Adapter/trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9989281886387995,
  "eval_steps": 59,
  "global_step": 467,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "eval_loss": 2.934493064880371,
      "eval_ppl": 18.81196,
      "eval_runtime": 181.0816,
      "eval_samples_per_second": 1.149,
      "eval_steps_per_second": 1.149,
      "memory/device_reserved (GiB)": 18.64,
      "memory/max_active (GiB)": 18.28,
      "memory/max_allocated (GiB)": 18.28,
      "step": 0
    },
    {
      "epoch": 0.004287245444801715,
      "grad_norm": 0.5241990685462952,
      "learning_rate": 0.0025,
      "loss": 3.2761871814727783,
      "memory/device_reserved (GiB)": 18.85,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 26.47464,
      "step": 1,
      "tokens/total": 2816,
      "tokens/train_per_sec_per_gpu": 1.6,
      "tokens/trainable": 1916
    },
    {
      "epoch": 0.00857449088960343,
      "grad_norm": 0.2679648697376251,
      "learning_rate": 0.0025,
      "loss": 2.743480920791626,
      "memory/device_reserved (GiB)": 18.94,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 15.54099,
      "step": 2,
      "tokens/total": 4416,
      "tokens/train_per_sec_per_gpu": 26.79,
      "tokens/trainable": 2568
    },
    {
      "epoch": 0.012861736334405145,
      "grad_norm": 2.105276346206665,
      "learning_rate": 0.0025,
      "loss": 4.918459415435791,
      "memory/device_reserved (GiB)": 18.94,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 136.79171,
      "step": 3,
      "tokens/total": 6272,
      "tokens/train_per_sec_per_gpu": 3.76,
      "tokens/trainable": 3571
    },
    {
      "epoch": 0.01714898177920686,
      "grad_norm": 1.7258063554763794,
      "learning_rate": 0.0025,
      "loss": 4.470706462860107,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 19.42,
      "memory/max_allocated (GiB)": 19.42,
      "ppl": 87.41846,
      "step": 4,
      "tokens/total": 9664,
      "tokens/train_per_sec_per_gpu": 4.21,
      "tokens/trainable": 6033
    },
    {
      "epoch": 0.021436227224008574,
      "grad_norm": 0.9528670907020569,
      "learning_rate": 0.0025,
      "loss": 3.6473782062530518,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 38.37393,
      "step": 5,
      "tokens/total": 11392,
      "tokens/train_per_sec_per_gpu": 22.21,
      "tokens/trainable": 6851
    },
    {
      "epoch": 0.02572347266881029,
      "grad_norm": 1.7168548107147217,
      "learning_rate": 0.0025,
      "loss": 3.641300678253174,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.63,
      "memory/max_allocated (GiB)": 18.63,
      "ppl": 38.14141,
      "step": 6,
      "tokens/total": 14464,
      "tokens/train_per_sec_per_gpu": 25.23,
      "tokens/trainable": 8951
    },
    {
      "epoch": 0.030010718113612004,
      "grad_norm": 9.308958053588867,
      "learning_rate": 0.0025,
      "loss": 11.468059539794922,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 95612.56656,
      "step": 7,
      "tokens/total": 16512,
      "tokens/train_per_sec_per_gpu": 43.04,
      "tokens/trainable": 9981
    },
    {
      "epoch": 0.03429796355841372,
      "grad_norm": 4.779436111450195,
      "learning_rate": 0.0025,
      "loss": 12.928940773010254,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.73,
      "memory/max_allocated (GiB)": 18.73,
      "ppl": 412066.80902,
      "step": 8,
      "tokens/total": 20160,
      "tokens/train_per_sec_per_gpu": 206.05,
      "tokens/trainable": 12656
    },
    {
      "epoch": 0.03858520900321544,
      "grad_norm": 11.870071411132812,
      "learning_rate": 0.0025,
      "loss": 26.585786819458008,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 351609428878.3472,
      "step": 9,
      "tokens/total": 22272,
      "tokens/train_per_sec_per_gpu": 21.53,
      "tokens/trainable": 13844
    },
    {
      "epoch": 0.04287245444801715,
      "grad_norm": 3.3879966735839844,
      "learning_rate": 0.0025,
      "loss": 21.875102996826172,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 3164000347.95887,
      "step": 10,
      "tokens/total": 24256,
      "tokens/train_per_sec_per_gpu": 53.92,
      "tokens/trainable": 15016
    },
    {
      "epoch": 0.04715969989281887,
      "grad_norm": 0.25865933299064636,
      "learning_rate": 0.0025,
      "loss": 14.399214744567871,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.53,
      "memory/max_allocated (GiB)": 18.53,
      "ppl": 1792666.51864,
      "step": 11,
      "tokens/total": 27008,
      "tokens/train_per_sec_per_gpu": 67.68,
      "tokens/trainable": 16945
    },
    {
      "epoch": 0.05144694533762058,
      "grad_norm": 3.67203950881958,
      "learning_rate": 0.0025,
      "loss": 12.23837947845459,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.59,
      "memory/max_allocated (GiB)": 18.59,
      "ppl": 206566.87286,
      "step": 12,
      "tokens/total": 29568,
      "tokens/train_per_sec_per_gpu": 20.56,
      "tokens/trainable": 18583
    },
    {
      "epoch": 0.055734190782422297,
      "grad_norm": 2.215568780899048,
      "learning_rate": 0.0025,
      "loss": 15.4072847366333,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 4912456.77743,
      "step": 13,
      "tokens/total": 31936,
      "tokens/train_per_sec_per_gpu": 30.37,
      "tokens/trainable": 19999
    },
    {
      "epoch": 0.06002143622722401,
      "grad_norm": 1.5136427879333496,
      "learning_rate": 0.0025,
      "loss": 11.96999454498291,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.37,
      "memory/max_allocated (GiB)": 18.37,
      "ppl": 157943.79881,
      "step": 14,
      "tokens/total": 33856,
      "tokens/train_per_sec_per_gpu": 4.74,
      "tokens/trainable": 21034
    },
    {
      "epoch": 0.06430868167202572,
      "grad_norm": 12.923323631286621,
      "learning_rate": 0.0025,
      "loss": 11.865056991577148,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.87,
      "memory/max_allocated (GiB)": 18.87,
      "ppl": 142209.55548,
      "step": 15,
      "tokens/total": 37632,
      "tokens/train_per_sec_per_gpu": 4.06,
      "tokens/trainable": 23846
    },
    {
      "epoch": 0.06859592711682744,
      "grad_norm": 1.347749948501587,
      "learning_rate": 0.0025,
      "loss": 9.3814115524292,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 11865.75207,
      "step": 16,
      "tokens/total": 39744,
      "tokens/train_per_sec_per_gpu": 78.77,
      "tokens/trainable": 24979
    },
    {
      "epoch": 0.07288317256162916,
      "grad_norm": 0.6687317490577698,
      "learning_rate": 0.0025,
      "loss": 8.9339599609375,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 7585.24325,
      "step": 17,
      "tokens/total": 42368,
      "tokens/train_per_sec_per_gpu": 30.07,
      "tokens/trainable": 26674
    },
    {
      "epoch": 0.07717041800643087,
      "grad_norm": 1.8986364603042603,
      "learning_rate": 0.0025,
      "loss": 10.20353889465332,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 26998.56228,
      "step": 18,
      "tokens/total": 44480,
      "tokens/train_per_sec_per_gpu": 10.04,
      "tokens/trainable": 27850
    },
    {
      "epoch": 0.08145766345123258,
      "grad_norm": 0.3103311359882355,
      "learning_rate": 0.0025,
      "loss": 8.43118953704834,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 4587.95439,
      "step": 19,
      "tokens/total": 45952,
      "tokens/train_per_sec_per_gpu": 4.36,
      "tokens/trainable": 28429
    },
    {
      "epoch": 0.0857449088960343,
      "grad_norm": 0.33680954575538635,
      "learning_rate": 0.0025,
      "loss": 8.632229804992676,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 5609.57257,
      "step": 20,
      "tokens/total": 48512,
      "tokens/train_per_sec_per_gpu": 157.2,
      "tokens/trainable": 29973
    },
    {
      "epoch": 0.09003215434083602,
      "grad_norm": 0.3231711983680725,
      "learning_rate": 0.0025,
      "loss": 8.600848197937012,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.86,
      "memory/max_allocated (GiB)": 18.86,
      "ppl": 5436.26867,
      "step": 21,
      "tokens/total": 51264,
      "tokens/train_per_sec_per_gpu": 2.82,
      "tokens/trainable": 31831
    },
    {
      "epoch": 0.09431939978563773,
      "grad_norm": 0.28503984212875366,
      "learning_rate": 0.0025,
      "loss": 8.879640579223633,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.38,
      "memory/max_allocated (GiB)": 18.38,
      "ppl": 7184.20812,
      "step": 22,
      "tokens/total": 52736,
      "tokens/train_per_sec_per_gpu": 52.3,
      "tokens/trainable": 32257
    },
    {
      "epoch": 0.09860664523043944,
      "grad_norm": 0.17652413249015808,
      "learning_rate": 0.0025,
      "loss": 8.288322448730469,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.66,
      "memory/max_allocated (GiB)": 18.66,
      "ppl": 3977.15671,
      "step": 23,
      "tokens/total": 55936,
      "tokens/train_per_sec_per_gpu": 2.04,
      "tokens/trainable": 34507
    },
    {
      "epoch": 0.10289389067524116,
      "grad_norm": 0.17270459234714508,
      "learning_rate": 0.0025,
      "loss": 7.758934020996094,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 2342.40632,
      "step": 24,
      "tokens/total": 57856,
      "tokens/train_per_sec_per_gpu": 52.59,
      "tokens/trainable": 35539
    },
    {
      "epoch": 0.10718113612004287,
      "grad_norm": 0.14134642481803894,
      "learning_rate": 0.0025,
      "loss": 7.277863502502441,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 1447.8913,
      "step": 25,
      "tokens/total": 60288,
      "tokens/train_per_sec_per_gpu": 175.7,
      "tokens/trainable": 37092
    },
    {
      "epoch": 0.11146838156484459,
      "grad_norm": 0.3017260730266571,
      "learning_rate": 0.0025,
      "loss": 7.528397083282471,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 1860.12149,
      "step": 26,
      "tokens/total": 62784,
      "tokens/train_per_sec_per_gpu": 67.24,
      "tokens/trainable": 38599
    },
    {
      "epoch": 0.1157556270096463,
      "grad_norm": 0.1806621253490448,
      "learning_rate": 0.0025,
      "loss": 7.274528503417969,
      "memory/device_reserved (GiB)": 19.79,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 1443.07063,
      "step": 27,
      "tokens/total": 65024,
      "tokens/train_per_sec_per_gpu": 22.31,
      "tokens/trainable": 39852
    },
    {
      "epoch": 0.12004287245444802,
      "grad_norm": 0.4423042833805084,
      "learning_rate": 0.0025,
      "loss": 6.991304874420166,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 19.54,
      "memory/max_allocated (GiB)": 19.54,
      "ppl": 1087.13913,
      "step": 28,
      "tokens/total": 70784,
      "tokens/train_per_sec_per_gpu": 70.63,
      "tokens/trainable": 44731
    },
    {
      "epoch": 0.12433011789924973,
      "grad_norm": 0.22539205849170685,
      "learning_rate": 0.0025,
      "loss": 6.762757778167725,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 865.02445,
      "step": 29,
      "tokens/total": 73024,
      "tokens/train_per_sec_per_gpu": 73.89,
      "tokens/trainable": 46095
    },
    {
      "epoch": 0.12861736334405144,
      "grad_norm": 0.585552990436554,
      "learning_rate": 0.0025,
      "loss": 6.865115165710449,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.64,
      "memory/max_allocated (GiB)": 18.64,
      "ppl": 958.25619,
      "step": 30,
      "tokens/total": 75456,
      "tokens/train_per_sec_per_gpu": 67.76,
      "tokens/trainable": 47570
    },
    {
      "epoch": 0.13290460878885316,
      "grad_norm": 0.9950224161148071,
      "learning_rate": 0.0025,
      "loss": 7.070884704589844,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 1177.18904,
      "step": 31,
      "tokens/total": 77824,
      "tokens/train_per_sec_per_gpu": 29.22,
      "tokens/trainable": 48877
    },
    {
      "epoch": 0.13719185423365488,
      "grad_norm": 1.1349307298660278,
      "learning_rate": 0.0025,
      "loss": 6.7641096115112305,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 866.19461,
      "step": 32,
      "tokens/total": 79680,
      "tokens/train_per_sec_per_gpu": 12.6,
      "tokens/trainable": 49878
    },
    {
      "epoch": 0.1414790996784566,
      "grad_norm": 0.19686993956565857,
      "learning_rate": 0.0025,
      "loss": 6.472617149353027,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 647.17527,
      "step": 33,
      "tokens/total": 81792,
      "tokens/train_per_sec_per_gpu": 110.67,
      "tokens/trainable": 51049
    },
    {
      "epoch": 0.1457663451232583,
      "grad_norm": 0.4460529685020447,
      "learning_rate": 0.0025,
      "loss": 6.628453254699707,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 756.31144,
      "step": 34,
      "tokens/total": 84032,
      "tokens/train_per_sec_per_gpu": 15.12,
      "tokens/trainable": 52250
    },
    {
      "epoch": 0.15005359056806003,
      "grad_norm": 0.09157463908195496,
      "learning_rate": 0.0025,
      "loss": 6.950263977050781,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 1043.42513,
      "step": 35,
      "tokens/total": 86656,
      "tokens/train_per_sec_per_gpu": 25.29,
      "tokens/trainable": 53911
    },
    {
      "epoch": 0.15434083601286175,
      "grad_norm": 1.2685779333114624,
      "learning_rate": 0.0025,
      "loss": 6.143320083618164,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 465.59683,
      "step": 36,
      "tokens/total": 88640,
      "tokens/train_per_sec_per_gpu": 15.77,
      "tokens/trainable": 54904
    },
    {
      "epoch": 0.15862808145766344,
      "grad_norm": 0.31893390417099,
      "learning_rate": 0.0025,
      "loss": 6.968283653259277,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.38,
      "memory/max_allocated (GiB)": 18.38,
      "ppl": 1062.39774,
      "step": 37,
      "tokens/total": 90496,
      "tokens/train_per_sec_per_gpu": 58.59,
      "tokens/trainable": 55743
    },
    {
      "epoch": 0.16291532690246516,
      "grad_norm": 1.0469295978546143,
      "learning_rate": 0.0025,
      "loss": 7.234709739685059,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 1386.73832,
      "step": 38,
      "tokens/total": 92352,
      "tokens/train_per_sec_per_gpu": 0.83,
      "tokens/trainable": 56555
    },
    {
      "epoch": 0.16720257234726688,
      "grad_norm": 0.2749118506908417,
      "learning_rate": 0.0025,
      "loss": 6.587377071380615,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 725.87445,
      "step": 39,
      "tokens/total": 94400,
      "tokens/train_per_sec_per_gpu": 33.93,
      "tokens/trainable": 57680
    },
    {
      "epoch": 0.1714898177920686,
      "grad_norm": 0.18221265077590942,
      "learning_rate": 0.0025,
      "loss": 6.830031871795654,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 925.2203,
