542388916,
      "learning_rate": 7.350794765155627e-07,
      "loss": 0.4024,
      "num_input_tokens_seen": 18826360,
      "step": 28585
    },
    {
      "epoch": 16.857311320754718,
      "grad_norm": 4.902193069458008,
      "learning_rate": 7.337372467516246e-07,
      "loss": 0.4051,
      "num_input_tokens_seen": 18829048,
      "step": 28590
    },
    {
      "epoch": 16.860259433962263,
      "grad_norm": 2.7453107833862305,
      "learning_rate": 7.323961464991369e-07,
      "loss": 0.3251,
      "num_input_tokens_seen": 18832632,
      "step": 28595
    },
    {
      "epoch": 16.86320754716981,
      "grad_norm": 6.596832752227783,
      "learning_rate": 7.310561761131601e-07,
      "loss": 0.2517,
      "num_input_tokens_seen": 18835480,
      "step": 28600
    },
    {
      "epoch": 16.86615566037736,
      "grad_norm": 2.842146873474121,
      "learning_rate": 7.297173359484605e-07,
      "loss": 0.3167,
      "num_input_tokens_seen": 18839384,
      "step": 28605
    },
    {
      "epoch": 16.869103773584907,
      "grad_norm": 2.4638924598693848,
      "learning_rate": 7.283796263595e-07,
      "loss": 0.2828,
      "num_input_tokens_seen": 18842136,
      "step": 28610
    },
    {
      "epoch": 16.872051886792452,
      "grad_norm": 8.432517051696777,
      "learning_rate": 7.270430477004431e-07,
      "loss": 0.2462,
      "num_input_tokens_seen": 18845336,
      "step": 28615
    },
    {
      "epoch": 16.875,
      "grad_norm": 2.7862906455993652,
      "learning_rate": 7.257076003251545e-07,
      "loss": 0.2611,
      "num_input_tokens_seen": 18848280,
      "step": 28620
    },
    {
      "epoch": 16.877948113207548,
      "grad_norm": 4.012385845184326,
      "learning_rate": 7.243732845871998e-07,
      "loss": 0.2586,
      "num_input_tokens_seen": 18851160,
      "step": 28625
    },
    {
      "epoch": 16.880896226415093,
      "grad_norm": 5.216960906982422,
      "learning_rate": 7.230401008398441e-07,
      "loss": 0.3454,
      "num_input_tokens_seen": 18854552,
      "step": 28630
    },
    {
      "epoch": 16.88384433962264,
      "grad_norm": 3.3255977630615234,
      "learning_rate": 7.217080494360546e-07,
      "loss": 0.3094,
      "num_input_tokens_seen": 18857816,
      "step": 28635
    },
    {
      "epoch": 16.88679245283019,
      "grad_norm": 2.7415771484375,
      "learning_rate": 7.20377130728498e-07,
      "loss": 0.2784,
      "num_input_tokens_seen": 18861592,
      "step": 28640
    },
    {
      "epoch": 16.889740566037737,
      "grad_norm": 2.9527549743652344,
      "learning_rate": 7.190473450695407e-07,
      "loss": 0.2352,
      "num_input_tokens_seen": 18864056,
      "step": 28645
    },
    {
      "epoch": 16.892688679245282,
      "grad_norm": 5.905776023864746,
      "learning_rate": 7.177186928112484e-07,
      "loss": 0.2812,
      "num_input_tokens_seen": 18866808,
      "step": 28650
    },
    {
      "epoch": 16.89563679245283,
      "grad_norm": 2.8307688236236572,
      "learning_rate": 7.163911743053876e-07,
      "loss": 0.3897,
      "num_input_tokens_seen": 18870488,
      "step": 28655
    },
    {
      "epoch": 16.89858490566038,
      "grad_norm": 2.8549771308898926,
      "learning_rate": 7.150647899034252e-07,
      "loss": 0.2308,
      "num_input_tokens_seen": 18873624,
      "step": 28660
    },
    {
      "epoch": 16.901533018867923,
      "grad_norm": 3.6679294109344482,
      "learning_rate": 7.13739539956525e-07,
      "loss": 0.4488,
      "num_input_tokens_seen": 18876696,
      "step": 28665
    },
    {
      "epoch": 16.90448113207547,
      "grad_norm": 3.8584022521972656,
      "learning_rate": 7.124154248155562e-07,
      "loss": 0.2335,
      "num_input_tokens_seen": 18879608,
      "step": 28670
    },
    {
      "epoch": 16.90742924528302,
      "grad_norm": 4.057493209838867,
      "learning_rate": 7.110924448310813e-07,
      "loss": 0.339,
      "num_input_tokens_seen": 18883576,
      "step": 28675
    },
    {
      "epoch": 16.910377358490567,
      "grad_norm": 4.394983291625977,
      "learning_rate": 7.097706003533666e-07,
      "loss": 0.2949,
      "num_input_tokens_seen": 18886296,
      "step": 28680
    },
    {
      "epoch": 16.913325471698112,
      "grad_norm": 8.983338356018066,
      "learning_rate": 7.084498917323751e-07,
      "loss": 0.305,
      "num_input_tokens_seen": 18888888,
      "step": 28685
    },
    {
      "epoch": 16.91627358490566,
      "grad_norm": 3.766512632369995,
      "learning_rate": 7.071303193177698e-07,
      "loss": 0.3394,
      "num_input_tokens_seen": 18892568,
      "step": 28690
    },
    {
      "epoch": 16.91922169811321,
      "grad_norm": 3.3280534744262695,
      "learning_rate": 7.058118834589133e-07,
      "loss": 0.2748,
      "num_input_tokens_seen": 18895544,
      "step": 28695
    },
    {
      "epoch": 16.922169811320753,
      "grad_norm": 2.32316255569458,
      "learning_rate": 7.044945845048684e-07,
      "loss": 0.2788,
      "num_input_tokens_seen": 18898520,
      "step": 28700
    },
    {
      "epoch": 16.9251179245283,
      "grad_norm": 5.377107620239258,
      "learning_rate": 7.031784228043948e-07,
      "loss": 0.3332,
      "num_input_tokens_seen": 18902168,
      "step": 28705
    },
    {
      "epoch": 16.92806603773585,
      "grad_norm": 4.215866565704346,
      "learning_rate": 7.01863398705952e-07,
      "loss": 0.3683,
      "num_input_tokens_seen": 18905208,
      "step": 28710
    },
    {
      "epoch": 16.931014150943398,
      "grad_norm": 4.07161283493042,
      "learning_rate": 7.005495125576983e-07,
      "loss": 0.3794,
      "num_input_tokens_seen": 18908120,
      "step": 28715
    },
    {
      "epoch": 16.933962264150942,
      "grad_norm": 4.284706115722656,
      "learning_rate": 6.99236764707491e-07,
      "loss": 0.3158,
      "num_input_tokens_seen": 18911800,
      "step": 28720
    },
    {
      "epoch": 16.93691037735849,
      "grad_norm": 3.0451526641845703,
      "learning_rate": 6.979251555028843e-07,
      "loss": 0.4096,
      "num_input_tokens_seen": 18915416,
      "step": 28725
    },
    {
      "epoch": 16.93985849056604,
      "grad_norm": 3.4537360668182373,
      "learning_rate": 6.966146852911332e-07,
      "loss": 0.2819,
      "num_input_tokens_seen": 18918328,
      "step": 28730
    },
    {
      "epoch": 16.942806603773583,
      "grad_norm": 3.584404945373535,
      "learning_rate": 6.953053544191923e-07,
      "loss": 0.2565,
      "num_input_tokens_seen": 18921784,
      "step": 28735
    },
    {
      "epoch": 16.94575471698113,
      "grad_norm": 3.4418082237243652,
      "learning_rate": 6.939971632337111e-07,
      "loss": 0.3185,
      "num_input_tokens_seen": 18925656,
      "step": 28740
    },
    {
      "epoch": 16.94870283018868,
      "grad_norm": 4.600634574890137,
      "learning_rate": 6.926901120810387e-07,
      "loss": 0.3391,
      "num_input_tokens_seen": 18928632,
      "step": 28745
    },
    {
      "epoch": 16.951650943396228,
      "grad_norm": 5.2548627853393555,
      "learning_rate": 6.91384201307222e-07,
      "loss": 0.3481,
      "num_input_tokens_seen": 18931192,
      "step": 28750
    },
    {
      "epoch": 16.954599056603772,
      "grad_norm": 3.1680102348327637,
      "learning_rate": 6.900794312580078e-07,
      "loss": 0.4799,
      "num_input_tokens_seen": 18933784,
      "step": 28755
    },
    {
      "epoch": 16.95754716981132,
      "grad_norm": 3.6733555793762207,
      "learning_rate": 6.887758022788377e-07,
      "loss": 0.3484,
      "num_input_tokens_seen": 18937560,
      "step": 28760
    },
    {
      "epoch": 16.96049528301887,
      "grad_norm": 3.482055187225342,
      "learning_rate": 6.874733147148549e-07,
      "loss": 0.3171,
      "num_input_tokens_seen": 18940536,
      "step": 28765
    },
    {
      "epoch": 16.963443396226417,
      "grad_norm": 3.26082181930542,
      "learning_rate": 6.861719689108987e-07,
      "loss": 0.3087,
      "num_input_tokens_seen": 18943384,
      "step": 28770
    },
    {
      "epoch": 16.96639150943396,
      "grad_norm": 6.608047962188721,
      "learning_rate": 6.84871765211505e-07,
      "loss": 0.294,
      "num_input_tokens_seen": 18947032,
      "step": 28775
    },
    {
      "epoch": 16.96933962264151,
      "grad_norm": 3.29699969291687,
      "learning_rate": 6.835727039609086e-07,
      "loss": 0.3294,
      "num_input_tokens_seen": 18950008,
      "step": 28780
    },
    {
      "epoch": 16.972287735849058,
      "grad_norm": 4.759889125823975,
      "learning_rate": 6.822747855030415e-07,
      "loss": 0.2198,
      "num_input_tokens_seen": 18952120,
      "step": 28785
    },
    {
      "epoch": 16.975235849056602,
      "grad_norm": 4.871650695800781,
      "learning_rate": 6.809780101815322e-07,
      "loss": 0.2426,
      "num_input_tokens_seen": 18955384,
      "step": 28790
    },
    {
      "epoch": 16.97818396226415,
      "grad_norm": 3.3428797721862793,
      "learning_rate": 6.796823783397099e-07,
      "loss": 0.4315,
      "num_input_tokens_seen": 18958328,
      "step": 28795
    },
    {
      "epoch": 16.9811320754717,
      "grad_norm": 4.5987420082092285,
      "learning_rate": 6.783878903205976e-07,
      "loss": 0.5022,
      "num_input_tokens_seen": 18961976,
      "step": 28800
    },
    {
      "epoch": 16.984080188679247,
      "grad_norm": 4.614760875701904,
      "learning_rate": 6.77094546466916e-07,
      "loss": 0.3399,
      "num_input_tokens_seen": 18965752,
      "step": 28805
    },
    {
      "epoch": 16.98702830188679,
      "grad_norm": 4.215668678283691,
      "learning_rate": 6.758023471210845e-07,
      "loss": 0.3825,
      "num_input_tokens_seen": 18969016,
      "step": 28810
    },
    {
      "epoch": 16.98997641509434,
      "grad_norm": 2.9908902645111084,
      "learning_rate": 6.745112926252162e-07,
      "loss": 0.224,
      "num_input_tokens_seen": 18971672,
      "step": 28815
    },
    {
      "epoch": 16.992924528301888,
      "grad_norm": 5.506106853485107,
      "learning_rate": 6.732213833211265e-07,
      "loss": 0.2555,
      "num_input_tokens_seen": 18974776,
      "step": 28820
    },
    {
      "epoch": 16.995872641509433,
      "grad_norm": 4.564915180206299,
      "learning_rate": 6.719326195503218e-07,
      "loss": 0.375,
      "num_input_tokens_seen": 18977528,
      "step": 28825
    },
    {
      "epoch": 16.99882075471698,
      "grad_norm": 4.007802963256836,
      "learning_rate": 6.706450016540094e-07,
      "loss": 0.303,
      "num_input_tokens_seen": 18981432,
      "step": 28830
    },
    {
      "epoch": 17.00176886792453,
      "grad_norm": 3.629765272140503,
      "learning_rate": 6.69358529973092e-07,
      "loss": 0.2273,
      "num_input_tokens_seen": 18984456,
      "step": 28835
    },
    {
      "epoch": 17.004716981132077,
      "grad_norm": 4.066451549530029,
      "learning_rate": 6.680732048481681e-07,
      "loss": 0.2474,
      "num_input_tokens_seen": 18987144,
      "step": 28840
    },
    {
      "epoch": 17.00766509433962,
      "grad_norm": 3.822641372680664,
      "learning_rate": 6.667890266195321e-07,
      "loss": 0.311,
      "num_input_tokens_seen": 18990280,
      "step": 28845
    },
    {
      "epoch": 17.01061320754717,
      "grad_norm": 4.115925312042236,
      "learning_rate": 6.655059956271759e-07,
      "loss": 0.2849,
      "num_input_tokens_seen": 18994376,
      "step": 28850
    },
    {
      "epoch": 17.013561320754718,
      "grad_norm": 5.251235485076904,
      "learning_rate": 6.642241122107884e-07,
      "loss": 0.2264,
      "num_input_tokens_seen": 18997128,
      "step": 28855
    },
    {
      "epoch": 17.016509433962263,
      "grad_norm": 4.801900863647461,
      "learning_rate": 6.629433767097537e-07,
      "loss": 0.2657,
      "num_input_tokens_seen": 18999784,
      "step": 28860
    },
    {
      "epoch": 17.01945754716981,
      "grad_norm": 5.395425319671631,
      "learning_rate": 6.616637894631517e-07,
      "loss": 0.3488,
      "num_input_tokens_seen": 19003208,
      "step": 28865
    },
    {
      "epoch": 17.02240566037736,
      "grad_norm": 4.204492092132568,
      "learning_rate": 6.603853508097591e-07,
      "loss": 0.3017,
      "num_input_tokens_seen": 19006504,
      "step": 28870
    },
    {
      "epoch": 17.025353773584907,
      "grad_norm": 2.8667423725128174,
      "learning_rate": 6.591080610880468e-07,
      "loss": 0.2404,
      "num_input_tokens_seen": 19009704,
      "step": 28875
    },
    {
      "epoch": 17.028301886792452,
      "grad_norm": 4.226739883422852,
      "learning_rate": 6.578319206361828e-07,
      "loss": 0.4329,
      "num_input_tokens_seen": 19013128,
      "step": 28880
    },
    {
      "epoch": 17.03125,
      "grad_norm": 5.551898956298828,
      "learning_rate": 6.565569297920327e-07,
      "loss": 0.3351,
      "num_input_tokens_seen": 19015976,
      "step": 28885
    },
    {
      "epoch": 17.034198113207548,
      "grad_norm": 3.1933698654174805,
      "learning_rate": 6.552830888931544e-07,
      "loss": 0.1988,
      "num_input_tokens_seen": 19019080,
      "step": 28890
    },
    {
      "epoch": 17.037146226415093,
      "grad_norm": 3.3837432861328125,
      "learning_rate": 6.540103982768031e-07,
      "loss": 0.2964,
      "num_input_tokens_seen": 19022312,
      "step": 28895
    },
    {
      "epoch": 17.04009433962264,
      "grad_norm": 5.0731520652771,
      "learning_rate": 6.527388582799293e-07,
      "loss": 0.3153,
      "num_input_tokens_seen": 19025064,
      "step": 28900
    },
    {
      "epoch": 17.04304245283019,
      "grad_norm": 3.469153642654419,
      "learning_rate": 6.514684692391782e-07,
      "loss": 0.2507,
      "num_input_tokens_seen": 19029128,
      "step": 28905
    },
    {
      "epoch": 17.045990566037737,
      "grad_norm": 5.417031764984131,
      "learning_rate": 6.501992314908895e-07,
      "loss": 0.3196,
      "num_input_tokens_seen": 19032168,
      "step": 28910
    },
    {
      "epoch": 17.048938679245282,
      "grad_norm": 2.4677956104278564,
      "learning_rate": 6.489311453711017e-07,
      "loss": 0.251,
      "num_input_tokens_seen": 19036456,
      "step": 28915
    },
    {
      "epoch": 17.05188679245283,
      "grad_norm": 6.521324157714844,
      "learning_rate": 6.476642112155457e-07,
      "loss": 0.3522,
      "num_input_tokens_seen": 19039848,
      "step": 28920
    },
    {
      "epoch": 17.05483490566038,
      "grad_norm": 4.398978233337402,
      "learning_rate": 6.463984293596476e-07,
      "loss": 0.2651,
      "num_input_tokens_seen": 19043752,
      "step": 28925
    },
    {
      "epoch": 17.057783018867923,
      "grad_norm": 3.5171890258789062,
      "learning_rate": 6.451338001385282e-07,
      "loss": 0.2463,
      "num_input_tokens_seen": 19049128,
      "step": 28930
    },
    {
      "epoch": 17.06073113207547,
      "grad_norm": 5.265621185302734,
      "learning_rate": 6.438703238870037e-07,
      "loss": 0.2951,
      "num_input_tokens_seen": 19051688,
      "step": 28935
    },
    {
      "epoch": 17.06367924528302,
      "grad_norm": 3.5978972911834717,
      "learning_rate": 6.426080009395846e-07,
      "loss": 0.2147,
      "num_input_tokens_seen": 19055208,
      "step": 28940
    },
    {
      "epoch": 17.066627358490567,
      "grad_norm": 4.823855876922607,
      "learning_rate": 6.413468316304755e-07,
      "loss": 0.2424,
      "num_input_tokens_seen": 19058152,
      "step": 28945
    },
    {
      "epoch": 17.069575471698112,
      "grad_norm": 3.9790167808532715,
      "learning_rate": 6.400868162935786e-07,
      "loss": 0.3135,
      "num_input_tokens_seen": 19060584,
      "step": 28950
    },
    {
      "epoch": 17.07252358490566,
      "grad_norm": 4.206318378448486,
      "learning_rate": 6.388279552624877e-07,
      "loss": 0.373,
      "num_input_tokens_seen": 19062888,
      "step": 28955
    },
    {
      "epoch": 17.07547169811321,
      "grad_norm": 4.334952354431152,
      "learning_rate": 6.37570248870491e-07,
      "loss": 0.296,
      "num_input_tokens_seen": 19066120,
      "step": 28960
    },
    {
      "epoch": 17.078419811320753,
      "grad_norm": 5.295276165008545,
      "learning_rate": 6.363136974505718e-07,
      "loss": 0.3195,
      "num_input_tokens_seen": 19068488,
      "step": 28965
    },
    {
      "epoch": 17.0813679245283,
      "grad_norm": 2.298973798751831,
      "learning_rate": 6.350583013354078e-07,
      "loss": 0.331,
      "num_input_tokens_seen": 19071816,
      "step": 28970
    },
    {
      "epoch": 17.08431603773585,
      "grad_norm": 4.2149810791015625,
      "learning_rate": 6.338040608573693e-07,
      "loss": 0.3813,
      "num_input_tokens_seen": 19074952,
      "step": 28975
    },
    {
      "epoch": 17.087264150943398,
      "grad_norm": 2.724093198776245,
      "learning_rate": 6.325509763485238e-07,
      "loss": 0.3138,
      "num_input_tokens_seen": 19077448,
      "step": 28980
    },
    {
      "epoch": 17.090212264150942,
      "grad_norm": 9.34151554107666,
      "learning_rate": 6.312990481406301e-07,
      "loss": 0.3523,
      "num_input_tokens_seen": 19080712,
      "step": 28985
    },
    {
      "epoch": 17.09316037735849,
      "grad_norm": 5.244448661804199,
      "learning_rate": 6.300482765651411e-07,
      "loss": 0.2523,
      "num_input_tokens_seen": 19083624,
      "step": 28990
    },
    {
      "epoch": 17.09610849056604,
      "grad_norm": 3.4472568035125732,
      "learning_rate": 6.28798661953205e-07,
      "loss": 0.2042,
      "num_input_tokens_seen": 19088040,
      "step": 28995
    },
    {
      "epoch": 17.099056603773583,
      "grad_norm": 3.1663014888763428,
      "learning_rate": 6.275502046356618e-07,
      "loss": 0.3421,
      "num_input_tokens_seen": 19090952,
      "step": 29000
    },
    {
      "epoch": 17.10200471698113,
      "grad_norm": 4.734785556793213,
      "learning_rate": 6.263029049430447e-07,
      "loss": 0.3413,
      "num_input_tokens_seen": 19094184,
      "step": 29005
    },
    {
      "epoch": 17.10495283018868,
      "grad_norm": 3.3045074939727783,
      "learning_rate": 6.250567632055832e-07,
      "loss": 0.3831,
      "num_input_tokens_seen": 19098824,
      "step": 29010
    },
    {
      "epoch": 17.107900943396228,
      "grad_norm": 4.207641124725342,
      "learning_rate": 6.238117797532e-07,
      "loss": 0.2814,
      "num_input_tokens_seen": 19102920,
      "step": 29015
    },
    {
      "epoch": 17.110849056603772,
      "grad_norm": 7.541811466217041,
      "learning_rate": 6.225679549155083e-07,
      "loss": 0.3795,
      "num_input_tokens_seen": 19105416,
      "step": 29020
    },
    {
      "epoch": 17.11379716981132,
      "grad_norm": 10.750240325927734,
      "learning_rate": 6.213252890218163e-07,
      "loss": 0.3093,
      "num_input_tokens_seen": 19109384,
      "step": 29025
    },
    {
      "epoch": 17.11674528301887,
      "grad_norm": 4.577506065368652,
      "learning_rate": 6.200837824011247e-07,
      "loss": 0.3172,
      "num_input_tokens_seen": 19112104,
      "step": 29030
    },
    {
      "epoch": 17.119693396226417,
      "grad_norm": 3.802475690841675,
      "learning_rate": 6.188434353821282e-07,
      "loss": 0.2909,
      "num_input_tokens_seen": 19115528,
      "step": 29035
    },
    {
      "epoch": 17.12264150943396,
      "grad_norm": 2.8041629791259766,
      "learning_rate": 6.176042482932132e-07,
      "loss": 0.2495,
      "num_input_tokens_seen": 19118984,
      "step": 29040
    },
    {
      "epoch": 17.12558962264151,
      "grad_norm": 3.020643472671509,
      "learning_rate": 6.163662214624616e-07,
      "loss": 0.3541,
      "num_input_tokens_seen": 19122856,
      "step": 29045
    },
    {
      "epoch": 17.128537735849058,
      "grad_norm": 4.110010623931885,
      "learning_rate": 6.151293552176451e-07,
      "loss": 0.3419,
      "num_input_tokens_seen": 19125768,
      "step": 29050
    },
    {
      "epoch": 17.131485849056602,
      "grad_norm": 5.032533645629883,
      "learning_rate": 6.138936498862291e-07,
      "loss": 0.2841,
      "num_input_tokens_seen": 19128712,
      "step": 29055
    },
    {
      "epoch": 17.13443396226415,
      "grad_norm": 5.3970255851745605,
      "learning_rate": 6.126591057953729e-07,
      "loss": 0.3984,
      "num_input_tokens_seen": 19131208,
      "step": 29060
    },
    {
      "epoch": 17.1373820754717,
      "grad_norm": 6.033452033996582,
      "learning_rate": 6.114257232719267e-07,
      "loss": 0.3307,
      "num_input_tokens_seen": 19135112,
      "step": 29065
    },
    {
      "epoch": 17.140330188679247,
      "grad_norm": 6.008227825164795,
      "learning_rate": 6.101935026424332e-07,
      "loss": 0.3551,
      "num_input_tokens_seen": 19137576,
      "step": 29070
    },
    {
      "epoch": 17.14327830188679,
      "grad_norm": 3.6999757289886475,
      "learning_rate": 6.089624442331293e-07,
      "loss": 0.2981,
      "num_input_tokens_seen": 19140584,
      "step": 29075
    },
    {
      "epoch": 17.14622641509434,
      "grad_norm": 4.418806552886963,
      "learning_rate": 6.077325483699432e-07,
      "loss": 0.2451,
      "num_input_tokens_seen": 19143816,
      "step": 29080
    },
    {
      "epoch": 17.149174528301888,
      "grad_norm": 3.2146553993225098,
      "learning_rate": 6.065038153784947e-07,
      "loss": 0.3054,
      "num_input_tokens_seen": 19147240,
      "step": 29085
    },
    {
      "epoch": 17.152122641509433,
      "grad_norm": 3.1978659629821777,
      "learning_rate": 6.052762455840955e-07,
      "loss": 0.351,
      "num_input_tokens_seen": 19150344,
      "step": 29090
    },
    {
      "epoch": 17.15507075471698,
      "grad_norm": 4.369086265563965,
      "learning_rate": 6.040498393117494e-07,
      "loss": 0.2785,
      "num_input_tokens_seen": 19153128,
      "step": 29095
    },
    {
      "epoch": 17.15801886792453,
      "grad_norm": 3.3388020992279053,
      "learning_rate": 6.028245968861551e-07,
      "loss": 0.3897,
      "num_input_tokens_seen": 19155592,
      "step": 29100
    },
    {
      "epoch": 17.160966981132077,
      "grad_norm": 3.121723175048828,
      "learning_rate": 6.016005186316987e-07,
      "loss": 0.3127,
      "num_input_tokens_seen": 19159528,
      "step": 29105
    },
    {
      "epoch": 17.16391509433962,
      "grad_norm": 5.467022895812988,
      "learning_rate": 6.003776048724614e-07,
      "loss": 0.3388,
      "num_input_tokens_seen": 19163464,
      "step": 29110
    },
    {
      "epoch": 17.16686320754717,
      "grad_norm": 4.134096145629883,
      "learning_rate": 5.991558559322152e-07,
      "loss": 0.3593,
      "num_input_tokens_seen": 19166632,
      "step": 29115
    },
    {
      "epoch": 17.169811320754718,
      "grad_norm": 4.204727649688721,
      "learning_rate": 5.979352721344223e-07,
      "loss": 0.3632,
      "num_input_tokens_seen": 19169608,
      "step": 29120
    },
    {
      "epoch": 17.172759433962263,
      "grad_norm": 4.6988844871521,
      "learning_rate": 5.967158538022383e-07,
      "loss": 0.3053,
      "num_input_tokens_seen": 19172616,
      "step": 29125
    },
    {
      "epoch": 17.17570754716981,
      "grad_norm": 3.9774045944213867,
      "learning_rate": 5.954976012585078e-07,
      "loss": 0.3277,
      "num_input_tokens_seen": 19175016,
      "step": 29130
    },
    {
      "epoch": 17.17865566037736,
      "grad_norm": 3.0399105548858643,
      "learning_rate": 5.942805148257713e-07,
      "loss": 0.297,
      "num_input_tokens_seen": 19177960,
      "step": 29135
    },
    {
      "epoch": 17.181603773584907,
      "grad_norm": 5.205494403839111,
      "learning_rate": 5.930645948262553e-07,
      "loss": 0.3035,
      "num_input_tokens_seen": 19180584,
      "step": 29140
    },
    {
      "epoch": 17.184551886792452,
      "grad_norm": 3.606154441833496,
      "learning_rate": 5.918498415818813e-07,
      "loss": 0.2727,
      "num_input_tokens_seen": 19183880,
      "step": 29145
    },
    {
      "epoch": 17.1875,
      "grad_norm": 4.335180282592773,
      "learning_rate": 5.906362554142592e-07,
      "loss": 0.2697,
      "num_input_tokens_seen": 19187944,
      "step": 29150
    },
    {
      "epoch": 17.190448113207548,
      "grad_norm": 5.392014026641846,
      "learning_rate": 5.894238366446925e-07,
      "loss": 0.3094,
      "num_input_tokens_seen": 19191912,
      "step": 29155
    },
    {
      "epoch": 17.193396226415093,
      "grad_norm": 3.467682361602783,
      "learning_rate": 5.882125855941723e-07,
      "loss": 0.2264,
      "num_input_tokens_seen": 19195336,
      "step": 29160
    },
    {
      "epoch": 17.19634433962264,
      "grad_norm": 2.458475112915039,
      "learning_rate": 5.870025025833842e-07,
      "loss": 0.2723,
      "num_input_tokens_seen": 19198856,
      "step": 29165
    },
    {
      "epoch": 17.19929245283019,
      "grad_norm": 3.1296072006225586,
      "learning_rate": 5.857935879327031e-07,
      "loss": 0.2278,
      "num_input_tokens_seen": 19201832,
      "step": 29170
    },
    {
      "epoch": 17.202240566037737,
      "grad_norm": 5.830741882324219,
      "learning_rate": 5.845858419621936e-07,
      "loss": 0.348,
      "num_input_tokens_seen": 19204872,
      "step": 29175
    },
    {
      "epoch": 17.205188679245282,
      "grad_norm": 2.988351821899414,
      "learning_rate": 5.83379264991612e-07,
      "loss": 0.3241,
      "num_input_tokens_seen": 19207688,
      "step": 29180
    },
    {
      "epoch": 17.20813679245283,
      "grad_norm": 3.7516775131225586,
      "learning_rate": 5.821738573404046e-07,
      "loss": 0.2574,
      "num_input_tokens_seen": 19210344,
      "step": 29185
    },
    {
      "epoch": 17.21108490566038,
      "grad_norm": 3.2243430614471436,
      "learning_rate": 5.80969619327707e-07,
      "loss": 0.2531,
      "num_input_tokens_seen": 19213768,
      "step": 29190
    },
    {
      "epoch": 17.214033018867923,
      "grad_norm": 3.6090569496154785,
      "learning_rate": 5.797665512723488e-07,
      "loss": 0.2591,
      "num_input_tokens_seen": 19216840,
      "step": 29195
    },
    {
      "epoch": 17.21698113207547,
      "grad_norm": 3.392146348953247,
      "learning_rate": 5.785646534928452e-07,
      "loss": 0.2949,
      "num_input_tokens_seen": 19220488,
      "step": 29200
    },
    {
      "epoch": 17.21992924528302,
      "grad_norm": 5.257474422454834,
      "learning_rate": 5.77363926307406e-07,
      "loss": 0.2233,
      "num_input_tokens_seen": 19224744,
      "step": 29205
    },
    {
      "epoch": 17.222877358490567,
      "grad_norm": 2.6907546520233154,
      "learning_rate": 5.761643700339281e-07,
      "loss": 0.2865,
      "num_input_tokens_seen": 19227912,
      "step": 29210
    },
    {
      "epoch": 17.225825471698112,
      "grad_norm": 4.239847660064697,
      "learning_rate": 5.749659849899985e-07,
      "loss": 0.3305,
      "num_input_tokens_seen": 19231080,
      "step": 29215
    },
    {
      "epoch": 17.22877358490566,
      "grad_norm": 2.083988904953003,
      "learning_rate": 5.737687714928953e-07,
      "loss": 0.3174,
      "num_input_tokens_seen": 19235176,
      "step": 29220
    },
    {
      "epoch": 17.23172169811321,
      "grad_norm": 3.7099568843841553,
      "learning_rate": 5.725727298595846e-07,
      "loss": 0.3281,
      "num_input_tokens_seen": 19237928,
      "step": 29225
    },
    {
      "epoch": 17.234669811320753,
      "grad_norm": 2.475501775741577,
      "learning_rate": 5.71377860406726e-07,
      "loss": 0.2727,
      "num_input_tokens_seen": 19241352,
      "step": 29230
    },
    {
      "epoch": 17.2376179245283,
      "grad_norm": 7.222882270812988,
      "learning_rate": 5.701841634506655e-07,
      "loss": 0.3739,
      "num_input_tokens_seen": 19244840,
      "step": 29235
    },
    {
      "epoch": 17.24056603773585,
      "grad_norm": 5.813836574554443,
      "learning_rate": 5.689916393074391e-07,
      "loss": 0.3448,
      "num_input_tokens_seen": 19247880,
      "step": 29240
    },
    {
      "epoch": 17.243514150943398,
      "grad_norm": 4.520518779754639,
      "learning_rate": 5.678002882927725e-07,
      "loss": 0.2501,
      "num_input_tokens_seen": 19250664,
      "step": 29245
    },
    {
      "epoch": 17.246462264150942,
      "grad_norm": 7.256585597991943,
      "learning_rate": 5.666101107220811e-07,
      "loss": 0.2344,
      "num_input_tokens_seen": 19253448,
      "step": 29250
    },
    {
      "epoch": 17.24941037735849,
      "grad_norm": 3.0063085556030273,
      "learning_rate": 5.654211069104693e-07,
      "loss": 0.3147,