      "step": 40,
      "tokens/total": 96512,
      "tokens/train_per_sec_per_gpu": 45.38,
      "tokens/trainable": 58955
    },
    {
      "epoch": 0.1757770632368703,
      "grad_norm": 0.24708712100982666,
      "learning_rate": 0.0025,
      "loss": 6.151037693023682,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 469.20402,
      "step": 41,
      "tokens/total": 98752,
      "tokens/train_per_sec_per_gpu": 74.94,
      "tokens/trainable": 60277
    },
    {
      "epoch": 0.18006430868167203,
      "grad_norm": 0.17541086673736572,
      "learning_rate": 0.0025,
      "loss": 6.18589973449707,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.63,
      "memory/max_allocated (GiB)": 18.63,
      "ppl": 485.8499,
      "step": 42,
      "tokens/total": 101440,
      "tokens/train_per_sec_per_gpu": 19.12,
      "tokens/trainable": 61928
    },
    {
      "epoch": 0.18435155412647375,
      "grad_norm": 0.5008364319801331,
      "learning_rate": 0.0025,
      "loss": 6.555770397186279,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 703.29075,
      "step": 43,
      "tokens/total": 103104,
      "tokens/train_per_sec_per_gpu": 8.28,
      "tokens/trainable": 62757
    },
    {
      "epoch": 0.18863879957127547,
      "grad_norm": 0.8753749132156372,
      "learning_rate": 0.0025,
      "loss": 6.545720100402832,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.54,
      "memory/max_allocated (GiB)": 18.54,
      "ppl": 696.25787,
      "step": 44,
      "tokens/total": 105536,
      "tokens/train_per_sec_per_gpu": 4.81,
      "tokens/trainable": 64177
    },
    {
      "epoch": 0.19292604501607716,
      "grad_norm": 0.22259370982646942,
      "learning_rate": 0.0025,
      "loss": 6.528131008148193,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 684.1184,
      "step": 45,
      "tokens/total": 107584,
      "tokens/train_per_sec_per_gpu": 14.82,
      "tokens/trainable": 65253
    },
    {
      "epoch": 0.19721329046087888,
      "grad_norm": 0.10767526179552078,
      "learning_rate": 0.0025,
      "loss": 6.292204856872559,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 540.3434,
      "step": 46,
      "tokens/total": 109632,
      "tokens/train_per_sec_per_gpu": 19.33,
      "tokens/trainable": 66464
    },
    {
      "epoch": 0.2015005359056806,
      "grad_norm": 0.12123644351959229,
      "learning_rate": 0.0025,
      "loss": 6.8791117668151855,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 971.76282,
      "step": 47,
      "tokens/total": 112192,
      "tokens/train_per_sec_per_gpu": 223.37,
      "tokens/trainable": 68117
    },
    {
      "epoch": 0.2057877813504823,
      "grad_norm": 0.1776631772518158,
      "learning_rate": 0.0025,
      "loss": 6.613104343414307,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.38,
      "memory/max_allocated (GiB)": 18.38,
      "ppl": 744.79152,
      "step": 48,
      "tokens/total": 113856,
      "tokens/train_per_sec_per_gpu": 54.59,
      "tokens/trainable": 68805
    },
    {
      "epoch": 0.21007502679528403,
      "grad_norm": 0.16078130900859833,
      "learning_rate": 0.0025,
      "loss": 6.84950065612793,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 943.4097,
      "step": 49,
      "tokens/total": 116032,
      "tokens/train_per_sec_per_gpu": 75.86,
      "tokens/trainable": 70058
    },
    {
      "epoch": 0.21436227224008575,
      "grad_norm": 0.10362584888935089,
      "learning_rate": 0.0025,
      "loss": 7.038058280944824,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.54,
      "memory/max_allocated (GiB)": 18.54,
      "ppl": 1139.1735,
      "step": 50,
      "tokens/total": 118336,
      "tokens/train_per_sec_per_gpu": 60.13,
      "tokens/trainable": 71390
    },
    {
      "epoch": 0.21864951768488747,
      "grad_norm": 0.14731702208518982,
      "learning_rate": 0.0025,
      "loss": 5.994701385498047,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 401.29683,
      "step": 51,
      "tokens/total": 120384,
      "tokens/train_per_sec_per_gpu": 3.02,
      "tokens/trainable": 72578
    },
    {
      "epoch": 0.22293676312968919,
      "grad_norm": 0.15322738885879517,
      "learning_rate": 0.0025,
      "loss": 7.148180961608887,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 1271.79041,
      "step": 52,
      "tokens/total": 122432,
      "tokens/train_per_sec_per_gpu": 49.03,
      "tokens/trainable": 73654
    },
    {
      "epoch": 0.22722400857449088,
      "grad_norm": 0.2865282893180847,
      "learning_rate": 0.0025,
      "loss": 6.537259578704834,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 690.39202,
      "step": 53,
      "tokens/total": 125056,
      "tokens/train_per_sec_per_gpu": 37.17,
      "tokens/trainable": 75448
    },
    {
      "epoch": 0.2315112540192926,
      "grad_norm": 0.19199238717556,
      "learning_rate": 0.0025,
      "loss": 6.6607537269592285,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 781.13948,
      "step": 54,
      "tokens/total": 127360,
      "tokens/train_per_sec_per_gpu": 153.39,
      "tokens/trainable": 76842
    },
    {
      "epoch": 0.2357984994640943,
      "grad_norm": 0.07870891690254211,
      "learning_rate": 0.0025,
      "loss": 6.618167400360107,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.88,
      "memory/max_allocated (GiB)": 18.88,
      "ppl": 748.57201,
      "step": 55,
      "tokens/total": 131008,
      "tokens/train_per_sec_per_gpu": 12.42,
      "tokens/trainable": 79539
    },
    {
      "epoch": 0.24008574490889603,
      "grad_norm": 0.10853379964828491,
      "learning_rate": 0.0025,
      "loss": 6.207864761352539,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 496.63967,
      "step": 56,
      "tokens/total": 132800,
      "tokens/train_per_sec_per_gpu": 9.35,
      "tokens/trainable": 80444
    },
    {
      "epoch": 0.24437299035369775,
      "grad_norm": 0.06515457481145859,
      "learning_rate": 0.0025,
      "loss": 6.508807182312012,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 671.02553,
      "step": 57,
      "tokens/total": 135040,
      "tokens/train_per_sec_per_gpu": 112.03,
      "tokens/trainable": 81812
    },
    {
      "epoch": 0.24866023579849947,
      "grad_norm": 0.12576992809772491,
      "learning_rate": 0.0025,
      "loss": 6.428314208984375,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 619.12935,
      "step": 58,
      "tokens/total": 137472,
      "tokens/train_per_sec_per_gpu": 117.23,
      "tokens/trainable": 83345
    },
    {
      "epoch": 0.2529474812433012,
      "grad_norm": 0.11791636049747467,
      "learning_rate": 0.0025,
      "loss": 6.194557189941406,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.38,
      "memory/max_allocated (GiB)": 18.38,
      "ppl": 490.07439,
      "step": 59,
      "tokens/total": 139136,
      "tokens/train_per_sec_per_gpu": 54.16,
      "tokens/trainable": 84003
    },
    {
      "epoch": 0.2529474812433012,
      "eval_loss": 6.30873441696167,
      "eval_ppl": 549.34926,
      "eval_runtime": 17.198,
      "eval_samples_per_second": 12.094,
      "eval_steps_per_second": 12.094,
      "memory/device_reserved (GiB)": 19.96,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "step": 59
    },
    {
      "epoch": 0.2572347266881029,
      "grad_norm": 0.20866750180721283,
      "learning_rate": 0.0025,
      "loss": 6.447090148925781,
      "memory/device_reserved (GiB)": 18.69,
      "memory/max_active (GiB)": 18.54,
      "memory/max_allocated (GiB)": 18.54,
      "ppl": 630.8639,
      "step": 60,
      "tokens/total": 142208,
      "tokens/train_per_sec_per_gpu": 37.47,
      "tokens/trainable": 86018
    },
    {
      "epoch": 0.2615219721329046,
      "grad_norm": 0.08183833956718445,
      "learning_rate": 0.0025,
      "loss": 6.008518218994141,
      "memory/device_reserved (GiB)": 18.73,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 406.87997,
      "step": 61,
      "tokens/total": 144000,
      "tokens/train_per_sec_per_gpu": 22.8,
      "tokens/trainable": 86897
    },
    {
      "epoch": 0.2658092175777063,
      "grad_norm": 0.21841929852962494,
      "learning_rate": 0.0025,
      "loss": 6.573157787322998,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.82,
      "memory/max_allocated (GiB)": 18.82,
      "ppl": 715.62607,
      "step": 62,
      "tokens/total": 147520,
      "tokens/train_per_sec_per_gpu": 19.02,
      "tokens/trainable": 89504
    },
    {
      "epoch": 0.27009646302250806,
      "grad_norm": 0.09511567652225494,
      "learning_rate": 0.0025,
      "loss": 6.555922985076904,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 703.39808,
      "step": 63,
      "tokens/total": 150016,
      "tokens/train_per_sec_per_gpu": 183.78,
      "tokens/trainable": 91081
    },
    {
      "epoch": 0.27438370846730975,
      "grad_norm": 0.07103318721055984,
      "learning_rate": 0.0025,
      "loss": 6.265772342681885,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.54,
      "memory/max_allocated (GiB)": 18.54,
      "ppl": 526.24787,
      "step": 64,
      "tokens/total": 152384,
      "tokens/train_per_sec_per_gpu": 50.64,
      "tokens/trainable": 92465
    },
    {
      "epoch": 0.27867095391211144,
      "grad_norm": 0.14229358732700348,
      "learning_rate": 0.0025,
      "loss": 6.940990924835205,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.38,
      "memory/max_allocated (GiB)": 18.38,
      "ppl": 1033.79412,
      "step": 65,
      "tokens/total": 153792,
      "tokens/train_per_sec_per_gpu": 2.2,
      "tokens/trainable": 93043
    },
    {
      "epoch": 0.2829581993569132,
      "grad_norm": 0.148279070854187,
      "learning_rate": 0.0025,
      "loss": 6.338968276977539,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 566.21184,
      "step": 66,
      "tokens/total": 156352,
      "tokens/train_per_sec_per_gpu": 56.15,
      "tokens/trainable": 94644
    },
    {
      "epoch": 0.2872454448017149,
      "grad_norm": 0.18289832770824432,
      "learning_rate": 0.0025,
      "loss": 6.198648929595947,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 492.08375,
      "step": 67,
      "tokens/total": 158144,
      "tokens/train_per_sec_per_gpu": 72.42,
      "tokens/trainable": 95506
    },
    {
      "epoch": 0.2915326902465166,
      "grad_norm": 0.11502089351415634,
      "learning_rate": 0.0025,
      "loss": 6.177389621734619,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 481.73281,
      "step": 68,
      "tokens/total": 159808,
      "tokens/train_per_sec_per_gpu": 13.1,
      "tokens/trainable": 96311
    },
    {
      "epoch": 0.2958199356913183,
      "grad_norm": 0.766052782535553,
      "learning_rate": 0.0025,
      "loss": 6.210302829742432,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 497.85199,
      "step": 69,
      "tokens/total": 162240,
      "tokens/train_per_sec_per_gpu": 15.54,
      "tokens/trainable": 97821
    },
    {
      "epoch": 0.30010718113612006,
      "grad_norm": 0.16312995553016663,
      "learning_rate": 0.0025,
      "loss": 6.322083950042725,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 556.73199,
      "step": 70,
      "tokens/total": 163968,
      "tokens/train_per_sec_per_gpu": 35.21,
      "tokens/trainable": 98557
    },
    {
      "epoch": 0.30439442658092175,
      "grad_norm": 0.09915313869714737,
      "learning_rate": 0.0025,
      "loss": 5.732677459716797,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 308.79495,
      "step": 71,
      "tokens/total": 165952,
      "tokens/train_per_sec_per_gpu": 126.46,
      "tokens/trainable": 99716
    },
    {
      "epoch": 0.3086816720257235,
      "grad_norm": 0.3138703405857086,
      "learning_rate": 0.0025,
      "loss": 5.753807067871094,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 315.38909,
      "step": 72,
      "tokens/total": 167872,
      "tokens/train_per_sec_per_gpu": 27.76,
      "tokens/trainable": 100641
    },
    {
      "epoch": 0.3129689174705252,
      "grad_norm": 0.2080921083688736,
      "learning_rate": 0.0025,
      "loss": 6.6848883628845215,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.6,
      "memory/max_allocated (GiB)": 18.6,
      "ppl": 800.22134,
      "step": 73,
      "tokens/total": 170304,
      "tokens/train_per_sec_per_gpu": 275.0,
      "tokens/trainable": 102108
    },
    {
      "epoch": 0.3172561629153269,
      "grad_norm": 0.09075198322534561,
      "learning_rate": 0.0025,
      "loss": 5.682041168212891,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 293.548,
      "step": 74,
      "tokens/total": 171840,
      "tokens/train_per_sec_per_gpu": 24.72,
      "tokens/trainable": 102828
    },
    {
      "epoch": 0.3215434083601286,
      "grad_norm": 0.19326482713222504,
      "learning_rate": 0.0025,
      "loss": 6.265585899353027,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 526.14977,
      "step": 75,
      "tokens/total": 173632,
      "tokens/train_per_sec_per_gpu": 9.51,
      "tokens/trainable": 103667
    },
    {
      "epoch": 0.3258306538049303,
      "grad_norm": 0.09544038772583008,
      "learning_rate": 0.0025,
      "loss": 5.699280261993408,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.42,
      "memory/max_allocated (GiB)": 18.42,
      "ppl": 298.65237,
      "step": 76,
      "tokens/total": 175808,
      "tokens/train_per_sec_per_gpu": 2.17,
      "tokens/trainable": 104870
    },
    {
      "epoch": 0.33011789924973206,
      "grad_norm": 0.7390451431274414,
      "learning_rate": 0.0025,
      "loss": 6.410941123962402,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 608.46605,
      "step": 77,
      "tokens/total": 177664,
      "tokens/train_per_sec_per_gpu": 10.88,
      "tokens/trainable": 105944
    },
    {
      "epoch": 0.33440514469453375,
      "grad_norm": 0.14168062806129456,
      "learning_rate": 0.0025,
      "loss": 7.039776802062988,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 1141.13288,
      "step": 78,
      "tokens/total": 179584,
      "tokens/train_per_sec_per_gpu": 103.93,
      "tokens/trainable": 106970
    },
    {
      "epoch": 0.3386923901393355,
      "grad_norm": 0.8850395679473877,
      "learning_rate": 0.0025,
      "loss": 6.027164459228516,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 414.53792,
      "step": 79,
      "tokens/total": 181312,
      "tokens/train_per_sec_per_gpu": 97.76,
      "tokens/trainable": 107821
    },
    {
      "epoch": 0.3429796355841372,
      "grad_norm": 0.17418481409549713,
      "learning_rate": 0.0025,
      "loss": 7.046483993530273,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.57,
      "memory/max_allocated (GiB)": 18.57,
      "ppl": 1148.8124,
      "step": 80,
      "tokens/total": 183936,
      "tokens/train_per_sec_per_gpu": 123.29,
      "tokens/trainable": 109492
    },
    {
      "epoch": 0.34726688102893893,
      "grad_norm": 0.09937312453985214,
      "learning_rate": 0.0025,
      "loss": 5.586414337158203,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 266.77733,
      "step": 81,
      "tokens/total": 185600,
      "tokens/train_per_sec_per_gpu": 23.19,
      "tokens/trainable": 110255
    },
    {
      "epoch": 0.3515541264737406,
      "grad_norm": 0.3117140233516693,
      "learning_rate": 0.0025,
      "loss": 6.475000381469727,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.54,
      "memory/max_allocated (GiB)": 18.54,
      "ppl": 648.71948,
      "step": 82,
      "tokens/total": 187776,
      "tokens/train_per_sec_per_gpu": 44.82,
      "tokens/trainable": 111472
    },
    {
      "epoch": 0.3558413719185423,
      "grad_norm": 0.3783544600009918,
      "learning_rate": 0.0025,
      "loss": 6.57240104675293,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 715.08474,
      "step": 83,
      "tokens/total": 189952,
      "tokens/train_per_sec_per_gpu": 106.76,
      "tokens/trainable": 112729
    },
    {
      "epoch": 0.36012861736334406,
      "grad_norm": 0.0920150876045227,
      "learning_rate": 0.0025,
      "loss": 6.143086910247803,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.54,
      "memory/max_allocated (GiB)": 18.54,
      "ppl": 465.48828,
      "step": 84,
      "tokens/total": 192320,
      "tokens/train_per_sec_per_gpu": 49.38,
      "tokens/trainable": 114215
    },
    {
      "epoch": 0.36441586280814575,
      "grad_norm": 0.09526661038398743,
      "learning_rate": 0.0025,
      "loss": 5.70522403717041,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 300.43278,
      "step": 85,
      "tokens/total": 194048,
      "tokens/train_per_sec_per_gpu": 31.47,
      "tokens/trainable": 114924
    },
    {
      "epoch": 0.3687031082529475,
      "grad_norm": 0.15185925364494324,
      "learning_rate": 0.0025,
      "loss": 6.073705673217773,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 434.28703,
      "step": 86,
      "tokens/total": 195392,
      "tokens/train_per_sec_per_gpu": 10.82,
      "tokens/trainable": 115406
    },
    {
      "epoch": 0.3729903536977492,
      "grad_norm": 0.11615428328514099,
      "learning_rate": 0.0025,
      "loss": 6.724306106567383,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 832.39418,
      "step": 87,
      "tokens/total": 197760,
      "tokens/train_per_sec_per_gpu": 32.79,
      "tokens/trainable": 116773
    },
    {
      "epoch": 0.37727759914255093,
      "grad_norm": 0.602443277835846,
      "learning_rate": 0.0025,
      "loss": 6.486382007598877,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 656.14514,
      "step": 88,
      "tokens/total": 200192,
      "tokens/train_per_sec_per_gpu": 45.96,
      "tokens/trainable": 118226
    },
    {
      "epoch": 0.3815648445873526,
      "grad_norm": 0.09665144979953766,
      "learning_rate": 0.0025,
      "loss": 6.018401622772217,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 410.92126,
      "step": 89,
      "tokens/total": 202816,
      "tokens/train_per_sec_per_gpu": 39.89,
      "tokens/trainable": 119912
    },
    {
      "epoch": 0.3858520900321543,
      "grad_norm": 0.11786024272441864,
      "learning_rate": 0.0025,
      "loss": 6.514355659484863,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 674.75905,
      "step": 90,
      "tokens/total": 204544,
      "tokens/train_per_sec_per_gpu": 16.48,
      "tokens/trainable": 120771
    },
    {
      "epoch": 0.39013933547695606,
      "grad_norm": 0.07515699416399002,
      "learning_rate": 0.0025,
      "loss": 5.9910969734191895,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 399.853,
      "step": 91,
      "tokens/total": 206464,
      "tokens/train_per_sec_per_gpu": 37.79,
      "tokens/trainable": 121829
    },
    {
      "epoch": 0.39442658092175775,
      "grad_norm": 0.23832163214683533,
      "learning_rate": 0.0025,
      "loss": 6.451230525970459,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 633.48133,
      "step": 92,
      "tokens/total": 208384,
      "tokens/train_per_sec_per_gpu": 6.91,
      "tokens/trainable": 122828
    },
    {
      "epoch": 0.3987138263665595,
      "grad_norm": 0.06281202286481857,
      "learning_rate": 0.0025,
      "loss": 5.861863613128662,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 351.37837,
      "step": 93,
      "tokens/total": 210688,
      "tokens/train_per_sec_per_gpu": 96.78,
      "tokens/trainable": 124211
    },
    {
      "epoch": 0.4030010718113612,
      "grad_norm": 0.1325235366821289,
      "learning_rate": 0.0025,
      "loss": 6.324355125427246,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.36,
      "memory/max_allocated (GiB)": 18.36,
      "ppl": 557.99786,
      "step": 94,
      "tokens/total": 212096,
      "tokens/train_per_sec_per_gpu": 32.79,
      "tokens/trainable": 124710
    },
    {
      "epoch": 0.40728831725616294,
      "grad_norm": 0.159224733710289,
      "learning_rate": 0.0025,
      "loss": 5.801607131958008,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 330.83082,
      "step": 95,
      "tokens/total": 213696,
      "tokens/train_per_sec_per_gpu": 21.94,
      "tokens/trainable": 125363
    },
    {
      "epoch": 0.4115755627009646,
      "grad_norm": 0.1464780569076538,
      "learning_rate": 0.0025,
      "loss": 6.5112481117248535,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 672.66546,
      "step": 96,
      "tokens/total": 216320,
      "tokens/train_per_sec_per_gpu": 106.96,
      "tokens/trainable": 127129
    },
    {
      "epoch": 0.41586280814576637,
      "grad_norm": 0.10386360436677933,
      "learning_rate": 0.0025,
      "loss": 6.267631530761719,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 527.22718,
      "step": 97,
      "tokens/total": 218368,
      "tokens/train_per_sec_per_gpu": 92.64,
      "tokens/trainable": 128258
    },
    {
      "epoch": 0.42015005359056806,
      "grad_norm": 0.08260782063007355,
      "learning_rate": 0.0025,
      "loss": 6.162295341491699,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.53,
      "memory/max_allocated (GiB)": 18.53,
      "ppl": 474.516,
      "step": 98,
      "tokens/total": 220288,
      "tokens/train_per_sec_per_gpu": 228.13,
      "tokens/trainable": 129259
    },
    {
      "epoch": 0.42443729903536975,
      "grad_norm": 0.15949538350105286,
      "learning_rate": 0.0025,
      "loss": 6.426211833953857,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.53,
      "memory/max_allocated (GiB)": 18.53,
      "ppl": 617.82907,
      "step": 99,
      "tokens/total": 222528,
      "tokens/train_per_sec_per_gpu": 27.38,
      "tokens/trainable": 130563
    },
    {
      "epoch": 0.4287245444801715,
      "grad_norm": 0.13608896732330322,
      "learning_rate": 0.0025,
      "loss": 5.752037048339844,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 314.83133,
      "step": 100,
      "tokens/total": 224320,
      "tokens/train_per_sec_per_gpu": 4.44,
      "tokens/trainable": 131392
    },
    {
      "epoch": 0.4330117899249732,
      "grad_norm": 0.0612945631146431,
      "learning_rate": 0.0025,
      "loss": 6.353605270385742,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.59,
      "memory/max_allocated (GiB)": 18.59,
      "ppl": 574.56043,
      "step": 101,
      "tokens/total": 226496,
      "tokens/train_per_sec_per_gpu": 64.63,
      "tokens/trainable": 132634
    },
    {
      "epoch": 0.43729903536977494,
      "grad_norm": 0.3771326243877411,
      "learning_rate": 0.0025,
      "loss": 5.794309616088867,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 328.42537,
      "step": 102,
      "tokens/total": 228736,
      "tokens/train_per_sec_per_gpu": 23.41,
      "tokens/trainable": 134004
    },
    {
      "epoch": 0.4415862808145766,
      "grad_norm": 0.05130897834897041,
      "learning_rate": 0.0025,
      "loss": 5.724153518676758,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.57,
      "memory/max_allocated (GiB)": 18.57,
      "ppl": 306.17398,
      "step": 103,
      "tokens/total": 231360,
      "tokens/train_per_sec_per_gpu": 36.23,
      "tokens/trainable": 135730
    },
    {
      "epoch": 0.44587352625937837,
      "grad_norm": 0.05757886916399002,
      "learning_rate": 0.0025,
      "loss": 5.929745674133301,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 376.05886,
      "step": 104,
      "tokens/total": 233664,
      "tokens/train_per_sec_per_gpu": 35.62,
      "tokens/trainable": 137092
    },
    {
      "epoch": 0.45016077170418006,
      "grad_norm": 0.17747093737125397,
      "learning_rate": 0.0025,
      "loss": 5.519756317138672,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 249.57421,
      "step": 105,
      "tokens/total": 235456,
      "tokens/train_per_sec_per_gpu": 6.16,
      "tokens/trainable": 138050
    },
    {
      "epoch": 0.45444801714898175,
      "grad_norm": 0.9483416080474854,
      "learning_rate": 0.0025,
      "loss": 6.132290363311768,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.53,
      "memory/max_allocated (GiB)": 18.53,
      "ppl": 460.48964,
      "step": 106,
      "tokens/total": 238016,
      "tokens/train_per_sec_per_gpu": 8.58,
      "tokens/trainable": 139739
    },
    {
      "epoch": 0.4587352625937835,
      "grad_norm": 0.0577179454267025,
      "learning_rate": 0.0025,
      "loss": 6.135179042816162,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 461.82177,
      "step": 107,
      "tokens/total": 240768,
      "tokens/train_per_sec_per_gpu": 197.15,
      "tokens/trainable": 141578
    },
    {
      "epoch": 0.4630225080385852,
      "grad_norm": 0.13632942736148834,
      "learning_rate": 0.0025,
      "loss": 5.787478923797607,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 326.18964,
      "step": 108,
      "tokens/total": 242624,
      "tokens/train_per_sec_per_gpu": 6.1,
      "tokens/trainable": 142512
    },
    {
      "epoch": 0.46730975348338694,
      "grad_norm": 0.06379847973585129,
      "learning_rate": 0.0025,
      "loss": 5.883008003234863,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.81,
      "memory/max_allocated (GiB)": 18.81,
      "ppl": 358.88715,
      "step": 109,
      "tokens/total": 246208,
      "tokens/train_per_sec_per_gpu": 80.44,
      "tokens/trainable": 145158
    },
    {
      "epoch": 0.4715969989281886,
      "grad_norm": 0.06732220202684402,
      "learning_rate": 0.0025,
      "loss": 6.3092498779296875,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 549.6325,
      "step": 110,
      "tokens/total": 248448,
      "tokens/train_per_sec_per_gpu": 111.87,
      "tokens/trainable": 146499
    },
    {
      "epoch": 0.4758842443729904,
      "grad_norm": 0.06807619333267212,
      "learning_rate": 0.0025,
      "loss": 5.3626203536987305,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.61,
      "memory/max_allocated (GiB)": 18.61,
      "ppl": 213.28309,
      "step": 111,
      "tokens/total": 251264,
      "tokens/train_per_sec_per_gpu": 30.5,
      "tokens/trainable": 148381
    },
    {
      "epoch": 0.48017148981779206,
      "grad_norm": 0.07463269680738449,
      "learning_rate": 0.0025,
      "loss": 6.311191558837891,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 550.70075,
      "step": 112,
      "tokens/total": 252928,
      "tokens/train_per_sec_per_gpu": 25.34,
      "tokens/trainable": 149181
    },
    {
      "epoch": 0.4844587352625938,
      "grad_norm": 0.08292581140995026,
      "learning_rate": 0.0025,
      "loss": 5.2816314697265625,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 196.69051,
      "step": 113,
      "tokens/total": 254464,
      "tokens/train_per_sec_per_gpu": 136.32,
      "tokens/trainable": 149924
    },
    {
      "epoch": 0.4887459807073955,
      "grad_norm": 0.061865709722042084,
      "learning_rate": 0.0025,
      "loss": 6.011983394622803,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 408.29232,
      "step": 114,
      "tokens/total": 256768,
      "tokens/train_per_sec_per_gpu": 10.89,
      "tokens/trainable": 151286
    },
    {
      "epoch": 0.4930332261521972,
      "grad_norm": 0.10802624374628067,
      "learning_rate": 0.0025,
      "loss": 5.940651893615723,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 380.18269,
      "step": 115,
      "tokens/total": 258688,
      "tokens/train_per_sec_per_gpu": 90.71,
      "tokens/trainable": 152264
    },
    {
      "epoch": 0.49732047159699894,
      "grad_norm": 0.06800372898578644,
      "learning_rate": 0.0025,
      "loss": 5.748905181884766,
      "memory/device_reserved (GiB)": 19.35,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 313.84687,
      "step": 116,
      "tokens/total": 260800,
      "tokens/train_per_sec_per_gpu": 10.94,
      "tokens/trainable": 153396
    },
    {
      "epoch": 0.5016077170418006,
      "grad_norm": 0.05444978550076485,
      "learning_rate": 0.0025,
      "loss": 5.818991661071777,
      "memory/device_reserved (GiB)": 19.39,
      "memory/max_active (GiB)": 19.1,
      "memory/max_allocated (GiB)": 19.1,
      "ppl": 336.63244,
      "step": 117,
      "tokens/total": 264320,
      "tokens/train_per_sec_per_gpu": 603.9,
      "tokens/trainable": 156011
    },
    {
      "epoch": 0.5058949624866024,
      "grad_norm": 0.05440564081072807,
      "learning_rate": 0.0025,
      "loss": 5.707052230834961,
      "memory/device_reserved (GiB)": 19.39,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 300.98253,
      "step": 118,
      "tokens/total": 266176,
      "tokens/train_per_sec_per_gpu": 31.98,
      "tokens/trainable": 156975
    },
    {
      "epoch": 0.5058949624866024,
      "eval_loss": 5.828575611114502,
      "eval_ppl": 339.87422,
      "eval_runtime": 17.0818,
      "eval_samples_per_second": 12.177,
      "eval_steps_per_second": 12.177,
      "memory/device_reserved (GiB)": 19.39,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "step": 118
    },
    {
      "epoch": 0.5101822079314041,
      "grad_norm": 0.10290460288524628,
      "learning_rate": 0.0025,
      "loss": 6.018277168273926,
      "memory/device_reserved (GiB)": 18.51,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 410.87013,
      "step": 119,
      "tokens/total": 267776,
      "tokens/train_per_sec_per_gpu": 11.96,
      "tokens/trainable": 157706
    },
    {
      "epoch": 0.5144694533762058,
      "grad_norm": 0.06775107234716415,
      "learning_rate": 0.0025,
      "loss": 6.1761064529418945,
      "memory/device_reserved (GiB)": 18.72,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 481.11506,
      "step": 120,
      "tokens/total": 270464,
      "tokens/train_per_sec_per_gpu": 63.79,
      "tokens/trainable": 159440
    },
    {
      "epoch": 0.5187566988210075,