      "num_input_tokens_seen": 19256520,
      "step": 29255
    },
    {
      "epoch": 17.25235849056604,
      "grad_norm": 3.7320215702056885,
      "learning_rate": 5.642332771727321e-07,
      "loss": 0.2885,
      "num_input_tokens_seen": 19259112,
      "step": 29260
    },
    {
      "epoch": 17.255306603773583,
      "grad_norm": 4.238996505737305,
      "learning_rate": 5.630466218233521e-07,
      "loss": 0.2681,
      "num_input_tokens_seen": 19262024,
      "step": 29265
    },
    {
      "epoch": 17.25825471698113,
      "grad_norm": 3.127275228500366,
      "learning_rate": 5.618611411765007e-07,
      "loss": 0.2663,
      "num_input_tokens_seen": 19266568,
      "step": 29270
    },
    {
      "epoch": 17.26120283018868,
      "grad_norm": 4.312900066375732,
      "learning_rate": 5.606768355460401e-07,
      "loss": 0.3633,
      "num_input_tokens_seen": 19269192,
      "step": 29275
    },
    {
      "epoch": 17.264150943396228,
      "grad_norm": 2.4792563915252686,
      "learning_rate": 5.594937052455191e-07,
      "loss": 0.2589,
      "num_input_tokens_seen": 19273096,
      "step": 29280
    },
    {
      "epoch": 17.267099056603772,
      "grad_norm": 7.178791046142578,
      "learning_rate": 5.583117505881764e-07,
      "loss": 0.208,
      "num_input_tokens_seen": 19276072,
      "step": 29285
    },
    {
      "epoch": 17.27004716981132,
      "grad_norm": 4.676227569580078,
      "learning_rate": 5.571309718869417e-07,
      "loss": 0.3923,
      "num_input_tokens_seen": 19278760,
      "step": 29290
    },
    {
      "epoch": 17.27299528301887,
      "grad_norm": 4.075265407562256,
      "learning_rate": 5.559513694544282e-07,
      "loss": 0.2616,
      "num_input_tokens_seen": 19282856,
      "step": 29295
    },
    {
      "epoch": 17.275943396226417,
      "grad_norm": 3.5127322673797607,
      "learning_rate": 5.547729436029442e-07,
      "loss": 0.3586,
      "num_input_tokens_seen": 19286408,
      "step": 29300
    },
    {
      "epoch": 17.27889150943396,
      "grad_norm": 4.33701753616333,
      "learning_rate": 5.535956946444809e-07,
      "loss": 0.5444,
      "num_input_tokens_seen": 19289768,
      "step": 29305
    },
    {
      "epoch": 17.28183962264151,
      "grad_norm": 5.16246223449707,
      "learning_rate": 5.524196228907203e-07,
      "loss": 0.304,
      "num_input_tokens_seen": 19292392,
      "step": 29310
    },
    {
      "epoch": 17.284787735849058,
      "grad_norm": 2.628082752227783,
      "learning_rate": 5.512447286530326e-07,
      "loss": 0.3056,
      "num_input_tokens_seen": 19296520,
      "step": 29315
    },
    {
      "epoch": 17.287735849056602,
      "grad_norm": 2.2496674060821533,
      "learning_rate": 5.500710122424746e-07,
      "loss": 0.3769,
      "num_input_tokens_seen": 19299976,
      "step": 29320
    },
    {
      "epoch": 17.29068396226415,
      "grad_norm": 2.6982271671295166,
      "learning_rate": 5.488984739697961e-07,
      "loss": 0.2944,
      "num_input_tokens_seen": 19303304,
      "step": 29325
    },
    {
      "epoch": 17.2936320754717,
      "grad_norm": 2.8845081329345703,
      "learning_rate": 5.477271141454294e-07,
      "loss": 0.3015,
      "num_input_tokens_seen": 19306312,
      "step": 29330
    },
    {
      "epoch": 17.296580188679247,
      "grad_norm": 2.7551987171173096,
      "learning_rate": 5.465569330794974e-07,
      "loss": 0.3237,
      "num_input_tokens_seen": 19309416,
      "step": 29335
    },
    {
      "epoch": 17.29952830188679,
      "grad_norm": 7.071700096130371,
      "learning_rate": 5.453879310818105e-07,
      "loss": 0.3546,
      "num_input_tokens_seen": 19312936,
      "step": 29340
    },
    {
      "epoch": 17.30247641509434,
      "grad_norm": 4.614994049072266,
      "learning_rate": 5.442201084618664e-07,
      "loss": 0.3274,
      "num_input_tokens_seen": 19316520,
      "step": 29345
    },
    {
      "epoch": 17.305424528301888,
      "grad_norm": 5.401861667633057,
      "learning_rate": 5.430534655288528e-07,
      "loss": 0.2335,
      "num_input_tokens_seen": 19319944,
      "step": 29350
    },
    {
      "epoch": 17.308372641509433,
      "grad_norm": 3.8011114597320557,
      "learning_rate": 5.418880025916428e-07,
      "loss": 0.3737,
      "num_input_tokens_seen": 19326696,
      "step": 29355
    },
    {
      "epoch": 17.31132075471698,
      "grad_norm": 3.827040672302246,
      "learning_rate": 5.407237199587973e-07,
      "loss": 0.2577,
      "num_input_tokens_seen": 19330248,
      "step": 29360
    },
    {
      "epoch": 17.31426886792453,
      "grad_norm": 1.876695990562439,
      "learning_rate": 5.395606179385654e-07,
      "loss": 0.3215,
      "num_input_tokens_seen": 19337224,
      "step": 29365
    },
    {
      "epoch": 17.317216981132077,
      "grad_norm": 7.1059651374816895,
      "learning_rate": 5.383986968388833e-07,
      "loss": 0.3201,
      "num_input_tokens_seen": 19339976,
      "step": 29370
    },
    {
      "epoch": 17.32016509433962,
      "grad_norm": 7.266638278961182,
      "learning_rate": 5.372379569673736e-07,
      "loss": 0.3633,
      "num_input_tokens_seen": 19342408,
      "step": 29375
    },
    {
      "epoch": 17.32311320754717,
      "grad_norm": 4.968816757202148,
      "learning_rate": 5.360783986313495e-07,
      "loss": 0.2557,
      "num_input_tokens_seen": 19346056,
      "step": 29380
    },
    {
      "epoch": 17.326061320754718,
      "grad_norm": 3.7884254455566406,
      "learning_rate": 5.349200221378076e-07,
      "loss": 0.2692,
      "num_input_tokens_seen": 19348808,
      "step": 29385
    },
    {
      "epoch": 17.329009433962263,
      "grad_norm": 2.5742886066436768,
      "learning_rate": 5.33762827793432e-07,
      "loss": 0.3213,
      "num_input_tokens_seen": 19352264,
      "step": 29390
    },
    {
      "epoch": 17.33195754716981,
      "grad_norm": 3.8796019554138184,
      "learning_rate": 5.326068159045978e-07,
      "loss": 0.3158,
      "num_input_tokens_seen": 19355336,
      "step": 29395
    },
    {
      "epoch": 17.33490566037736,
      "grad_norm": 2.8094427585601807,
      "learning_rate": 5.314519867773621e-07,
      "loss": 0.3525,
      "num_input_tokens_seen": 19359912,
      "step": 29400
    },
    {
      "epoch": 17.337853773584907,
      "grad_norm": 3.3315675258636475,
      "learning_rate": 5.302983407174711e-07,
      "loss": 0.2654,
      "num_input_tokens_seen": 19363912,
      "step": 29405
    },
    {
      "epoch": 17.340801886792452,
      "grad_norm": 3.567293882369995,
      "learning_rate": 5.291458780303572e-07,
      "loss": 0.2146,
      "num_input_tokens_seen": 19366856,
      "step": 29410
    },
    {
      "epoch": 17.34375,
      "grad_norm": 3.330280065536499,
      "learning_rate": 5.279945990211411e-07,
      "loss": 0.3007,
      "num_input_tokens_seen": 19370248,
      "step": 29415
    },
    {
      "epoch": 17.346698113207548,
      "grad_norm": 3.7482454776763916,
      "learning_rate": 5.26844503994628e-07,
      "loss": 0.2787,
      "num_input_tokens_seen": 19373224,
      "step": 29420
    },
    {
      "epoch": 17.349646226415093,
      "grad_norm": 7.059601783752441,
      "learning_rate": 5.25695593255311e-07,
      "loss": 0.4259,
      "num_input_tokens_seen": 19376328,
      "step": 29425
    },
    {
      "epoch": 17.35259433962264,
      "grad_norm": 5.001880645751953,
      "learning_rate": 5.24547867107369e-07,
      "loss": 0.2093,
      "num_input_tokens_seen": 19379432,
      "step": 29430
    },
    {
      "epoch": 17.35554245283019,
      "grad_norm": 4.528671741485596,
      "learning_rate": 5.234013258546672e-07,
      "loss": 0.387,
      "num_input_tokens_seen": 19382504,
      "step": 29435
    },
    {
      "epoch": 17.358490566037737,
      "grad_norm": 7.720860481262207,
      "learning_rate": 5.222559698007563e-07,
      "loss": 0.2679,
      "num_input_tokens_seen": 19385480,
      "step": 29440
    },
    {
      "epoch": 17.361438679245282,
      "grad_norm": 2.4703900814056396,
      "learning_rate": 5.211117992488763e-07,
      "loss": 0.309,
      "num_input_tokens_seen": 19389736,
      "step": 29445
    },
    {
      "epoch": 17.36438679245283,
      "grad_norm": 6.030385494232178,
      "learning_rate": 5.199688145019505e-07,
      "loss": 0.3257,
      "num_input_tokens_seen": 19393224,
      "step": 29450
    },
    {
      "epoch": 17.36733490566038,
      "grad_norm": 3.3421385288238525,
      "learning_rate": 5.188270158625891e-07,
      "loss": 0.3853,
      "num_input_tokens_seen": 19396520,
      "step": 29455
    },
    {
      "epoch": 17.370283018867923,
      "grad_norm": 3.882086753845215,
      "learning_rate": 5.176864036330875e-07,
      "loss": 0.431,
      "num_input_tokens_seen": 19399368,
      "step": 29460
    },
    {
      "epoch": 17.37323113207547,
      "grad_norm": 2.797380208969116,
      "learning_rate": 5.165469781154287e-07,
      "loss": 0.3145,
      "num_input_tokens_seen": 19402760,
      "step": 29465
    },
    {
      "epoch": 17.37617924528302,
      "grad_norm": 3.8866732120513916,
      "learning_rate": 5.154087396112789e-07,
      "loss": 0.2619,
      "num_input_tokens_seen": 19405576,
      "step": 29470
    },
    {
      "epoch": 17.379127358490567,
      "grad_norm": 3.124830484390259,
      "learning_rate": 5.142716884219939e-07,
      "loss": 0.2536,
      "num_input_tokens_seen": 19408424,
      "step": 29475
    },
    {
      "epoch": 17.382075471698112,
      "grad_norm": 4.029104709625244,
      "learning_rate": 5.131358248486118e-07,
      "loss": 0.2993,
      "num_input_tokens_seen": 19411464,
      "step": 29480
    },
    {
      "epoch": 17.38502358490566,
      "grad_norm": 4.295344352722168,
      "learning_rate": 5.120011491918564e-07,
      "loss": 0.3543,
      "num_input_tokens_seen": 19414792,
      "step": 29485
    },
    {
      "epoch": 17.38797169811321,
      "grad_norm": 2.833376169204712,
      "learning_rate": 5.108676617521402e-07,
      "loss": 0.3289,
      "num_input_tokens_seen": 19417480,
      "step": 29490
    },
    {
      "epoch": 17.390919811320753,
      "grad_norm": 4.86480712890625,
      "learning_rate": 5.097353628295571e-07,
      "loss": 0.2378,
      "num_input_tokens_seen": 19421096,
      "step": 29495
    },
    {
      "epoch": 17.3938679245283,
      "grad_norm": 3.5910143852233887,
      "learning_rate": 5.086042527238893e-07,
      "loss": 0.2667,
      "num_input_tokens_seen": 19424360,
      "step": 29500
    },
    {
      "epoch": 17.39681603773585,
      "grad_norm": 4.161156177520752,
      "learning_rate": 5.074743317346009e-07,
      "loss": 0.2091,
      "num_input_tokens_seen": 19427048,
      "step": 29505
    },
    {
      "epoch": 17.399764150943398,
      "grad_norm": 5.885505676269531,
      "learning_rate": 5.063456001608458e-07,
      "loss": 0.2467,
      "num_input_tokens_seen": 19431144,
      "step": 29510
    },
    {
      "epoch": 17.402712264150942,
      "grad_norm": 1.8738832473754883,
      "learning_rate": 5.052180583014599e-07,
      "loss": 0.332,
      "num_input_tokens_seen": 19434664,
      "step": 29515
    },
    {
      "epoch": 17.40566037735849,
      "grad_norm": 4.110084056854248,
      "learning_rate": 5.04091706454965e-07,
      "loss": 0.3449,
      "num_input_tokens_seen": 19437992,
      "step": 29520
    },
    {
      "epoch": 17.40860849056604,
      "grad_norm": 4.678400993347168,
      "learning_rate": 5.029665449195665e-07,
      "loss": 0.4072,
      "num_input_tokens_seen": 19440552,
      "step": 29525
    },
    {
      "epoch": 17.411556603773583,
      "grad_norm": 6.099027156829834,
      "learning_rate": 5.018425739931559e-07,
      "loss": 0.3356,
      "num_input_tokens_seen": 19443464,
      "step": 29530
    },
    {
      "epoch": 17.41450471698113,
      "grad_norm": 4.0941996574401855,
      "learning_rate": 5.007197939733099e-07,
      "loss": 0.2823,
      "num_input_tokens_seen": 19446280,
      "step": 29535
    },
    {
      "epoch": 17.41745283018868,
      "grad_norm": 2.156578302383423,
      "learning_rate": 4.995982051572895e-07,
      "loss": 0.3229,
      "num_input_tokens_seen": 19450344,
      "step": 29540
    },
    {
      "epoch": 17.420400943396228,
      "grad_norm": 5.555599212646484,
      "learning_rate": 4.984778078420405e-07,
      "loss": 0.3578,
      "num_input_tokens_seen": 19453384,
      "step": 29545
    },
    {
      "epoch": 17.423349056603772,
      "grad_norm": 3.5067882537841797,
      "learning_rate": 4.973586023241917e-07,
      "loss": 0.3199,
      "num_input_tokens_seen": 19457512,
      "step": 29550
    },
    {
      "epoch": 17.42629716981132,
      "grad_norm": 6.904305458068848,
      "learning_rate": 4.962405889000588e-07,
      "loss": 0.2731,
      "num_input_tokens_seen": 19460776,
      "step": 29555
    },
    {
      "epoch": 17.42924528301887,
      "grad_norm": 6.336822509765625,
      "learning_rate": 4.951237678656396e-07,
      "loss": 0.3993,
      "num_input_tokens_seen": 19464040,
      "step": 29560
    },
    {
      "epoch": 17.432193396226417,
      "grad_norm": 4.64253044128418,
      "learning_rate": 4.940081395166174e-07,
      "loss": 0.3206,
      "num_input_tokens_seen": 19467592,
      "step": 29565
    },
    {
      "epoch": 17.43514150943396,
      "grad_norm": 7.134698867797852,
      "learning_rate": 4.928937041483606e-07,
      "loss": 0.2979,
      "num_input_tokens_seen": 19470312,
      "step": 29570
    },
    {
      "epoch": 17.43808962264151,
      "grad_norm": 4.73258638381958,
      "learning_rate": 4.917804620559202e-07,
      "loss": 0.3195,
      "num_input_tokens_seen": 19473992,
      "step": 29575
    },
    {
      "epoch": 17.441037735849058,
      "grad_norm": 3.845017671585083,
      "learning_rate": 4.906684135340317e-07,
      "loss": 0.2788,
      "num_input_tokens_seen": 19479656,
      "step": 29580
    },
    {
      "epoch": 17.443985849056602,
      "grad_norm": 3.723254680633545,
      "learning_rate": 4.89557558877114e-07,
      "loss": 0.2832,
      "num_input_tokens_seen": 19482184,
      "step": 29585
    },
    {
      "epoch": 17.44693396226415,
      "grad_norm": 6.7167816162109375,
      "learning_rate": 4.884478983792728e-07,
      "loss": 0.3299,
      "num_input_tokens_seen": 19484968,
      "step": 29590
    },
    {
      "epoch": 17.4498820754717,
      "grad_norm": 4.189873218536377,
      "learning_rate": 4.873394323342939e-07,
      "loss": 0.2405,
      "num_input_tokens_seen": 19487816,
      "step": 29595
    },
    {
      "epoch": 17.452830188679247,
      "grad_norm": 3.174276351928711,
      "learning_rate": 4.86232161035648e-07,
      "loss": 0.2785,
      "num_input_tokens_seen": 19490600,
      "step": 29600
    },
    {
      "epoch": 17.45577830188679,
      "grad_norm": 5.178730487823486,
      "learning_rate": 4.851260847764916e-07,
      "loss": 0.3146,
      "num_input_tokens_seen": 19493064,
      "step": 29605
    },
    {
      "epoch": 17.45872641509434,
      "grad_norm": 4.461727142333984,
      "learning_rate": 4.840212038496622e-07,
      "loss": 0.2817,
      "num_input_tokens_seen": 19497768,
      "step": 29610
    },
    {
      "epoch": 17.461674528301888,
      "grad_norm": 2.595317840576172,
      "learning_rate": 4.82917518547682e-07,
      "loss": 0.4113,
      "num_input_tokens_seen": 19501512,
      "step": 29615
    },
    {
      "epoch": 17.464622641509433,
      "grad_norm": 3.2349748611450195,
      "learning_rate": 4.81815029162756e-07,
      "loss": 0.3375,
      "num_input_tokens_seen": 19504520,
      "step": 29620
    },
    {
      "epoch": 17.46757075471698,
      "grad_norm": 4.250657558441162,
      "learning_rate": 4.807137359867725e-07,
      "loss": 0.3,
      "num_input_tokens_seen": 19507368,
      "step": 29625
    },
    {
      "epoch": 17.47051886792453,
      "grad_norm": 4.991415500640869,
      "learning_rate": 4.79613639311306e-07,
      "loss": 0.3118,
      "num_input_tokens_seen": 19510536,
      "step": 29630
    },
    {
      "epoch": 17.473466981132077,
      "grad_norm": 4.052032947540283,
      "learning_rate": 4.785147394276096e-07,
      "loss": 0.3678,
      "num_input_tokens_seen": 19513128,
      "step": 29635
    },
    {
      "epoch": 17.47641509433962,
      "grad_norm": 3.344099998474121,
      "learning_rate": 4.774170366266223e-07,
      "loss": 0.3211,
      "num_input_tokens_seen": 19515880,
      "step": 29640
    },
    {
      "epoch": 17.47936320754717,
      "grad_norm": 5.393265247344971,
      "learning_rate": 4.763205311989666e-07,
      "loss": 0.2835,
      "num_input_tokens_seen": 19518600,
      "step": 29645
    },
    {
      "epoch": 17.482311320754718,
      "grad_norm": 2.295955181121826,
      "learning_rate": 4.752252234349458e-07,
      "loss": 0.2592,
      "num_input_tokens_seen": 19521800,
      "step": 29650
    },
    {
      "epoch": 17.485259433962263,
      "grad_norm": 4.238221168518066,
      "learning_rate": 4.7413111362454634e-07,
      "loss": 0.3703,
      "num_input_tokens_seen": 19525960,
      "step": 29655
    },
    {
      "epoch": 17.48820754716981,
      "grad_norm": 3.3556714057922363,
      "learning_rate": 4.7303820205744143e-07,
      "loss": 0.341,
      "num_input_tokens_seen": 19531336,
      "step": 29660
    },
    {
      "epoch": 17.49115566037736,
      "grad_norm": 5.571547985076904,
      "learning_rate": 4.7194648902298303e-07,
      "loss": 0.2802,
      "num_input_tokens_seen": 19534120,
      "step": 29665
    },
    {
      "epoch": 17.494103773584907,
      "grad_norm": 5.80798864364624,
      "learning_rate": 4.7085597481020594e-07,
      "loss": 0.3529,
      "num_input_tokens_seen": 19537864,
      "step": 29670
    },
    {
      "epoch": 17.497051886792452,
      "grad_norm": 2.9912471771240234,
      "learning_rate": 4.697666597078293e-07,
      "loss": 0.3253,
      "num_input_tokens_seen": 19541224,
      "step": 29675
    },
    {
      "epoch": 17.5,
      "grad_norm": 3.7187981605529785,
      "learning_rate": 4.6867854400425237e-07,
      "loss": 0.311,
      "num_input_tokens_seen": 19544104,
      "step": 29680
    },
    {
      "epoch": 17.502948113207548,
      "grad_norm": 2.143806219100952,
      "learning_rate": 4.6759162798756084e-07,
      "loss": 0.2602,
      "num_input_tokens_seen": 19548168,
      "step": 29685
    },
    {
      "epoch": 17.505896226415093,
      "grad_norm": 4.705226898193359,
      "learning_rate": 4.6650591194551895e-07,
      "loss": 0.2601,
      "num_input_tokens_seen": 19551976,
      "step": 29690
    },
    {
      "epoch": 17.50884433962264,
      "grad_norm": 3.715590715408325,
      "learning_rate": 4.654213961655757e-07,
      "loss": 0.3683,
      "num_input_tokens_seen": 19554824,
      "step": 29695
    },
    {
      "epoch": 17.51179245283019,
      "grad_norm": 4.254121780395508,
      "learning_rate": 4.6433808093486075e-07,
      "loss": 0.4422,
      "num_input_tokens_seen": 19557608,
      "step": 29700
    },
    {
      "epoch": 17.514740566037737,
      "grad_norm": 3.0493977069854736,
      "learning_rate": 4.63255966540187e-07,
      "loss": 0.2603,
      "num_input_tokens_seen": 19561000,
      "step": 29705
    },
    {
      "epoch": 17.517688679245282,
      "grad_norm": 3.500016450881958,
      "learning_rate": 4.62175053268048e-07,
      "loss": 0.2645,
      "num_input_tokens_seen": 19565576,
      "step": 29710
    },
    {
      "epoch": 17.52063679245283,
      "grad_norm": 3.223698616027832,
      "learning_rate": 4.6109534140462045e-07,
      "loss": 0.3407,
      "num_input_tokens_seen": 19568936,
      "step": 29715
    },
    {
      "epoch": 17.52358490566038,
      "grad_norm": 5.738717555999756,
      "learning_rate": 4.6001683123576226e-07,
      "loss": 0.3318,
      "num_input_tokens_seen": 19572904,
      "step": 29720
    },
    {
      "epoch": 17.526533018867923,
      "grad_norm": 4.319509506225586,
      "learning_rate": 4.589395230470145e-07,
      "loss": 0.2821,
      "num_input_tokens_seen": 19575816,
      "step": 29725
    },
    {
      "epoch": 17.52948113207547,
      "grad_norm": 4.96510648727417,
      "learning_rate": 4.578634171235996e-07,
      "loss": 0.2672,
      "num_input_tokens_seen": 19579016,
      "step": 29730
    },
    {
      "epoch": 17.53242924528302,
      "grad_norm": 4.598548412322998,
      "learning_rate": 4.567885137504202e-07,
      "loss": 0.3453,
      "num_input_tokens_seen": 19582984,
      "step": 29735
    },
    {
      "epoch": 17.535377358490567,
      "grad_norm": 4.397843360900879,
      "learning_rate": 4.55714813212062e-07,
      "loss": 0.3727,
      "num_input_tokens_seen": 19585832,
      "step": 29740
    },
    {
      "epoch": 17.538325471698112,
      "grad_norm": 5.658368110656738,
      "learning_rate": 4.5464231579279206e-07,
      "loss": 0.4499,
      "num_input_tokens_seen": 19588712,
      "step": 29745
    },
    {
      "epoch": 17.54127358490566,
      "grad_norm": 3.6007158756256104,
      "learning_rate": 4.535710217765571e-07,
      "loss": 0.2943,
      "num_input_tokens_seen": 19592328,
      "step": 29750
    },
    {
      "epoch": 17.54422169811321,
      "grad_norm": 4.6998138427734375,
      "learning_rate": 4.5250093144698913e-07,
      "loss": 0.4056,
      "num_input_tokens_seen": 19595272,
      "step": 29755
    },
    {
      "epoch": 17.547169811320753,
      "grad_norm": 4.028368949890137,
      "learning_rate": 4.514320450873988e-07,
      "loss": 0.2969,
      "num_input_tokens_seen": 19598248,
      "step": 29760
    },
    {
      "epoch": 17.5501179245283,
      "grad_norm": 2.417860984802246,
      "learning_rate": 4.503643629807769e-07,
      "loss": 0.2589,
      "num_input_tokens_seen": 19602632,
      "step": 29765
    },
    {
      "epoch": 17.55306603773585,
      "grad_norm": 3.2854342460632324,
      "learning_rate": 4.4929788540979844e-07,
      "loss": 0.357,
      "num_input_tokens_seen": 19606376,
      "step": 29770
    },
    {
      "epoch": 17.556014150943398,
      "grad_norm": 3.2687926292419434,
      "learning_rate": 4.4823261265681596e-07,
      "loss": 0.2627,
      "num_input_tokens_seen": 19609736,
      "step": 29775
    },
    {
      "epoch": 17.558962264150942,
      "grad_norm": 3.065255641937256,
      "learning_rate": 4.471685450038671e-07,
      "loss": 0.3106,
      "num_input_tokens_seen": 19613704,
      "step": 29780
    },
    {
      "epoch": 17.56191037735849,
      "grad_norm": 6.990057468414307,
      "learning_rate": 4.4610568273266706e-07,
      "loss": 0.3188,
      "num_input_tokens_seen": 19616584,
      "step": 29785
    },
    {
      "epoch": 17.56485849056604,
      "grad_norm": 6.237834453582764,
      "learning_rate": 4.450440261246142e-07,
      "loss": 0.2875,
      "num_input_tokens_seen": 19619240,
      "step": 29790
    },
    {
      "epoch": 17.567806603773583,
      "grad_norm": 4.061367034912109,
      "learning_rate": 4.439835754607863e-07,
      "loss": 0.3021,
      "num_input_tokens_seen": 19622312,
      "step": 29795
    },
    {
      "epoch": 17.57075471698113,
      "grad_norm": 2.629948377609253,
      "learning_rate": 4.429243310219422e-07,
      "loss": 0.3804,
      "num_input_tokens_seen": 19629288,
      "step": 29800
    },
    {
      "epoch": 17.57370283018868,
      "grad_norm": 4.159302234649658,
      "learning_rate": 4.418662930885215e-07,
      "loss": 0.2913,
      "num_input_tokens_seen": 19632680,
      "step": 29805
    },
    {
      "epoch": 17.576650943396228,
      "grad_norm": 2.6110029220581055,
      "learning_rate": 4.408094619406439e-07,
      "loss": 0.2566,
      "num_input_tokens_seen": 19635880,
      "step": 29810
    },
    {
      "epoch": 17.579599056603772,
      "grad_norm": 3.5590932369232178,
      "learning_rate": 4.3975383785810954e-07,
      "loss": 0.3008,
      "num_input_tokens_seen": 19638536,
      "step": 29815
    },
    {
      "epoch": 17.58254716981132,
      "grad_norm": 4.521186351776123,
      "learning_rate": 4.3869942112040096e-07,
      "loss": 0.2424,
      "num_input_tokens_seen": 19642472,
      "step": 29820
    },
    {
      "epoch": 17.58549528301887,
      "grad_norm": 3.793438196182251,
      "learning_rate": 4.3764621200667936e-07,
      "loss": 0.2496,
      "num_input_tokens_seen": 19645928,
      "step": 29825
    },
    {
      "epoch": 17.588443396226417,
      "grad_norm": 5.112852573394775,
      "learning_rate": 4.365942107957849e-07,
      "loss": 0.5703,
      "num_input_tokens_seen": 19649128,
      "step": 29830
    },
    {
      "epoch": 17.59139150943396,
      "grad_norm": 3.8940229415893555,
      "learning_rate": 4.35543417766241e-07,
      "loss": 0.2541,
      "num_input_tokens_seen": 19651816,
      "step": 29835
    },
    {
      "epoch": 17.59433962264151,
      "grad_norm": 3.222027540206909,
      "learning_rate": 4.3449383319624785e-07,
      "loss": 0.3981,
      "num_input_tokens_seen": 19655336,
      "step": 29840
    },
    {
      "epoch": 17.597287735849058,
      "grad_norm": 2.8867931365966797,
      "learning_rate": 4.3344545736368926e-07,
      "loss": 0.347,
      "num_input_tokens_seen": 19658792,
      "step": 29845
    },
    {
      "epoch": 17.600235849056602,
      "grad_norm": 2.871699810028076,
      "learning_rate": 4.323982905461266e-07,
      "loss": 0.358,
      "num_input_tokens_seen": 19662056,
      "step": 29850
    },
    {
      "epoch": 17.60318396226415,
      "grad_norm": 6.089107513427734,
      "learning_rate": 4.313523330208019e-07,
      "loss": 0.2482,
      "num_input_tokens_seen": 19664712,
      "step": 29855
    },
    {
      "epoch": 17.6061320754717,
      "grad_norm": 5.088212966918945,
      "learning_rate": 4.303075850646371e-07,
      "loss": 0.2477,
      "num_input_tokens_seen": 19671336,
      "step": 29860
    },
    {
      "epoch": 17.609080188679247,
      "grad_norm": 3.831132650375366,
      "learning_rate": 4.2926404695423305e-07,
      "loss": 0.3539,
      "num_input_tokens_seen": 19674504,
      "step": 29865
    },
    {
      "epoch": 17.61202830188679,
      "grad_norm": 3.865968942642212,
      "learning_rate": 4.282217189658705e-07,
      "loss": 0.3731,
      "num_input_tokens_seen": 19676648,
      "step": 29870
    },
    {
      "epoch": 17.61497641509434,
      "grad_norm": 2.994279384613037,
      "learning_rate": 4.27180601375512e-07,
      "loss": 0.2962,
      "num_input_tokens_seen": 19680104,
      "step": 29875
    },
    {
      "epoch": 17.617924528301888,
      "grad_norm": 4.240746974945068,
      "learning_rate": 4.2614069445879646e-07,
      "loss": 0.3189,
      "num_input_tokens_seen": 19683240,
      "step": 29880
    },
    {
      "epoch": 17.620872641509433,
      "grad_norm": 7.611937999725342,
      "learning_rate": 4.251019984910448e-07,
      "loss": 0.3343,
      "num_input_tokens_seen": 19688040,
      "step": 29885
    },
    {
      "epoch": 17.62382075471698,
      "grad_norm": 3.59690260887146,
      "learning_rate": 4.2406451374725597e-07,
      "loss": 0.2138,
      "num_input_tokens_seen": 19690664,
      "step": 29890
    },
    {
      "epoch": 17.62676886792453,
      "grad_norm": 3.509387969970703,
      "learning_rate": 4.2302824050210855e-07,
      "loss": 0.2724,
      "num_input_tokens_seen": 19694248,
      "step": 29895
    },
    {
      "epoch": 17.629716981132077,
      "grad_norm": 3.3970696926116943,
      "learning_rate": 4.2199317902995974e-07,
      "loss": 0.322,
      "num_input_tokens_seen": 19696552,
      "step": 29900
    },
    {
      "epoch": 17.63266509433962,
      "grad_norm": 3.1113932132720947,
      "learning_rate": 4.209593296048459e-07,
      "loss": 0.3478,
      "num_input_tokens_seen": 19700296,
      "step": 29905
    },
    {
      "epoch": 17.63561320754717,
      "grad_norm": 4.663711071014404,
      "learning_rate": 4.1992669250048524e-07,
      "loss": 0.3821,
      "num_input_tokens_seen": 19702856,
      "step": 29910
    },
    {
      "epoch": 17.638561320754718,
      "grad_norm": 2.97101092338562,
      "learning_rate": 4.188952679902719e-07,
      "loss": 0.2874,
      "num_input_tokens_seen": 19705544,
      "step": 29915
    },
    {
      "epoch": 17.641509433962263,
      "grad_norm": 4.602298736572266,
      "learning_rate": 4.178650563472797e-07,
      "loss": 0.4021,
      "num_input_tokens_seen": 19708520,
      "step": 29920
    },
    {
      "epoch": 17.64445754716981,
      "grad_norm": 10.533122062683105,
      "learning_rate": 4.168360578442615e-07,
      "loss": 0.4028,
      "num_input_tokens_seen": 19711720,