      "grad_norm": 0.09368869662284851,
      "learning_rate": 0.0025,
      "loss": 5.502309799194336,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.69,
      "memory/max_allocated (GiB)": 18.69,
      "ppl": 245.25777,
      "step": 121,
      "tokens/total": 273024,
      "tokens/train_per_sec_per_gpu": 0.87,
      "tokens/trainable": 161081
    },
    {
      "epoch": 0.5230439442658092,
      "grad_norm": 0.05680066719651222,
      "learning_rate": 0.0025,
      "loss": 5.426424503326416,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 227.33496,
      "step": 122,
      "tokens/total": 274944,
      "tokens/train_per_sec_per_gpu": 7.41,
      "tokens/trainable": 162094
    },
    {
      "epoch": 0.5273311897106109,
      "grad_norm": 0.05151861906051636,
      "learning_rate": 0.0025,
      "loss": 5.80197811126709,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 330.95358,
      "step": 123,
      "tokens/total": 276864,
      "tokens/train_per_sec_per_gpu": 49.9,
      "tokens/trainable": 163090
    },
    {
      "epoch": 0.5316184351554126,
      "grad_norm": 0.08232463896274567,
      "learning_rate": 0.0025,
      "loss": 5.85071325302124,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.63,
      "memory/max_allocated (GiB)": 18.63,
      "ppl": 347.48213,
      "step": 124,
      "tokens/total": 279360,
      "tokens/train_per_sec_per_gpu": 1.75,
      "tokens/trainable": 164564
    },
    {
      "epoch": 0.5359056806002144,
      "grad_norm": 0.08689901232719421,
      "learning_rate": 0.0025,
      "loss": 5.9564080238342285,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 386.22034,
      "step": 125,
      "tokens/total": 281984,
      "tokens/train_per_sec_per_gpu": 192.74,
      "tokens/trainable": 166194
    },
    {
      "epoch": 0.5401929260450161,
      "grad_norm": 0.06799639761447906,
      "learning_rate": 0.0025,
      "loss": 5.424775123596191,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 226.9603,
      "step": 126,
      "tokens/total": 283584,
      "tokens/train_per_sec_per_gpu": 20.32,
      "tokens/trainable": 166912
    },
    {
      "epoch": 0.5444801714898178,
      "grad_norm": 0.06623026728630066,
      "learning_rate": 0.0025,
      "loss": 5.536446571350098,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 253.77463,
      "step": 127,
      "tokens/total": 286400,
      "tokens/train_per_sec_per_gpu": 112.92,
      "tokens/trainable": 168763
    },
    {
      "epoch": 0.5487674169346195,
      "grad_norm": 0.07652036845684052,
      "learning_rate": 0.0025,
      "loss": 4.905512809753418,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 135.03214,
      "step": 128,
      "tokens/total": 288448,
      "tokens/train_per_sec_per_gpu": 213.64,
      "tokens/trainable": 169869
    },
    {
      "epoch": 0.5530546623794212,
      "grad_norm": 0.07232151925563812,
      "learning_rate": 0.0025,
      "loss": 5.730243682861328,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 308.04432,
      "step": 129,
      "tokens/total": 290560,
      "tokens/train_per_sec_per_gpu": 8.69,
      "tokens/trainable": 171082
    },
    {
      "epoch": 0.5573419078242229,
      "grad_norm": 0.1090153232216835,
      "learning_rate": 0.0025,
      "loss": 4.770911693572998,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 118.0268,
      "step": 130,
      "tokens/total": 292096,
      "tokens/train_per_sec_per_gpu": 2.14,
      "tokens/trainable": 171701
    },
    {
      "epoch": 0.5616291532690246,
      "grad_norm": 0.07109065353870392,
      "learning_rate": 0.0025,
      "loss": 6.074782371520996,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 434.75488,
      "step": 131,
      "tokens/total": 293952,
      "tokens/train_per_sec_per_gpu": 11.1,
      "tokens/trainable": 172572
    },
    {
      "epoch": 0.5659163987138264,
      "grad_norm": 0.06394513696432114,
      "learning_rate": 0.0025,
      "loss": 5.415735244750977,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 224.91785,
      "step": 132,
      "tokens/total": 295424,
      "tokens/train_per_sec_per_gpu": 25.67,
      "tokens/trainable": 173255
    },
    {
      "epoch": 0.5702036441586281,
      "grad_norm": 0.07912840694189072,
      "learning_rate": 0.0025,
      "loss": 5.249301910400391,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.67,
      "memory/max_allocated (GiB)": 18.67,
      "ppl": 190.43328,
      "step": 133,
      "tokens/total": 298112,
      "tokens/train_per_sec_per_gpu": 6.48,
      "tokens/trainable": 175036
    },
    {
      "epoch": 0.5744908896034298,
      "grad_norm": 0.05524107813835144,
      "learning_rate": 0.0025,
      "loss": 6.232509613037109,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 509.03135,
      "step": 134,
      "tokens/total": 300288,
      "tokens/train_per_sec_per_gpu": 114.55,
      "tokens/trainable": 176200
    },
    {
      "epoch": 0.5787781350482315,
      "grad_norm": 0.08867005258798599,
      "learning_rate": 0.0025,
      "loss": 5.501221656799316,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 244.99104,
      "step": 135,
      "tokens/total": 302016,
      "tokens/train_per_sec_per_gpu": 8.77,
      "tokens/trainable": 176997
    },
    {
      "epoch": 0.5830653804930332,
      "grad_norm": 0.0581279955804348,
      "learning_rate": 0.0025,
      "loss": 6.093716144561768,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 443.06485,
      "step": 136,
      "tokens/total": 303936,
      "tokens/train_per_sec_per_gpu": 49.42,
      "tokens/trainable": 177895
    },
    {
      "epoch": 0.587352625937835,
      "grad_norm": 0.24256502091884613,
      "learning_rate": 0.0025,
      "loss": 5.912446975708008,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 369.60948,
      "step": 137,
      "tokens/total": 306240,
      "tokens/train_per_sec_per_gpu": 117.87,
      "tokens/trainable": 179285
    },
    {
      "epoch": 0.5916398713826366,
      "grad_norm": 0.08909733593463898,
      "learning_rate": 0.0025,
      "loss": 7.02211332321167,
      "memory/device_reserved (GiB)": 19.02,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 1121.15348,
      "step": 138,
      "tokens/total": 308672,
      "tokens/train_per_sec_per_gpu": 145.17,
      "tokens/trainable": 180732
    },
    {
      "epoch": 0.5959271168274384,
      "grad_norm": 0.11382216215133667,
      "learning_rate": 0.0025,
      "loss": 6.5203537940979,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.92,
      "memory/max_allocated (GiB)": 18.92,
      "ppl": 678.8185,
      "step": 139,
      "tokens/total": 311552,
      "tokens/train_per_sec_per_gpu": 17.16,
      "tokens/trainable": 182719
    },
    {
      "epoch": 0.6002143622722401,
      "grad_norm": 0.09745891392230988,
      "learning_rate": 0.0025,
      "loss": 5.971193790435791,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 391.97333,
      "step": 140,
      "tokens/total": 313856,
      "tokens/train_per_sec_per_gpu": 150.24,
      "tokens/trainable": 184142
    },
    {
      "epoch": 0.6045016077170418,
      "grad_norm": 0.0861353650689125,
      "learning_rate": 0.0025,
      "loss": 5.941656112670898,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 380.56467,
      "step": 141,
      "tokens/total": 315392,
      "tokens/train_per_sec_per_gpu": 68.11,
      "tokens/trainable": 184842
    },
    {
      "epoch": 0.6087888531618435,
      "grad_norm": 0.07818640768527985,
      "learning_rate": 0.0025,
      "loss": 5.195644378662109,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 180.48441,
      "step": 142,
      "tokens/total": 317248,
      "tokens/train_per_sec_per_gpu": 24.45,
      "tokens/trainable": 185810
    },
    {
      "epoch": 0.6130760986066452,
      "grad_norm": 0.06124793365597725,
      "learning_rate": 0.0025,
      "loss": 5.502886772155762,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 245.39932,
      "step": 143,
      "tokens/total": 319232,
      "tokens/train_per_sec_per_gpu": 22.21,
      "tokens/trainable": 186800
    },
    {
      "epoch": 0.617363344051447,
      "grad_norm": 0.0705777034163475,
      "learning_rate": 0.0025,
      "loss": 6.341245651245117,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.72,
      "memory/max_allocated (GiB)": 18.72,
      "ppl": 567.50278,
      "step": 144,
      "tokens/total": 322304,
      "tokens/train_per_sec_per_gpu": 19.8,
      "tokens/trainable": 189022
    },
    {
      "epoch": 0.6216505894962486,
      "grad_norm": 0.07551176846027374,
      "learning_rate": 0.0025,
      "loss": 5.537265777587891,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 253.9826,
      "step": 145,
      "tokens/total": 324608,
      "tokens/train_per_sec_per_gpu": 14.15,
      "tokens/trainable": 190415
    },
    {
      "epoch": 0.6259378349410504,
      "grad_norm": 0.04925059527158737,
      "learning_rate": 0.0025,
      "loss": 5.8168511390686035,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 335.91264,
      "step": 146,
      "tokens/total": 327168,
      "tokens/train_per_sec_per_gpu": 28.95,
      "tokens/trainable": 192079
    },
    {
      "epoch": 0.6302250803858521,
      "grad_norm": 0.06655045598745346,
      "learning_rate": 0.0025,
      "loss": 5.837461471557617,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 342.90775,
      "step": 147,
      "tokens/total": 328640,
      "tokens/train_per_sec_per_gpu": 33.92,
      "tokens/trainable": 192665
    },
    {
      "epoch": 0.6345123258306538,
      "grad_norm": 0.07563883811235428,
      "learning_rate": 0.0025,
      "loss": 5.163289546966553,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 174.73832,
      "step": 148,
      "tokens/total": 330432,
      "tokens/train_per_sec_per_gpu": 112.65,
      "tokens/trainable": 193531
    },
    {
      "epoch": 0.6387995712754555,
      "grad_norm": 0.05340481176972389,
      "learning_rate": 0.0025,
      "loss": 5.348882675170898,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 210.37311,
      "step": 149,
      "tokens/total": 332032,
      "tokens/train_per_sec_per_gpu": 35.17,
      "tokens/trainable": 194286
    },
    {
      "epoch": 0.6430868167202572,
      "grad_norm": 0.07053768634796143,
      "learning_rate": 0.0025,
      "loss": 6.0246500968933105,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 413.49693,
      "step": 150,
      "tokens/total": 334592,
      "tokens/train_per_sec_per_gpu": 53.06,
      "tokens/trainable": 195894
    },
    {
      "epoch": 0.647374062165059,
      "grad_norm": 0.10346148163080215,
      "learning_rate": 0.0025,
      "loss": 6.111691474914551,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 451.1011,
      "step": 151,
      "tokens/total": 336896,
      "tokens/train_per_sec_per_gpu": 84.27,
      "tokens/trainable": 197140
    },
    {
      "epoch": 0.6516613076098606,
      "grad_norm": 0.05668232962489128,
      "learning_rate": 0.0025,
      "loss": 5.328207015991211,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 206.06817,
      "step": 152,
      "tokens/total": 339392,
      "tokens/train_per_sec_per_gpu": 68.85,
      "tokens/trainable": 198641
    },
    {
      "epoch": 0.6559485530546624,
      "grad_norm": 0.07267311960458755,
      "learning_rate": 0.0025,
      "loss": 5.421064853668213,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 226.11978,
      "step": 153,
      "tokens/total": 341312,
      "tokens/train_per_sec_per_gpu": 108.38,
      "tokens/trainable": 199572
    },
    {
      "epoch": 0.6602357984994641,
      "grad_norm": 0.05049528181552887,
      "learning_rate": 0.0025,
      "loss": 5.701568603515625,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.63,
      "memory/max_allocated (GiB)": 18.63,
      "ppl": 299.33657,
      "step": 154,
      "tokens/total": 344064,
      "tokens/train_per_sec_per_gpu": 300.52,
      "tokens/trainable": 201242
    },
    {
      "epoch": 0.6645230439442658,
      "grad_norm": 0.056077949702739716,
      "learning_rate": 0.0025,
      "loss": 5.401850700378418,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 221.81655,
      "step": 155,
      "tokens/total": 346432,
      "tokens/train_per_sec_per_gpu": 258.94,
      "tokens/trainable": 202788
    },
    {
      "epoch": 0.6688102893890675,
      "grad_norm": 0.11511314660310745,
      "learning_rate": 0.0025,
      "loss": 5.723288536071777,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 305.90926,
      "step": 156,
      "tokens/total": 348160,
      "tokens/train_per_sec_per_gpu": 129.3,
      "tokens/trainable": 203548
    },
    {
      "epoch": 0.6730975348338692,
      "grad_norm": 0.04815061762928963,
      "learning_rate": 0.0025,
      "loss": 5.29791259765625,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 199.91906,
      "step": 157,
      "tokens/total": 349888,
      "tokens/train_per_sec_per_gpu": 61.9,
      "tokens/trainable": 204408
    },
    {
      "epoch": 0.677384780278671,
      "grad_norm": 0.0801524966955185,
      "learning_rate": 0.0025,
      "loss": 6.227181911468506,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 506.3266,
      "step": 158,
      "tokens/total": 351744,
      "tokens/train_per_sec_per_gpu": 102.56,
      "tokens/trainable": 205262
    },
    {
      "epoch": 0.6816720257234726,
      "grad_norm": 0.059293024241924286,
      "learning_rate": 0.0025,
      "loss": 5.615620136260986,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 274.68367,
      "step": 159,
      "tokens/total": 354176,
      "tokens/train_per_sec_per_gpu": 31.25,
      "tokens/trainable": 206828
    },
    {
      "epoch": 0.6859592711682744,
      "grad_norm": 0.07457486540079117,
      "learning_rate": 0.0025,
      "loss": 5.283931255340576,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 197.14337,
      "step": 160,
      "tokens/total": 356416,
      "tokens/train_per_sec_per_gpu": 51.94,
      "tokens/trainable": 208192
    },
    {
      "epoch": 0.6902465166130761,
      "grad_norm": 0.06552717089653015,
      "learning_rate": 0.0025,
      "loss": 5.763904571533203,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 318.58986,
      "step": 161,
      "tokens/total": 358336,
      "tokens/train_per_sec_per_gpu": 169.49,
      "tokens/trainable": 209238
    },
    {
      "epoch": 0.6945337620578779,
      "grad_norm": 0.07754746079444885,
      "learning_rate": 0.0025,
      "loss": 5.163079261779785,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 174.70158,
      "step": 162,
      "tokens/total": 360576,
      "tokens/train_per_sec_per_gpu": 17.68,
      "tokens/trainable": 210560
    },
    {
      "epoch": 0.6988210075026795,
      "grad_norm": 0.11694307625293732,
      "learning_rate": 0.0025,
      "loss": 6.16943359375,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 477.91534,
      "step": 163,
      "tokens/total": 362560,
      "tokens/train_per_sec_per_gpu": 6.89,
      "tokens/trainable": 211646
    },
    {
      "epoch": 0.7031082529474812,
      "grad_norm": 0.3541814684867859,
      "learning_rate": 0.0025,
      "loss": 6.192336082458496,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 488.98709,
      "step": 164,
      "tokens/total": 364032,
      "tokens/train_per_sec_per_gpu": 68.77,
      "tokens/trainable": 212233
    },
    {
      "epoch": 0.707395498392283,
      "grad_norm": 0.0691906213760376,
      "learning_rate": 0.0025,
      "loss": 5.444247722625732,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 231.42312,
      "step": 165,
      "tokens/total": 366208,
      "tokens/train_per_sec_per_gpu": 74.88,
      "tokens/trainable": 213482
    },
    {
      "epoch": 0.7116827438370846,
      "grad_norm": 0.06896678358316422,
      "learning_rate": 0.0025,
      "loss": 5.45513391494751,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 233.9562,
      "step": 166,
      "tokens/total": 368192,
      "tokens/train_per_sec_per_gpu": 150.94,
      "tokens/trainable": 214570
    },
    {
      "epoch": 0.7159699892818864,
      "grad_norm": 0.07068932056427002,
      "learning_rate": 0.0025,
      "loss": 5.088237762451172,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 162.10394,
      "step": 167,
      "tokens/total": 370432,
      "tokens/train_per_sec_per_gpu": 37.23,
      "tokens/trainable": 215958
    },
    {
      "epoch": 0.7202572347266881,
      "grad_norm": 0.07037021219730377,
      "learning_rate": 0.0025,
      "loss": 5.780329704284668,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 323.86595,
      "step": 168,
      "tokens/total": 373248,
      "tokens/train_per_sec_per_gpu": 48.79,
      "tokens/trainable": 217782
    },
    {
      "epoch": 0.7245444801714899,
      "grad_norm": 0.09113272279500961,
      "learning_rate": 0.0025,
      "loss": 5.445977210998535,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 231.82371,
      "step": 169,
      "tokens/total": 375296,
      "tokens/train_per_sec_per_gpu": 135.42,
      "tokens/trainable": 218943
    },
    {
      "epoch": 0.7288317256162915,
      "grad_norm": 0.04421088844537735,
      "learning_rate": 0.0025,
      "loss": 5.593143463134766,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 268.57856,
      "step": 170,
      "tokens/total": 377280,
      "tokens/train_per_sec_per_gpu": 32.25,
      "tokens/trainable": 220097
    },
    {
      "epoch": 0.7331189710610932,
      "grad_norm": 0.07146560400724411,
      "learning_rate": 0.0025,
      "loss": 5.347398281097412,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 210.06107,
      "step": 171,
      "tokens/total": 379392,
      "tokens/train_per_sec_per_gpu": 15.36,
      "tokens/trainable": 221250
    },
    {
      "epoch": 0.737406216505895,
      "grad_norm": 0.058737654238939285,
      "learning_rate": 0.0025,
      "loss": 5.430908203125,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.58,
      "memory/max_allocated (GiB)": 18.58,
      "ppl": 228.35655,
      "step": 172,
      "tokens/total": 382144,
      "tokens/train_per_sec_per_gpu": 17.26,
      "tokens/trainable": 223081
    },
    {
      "epoch": 0.7416934619506966,
      "grad_norm": 0.08348660171031952,
      "learning_rate": 0.0025,
      "loss": 5.785150527954102,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 325.43102,
      "step": 173,
      "tokens/total": 384768,
      "tokens/train_per_sec_per_gpu": 56.26,
      "tokens/trainable": 224769
    },
    {
      "epoch": 0.7459807073954984,
      "grad_norm": 0.04298631101846695,
      "learning_rate": 0.0025,
      "loss": 5.393362998962402,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 219.94181,
      "step": 174,
      "tokens/total": 387136,
      "tokens/train_per_sec_per_gpu": 37.67,
      "tokens/trainable": 226216
    },
    {
      "epoch": 0.7502679528403001,
      "grad_norm": 0.08122890442609787,
      "learning_rate": 0.0025,
      "loss": 5.101113796234131,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 164.2047,
      "step": 175,
      "tokens/total": 388608,
      "tokens/train_per_sec_per_gpu": 44.01,
      "tokens/trainable": 226770
    },
    {
      "epoch": 0.7545551982851019,
      "grad_norm": 0.11935891956090927,
      "learning_rate": 0.0025,
      "loss": 5.519252300262451,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 249.44845,
      "step": 176,
      "tokens/total": 390784,
      "tokens/train_per_sec_per_gpu": 39.54,
      "tokens/trainable": 228098
    },
    {
      "epoch": 0.7588424437299035,
      "grad_norm": 0.15656660497188568,
      "learning_rate": 0.0025,
      "loss": 4.941908836364746,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 140.0373,
      "step": 177,
      "tokens/total": 392512,
      "tokens/train_per_sec_per_gpu": 36.08,
      "tokens/trainable": 228809
    },
    {
      "epoch": 0.7588424437299035,
      "eval_loss": 5.460564136505127,
      "eval_ppl": 235.23009,
      "eval_runtime": 17.2516,
      "eval_samples_per_second": 12.057,
      "eval_steps_per_second": 12.057,
      "memory/device_reserved (GiB)": 19.56,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "step": 177
    },
    {
      "epoch": 0.7631296891747053,
      "grad_norm": 0.05968443304300308,
      "learning_rate": 0.0025,
      "loss": 5.653349876403809,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 19.12,
      "memory/max_allocated (GiB)": 19.12,
      "ppl": 285.2454,
      "step": 178,
      "tokens/total": 396416,
      "tokens/train_per_sec_per_gpu": 110.71,
      "tokens/trainable": 231745
    },
    {
      "epoch": 0.767416934619507,
      "grad_norm": 0.11675012111663818,
      "learning_rate": 0.0025,
      "loss": 5.658262252807617,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 286.65008,
      "step": 179,
      "tokens/total": 399040,
      "tokens/train_per_sec_per_gpu": 101.59,
      "tokens/trainable": 233370
    },
    {
      "epoch": 0.7717041800643086,
      "grad_norm": 0.07265754044055939,
      "learning_rate": 0.0025,
      "loss": 5.7116546630859375,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 302.37098,
      "step": 180,
      "tokens/total": 400896,
      "tokens/train_per_sec_per_gpu": 5.2,
      "tokens/trainable": 234357
    },
    {
      "epoch": 0.7759914255091104,
      "grad_norm": 0.06884697079658508,
      "learning_rate": 0.0025,
      "loss": 5.63464879989624,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 279.96058,
      "step": 181,
      "tokens/total": 403072,
      "tokens/train_per_sec_per_gpu": 105.0,
      "tokens/trainable": 235702
    },
    {
      "epoch": 0.7802786709539121,
      "grad_norm": 0.12419719249010086,
      "learning_rate": 0.0025,
      "loss": 5.239850997924805,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 188.64199,
      "step": 182,
      "tokens/total": 404928,
      "tokens/train_per_sec_per_gpu": 19.58,
      "tokens/trainable": 236615
    },
    {
      "epoch": 0.7845659163987139,
      "grad_norm": 0.10391955822706223,
      "learning_rate": 0.0025,
      "loss": 4.728728294372559,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 113.15158,
      "step": 183,
      "tokens/total": 406272,
      "tokens/train_per_sec_per_gpu": 24.51,
      "tokens/trainable": 237082
    },
    {
      "epoch": 0.7888531618435155,
      "grad_norm": 0.07022784650325775,
      "learning_rate": 0.0025,
      "loss": 5.611477375030518,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 273.54807,
      "step": 184,
      "tokens/total": 408384,
      "tokens/train_per_sec_per_gpu": 110.02,
      "tokens/trainable": 238396
    },
    {
      "epoch": 0.7931404072883173,
      "grad_norm": 0.11678767204284668,
      "learning_rate": 0.0025,
      "loss": 6.098667144775391,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 445.2639,
      "step": 185,
      "tokens/total": 410560,
      "tokens/train_per_sec_per_gpu": 2.18,
      "tokens/trainable": 239649
    },
    {
      "epoch": 0.797427652733119,
      "grad_norm": 0.06710375845432281,
      "learning_rate": 0.0025,
      "loss": 5.643878936767578,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.65,
      "memory/max_allocated (GiB)": 18.65,
      "ppl": 282.55661,
      "step": 186,
      "tokens/total": 412800,
      "tokens/train_per_sec_per_gpu": 12.9,
      "tokens/trainable": 240972
    },
    {
      "epoch": 0.8017148981779206,
      "grad_norm": 0.09496990591287613,
      "learning_rate": 0.0025,
      "loss": 5.7106122970581055,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 302.05596,
      "step": 187,
      "tokens/total": 415296,
      "tokens/train_per_sec_per_gpu": 67.78,
      "tokens/trainable": 242450
    },
    {
      "epoch": 0.8060021436227224,
      "grad_norm": 0.06658744066953659,
      "learning_rate": 0.0025,
      "loss": 5.754177093505859,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 315.50581,
      "step": 188,
      "tokens/total": 417856,
      "tokens/train_per_sec_per_gpu": 14.18,
      "tokens/trainable": 244076
    },
    {
      "epoch": 0.8102893890675241,
      "grad_norm": 0.054122067987918854,
      "learning_rate": 0.0025,
      "loss": 5.310729026794434,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 202.4978,
      "step": 189,
      "tokens/total": 419584,
      "tokens/train_per_sec_per_gpu": 11.57,
      "tokens/trainable": 244832
    },
    {
      "epoch": 0.8145766345123259,
      "grad_norm": 0.11224393546581268,
      "learning_rate": 0.0025,
      "loss": 5.699178695678711,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 298.62204,
      "step": 190,
      "tokens/total": 420928,
      "tokens/train_per_sec_per_gpu": 8.17,
      "tokens/trainable": 245280
    },
    {
      "epoch": 0.8188638799571275,
      "grad_norm": 0.05273159593343735,
      "learning_rate": 0.0025,
      "loss": 5.184141159057617,
      "memory/device_reserved (GiB)": 19.59,
      "memory/max_active (GiB)": 18.73,
      "memory/max_allocated (GiB)": 18.73,
      "ppl": 178.42015,
      "step": 191,
      "tokens/total": 424000,
      "tokens/train_per_sec_per_gpu": 53.69,
      "tokens/trainable": 247590
    },
    {
      "epoch": 0.8231511254019293,
      "grad_norm": 0.03733557090163231,
      "learning_rate": 0.0025,
      "loss": 5.854153633117676,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 19.12,
      "memory/max_allocated (GiB)": 19.12,
      "ppl": 348.67966,
      "step": 192,
      "tokens/total": 428480,
      "tokens/train_per_sec_per_gpu": 52.36,
      "tokens/trainable": 251024
    },
    {
      "epoch": 0.827438370846731,
      "grad_norm": 0.06321356445550919,
      "learning_rate": 0.0025,
      "loss": 5.20097017288208,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 181.44819,
      "step": 193,
      "tokens/total": 430656,
      "tokens/train_per_sec_per_gpu": 1.74,
      "tokens/trainable": 252277
    },
    {
      "epoch": 0.8317256162915327,
      "grad_norm": 0.1052091047167778,
      "learning_rate": 0.0025,
      "loss": 4.629343509674072,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 102.44679,
      "step": 194,
      "tokens/total": 432064,
      "tokens/train_per_sec_per_gpu": 21.14,
      "tokens/trainable": 252834
    },
    {
      "epoch": 0.8360128617363344,
      "grad_norm": 0.06142156571149826,
      "learning_rate": 0.0025,
      "loss": 5.491484642028809,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 242.61714,
      "step": 195,
      "tokens/total": 434624,
      "tokens/train_per_sec_per_gpu": 23.29,
      "tokens/trainable": 254420
    },
    {
      "epoch": 0.8403001071811361,
      "grad_norm": 0.05292431265115738,
      "learning_rate": 0.0025,
      "loss": 5.481139183044434,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.62,
      "memory/max_allocated (GiB)": 18.62,
      "ppl": 240.12009,
      "step": 196,
      "tokens/total": 437440,
      "tokens/train_per_sec_per_gpu": 52.86,
      "tokens/trainable": 256271
    },
    {
      "epoch": 0.8445873526259379,
      "grad_norm": 0.0795208215713501,
      "learning_rate": 0.0025,
      "loss": 5.259885787963867,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 192.45951,
      "step": 197,
      "tokens/total": 439744,
      "tokens/train_per_sec_per_gpu": 64.52,
      "tokens/trainable": 257576
    },
    {
      "epoch": 0.8488745980707395,
      "grad_norm": 0.08473947644233704,
      "learning_rate": 0.0025,
      "loss": 5.696974754333496,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.59,
      "memory/max_allocated (GiB)": 18.59,
      "ppl": 297.96462,
      "step": 198,
      "tokens/total": 442816,
      "tokens/train_per_sec_per_gpu": 36.19,
      "tokens/trainable": 259567
    },
    {
      "epoch": 0.8531618435155413,
      "grad_norm": 0.057201892137527466,
      "learning_rate": 0.0025,
      "loss": 5.780117034912109,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.62,
      "memory/max_allocated (GiB)": 18.62,
      "ppl": 323.79708,
      "step": 199,
      "tokens/total": 445888,
      "tokens/train_per_sec_per_gpu": 81.99,
      "tokens/trainable": 261637
    },
    {
      "epoch": 0.857449088960343,
      "grad_norm": 0.09185982495546341,
      "learning_rate": 0.0025,
      "loss": 6.029691219329834,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 415.58668,
      "step": 200,
      "tokens/total": 449408,
      "tokens/train_per_sec_per_gpu": 163.44,
      "tokens/trainable": 264161
    },
    {
      "epoch": 0.8617363344051447,
      "grad_norm": 0.055811040103435516,
      "learning_rate": 0.0025,
      "loss": 5.37666654586792,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 216.30005,
      "step": 201,
      "tokens/total": 452032,
      "tokens/train_per_sec_per_gpu": 14.76,
      "tokens/trainable": 265933
    },
    {
      "epoch": 0.8660235798499464,
      "grad_norm": 0.06049516424536705,
      "learning_rate": 0.0025,
      "loss": 5.930117130279541,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 376.19858,
      "step": 202,
      "tokens/total": 455232,
      "tokens/train_per_sec_per_gpu": 34.59,
      "tokens/trainable": 268176
    },
    {
      "epoch": 0.8703108252947481,
      "grad_norm": 0.04760754853487015,
      "learning_rate": 0.0025,
      "loss": 5.988245487213135,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.54,
      "memory/max_allocated (GiB)": 18.54,
      "ppl": 398.71445,
      "step": 203,
      "tokens/total": 457792,
      "tokens/train_per_sec_per_gpu": 23.31,
      "tokens/trainable": 269826
    },
    {
      "epoch": 0.8745980707395499,
      "grad_norm": 0.3524979054927826,
      "learning_rate": 0.0025,
      "loss": 5.5907745361328125,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 267.94307,
      "step": 204,