      "step": 29925
    },
    {
      "epoch": 17.64740566037736,
      "grad_norm": 2.6649858951568604,
      "learning_rate": 4.1580827275365e-07,
      "loss": 0.4136,
      "num_input_tokens_seen": 19716200,
      "step": 29930
    },
    {
      "epoch": 17.650353773584907,
      "grad_norm": 3.9207990169525146,
      "learning_rate": 4.147817013475536e-07,
      "loss": 0.302,
      "num_input_tokens_seen": 19719624,
      "step": 29935
    },
    {
      "epoch": 17.653301886792452,
      "grad_norm": 2.148232936859131,
      "learning_rate": 4.1375634389776375e-07,
      "loss": 0.3181,
      "num_input_tokens_seen": 19725896,
      "step": 29940
    },
    {
      "epoch": 17.65625,
      "grad_norm": 2.9869701862335205,
      "learning_rate": 4.127322006757478e-07,
      "loss": 0.2546,
      "num_input_tokens_seen": 19728968,
      "step": 29945
    },
    {
      "epoch": 17.659198113207548,
      "grad_norm": 2.1225473880767822,
      "learning_rate": 4.1170927195265163e-07,
      "loss": 0.1704,
      "num_input_tokens_seen": 19732008,
      "step": 29950
    },
    {
      "epoch": 17.662146226415093,
      "grad_norm": 5.6369805335998535,
      "learning_rate": 4.1068755799930026e-07,
      "loss": 0.2706,
      "num_input_tokens_seen": 19734792,
      "step": 29955
    },
    {
      "epoch": 17.66509433962264,
      "grad_norm": 3.763791084289551,
      "learning_rate": 4.096670590861962e-07,
      "loss": 0.2953,
      "num_input_tokens_seen": 19738728,
      "step": 29960
    },
    {
      "epoch": 17.66804245283019,
      "grad_norm": 3.8790884017944336,
      "learning_rate": 4.086477754835211e-07,
      "loss": 0.3379,
      "num_input_tokens_seen": 19741800,
      "step": 29965
    },
    {
      "epoch": 17.670990566037737,
      "grad_norm": 3.922231674194336,
      "learning_rate": 4.0762970746113517e-07,
      "loss": 0.3077,
      "num_input_tokens_seen": 19744520,
      "step": 29970
    },
    {
      "epoch": 17.673938679245282,
      "grad_norm": 5.03695011138916,
      "learning_rate": 4.0661285528857676e-07,
      "loss": 0.3189,
      "num_input_tokens_seen": 19746632,
      "step": 29975
    },
    {
      "epoch": 17.67688679245283,
      "grad_norm": 4.171802520751953,
      "learning_rate": 4.0559721923506155e-07,
      "loss": 0.3724,
      "num_input_tokens_seen": 19750088,
      "step": 29980
    },
    {
      "epoch": 17.67983490566038,
      "grad_norm": 5.3741679191589355,
      "learning_rate": 4.045827995694834e-07,
      "loss": 0.3247,
      "num_input_tokens_seen": 19752840,
      "step": 29985
    },
    {
      "epoch": 17.682783018867923,
      "grad_norm": 3.56262469291687,
      "learning_rate": 4.035695965604142e-07,
      "loss": 0.2222,
      "num_input_tokens_seen": 19756392,
      "step": 29990
    },
    {
      "epoch": 17.68573113207547,
      "grad_norm": 3.471184015274048,
      "learning_rate": 4.0255761047610365e-07,
      "loss": 0.2544,
      "num_input_tokens_seen": 19759368,
      "step": 29995
    },
    {
      "epoch": 17.68867924528302,
      "grad_norm": 3.6051154136657715,
      "learning_rate": 4.0154684158447864e-07,
      "loss": 0.3438,
      "num_input_tokens_seen": 19763496,
      "step": 30000
    },
    {
      "epoch": 17.691627358490567,
      "grad_norm": 3.8744142055511475,
      "learning_rate": 4.0053729015314623e-07,
      "loss": 0.2484,
      "num_input_tokens_seen": 19767560,
      "step": 30005
    },
    {
      "epoch": 17.694575471698112,
      "grad_norm": 4.841684341430664,
      "learning_rate": 3.9952895644938926e-07,
      "loss": 0.4186,
      "num_input_tokens_seen": 19770664,
      "step": 30010
    },
    {
      "epoch": 17.69752358490566,
      "grad_norm": 4.028140544891357,
      "learning_rate": 3.985218407401681e-07,
      "loss": 0.2302,
      "num_input_tokens_seen": 19772936,
      "step": 30015
    },
    {
      "epoch": 17.70047169811321,
      "grad_norm": 4.625421047210693,
      "learning_rate": 3.975159432921205e-07,
      "loss": 0.2621,
      "num_input_tokens_seen": 19775784,
      "step": 30020
    },
    {
      "epoch": 17.703419811320753,
      "grad_norm": 5.420171737670898,
      "learning_rate": 3.9651126437156294e-07,
      "loss": 0.2319,
      "num_input_tokens_seen": 19778120,
      "step": 30025
    },
    {
      "epoch": 17.7063679245283,
      "grad_norm": 4.396236896514893,
      "learning_rate": 3.9550780424448653e-07,
      "loss": 0.2403,
      "num_input_tokens_seen": 19780872,
      "step": 30030
    },
    {
      "epoch": 17.70931603773585,
      "grad_norm": 5.323135852813721,
      "learning_rate": 3.9450556317656487e-07,
      "loss": 0.3197,
      "num_input_tokens_seen": 19783624,
      "step": 30035
    },
    {
      "epoch": 17.712264150943398,
      "grad_norm": 4.183505535125732,
      "learning_rate": 3.935045414331434e-07,
      "loss": 0.2392,
      "num_input_tokens_seen": 19786856,
      "step": 30040
    },
    {
      "epoch": 17.715212264150942,
      "grad_norm": 3.401715040206909,
      "learning_rate": 3.925047392792475e-07,
      "loss": 0.2502,
      "num_input_tokens_seen": 19789736,
      "step": 30045
    },
    {
      "epoch": 17.71816037735849,
      "grad_norm": 2.9759464263916016,
      "learning_rate": 3.9150615697957917e-07,
      "loss": 0.359,
      "num_input_tokens_seen": 19792680,
      "step": 30050
    },
    {
      "epoch": 17.72110849056604,
      "grad_norm": 6.347644329071045,
      "learning_rate": 3.9050879479851753e-07,
      "loss": 0.3283,
      "num_input_tokens_seen": 19795464,
      "step": 30055
    },
    {
      "epoch": 17.724056603773583,
      "grad_norm": 3.1435301303863525,
      "learning_rate": 3.89512653000117e-07,
      "loss": 0.2638,
      "num_input_tokens_seen": 19798696,
      "step": 30060
    },
    {
      "epoch": 17.72700471698113,
      "grad_norm": 4.00762414932251,
      "learning_rate": 3.8851773184811203e-07,
      "loss": 0.4844,
      "num_input_tokens_seen": 19801256,
      "step": 30065
    },
    {
      "epoch": 17.72995283018868,
      "grad_norm": 3.8779051303863525,
      "learning_rate": 3.8752403160591255e-07,
      "loss": 0.3087,
      "num_input_tokens_seen": 19804616,
      "step": 30070
    },
    {
      "epoch": 17.732900943396228,
      "grad_norm": 3.928114414215088,
      "learning_rate": 3.8653155253660477e-07,
      "loss": 0.377,
      "num_input_tokens_seen": 19808072,
      "step": 30075
    },
    {
      "epoch": 17.735849056603772,
      "grad_norm": 3.8247318267822266,
      "learning_rate": 3.8554029490295073e-07,
      "loss": 0.3582,
      "num_input_tokens_seen": 19810440,
      "step": 30080
    },
    {
      "epoch": 17.73879716981132,
      "grad_norm": 4.218780517578125,
      "learning_rate": 3.8455025896739164e-07,
      "loss": 0.3507,
      "num_input_tokens_seen": 19813736,
      "step": 30085
    },
    {
      "epoch": 17.74174528301887,
      "grad_norm": 3.1730847358703613,
      "learning_rate": 3.8356144499204215e-07,
      "loss": 0.2504,
      "num_input_tokens_seen": 19816552,
      "step": 30090
    },
    {
      "epoch": 17.744693396226417,
      "grad_norm": 3.107004165649414,
      "learning_rate": 3.8257385323869576e-07,
      "loss": 0.3446,
      "num_input_tokens_seen": 19819432,
      "step": 30095
    },
    {
      "epoch": 17.74764150943396,
      "grad_norm": 3.7715349197387695,
      "learning_rate": 3.815874839688222e-07,
      "loss": 0.2953,
      "num_input_tokens_seen": 19823240,
      "step": 30100
    },
    {
      "epoch": 17.75058962264151,
      "grad_norm": 5.338054656982422,
      "learning_rate": 3.8060233744356634e-07,
      "loss": 0.3521,
      "num_input_tokens_seen": 19826120,
      "step": 30105
    },
    {
      "epoch": 17.753537735849058,
      "grad_norm": 3.4981753826141357,
      "learning_rate": 3.796184139237502e-07,
      "loss": 0.2488,
      "num_input_tokens_seen": 19829480,
      "step": 30110
    },
    {
      "epoch": 17.756485849056602,
      "grad_norm": 2.4044511318206787,
      "learning_rate": 3.7863571366987206e-07,
      "loss": 0.3368,
      "num_input_tokens_seen": 19833160,
      "step": 30115
    },
    {
      "epoch": 17.75943396226415,
      "grad_norm": 3.878664970397949,
      "learning_rate": 3.776542369421049e-07,
      "loss": 0.2812,
      "num_input_tokens_seen": 19836744,
      "step": 30120
    },
    {
      "epoch": 17.7623820754717,
      "grad_norm": 4.030726909637451,
      "learning_rate": 3.766739840003003e-07,
      "loss": 0.3167,
      "num_input_tokens_seen": 19839016,
      "step": 30125
    },
    {
      "epoch": 17.765330188679247,
      "grad_norm": 4.084760665893555,
      "learning_rate": 3.756949551039835e-07,
      "loss": 0.3715,
      "num_input_tokens_seen": 19842344,
      "step": 30130
    },
    {
      "epoch": 17.76827830188679,
      "grad_norm": 3.442204236984253,
      "learning_rate": 3.7471715051235757e-07,
      "loss": 0.2821,
      "num_input_tokens_seen": 19846728,
      "step": 30135
    },
    {
      "epoch": 17.77122641509434,
      "grad_norm": 2.6435577869415283,
      "learning_rate": 3.7374057048429947e-07,
      "loss": 0.2732,
      "num_input_tokens_seen": 19849960,
      "step": 30140
    },
    {
      "epoch": 17.774174528301888,
      "grad_norm": 2.47463059425354,
      "learning_rate": 3.7276521527836396e-07,
      "loss": 0.2152,
      "num_input_tokens_seen": 19853480,
      "step": 30145
    },
    {
      "epoch": 17.777122641509433,
      "grad_norm": 3.2162978649139404,
      "learning_rate": 3.717910851527784e-07,
      "loss": 0.3317,
      "num_input_tokens_seen": 19856168,
      "step": 30150
    },
    {
      "epoch": 17.78007075471698,
      "grad_norm": 2.2225043773651123,
      "learning_rate": 3.708181803654498e-07,
      "loss": 0.3423,
      "num_input_tokens_seen": 19859880,
      "step": 30155
    },
    {
      "epoch": 17.78301886792453,
      "grad_norm": 4.132512092590332,
      "learning_rate": 3.6984650117395993e-07,
      "loss": 0.2323,
      "num_input_tokens_seen": 19863080,
      "step": 30160
    },
    {
      "epoch": 17.785966981132077,
      "grad_norm": 3.6657071113586426,
      "learning_rate": 3.688760478355635e-07,
      "loss": 0.1986,
      "num_input_tokens_seen": 19866056,
      "step": 30165
    },
    {
      "epoch": 17.78891509433962,
      "grad_norm": 4.471585750579834,
      "learning_rate": 3.679068206071923e-07,
      "loss": 0.2789,
      "num_input_tokens_seen": 19869192,
      "step": 30170
    },
    {
      "epoch": 17.79186320754717,
      "grad_norm": 4.335415363311768,
      "learning_rate": 3.669388197454532e-07,
      "loss": 0.3625,
      "num_input_tokens_seen": 19872648,
      "step": 30175
    },
    {
      "epoch": 17.794811320754718,
      "grad_norm": 5.638824939727783,
      "learning_rate": 3.6597204550662956e-07,
      "loss": 0.2976,
      "num_input_tokens_seen": 19875432,
      "step": 30180
    },
    {
      "epoch": 17.797759433962263,
      "grad_norm": 2.1408536434173584,
      "learning_rate": 3.650064981466772e-07,
      "loss": 0.2565,
      "num_input_tokens_seen": 19878696,
      "step": 30185
    },
    {
      "epoch": 17.80070754716981,
      "grad_norm": 6.525670528411865,
      "learning_rate": 3.640421779212311e-07,
      "loss": 0.269,
      "num_input_tokens_seen": 19881896,
      "step": 30190
    },
    {
      "epoch": 17.80365566037736,
      "grad_norm": 4.5104498863220215,
      "learning_rate": 3.630790850855986e-07,
      "loss": 0.3456,
      "num_input_tokens_seen": 19884808,
      "step": 30195
    },
    {
      "epoch": 17.806603773584907,
      "grad_norm": 4.072718143463135,
      "learning_rate": 3.62117219894762e-07,
      "loss": 0.25,
      "num_input_tokens_seen": 19887528,
      "step": 30200
    },
    {
      "epoch": 17.809551886792452,
      "grad_norm": 4.165245056152344,
      "learning_rate": 3.611565826033797e-07,
      "loss": 0.2629,
      "num_input_tokens_seen": 19890440,
      "step": 30205
    },
    {
      "epoch": 17.8125,
      "grad_norm": 3.3222293853759766,
      "learning_rate": 3.6019717346578445e-07,
      "loss": 0.2757,
      "num_input_tokens_seen": 19893288,
      "step": 30210
    },
    {
      "epoch": 17.815448113207548,
      "grad_norm": 4.295966625213623,
      "learning_rate": 3.5923899273598293e-07,
      "loss": 0.2749,
      "num_input_tokens_seen": 19896136,
      "step": 30215
    },
    {
      "epoch": 17.818396226415093,
      "grad_norm": 2.7031846046447754,
      "learning_rate": 3.582820406676596e-07,
      "loss": 0.2919,
      "num_input_tokens_seen": 19899880,
      "step": 30220
    },
    {
      "epoch": 17.82134433962264,
      "grad_norm": 3.1243996620178223,
      "learning_rate": 3.5732631751417056e-07,
      "loss": 0.2423,
      "num_input_tokens_seen": 19903944,
      "step": 30225
    },
    {
      "epoch": 17.82429245283019,
      "grad_norm": 5.292044162750244,
      "learning_rate": 3.563718235285485e-07,
      "loss": 0.3418,
      "num_input_tokens_seen": 19906952,
      "step": 30230
    },
    {
      "epoch": 17.827240566037737,
      "grad_norm": 4.170035362243652,
      "learning_rate": 3.5541855896349844e-07,
      "loss": 0.3596,
      "num_input_tokens_seen": 19909512,
      "step": 30235
    },
    {
      "epoch": 17.830188679245282,
      "grad_norm": 3.3870558738708496,
      "learning_rate": 3.544665240714018e-07,
      "loss": 0.4664,
      "num_input_tokens_seen": 19912392,
      "step": 30240
    },
    {
      "epoch": 17.83313679245283,
      "grad_norm": 4.486199855804443,
      "learning_rate": 3.535157191043137e-07,
      "loss": 0.2597,
      "num_input_tokens_seen": 19916104,
      "step": 30245
    },
    {
      "epoch": 17.83608490566038,
      "grad_norm": 4.541314125061035,
      "learning_rate": 3.5256614431396385e-07,
      "loss": 0.3624,
      "num_input_tokens_seen": 19919016,
      "step": 30250
    },
    {
      "epoch": 17.839033018867923,
      "grad_norm": 5.998143672943115,
      "learning_rate": 3.516177999517578e-07,
      "loss": 0.2706,
      "num_input_tokens_seen": 19921992,
      "step": 30255
    },
    {
      "epoch": 17.84198113207547,
      "grad_norm": 3.07456374168396,
      "learning_rate": 3.50670686268772e-07,
      "loss": 0.2668,
      "num_input_tokens_seen": 19925800,
      "step": 30260
    },
    {
      "epoch": 17.84492924528302,
      "grad_norm": 4.133420467376709,
      "learning_rate": 3.497248035157602e-07,
      "loss": 0.2921,
      "num_input_tokens_seen": 19928104,
      "step": 30265
    },
    {
      "epoch": 17.847877358490567,
      "grad_norm": 6.266748905181885,
      "learning_rate": 3.4878015194314773e-07,
      "loss": 0.3811,
      "num_input_tokens_seen": 19930600,
      "step": 30270
    },
    {
      "epoch": 17.850825471698112,
      "grad_norm": 3.6210923194885254,
      "learning_rate": 3.4783673180103617e-07,
      "loss": 0.3804,
      "num_input_tokens_seen": 19934504,
      "step": 30275
    },
    {
      "epoch": 17.85377358490566,
      "grad_norm": 3.5301764011383057,
      "learning_rate": 3.468945433391985e-07,
      "loss": 0.2882,
      "num_input_tokens_seen": 19938024,
      "step": 30280
    },
    {
      "epoch": 17.85672169811321,
      "grad_norm": 3.8964638710021973,
      "learning_rate": 3.459535868070851e-07,
      "loss": 0.2419,
      "num_input_tokens_seen": 19941576,
      "step": 30285
    },
    {
      "epoch": 17.859669811320753,
      "grad_norm": 3.253112554550171,
      "learning_rate": 3.450138624538174e-07,
      "loss": 0.3102,
      "num_input_tokens_seen": 19945384,
      "step": 30290
    },
    {
      "epoch": 17.8626179245283,
      "grad_norm": 8.481961250305176,
      "learning_rate": 3.440753705281913e-07,
      "loss": 0.309,
      "num_input_tokens_seen": 19948904,
      "step": 30295
    },
    {
      "epoch": 17.86556603773585,
      "grad_norm": 1.9623233079910278,
      "learning_rate": 3.4313811127867693e-07,
      "loss": 0.3372,
      "num_input_tokens_seen": 19952200,
      "step": 30300
    },
    {
      "epoch": 17.868514150943398,
      "grad_norm": 3.3775486946105957,
      "learning_rate": 3.4220208495341745e-07,
      "loss": 0.2357,
      "num_input_tokens_seen": 19956072,
      "step": 30305
    },
    {
      "epoch": 17.871462264150942,
      "grad_norm": 6.904544830322266,
      "learning_rate": 3.412672918002291e-07,
      "loss": 0.3974,
      "num_input_tokens_seen": 19959304,
      "step": 30310
    },
    {
      "epoch": 17.87441037735849,
      "grad_norm": 9.38853645324707,
      "learning_rate": 3.403337320666045e-07,
      "loss": 0.4897,
      "num_input_tokens_seen": 19961736,
      "step": 30315
    },
    {
      "epoch": 17.87735849056604,
      "grad_norm": 4.800779342651367,
      "learning_rate": 3.394014059997064e-07,
      "loss": 0.3162,
      "num_input_tokens_seen": 19964776,
      "step": 30320
    },
    {
      "epoch": 17.880306603773583,
      "grad_norm": 4.531450271606445,
      "learning_rate": 3.3847031384637185e-07,
      "loss": 0.2975,
      "num_input_tokens_seen": 19967688,
      "step": 30325
    },
    {
      "epoch": 17.88325471698113,
      "grad_norm": 3.012280225753784,
      "learning_rate": 3.3754045585311147e-07,
      "loss": 0.2639,
      "num_input_tokens_seen": 19971112,
      "step": 30330
    },
    {
      "epoch": 17.88620283018868,
      "grad_norm": 2.7669730186462402,
      "learning_rate": 3.366118322661094e-07,
      "loss": 0.5017,
      "num_input_tokens_seen": 19975528,
      "step": 30335
    },
    {
      "epoch": 17.889150943396228,
      "grad_norm": 5.074180603027344,
      "learning_rate": 3.3568444333122283e-07,
      "loss": 0.4853,
      "num_input_tokens_seen": 19979048,
      "step": 30340
    },
    {
      "epoch": 17.892099056603772,
      "grad_norm": 3.1054131984710693,
      "learning_rate": 3.347582892939816e-07,
      "loss": 0.3321,
      "num_input_tokens_seen": 19982376,
      "step": 30345
    },
    {
      "epoch": 17.89504716981132,
      "grad_norm": 3.942824363708496,
      "learning_rate": 3.338333703995905e-07,
      "loss": 0.5144,
      "num_input_tokens_seen": 19985576,
      "step": 30350
    },
    {
      "epoch": 17.89799528301887,
      "grad_norm": 3.834005355834961,
      "learning_rate": 3.329096868929238e-07,
      "loss": 0.2666,
      "num_input_tokens_seen": 19988456,
      "step": 30355
    },
    {
      "epoch": 17.900943396226417,
      "grad_norm": 6.860472202301025,
      "learning_rate": 3.319872390185325e-07,
      "loss": 0.3393,
      "num_input_tokens_seen": 19991784,
      "step": 30360
    },
    {
      "epoch": 17.90389150943396,
      "grad_norm": 2.2435178756713867,
      "learning_rate": 3.3106602702063727e-07,
      "loss": 0.3493,
      "num_input_tokens_seen": 19995080,
      "step": 30365
    },
    {
      "epoch": 17.90683962264151,
      "grad_norm": 3.7399089336395264,
      "learning_rate": 3.3014605114313316e-07,
      "loss": 0.2601,
      "num_input_tokens_seen": 19997576,
      "step": 30370
    },
    {
      "epoch": 17.909787735849058,
      "grad_norm": 3.4127752780914307,
      "learning_rate": 3.2922731162958744e-07,
      "loss": 0.32,
      "num_input_tokens_seen": 20000968,
      "step": 30375
    },
    {
      "epoch": 17.912735849056602,
      "grad_norm": 3.4336726665496826,
      "learning_rate": 3.2830980872324114e-07,
      "loss": 0.3419,
      "num_input_tokens_seen": 20005000,
      "step": 30380
    },
    {
      "epoch": 17.91568396226415,
      "grad_norm": 4.5645623207092285,
      "learning_rate": 3.2739354266700775e-07,
      "loss": 0.4541,
      "num_input_tokens_seen": 20009288,
      "step": 30385
    },
    {
      "epoch": 17.9186320754717,
      "grad_norm": 3.7522683143615723,
      "learning_rate": 3.264785137034709e-07,
      "loss": 0.3424,
      "num_input_tokens_seen": 20012584,
      "step": 30390
    },
    {
      "epoch": 17.921580188679247,
      "grad_norm": 3.012054920196533,
      "learning_rate": 3.2556472207488977e-07,
      "loss": 0.2917,
      "num_input_tokens_seen": 20015624,
      "step": 30395
    },
    {
      "epoch": 17.92452830188679,
      "grad_norm": 2.518437623977661,
      "learning_rate": 3.246521680231934e-07,
      "loss": 0.2219,
      "num_input_tokens_seen": 20018248,
      "step": 30400
    },
    {
      "epoch": 17.92747641509434,
      "grad_norm": 3.0627400875091553,
      "learning_rate": 3.2374085178998594e-07,
      "loss": 0.2703,
      "num_input_tokens_seen": 20021960,
      "step": 30405
    },
    {
      "epoch": 17.930424528301888,
      "grad_norm": 3.220285654067993,
      "learning_rate": 3.2283077361654145e-07,
      "loss": 0.2776,
      "num_input_tokens_seen": 20024904,
      "step": 30410
    },
    {
      "epoch": 17.933372641509433,
      "grad_norm": 5.04200553894043,
      "learning_rate": 3.2192193374380677e-07,
      "loss": 0.4731,
      "num_input_tokens_seen": 20028008,
      "step": 30415
    },
    {
      "epoch": 17.93632075471698,
      "grad_norm": 4.492512226104736,
      "learning_rate": 3.210143324124021e-07,
      "loss": 0.3391,
      "num_input_tokens_seen": 20030632,
      "step": 30420
    },
    {
      "epoch": 17.93926886792453,
      "grad_norm": 3.0091679096221924,
      "learning_rate": 3.2010796986261805e-07,
      "loss": 0.2719,
      "num_input_tokens_seen": 20033704,
      "step": 30425
    },
    {
      "epoch": 17.942216981132077,
      "grad_norm": 6.948009490966797,
      "learning_rate": 3.1920284633441713e-07,
      "loss": 0.2505,
      "num_input_tokens_seen": 20037160,
      "step": 30430
    },
    {
      "epoch": 17.94516509433962,
      "grad_norm": 2.2780392169952393,
      "learning_rate": 3.1829896206743704e-07,
      "loss": 0.4609,
      "num_input_tokens_seen": 20040616,
      "step": 30435
    },
    {
      "epoch": 17.94811320754717,
      "grad_norm": 6.193571090698242,
      "learning_rate": 3.173963173009825e-07,
      "loss": 0.3146,
      "num_input_tokens_seen": 20043112,
      "step": 30440
    },
    {
      "epoch": 17.951061320754718,
      "grad_norm": 5.058977127075195,
      "learning_rate": 3.164949122740352e-07,
      "loss": 0.2848,
      "num_input_tokens_seen": 20045864,
      "step": 30445
    },
    {
      "epoch": 17.954009433962263,
      "grad_norm": 11.123414993286133,
      "learning_rate": 3.1559474722524406e-07,
      "loss": 0.2781,
      "num_input_tokens_seen": 20049064,
      "step": 30450
    },
    {
      "epoch": 17.95695754716981,
      "grad_norm": 3.662850856781006,
      "learning_rate": 3.146958223929325e-07,
      "loss": 0.2871,
      "num_input_tokens_seen": 20052552,
      "step": 30455
    },
    {
      "epoch": 17.95990566037736,
      "grad_norm": 4.477660179138184,
      "learning_rate": 3.1379813801509454e-07,
      "loss": 0.381,
      "num_input_tokens_seen": 20055336,
      "step": 30460
    },
    {
      "epoch": 17.962853773584907,
      "grad_norm": 6.313104152679443,
      "learning_rate": 3.1290169432939556e-07,
      "loss": 0.4078,
      "num_input_tokens_seen": 20058824,
      "step": 30465
    },
    {
      "epoch": 17.965801886792452,
      "grad_norm": 4.574292182922363,
      "learning_rate": 3.120064915731735e-07,
      "loss": 0.2885,
      "num_input_tokens_seen": 20061512,
      "step": 30470
    },
    {
      "epoch": 17.96875,
      "grad_norm": 4.119393348693848,
      "learning_rate": 3.1111252998343723e-07,
      "loss": 0.3101,
      "num_input_tokens_seen": 20064040,
      "step": 30475
    },
    {
      "epoch": 17.971698113207548,
      "grad_norm": 4.636132717132568,
      "learning_rate": 3.102198097968662e-07,
      "loss": 0.2692,
      "num_input_tokens_seen": 20066920,
      "step": 30480
    },
    {
      "epoch": 17.974646226415093,
      "grad_norm": 3.995454788208008,
      "learning_rate": 3.093283312498124e-07,
      "loss": 0.3757,
      "num_input_tokens_seen": 20070312,
      "step": 30485
    },
    {
      "epoch": 17.97759433962264,
      "grad_norm": 2.2479777336120605,
      "learning_rate": 3.084380945782989e-07,
      "loss": 0.2376,
      "num_input_tokens_seen": 20073864,
      "step": 30490
    },
    {
      "epoch": 17.98054245283019,
      "grad_norm": 4.480945110321045,
      "learning_rate": 3.0754910001801866e-07,
      "loss": 0.3591,
      "num_input_tokens_seen": 20077224,
      "step": 30495
    },
    {
      "epoch": 17.983490566037737,
      "grad_norm": 4.237768173217773,
      "learning_rate": 3.0666134780433786e-07,
      "loss": 0.3332,
      "num_input_tokens_seen": 20080520,
      "step": 30500
    },
    {
      "epoch": 17.986438679245282,
      "grad_norm": 3.763822078704834,
      "learning_rate": 3.0577483817229306e-07,
      "loss": 0.3545,
      "num_input_tokens_seen": 20084008,
      "step": 30505
    },
    {
      "epoch": 17.98938679245283,
      "grad_norm": 5.610185623168945,
      "learning_rate": 3.0488957135659023e-07,
      "loss": 0.2566,
      "num_input_tokens_seen": 20087080,
      "step": 30510
    },
    {
      "epoch": 17.99233490566038,
      "grad_norm": 3.0708556175231934,
      "learning_rate": 3.040055475916087e-07,
      "loss": 0.2916,
      "num_input_tokens_seen": 20090728,
      "step": 30515
    },
    {
      "epoch": 17.995283018867923,
      "grad_norm": 3.7443151473999023,
      "learning_rate": 3.0312276711139675e-07,
      "loss": 0.2435,
      "num_input_tokens_seen": 20093832,
      "step": 30520
    },
    {
      "epoch": 17.99823113207547,
      "grad_norm": 2.6728296279907227,
      "learning_rate": 3.0224123014967353e-07,
      "loss": 0.3118,
      "num_input_tokens_seen": 20096936,
      "step": 30525
    },
    {
      "epoch": 18.0,
      "eval_loss": 0.6042706966400146,
      "eval_runtime": 18.7352,
      "eval_samples_per_second": 90.525,
      "eval_steps_per_second": 22.631,
      "num_input_tokens_seen": 20098848,
      "step": 30528
    },
    {
      "epoch": 18.00117924528302,
      "grad_norm": 3.8188984394073486,
      "learning_rate": 3.013609369398324e-07,
      "loss": 0.237,
      "num_input_tokens_seen": 20100064,
      "step": 30530
    },
    {
      "epoch": 18.004127358490567,
      "grad_norm": 3.6292078495025635,
      "learning_rate": 3.004818877149318e-07,
      "loss": 0.2298,
      "num_input_tokens_seen": 20102912,
      "step": 30535
    },
    {
      "epoch": 18.007075471698112,
      "grad_norm": 3.201230525970459,
      "learning_rate": 2.9960408270770624e-07,
      "loss": 0.3178,
      "num_input_tokens_seen": 20105408,
      "step": 30540
    },
    {
      "epoch": 18.01002358490566,
      "grad_norm": 3.5176172256469727,
      "learning_rate": 2.9872752215055755e-07,
      "loss": 0.3001,
      "num_input_tokens_seen": 20108160,
      "step": 30545
    },
    {
      "epoch": 18.01297169811321,
      "grad_norm": 2.853154420852661,
      "learning_rate": 2.9785220627555844e-07,
      "loss": 0.3134,
      "num_input_tokens_seen": 20112736,
      "step": 30550
    },
    {
      "epoch": 18.015919811320753,
      "grad_norm": 4.791630744934082,
      "learning_rate": 2.9697813531445295e-07,
      "loss": 0.3682,
      "num_input_tokens_seen": 20115840,
      "step": 30555
    },
    {
      "epoch": 18.0188679245283,
      "grad_norm": 6.019406795501709,
      "learning_rate": 2.9610530949865433e-07,
      "loss": 0.2745,
      "num_input_tokens_seen": 20119200,
      "step": 30560
    },
    {
      "epoch": 18.02181603773585,
      "grad_norm": 3.148571014404297,
      "learning_rate": 2.952337290592483e-07,
      "loss": 0.3283,
      "num_input_tokens_seen": 20122720,
      "step": 30565
    },
    {
      "epoch": 18.024764150943398,
      "grad_norm": 4.8672776222229,
      "learning_rate": 2.9436339422698913e-07,
      "loss": 0.266,
      "num_input_tokens_seen": 20125728,
      "step": 30570
    },
    {
      "epoch": 18.027712264150942,
      "grad_norm": 4.219769477844238,
      "learning_rate": 2.934943052323008e-07,
      "loss": 0.2899,
      "num_input_tokens_seen": 20129216,
      "step": 30575
    },
    {
      "epoch": 18.03066037735849,
      "grad_norm": 1.7142560482025146,
      "learning_rate": 2.926264623052799e-07,
      "loss": 0.4901,
      "num_input_tokens_seen": 20135008,
      "step": 30580
    },
    {
      "epoch": 18.03360849056604,
      "grad_norm": 4.376192569732666,
      "learning_rate": 2.9175986567569036e-07,
      "loss": 0.2576,
      "num_input_tokens_seen": 20138272,
      "step": 30585
    },
    {
      "epoch": 18.036556603773583,
      "grad_norm": 3.883464813232422,
      "learning_rate": 2.9089451557296755e-07,