      "tokens/total": 460224,
      "tokens/train_per_sec_per_gpu": 112.76,
      "tokens/trainable": 271226
    },
    {
      "epoch": 0.8788853161843515,
      "grad_norm": 0.07816269993782043,
      "learning_rate": 0.0025,
      "loss": 5.5978217124938965,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.53,
      "memory/max_allocated (GiB)": 18.53,
      "ppl": 269.83798,
      "step": 205,
      "tokens/total": 463232,
      "tokens/train_per_sec_per_gpu": 83.11,
      "tokens/trainable": 273301
    },
    {
      "epoch": 0.8831725616291533,
      "grad_norm": 0.06009744852781296,
      "learning_rate": 0.0025,
      "loss": 5.7178802490234375,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 304.25928,
      "step": 206,
      "tokens/total": 464960,
      "tokens/train_per_sec_per_gpu": 16.0,
      "tokens/trainable": 274259
    },
    {
      "epoch": 0.887459807073955,
      "grad_norm": 0.09694831818342209,
      "learning_rate": 0.0025,
      "loss": 5.388084888458252,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 218.78399,
      "step": 207,
      "tokens/total": 467008,
      "tokens/train_per_sec_per_gpu": 34.61,
      "tokens/trainable": 275392
    },
    {
      "epoch": 0.8917470525187567,
      "grad_norm": 0.06148134917020798,
      "learning_rate": 0.0025,
      "loss": 5.084897041320801,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 161.5633,
      "step": 208,
      "tokens/total": 468608,
      "tokens/train_per_sec_per_gpu": 9.12,
      "tokens/trainable": 276045
    },
    {
      "epoch": 0.8960342979635584,
      "grad_norm": 0.09476284682750702,
      "learning_rate": 0.0025,
      "loss": 5.941885471343994,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.6,
      "memory/max_allocated (GiB)": 18.6,
      "ppl": 380.65196,
      "step": 209,
      "tokens/total": 470912,
      "tokens/train_per_sec_per_gpu": 31.97,
      "tokens/trainable": 277484
    },
    {
      "epoch": 0.9003215434083601,
      "grad_norm": 0.07026304304599762,
      "learning_rate": 0.0025,
      "loss": 5.341467380523682,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.7,
      "memory/max_allocated (GiB)": 18.7,
      "ppl": 208.8189,
      "step": 210,
      "tokens/total": 473920,
      "tokens/train_per_sec_per_gpu": 114.73,
      "tokens/trainable": 279470
    },
    {
      "epoch": 0.9046087888531619,
      "grad_norm": 0.046190474182367325,
      "learning_rate": 0.0025,
      "loss": 5.205893516540527,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.54,
      "memory/max_allocated (GiB)": 18.54,
      "ppl": 182.34373,
      "step": 211,
      "tokens/total": 476480,
      "tokens/train_per_sec_per_gpu": 112.77,
      "tokens/trainable": 280947
    },
    {
      "epoch": 0.9088960342979635,
      "grad_norm": 0.06688795238733292,
      "learning_rate": 0.0025,
      "loss": 4.937982559204102,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 139.48856,
      "step": 212,
      "tokens/total": 478016,
      "tokens/train_per_sec_per_gpu": 101.68,
      "tokens/trainable": 281612
    },
    {
      "epoch": 0.9131832797427653,
      "grad_norm": 0.07849342375993729,
      "learning_rate": 0.0025,
      "loss": 5.182241439819336,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 178.08152,
      "step": 213,
      "tokens/total": 479680,
      "tokens/train_per_sec_per_gpu": 150.05,
      "tokens/trainable": 282478
    },
    {
      "epoch": 0.917470525187567,
      "grad_norm": 0.05929256230592728,
      "learning_rate": 0.0025,
      "loss": 4.967282295227051,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.42,
      "memory/max_allocated (GiB)": 18.42,
      "ppl": 143.636,
      "step": 214,
      "tokens/total": 481728,
      "tokens/train_per_sec_per_gpu": 128.72,
      "tokens/trainable": 283641
    },
    {
      "epoch": 0.9217577706323687,
      "grad_norm": 0.07487839460372925,
      "learning_rate": 0.0025,
      "loss": 5.071913719177246,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 159.47923,
      "step": 215,
      "tokens/total": 483200,
      "tokens/train_per_sec_per_gpu": 13.64,
      "tokens/trainable": 284253
    },
    {
      "epoch": 0.9260450160771704,
      "grad_norm": 0.08760891854763031,
      "learning_rate": 0.0025,
      "loss": 5.086419582366943,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 161.80948,
      "step": 216,
      "tokens/total": 485184,
      "tokens/train_per_sec_per_gpu": 33.77,
      "tokens/trainable": 285350
    },
    {
      "epoch": 0.9303322615219721,
      "grad_norm": 0.058040693402290344,
      "learning_rate": 0.0025,
      "loss": 5.140557289123535,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 170.81093,
      "step": 217,
      "tokens/total": 487488,
      "tokens/train_per_sec_per_gpu": 148.26,
      "tokens/trainable": 286725
    },
    {
      "epoch": 0.9346195069667739,
      "grad_norm": 0.051069747656583786,
      "learning_rate": 0.0025,
      "loss": 5.112553119659424,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.54,
      "memory/max_allocated (GiB)": 18.54,
      "ppl": 166.09387,
      "step": 218,
      "tokens/total": 490176,
      "tokens/train_per_sec_per_gpu": 7.43,
      "tokens/trainable": 288433
    },
    {
      "epoch": 0.9389067524115756,
      "grad_norm": 0.04658494517207146,
      "learning_rate": 0.0025,
      "loss": 5.061060428619385,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 157.75772,
      "step": 219,
      "tokens/total": 492288,
      "tokens/train_per_sec_per_gpu": 18.83,
      "tokens/trainable": 289637
    },
    {
      "epoch": 0.9431939978563773,
      "grad_norm": 0.058834441006183624,
      "learning_rate": 0.0025,
      "loss": 5.130789756774902,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 169.15065,
      "step": 220,
      "tokens/total": 494848,
      "tokens/train_per_sec_per_gpu": 137.69,
      "tokens/trainable": 291132
    },
    {
      "epoch": 0.947481243301179,
      "grad_norm": 0.10351614654064178,
      "learning_rate": 0.0025,
      "loss": 5.497127532958984,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 243.99007,
      "step": 221,
      "tokens/total": 497088,
      "tokens/train_per_sec_per_gpu": 195.46,
      "tokens/trainable": 292518
    },
    {
      "epoch": 0.9517684887459807,
      "grad_norm": 0.14364013075828552,
      "learning_rate": 0.0025,
      "loss": 5.386436939239502,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 218.42374,
      "step": 222,
      "tokens/total": 499520,
      "tokens/train_per_sec_per_gpu": 4.31,
      "tokens/trainable": 293866
    },
    {
      "epoch": 0.9560557341907824,
      "grad_norm": 0.06514472514390945,
      "learning_rate": 0.0025,
      "loss": 5.5871148109436035,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.53,
      "memory/max_allocated (GiB)": 18.53,
      "ppl": 266.96427,
      "step": 223,
      "tokens/total": 502208,
      "tokens/train_per_sec_per_gpu": 19.56,
      "tokens/trainable": 295684
    },
    {
      "epoch": 0.9603429796355841,
      "grad_norm": 0.05746331810951233,
      "learning_rate": 0.0025,
      "loss": 5.019771099090576,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.57,
      "memory/max_allocated (GiB)": 18.57,
      "ppl": 151.37665,
      "step": 224,
      "tokens/total": 504576,
      "tokens/train_per_sec_per_gpu": 133.41,
      "tokens/trainable": 297180
    },
    {
      "epoch": 0.9646302250803859,
      "grad_norm": 0.06428291648626328,
      "learning_rate": 0.0025,
      "loss": 5.374805450439453,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.77,
      "memory/max_allocated (GiB)": 18.77,
      "ppl": 215.89787,
      "step": 225,
      "tokens/total": 508224,
      "tokens/train_per_sec_per_gpu": 174.74,
      "tokens/trainable": 299827
    },
    {
      "epoch": 0.9689174705251876,
      "grad_norm": 0.06595566868782043,
      "learning_rate": 0.0025,
      "loss": 5.579074859619141,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 264.82649,
      "step": 226,
      "tokens/total": 509824,
      "tokens/train_per_sec_per_gpu": 6.15,
      "tokens/trainable": 300582
    },
    {
      "epoch": 0.9732047159699893,
      "grad_norm": 0.06129618361592293,
      "learning_rate": 0.0025,
      "loss": 4.87337064743042,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.67,
      "memory/max_allocated (GiB)": 18.67,
      "ppl": 130.76092,
      "step": 227,
      "tokens/total": 512768,
      "tokens/train_per_sec_per_gpu": 14.39,
      "tokens/trainable": 302643
    },
    {
      "epoch": 0.977491961414791,
      "grad_norm": 0.07292018085718155,
      "learning_rate": 0.0025,
      "loss": 5.493099212646484,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.55,
      "memory/max_allocated (GiB)": 18.55,
      "ppl": 243.00918,
      "step": 228,
      "tokens/total": 515328,
      "tokens/train_per_sec_per_gpu": 8.6,
      "tokens/trainable": 304358
    },
    {
      "epoch": 0.9817792068595927,
      "grad_norm": 0.04551401734352112,
      "learning_rate": 0.0025,
      "loss": 5.522790908813477,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.54,
      "memory/max_allocated (GiB)": 18.54,
      "ppl": 250.33272,
      "step": 229,
      "tokens/total": 518144,
      "tokens/train_per_sec_per_gpu": 36.31,
      "tokens/trainable": 306242
    },
    {
      "epoch": 0.9860664523043944,
      "grad_norm": 0.08372899889945984,
      "learning_rate": 0.0025,
      "loss": 5.140377998352051,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 170.78031,
      "step": 230,
      "tokens/total": 520000,
      "tokens/train_per_sec_per_gpu": 20.62,
      "tokens/trainable": 307298
    },
    {
      "epoch": 0.9903536977491961,
      "grad_norm": 0.0876646488904953,
      "learning_rate": 0.0025,
      "loss": 5.655140399932861,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.96,
      "memory/max_allocated (GiB)": 18.96,
      "ppl": 285.7566,
      "step": 231,
      "tokens/total": 523456,
      "tokens/train_per_sec_per_gpu": 176.78,
      "tokens/trainable": 309825
    },
    {
      "epoch": 0.9946409431939979,
      "grad_norm": 0.11478639394044876,
      "learning_rate": 0.0025,
      "loss": 5.731156826019287,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 308.32574,
      "step": 232,
      "tokens/total": 524928,
      "tokens/train_per_sec_per_gpu": 73.31,
      "tokens/trainable": 310402
    },
    {
      "epoch": 0.9989281886387996,
      "grad_norm": 0.05332854762673378,
      "learning_rate": 0.0025,
      "loss": 4.958339214324951,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 142.35717,
      "step": 233,
      "tokens/total": 527040,
      "tokens/train_per_sec_per_gpu": 168.24,
      "tokens/trainable": 311695
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.09713005274534225,
      "learning_rate": 0.0025,
      "loss": 4.436938285827637,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 84.51578,
      "step": 234,
      "tokens/total": 527680,
      "tokens/train_per_sec_per_gpu": 682.75,
      "tokens/trainable": 312130
    },
    {
      "epoch": 1.0042872454448017,
      "grad_norm": 0.05925685912370682,
      "learning_rate": 0.0025,
      "loss": 4.939643859863281,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 139.72048,
      "step": 235,
      "tokens/total": 529920,
      "tokens/train_per_sec_per_gpu": 77.24,
      "tokens/trainable": 313388
    },
    {
      "epoch": 1.0085744908896035,
      "grad_norm": 0.09591115266084671,
      "learning_rate": 0.0025,
      "loss": 5.589860916137695,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 267.69838,
      "step": 236,
      "tokens/total": 531648,
      "tokens/train_per_sec_per_gpu": 11.46,
      "tokens/trainable": 314241
    },
    {
      "epoch": 1.0085744908896035,
      "eval_loss": 5.21028470993042,
      "eval_ppl": 183.14619,
      "eval_runtime": 17.0244,
      "eval_samples_per_second": 12.218,
      "eval_steps_per_second": 12.218,
      "memory/device_reserved (GiB)": 19.84,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "step": 236
    },
    {
      "epoch": 1.0128617363344052,
      "grad_norm": 0.07797813415527344,
      "learning_rate": 0.0025,
      "loss": 5.490115165710449,
      "memory/device_reserved (GiB)": 18.62,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 242.28511,
      "step": 237,
      "tokens/total": 533760,
      "tokens/train_per_sec_per_gpu": 66.73,
      "tokens/trainable": 315528
    },
    {
      "epoch": 1.0171489817792068,
      "grad_norm": 0.058185361325740814,
      "learning_rate": 0.0025,
      "loss": 5.693612098693848,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.75,
      "memory/max_allocated (GiB)": 18.75,
      "ppl": 296.96435,
      "step": 238,
      "tokens/total": 536832,
      "tokens/train_per_sec_per_gpu": 6.22,
      "tokens/trainable": 317658
    },
    {
      "epoch": 1.0214362272240085,
      "grad_norm": 0.06836100667715073,
      "learning_rate": 0.0025,
      "loss": 5.170307159423828,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 175.96888,
      "step": 239,
      "tokens/total": 539648,
      "tokens/train_per_sec_per_gpu": 124.58,
      "tokens/trainable": 319553
    },
    {
      "epoch": 1.0257234726688103,
      "grad_norm": 0.063715860247612,
      "learning_rate": 0.0025,
      "loss": 5.572962760925293,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 263.21278,
      "step": 240,
      "tokens/total": 542144,
      "tokens/train_per_sec_per_gpu": 16.68,
      "tokens/trainable": 321099
    },
    {
      "epoch": 1.030010718113612,
      "grad_norm": 0.04595501348376274,
      "learning_rate": 0.0025,
      "loss": 5.575761795043945,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 263.95056,
      "step": 241,
      "tokens/total": 544768,
      "tokens/train_per_sec_per_gpu": 42.37,
      "tokens/trainable": 322797
    },
    {
      "epoch": 1.0342979635584137,
      "grad_norm": 0.043023187667131424,
      "learning_rate": 0.0025,
      "loss": 5.4010138511657715,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 221.631,
      "step": 242,
      "tokens/total": 547392,
      "tokens/train_per_sec_per_gpu": 64.52,
      "tokens/trainable": 324497
    },
    {
      "epoch": 1.0385852090032155,
      "grad_norm": 0.10789299011230469,
      "learning_rate": 0.0025,
      "loss": 5.370054244995117,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 214.87452,
      "step": 243,
      "tokens/total": 548736,
      "tokens/train_per_sec_per_gpu": 3.01,
      "tokens/trainable": 324950
    },
    {
      "epoch": 1.0428724544480172,
      "grad_norm": 0.043591853231191635,
      "learning_rate": 0.0025,
      "loss": 5.297659397125244,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.59,
      "memory/max_allocated (GiB)": 18.59,
      "ppl": 199.86845,