      "loss": 0.3203,
      "num_input_tokens_seen": 20141536,
      "step": 30590
    },
    {
      "epoch": 18.03950471698113,
      "grad_norm": 3.6544694900512695,
      "learning_rate": 2.9003041222621706e-07,
      "loss": 0.2677,
      "num_input_tokens_seen": 20145728,
      "step": 30595
    },
    {
      "epoch": 18.04245283018868,
      "grad_norm": 4.156112194061279,
      "learning_rate": 2.8916755586421375e-07,
      "loss": 0.3314,
      "num_input_tokens_seen": 20148544,
      "step": 30600
    },
    {
      "epoch": 18.045400943396228,
      "grad_norm": 2.4669198989868164,
      "learning_rate": 2.883059467154031e-07,
      "loss": 0.333,
      "num_input_tokens_seen": 20151904,
      "step": 30605
    },
    {
      "epoch": 18.048349056603772,
      "grad_norm": 5.6794328689575195,
      "learning_rate": 2.8744558500789887e-07,
      "loss": 0.335,
      "num_input_tokens_seen": 20155264,
      "step": 30610
    },
    {
      "epoch": 18.05129716981132,
      "grad_norm": 4.428301811218262,
      "learning_rate": 2.8658647096948546e-07,
      "loss": 0.2517,
      "num_input_tokens_seen": 20158240,
      "step": 30615
    },
    {
      "epoch": 18.05424528301887,
      "grad_norm": 4.884956359863281,
      "learning_rate": 2.8572860482761813e-07,
      "loss": 0.2661,
      "num_input_tokens_seen": 20161344,
      "step": 30620
    },
    {
      "epoch": 18.057193396226417,
      "grad_norm": 4.127658843994141,
      "learning_rate": 2.8487198680942017e-07,
      "loss": 0.282,
      "num_input_tokens_seen": 20164192,
      "step": 30625
    },
    {
      "epoch": 18.06014150943396,
      "grad_norm": 3.39548921585083,
      "learning_rate": 2.840166171416836e-07,
      "loss": 0.3425,
      "num_input_tokens_seen": 20167456,
      "step": 30630
    },
    {
      "epoch": 18.06308962264151,
      "grad_norm": 2.6275031566619873,
      "learning_rate": 2.8316249605087386e-07,
      "loss": 0.2527,
      "num_input_tokens_seen": 20169920,
      "step": 30635
    },
    {
      "epoch": 18.066037735849058,
      "grad_norm": 4.928817272186279,
      "learning_rate": 2.823096237631212e-07,
      "loss": 0.494,
      "num_input_tokens_seen": 20172896,
      "step": 30640
    },
    {
      "epoch": 18.068985849056602,
      "grad_norm": 3.5940043926239014,
      "learning_rate": 2.814580005042283e-07,
      "loss": 0.2346,
      "num_input_tokens_seen": 20177024,
      "step": 30645
    },
    {
      "epoch": 18.07193396226415,
      "grad_norm": 3.818828821182251,
      "learning_rate": 2.8060762649966435e-07,
      "loss": 0.3488,
      "num_input_tokens_seen": 20180416,
      "step": 30650
    },
    {
      "epoch": 18.0748820754717,
      "grad_norm": 2.8055102825164795,
      "learning_rate": 2.797585019745713e-07,
      "loss": 0.2322,
      "num_input_tokens_seen": 20183840,
      "step": 30655
    },
    {
      "epoch": 18.077830188679247,
      "grad_norm": 3.75308895111084,
      "learning_rate": 2.789106271537584e-07,
      "loss": 0.4089,
      "num_input_tokens_seen": 20186784,
      "step": 30660
    },
    {
      "epoch": 18.08077830188679,
      "grad_norm": 2.7944698333740234,
      "learning_rate": 2.780640022617037e-07,
      "loss": 0.2981,
      "num_input_tokens_seen": 20189440,
      "step": 30665
    },
    {
      "epoch": 18.08372641509434,
      "grad_norm": 3.6440486907958984,
      "learning_rate": 2.772186275225547e-07,
      "loss": 0.3085,
      "num_input_tokens_seen": 20191936,
      "step": 30670
    },
    {
      "epoch": 18.086674528301888,
      "grad_norm": 3.681472063064575,
      "learning_rate": 2.7637450316012836e-07,
      "loss": 0.3451,
      "num_input_tokens_seen": 20196288,
      "step": 30675
    },
    {
      "epoch": 18.089622641509433,
      "grad_norm": 4.540743350982666,
      "learning_rate": 2.755316293979088e-07,
      "loss": 0.3475,
      "num_input_tokens_seen": 20203232,
      "step": 30680
    },
    {
      "epoch": 18.09257075471698,
      "grad_norm": 2.684149742126465,
      "learning_rate": 2.7469000645905295e-07,
      "loss": 0.3314,
      "num_input_tokens_seen": 20206752,
      "step": 30685
    },
    {
      "epoch": 18.09551886792453,
      "grad_norm": 3.48811936378479,
      "learning_rate": 2.738496345663827e-07,
      "loss": 0.3292,
      "num_input_tokens_seen": 20210400,
      "step": 30690
    },
    {
      "epoch": 18.098466981132077,
      "grad_norm": 2.8548920154571533,
      "learning_rate": 2.7301051394239e-07,
      "loss": 0.2567,
      "num_input_tokens_seen": 20213664,
      "step": 30695
    },
    {
      "epoch": 18.10141509433962,
      "grad_norm": 3.4588208198547363,
      "learning_rate": 2.72172644809236e-07,
      "loss": 0.2703,
      "num_input_tokens_seen": 20217056,
      "step": 30700
    },
    {
      "epoch": 18.10436320754717,
      "grad_norm": 3.721419095993042,
      "learning_rate": 2.7133602738875e-07,
      "loss": 0.242,
      "num_input_tokens_seen": 20220480,
      "step": 30705
    },
    {
      "epoch": 18.107311320754718,
      "grad_norm": 5.749950408935547,
      "learning_rate": 2.7050066190242976e-07,
      "loss": 0.315,
      "num_input_tokens_seen": 20227424,
      "step": 30710
    },
    {
      "epoch": 18.110259433962263,
      "grad_norm": 3.5700979232788086,
      "learning_rate": 2.696665485714428e-07,
      "loss": 0.309,
      "num_input_tokens_seen": 20231424,
      "step": 30715
    },
    {
      "epoch": 18.11320754716981,
      "grad_norm": 2.326014995574951,
      "learning_rate": 2.6883368761662367e-07,
      "loss": 0.295,
      "num_input_tokens_seen": 20234464,
      "step": 30720
    },
    {
      "epoch": 18.11615566037736,
      "grad_norm": 3.306551694869995,
      "learning_rate": 2.680020792584759e-07,
      "loss": 0.435,
      "num_input_tokens_seen": 20237888,
      "step": 30725
    },
    {
      "epoch": 18.119103773584907,
      "grad_norm": 3.4027562141418457,
      "learning_rate": 2.6717172371717113e-07,
      "loss": 0.3982,
      "num_input_tokens_seen": 20240640,
      "step": 30730
    },
    {
      "epoch": 18.122051886792452,
      "grad_norm": 3.5082497596740723,
      "learning_rate": 2.663426212125503e-07,
      "loss": 0.4454,
      "num_input_tokens_seen": 20244224,
      "step": 30735
    },
    {
      "epoch": 18.125,
      "grad_norm": 3.5127034187316895,
      "learning_rate": 2.655147719641216e-07,
      "loss": 0.3238,
      "num_input_tokens_seen": 20248128,
      "step": 30740
    },
    {
      "epoch": 18.127948113207548,
      "grad_norm": 8.180127143859863,
      "learning_rate": 2.646881761910602e-07,
      "loss": 0.3743,
      "num_input_tokens_seen": 20252160,
      "step": 30745
    },
    {
      "epoch": 18.130896226415093,
      "grad_norm": 5.3620123863220215,
      "learning_rate": 2.638628341122135e-07,
      "loss": 0.2597,
      "num_input_tokens_seen": 20255488,
      "step": 30750
    },
    {
      "epoch": 18.13384433962264,
      "grad_norm": 3.4856011867523193,
      "learning_rate": 2.6303874594609314e-07,
      "loss": 0.2966,
      "num_input_tokens_seen": 20258720,
      "step": 30755
    },
    {
      "epoch": 18.13679245283019,
      "grad_norm": 3.5219902992248535,
      "learning_rate": 2.622159119108797e-07,
      "loss": 0.3143,
      "num_input_tokens_seen": 20261728,
      "step": 30760
    },
    {
      "epoch": 18.139740566037737,
      "grad_norm": 3.437431812286377,
      "learning_rate": 2.6139433222442226e-07,
      "loss": 0.1921,
      "num_input_tokens_seen": 20265696,
      "step": 30765
    },
    {
      "epoch": 18.142688679245282,
      "grad_norm": 4.516801834106445,
      "learning_rate": 2.6057400710423787e-07,
      "loss": 0.2661,
      "num_input_tokens_seen": 20269248,
      "step": 30770
    },
    {
      "epoch": 18.14563679245283,
      "grad_norm": 4.103306770324707,
      "learning_rate": 2.5975493676751004e-07,
      "loss": 0.3622,
      "num_input_tokens_seen": 20272896,
      "step": 30775
    },
    {
      "epoch": 18.14858490566038,
      "grad_norm": 3.567370891571045,
      "learning_rate": 2.589371214310926e-07,
      "loss": 0.2917,
      "num_input_tokens_seen": 20275744,
      "step": 30780
    },
    {
      "epoch": 18.151533018867923,
      "grad_norm": 3.456923246383667,
      "learning_rate": 2.581205613115051e-07,
      "loss": 0.2928,
      "num_input_tokens_seen": 20278368,
      "step": 30785
    },
    {
      "epoch": 18.15448113207547,
      "grad_norm": 3.9619452953338623,
      "learning_rate": 2.573052566249357e-07,
      "loss": 0.3468,
      "num_input_tokens_seen": 20281120,
      "step": 30790
    },
    {
      "epoch": 18.15742924528302,
      "grad_norm": 3.9924521446228027,
      "learning_rate": 2.5649120758723945e-07,
      "loss": 0.3574,
      "num_input_tokens_seen": 20283712,
      "step": 30795
    },
    {
      "epoch": 18.160377358490567,
      "grad_norm": 3.767536163330078,
      "learning_rate": 2.5567841441393906e-07,
      "loss": 0.2377,
      "num_input_tokens_seen": 20287072,
      "step": 30800
    },
    {
      "epoch": 18.163325471698112,
      "grad_norm": 5.446342468261719,
      "learning_rate": 2.548668773202245e-07,
      "loss": 0.3674,
      "num_input_tokens_seen": 20290816,
      "step": 30805
    },
    {
      "epoch": 18.16627358490566,
      "grad_norm": 6.483790874481201,
      "learning_rate": 2.5405659652095573e-07,
      "loss": 0.2954,
      "num_input_tokens_seen": 20293920,
      "step": 30810
    },
    {
      "epoch": 18.16922169811321,
      "grad_norm": 3.2987546920776367,
      "learning_rate": 2.5324757223065655e-07,
      "loss": 0.3877,
      "num_input_tokens_seen": 20298720,
      "step": 30815
    },
    {
      "epoch": 18.172169811320753,
      "grad_norm": 2.680154800415039,
      "learning_rate": 2.524398046635207e-07,
      "loss": 0.2492,
      "num_input_tokens_seen": 20303424,
      "step": 30820
    },
    {
      "epoch": 18.1751179245283,
      "grad_norm": 3.6693553924560547,
      "learning_rate": 2.51633294033406e-07,
      "loss": 0.2431,
      "num_input_tokens_seen": 20306400,
      "step": 30825
    },
    {
      "epoch": 18.17806603773585,
      "grad_norm": 9.233122825622559,
      "learning_rate": 2.5082804055384214e-07,
      "loss": 0.3308,
      "num_input_tokens_seen": 20309440,
      "step": 30830
    },
    {
      "epoch": 18.181014150943398,
      "grad_norm": 3.578449249267578,
      "learning_rate": 2.50024044438022e-07,
      "loss": 0.3599,
      "num_input_tokens_seen": 20312736,
      "step": 30835
    },
    {
      "epoch": 18.183962264150942,
      "grad_norm": 6.937576770782471,
      "learning_rate": 2.492213058988069e-07,
      "loss": 0.2549,
      "num_input_tokens_seen": 20315424,
      "step": 30840
    },
    {
      "epoch": 18.18691037735849,
      "grad_norm": 4.6421074867248535,
      "learning_rate": 2.4841982514872633e-07,
      "loss": 0.3904,
      "num_input_tokens_seen": 20318176,
      "step": 30845
    },
    {
      "epoch": 18.18985849056604,
      "grad_norm": 3.042956590652466,
      "learning_rate": 2.4761960239997497e-07,
      "loss": 0.4196,
      "num_input_tokens_seen": 20324320,
      "step": 30850
    },
    {
      "epoch": 18.192806603773583,
      "grad_norm": 3.5896599292755127,
      "learning_rate": 2.4682063786441556e-07,
      "loss": 0.1906,
      "num_input_tokens_seen": 20327136,
      "step": 30855
    },
    {
      "epoch": 18.19575471698113,
      "grad_norm": 4.945327281951904,
      "learning_rate": 2.460229317535778e-07,
      "loss": 0.2517,
      "num_input_tokens_seen": 20330816,
      "step": 30860
    },
    {
      "epoch": 18.19870283018868,
      "grad_norm": 4.5661516189575195,
      "learning_rate": 2.4522648427865725e-07,
      "loss": 0.2761,
      "num_input_tokens_seen": 20333696,
      "step": 30865
    },
    {
      "epoch": 18.201650943396228,
      "grad_norm": 2.898955821990967,
      "learning_rate": 2.444312956505163e-07,
      "loss": 0.4167,
      "num_input_tokens_seen": 20336768,
      "step": 30870
    },
    {
      "epoch": 18.204599056603772,
      "grad_norm": 4.249890327453613,
      "learning_rate": 2.4363736607968537e-07,
      "loss": 0.2687,
      "num_input_tokens_seen": 20339552,
      "step": 30875
    },
    {
      "epoch": 18.20754716981132,
      "grad_norm": 3.0976948738098145,
      "learning_rate": 2.428446957763608e-07,
      "loss": 0.2761,
      "num_input_tokens_seen": 20343136,
      "step": 30880
    },
    {
      "epoch": 18.21049528301887,
      "grad_norm": 4.360994338989258,
      "learning_rate": 2.4205328495040535e-07,
      "loss": 0.3296,
      "num_input_tokens_seen": 20345728,
      "step": 30885
    },
    {
      "epoch": 18.213443396226417,
      "grad_norm": 6.214707374572754,
      "learning_rate": 2.412631338113486e-07,
      "loss": 0.2712,
      "num_input_tokens_seen": 20348832,
      "step": 30890
    },
    {
      "epoch": 18.21639150943396,
      "grad_norm": 4.472646236419678,
      "learning_rate": 2.404742425683848e-07,
      "loss": 0.3167,
      "num_input_tokens_seen": 20352896,
      "step": 30895
    },
    {
      "epoch": 18.21933962264151,
      "grad_norm": 3.883723735809326,
      "learning_rate": 2.3968661143037864e-07,
      "loss": 0.3657,
      "num_input_tokens_seen": 20355584,
      "step": 30900
    },
    {
      "epoch": 18.222287735849058,
      "grad_norm": 3.0029518604278564,
      "learning_rate": 2.3890024060585823e-07,
      "loss": 0.393,
      "num_input_tokens_seen": 20359072,
      "step": 30905
    },
    {
      "epoch": 18.225235849056602,
      "grad_norm": 3.6400465965270996,
      "learning_rate": 2.3811513030301826e-07,
      "loss": 0.3193,
      "num_input_tokens_seen": 20361696,
      "step": 30910
    },
    {
      "epoch": 18.22818396226415,
      "grad_norm": 6.030528545379639,
      "learning_rate": 2.373312807297201e-07,
      "loss": 0.2675,
      "num_input_tokens_seen": 20365888,
      "step": 30915
    },
    {
      "epoch": 18.2311320754717,
      "grad_norm": 4.163887023925781,
      "learning_rate": 2.3654869209349007e-07,
      "loss": 0.3274,
      "num_input_tokens_seen": 20369120,
      "step": 30920
    },
    {
      "epoch": 18.234080188679247,
      "grad_norm": 3.2333908081054688,
      "learning_rate": 2.357673646015246e-07,
      "loss": 0.2737,
      "num_input_tokens_seen": 20372000,
      "step": 30925
    },
    {
      "epoch": 18.23702830188679,
      "grad_norm": 3.4262094497680664,
      "learning_rate": 2.3498729846068103e-07,
      "loss": 0.3645,
      "num_input_tokens_seen": 20375104,
      "step": 30930
    },
    {
      "epoch": 18.23997641509434,
      "grad_norm": 2.8185036182403564,
      "learning_rate": 2.342084938774869e-07,
      "loss": 0.2738,
      "num_input_tokens_seen": 20377472,
      "step": 30935
    },
    {
      "epoch": 18.242924528301888,
      "grad_norm": 5.589546203613281,
      "learning_rate": 2.334309510581334e-07,
      "loss": 0.2884,
      "num_input_tokens_seen": 20379744,
      "step": 30940
    },
    {
      "epoch": 18.245872641509433,
      "grad_norm": 3.081677198410034,
      "learning_rate": 2.3265467020847864e-07,
      "loss": 0.2372,
      "num_input_tokens_seen": 20382144,
      "step": 30945
    },
    {
      "epoch": 18.24882075471698,
      "grad_norm": 5.550468921661377,
      "learning_rate": 2.31879651534046e-07,
      "loss": 0.3478,
      "num_input_tokens_seen": 20385344,
      "step": 30950
    },
    {
      "epoch": 18.25176886792453,
      "grad_norm": 6.823090553283691,
      "learning_rate": 2.311058952400247e-07,
      "loss": 0.3735,
      "num_input_tokens_seen": 20388224,
      "step": 30955
    },
    {
      "epoch": 18.254716981132077,
      "grad_norm": 4.398336410522461,
      "learning_rate": 2.3033340153127026e-07,
      "loss": 0.2429,
      "num_input_tokens_seen": 20391040,
      "step": 30960
    },
    {
      "epoch": 18.25766509433962,
      "grad_norm": 4.11381721496582,
      "learning_rate": 2.295621706123041e-07,
      "loss": 0.2721,
      "num_input_tokens_seen": 20394208,
      "step": 30965
    },
    {
      "epoch": 18.26061320754717,
      "grad_norm": 2.8214895725250244,
      "learning_rate": 2.287922026873135e-07,
      "loss": 0.3591,
      "num_input_tokens_seen": 20397248,
      "step": 30970
    },
    {
      "epoch": 18.263561320754718,
      "grad_norm": 10.338888168334961,
      "learning_rate": 2.2802349796014923e-07,
      "loss": 0.303,
      "num_input_tokens_seen": 20401408,
      "step": 30975
    },
    {
      "epoch": 18.266509433962263,
      "grad_norm": 4.391123294830322,
      "learning_rate": 2.2725605663433013e-07,
      "loss": 0.308,
      "num_input_tokens_seen": 20405824,
      "step": 30980
    },
    {
      "epoch": 18.26945754716981,
      "grad_norm": 4.176970958709717,
      "learning_rate": 2.264898789130393e-07,
      "loss": 0.2213,
      "num_input_tokens_seen": 20408992,
      "step": 30985
    },
    {
      "epoch": 18.27240566037736,
      "grad_norm": 4.413639545440674,
      "learning_rate": 2.2572496499912554e-07,
      "loss": 0.2893,
      "num_input_tokens_seen": 20412320,
      "step": 30990
    },
    {
      "epoch": 18.275353773584907,
      "grad_norm": 4.422268867492676,
      "learning_rate": 2.2496131509510354e-07,
      "loss": 0.272,
      "num_input_tokens_seen": 20415488,
      "step": 30995
    },
    {
      "epoch": 18.278301886792452,
      "grad_norm": 3.1491551399230957,
      "learning_rate": 2.2419892940315268e-07,
      "loss": 0.2071,
      "num_input_tokens_seen": 20419136,
      "step": 31000
    },
    {
      "epoch": 18.28125,
      "grad_norm": 3.144390106201172,
      "learning_rate": 2.2343780812511819e-07,
      "loss": 0.284,
      "num_input_tokens_seen": 20422112,
      "step": 31005
    },
    {
      "epoch": 18.284198113207548,
      "grad_norm": 2.7935500144958496,
      "learning_rate": 2.2267795146250936e-07,
      "loss": 0.3187,
      "num_input_tokens_seen": 20424896,
      "step": 31010
    },
    {
      "epoch": 18.287146226415093,
      "grad_norm": 1.3718384504318237,
      "learning_rate": 2.2191935961650146e-07,
      "loss": 0.3575,
      "num_input_tokens_seen": 20427392,
      "step": 31015
    },
    {
      "epoch": 18.29009433962264,
      "grad_norm": 5.452826023101807,
      "learning_rate": 2.2116203278793603e-07,
      "loss": 0.3512,
      "num_input_tokens_seen": 20431168,
      "step": 31020
    },
    {
      "epoch": 18.29304245283019,
      "grad_norm": 4.376065254211426,
      "learning_rate": 2.2040597117731766e-07,
      "loss": 0.2465,
      "num_input_tokens_seen": 20433824,
      "step": 31025
    },
    {
      "epoch": 18.295990566037737,
      "grad_norm": 3.346956253051758,
      "learning_rate": 2.1965117498481793e-07,
      "loss": 0.279,
      "num_input_tokens_seen": 20436864,
      "step": 31030
    },
    {
      "epoch": 18.298938679245282,
      "grad_norm": 3.6924564838409424,
      "learning_rate": 2.188976444102714e-07,
      "loss": 0.3014,
      "num_input_tokens_seen": 20440224,
      "step": 31035
    },
    {
      "epoch": 18.30188679245283,
      "grad_norm": 4.471794128417969,
      "learning_rate": 2.181453796531796e-07,
      "loss": 0.4008,
      "num_input_tokens_seen": 20443232,
      "step": 31040
    },
    {
      "epoch": 18.30483490566038,
      "grad_norm": 5.092248439788818,
      "learning_rate": 2.1739438091270658e-07,
      "loss": 0.2953,
      "num_input_tokens_seen": 20445920,
      "step": 31045
    },
    {
      "epoch": 18.307783018867923,
      "grad_norm": 3.185288429260254,
      "learning_rate": 2.1664464838768329e-07,
      "loss": 0.3937,
      "num_input_tokens_seen": 20449088,
      "step": 31050
    },
    {
      "epoch": 18.31073113207547,
      "grad_norm": 3.649665355682373,
      "learning_rate": 2.1589618227660426e-07,
      "loss": 0.2997,
      "num_input_tokens_seen": 20452224,
      "step": 31055
    },
    {
      "epoch": 18.31367924528302,
      "grad_norm": 4.172647476196289,
      "learning_rate": 2.151489827776293e-07,
      "loss": 0.2861,
      "num_input_tokens_seen": 20455392,
      "step": 31060
    },
    {
      "epoch": 18.316627358490567,
      "grad_norm": 2.584599733352661,
      "learning_rate": 2.1440305008858298e-07,
      "loss": 0.2955,
      "num_input_tokens_seen": 20458336,
      "step": 31065
    },
    {
      "epoch": 18.319575471698112,
      "grad_norm": 3.016732931137085,
      "learning_rate": 2.1365838440695397e-07,
      "loss": 0.3376,
      "num_input_tokens_seen": 20460960,
      "step": 31070
    },
    {
      "epoch": 18.32252358490566,
      "grad_norm": 2.2386207580566406,
      "learning_rate": 2.129149859298957e-07,
      "loss": 0.2725,
      "num_input_tokens_seen": 20463744,
      "step": 31075
    },
    {
      "epoch": 18.32547169811321,
      "grad_norm": 3.2483222484588623,
      "learning_rate": 2.1217285485422622e-07,
      "loss": 0.3473,
      "num_input_tokens_seen": 20466784,
      "step": 31080
    },
    {
      "epoch": 18.328419811320753,
      "grad_norm": 3.1848278045654297,
      "learning_rate": 2.114319913764268e-07,
      "loss": 0.1903,
      "num_input_tokens_seen": 20470784,
      "step": 31085
    },
    {
      "epoch": 18.3313679245283,
      "grad_norm": 6.460749626159668,
      "learning_rate": 2.10692395692646e-07,
      "loss": 0.3881,
      "num_input_tokens_seen": 20473568,
      "step": 31090
    },
    {
      "epoch": 18.33431603773585,
      "grad_norm": 3.343663215637207,
      "learning_rate": 2.0995406799869444e-07,
      "loss": 0.2965,
      "num_input_tokens_seen": 20476896,
      "step": 31095
    },
    {
      "epoch": 18.337264150943398,
      "grad_norm": 2.578573703765869,
      "learning_rate": 2.0921700849004743e-07,
      "loss": 0.3841,
      "num_input_tokens_seen": 20480288,
      "step": 31100
    },
    {
      "epoch": 18.340212264150942,
      "grad_norm": 3.473708391189575,
      "learning_rate": 2.084812173618439e-07,
      "loss": 0.3356,
      "num_input_tokens_seen": 20483456,
      "step": 31105
    },
    {
      "epoch": 18.34316037735849,
      "grad_norm": 5.837169647216797,
      "learning_rate": 2.0774669480888853e-07,
      "loss": 0.2546,
      "num_input_tokens_seen": 20486048,
      "step": 31110
    },
    {
      "epoch": 18.34610849056604,
      "grad_norm": 3.382554769515991,
      "learning_rate": 2.0701344102564912e-07,
      "loss": 0.2331,
      "num_input_tokens_seen": 20489216,
      "step": 31115
    },
    {
      "epoch": 18.349056603773583,
      "grad_norm": 3.5535361766815186,
      "learning_rate": 2.062814562062576e-07,
      "loss": 0.2636,
      "num_input_tokens_seen": 20492576,
      "step": 31120
    },
    {
      "epoch": 18.35200471698113,
      "grad_norm": 3.515958070755005,
      "learning_rate": 2.0555074054451063e-07,
      "loss": 0.3102,
      "num_input_tokens_seen": 20496032,
      "step": 31125
    },
    {
      "epoch": 18.35495283018868,
      "grad_norm": 5.191166877746582,
      "learning_rate": 2.0482129423386843e-07,
      "loss": 0.2613,
      "num_input_tokens_seen": 20499008,
      "step": 31130
    },
    {
      "epoch": 18.357900943396228,
      "grad_norm": 4.360015392303467,
      "learning_rate": 2.040931174674543e-07,
      "loss": 0.2769,
      "num_input_tokens_seen": 20503584,
      "step": 31135
    },
    {
      "epoch": 18.360849056603772,
      "grad_norm": 4.602005481719971,
      "learning_rate": 2.0336621043805682e-07,
      "loss": 0.268,
      "num_input_tokens_seen": 20507264,
      "step": 31140
    },
    {
      "epoch": 18.36379716981132,
      "grad_norm": 5.071179389953613,
      "learning_rate": 2.0264057333812704e-07,
      "loss": 0.2399,
      "num_input_tokens_seen": 20509728,
      "step": 31145
    },
    {
      "epoch": 18.36674528301887,
      "grad_norm": 4.038583278656006,
      "learning_rate": 2.0191620635978127e-07,
      "loss": 0.3001,
      "num_input_tokens_seen": 20512480,
      "step": 31150
    },
    {
      "epoch": 18.369693396226417,
      "grad_norm": 3.3885366916656494,
      "learning_rate": 2.0119310969479833e-07,
      "loss": 0.3345,
      "num_input_tokens_seen": 20515520,
      "step": 31155
    },
    {
      "epoch": 18.37264150943396,
      "grad_norm": 8.43363094329834,
      "learning_rate": 2.004712835346212e-07,
      "loss": 0.2886,
      "num_input_tokens_seen": 20518400,
      "step": 31160
    },
    {
      "epoch": 18.37558962264151,
      "grad_norm": 5.76726770401001,
      "learning_rate": 1.99750728070357e-07,
      "loss": 0.2344,
      "num_input_tokens_seen": 20522048,
      "step": 31165
    },
    {
      "epoch": 18.378537735849058,
      "grad_norm": 3.9440267086029053,
      "learning_rate": 1.9903144349277536e-07,
      "loss": 0.3083,
      "num_input_tokens_seen": 20525312,
      "step": 31170
    },
    {
      "epoch": 18.381485849056602,
      "grad_norm": 8.108642578125,
      "learning_rate": 1.983134299923095e-07,
      "loss": 0.316,
      "num_input_tokens_seen": 20528480,
      "step": 31175
    },
    {
      "epoch": 18.38443396226415,
      "grad_norm": 0.7628658413887024,
      "learning_rate": 1.9759668775905737e-07,
      "loss": 0.1829,
      "num_input_tokens_seen": 20534944,
      "step": 31180
    },
    {
      "epoch": 18.3873820754717,
      "grad_norm": 2.8172054290771484,
      "learning_rate": 1.9688121698277995e-07,
      "loss": 0.2321,
      "num_input_tokens_seen": 20537568,
      "step": 31185
    },
    {
      "epoch": 18.390330188679247,
      "grad_norm": 2.530588150024414,
      "learning_rate": 1.9616701785290015e-07,
      "loss": 0.2692,
      "num_input_tokens_seen": 20541248,
      "step": 31190
    },
    {
      "epoch": 18.39327830188679,
      "grad_norm": 4.619386196136475,
      "learning_rate": 1.954540905585056e-07,
      "loss": 0.3523,
      "num_input_tokens_seen": 20544384,
      "step": 31195
    },
    {
      "epoch": 18.39622641509434,
      "grad_norm": 3.7644381523132324,
      "learning_rate": 1.9474243528834757e-07,
      "loss": 0.3126,
      "num_input_tokens_seen": 20548096,
      "step": 31200
    },
    {
      "epoch": 18.399174528301888,
      "grad_norm": 3.5802524089813232,
      "learning_rate": 1.9403205223083866e-07,
      "loss": 0.2781,
      "num_input_tokens_seen": 20551328,
      "step": 31205
    },
    {
      "epoch": 18.402122641509433,
      "grad_norm": 3.341823101043701,
      "learning_rate": 1.9332294157405619e-07,
      "loss": 0.3699,
      "num_input_tokens_seen": 20555680,
      "step": 31210
    },
    {
      "epoch": 18.40507075471698,
      "grad_norm": 3.387204885482788,
      "learning_rate": 1.926151035057411e-07,
      "loss": 0.2894,
      "num_input_tokens_seen": 20559680,
      "step": 31215
    },
    {
      "epoch": 18.40801886792453,
      "grad_norm": 3.435789108276367,
      "learning_rate": 1.9190853821329626e-07,
      "loss": 0.3567,
      "num_input_tokens_seen": 20563136,
      "step": 31220
    },
    {
      "epoch": 18.410966981132077,
      "grad_norm": 5.536624908447266,
      "learning_rate": 1.9120324588378757e-07,
      "loss": 0.2883,
      "num_input_tokens_seen": 20566272,
      "step": 31225
    },
    {
      "epoch": 18.41391509433962,
      "grad_norm": 4.876430034637451,
      "learning_rate": 1.9049922670394461e-07,
      "loss": 0.301,
      "num_input_tokens_seen": 20569312,
      "step": 31230
    },
    {
      "epoch": 18.41686320754717,
      "grad_norm": 3.3652756214141846,
      "learning_rate": 1.897964808601588e-07,
      "loss": 0.2339,
      "num_input_tokens_seen": 20572224,
      "step": 31235
    },
    {
      "epoch": 18.419811320754718,
      "grad_norm": 2.580806016921997,
      "learning_rate": 1.8909500853848517e-07,
      "loss": 0.2335,
      "num_input_tokens_seen": 20574816,
      "step": 31240
    },
    {
      "epoch": 18.422759433962263,
      "grad_norm": 5.255645275115967,
      "learning_rate": 1.8839480992464243e-07,
      "loss": 0.3102,
      "num_input_tokens_seen": 20577920,
      "step": 31245
    },
    {
      "epoch": 18.42570754716981,
      "grad_norm": 2.981544017791748,
      "learning_rate": 1.8769588520401005e-07,
      "loss": 0.3214,
      "num_input_tokens_seen": 20580672,
      "step": 31250
    },
    {
      "epoch": 18.42865566037736,
      "grad_norm": 4.2234721183776855,
      "learning_rate": 1.8699823456163279e-07,
      "loss": 0.2898,
      "num_input_tokens_seen": 20583392,
      "step": 31255
    },
    {
      "epoch": 18.431603773584907,
      "grad_norm": 5.549757480621338,
      "learning_rate": 1.8630185818221514e-07,