      "step": 244,
      "tokens/total": 551872,
      "tokens/train_per_sec_per_gpu": 99.52,
      "tokens/trainable": 327123
    },
    {
      "epoch": 1.0471596998928188,
      "grad_norm": 0.07090502977371216,
      "learning_rate": 0.0025,
      "loss": 5.015192985534668,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 150.68521,
      "step": 245,
      "tokens/total": 554048,
      "tokens/train_per_sec_per_gpu": 4.32,
      "tokens/trainable": 328432
    },
    {
      "epoch": 1.0514469453376205,
      "grad_norm": 0.06152981519699097,
      "learning_rate": 0.0025,
      "loss": 4.667083740234375,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 106.38704,
      "step": 246,
      "tokens/total": 555712,
      "tokens/train_per_sec_per_gpu": 45.68,
      "tokens/trainable": 329309
    },
    {
      "epoch": 1.0557341907824223,
      "grad_norm": 0.06200568005442619,
      "learning_rate": 0.0025,
      "loss": 5.22236967086792,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 185.37294,
      "step": 247,
      "tokens/total": 557632,
      "tokens/train_per_sec_per_gpu": 6.58,
      "tokens/trainable": 330316
    },
    {
      "epoch": 1.060021436227224,
      "grad_norm": 0.0687415823340416,
      "learning_rate": 0.0025,
      "loss": 4.802757740020752,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.59,
      "memory/max_allocated (GiB)": 18.59,
      "ppl": 121.84597,
      "step": 248,
      "tokens/total": 560320,
      "tokens/train_per_sec_per_gpu": 18.48,
      "tokens/trainable": 331985
    },
    {
      "epoch": 1.0643086816720257,
      "grad_norm": 0.07006296515464783,
      "learning_rate": 0.0025,
      "loss": 5.4001383781433105,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 221.43706,
      "step": 249,
      "tokens/total": 562304,
      "tokens/train_per_sec_per_gpu": 15.87,
      "tokens/trainable": 333042
    },
    {
      "epoch": 1.0685959271168275,
      "grad_norm": 0.057536471635103226,
      "learning_rate": 0.0025,
      "loss": 4.920461654663086,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 137.06588,
      "step": 250,
      "tokens/total": 564096,
      "tokens/train_per_sec_per_gpu": 97.41,
      "tokens/trainable": 334009
    },
    {
      "epoch": 1.0728831725616292,
      "grad_norm": 0.2622619867324829,
      "learning_rate": 0.0025,
      "loss": 6.004157543182373,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.38,
      "memory/max_allocated (GiB)": 18.38,
      "ppl": 405.10956,
      "step": 251,
      "tokens/total": 565376,
      "tokens/train_per_sec_per_gpu": 11.03,
      "tokens/trainable": 334450
    },
    {
      "epoch": 1.077170418006431,
      "grad_norm": 0.08943215012550354,
      "learning_rate": 0.0025,
      "loss": 5.477893829345703,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.62,
      "memory/max_allocated (GiB)": 18.62,
      "ppl": 239.34208,
      "step": 252,
      "tokens/total": 567744,
      "tokens/train_per_sec_per_gpu": 28.99,
      "tokens/trainable": 335890
    },
    {
      "epoch": 1.0814576634512325,
      "grad_norm": 0.05865178629755974,
      "learning_rate": 0.0025,
      "loss": 4.780120849609375,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 119.11874,
      "step": 253,
      "tokens/total": 570240,
      "tokens/train_per_sec_per_gpu": 32.08,
      "tokens/trainable": 337404
    },
    {
      "epoch": 1.0857449088960343,
      "grad_norm": 0.08746153116226196,
      "learning_rate": 0.0025,
      "loss": 5.729345798492432,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 307.76786,
      "step": 254,
      "tokens/total": 572416,
      "tokens/train_per_sec_per_gpu": 37.32,
      "tokens/trainable": 338644
    },
    {
      "epoch": 1.090032154340836,
      "grad_norm": 0.0820365622639656,
      "learning_rate": 0.0025,
      "loss": 4.818498611450195,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 123.77911,
      "step": 255,
      "tokens/total": 574144,
      "tokens/train_per_sec_per_gpu": 204.86,
      "tokens/trainable": 339546
    },
    {
      "epoch": 1.0943193997856377,
      "grad_norm": 0.05401737242937088,
      "learning_rate": 0.0025,
      "loss": 4.729460716247559,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 113.23448,
      "step": 256,
      "tokens/total": 576448,
      "tokens/train_per_sec_per_gpu": 62.54,
      "tokens/trainable": 340927
    },
    {
      "epoch": 1.0986066452304395,
      "grad_norm": 0.08213179558515549,
      "learning_rate": 0.0025,
      "loss": 4.536296844482422,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 93.34449,
      "step": 257,
      "tokens/total": 577856,
      "tokens/train_per_sec_per_gpu": 25.55,
      "tokens/trainable": 341479
    },
    {
      "epoch": 1.1028938906752412,
      "grad_norm": 0.09325698018074036,
      "learning_rate": 0.0025,
      "loss": 6.08807897567749,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.58,
      "memory/max_allocated (GiB)": 18.58,
      "ppl": 440.57424,
      "step": 258,
      "tokens/total": 580928,
      "tokens/train_per_sec_per_gpu": 30.01,
      "tokens/trainable": 343497
    },
    {
      "epoch": 1.107181136120043,
      "grad_norm": 0.06686203926801682,
      "learning_rate": 0.0025,
      "loss": 5.119194984436035,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 167.20072,
      "step": 259,
      "tokens/total": 583040,
      "tokens/train_per_sec_per_gpu": 112.68,
      "tokens/trainable": 344661
    },
    {
      "epoch": 1.1114683815648445,
      "grad_norm": 0.08321559429168701,
      "learning_rate": 0.0025,
      "loss": 5.304489612579346,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 201.23827,
      "step": 260,
      "tokens/total": 585472,
      "tokens/train_per_sec_per_gpu": 2.12,
      "tokens/trainable": 346087
    },
    {
      "epoch": 1.1157556270096463,
      "grad_norm": 0.05973471328616142,
      "learning_rate": 0.0025,
      "loss": 5.243081092834473,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.42,
      "memory/max_allocated (GiB)": 18.42,
      "ppl": 189.25231,
      "step": 261,
      "tokens/total": 587520,
      "tokens/train_per_sec_per_gpu": 118.72,
      "tokens/trainable": 347244
    },
    {
      "epoch": 1.120042872454448,
      "grad_norm": 0.06643401831388474,
      "learning_rate": 0.0025,
      "loss": 4.716557025909424,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.6,
      "memory/max_allocated (GiB)": 18.6,
      "ppl": 111.78272,
      "step": 262,
      "tokens/total": 589504,
      "tokens/train_per_sec_per_gpu": 7.82,
      "tokens/trainable": 348354
    },
    {
      "epoch": 1.1243301178992497,
      "grad_norm": 0.05456831306219101,
      "learning_rate": 0.0025,
      "loss": 5.202316761016846,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 181.69269,
      "step": 263,
      "tokens/total": 592448,
      "tokens/train_per_sec_per_gpu": 247.8,
      "tokens/trainable": 350338
    },
    {
      "epoch": 1.1286173633440515,
      "grad_norm": 0.05614905431866646,
      "learning_rate": 0.0025,
      "loss": 4.99215030670166,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 147.25272,
      "step": 264,
      "tokens/total": 594432,
      "tokens/train_per_sec_per_gpu": 2.18,
      "tokens/trainable": 351396
    },
    {
      "epoch": 1.1329046087888532,
      "grad_norm": 0.06128396466374397,
      "learning_rate": 0.0025,
      "loss": 4.919600009918213,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 136.94782,
      "step": 265,
      "tokens/total": 596864,
      "tokens/train_per_sec_per_gpu": 143.33,
      "tokens/trainable": 352905
    },
    {
      "epoch": 1.137191854233655,
      "grad_norm": 0.09635547548532486,
      "learning_rate": 0.0025,
      "loss": 4.9075703620910645,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 135.31026,
      "step": 266,
      "tokens/total": 598784,
      "tokens/train_per_sec_per_gpu": 95.18,
      "tokens/trainable": 353980
    },
    {
      "epoch": 1.1414790996784565,
      "grad_norm": 0.07255875319242477,
      "learning_rate": 0.0025,
      "loss": 5.526827812194824,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 251.34533,
      "step": 267,
      "tokens/total": 600704,
      "tokens/train_per_sec_per_gpu": 208.93,
      "tokens/trainable": 355034
    },
    {
      "epoch": 1.1457663451232583,
      "grad_norm": 0.07017937302589417,
      "learning_rate": 0.0025,
      "loss": 5.108213424682617,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.59,
      "memory/max_allocated (GiB)": 18.59,
      "ppl": 165.37464,
      "step": 268,
      "tokens/total": 603136,
      "tokens/train_per_sec_per_gpu": 169.95,
      "tokens/trainable": 356502
    },
    {
      "epoch": 1.15005359056806,
      "grad_norm": 0.05392616242170334,
      "learning_rate": 0.0025,
      "loss": 4.911001205444336,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 135.77529,
      "step": 269,
      "tokens/total": 605696,
      "tokens/train_per_sec_per_gpu": 98.87,
      "tokens/trainable": 358164
    },
    {
      "epoch": 1.1543408360128617,
      "grad_norm": 0.06459183990955353,
      "learning_rate": 0.0025,
      "loss": 5.657564640045166,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.67,
      "memory/max_allocated (GiB)": 18.67,
      "ppl": 286.45018,
      "step": 270,
      "tokens/total": 608256,
      "tokens/train_per_sec_per_gpu": 34.31,
      "tokens/trainable": 359715
    },
    {
      "epoch": 1.1586280814576635,
      "grad_norm": 0.05815054103732109,
      "learning_rate": 0.0025,
      "loss": 5.023680210113525,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 151.96956,
      "step": 271,
      "tokens/total": 610752,
      "tokens/train_per_sec_per_gpu": 256.08,
      "tokens/trainable": 361290
    },
    {
      "epoch": 1.1629153269024652,
      "grad_norm": 0.08935742825269699,
      "learning_rate": 0.0025,
      "loss": 5.135861396789551,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 170.0107,
      "step": 272,
      "tokens/total": 612480,
      "tokens/train_per_sec_per_gpu": 3.04,
      "tokens/trainable": 362007
    },
    {
      "epoch": 1.167202572347267,
      "grad_norm": 0.07162267714738846,
      "learning_rate": 0.0025,
      "loss": 4.85862922668457,
      "memory/device_reserved (GiB)": 19.05,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 128.84746,
      "step": 273,
      "tokens/total": 614144,
      "tokens/train_per_sec_per_gpu": 10.47,
      "tokens/trainable": 362762
    },
    {
      "epoch": 1.1714898177920685,
      "grad_norm": 0.048264019191265106,
      "learning_rate": 0.0025,
      "loss": 5.584687232971191,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.85,
      "memory/max_allocated (GiB)": 18.85,
      "ppl": 266.31697,
      "step": 274,
      "tokens/total": 617792,
      "tokens/train_per_sec_per_gpu": 37.41,
      "tokens/trainable": 365433
    },
    {
      "epoch": 1.1757770632368703,
      "grad_norm": 0.05624736472964287,
      "learning_rate": 0.0025,
      "loss": 5.056115627288818,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 156.97956,
      "step": 275,
      "tokens/total": 619712,
      "tokens/train_per_sec_per_gpu": 177.31,
      "tokens/trainable": 366496
    },
    {
      "epoch": 1.180064308681672,
      "grad_norm": 0.0889284759759903,
      "learning_rate": 0.0025,
      "loss": 5.753776550292969,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 315.37946,
      "step": 276,
      "tokens/total": 621248,
      "tokens/train_per_sec_per_gpu": 30.13,
      "tokens/trainable": 367083
    },
    {
      "epoch": 1.1843515541264737,
      "grad_norm": 0.07002771645784378,
      "learning_rate": 0.0025,
      "loss": 5.0761637687683105,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 160.15847,
      "step": 277,
      "tokens/total": 623552,
      "tokens/train_per_sec_per_gpu": 124.59,
      "tokens/trainable": 368541
    },
    {
      "epoch": 1.1886387995712755,
      "grad_norm": 0.04624473676085472,
      "learning_rate": 0.0025,
      "loss": 5.2077860832214355,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 182.68915,
      "step": 278,
      "tokens/total": 625600,
      "tokens/train_per_sec_per_gpu": 20.49,
      "tokens/trainable": 369666
    },
    {
      "epoch": 1.1929260450160772,
      "grad_norm": 0.08523814380168915,
      "learning_rate": 0.0025,
      "loss": 4.794674873352051,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 120.86508,
      "step": 279,
      "tokens/total": 627008,
      "tokens/train_per_sec_per_gpu": 6.56,
      "tokens/trainable": 370220
    },
    {
      "epoch": 1.197213290460879,
      "grad_norm": 0.04043371230363846,
      "learning_rate": 0.0025,
      "loss": 4.8553667068481445,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.6,
      "memory/max_allocated (GiB)": 18.6,
      "ppl": 128.42778,
      "step": 280,
      "tokens/total": 629824,
      "tokens/train_per_sec_per_gpu": 277.58,
      "tokens/trainable": 372088
    },
    {
      "epoch": 1.2015005359056805,
      "grad_norm": 0.05826675891876221,
      "learning_rate": 0.0025,
      "loss": 5.2880730628967285,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 197.9616,
      "step": 281,
      "tokens/total": 632128,
      "tokens/train_per_sec_per_gpu": 58.45,
      "tokens/trainable": 373418
    },
    {
      "epoch": 1.2057877813504823,
      "grad_norm": 0.055210962891578674,
      "learning_rate": 0.0025,
      "loss": 5.282061576843262,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 196.77512,
      "step": 282,
      "tokens/total": 634880,
      "tokens/train_per_sec_per_gpu": 43.45,
      "tokens/trainable": 375160
    },
    {
      "epoch": 1.210075026795284,
      "grad_norm": 0.05953294038772583,
      "learning_rate": 0.0025,
      "loss": 4.843015193939209,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 126.85126,
      "step": 283,
      "tokens/total": 637120,
      "tokens/train_per_sec_per_gpu": 11.03,
      "tokens/trainable": 376400
    },
    {
      "epoch": 1.2143622722400857,
      "grad_norm": 0.09921937435865402,
      "learning_rate": 0.0025,
      "loss": 5.3062543869018555,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.68,
      "memory/max_allocated (GiB)": 18.68,
      "ppl": 201.59372,
      "step": 284,
      "tokens/total": 640192,
      "tokens/train_per_sec_per_gpu": 26.81,
      "tokens/trainable": 378440
    },
    {
      "epoch": 1.2186495176848875,
      "grad_norm": 0.07062297314405441,
      "learning_rate": 0.0025,
      "loss": 5.3862786293029785,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 218.38916,
      "step": 285,
      "tokens/total": 642432,
      "tokens/train_per_sec_per_gpu": 106.98,
      "tokens/trainable": 379692
    },
    {
      "epoch": 1.2229367631296892,
      "grad_norm": 0.061749882996082306,