      "loss": 0.3057,
      "num_input_tokens_seen": 20586688,
      "step": 31260
    },
    {
      "epoch": 18.434551886792452,
      "grad_norm": 2.9385437965393066,
      "learning_rate": 1.856067562501268e-07,
      "loss": 0.3355,
      "num_input_tokens_seen": 20590272,
      "step": 31265
    },
    {
      "epoch": 18.4375,
      "grad_norm": 2.754838228225708,
      "learning_rate": 1.8491292894939837e-07,
      "loss": 0.2169,
      "num_input_tokens_seen": 20594976,
      "step": 31270
    },
    {
      "epoch": 18.440448113207548,
      "grad_norm": 4.947962760925293,
      "learning_rate": 1.8422037646372405e-07,
      "loss": 0.255,
      "num_input_tokens_seen": 20599040,
      "step": 31275
    },
    {
      "epoch": 18.443396226415093,
      "grad_norm": 5.854122161865234,
      "learning_rate": 1.8352909897645989e-07,
      "loss": 0.4638,
      "num_input_tokens_seen": 20601600,
      "step": 31280
    },
    {
      "epoch": 18.44634433962264,
      "grad_norm": 3.028515100479126,
      "learning_rate": 1.8283909667062448e-07,
      "loss": 0.2314,
      "num_input_tokens_seen": 20605376,
      "step": 31285
    },
    {
      "epoch": 18.44929245283019,
      "grad_norm": 4.523483753204346,
      "learning_rate": 1.82150369728899e-07,
      "loss": 0.3108,
      "num_input_tokens_seen": 20609440,
      "step": 31290
    },
    {
      "epoch": 18.452240566037737,
      "grad_norm": 3.2685928344726562,
      "learning_rate": 1.814629183336275e-07,
      "loss": 0.5015,
      "num_input_tokens_seen": 20612672,
      "step": 31295
    },
    {
      "epoch": 18.455188679245282,
      "grad_norm": 3.4127085208892822,
      "learning_rate": 1.807767426668139e-07,
      "loss": 0.3402,
      "num_input_tokens_seen": 20615744,
      "step": 31300
    },
    {
      "epoch": 18.45813679245283,
      "grad_norm": 3.769554615020752,
      "learning_rate": 1.8009184291012783e-07,
      "loss": 0.3054,
      "num_input_tokens_seen": 20618336,
      "step": 31305
    },
    {
      "epoch": 18.46108490566038,
      "grad_norm": 2.962156057357788,
      "learning_rate": 1.7940821924489926e-07,
      "loss": 0.2605,
      "num_input_tokens_seen": 20621472,
      "step": 31310
    },
    {
      "epoch": 18.464033018867923,
      "grad_norm": 3.5438873767852783,
      "learning_rate": 1.7872587185212009e-07,
      "loss": 0.2885,
      "num_input_tokens_seen": 20624352,
      "step": 31315
    },
    {
      "epoch": 18.46698113207547,
      "grad_norm": 4.742137432098389,
      "learning_rate": 1.7804480091244524e-07,
      "loss": 0.1905,
      "num_input_tokens_seen": 20627136,
      "step": 31320
    },
    {
      "epoch": 18.46992924528302,
      "grad_norm": 4.170551776885986,
      "learning_rate": 1.7736500660619104e-07,
      "loss": 0.144,
      "num_input_tokens_seen": 20630976,
      "step": 31325
    },
    {
      "epoch": 18.472877358490567,
      "grad_norm": 4.92624568939209,
      "learning_rate": 1.766864891133352e-07,
      "loss": 0.4279,
      "num_input_tokens_seen": 20633568,
      "step": 31330
    },
    {
      "epoch": 18.475825471698112,
      "grad_norm": 2.6373305320739746,
      "learning_rate": 1.7600924861351843e-07,
      "loss": 0.3558,
      "num_input_tokens_seen": 20636992,
      "step": 31335
    },
    {
      "epoch": 18.47877358490566,
      "grad_norm": 6.627933502197266,
      "learning_rate": 1.7533328528604398e-07,
      "loss": 0.2553,
      "num_input_tokens_seen": 20640032,
      "step": 31340
    },
    {
      "epoch": 18.48172169811321,
      "grad_norm": 3.4831786155700684,
      "learning_rate": 1.746585993098754e-07,
      "loss": 0.3101,
      "num_input_tokens_seen": 20642688,
      "step": 31345
    },
    {
      "epoch": 18.484669811320753,
      "grad_norm": 4.131565570831299,
      "learning_rate": 1.7398519086363864e-07,
      "loss": 0.2545,
      "num_input_tokens_seen": 20646528,
      "step": 31350
    },
    {
      "epoch": 18.4876179245283,
      "grad_norm": 5.421937465667725,
      "learning_rate": 1.733130601256211e-07,
      "loss": 0.3438,
      "num_input_tokens_seen": 20649376,
      "step": 31355
    },
    {
      "epoch": 18.49056603773585,
      "grad_norm": 3.1420092582702637,
      "learning_rate": 1.7264220727377323e-07,
      "loss": 0.3083,
      "num_input_tokens_seen": 20652448,
      "step": 31360
    },
    {
      "epoch": 18.493514150943398,
      "grad_norm": 3.482408285140991,
      "learning_rate": 1.7197263248570517e-07,
      "loss": 0.1776,
      "num_input_tokens_seen": 20655680,
      "step": 31365
    },
    {
      "epoch": 18.496462264150942,
      "grad_norm": 3.1804141998291016,
      "learning_rate": 1.7130433593869124e-07,
      "loss": 0.2724,
      "num_input_tokens_seen": 20658592,
      "step": 31370
    },
    {
      "epoch": 18.49941037735849,
      "grad_norm": 5.911397457122803,
      "learning_rate": 1.706373178096643e-07,
      "loss": 0.3661,
      "num_input_tokens_seen": 20661760,
      "step": 31375
    },
    {
      "epoch": 18.50235849056604,
      "grad_norm": 5.4699554443359375,
      "learning_rate": 1.6997157827522092e-07,
      "loss": 0.3047,
      "num_input_tokens_seen": 20664928,
      "step": 31380
    },
    {
      "epoch": 18.505306603773583,
      "grad_norm": 4.213530540466309,
      "learning_rate": 1.6930711751161843e-07,
      "loss": 0.3332,
      "num_input_tokens_seen": 20669152,
      "step": 31385
    },
    {
      "epoch": 18.50825471698113,
      "grad_norm": 8.019575119018555,
      "learning_rate": 1.6864393569477556e-07,
      "loss": 0.4271,
      "num_input_tokens_seen": 20672448,
      "step": 31390
    },
    {
      "epoch": 18.51120283018868,
      "grad_norm": 3.305497407913208,
      "learning_rate": 1.6798203300027295e-07,
      "loss": 0.2415,
      "num_input_tokens_seen": 20675296,
      "step": 31395
    },
    {
      "epoch": 18.514150943396228,
      "grad_norm": 3.2649123668670654,
      "learning_rate": 1.6732140960335152e-07,
      "loss": 0.1918,
      "num_input_tokens_seen": 20678368,
      "step": 31400
    },
    {
      "epoch": 18.517099056603772,
      "grad_norm": 7.204807758331299,
      "learning_rate": 1.666620656789153e-07,
      "loss": 0.219,
      "num_input_tokens_seen": 20681088,
      "step": 31405
    },
    {
      "epoch": 18.52004716981132,
      "grad_norm": 2.9128291606903076,
      "learning_rate": 1.660040014015274e-07,
      "loss": 0.3178,
      "num_input_tokens_seen": 20684800,
      "step": 31410
    },
    {
      "epoch": 18.52299528301887,
      "grad_norm": 4.086798667907715,
      "learning_rate": 1.6534721694541344e-07,
      "loss": 0.3978,
      "num_input_tokens_seen": 20688384,
      "step": 31415
    },
    {
      "epoch": 18.525943396226417,
      "grad_norm": 4.137607574462891,
      "learning_rate": 1.6469171248445993e-07,
      "loss": 0.2508,
      "num_input_tokens_seen": 20691456,
      "step": 31420
    },
    {
      "epoch": 18.52889150943396,
      "grad_norm": 9.248373985290527,
      "learning_rate": 1.6403748819221464e-07,
      "loss": 0.2799,
      "num_input_tokens_seen": 20694912,
      "step": 31425
    },
    {
      "epoch": 18.53183962264151,
      "grad_norm": 3.1591176986694336,
      "learning_rate": 1.6338454424188632e-07,
      "loss": 0.3064,
      "num_input_tokens_seen": 20698848,
      "step": 31430
    },
    {
      "epoch": 18.534787735849058,
      "grad_norm": 2.9485862255096436,
      "learning_rate": 1.6273288080634442e-07,
      "loss": 0.3453,
      "num_input_tokens_seen": 20702816,
      "step": 31435
    },
    {
      "epoch": 18.537735849056602,
      "grad_norm": 2.7948946952819824,
      "learning_rate": 1.6208249805811982e-07,
      "loss": 0.3503,
      "num_input_tokens_seen": 20706720,
      "step": 31440
    },
    {
      "epoch": 18.54068396226415,
      "grad_norm": 5.038859844207764,
      "learning_rate": 1.6143339616940423e-07,
      "loss": 0.3267,
      "num_input_tokens_seen": 20709376,
      "step": 31445
    },
    {
      "epoch": 18.5436320754717,
      "grad_norm": 5.11807918548584,
      "learning_rate": 1.6078557531205018e-07,
      "loss": 0.3574,
      "num_input_tokens_seen": 20713216,
      "step": 31450
    },
    {
      "epoch": 18.546580188679247,
      "grad_norm": 3.245126962661743,
      "learning_rate": 1.601390356575705e-07,
      "loss": 0.2175,
      "num_input_tokens_seen": 20715936,
      "step": 31455
    },
    {
      "epoch": 18.54952830188679,
      "grad_norm": 3.481942653656006,
      "learning_rate": 1.5949377737713988e-07,
      "loss": 0.2481,
      "num_input_tokens_seen": 20719296,
      "step": 31460
    },
    {
      "epoch": 18.55247641509434,
      "grad_norm": 5.866299152374268,
      "learning_rate": 1.5884980064159338e-07,
      "loss": 0.3427,
      "num_input_tokens_seen": 20722144,
      "step": 31465
    },
    {
      "epoch": 18.555424528301888,
      "grad_norm": 2.8508875370025635,
      "learning_rate": 1.5820710562142627e-07,
      "loss": 0.3431,
      "num_input_tokens_seen": 20725568,
      "step": 31470
    },
    {
      "epoch": 18.558372641509433,
      "grad_norm": 5.760117530822754,
      "learning_rate": 1.575656924867952e-07,
      "loss": 0.372,
      "num_input_tokens_seen": 20728480,
      "step": 31475
    },
    {
      "epoch": 18.56132075471698,
      "grad_norm": 3.7278289794921875,
      "learning_rate": 1.5692556140751658e-07,
      "loss": 0.2229,
      "num_input_tokens_seen": 20733504,
      "step": 31480
    },
    {
      "epoch": 18.56426886792453,
      "grad_norm": 3.135848045349121,
      "learning_rate": 1.5628671255306706e-07,
      "loss": 0.1943,
      "num_input_tokens_seen": 20738848,
      "step": 31485
    },
    {
      "epoch": 18.567216981132077,
      "grad_norm": 3.7599825859069824,
      "learning_rate": 1.556491460925863e-07,
      "loss": 0.2849,
      "num_input_tokens_seen": 20742080,
      "step": 31490
    },
    {
      "epoch": 18.57016509433962,
      "grad_norm": 3.206240653991699,
      "learning_rate": 1.550128621948721e-07,
      "loss": 0.3311,
      "num_input_tokens_seen": 20744832,
      "step": 31495
    },
    {
      "epoch": 18.57311320754717,
      "grad_norm": 5.221689224243164,
      "learning_rate": 1.5437786102838413e-07,
      "loss": 0.2841,
      "num_input_tokens_seen": 20748064,
      "step": 31500
    },
    {
      "epoch": 18.576061320754718,
      "grad_norm": 2.4239017963409424,
      "learning_rate": 1.5374414276124017e-07,
      "loss": 0.2131,
      "num_input_tokens_seen": 20751136,
      "step": 31505
    },
    {
      "epoch": 18.579009433962263,
      "grad_norm": 4.003374099731445,
      "learning_rate": 1.5311170756122095e-07,
      "loss": 0.2703,
      "num_input_tokens_seen": 20755072,
      "step": 31510
    },
    {
      "epoch": 18.58195754716981,
      "grad_norm": 3.2510509490966797,
      "learning_rate": 1.5248055559576647e-07,
      "loss": 0.2238,
      "num_input_tokens_seen": 20757728,
      "step": 31515
    },
    {
      "epoch": 18.58490566037736,
      "grad_norm": 5.6493306159973145,
      "learning_rate": 1.5185068703197526e-07,
      "loss": 0.2602,
      "num_input_tokens_seen": 20760960,
      "step": 31520
    },
    {
      "epoch": 18.587853773584907,
      "grad_norm": 4.758347034454346,
      "learning_rate": 1.5122210203661004e-07,
      "loss": 0.3476,
      "num_input_tokens_seen": 20763968,
      "step": 31525
    },
    {
      "epoch": 18.590801886792452,
      "grad_norm": 5.461327075958252,
      "learning_rate": 1.505948007760899e-07,
      "loss": 0.3324,
      "num_input_tokens_seen": 20767296,
      "step": 31530
    },
    {
      "epoch": 18.59375,
      "grad_norm": 4.437388896942139,
      "learning_rate": 1.4996878341649647e-07,
      "loss": 0.2508,
      "num_input_tokens_seen": 20770304,
      "step": 31535
    },
    {
      "epoch": 18.596698113207548,
      "grad_norm": 5.735592842102051,
      "learning_rate": 1.493440501235699e-07,
      "loss": 0.2714,
      "num_input_tokens_seen": 20774144,
      "step": 31540
    },
    {
      "epoch": 18.599646226415093,
      "grad_norm": 3.1009209156036377,
      "learning_rate": 1.487206010627118e-07,
      "loss": 0.3096,
      "num_input_tokens_seen": 20776704,
      "step": 31545
    },
    {
      "epoch": 18.60259433962264,
      "grad_norm": 6.197736740112305,
      "learning_rate": 1.4809843639898124e-07,
      "loss": 0.1854,
      "num_input_tokens_seen": 20778816,
      "step": 31550
    },
    {
      "epoch": 18.60554245283019,
      "grad_norm": 4.4898810386657715,
      "learning_rate": 1.4747755629710093e-07,
      "loss": 0.2772,
      "num_input_tokens_seen": 20782144,
      "step": 31555
    },
    {
      "epoch": 18.608490566037737,
      "grad_norm": 3.608382225036621,
      "learning_rate": 1.4685796092145045e-07,
      "loss": 0.2522,
      "num_input_tokens_seen": 20785824,
      "step": 31560
    },
    {
      "epoch": 18.611438679245282,
      "grad_norm": 3.6139631271362305,
      "learning_rate": 1.4623965043607135e-07,
      "loss": 0.1991,
      "num_input_tokens_seen": 20788512,
      "step": 31565
    },
    {
      "epoch": 18.61438679245283,
      "grad_norm": 2.9461560249328613,
      "learning_rate": 1.4562262500466273e-07,
      "loss": 0.2738,
      "num_input_tokens_seen": 20791072,
      "step": 31570
    },
    {
      "epoch": 18.61733490566038,
      "grad_norm": 3.6773524284362793,
      "learning_rate": 1.4500688479058556e-07,
      "loss": 0.3125,
      "num_input_tokens_seen": 20794112,
      "step": 31575
    },
    {
      "epoch": 18.620283018867923,
      "grad_norm": 3.2475693225860596,
      "learning_rate": 1.4439242995685943e-07,
      "loss": 0.3096,
      "num_input_tokens_seen": 20797600,
      "step": 31580
    },
    {
      "epoch": 18.62323113207547,
      "grad_norm": 4.367268085479736,
      "learning_rate": 1.4377926066616364e-07,
      "loss": 0.3893,
      "num_input_tokens_seen": 20800640,
      "step": 31585
    },
    {
      "epoch": 18.62617924528302,
      "grad_norm": 3.264085531234741,
      "learning_rate": 1.4316737708083783e-07,
      "loss": 0.238,
      "num_input_tokens_seen": 20804384,
      "step": 31590
    },
    {
      "epoch": 18.629127358490567,
      "grad_norm": 3.8310344219207764,
      "learning_rate": 1.4255677936288127e-07,
      "loss": 0.3017,
      "num_input_tokens_seen": 20807456,
      "step": 31595
    },
    {
      "epoch": 18.632075471698112,
      "grad_norm": 5.724236965179443,
      "learning_rate": 1.4194746767395184e-07,
      "loss": 0.3664,
      "num_input_tokens_seen": 20810176,
      "step": 31600
    },
    {
      "epoch": 18.63502358490566,
      "grad_norm": 2.493313789367676,
      "learning_rate": 1.4133944217536722e-07,
      "loss": 0.2649,
      "num_input_tokens_seen": 20812992,
      "step": 31605
    },
    {
      "epoch": 18.63797169811321,
      "grad_norm": 2.8976597785949707,
      "learning_rate": 1.4073270302810471e-07,
      "loss": 0.2098,
      "num_input_tokens_seen": 20816576,
      "step": 31610
    },
    {
      "epoch": 18.640919811320753,
      "grad_norm": 1.9183794260025024,
      "learning_rate": 1.4012725039280084e-07,
      "loss": 0.2744,
      "num_input_tokens_seen": 20819808,
      "step": 31615
    },
    {
      "epoch": 18.6438679245283,
      "grad_norm": 4.081894397735596,
      "learning_rate": 1.3952308442975292e-07,
      "loss": 0.2978,
      "num_input_tokens_seen": 20823552,
      "step": 31620
    },
    {
      "epoch": 18.64681603773585,
      "grad_norm": 5.020522117614746,
      "learning_rate": 1.3892020529891637e-07,
      "loss": 0.2867,
      "num_input_tokens_seen": 20826880,
      "step": 31625
    },
    {
      "epoch": 18.649764150943398,
      "grad_norm": 9.615575790405273,
      "learning_rate": 1.3831861315990514e-07,
      "loss": 0.3328,
      "num_input_tokens_seen": 20829184,
      "step": 31630
    },
    {
      "epoch": 18.652712264150942,
      "grad_norm": 3.9860572814941406,
      "learning_rate": 1.377183081719935e-07,
      "loss": 0.2415,
      "num_input_tokens_seen": 20831808,
      "step": 31635
    },
    {
      "epoch": 18.65566037735849,
      "grad_norm": 3.1889400482177734,
      "learning_rate": 1.3711929049411544e-07,
      "loss": 0.2325,
      "num_input_tokens_seen": 20835328,
      "step": 31640
    },
    {
      "epoch": 18.65860849056604,
      "grad_norm": 3.406130075454712,
      "learning_rate": 1.365215602848624e-07,
      "loss": 0.2208,
      "num_input_tokens_seen": 20838464,
      "step": 31645
    },
    {
      "epoch": 18.661556603773583,
      "grad_norm": 3.1452643871307373,
      "learning_rate": 1.3592511770248727e-07,
      "loss": 0.2708,
      "num_input_tokens_seen": 20841472,
      "step": 31650
    },
    {
      "epoch": 18.66450471698113,
      "grad_norm": 6.327495574951172,
      "learning_rate": 1.3532996290490041e-07,
      "loss": 0.3102,
      "num_input_tokens_seen": 20844128,
      "step": 31655
    },
    {
      "epoch": 18.66745283018868,
      "grad_norm": 8.49341106414795,
      "learning_rate": 1.347360960496713e-07,
      "loss": 0.313,
      "num_input_tokens_seen": 20847424,
      "step": 31660
    },
    {
      "epoch": 18.670400943396228,
      "grad_norm": 3.115478277206421,
      "learning_rate": 1.3414351729402862e-07,
      "loss": 0.2812,
      "num_input_tokens_seen": 20851616,
      "step": 31665
    },
    {
      "epoch": 18.673349056603772,
      "grad_norm": 4.571174144744873,
      "learning_rate": 1.3355222679486025e-07,
      "loss": 0.3361,
      "num_input_tokens_seen": 20854880,
      "step": 31670
    },
    {
      "epoch": 18.67629716981132,
      "grad_norm": 3.902445077896118,
      "learning_rate": 1.3296222470871367e-07,
      "loss": 0.3138,
      "num_input_tokens_seen": 20857888,
      "step": 31675
    },
    {
      "epoch": 18.67924528301887,
      "grad_norm": 4.276546478271484,
      "learning_rate": 1.3237351119179287e-07,
      "loss": 0.2537,
      "num_input_tokens_seen": 20861536,
      "step": 31680
    },
    {
      "epoch": 18.682193396226417,
      "grad_norm": 4.376583576202393,
      "learning_rate": 1.3178608639996425e-07,
      "loss": 0.2684,
      "num_input_tokens_seen": 20865024,
      "step": 31685
    },
    {
      "epoch": 18.68514150943396,
      "grad_norm": 3.8199117183685303,
      "learning_rate": 1.3119995048874957e-07,
      "loss": 0.2398,
      "num_input_tokens_seen": 20868480,
      "step": 31690
    },
    {
      "epoch": 18.68808962264151,
      "grad_norm": 4.324284076690674,
      "learning_rate": 1.3061510361333186e-07,
      "loss": 0.2951,
      "num_input_tokens_seen": 20871200,
      "step": 31695
    },
    {
      "epoch": 18.691037735849058,
      "grad_norm": 4.0500288009643555,
      "learning_rate": 1.3003154592855116e-07,
      "loss": 0.3776,
      "num_input_tokens_seen": 20875584,
      "step": 31700
    },
    {
      "epoch": 18.693985849056602,
      "grad_norm": 4.489680767059326,
      "learning_rate": 1.2944927758890668e-07,
      "loss": 0.3801,
      "num_input_tokens_seen": 20878368,
      "step": 31705
    },
    {
      "epoch": 18.69693396226415,
      "grad_norm": 4.11642599105835,
      "learning_rate": 1.2886829874855733e-07,
      "loss": 0.3346,
      "num_input_tokens_seen": 20881056,
      "step": 31710
    },
    {
      "epoch": 18.6998820754717,
      "grad_norm": 5.96708869934082,
      "learning_rate": 1.2828860956131894e-07,
      "loss": 0.3651,
      "num_input_tokens_seen": 20883552,
      "step": 31715
    },
    {
      "epoch": 18.702830188679247,
      "grad_norm": 4.011523723602295,
      "learning_rate": 1.2771021018066765e-07,
      "loss": 0.4373,
      "num_input_tokens_seen": 20887328,
      "step": 31720
    },
    {
      "epoch": 18.70577830188679,
      "grad_norm": 5.115931034088135,
      "learning_rate": 1.271331007597365e-07,
      "loss": 0.3102,
      "num_input_tokens_seen": 20889984,
      "step": 31725
    },
    {
      "epoch": 18.70872641509434,
      "grad_norm": 3.0553956031799316,
      "learning_rate": 1.2655728145131774e-07,
      "loss": 0.3166,
      "num_input_tokens_seen": 20892864,
      "step": 31730
    },
    {
      "epoch": 18.711674528301888,
      "grad_norm": 3.3029751777648926,
      "learning_rate": 1.2598275240786105e-07,
      "loss": 0.3619,
      "num_input_tokens_seen": 20896704,
      "step": 31735
    },
    {
      "epoch": 18.714622641509433,
      "grad_norm": 3.6855485439300537,
      "learning_rate": 1.254095137814776e-07,
      "loss": 0.2886,
      "num_input_tokens_seen": 20900320,
      "step": 31740
    },
    {
      "epoch": 18.71757075471698,
      "grad_norm": 3.4328532218933105,
      "learning_rate": 1.2483756572393368e-07,
      "loss": 0.318,
      "num_input_tokens_seen": 20905184,
      "step": 31745
    },
    {
      "epoch": 18.72051886792453,
      "grad_norm": 5.887598514556885,
      "learning_rate": 1.242669083866549e-07,
      "loss": 0.3058,
      "num_input_tokens_seen": 20908320,
      "step": 31750
    },
    {
      "epoch": 18.723466981132077,
      "grad_norm": 6.546753883361816,
      "learning_rate": 1.2369754192072537e-07,
      "loss": 0.3514,
      "num_input_tokens_seen": 20910624,
      "step": 31755
    },
    {
      "epoch": 18.72641509433962,
      "grad_norm": 3.494891881942749,
      "learning_rate": 1.231294664768873e-07,
      "loss": 0.2971,
      "num_input_tokens_seen": 20914016,
      "step": 31760
    },
    {
      "epoch": 18.72936320754717,
      "grad_norm": 4.4511847496032715,
      "learning_rate": 1.225626822055409e-07,
      "loss": 0.4624,
      "num_input_tokens_seen": 20917088,
      "step": 31765
    },
    {
      "epoch": 18.732311320754718,
      "grad_norm": 4.15349006652832,
      "learning_rate": 1.2199718925674508e-07,
      "loss": 0.2951,
      "num_input_tokens_seen": 20920192,
      "step": 31770
    },
    {
      "epoch": 18.735259433962263,
      "grad_norm": 6.977143287658691,
      "learning_rate": 1.2143298778021616e-07,
      "loss": 0.3457,
      "num_input_tokens_seen": 20922912,
      "step": 31775
    },
    {
      "epoch": 18.73820754716981,
      "grad_norm": 2.9361095428466797,
      "learning_rate": 1.2087007792532967e-07,
      "loss": 0.2978,
      "num_input_tokens_seen": 20926912,
      "step": 31780
    },
    {
      "epoch": 18.74115566037736,
      "grad_norm": 4.062680721282959,
      "learning_rate": 1.203084598411175e-07,
      "loss": 0.4046,
      "num_input_tokens_seen": 20930112,
      "step": 31785
    },
    {
      "epoch": 18.744103773584907,
      "grad_norm": 4.63753604888916,
      "learning_rate": 1.1974813367627124e-07,
      "loss": 0.2565,
      "num_input_tokens_seen": 20933984,
      "step": 31790
    },
    {
      "epoch": 18.747051886792452,
      "grad_norm": 4.693019390106201,
      "learning_rate": 1.1918909957913949e-07,
      "loss": 0.3247,
      "num_input_tokens_seen": 20936512,
      "step": 31795
    },
    {
      "epoch": 18.75,
      "grad_norm": 5.011508941650391,
      "learning_rate": 1.1863135769772827e-07,
      "loss": 0.3118,
      "num_input_tokens_seen": 20939072,
      "step": 31800
    },
    {
      "epoch": 18.752948113207548,
      "grad_norm": 7.069087982177734,
      "learning_rate": 1.1807490817970279e-07,
      "loss": 0.2144,
      "num_input_tokens_seen": 20942784,
      "step": 31805
    },
    {
      "epoch": 18.755896226415093,
      "grad_norm": 2.8182363510131836,
      "learning_rate": 1.1751975117238578e-07,
      "loss": 0.4106,
      "num_input_tokens_seen": 20945696,
      "step": 31810
    },
    {
      "epoch": 18.75884433962264,
      "grad_norm": 7.110505104064941,
      "learning_rate": 1.1696588682275633e-07,
      "loss": 0.3534,
      "num_input_tokens_seen": 20949152,
      "step": 31815
    },
    {
      "epoch": 18.76179245283019,
      "grad_norm": 5.114602565765381,
      "learning_rate": 1.1641331527745325e-07,
      "loss": 0.418,
      "num_input_tokens_seen": 20952064,
      "step": 31820
    },
    {
      "epoch": 18.764740566037737,
      "grad_norm": 6.656158924102783,
      "learning_rate": 1.1586203668277229e-07,
      "loss": 0.2496,
      "num_input_tokens_seen": 20954752,
      "step": 31825
    },
    {
      "epoch": 18.767688679245282,
      "grad_norm": 2.197567939758301,
      "learning_rate": 1.1531205118466615e-07,
      "loss": 0.1909,
      "num_input_tokens_seen": 20957568,
      "step": 31830
    },
    {
      "epoch": 18.77063679245283,
      "grad_norm": 6.692008972167969,
      "learning_rate": 1.1476335892874669e-07,
      "loss": 0.3017,
      "num_input_tokens_seen": 20960640,
      "step": 31835
    },
    {
      "epoch": 18.77358490566038,
      "grad_norm": 3.9671754837036133,
      "learning_rate": 1.1421596006028157e-07,
      "loss": 0.3018,
      "num_input_tokens_seen": 20963296,
      "step": 31840
    },
    {
      "epoch": 18.776533018867923,
      "grad_norm": 7.727102279663086,
      "learning_rate": 1.1366985472419823e-07,
      "loss": 0.3012,
      "num_input_tokens_seen": 20966304,
      "step": 31845
    },
    {
      "epoch": 18.77948113207547,
      "grad_norm": 1.6524502038955688,
      "learning_rate": 1.1312504306507987e-07,
      "loss": 0.2614,
      "num_input_tokens_seen": 20969920,
      "step": 31850
    },
    {
      "epoch": 18.78242924528302,
      "grad_norm": 9.169069290161133,
      "learning_rate": 1.1258152522716725e-07,
      "loss": 0.279,
      "num_input_tokens_seen": 20972576,
      "step": 31855
    },
    {
      "epoch": 18.785377358490567,
      "grad_norm": 2.7980551719665527,
      "learning_rate": 1.1203930135435914e-07,
      "loss": 0.3219,
      "num_input_tokens_seen": 20976384,
      "step": 31860
    },
    {
      "epoch": 18.788325471698112,
      "grad_norm": 3.321323871612549,
      "learning_rate": 1.1149837159021238e-07,
      "loss": 0.3315,
      "num_input_tokens_seen": 20980640,
      "step": 31865
    },
    {
      "epoch": 18.79127358490566,
      "grad_norm": 5.540903568267822,
      "learning_rate": 1.1095873607793961e-07,
      "loss": 0.2896,
      "num_input_tokens_seen": 20983424,
      "step": 31870
    },
    {
      "epoch": 18.79422169811321,
      "grad_norm": 4.377302169799805,
      "learning_rate": 1.1042039496041212e-07,
      "loss": 0.3201,
      "num_input_tokens_seen": 20986336,
      "step": 31875
    },
    {
      "epoch": 18.797169811320753,
      "grad_norm": 1.7409120798110962,
      "learning_rate": 1.0988334838015812e-07,
      "loss": 0.2905,
      "num_input_tokens_seen": 20989696,
      "step": 31880
    },
    {
      "epoch": 18.8001179245283,
      "grad_norm": 5.265538692474365,
      "learning_rate": 1.0934759647936333e-07,
      "loss": 0.3561,
      "num_input_tokens_seen": 20992736,
      "step": 31885
    },
    {
      "epoch": 18.80306603773585,
      "grad_norm": 3.6148386001586914,
      "learning_rate": 1.0881313939986926e-07,
      "loss": 0.3549,
      "num_input_tokens_seen": 20996544,
      "step": 31890
    },
    {
      "epoch": 18.806014150943398,
      "grad_norm": 4.893805027008057,
      "learning_rate": 1.0827997728317662e-07,
      "loss": 0.3562,
      "num_input_tokens_seen": 21001152,
      "step": 31895
    },
    {
      "epoch": 18.808962264150942,
      "grad_norm": 3.538806676864624,
      "learning_rate": 1.0774811027044196e-07,
      "loss": 0.3298,
      "num_input_tokens_seen": 21003840,
      "step": 31900
    },
    {
      "epoch": 18.81191037735849,
      "grad_norm": 6.978827953338623,
      "learning_rate": 1.0721753850247984e-07,
      "loss": 0.3496,
      "num_input_tokens_seen": 21007168,
      "step": 31905
    },
    {
      "epoch": 18.81485849056604,
      "grad_norm": 3.5692412853240967,
      "learning_rate": 1.0668826211976124e-07,
      "loss": 0.2527,
      "num_input_tokens_seen": 21010112,
      "step": 31910
    },
    {
      "epoch": 18.817806603773583,
      "grad_norm": 5.8188371658325195,
      "learning_rate": 1.0616028126241407e-07,
      "loss": 0.3332,
      "num_input_tokens_seen": 21012896,
      "step": 31915
    },
    {
      "epoch": 18.82075471698113,
      "grad_norm": 3.339078903198242,
      "learning_rate": 1.0563359607022372e-07,
      "loss": 0.3161,
      "num_input_tokens_seen": 21015840,
      "step": 31920
    },
    {
      "epoch": 18.82370283018868,
      "grad_norm": 5.761385440826416,
      "learning_rate": 1.05108206682632e-07,
      "loss": 0.3291,
      "num_input_tokens_seen": 21018496,
      "step": 31925
    },
    {
      "epoch": 18.826650943396228,
      "grad_norm": 3.2524750232696533,
      "learning_rate": 1.0458411323873874e-07,
      "loss": 0.416,
      "num_input_tokens_seen": 21022144,
      "step": 31930
    },
    {
      "epoch": 18.829599056603772,
      "grad_norm": 4.434716701507568,
      "learning_rate": 1.0406131587729962e-07,
      "loss": 0.3393,
      "num_input_tokens_seen": 21025408,
      "step": 31935
    },
    {
      "epoch": 18.83254716981132,
      "grad_norm": 5.129987716674805,
      "learning_rate": 1.035398147367278e-07,
      "loss": 0.2728,
      "num_input_tokens_seen": 21028864,
      "step": 31940
    },
    {
      "epoch": 18.83549528301887,
      "grad_norm": 5.315784931182861,
      "learning_rate": 1.030196099550923e-07,
      "loss": 0.3146,
      "num_input_tokens_seen": 21032032,
      "step": 31945
    },
    {
      "epoch": 18.838443396226417,
      "grad_norm": 3.251617670059204,
      "learning_rate": 1.0250070167011905e-07,
      "loss": 0.3748,
      "num_input_tokens_seen": 21035808,
      "step": 31950
    },
    {
      "epoch": 18.84139150943396,
      "grad_norm": 3.6528632640838623,
      "learning_rate": 1.0198309001919315e-07,
      "loss": 0.3052,
      "num_input_tokens_seen": 21039616,
      "step": 31955
    },
    {
      "epoch": 18.84433962264151,
      "grad_norm": 4.727553367614746,
      "learning_rate": 1.0146677513935277e-07,
      "loss": 0.4245,
      "num_input_tokens_seen": 21042368,
      "step": 31960
    },
    {
      "epoch": 18.847287735849058,
      "grad_norm": 5.430485248565674,
      "learning_rate": 1.0095175716729578e-07,
      "loss": 0.2756,
      "num_input_tokens_seen": 21045984,
      "step": 31965
    },
    {
      "epoch": 18.850235849056602,
      "grad_norm": 3.3533363342285156,
      "learning_rate": 1.004380362393742e-07,
      "loss": 0.3144,
      "num_input_tokens_seen": 21049056,
      "step": 31970
    },
    {
      "epoch": 18.85318396226415,
      "grad_norm": 3.6844961643218994,
      "learning_rate": 9.99256124915987e-08,
      "loss": 0.2716,
      "num_input_tokens_seen": 21051712,
      "step": 31975
    },
    {
      "epoch": 18.8561320754717,
      "grad_norm": 5.12249231338501,
      "learning_rate": 9.941448605963577e-08,
      "loss": 0.2737,
      "num_input_tokens_seen": 21054336,
      "step": 31980
    },
    {
      "epoch": 18.859080188679247,
      "grad_norm": 3.514850616455078,
      "learning_rate": 9.890465707880715e-08,
      "loss": 0.2659,
      "num_input_tokens_seen": 21057536,
      "step": 31985
    },
    {
      "epoch": 18.86202830188679,
      "grad_norm": 4.666265964508057,
      "learning_rate": 9.839612568409374e-08,
      "loss": 0.268,
      "num_input_tokens_seen": 21060448,
      "step": 31990
    },
    {
      "epoch": 18.86497641509434,
      "grad_norm": 4.496227741241455,
      "learning_rate": 9.788889201013119e-08,
      "loss": 0.2773,
      "num_input_tokens_seen": 21062976,
      "step": 31995
    },
    {
      "epoch": 18.867924528301888,
      "grad_norm": 2.919137477874756,
      "learning_rate": 9.738295619121097e-08,
      "loss": 0.1921,
      "num_input_tokens_seen": 21066144,
      "step": 32000
    },
    {
      "epoch": 18.870872641509433,
      "grad_norm": 4.782341003417969,
      "learning_rate": 9.687831836128203e-08,
      "loss": 0.2307,
      "num_input_tokens_seen": 21069440,
      "step": 32005
    },
    {
      "epoch": 18.87382075471698,
      "grad_norm": 4.0959625244140625,
      "learning_rate": 9.637497865395029e-08,
      "loss": 0.3318,
      "num_input_tokens_seen": 21073536,
      "step": 32010
    },
    {
      "epoch": 18.87676886792453,
      "grad_norm": 3.579558849334717,
      "learning_rate": 9.587293720247526e-08,
      "loss": 0.2782,
      "num_input_tokens_seen": 21076672,
      "step": 32015
    },
    {
      "epoch": 18.879716981132077,
      "grad_norm": 3.4353606700897217,
      "learning_rate": 9.537219413977672e-08,
      "loss": 0.3116,
      "num_input_tokens_seen": 21080672,
      "step": 32020
    },
    {
      "epoch": 18.88266509433962,
      "grad_norm": 3.08475923538208,
      "learning_rate": 9.487274959842696e-08,
      "loss": 0.2796,
      "num_input_tokens_seen": 21084384,
      "step": 32025
    },
    {
      "epoch": 18.88561320754717,
      "grad_norm": 4.317970275878906,
      "learning_rate": 9.437460371065687e-08,
      "loss": 0.3324,
      "num_input_tokens_seen": 21087200,
      "step": 32030
    },
    {
      "epoch": 18.888561320754718,
      "grad_norm": 3.62984037399292,
      "learning_rate": 9.387775660835263e-08,
      "loss": 0.2136,
      "num_input_tokens_seen": 21090560,
      "step": 32035
    },
    {
      "epoch": 18.891509433962263,
      "grad_norm": 5.6916823387146,
      "learning_rate": 9.338220842305678e-08,
      "loss": 0.2969,
      "num_input_tokens_seen": 21093728,
      "step": 32040
    },
    {
      "epoch": 18.89445754716981,
      "grad_norm": 3.577956199645996,
      "learning_rate": 9.288795928596661e-08,
      "loss": 0.3623,
      "num_input_tokens_seen": 21097472,
      "step": 32045
    },
    {
      "epoch": 18.89740566037736,
      "grad_norm": 3.4244444370269775,
      "learning_rate": 9.239500932793854e-08,
      "loss": 0.3069,
      "num_input_tokens_seen": 21101920,
      "step": 32050
    },
    {
      "epoch": 18.900353773584907,
      "grad_norm": 4.235383033752441,
      "learning_rate": 9.190335867948263e-08,
      "loss": 0.3185,
      "num_input_tokens_seen": 21104672,
      "step": 32055
    },
    {
      "epoch": 18.903301886792452,
      "grad_norm": 4.040127277374268,
      "learning_rate": 9.141300747076476e-08,
      "loss": 0.271,
      "num_input_tokens_seen": 21106944,
      "step": 32060
    },
    {
      "epoch": 18.90625,
      "grad_norm": 9.317220687866211,
      "learning_rate": 9.092395583160773e-08,
      "loss": 0.2833,
      "num_input_tokens_seen": 21109600,
      "step": 32065
    },
    {
      "epoch": 18.909198113207548,
      "grad_norm": 3.0058538913726807,
      "learning_rate": 9.043620389149021e-08,
      "loss": 0.3279,
      "num_input_tokens_seen": 21112960,
      "step": 32070
    },
    {
      "epoch": 18.912146226415093,
      "grad_norm": 3.0771429538726807,
      "learning_rate": 8.994975177954723e-08,
      "loss": 0.3178,
      "num_input_tokens_seen": 21115776,
      "step": 32075
    },
    {
      "epoch": 18.91509433962264,
      "grad_norm": 4.508816242218018,
      "learning_rate": 8.946459962456855e-08,
      "loss": 0.3191,
      "num_input_tokens_seen": 21119296,
      "step": 32080
    },
    {
      "epoch": 18.91804245283019,
      "grad_norm": 6.937246322631836,
      "learning_rate": 8.89807475550003e-08,
      "loss": 0.2418,
      "num_input_tokens_seen": 21122208,
      "step": 32085
    },
    {
      "epoch": 18.920990566037737,
      "grad_norm": 4.784231185913086,
      "learning_rate": 8.849819569894447e-08,
      "loss": 0.3162,
      "num_input_tokens_seen": 21125248,
      "step": 32090
    },
    {
      "epoch": 18.923938679245282,
      "grad_norm": 5.082605361938477,
      "learning_rate": 8.801694418415884e-08,
      "loss": 0.2741,
      "num_input_tokens_seen": 21128192,
      "step": 32095
    },
    {
      "epoch": 18.92688679245283,
      "grad_norm": 10.354985237121582,
      "learning_rate": 8.753699313805708e-08,
      "loss": 0.4065,
      "num_input_tokens_seen": 21131200,
      "step": 32100
    },
    {
      "epoch": 18.92983490566038,
      "grad_norm": 6.095724105834961,
      "learning_rate": 8.705834268770753e-08,
      "loss": 0.3387,
      "num_input_tokens_seen": 21134144,
      "step": 32105
    },
    {
      "epoch": 18.932783018867923,
      "grad_norm": 3.7051875591278076,
      "learning_rate": 8.65809929598349e-08,
      "loss": 0.389,
      "num_input_tokens_seen": 21137344,
      "step": 32110
    },
    {
      "epoch": 18.93573113207547,
      "grad_norm": 6.291484832763672,
      "learning_rate": 8.610494408082037e-08,
      "loss": 0.2981,
      "num_input_tokens_seen": 21140512,
      "step": 32115
    },
    {
      "epoch": 18.93867924528302,
      "grad_norm": 3.587689161300659,
      "learning_rate": 8.563019617669977e-08,
      "loss": 0.302,
      "num_input_tokens_seen": 21144032,
      "step": 32120
    },
    {
      "epoch": 18.941627358490567,
      "grad_norm": 3.005415916442871,
      "learning_rate": 8.51567493731642e-08,
      "loss": 0.4552,
      "num_input_tokens_seen": 21146912,
      "step": 32125
    },
    {
      "epoch": 18.944575471698112,
      "grad_norm": 3.0800535678863525,
      "learning_rate": 8.468460379556176e-08,
      "loss": 0.2838,
      "num_input_tokens_seen": 21150848,
      "step": 32130
    },
    {
      "epoch": 18.94752358490566,
      "grad_norm": 4.905485153198242,
      "learning_rate": 8.421375956889355e-08,
      "loss": 0.3783,
      "num_input_tokens_seen": 21154912,
      "step": 32135
    },
    {
      "epoch": 18.95047169811321,
      "grad_norm": 2.1760239601135254,
      "learning_rate": 8.374421681781819e-08,
      "loss": 0.2869,
      "num_input_tokens_seen": 21159136,
      "step": 32140
    },
    {
      "epoch": 18.953419811320753,
      "grad_norm": 3.170729160308838,
      "learning_rate": 8.327597566665013e-08,
      "loss": 0.4012,
      "num_input_tokens_seen": 21162688,
      "step": 32145
    },
    {
      "epoch": 18.9563679245283,
      "grad_norm": 6.4596428871154785,
      "learning_rate": 8.280903623935688e-08,
      "loss": 0.283,
      "num_input_tokens_seen": 21166400,
      "step": 32150
    },
    {
      "epoch": 18.95931603773585,
      "grad_norm": 2.7897789478302,
      "learning_rate": 8.234339865956342e-08,
      "loss": 0.4112,
      "num_input_tokens_seen": 21170880,
      "step": 32155
    },
    {
      "epoch": 18.962264150943398,
      "grad_norm": 3.8774032592773438,
      "learning_rate": 8.187906305054838e-08,
      "loss": 0.3218,
      "num_input_tokens_seen": 21173856,
      "step": 32160
    },
    {
      "epoch": 18.965212264150942,
      "grad_norm": 3.235830545425415,
      "learning_rate": 8.141602953524841e-08,
      "loss": 0.2773,
      "num_input_tokens_seen": 21176896,
      "step": 32165
    },
    {
      "epoch": 18.96816037735849,
      "grad_norm": 4.469286918640137,
      "learning_rate": 8.095429823625212e-08,
      "loss": 0.2673,
      "num_input_tokens_seen": 21180576,
      "step": 32170
    },
    {
      "epoch": 18.97110849056604,
      "grad_norm": 3.184729814529419,
      "learning_rate": 8.04938692758045e-08,
      "loss": 0.2532,
      "num_input_tokens_seen": 21184160,
      "step": 32175
    },
    {
      "epoch": 18.974056603773583,
      "grad_norm": 3.7705516815185547,
      "learning_rate": 8.003474277580803e-08,
      "loss": 0.3168,
      "num_input_tokens_seen": 21187584,
      "step": 32180
    },
    {
      "epoch": 18.97700471698113,
      "grad_norm": 2.5850958824157715,
      "learning_rate": 7.95769188578166e-08,
      "loss": 0.2607,
      "num_input_tokens_seen": 21190592,
      "step": 32185
    },
    {
      "epoch": 18.97995283018868,
      "grad_norm": 4.869413375854492,
      "learning_rate": 7.912039764304213e-08,
      "loss": 0.2441,
      "num_input_tokens_seen": 21193472,
      "step": 32190
    },
    {
      "epoch": 18.982900943396228,
      "grad_norm": 5.180572032928467,
      "learning_rate": 7.866517925235017e-08,
      "loss": 0.378,
      "num_input_tokens_seen": 21196640,
      "step": 32195
    },
    {
      "epoch": 18.985849056603772,
      "grad_norm": 4.420051097869873,
      "learning_rate": 7.821126380626154e-08,
      "loss": 0.388,
      "num_input_tokens_seen": 21199168,
      "step": 32200
    },
    {
      "epoch": 18.98879716981132,
      "grad_norm": 5.532229900360107,
      "learning_rate": 7.775865142495286e-08,
      "loss": 0.2318,
      "num_input_tokens_seen": 21202592,
      "step": 32205
    },
    {
      "epoch": 18.99174528301887,
      "grad_norm": 5.263530254364014,
      "learning_rate": 7.730734222825442e-08,
      "loss": 0.3996,
      "num_input_tokens_seen": 21206176,
      "step": 32210
    },
    {
      "epoch": 18.994693396226417,
      "grad_norm": 3.4844753742218018,
      "learning_rate": 7.68573363356534e-08,
      "loss": 0.377,
      "num_input_tokens_seen": 21209952,
      "step": 32215
    },
    {
      "epoch": 18.99764150943396,
      "grad_norm": 3.423243761062622,
      "learning_rate": 7.640863386629005e-08,
      "loss": 0.3378,
      "num_input_tokens_seen": 21212960,
      "step": 32220
    },
    {
      "epoch": 19.00058962264151,
      "grad_norm": 4.450048446655273,
      "learning_rate": 7.59612349389599e-08,
      "loss": 0.2898,
      "num_input_tokens_seen": 21215152,
      "step": 32225
    },
    {
      "epoch": 19.003537735849058,
      "grad_norm": 2.8897457122802734,
      "learning_rate": 7.551513967211433e-08,
      "loss": 0.3032,
      "num_input_tokens_seen": 21218544,
      "step": 32230
    },
    {
      "epoch": 19.006485849056602,
      "grad_norm": 4.931918144226074,
      "learning_rate": 7.507034818385883e-08,
      "loss": 0.2807,
      "num_input_tokens_seen": 21222000,
      "step": 32235
    },
    {
      "epoch": 19.00943396226415,
      "grad_norm": 3.286038398742676,
      "learning_rate": 7.462686059195423e-08,
      "loss": 0.3521,
      "num_input_tokens_seen": 21225840,
      "step": 32240
    },
    {
      "epoch": 19.0123820754717,
      "grad_norm": 4.063599586486816,
      "learning_rate": 7.418467701381548e-08,
      "loss": 0.2847,
      "num_input_tokens_seen": 21229360,
      "step": 32245
    },
    {
      "epoch": 19.015330188679247,
      "grad_norm": 3.875041961669922,
      "learning_rate": 7.374379756651285e-08,
      "loss": 0.3094,
      "num_input_tokens_seen": 21232176,
      "step": 32250
    },
    {
      "epoch": 19.01827830188679,
      "grad_norm": 3.0990335941314697,
      "learning_rate": 7.330422236677015e-08,
      "loss": 0.2887,
      "num_input_tokens_seen": 21235216,
      "step": 32255
    },
    {
      "epoch": 19.02122641509434,
      "grad_norm": 3.228283405303955,
      "learning_rate": 7.286595153096765e-08,
      "loss": 0.3306,
      "num_input_tokens_seen": 21238256,
      "step": 32260
    },
    {
      "epoch": 19.024174528301888,
      "grad_norm": 3.612644672393799,
      "learning_rate": 7.242898517513864e-08,
      "loss": 0.2561,
      "num_input_tokens_seen": 21241296,
      "step": 32265
    },
    {
      "epoch": 19.027122641509433,
      "grad_norm": 3.3365607261657715,
      "learning_rate": 7.199332341497333e-08,
      "loss": 0.364,
      "num_input_tokens_seen": 21243600,
      "step": 32270
    },
    {
      "epoch": 19.03007075471698,
      "grad_norm": 4.85602331161499,
      "learning_rate": 7.155896636581394e-08,
      "loss": 0.229,
      "num_input_tokens_seen": 21246544,
      "step": 32275
    },
    {
      "epoch": 19.03301886792453,
      "grad_norm": 4.502285480499268,
      "learning_rate": 7.112591414265901e-08,
      "loss": 0.2356,
      "num_input_tokens_seen": 21250576,
      "step": 32280
    },
    {
      "epoch": 19.035966981132077,
      "grad_norm": 3.2664289474487305,
      "learning_rate": 7.069416686016018e-08,
      "loss": 0.2541,
      "num_input_tokens_seen": 21253392,
      "step": 32285
    },
    {
      "epoch": 19.03891509433962,
      "grad_norm": 3.3822925090789795,
      "learning_rate": 7.026372463262488e-08,
      "loss": 0.2761,
      "num_input_tokens_seen": 21258064,
      "step": 32290
    },
    {
      "epoch": 19.04186320754717,
      "grad_norm": 4.170190811157227,
      "learning_rate": 6.983458757401418e-08,
      "loss": 0.3237,
      "num_input_tokens_seen": 21261488,
      "step": 32295
    },
    {
      "epoch": 19.044811320754718,
      "grad_norm": 7.180066108703613,
      "learning_rate": 6.940675579794443e-08,
      "loss": 0.3082,
      "num_input_tokens_seen": 21264624,
      "step": 32300
    },
    {
      "epoch": 19.047759433962263,
      "grad_norm": 4.67556619644165,
      "learning_rate": 6.898022941768612e-08,
      "loss": 0.2444,
      "num_input_tokens_seen": 21267184,
      "step": 32305
    },
    {
      "epoch": 19.05070754716981,
      "grad_norm": 2.686640977859497,
      "learning_rate": 6.855500854616337e-08,
      "loss": 0.3667,
      "num_input_tokens_seen": 21270704,
      "step": 32310
    },
    {
      "epoch": 19.05365566037736,
      "grad_norm": 3.726238489151001,
      "learning_rate": 6.813109329595557e-08,
      "loss": 0.3057,
      "num_input_tokens_seen": 21273520,
      "step": 32315
    },
    {
      "epoch": 19.056603773584907,
      "grad_norm": 3.3229260444641113,
      "learning_rate": 6.770848377929573e-08,
      "loss": 0.3025,
      "num_input_tokens_seen": 21277040,
      "step": 32320
    },
    {
      "epoch": 19.059551886792452,
      "grad_norm": 4.1956634521484375,
      "learning_rate": 6.728718010807156e-08,
      "loss": 0.2931,
      "num_input_tokens_seen": 21279888,
      "step": 32325
    },
    {
      "epoch": 19.0625,
      "grad_norm": 3.006791114807129,
      "learning_rate": 6.68671823938255e-08,
      "loss": 0.382,
      "num_input_tokens_seen": 21282320,
      "step": 32330
    },
    {
      "epoch": 19.065448113207548,
      "grad_norm": 3.304567813873291,
      "learning_rate": 6.644849074775361e-08,
      "loss": 0.2964,
      "num_input_tokens_seen": 21285296,
      "step": 32335
    },
    {
      "epoch": 19.068396226415093,
      "grad_norm": 8.171113014221191,
      "learning_rate": 6.603110528070667e-08,
      "loss": 0.424,
      "num_input_tokens_seen": 21289008,
      "step": 32340
    },
    {
      "epoch": 19.07134433962264,
      "grad_norm": 2.8853719234466553,
      "learning_rate": 6.561502610318849e-08,
      "loss": 0.2611,
      "num_input_tokens_seen": 21291952,
      "step": 32345
    },
    {
      "epoch": 19.07429245283019,
      "grad_norm": 4.270910263061523,
      "learning_rate": 6.520025332535762e-08,
      "loss": 0.2914,
      "num_input_tokens_seen": 21295632,
      "step": 32350
    },
    {
      "epoch": 19.077240566037737,
      "grad_norm": 2.4229507446289062,
      "learning_rate": 6.47867870570279e-08,
      "loss": 0.419,
      "num_input_tokens_seen": 21298448,
      "step": 32355
    },
    {
      "epoch": 19.080188679245282,
      "grad_norm": 6.628303050994873,
      "learning_rate": 6.437462740766564e-08,
      "loss": 0.3058,
      "num_input_tokens_seen": 21301776,
      "step": 32360
    },
    {
      "epoch": 19.08313679245283,
      "grad_norm": 3.9984376430511475,
      "learning_rate": 6.396377448639246e-08,
      "loss": 0.3908,
      "num_input_tokens_seen": 21305168,
      "step": 32365
    },
    {
      "epoch": 19.08608490566038,
      "grad_norm": 3.5370075702667236,
      "learning_rate": 6.3554228401983e-08,
      "loss": 0.2703,
      "num_input_tokens_seen": 21307856,
      "step": 32370
    },
    {
      "epoch": 19.089033018867923,
      "grad_norm": 3.4733757972717285,
      "learning_rate": 6.314598926286663e-08,
      "loss": 0.2725,
      "num_input_tokens_seen": 21311088,
      "step": 32375
    },
    {
      "epoch": 19.09198113207547,
      "grad_norm": 8.322453498840332,
      "learning_rate": 6.273905717712637e-08,
      "loss": 0.3902,
      "num_input_tokens_seen": 21314544,
      "step": 32380
    },
    {
      "epoch": 19.09492924528302,
      "grad_norm": 3.7995259761810303,
      "learning_rate": 6.233343225249933e-08,
      "loss": 0.2765,
      "num_input_tokens_seen": 21317168,
      "step": 32385
    },
    {
      "epoch": 19.097877358490567,
      "grad_norm": 4.714277744293213,
      "learning_rate": 6.192911459637519e-08,
      "loss": 0.348,
      "num_input_tokens_seen": 21320208,
      "step": 32390
    },
    {
      "epoch": 19.100825471698112,
      "grad_norm": 3.7923521995544434,
      "learning_rate": 6.152610431580052e-08,
      "loss": 0.3421,
      "num_input_tokens_seen": 21323152,
      "step": 32395
    },
    {
      "epoch": 19.10377358490566,
      "grad_norm": 6.024499416351318,
      "learning_rate": 6.112440151747389e-08,
      "loss": 0.3556,
      "num_input_tokens_seen": 21326160,
      "step": 32400
    },
    {
      "epoch": 19.10672169811321,
      "grad_norm": 3.98486590385437,
      "learning_rate": 6.072400630774689e-08,
      "loss": 0.3095,
      "num_input_tokens_seen": 21329264,
      "step": 32405
    },
    {
      "epoch": 19.109669811320753,
      "grad_norm": 4.058884143829346,
      "learning_rate": 6.032491879262637e-08,
      "loss": 0.2777,
      "num_input_tokens_seen": 21332016,
      "step": 32410
    },
    {
      "epoch": 19.1126179245283,
      "grad_norm": 3.014965772628784,
      "learning_rate": 5.99271390777717e-08,
      "loss": 0.2691,
      "num_input_tokens_seen": 21334992,
      "step": 32415
    },
    {
      "epoch": 19.11556603773585,
      "grad_norm": 3.3017826080322266,
      "learning_rate": 5.953066726849865e-08,
      "loss": 0.2207,
      "num_input_tokens_seen": 21338032,
      "step": 32420
    },
    {
      "epoch": 19.118514150943398,
      "grad_norm": 3.427173137664795,
      "learning_rate": 5.913550346977326e-08,
      "loss": 0.3282,
      "num_input_tokens_seen": 21342032,
      "step": 32425
    },
    {
      "epoch": 19.121462264150942,
      "grad_norm": 1.9431265592575073,
      "learning_rate": 5.874164778621683e-08,
      "loss": 0.2933,
      "num_input_tokens_seen": 21346736,
      "step": 32430
    },
    {
      "epoch": 19.12441037735849,
      "grad_norm": 3.4423463344573975,
      "learning_rate": 5.834910032210539e-08,
      "loss": 0.2501,
      "num_input_tokens_seen": 21349488,
      "step": 32435
    },
    {
      "epoch": 19.12735849056604,
      "grad_norm": 4.3906331062316895,
      "learning_rate": 5.795786118136693e-08,
      "loss": 0.4349,
      "num_input_tokens_seen": 21352784,
      "step": 32440
    },
    {
      "epoch": 19.130306603773583,
      "grad_norm": 5.205923080444336,
      "learning_rate": 5.756793046758302e-08,
      "loss": 0.3724,
      "num_input_tokens_seen": 21355248,
      "step": 32445
    },
    {
      "epoch": 19.13325471698113,
      "grad_norm": 4.395692825317383,
      "learning_rate": 5.7179308283990544e-08,
      "loss": 0.276,
      "num_input_tokens_seen": 21358832,
      "step": 32450
    },
    {
      "epoch": 19.13620283018868,
      "grad_norm": 5.382297039031982,
      "learning_rate": 5.679199473347885e-08,
      "loss": 0.427,
      "num_input_tokens_seen": 21361552,
      "step": 32455
    },
    {
      "epoch": 19.139150943396228,
      "grad_norm": 3.999316930770874,
      "learning_rate": 5.6405989918590366e-08,
      "loss": 0.2596,
      "num_input_tokens_seen": 21365648,
      "step": 32460
    },
    {
      "epoch": 19.142099056603772,
      "grad_norm": 2.8459649085998535,
      "learning_rate": 5.6021293941522225e-08,
      "loss": 0.2447,
      "num_input_tokens_seen": 21369136,
      "step": 32465
    },
    {
      "epoch": 19.14504716981132,
      "grad_norm": 5.482738971710205,
      "learning_rate": 5.563790690412352e-08,
      "loss": 0.2511,
      "num_input_tokens_seen": 21372304,
      "step": 32470
    },
    {
      "epoch": 19.14799528301887,
      "grad_norm": 4.241137504577637,
      "learning_rate": 5.525582890789805e-08,
      "loss": 0.4385,
      "num_input_tokens_seen": 21375184,
      "step": 32475
    },
    {
      "epoch": 19.150943396226417,
      "grad_norm": 2.693159341812134,
      "learning_rate": 5.4875060054002115e-08,
      "loss": 0.3633,
      "num_input_tokens_seen": 21378960,
      "step": 32480
    },
    {
      "epoch": 19.15389150943396,
      "grad_norm": 4.197110652923584,
      "learning_rate": 5.4495600443246755e-08,
      "loss": 0.3565,
      "num_input_tokens_seen": 21381808,
      "step": 32485
    },
    {
      "epoch": 19.15683962264151,
      "grad_norm": 4.993841648101807,
      "learning_rate": 5.411745017609493e-08,
      "loss": 0.2618,
      "num_input_tokens_seen": 21384592,
      "step": 32490
    },
    {
      "epoch": 19.159787735849058,
      "grad_norm": 3.4104506969451904,
      "learning_rate": 5.374060935266434e-08,
      "loss": 0.2724,
      "num_input_tokens_seen": 21388304,
      "step": 32495
    },
    {
      "epoch": 19.162735849056602,
      "grad_norm": 5.683518409729004,
      "learning_rate": 5.3365078072724065e-08,
      "loss": 0.313,
      "num_input_tokens_seen": 21391824,
      "step": 32500
    },
    {
      "epoch": 19.16568396226415,
      "grad_norm": 2.3511602878570557,
      "learning_rate": 5.299085643569846e-08,
      "loss": 0.2316,
      "num_input_tokens_seen": 21395056,
      "step": 32505
    },
    {
      "epoch": 19.1686320754717,
      "grad_norm": 3.363922595977783,
      "learning_rate": 5.261794454066327e-08,
      "loss": 0.2785,
      "num_input_tokens_seen": 21398608,
      "step": 32510
    },
    {
      "epoch": 19.171580188679247,
      "grad_norm": 2.594398021697998,
      "learning_rate": 5.224634248635008e-08,
      "loss": 0.2068,
      "num_input_tokens_seen": 21402704,
      "step": 32515
    },
    {
      "epoch": 19.17452830188679,
      "grad_norm": 5.2465901374816895,
      "learning_rate": 5.187605037114129e-08,
      "loss": 0.3764,
      "num_input_tokens_seen": 21406352,
      "step": 32520
    },
    {
      "epoch": 19.17747641509434,
      "grad_norm": 3.2278695106506348,
      "learning_rate": 5.15070682930735e-08,
      "loss": 0.2558,
      "num_input_tokens_seen": 21409584,
      "step": 32525
    },
    {
      "epoch": 19.180424528301888,
      "grad_norm": 15.956546783447266,
      "learning_rate": 5.113939634983578e-08,
      "loss": 0.3528,
      "num_input_tokens_seen": 21413168,
      "step": 32530
    },
    {
      "epoch": 19.183372641509433,
      "grad_norm": 2.279872179031372,
      "learning_rate": 5.077303463877192e-08,
      "loss": 0.2476,
      "num_input_tokens_seen": 21416848,
      "step": 32535
    },
    {
      "epoch": 19.18632075471698,
      "grad_norm": 4.9924798011779785,
      "learning_rate": 5.040798325687601e-08,
      "loss": 0.374,
      "num_input_tokens_seen": 21420016,
      "step": 32540
    },
    {
      "epoch": 19.18926886792453,
      "grad_norm": 3.178415298461914,
      "learning_rate": 5.004424230079852e-08,
      "loss": 0.2976,
      "num_input_tokens_seen": 21424656,
      "step": 32545
    },
    {
      "epoch": 19.192216981132077,
      "grad_norm": 4.079249382019043,
      "learning_rate": 4.968181186684129e-08,
      "loss": 0.3522,
      "num_input_tokens_seen": 21427664,
      "step": 32550
    },
    {
      "epoch": 19.19516509433962,
      "grad_norm": 6.391392707824707,
      "learning_rate": 4.932069205095924e-08,
      "loss": 0.3396,
      "num_input_tokens_seen": 21430448,
      "step": 32555
    },
    {
      "epoch": 19.19811320754717,
      "grad_norm": 3.7254137992858887,
      "learning_rate": 4.896088294875978e-08,
      "loss": 0.3152,
      "num_input_tokens_seen": 21433616,
      "step": 32560
    },
    {
      "epoch": 19.201061320754718,
      "grad_norm": 3.066105842590332,
      "learning_rate": 4.8602384655505044e-08,
      "loss": 0.3001,
      "num_input_tokens_seen": 21436496,
      "step": 32565
    },
    {
      "epoch": 19.204009433962263,
      "grad_norm": 8.044846534729004,
      "learning_rate": 4.824519726610744e-08,
      "loss": 0.2978,
      "num_input_tokens_seen": 21439792,
      "step": 32570
    },
    {
      "epoch": 19.20695754716981,
      "grad_norm": 4.160444736480713,
      "learning_rate": 4.7889320875135206e-08,
      "loss": 0.3449,
      "num_input_tokens_seen": 21443024,
      "step": 32575
    },
    {
      "epoch": 19.20990566037736,
      "grad_norm": 3.9880566596984863,
      "learning_rate": 4.753475557680742e-08,
      "loss": 0.2996,
      "num_input_tokens_seen": 21446096,
      "step": 32580
    },
    {
      "epoch": 19.212853773584907,
      "grad_norm": 2.2955808639526367,
      "learning_rate": 4.718150146499734e-08,
      "loss": 0.35,
      "num_input_tokens_seen": 21450480,
      "step": 32585
    },
    {
      "epoch": 19.215801886792452,
      "grad_norm": 3.6039063930511475,
      "learning_rate": 4.682955863323013e-08,
      "loss": 0.2721,
      "num_input_tokens_seen": 21453360,
      "step": 32590
    },
    {
      "epoch": 19.21875,
      "grad_norm": 7.154551982879639,
      "learning_rate": 4.6478927174684606e-08,
      "loss": 0.3261,
      "num_input_tokens_seen": 21456112,
      "step": 32595
    },
    {
      "epoch": 19.221698113207548,
      "grad_norm": 2.7115204334259033,
      "learning_rate": 4.612960718219095e-08,
      "loss": 0.3093,
      "num_input_tokens_seen": 21458864,
      "step": 32600
    },
    {
      "epoch": 19.224646226415093,
      "grad_norm": 5.209589004516602,
      "learning_rate": 4.578159874823407e-08,
      "loss": 0.2758,
      "num_input_tokens_seen": 21462192,
      "step": 32605
    },
    {
      "epoch": 19.22759433962264,
      "grad_norm": 2.6899969577789307,
      "learning_rate": 4.5434901964950264e-08,
      "loss": 0.2303,
      "num_input_tokens_seen": 21465328,
      "step": 32610
    },
    {
      "epoch": 19.23054245283019,
      "grad_norm": 2.7749316692352295,
      "learning_rate": 4.508951692412944e-08,
      "loss": 0.3314,
      "num_input_tokens_seen": 21468752,
      "step": 32615
    },
    {
      "epoch": 19.233490566037737,
      "grad_norm": 2.3392293453216553,
      "learning_rate": 4.4745443717213455e-08,
      "loss": 0.1805,
      "num_input_tokens_seen": 21471696,
      "step": 32620
    },
    {
      "epoch": 19.236438679245282,
      "grad_norm": 2.5304219722747803,
      "learning_rate": 4.4402682435296666e-08,
      "loss": 0.3073,
      "num_input_tokens_seen": 21474928,
      "step": 32625
    },
    {
      "epoch": 19.23938679245283,
      "grad_norm": 3.3401236534118652,
      "learning_rate": 4.406123316912758e-08,
      "loss": 0.328,
      "num_input_tokens_seen": 21478448,
      "step": 32630
    },
    {
      "epoch": 19.24233490566038,
      "grad_norm": 8.572898864746094,
      "learning_rate": 4.372109600910612e-08,
      "loss": 0.3637,
      "num_input_tokens_seen": 21481456,
      "step": 32635
    },
    {
      "epoch": 19.245283018867923,
      "grad_norm": 3.776031255722046,
      "learning_rate": 4.338227104528414e-08,
      "loss": 0.3251,
      "num_input_tokens_seen": 21486000,
      "step": 32640
    },
    {
      "epoch": 19.24823113207547,
      "grad_norm": 3.9124088287353516,
      "learning_rate": 4.304475836736821e-08,
      "loss": 0.2865,
      "num_input_tokens_seen": 21489520,
      "step": 32645
    },
    {
      "epoch": 19.25117924528302,
      "grad_norm": 3.358508586883545,
      "learning_rate": 4.27085580647163e-08,
      "loss": 0.3941,
      "num_input_tokens_seen": 21493104,
      "step": 32650
    },
    {
      "epoch": 19.254127358490567,
      "grad_norm": 3.7605738639831543,
      "learning_rate": 4.237367022633776e-08,
      "loss": 0.3132,
      "num_input_tokens_seen": 21496016,
      "step": 32655
    },
    {
      "epoch": 19.257075471698112,
      "grad_norm": 5.3355631828308105,
      "learning_rate": 4.204009494089612e-08,
      "loss": 0.2877,
      "num_input_tokens_seen": 21499632,
      "step": 32660
    },
    {
      "epoch": 19.26002358490566,
      "grad_norm": 2.8282406330108643,
      "learning_rate": 4.170783229670739e-08,
      "loss": 0.2605,
      "num_input_tokens_seen": 21502736,
      "step": 32665
    },
    {
      "epoch": 19.26297169811321,
      "grad_norm": 2.4860422611236572,
      "learning_rate": 4.137688238173898e-08,
      "loss": 0.3163,
      "num_input_tokens_seen": 21506000,
      "step": 32670
    },
    {
      "epoch": 19.265919811320753,
      "grad_norm": 6.191598892211914,
      "learning_rate": 4.104724528361137e-08,
      "loss": 0.2895,
      "num_input_tokens_seen": 21509360,
      "step": 32675
    },
    {
      "epoch": 19.2688679245283,
      "grad_norm": 4.377939701080322,
      "learning_rate": 4.071892108959752e-08,
      "loss": 0.274,
      "num_input_tokens_seen": 21513232,
      "step": 32680
    },
    {
      "epoch": 19.27181603773585,
      "grad_norm": 3.1368260383605957,
      "learning_rate": 4.039190988662234e-08,
      "loss": 0.4243,
      "num_input_tokens_seen": 21516592,
      "step": 32685
    },
    {
      "epoch": 19.274764150943398,
      "grad_norm": 4.228139877319336,
      "learning_rate": 4.006621176126435e-08,
      "loss": 0.4868,
      "num_input_tokens_seen": 21522864,
      "step": 32690
    },
    {
      "epoch": 19.277712264150942,
      "grad_norm": 3.7591402530670166,
      "learning_rate": 3.974182679975236e-08,
      "loss": 0.3448,
      "num_input_tokens_seen": 21526544,
      "step": 32695
    },
    {
      "epoch": 19.28066037735849,
      "grad_norm": 4.710849285125732,
      "learning_rate": 3.941875508796933e-08,
      "loss": 0.3236,
      "num_input_tokens_seen": 21529936,
      "step": 32700
    },
    {
      "epoch": 19.28360849056604,
      "grad_norm": 2.730416774749756,
      "learning_rate": 3.909699671145017e-08,
      "loss": 0.3061,
      "num_input_tokens_seen": 21533168,
      "step": 32705
    },
    {
      "epoch": 19.286556603773583,
      "grad_norm": 2.1819090843200684,
      "learning_rate": 3.87765517553812e-08,
      "loss": 0.2455,
      "num_input_tokens_seen": 21535984,
      "step": 32710
    },
    {
      "epoch": 19.28950471698113,
      "grad_norm": 5.429643630981445,
      "learning_rate": 3.8457420304601756e-08,
      "loss": 0.3099,
      "num_input_tokens_seen": 21538928,
      "step": 32715
    },
    {
      "epoch": 19.29245283018868,
      "grad_norm": 2.9277901649475098,
      "learning_rate": 3.813960244360371e-08,
      "loss": 0.3078,
      "num_input_tokens_seen": 21543152,
      "step": 32720
    },
    {
      "epoch": 19.295400943396228,
      "grad_norm": 3.2288620471954346,
      "learning_rate": 3.7823098256529744e-08,
      "loss": 0.3203,
      "num_input_tokens_seen": 21546448,
      "step": 32725
    },
    {
      "epoch": 19.298349056603772,
      "grad_norm": 4.047641754150391,
      "learning_rate": 3.750790782717673e-08,
      "loss": 0.3431,
      "num_input_tokens_seen": 21549520,
      "step": 32730
    },
    {
      "epoch": 19.30129716981132,
      "grad_norm": 5.900483131408691,
      "learning_rate": 3.719403123899179e-08,
      "loss": 0.3832,
      "num_input_tokens_seen": 21551888,
      "step": 32735
    },
    {
      "epoch": 19.30424528301887,
      "grad_norm": 2.6652796268463135,
      "learning_rate": 3.688146857507624e-08,
      "loss": 0.3341,
      "num_input_tokens_seen": 21554896,
      "step": 32740
    },
    {
      "epoch": 19.307193396226417,
      "grad_norm": 2.82135009765625,
      "learning_rate": 3.657021991818166e-08,
      "loss": 0.4191,
      "num_input_tokens_seen": 21558960,
      "step": 32745
    },
    {
      "epoch": 19.31014150943396,
      "grad_norm": 4.203566551208496,
      "learning_rate": 3.626028535071213e-08,
      "loss": 0.2568,
      "num_input_tokens_seen": 21562064,
      "step": 32750
    },
    {
      "epoch": 19.31308962264151,
      "grad_norm": 3.810723066329956,
      "learning_rate": 3.59516649547248e-08,
      "loss": 0.3213,
      "num_input_tokens_seen": 21564624,
      "step": 32755
    },
    {
      "epoch": 19.316037735849058,
      "grad_norm": 3.3215043544769287,
      "learning_rate": 3.564435881192818e-08,
      "loss": 0.2942,
      "num_input_tokens_seen": 21567312,
      "step": 32760
    },
    {
      "epoch": 19.318985849056602,
      "grad_norm": 4.54376745223999,
      "learning_rate": 3.5338367003682763e-08,
      "loss": 0.284,
      "num_input_tokens_seen": 21569904,
      "step": 32765
    },
    {
      "epoch": 19.32193396226415,
      "grad_norm": 3.0733978748321533,
      "learning_rate": 3.5033689611000954e-08,
      "loss": 0.2839,
      "num_input_tokens_seen": 21573008,
      "step": 32770
    },
    {
      "epoch": 19.3248820754717,
      "grad_norm": 5.8661088943481445,
      "learning_rate": 3.473032671454768e-08,
      "loss": 0.5003,
      "num_input_tokens_seen": 21576240,
      "step": 32775
    },
    {
      "epoch": 19.327830188679247,
      "grad_norm": 3.543764352798462,
      "learning_rate": 3.44282783946398e-08,
      "loss": 0.2326,
      "num_input_tokens_seen": 21578960,
      "step": 32780
    },
    {
      "epoch": 19.33077830188679,
      "grad_norm": 5.593371868133545,
      "learning_rate": 3.4127544731245575e-08,
      "loss": 0.2679,
      "num_input_tokens_seen": 21582128,
      "step": 32785
    },
    {
      "epoch": 19.33372641509434,
      "grad_norm": 4.65841007232666,
      "learning_rate": 3.38281258039852e-08,
      "loss": 0.2334,
      "num_input_tokens_seen": 21585264,
      "step": 32790
    },
    {
      "epoch": 19.336674528301888,
      "grad_norm": 3.422523260116577,
      "learning_rate": 3.353002169213193e-08,
      "loss": 0.3433,
      "num_input_tokens_seen": 21588688,
      "step": 32795
    },
    {
      "epoch": 19.339622641509433,
      "grad_norm": 5.698583602905273,
      "learning_rate": 3.32332324746093e-08,
      "loss": 0.3634,
      "num_input_tokens_seen": 21592720,
      "step": 32800
    },
    {
      "epoch": 19.34257075471698,
      "grad_norm": 3.1233718395233154,
      "learning_rate": 3.2937758229994455e-08,
      "loss": 0.3475,
      "num_input_tokens_seen": 21595568,
      "step": 32805
    },
    {
      "epoch": 19.34551886792453,
      "grad_norm": 3.2980191707611084,
      "learning_rate": 3.2643599036514815e-08,
      "loss": 0.3581,
      "num_input_tokens_seen": 21599120,
      "step": 32810
    },
    {
      "epoch": 19.348466981132077,
      "grad_norm": 3.699122667312622,
      "learning_rate": 3.2350754972050316e-08,
      "loss": 0.3286,
      "num_input_tokens_seen": 21603600,
      "step": 32815
    },
    {
      "epoch": 19.35141509433962,
      "grad_norm": 3.080320358276367,
      "learning_rate": 3.2059226114132815e-08,
      "loss": 0.3051,
      "num_input_tokens_seen": 21607152,
      "step": 32820
    },
    {
      "epoch": 19.35436320754717,
      "grad_norm": 8.699760437011719,
      "learning_rate": 3.1769012539945575e-08,
      "loss": 0.3464,
      "num_input_tokens_seen": 21609936,
      "step": 32825
    },
    {
      "epoch": 19.357311320754718,
      "grad_norm": 6.173746109008789,
      "learning_rate": 3.1480114326324364e-08,
      "loss": 0.3146,
      "num_input_tokens_seen": 21614384,
      "step": 32830
    },
    {
      "epoch": 19.360259433962263,
      "grad_norm": 5.0818986892700195,
      "learning_rate": 3.1192531549756325e-08,
      "loss": 0.3227,
      "num_input_tokens_seen": 21617872,
      "step": 32835
    },
    {
      "epoch": 19.36320754716981,
      "grad_norm": 3.7998483180999756,
      "learning_rate": 3.090626428638e-08,
      "loss": 0.3422,
      "num_input_tokens_seen": 21621328,
      "step": 32840
    },
    {
      "epoch": 19.36615566037736,
      "grad_norm": 2.093597888946533,
      "learning_rate": 3.062131261198531e-08,
      "loss": 0.3451,
      "num_input_tokens_seen": 21624432,
      "step": 32845
    },
    {
      "epoch": 19.369103773584907,
      "grad_norm": 3.480013370513916,
      "learning_rate": 3.033767660201525e-08,
      "loss": 0.2737,
      "num_input_tokens_seen": 21627248,
      "step": 32850
    },
    {
      "epoch": 19.372051886792452,
      "grad_norm": 3.2013187408447266,
      "learning_rate": 3.005535633156309e-08,
      "loss": 0.2994,
      "num_input_tokens_seen": 21630064,
      "step": 32855
    },
    {
      "epoch": 19.375,
      "grad_norm": 5.60385274887085,
      "learning_rate": 2.977435187537514e-08,
      "loss": 0.2403,
      "num_input_tokens_seen": 21633264,
      "step": 32860
    },
    {
      "epoch": 19.377948113207548,
      "grad_norm": 1.8953135013580322,
      "learning_rate": 2.949466330784745e-08,
      "loss": 0.2793,
      "num_input_tokens_seen": 21636656,
      "step": 32865
    },
    {
      "epoch": 19.380896226415093,
      "grad_norm": 2.701587677001953,
      "learning_rate": 2.921629070302967e-08,
      "loss": 0.2044,
      "num_input_tokens_seen": 21640752,
      "step": 32870
    },
    {
      "epoch": 19.38384433962264,
      "grad_norm": 4.137882709503174,
      "learning_rate": 2.893923413462174e-08,
      "loss": 0.2555,
      "num_input_tokens_seen": 21644144,
      "step": 32875
    },
    {
      "epoch": 19.38679245283019,
      "grad_norm": 4.986118316650391,
      "learning_rate": 2.866349367597554e-08,
      "loss": 0.3363,
      "num_input_tokens_seen": 21646928,
      "step": 32880
    },
    {
      "epoch": 19.389740566037737,
      "grad_norm": 3.912252902984619,
      "learning_rate": 2.8389069400094893e-08,
      "loss": 0.3627,
      "num_input_tokens_seen": 21649296,
      "step": 32885
    },
    {
      "epoch": 19.392688679245282,
      "grad_norm": 4.870301723480225,
      "learning_rate": 2.811596137963446e-08,
      "loss": 0.1841,
      "num_input_tokens_seen": 21651984,
      "step": 32890
    },
    {
      "epoch": 19.39563679245283,
      "grad_norm": 3.4312751293182373,
      "learning_rate": 2.7844169686900844e-08,
      "loss": 0.4163,
      "num_input_tokens_seen": 21655120,
      "step": 32895
    },
    {
      "epoch": 19.39858490566038,
      "grad_norm": 6.080343246459961,
      "learning_rate": 2.7573694393852047e-08,
      "loss": 0.2757,
      "num_input_tokens_seen": 21658224,
      "step": 32900
    },
    {
      "epoch": 19.401533018867923,
      "grad_norm": 2.986525535583496,
      "learning_rate": 2.7304535572098e-08,
      "loss": 0.2021,
      "num_input_tokens_seen": 21661456,
      "step": 32905
    },
    {
      "epoch": 19.40448113207547,
      "grad_norm": 5.440608024597168,
      "learning_rate": 2.703669329289893e-08,
      "loss": 0.3331,
      "num_input_tokens_seen": 21664272,
      "step": 32910
    },
    {
      "epoch": 19.40742924528302,
      "grad_norm": 4.344691276550293,
      "learning_rate": 2.6770167627167554e-08,
      "loss": 0.2606,
      "num_input_tokens_seen": 21667088,
      "step": 32915
    },
    {
      "epoch": 19.410377358490567,
      "grad_norm": 3.8293631076812744,
      "learning_rate": 2.6504958645467426e-08,
      "loss": 0.4221,
      "num_input_tokens_seen": 21669744,
      "step": 32920
    },
    {
      "epoch": 19.413325471698112,
      "grad_norm": 3.1105282306671143,
      "learning_rate": 2.6241066418014605e-08,
      "loss": 0.3385,
      "num_input_tokens_seen": 21673008,
      "step": 32925
    },
    {
      "epoch": 19.41627358490566,
      "grad_norm": 4.527070045471191,
      "learning_rate": 2.5978491014674866e-08,
      "loss": 0.2689,
      "num_input_tokens_seen": 21676080,
      "step": 32930
    },
    {
      "epoch": 19.41922169811321,
      "grad_norm": 1.8748838901519775,
      "learning_rate": 2.571723250496705e-08,
      "loss": 0.2322,
      "num_input_tokens_seen": 21679216,
      "step": 32935
    },
    {
      "epoch": 19.422169811320753,
      "grad_norm": 5.888749122619629,
      "learning_rate": 2.5457290958059155e-08,
      "loss": 0.2334,
      "num_input_tokens_seen": 21682064,
      "step": 32940
    },
    {
      "epoch": 19.4251179245283,
      "grad_norm": 6.817114353179932,
      "learning_rate": 2.51986664427728e-08,
      "loss": 0.3414,
      "num_input_tokens_seen": 21684912,
      "step": 32945
    },
    {
      "epoch": 19.42806603773585,
      "grad_norm": 3.5051259994506836,
      "learning_rate": 2.4941359027579883e-08,
      "loss": 0.2539,
      "num_input_tokens_seen": 21687920,
      "step": 32950
    },
    {
      "epoch": 19.431014150943398,
      "grad_norm": 3.621793508529663,
      "learning_rate": 2.468536878060368e-08,
      "loss": 0.2879,
      "num_input_tokens_seen": 21690864,
      "step": 32955
    },
    {
      "epoch": 19.433962264150942,
      "grad_norm": 10.94771671295166,
      "learning_rate": 2.443069576961832e-08,
      "loss": 0.2686,
      "num_input_tokens_seen": 21693648,
      "step": 32960
    },
    {
      "epoch": 19.43691037735849,
      "grad_norm": 6.247159004211426,
      "learning_rate": 2.4177340062049304e-08,
      "loss": 0.3295,
      "num_input_tokens_seen": 21696976,
      "step": 32965
    },
    {
      "epoch": 19.43985849056604,
      "grad_norm": 3.7519142627716064,
      "learning_rate": 2.3925301724974647e-08,
      "loss": 0.2582,
      "num_input_tokens_seen": 21700912,
      "step": 32970
    },
    {
      "epoch": 19.442806603773583,
      "grad_norm": 3.593998670578003,
      "learning_rate": 2.367458082512153e-08,
      "loss": 0.1738,
      "num_input_tokens_seen": 21704112,
      "step": 32975
    },
    {
      "epoch": 19.44575471698113,
      "grad_norm": 5.1533074378967285,
      "learning_rate": 2.3425177428870737e-08,
      "loss": 0.2444,
      "num_input_tokens_seen": 21708176,
      "step": 32980
    },
    {
      "epoch": 19.44870283018868,
      "grad_norm": 3.955094814300537,
      "learning_rate": 2.3177091602251677e-08,
      "loss": 0.2761,
      "num_input_tokens_seen": 21711344,
      "step": 32985
    },
    {
      "epoch": 19.451650943396228,
      "grad_norm": 5.484734535217285,
      "learning_rate": 2.2930323410946254e-08,
      "loss": 0.3634,
      "num_input_tokens_seen": 21714096,
      "step": 32990
    },
    {
      "epoch": 19.454599056603772,
      "grad_norm": 5.268672466278076,
      "learning_rate": 2.2684872920287758e-08,
      "loss": 0.2577,
      "num_input_tokens_seen": 21717744,
      "step": 32995
    },
    {
      "epoch": 19.45754716981132,
      "grad_norm": 3.444180488586426,
      "learning_rate": 2.2440740195260323e-08,
      "loss": 0.4002,
      "num_input_tokens_seen": 21720752,
      "step": 33000
    },
    {
      "epoch": 19.46049528301887,
      "grad_norm": 3.771821975708008,
      "learning_rate": 2.219792530049891e-08,
      "loss": 0.3104,
      "num_input_tokens_seen": 21724784,
      "step": 33005
    },
    {
      "epoch": 19.463443396226417,
      "grad_norm": 4.599825859069824,
      "learning_rate": 2.1956428300290434e-08,
      "loss": 0.2844,
      "num_input_tokens_seen": 21727728,
      "step": 33010
    },
    {
      "epoch": 19.46639150943396,
      "grad_norm": 3.0105531215667725,
      "learning_rate": 2.1716249258570966e-08,
      "loss": 0.3298,
      "num_input_tokens_seen": 21730960,
      "step": 33015
    },
    {
      "epoch": 19.46933962264151,
      "grad_norm": 2.413891553878784,
      "learning_rate": 2.1477388238930196e-08,
      "loss": 0.2423,
      "num_input_tokens_seen": 21734480,
      "step": 33020
    },
    {
      "epoch": 19.472287735849058,
      "grad_norm": 3.628007411956787,
      "learning_rate": 2.1239845304606988e-08,
      "loss": 0.2789,
      "num_input_tokens_seen": 21737392,
      "step": 33025
    },
    {
      "epoch": 19.475235849056602,
      "grad_norm": 1.9162883758544922,
      "learning_rate": 2.100362051849214e-08,
      "loss": 0.3375,
      "num_input_tokens_seen": 21741424,
      "step": 33030
    },
    {
      "epoch": 19.47818396226415,
      "grad_norm": 3.2622172832489014,
      "learning_rate": 2.076871394312674e-08,
      "loss": 0.3623,
      "num_input_tokens_seen": 21744560,
      "step": 33035
    },
    {
      "epoch": 19.4811320754717,
      "grad_norm": 2.875668525695801,
      "learning_rate": 2.0535125640703813e-08,
      "loss": 0.3243,
      "num_input_tokens_seen": 21747568,
      "step": 33040
    },
    {
      "epoch": 19.484080188679247,
      "grad_norm": 4.9066901206970215,
      "learning_rate": 2.0302855673066667e-08,
      "loss": 0.2068,
      "num_input_tokens_seen": 21750224,
      "step": 33045
    },
    {
      "epoch": 19.48702830188679,
      "grad_norm": 3.9858763217926025,
      "learning_rate": 2.0071904101710004e-08,
      "loss": 0.2446,
      "num_input_tokens_seen": 21756432,
      "step": 33050
    },
    {
      "epoch": 19.48997641509434,
      "grad_norm": 5.732495307922363,
      "learning_rate": 1.98422709877788e-08,
      "loss": 0.2636,
      "num_input_tokens_seen": 21759568,
      "step": 33055
    },
    {
      "epoch": 19.492924528301888,
      "grad_norm": 5.350522994995117,
      "learning_rate": 1.961395639206942e-08,
      "loss": 0.3139,
      "num_input_tokens_seen": 21762640,
      "step": 33060
    },
    {
      "epoch": 19.495872641509433,
      "grad_norm": 5.364588737487793,
      "learning_rate": 1.9386960375029628e-08,
      "loss": 0.2707,
      "num_input_tokens_seen": 21765616,
      "step": 33065
    },
    {
      "epoch": 19.49882075471698,
      "grad_norm": 3.3591549396514893,
      "learning_rate": 1.9161282996757458e-08,
      "loss": 0.2647,
      "num_input_tokens_seen": 21769552,
      "step": 33070
    },
    {
      "epoch": 19.50176886792453,
      "grad_norm": 4.25196647644043,
      "learning_rate": 1.8936924317001225e-08,
      "loss": 0.308,
      "num_input_tokens_seen": 21772880,
      "step": 33075
    },
    {
      "epoch": 19.504716981132077,
      "grad_norm": 2.9450671672821045,
      "learning_rate": 1.8713884395162308e-08,
      "loss": 0.2911,
      "num_input_tokens_seen": 21776208,
      "step": 33080
    },
    {
      "epoch": 19.50766509433962,
      "grad_norm": 3.5320394039154053,
      "learning_rate": 1.8492163290290132e-08,
      "loss": 0.216,
      "num_input_tokens_seen": 21778608,
      "step": 33085
    },
    {
      "epoch": 19.51061320754717,
      "grad_norm": 3.5144801139831543,
      "learning_rate": 1.827176106108719e-08,
      "loss": 0.2731,
      "num_input_tokens_seen": 21781264,
      "step": 33090
    },
    {
      "epoch": 19.513561320754718,
      "grad_norm": 5.058701992034912,
      "learning_rate": 1.8052677765905137e-08,
      "loss": 0.3999,
      "num_input_tokens_seen": 21784208,
      "step": 33095
    },
    {
      "epoch": 19.516509433962263,
      "grad_norm": 5.439336776733398,
      "learning_rate": 1.783491346274757e-08,
      "loss": 0.3224,
      "num_input_tokens_seen": 21787088,
      "step": 33100
    },
    {
      "epoch": 19.51945754716981,
      "grad_norm": 2.542010545730591,
      "learning_rate": 1.7618468209268936e-08,
      "loss": 0.2853,
      "num_input_tokens_seen": 21790768,
      "step": 33105
    },
    {
      "epoch": 19.52240566037736,
      "grad_norm": 4.11661434173584,
      "learning_rate": 1.7403342062773943e-08,
      "loss": 0.2907,
      "num_input_tokens_seen": 21793488,
      "step": 33110
    },
    {
      "epoch": 19.525353773584907,
      "grad_norm": 2.832939863204956,
      "learning_rate": 1.718953508021759e-08,
      "loss": 0.3645,
      "num_input_tokens_seen": 21796656,
      "step": 33115
    },
    {
      "epoch": 19.528301886792452,
      "grad_norm": 2.7160298824310303,
      "learning_rate": 1.6977047318206262e-08,
      "loss": 0.2268,
      "num_input_tokens_seen": 21799600,
      "step": 33120
    },
    {
      "epoch": 19.53125,
      "grad_norm": 3.661931037902832,
      "learning_rate": 1.676587883299774e-08,
      "loss": 0.3875,
      "num_input_tokens_seen": 21802992,
      "step": 33125
    },
    {
      "epoch": 19.534198113207548,
      "grad_norm": 4.054445266723633,
      "learning_rate": 1.655602968049952e-08,
      "loss": 0.2487,
      "num_input_tokens_seen": 21805552,
      "step": 33130
    },
    {
      "epoch": 19.537146226415093,
      "grad_norm": 9.847738265991211,
      "learning_rate": 1.634749991626938e-08,
      "loss": 0.5729,
      "num_input_tokens_seen": 21807856,
      "step": 33135
    },
    {
      "epoch": 19.54009433962264,
      "grad_norm": 3.4979493618011475,
      "learning_rate": 1.6140289595517056e-08,
      "loss": 0.2389,
      "num_input_tokens_seen": 21810928,
      "step": 33140
    },
    {
      "epoch": 19.54304245283019,
      "grad_norm": 4.454919338226318,
      "learning_rate": 1.5934398773102545e-08,
      "loss": 0.278,
      "num_input_tokens_seen": 21813296,
      "step": 33145
    },
    {
      "epoch": 19.545990566037737,
      "grad_norm": 1.7939752340316772,
      "learning_rate": 1.5729827503536133e-08,
      "loss": 0.2338,
      "num_input_tokens_seen": 21817264,
      "step": 33150
    },
    {
      "epoch": 19.548938679245282,
      "grad_norm": 3.9831128120422363,
      "learning_rate": 1.5526575840978942e-08,
      "loss": 0.4554,
      "num_input_tokens_seen": 21820048,
      "step": 33155
    },
    {
      "epoch": 19.55188679245283,
      "grad_norm": 3.663813352584839,
      "learning_rate": 1.532464383924237e-08,
      "loss": 0.2952,
      "num_input_tokens_seen": 21824112,
      "step": 33160
    },
    {
      "epoch": 19.55483490566038,
      "grad_norm": 4.076108932495117,
      "learning_rate": 1.5124031551789208e-08,
      "loss": 0.3902,
      "num_input_tokens_seen": 21828048,
      "step": 33165
    },
    {
      "epoch": 19.557783018867923,
      "grad_norm": 2.813528537750244,
      "learning_rate": 1.4924739031732527e-08,
      "loss": 0.4169,
      "num_input_tokens_seen": 21830928,
      "step": 33170
    },
    {
      "epoch": 19.56073113207547,
      "grad_norm": 2.4708945751190186,
      "learning_rate": 1.4726766331835118e-08,
      "loss": 0.3696,
      "num_input_tokens_seen": 21835792,
      "step": 33175
    },
    {
      "epoch": 19.56367924528302,
      "grad_norm": 3.0704009532928467,
      "learning_rate": 1.4530113504512278e-08,
      "loss": 0.2224,
      "num_input_tokens_seen": 21838544,
      "step": 33180
    },
    {
      "epoch": 19.566627358490567,
      "grad_norm": 2.2923977375030518,
      "learning_rate": 1.4334780601827914e-08,
      "loss": 0.216,
      "num_input_tokens_seen": 21842096,
      "step": 33185
    },
    {
      "epoch": 19.569575471698112,
      "grad_norm": 3.1126158237457275,
      "learning_rate": 1.4140767675497325e-08,
      "loss": 0.3483,
      "num_input_tokens_seen": 21845776,
      "step": 33190
    },
    {
      "epoch": 19.57252358490566,
      "grad_norm": 3.1400578022003174,
      "learning_rate": 1.394807477688609e-08,
      "loss": 0.3862,
      "num_input_tokens_seen": 21849328,
      "step": 33195
    },
    {
      "epoch": 19.57547169811321,
      "grad_norm": 4.118099212646484,
      "learning_rate": 1.3756701957011177e-08,
      "loss": 0.2353,
      "num_input_tokens_seen": 21852464,
      "step": 33200
    },
    {
      "epoch": 19.578419811320753,
      "grad_norm": 4.612195014953613,
      "learning_rate": 1.3566649266538723e-08,
      "loss": 0.2538,
      "num_input_tokens_seen": 21855056,
      "step": 33205
    },
    {
      "epoch": 19.5813679245283,
      "grad_norm": 3.7246270179748535,
      "learning_rate": 1.3377916755786257e-08,
      "loss": 0.4268,
      "num_input_tokens_seen": 21858512,
      "step": 33210
    },
    {
      "epoch": 19.58431603773585,
      "grad_norm": 3.6216540336608887,
      "learning_rate": 1.3190504474721588e-08,
      "loss": 0.3171,
      "num_input_tokens_seen": 21862128,
      "step": 33215
    },
    {
      "epoch": 19.587264150943398,
      "grad_norm": 3.371044397354126,
      "learning_rate": 1.3004412472962802e-08,
      "loss": 0.2404,
      "num_input_tokens_seen": 21864944,
      "step": 33220
    },
    {
      "epoch": 19.590212264150942,
      "grad_norm": 5.70059061050415,
      "learning_rate": 1.2819640799778266e-08,
      "loss": 0.2404,
      "num_input_tokens_seen": 21867952,
      "step": 33225
    },
    {
      "epoch": 19.59316037735849,
      "grad_norm": 4.59685754776001,
      "learning_rate": 1.2636189504087737e-08,
      "loss": 0.3266,
      "num_input_tokens_seen": 21870352,
      "step": 33230
    },
    {
      "epoch": 19.59610849056604,
      "grad_norm": 3.030513286590576,
      "learning_rate": 1.2454058634460142e-08,
      "loss": 0.3808,
      "num_input_tokens_seen": 21873488,
      "step": 33235
    },
    {
      "epoch": 19.599056603773583,
      "grad_norm": 3.2464540004730225,
      "learning_rate": 1.2273248239115798e-08,
      "loss": 0.2503,
      "num_input_tokens_seen": 21876208,
      "step": 33240
    },
    {
      "epoch": 19.60200471698113,
      "grad_norm": 4.032291412353516,
      "learning_rate": 1.2093758365924746e-08,
      "loss": 0.3629,
      "num_input_tokens_seen": 21879632,
      "step": 33245
    },
    {
      "epoch": 19.60495283018868,
      "grad_norm": 8.33803653717041,
      "learning_rate": 1.1915589062408417e-08,
      "loss": 0.5949,
      "num_input_tokens_seen": 21882576,
      "step": 33250
    },
    {
      "epoch": 19.607900943396228,
      "grad_norm": 2.5552165508270264,
      "learning_rate": 1.1738740375736301e-08,
      "loss": 0.2428,
      "num_input_tokens_seen": 21885936,
      "step": 33255
    },
    {
      "epoch": 19.610849056603772,
      "grad_norm": 3.129723072052002,
      "learning_rate": 1.15632123527315e-08,
      "loss": 0.2147,
      "num_input_tokens_seen": 21888400,
      "step": 33260
    },
    {
      "epoch": 19.61379716981132,
      "grad_norm": 4.269466876983643,
      "learning_rate": 1.1389005039865176e-08,
      "loss": 0.2427,
      "num_input_tokens_seen": 21890960,
      "step": 33265
    },
    {
      "epoch": 19.61674528301887,
      "grad_norm": 5.059965133666992,
      "learning_rate": 1.1216118483259875e-08,
      "loss": 0.2302,
      "num_input_tokens_seen": 21893712,
      "step": 33270
    },
    {
      "epoch": 19.619693396226417,
      "grad_norm": 3.5500428676605225,
      "learning_rate": 1.1044552728687319e-08,
      "loss": 0.312,
      "num_input_tokens_seen": 21896304,
      "step": 33275
    },
    {
      "epoch": 19.62264150943396,
      "grad_norm": 3.5952975749969482,
      "learning_rate": 1.0874307821570618e-08,
      "loss": 0.1781,
      "num_input_tokens_seen": 21899504,
      "step": 33280
    },
    {
      "epoch": 19.62558962264151,
      "grad_norm": 2.671367645263672,
      "learning_rate": 1.0705383806982606e-08,
      "loss": 0.2039,
      "num_input_tokens_seen": 21902448,
      "step": 33285
    },
    {
      "epoch": 19.628537735849058,
      "grad_norm": 3.1334474086761475,
      "learning_rate": 1.0537780729646952e-08,
      "loss": 0.3002,
      "num_input_tokens_seen": 21905200,
      "step": 33290
    },
    {
      "epoch": 19.631485849056602,
      "grad_norm": 3.290881633758545,
      "learning_rate": 1.0371498633937605e-08,
      "loss": 0.2632,
      "num_input_tokens_seen": 21908432,
      "step": 33295
    },
    {
      "epoch": 19.63443396226415,
      "grad_norm": 3.439811944961548,
      "learning_rate": 1.0206537563877683e-08,
      "loss": 0.232,
      "num_input_tokens_seen": 21912304,
      "step": 33300
    },
    {
      "epoch": 19.6373820754717,
      "grad_norm": 2.8922455310821533,
      "learning_rate": 1.0042897563141695e-08,
      "loss": 0.3567,
      "num_input_tokens_seen": 21915408,
      "step": 33305
    },
    {
      "epoch": 19.640330188679247,
      "grad_norm": 6.978609561920166,
      "learning_rate": 9.88057867505443e-09,
      "loss": 0.3008,
      "num_input_tokens_seen": 21918288,
      "step": 33310
    },
    {
      "epoch": 19.64327830188679,
      "grad_norm": 3.0820319652557373,
      "learning_rate": 9.7195809425904e-09,
      "loss": 0.3639,
      "num_input_tokens_seen": 21922480,
      "step": 33315
    },
    {
      "epoch": 19.64622641509434,
      "grad_norm": 3.548482894897461,
      "learning_rate": 9.559904408373844e-09,
      "loss": 0.3291,
      "num_input_tokens_seen": 21925136,
      "step": 33320
    },
    {
      "epoch": 19.649174528301888,
      "grad_norm": 3.4397706985473633,
      "learning_rate": 9.401549114680387e-09,
      "loss": 0.2658,
      "num_input_tokens_seen": 21929296,
      "step": 33325
    },
    {
      "epoch": 19.652122641509433,
      "grad_norm": 5.406820774078369,
      "learning_rate": 9.244515103434826e-09,
      "loss": 0.4028,
      "num_input_tokens_seen": 21932048,
      "step": 33330
    },
    {
      "epoch": 19.65507075471698,
      "grad_norm": 4.270076751708984,
      "learning_rate": 9.08880241621335e-09,
      "loss": 0.3923,
      "num_input_tokens_seen": 21941648,
      "step": 33335
    },
    {
      "epoch": 19.65801886792453,
      "grad_norm": 2.8622243404388428,
      "learning_rate": 8.934411094240758e-09,
      "loss": 0.2242,
      "num_input_tokens_seen": 21945680,
      "step": 33340
    },
    {
      "epoch": 19.660966981132077,
      "grad_norm": 3.5761406421661377,
      "learning_rate": 8.781341178393244e-09,
      "loss": 0.2329,
      "num_input_tokens_seen": 21949936,
      "step": 33345
    },
    {
      "epoch": 19.66391509433962,
      "grad_norm": 8.338662147521973,
      "learning_rate": 8.629592709196167e-09,
      "loss": 0.2693,
      "num_input_tokens_seen": 21952944,
      "step": 33350
    },
    {
      "epoch": 19.66686320754717,
      "grad_norm": 5.499160289764404,
      "learning_rate": 8.479165726826277e-09,
      "loss": 0.2872,
      "num_input_tokens_seen": 21955632,
      "step": 33355
    },
    {
      "epoch": 19.669811320754718,
      "grad_norm": 3.6245574951171875,
      "learning_rate": 8.330060271109496e-09,
      "loss": 0.2962,
      "num_input_tokens_seen": 21959088,
      "step": 33360
    },
    {
      "epoch": 19.672759433962263,
      "grad_norm": 4.02509880065918,
      "learning_rate": 8.18227638152258e-09,
      "loss": 0.2854,
      "num_input_tokens_seen": 21962544,
      "step": 33365
    },
    {
      "epoch": 19.67570754716981,
      "grad_norm": 3.0707826614379883,
      "learning_rate": 8.035814097191452e-09,
      "loss": 0.2024,
      "num_input_tokens_seen": 21965808,
      "step": 33370
    },
    {
      "epoch": 19.67865566037736,
      "grad_norm": 3.3304855823516846,
      "learning_rate": 7.890673456892317e-09,
      "loss": 0.302,
      "num_input_tokens_seen": 21969232,
      "step": 33375
    },
    {
      "epoch": 19.681603773584907,
      "grad_norm": 4.168802261352539,
      "learning_rate": 7.746854499052215e-09,
      "loss": 0.3199,
      "num_input_tokens_seen": 21973008,
      "step": 33380
    },
    {
      "epoch": 19.684551886792452,
      "grad_norm": 17.163881301879883,
      "learning_rate": 7.604357261747907e-09,
      "loss": 0.3323,
      "num_input_tokens_seen": 21975888,
      "step": 33385
    },
    {
      "epoch": 19.6875,
      "grad_norm": 4.095640659332275,
      "learning_rate": 7.463181782705886e-09,
      "loss": 0.2802,
      "num_input_tokens_seen": 21978960,
      "step": 33390
    },
    {
      "epoch": 19.690448113207548,
      "grad_norm": 5.29165506362915,
      "learning_rate": 7.3233280993034726e-09,
      "loss": 0.5767,
      "num_input_tokens_seen": 21982096,
      "step": 33395
    },
    {
      "epoch": 19.693396226415093,
      "grad_norm": 3.641721487045288,
      "learning_rate": 7.184796248567161e-09,
      "loss": 0.2984,
      "num_input_tokens_seen": 21986480,
      "step": 33400
    },
    {
      "epoch": 19.69634433962264,
      "grad_norm": 4.412236213684082,
      "learning_rate": 7.047586267173723e-09,
      "loss": 0.1844,
      "num_input_tokens_seen": 21989328,
      "step": 33405
    },
    {
      "epoch": 19.69929245283019,
      "grad_norm": 8.34953498840332,
      "learning_rate": 6.9116981914502114e-09,
      "loss": 0.3878,
      "num_input_tokens_seen": 21993840,
      "step": 33410
    },
    {
      "epoch": 19.702240566037737,
      "grad_norm": 5.094158172607422,
      "learning_rate": 6.7771320573734036e-09,
      "loss": 0.2825,
      "num_input_tokens_seen": 21996944,
      "step": 33415
    },
    {
      "epoch": 19.705188679245282,
      "grad_norm": 2.4464001655578613,
      "learning_rate": 6.6438879005709114e-09,
      "loss": 0.2576,
      "num_input_tokens_seen": 22001392,
      "step": 33420
    },
    {
      "epoch": 19.70813679245283,
      "grad_norm": 7.755564212799072,
      "learning_rate": 6.511965756318961e-09,
      "loss": 0.3424,
      "num_input_tokens_seen": 22004336,
      "step": 33425
    },
    {
      "epoch": 19.71108490566038,
      "grad_norm": 5.430515766143799,
      "learning_rate": 6.381365659545169e-09,
      "loss": 0.3734,
      "num_input_tokens_seen": 22006768,
      "step": 33430
    },
    {
      "epoch": 19.714033018867923,
      "grad_norm": 4.830272674560547,
      "learning_rate": 6.252087644825766e-09,
      "loss": 0.3722,
      "num_input_tokens_seen": 22010160,
      "step": 33435
    },
    {
      "epoch": 19.71698113207547,
      "grad_norm": 2.892995595932007,
      "learning_rate": 6.124131746388373e-09,
      "loss": 0.2388,
      "num_input_tokens_seen": 22013648,
      "step": 33440
    },
    {
      "epoch": 19.71992924528302,
      "grad_norm": 2.4305663108825684,
      "learning_rate": 5.997497998109225e-09,
      "loss": 0.2748,
      "num_input_tokens_seen": 22016240,
      "step": 33445
    },
    {
      "epoch": 19.722877358490567,
      "grad_norm": 5.350555896759033,
      "learning_rate": 5.8721864335153925e-09,
      "loss": 0.2211,
      "num_input_tokens_seen": 22019376,
      "step": 33450
    },
    {
      "epoch": 19.725825471698112,
      "grad_norm": 3.699831962585449,
      "learning_rate": 5.748197085784224e-09,
      "loss": 0.2945,
      "num_input_tokens_seen": 22022416,
      "step": 33455
    },
    {
      "epoch": 19.72877358490566,
      "grad_norm": 1.709697961807251,
      "learning_rate": 5.62552998774113e-09,
      "loss": 0.216,
      "num_input_tokens_seen": 22026064,
      "step": 33460
    },
    {
      "epoch": 19.73172169811321,
      "grad_norm": 3.725552797317505,
      "learning_rate": 5.504185171864018e-09,
      "loss": 0.3391,
      "num_input_tokens_seen": 22028816,
      "step": 33465
    },
    {
      "epoch": 19.734669811320753,
      "grad_norm": 1.7166215181350708,
      "learning_rate": 5.384162670278858e-09,
      "loss": 0.3647,
      "num_input_tokens_seen": 22032432,
      "step": 33470
    },
    {
      "epoch": 19.7376179245283,
      "grad_norm": 4.447828769683838,
      "learning_rate": 5.265462514762454e-09,
      "loss": 0.3746,
      "num_input_tokens_seen": 22036528,
      "step": 33475
    },
    {
      "epoch": 19.74056603773585,
      "grad_norm": 4.044922828674316,
      "learning_rate": 5.148084736740777e-09,
      "loss": 0.3347,
      "num_input_tokens_seen": 22039632,
      "step": 33480
    },
    {
      "epoch": 19.743514150943398,
      "grad_norm": 3.838886022567749,
      "learning_rate": 5.032029367290081e-09,
      "loss": 0.3137,
      "num_input_tokens_seen": 22042672,
      "step": 33485
    },
    {
      "epoch": 19.746462264150942,
      "grad_norm": 4.052633285522461,
      "learning_rate": 4.917296437136898e-09,
      "loss": 0.3739,
      "num_input_tokens_seen": 22045968,
      "step": 33490
    },
    {
      "epoch": 19.74941037735849,
      "grad_norm": 4.024411201477051,
      "learning_rate": 4.8038859766569305e-09,
      "loss": 0.3236,
      "num_input_tokens_seen": 22049680,
      "step": 33495
    },
    {
      "epoch": 19.75235849056604,
      "grad_norm": 4.350601673126221,
      "learning_rate": 4.691798015876714e-09,
      "loss": 0.2395,
      "num_input_tokens_seen": 22053104,
      "step": 33500
    },
    {
      "epoch": 19.755306603773583,
      "grad_norm": 3.6485471725463867,
      "learning_rate": 4.581032584470846e-09,
      "loss": 0.2529,
      "num_input_tokens_seen": 22056304,
      "step": 33505
    },
    {
      "epoch": 19.75825471698113,
      "grad_norm": 3.912362813949585,
      "learning_rate": 4.471589711766422e-09,
      "loss": 0.3302,
      "num_input_tokens_seen": 22060144,
      "step": 33510
    },
    {
      "epoch": 19.76120283018868,
      "grad_norm": 3.7051432132720947,
      "learning_rate": 4.363469426737487e-09,
      "loss": 0.2151,
      "num_input_tokens_seen": 22062960,
      "step": 33515
    },
    {
      "epoch": 19.764150943396228,
      "grad_norm": 5.5879034996032715,
      "learning_rate": 4.256671758010588e-09,
      "loss": 0.2887,
      "num_input_tokens_seen": 22065776,
      "step": 33520
    },
    {
      "epoch": 19.767099056603772,
      "grad_norm": 2.478210926055908,
      "learning_rate": 4.151196733859775e-09,
      "loss": 0.2253,
      "num_input_tokens_seen": 22070000,
      "step": 33525
    },
    {
      "epoch": 19.77004716981132,
      "grad_norm": 3.304633855819702,
      "learning_rate": 4.047044382211041e-09,
      "loss": 0.24,
      "num_input_tokens_seen": 22073264,
      "step": 33530
    },
    {
      "epoch": 19.77299528301887,
      "grad_norm": 4.968329906463623,
      "learning_rate": 3.94421473063844e-09,
      "loss": 0.2669,
      "num_input_tokens_seen": 22076368,
      "step": 33535
    },
    {
      "epoch": 19.775943396226417,
      "grad_norm": 5.269434928894043,
      "learning_rate": 3.842707806366863e-09,
      "loss": 0.4777,
      "num_input_tokens_seen": 22079632,
      "step": 33540
    },
    {
      "epoch": 19.77889150943396,
      "grad_norm": 4.913489818572998,
      "learning_rate": 3.742523636270368e-09,
      "loss": 0.3285,
      "num_input_tokens_seen": 22082512,
      "step": 33545
    },
    {
      "epoch": 19.78183962264151,
      "grad_norm": 3.280696392059326,
      "learning_rate": 3.6436622468738468e-09,
      "loss": 0.2789,
      "num_input_tokens_seen": 22086896,
      "step": 33550
    },
    {
      "epoch": 19.784787735849058,
      "grad_norm": 5.069036960601807,
      "learning_rate": 3.546123664350254e-09,
      "loss": 0.3315,
      "num_input_tokens_seen": 22089456,
      "step": 33555
    },
    {
      "epoch": 19.787735849056602,
      "grad_norm": 3.9664299488067627,
      "learning_rate": 3.449907914524486e-09,
      "loss": 0.248,
      "num_input_tokens_seen": 22092496,
      "step": 33560
    },
    {
      "epoch": 19.79068396226415,
      "grad_norm": 2.525017023086548,
      "learning_rate": 3.355015022869501e-09,
      "loss": 0.2253,
      "num_input_tokens_seen": 22095216,
      "step": 33565
    },
    {
      "epoch": 19.7936320754717,
      "grad_norm": 4.582206726074219,
      "learning_rate": 3.261445014508535e-09,
      "loss": 0.3973,
      "num_input_tokens_seen": 22097872,
      "step": 33570
    },
    {
      "epoch": 19.796580188679247,
      "grad_norm": 4.249237537384033,
      "learning_rate": 3.1691979142145503e-09,
      "loss": 0.2419,
      "num_input_tokens_seen": 22100624,
      "step": 33575
    },
    {
      "epoch": 19.79952830188679,
      "grad_norm": 4.398886680603027,
      "learning_rate": 3.078273746410787e-09,
      "loss": 0.3073,
      "num_input_tokens_seen": 22105808,
      "step": 33580
    },
    {
      "epoch": 19.80247641509434,
      "grad_norm": 4.039735317230225,
      "learning_rate": 2.988672535169657e-09,
      "loss": 0.426,
      "num_input_tokens_seen": 22112336,
      "step": 33585
    },
    {
      "epoch": 19.805424528301888,
      "grad_norm": 5.33636474609375,
      "learning_rate": 2.9003943042127393e-09,
      "loss": 0.3055,
      "num_input_tokens_seen": 22114736,
      "step": 33590
    },
    {
      "epoch": 19.808372641509433,
      "grad_norm": 2.4708192348480225,
      "learning_rate": 2.8134390769135598e-09,
      "loss": 0.2263,
      "num_input_tokens_seen": 22117680,
      "step": 33595
    },
    {
      "epoch": 19.81132075471698,
      "grad_norm": 8.0903902053833,
      "learning_rate": 2.7278068762925935e-09,
      "loss": 0.3221,
      "num_input_tokens_seen": 22120432,
      "step": 33600
    },
    {
      "epoch": 19.81426886792453,
      "grad_norm": 3.7265799045562744,
      "learning_rate": 2.6434977250217043e-09,
      "loss": 0.2959,
      "num_input_tokens_seen": 22124080,
      "step": 33605
    },
    {
      "epoch": 19.817216981132077,
      "grad_norm": 3.718052864074707,
      "learning_rate": 2.560511645422481e-09,
      "loss": 0.3205,
      "num_input_tokens_seen": 22127184,
      "step": 33610
    },
    {
      "epoch": 19.82016509433962,
      "grad_norm": 4.53712797164917,
      "learning_rate": 2.4788486594656825e-09,
      "loss": 0.2803,
      "num_input_tokens_seen": 22130704,
      "step": 33615
    },
    {
      "epoch": 19.82311320754717,
      "grad_norm": 3.8216962814331055,
      "learning_rate": 2.398508788771792e-09,
      "loss": 0.3116,
      "num_input_tokens_seen": 22133648,
      "step": 33620
    },
    {
      "epoch": 19.826061320754718,
      "grad_norm": 3.9864277839660645,
      "learning_rate": 2.3194920546110166e-09,
      "loss": 0.3171,
      "num_input_tokens_seen": 22137712,
      "step": 33625
    },
    {
      "epoch": 19.829009433962263,
      "grad_norm": 4.096879482269287,
      "learning_rate": 2.2417984779032896e-09,
      "loss": 0.3506,
      "num_input_tokens_seen": 22142192,
      "step": 33630
    },
    {
      "epoch": 19.83195754716981,
      "grad_norm": 3.413534164428711,
      "learning_rate": 2.1654280792193782e-09,
      "loss": 0.274,
      "num_input_tokens_seen": 22146864,
      "step": 33635
    },
    {
      "epoch": 19.83490566037736,
      "grad_norm": 3.666902542114258,
      "learning_rate": 2.0903808787769987e-09,
      "loss": 0.3359,
      "num_input_tokens_seen": 22150704,
      "step": 33640
    },
    {
      "epoch": 19.837853773584907,
      "grad_norm": 5.231522560119629,
      "learning_rate": 2.0166568964463673e-09,
      "loss": 0.2749,
      "num_input_tokens_seen": 22153968,
      "step": 33645
    },
    {
      "epoch": 19.840801886792452,
      "grad_norm": 3.7553586959838867,
      "learning_rate": 1.9442561517463153e-09,
      "loss": 0.254,
      "num_input_tokens_seen": 22157424,
      "step": 33650
    },
    {
      "epoch": 19.84375,
      "grad_norm": 4.482961177825928,
      "learning_rate": 1.8731786638442886e-09,
      "loss": 0.2873,
      "num_input_tokens_seen": 22162096,
      "step": 33655
    },
    {
      "epoch": 19.846698113207548,
      "grad_norm": 4.586773872375488,
      "learning_rate": 1.8034244515591214e-09,
      "loss": 0.279,
      "num_input_tokens_seen": 22164336,
      "step": 33660
    },
    {
      "epoch": 19.849646226415093,
      "grad_norm": 3.938096284866333,
      "learning_rate": 1.7349935333582646e-09,
      "loss": 0.2844,
      "num_input_tokens_seen": 22167536,
      "step": 33665
    },
    {
      "epoch": 19.85259433962264,
      "grad_norm": 3.231001615524292,
      "learning_rate": 1.6678859273594471e-09,
      "loss": 0.241,
      "num_input_tokens_seen": 22170096,
      "step": 33670
    },
    {
      "epoch": 19.85554245283019,
      "grad_norm": 5.322887897491455,
      "learning_rate": 1.6021016513295683e-09,
      "loss": 0.3039,
      "num_input_tokens_seen": 22172624,
      "step": 33675
    },
    {
      "epoch": 19.858490566037737,
      "grad_norm": 4.332460880279541,
      "learning_rate": 1.5376407226846968e-09,
      "loss": 0.2798,
      "num_input_tokens_seen": 22174832,
      "step": 33680
    },
    {
      "epoch": 19.861438679245282,
      "grad_norm": 3.656841278076172,
      "learning_rate": 1.4745031584917357e-09,
      "loss": 0.3391,
      "num_input_tokens_seen": 22177936,
      "step": 33685
    },
    {
      "epoch": 19.86438679245283,
      "grad_norm": 4.663623809814453,
      "learning_rate": 1.4126889754667583e-09,
      "loss": 0.3127,
      "num_input_tokens_seen": 22181648,
      "step": 33690
    },
    {
      "epoch": 19.86733490566038,
      "grad_norm": 3.0955331325531006,
      "learning_rate": 1.3521981899750069e-09,
      "loss": 0.3793,
      "num_input_tokens_seen": 22185232,
      "step": 33695
    },
    {
      "epoch": 19.870283018867923,
      "grad_norm": 3.6548914909362793,
      "learning_rate": 1.293030818032004e-09,
      "loss": 0.1927,
      "num_input_tokens_seen": 22188656,
      "step": 33700
    },
    {
      "epoch": 19.87323113207547,
      "grad_norm": 3.526951789855957,
      "learning_rate": 1.2351868753018858e-09,
      "loss": 0.1966,
      "num_input_tokens_seen": 22191600,
      "step": 33705
    },
    {
      "epoch": 19.87617924528302,
      "grad_norm": 2.485731840133667,
      "learning_rate": 1.1786663770996242e-09,
      "loss": 0.2975,
      "num_input_tokens_seen": 22195120,
      "step": 33710
    },
    {
      "epoch": 19.879127358490567,
      "grad_norm": 4.43695068359375,
      "learning_rate": 1.1234693383893602e-09,
      "loss": 0.2904,
      "num_input_tokens_seen": 22198928,
      "step": 33715
    },
    {
      "epoch": 19.882075471698112,
      "grad_norm": 4.119534492492676,
      "learning_rate": 1.0695957737844043e-09,
      "loss": 0.2771,
      "num_input_tokens_seen": 22201904,
      "step": 33720
    },
    {
      "epoch": 19.88502358490566,
      "grad_norm": 5.352659225463867,
      "learning_rate": 1.0170456975483467e-09,
      "loss": 0.4274,
      "num_input_tokens_seen": 22205104,
      "step": 33725
    },
    {
      "epoch": 19.88797169811321,
      "grad_norm": 5.819039821624756,
      "learning_rate": 9.658191235933922e-10,
      "loss": 0.3181,
      "num_input_tokens_seen": 22208400,
      "step": 33730
    },
    {
      "epoch": 19.890919811320753,
      "grad_norm": 3.774292469024658,
      "learning_rate": 9.159160654825805e-10,
      "loss": 0.25,
      "num_input_tokens_seen": 22212816,
      "step": 33735
    },
    {
      "epoch": 19.8938679245283,
      "grad_norm": 3.9455204010009766,
      "learning_rate": 8.673365364281205e-10,
      "loss": 0.3072,
      "num_input_tokens_seen": 22215824,
      "step": 33740
    },
    {
      "epoch": 19.89681603773585,
      "grad_norm": 4.732486724853516,
      "learning_rate": 8.200805492913911e-10,
      "loss": 0.1986,
      "num_input_tokens_seen": 22218352,
      "step": 33745
    },
    {
      "epoch": 19.899764150943398,
      "grad_norm": 7.313977241516113,
      "learning_rate": 7.741481165834952e-10,
      "loss": 0.2853,
      "num_input_tokens_seen": 22222096,
      "step": 33750
    },
    {
      "epoch": 19.902712264150942,
      "grad_norm": 2.3822150230407715,
      "learning_rate": 7.29539250465261e-10,
      "loss": 0.2518,
      "num_input_tokens_seen": 22225712,
      "step": 33755
    },
    {
      "epoch": 19.90566037735849,
      "grad_norm": 4.396801471710205,
      "learning_rate": 6.862539627472409e-10,
      "loss": 0.2716,
      "num_input_tokens_seen": 22228752,
      "step": 33760
    },
    {
      "epoch": 19.90860849056604,
      "grad_norm": 4.8268723487854,
      "learning_rate": 6.442922648897121e-10,
      "loss": 0.4382,
      "num_input_tokens_seen": 22231088,
      "step": 33765
    },
    {
      "epoch": 19.911556603773583,
      "grad_norm": 4.775531768798828,
      "learning_rate": 6.036541680015662e-10,
      "loss": 0.3201,
      "num_input_tokens_seen": 22233552,
      "step": 33770
    },
    {
      "epoch": 19.91450471698113,
      "grad_norm": 3.373725652694702,
      "learning_rate": 5.643396828419745e-10,
      "loss": 0.2174,
      "num_input_tokens_seen": 22236752,
      "step": 33775
    },
    {
      "epoch": 19.91745283018868,
      "grad_norm": 3.616190195083618,
      "learning_rate": 5.26348819819833e-10,
      "loss": 0.2702,
      "num_input_tokens_seen": 22240272,
      "step": 33780
    },
    {
      "epoch": 19.920400943396228,
      "grad_norm": 6.677011966705322,
      "learning_rate": 4.896815889937622e-10,
      "loss": 0.2708,
      "num_input_tokens_seen": 22243504,
      "step": 33785
    },
    {
      "epoch": 19.923349056603772,
      "grad_norm": 6.920074939727783,
      "learning_rate": 4.543380000704423e-10,
      "loss": 0.3351,
      "num_input_tokens_seen": 22247056,
      "step": 33790
    },
    {
      "epoch": 19.92629716981132,
      "grad_norm": 4.731483459472656,
      "learning_rate": 4.203180624084979e-10,
      "loss": 0.3012,
      "num_input_tokens_seen": 22249616,
      "step": 33795
    },
    {
      "epoch": 19.92924528301887,
      "grad_norm": 3.9257538318634033,
      "learning_rate": 3.876217850146136e-10,
      "loss": 0.2986,
      "num_input_tokens_seen": 22252496,
      "step": 33800
    },
    {
      "epoch": 19.932193396226417,
      "grad_norm": 4.87930154800415,
      "learning_rate": 3.56249176544643e-10,
      "loss": 0.3177,
      "num_input_tokens_seen": 22255792,
      "step": 33805
    },
    {
      "epoch": 19.93514150943396,
      "grad_norm": 9.57633113861084,
      "learning_rate": 3.262002453047197e-10,
      "loss": 0.26,
      "num_input_tokens_seen": 22258576,
      "step": 33810
    },
    {
      "epoch": 19.93808962264151,
      "grad_norm": 4.37871789932251,
      "learning_rate": 2.974749992512571e-10,
      "loss": 0.3009,
      "num_input_tokens_seen": 22261328,
      "step": 33815
    },
    {
      "epoch": 19.941037735849058,
      "grad_norm": 6.5309858322143555,
      "learning_rate": 2.700734459881726e-10,
      "loss": 0.3519,
      "num_input_tokens_seen": 22264112,
      "step": 33820
    },
    {
      "epoch": 19.943985849056602,
      "grad_norm": 2.673145294189453,
      "learning_rate": 2.439955927713289e-10,
      "loss": 0.5742,
      "num_input_tokens_seen": 22268176,
      "step": 33825
    },
    {
      "epoch": 19.94693396226415,
      "grad_norm": 4.27263879776001,
      "learning_rate": 2.1924144650409263e-10,
      "loss": 0.2232,
      "num_input_tokens_seen": 22271120,
      "step": 33830
    },
    {
      "epoch": 19.9498820754717,
      "grad_norm": 3.1744532585144043,
      "learning_rate": 1.9581101374066546e-10,
      "loss": 0.2776,
      "num_input_tokens_seen": 22273424,
      "step": 33835
    },
    {
      "epoch": 19.952830188679247,
      "grad_norm": 5.813560962677002,
      "learning_rate": 1.7370430068441858e-10,
      "loss": 0.2911,
      "num_input_tokens_seen": 22276816,
      "step": 33840
    },
    {
      "epoch": 19.95577830188679,
      "grad_norm": 3.888279914855957,
      "learning_rate": 1.529213131878926e-10,
      "loss": 0.4486,
      "num_input_tokens_seen": 22280016,
      "step": 33845
    },
    {
      "epoch": 19.95872641509434,
      "grad_norm": 4.288168430328369,
      "learning_rate": 1.3346205675335288e-10,
      "loss": 0.4143,
      "num_input_tokens_seen": 22283408,
      "step": 33850
    },
    {
      "epoch": 19.961674528301888,
      "grad_norm": 4.730678081512451,
      "learning_rate": 1.1532653653334447e-10,
      "loss": 0.2756,
      "num_input_tokens_seen": 22286032,
      "step": 33855
    },
    {
      "epoch": 19.964622641509433,
      "grad_norm": 2.9191572666168213,
      "learning_rate": 9.85147573284717e-11,
      "loss": 0.3083,
      "num_input_tokens_seen": 22289744,
      "step": 33860
    },
    {
      "epoch": 19.96757075471698,
      "grad_norm": 2.6264615058898926,
      "learning_rate": 8.302672359072894e-11,
      "loss": 0.2281,
      "num_input_tokens_seen": 22292656,
      "step": 33865
    },
    {
      "epoch": 19.97051886792453,
      "grad_norm": 6.599427700042725,
      "learning_rate": 6.886243941961468e-11,
      "loss": 0.2721,
      "num_input_tokens_seen": 22295472,
      "step": 33870
    },
    {
      "epoch": 19.973466981132077,
      "grad_norm": 5.872149467468262,
      "learning_rate": 5.602190856601741e-11,
      "loss": 0.2705,
      "num_input_tokens_seen": 22299216,
      "step": 33875
    },
    {
      "epoch": 19.97641509433962,
      "grad_norm": 3.0358333587646484,
      "learning_rate": 4.450513442888493e-11,
      "loss": 0.3063,
      "num_input_tokens_seen": 22302512,
      "step": 33880
    },
    {
      "epoch": 19.97936320754717,
      "grad_norm": 3.5732314586639404,
      "learning_rate": 3.4312120057999886e-11,
      "loss": 0.3465,
      "num_input_tokens_seen": 22305008,
      "step": 33885
    },
    {
      "epoch": 19.982311320754718,
      "grad_norm": 5.368415355682373,
      "learning_rate": 2.5442868151204224e-11,
      "loss": 0.3758,
      "num_input_tokens_seen": 22308336,
      "step": 33890
    },
    {
      "epoch": 19.985259433962263,
      "grad_norm": 2.3642287254333496,
      "learning_rate": 1.7897381057729867e-11,
      "loss": 0.2291,
      "num_input_tokens_seen": 22311120,
      "step": 33895
    },
    {
      "epoch": 19.98820754716981,
      "grad_norm": 6.640480041503906,
      "learning_rate": 1.1675660773757813e-11,
      "loss": 0.238,
      "num_input_tokens_seen": 22313840,
      "step": 33900
    },
    {
      "epoch": 19.99115566037736,
      "grad_norm": 5.5302934646606445,
      "learning_rate": 6.777708947969253e-12,
      "loss": 0.2988,
      "num_input_tokens_seen": 22317104,
      "step": 33905
    },
    {
      "epoch": 19.994103773584907,
      "grad_norm": 3.577695608139038,
      "learning_rate": 3.2035268765495674e-12,
      "loss": 0.3082,
      "num_input_tokens_seen": 22321392,
      "step": 33910
    },
    {
      "epoch": 19.997051886792452,
      "grad_norm": 6.6892852783203125,
      "learning_rate": 9.531155059638863e-13,
      "loss": 0.3209,
      "num_input_tokens_seen": 22324176,
      "step": 33915
    },
    {
      "epoch": 20.0,
      "grad_norm": 3.4563357830047607,
      "learning_rate": 2.6475431291750342e-14,
      "loss": 0.3651,
      "num_input_tokens_seen": 22326528,
      "step": 33920
    },
    {
      "epoch": 20.0,
      "eval_loss": 0.6114597916603088,
      "eval_runtime": 18.6561,
      "eval_samples_per_second": 90.909,
      "eval_steps_per_second": 22.727,
      "num_input_tokens_seen": 22326528,
      "step": 33920
    },
    {
      "epoch": 20.0,
      "num_input_tokens_seen": 22326528,
      "step": 33920,
      "total_flos": 1.0053534649637929e+18,
      "train_loss": 0.4641511028790671,
      "train_runtime": 3972.823,
      "train_samples_per_second": 34.137,
      "train_steps_per_second": 8.538
    }
  ],
  "logging_steps": 5,
  "max_steps": 33920,
  "num_input_tokens_seen": 22326528,
  "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": 1.0053534649637929e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}