      "learning_rate": 0.0025,
      "loss": 4.912004470825195,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 135.91157,
      "step": 286,
      "tokens/total": 644224,
      "tokens/train_per_sec_per_gpu": 97.43,
      "tokens/trainable": 380455
    },
    {
      "epoch": 1.227224008574491,
      "grad_norm": 0.08968321979045868,
      "learning_rate": 0.0025,
      "loss": 5.269050598144531,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.77,
      "memory/max_allocated (GiB)": 18.77,
      "ppl": 194.23147,
      "step": 287,
      "tokens/total": 647488,
      "tokens/train_per_sec_per_gpu": 11.93,
      "tokens/trainable": 382717
    },
    {
      "epoch": 1.2315112540192925,
      "grad_norm": 0.06253078579902649,
      "learning_rate": 0.0025,
      "loss": 4.804765701293945,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 122.09088,
      "step": 288,
      "tokens/total": 649216,
      "tokens/train_per_sec_per_gpu": 112.0,
      "tokens/trainable": 383607
    },
    {
      "epoch": 1.2357984994640943,
      "grad_norm": 0.07760690897703171,
      "learning_rate": 0.0025,
      "loss": 4.299499034881592,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.66,
      "memory/max_allocated (GiB)": 18.66,
      "ppl": 73.66288,
      "step": 289,
      "tokens/total": 652160,
      "tokens/train_per_sec_per_gpu": 115.01,
      "tokens/trainable": 385609
    },
    {
      "epoch": 1.240085744908896,
      "grad_norm": 0.05732857063412666,
      "learning_rate": 0.0025,
      "loss": 4.712733268737793,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 111.35611,
      "step": 290,
      "tokens/total": 655104,
      "tokens/train_per_sec_per_gpu": 26.12,
      "tokens/trainable": 387600
    },
    {
      "epoch": 1.2443729903536977,
      "grad_norm": 0.08104580640792847,
      "learning_rate": 0.0025,
      "loss": 5.645717620849609,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 283.07663,
      "step": 291,
      "tokens/total": 657600,
      "tokens/train_per_sec_per_gpu": 134.65,
      "tokens/trainable": 389160
    },
    {
      "epoch": 1.2486602357984995,
      "grad_norm": 0.07328224182128906,
      "learning_rate": 0.0025,
      "loss": 4.8165693283081055,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 123.54054,
      "step": 292,
      "tokens/total": 659520,
      "tokens/train_per_sec_per_gpu": 29.84,
      "tokens/trainable": 389993
    },
    {
      "epoch": 1.2529474812433012,
      "grad_norm": 0.05570969730615616,
      "learning_rate": 0.0025,
      "loss": 4.818185806274414,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 123.7404,
      "step": 293,
      "tokens/total": 661824,
      "tokens/train_per_sec_per_gpu": 58.03,
      "tokens/trainable": 391387
    },
    {
      "epoch": 1.257234726688103,
      "grad_norm": 0.06483778357505798,
      "learning_rate": 0.0025,
      "loss": 4.867541313171387,
      "memory/device_reserved (GiB)": 19.21,
      "memory/max_active (GiB)": 18.86,
      "memory/max_allocated (GiB)": 18.86,
      "ppl": 130.00089,
      "step": 294,
      "tokens/total": 664384,
      "tokens/train_per_sec_per_gpu": 58.74,
      "tokens/trainable": 393119
    },
    {
      "epoch": 1.2615219721329045,
      "grad_norm": 0.06346052885055542,
      "learning_rate": 0.0025,
      "loss": 4.831150054931641,
      "memory/device_reserved (GiB)": 19.34,
      "memory/max_active (GiB)": 18.91,
      "memory/max_allocated (GiB)": 18.91,
      "ppl": 125.35504,
      "step": 295,
      "tokens/total": 667456,
      "tokens/train_per_sec_per_gpu": 105.27,
      "tokens/trainable": 395350
    },
    {
      "epoch": 1.2615219721329045,
      "eval_loss": 4.9676361083984375,
      "eval_ppl": 143.68683,
      "eval_runtime": 17.3088,
      "eval_samples_per_second": 12.017,
      "eval_steps_per_second": 12.017,
      "memory/device_reserved (GiB)": 19.34,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "step": 295
    },
    {
      "epoch": 1.2658092175777063,
      "grad_norm": 0.06215736269950867,
      "learning_rate": 0.0025,
      "loss": 5.038169860839844,
      "memory/device_reserved (GiB)": 18.81,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 154.18757,
      "step": 296,
      "tokens/total": 669568,
      "tokens/train_per_sec_per_gpu": 85.6,
      "tokens/trainable": 396670
    },
    {
      "epoch": 1.270096463022508,
      "grad_norm": 0.07183429598808289,
      "learning_rate": 0.0025,
      "loss": 5.635323524475098,
      "memory/device_reserved (GiB)": 18.81,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 280.14954,
      "step": 297,
      "tokens/total": 671360,
      "tokens/train_per_sec_per_gpu": 3.91,
      "tokens/trainable": 397597
    },
    {
      "epoch": 1.2743837084673098,
      "grad_norm": 0.08423589169979095,
      "learning_rate": 0.0025,
      "loss": 4.927524089813232,
      "memory/device_reserved (GiB)": 18.81,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 138.03732,
      "step": 298,
      "tokens/total": 672896,
      "tokens/train_per_sec_per_gpu": 21.59,
      "tokens/trainable": 398137
    },
    {
      "epoch": 1.2786709539121115,
      "grad_norm": 0.050887443125247955,
      "learning_rate": 0.0025,
      "loss": 4.658658981323242,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.67,
      "memory/max_allocated (GiB)": 18.67,
      "ppl": 105.49452,
      "step": 299,
      "tokens/total": 675968,
      "tokens/train_per_sec_per_gpu": 343.04,
      "tokens/trainable": 400317
    },
    {
      "epoch": 1.2829581993569132,
      "grad_norm": 0.0761413648724556,
      "learning_rate": 0.0025,
      "loss": 5.080306529998779,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 160.82335,
      "step": 300,
      "tokens/total": 678400,
      "tokens/train_per_sec_per_gpu": 112.46,
      "tokens/trainable": 401891
    },
    {
      "epoch": 1.287245444801715,
      "grad_norm": 0.08556882292032242,
      "learning_rate": 0.0025,
      "loss": 5.198375225067139,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 180.97795,
      "step": 301,
      "tokens/total": 680576,
      "tokens/train_per_sec_per_gpu": 57.45,
      "tokens/trainable": 403197
    },
    {
      "epoch": 1.2915326902465165,
      "grad_norm": 0.05796833708882332,
      "learning_rate": 0.0025,
      "loss": 5.415544509887695,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 224.87496,
      "step": 302,
      "tokens/total": 682688,
      "tokens/train_per_sec_per_gpu": 111.88,
      "tokens/trainable": 404402
    },
    {
      "epoch": 1.2958199356913183,
      "grad_norm": 0.054447028785943985,
      "learning_rate": 0.0025,
      "loss": 5.261867523193359,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 192.84129,
      "step": 303,
      "tokens/total": 684864,
      "tokens/train_per_sec_per_gpu": 103.18,
      "tokens/trainable": 405715
    },
    {
      "epoch": 1.30010718113612,
      "grad_norm": 0.060531970113515854,
      "learning_rate": 0.0025,
      "loss": 5.523534297943115,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 250.51888,
      "step": 304,
      "tokens/total": 687616,
      "tokens/train_per_sec_per_gpu": 106.29,
      "tokens/trainable": 407456
    },
    {
      "epoch": 1.3043944265809218,
      "grad_norm": 0.050943512469530106,
      "learning_rate": 0.0025,
      "loss": 5.201202392578125,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 181.49033,
      "step": 305,
      "tokens/total": 689984,
      "tokens/train_per_sec_per_gpu": 146.93,
      "tokens/trainable": 408956
    },
    {
      "epoch": 1.3086816720257235,
      "grad_norm": 0.0500502735376358,
      "learning_rate": 0.0025,
      "loss": 4.794355392456055,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.63,
      "memory/max_allocated (GiB)": 18.63,
      "ppl": 120.82647,
      "step": 306,
      "tokens/total": 693440,
      "tokens/train_per_sec_per_gpu": 298.58,
      "tokens/trainable": 411585
    },
    {
      "epoch": 1.3129689174705252,
      "grad_norm": 0.055616557598114014,
      "learning_rate": 0.0025,
      "loss": 5.319107532501221,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 204.20156,
      "step": 307,
      "tokens/total": 695872,
      "tokens/train_per_sec_per_gpu": 88.09,
      "tokens/trainable": 413098
    },
    {
      "epoch": 1.317256162915327,
      "grad_norm": 0.1226491928100586,
      "learning_rate": 0.0025,
      "loss": 4.878961563110352,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 131.49404,
      "step": 308,
      "tokens/total": 697664,
      "tokens/train_per_sec_per_gpu": 14.26,
      "tokens/trainable": 414049
    },
    {
      "epoch": 1.3215434083601285,
      "grad_norm": 0.044849809259176254,
      "learning_rate": 0.0025,
      "loss": 4.39945650100708,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 81.40661,
      "step": 309,
      "tokens/total": 699456,
      "tokens/train_per_sec_per_gpu": 35.77,
      "tokens/trainable": 414939
    },
    {
      "epoch": 1.3258306538049303,
      "grad_norm": 0.10974457114934921,
      "learning_rate": 0.0025,
      "loss": 4.777448654174805,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 118.80086,
      "step": 310,
      "tokens/total": 701312,
      "tokens/train_per_sec_per_gpu": 35.79,
      "tokens/trainable": 415891
    },
    {
      "epoch": 1.330117899249732,
      "grad_norm": 0.06229991093277931,
      "learning_rate": 0.0025,
      "loss": 4.938532829284668,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 139.56533,
      "step": 311,
      "tokens/total": 703744,
      "tokens/train_per_sec_per_gpu": 57.07,
      "tokens/trainable": 417446
    },
    {
      "epoch": 1.3344051446945338,
      "grad_norm": 0.06946682929992676,
      "learning_rate": 0.0025,
      "loss": 5.452759265899658,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 233.40129,
      "step": 312,
      "tokens/total": 706112,
      "tokens/train_per_sec_per_gpu": 213.08,
      "tokens/trainable": 418846
    },
    {
      "epoch": 1.3386923901393355,
      "grad_norm": 0.0820111483335495,
      "learning_rate": 0.0025,
      "loss": 4.500866413116455,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.42,
      "memory/max_allocated (GiB)": 18.42,
      "ppl": 90.09516,
      "step": 313,
      "tokens/total": 707520,
      "tokens/train_per_sec_per_gpu": 99.78,
      "tokens/trainable": 419352
    },
    {
      "epoch": 1.3429796355841372,
      "grad_norm": 0.16722634434700012,
      "learning_rate": 0.0025,
      "loss": 4.281378269195557,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.59,
      "memory/max_allocated (GiB)": 18.59,
      "ppl": 72.34008,
      "step": 314,
      "tokens/total": 709440,
      "tokens/train_per_sec_per_gpu": 15.86,
      "tokens/trainable": 420321
    },
    {
      "epoch": 1.347266881028939,
      "grad_norm": 0.06021692231297493,
      "learning_rate": 0.0025,
      "loss": 5.971747875213623,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 392.19057,
      "step": 315,
      "tokens/total": 711936,
      "tokens/train_per_sec_per_gpu": 77.5,
      "tokens/trainable": 421997
    },
    {
      "epoch": 1.3515541264737405,
      "grad_norm": 0.048991329967975616,
      "learning_rate": 0.0025,
      "loss": 5.36166524887085,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 213.07948,
      "step": 316,
      "tokens/total": 714432,
      "tokens/train_per_sec_per_gpu": 144.42,
      "tokens/trainable": 423421
    },
    {
      "epoch": 1.3558413719185423,
      "grad_norm": 0.11430433392524719,
      "learning_rate": 0.0025,
      "loss": 4.600022792816162,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.36,
      "memory/max_allocated (GiB)": 18.36,
      "ppl": 99.48658,
      "step": 317,
      "tokens/total": 715648,
      "tokens/train_per_sec_per_gpu": 0.87,
      "tokens/trainable": 423767
    },
    {
      "epoch": 1.360128617363344,
      "grad_norm": 0.08438309282064438,
      "learning_rate": 0.0025,
      "loss": 5.151257038116455,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 172.64838,
      "step": 318,
      "tokens/total": 718272,
      "tokens/train_per_sec_per_gpu": 259.31,
      "tokens/trainable": 425467
    },
    {
      "epoch": 1.3644158628081458,
      "grad_norm": 0.056045304983854294,
      "learning_rate": 0.0025,
      "loss": 4.737667560577393,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 114.1676,
      "step": 319,
      "tokens/total": 720384,
      "tokens/train_per_sec_per_gpu": 30.87,
      "tokens/trainable": 426694
    },
    {
      "epoch": 1.3687031082529475,
      "grad_norm": 0.06773889809846878,
      "learning_rate": 0.0025,
      "loss": 5.229560852050781,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.57,
      "memory/max_allocated (GiB)": 18.57,
      "ppl": 186.71079,
      "step": 320,
      "tokens/total": 722688,
      "tokens/train_per_sec_per_gpu": 10.25,
      "tokens/trainable": 428099
    },
    {
      "epoch": 1.3729903536977492,
      "grad_norm": 0.06447850167751312,
      "learning_rate": 0.0025,
      "loss": 5.048516750335693,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.42,
      "memory/max_allocated (GiB)": 18.42,
      "ppl": 155.79122,
      "step": 321,
      "tokens/total": 724736,
      "tokens/train_per_sec_per_gpu": 14.38,
      "tokens/trainable": 429136
    },
    {
      "epoch": 1.377277599142551,
      "grad_norm": 0.07881579548120499,
      "learning_rate": 0.0025,
      "loss": 5.187434673309326,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.57,
      "memory/max_allocated (GiB)": 18.57,
      "ppl": 179.00875,
      "step": 322,
      "tokens/total": 726784,
      "tokens/train_per_sec_per_gpu": 25.14,
      "tokens/trainable": 430417
    },
    {
      "epoch": 1.3815648445873525,
      "grad_norm": 0.06841576844453812,
      "learning_rate": 0.0025,
      "loss": 5.470486640930176,
      "memory/device_reserved (GiB)": 19.0,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 237.57578,
      "step": 323,
      "tokens/total": 728512,
      "tokens/train_per_sec_per_gpu": 156.65,
      "tokens/trainable": 431369
    },
    {
      "epoch": 1.3858520900321543,
      "grad_norm": 0.11785981059074402,
      "learning_rate": 0.0025,
      "loss": 4.845800399780273,
      "memory/device_reserved (GiB)": 19.14,
      "memory/max_active (GiB)": 18.69,
      "memory/max_allocated (GiB)": 18.69,
      "ppl": 127.20506,
      "step": 324,
      "tokens/total": 731200,
      "tokens/train_per_sec_per_gpu": 6.44,
      "tokens/trainable": 433078
    },
    {
      "epoch": 1.390139335476956,
      "grad_norm": 0.07563474774360657,
      "learning_rate": 0.0025,
      "loss": 4.149139404296875,
      "memory/device_reserved (GiB)": 19.14,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 63.37943,
      "step": 325,
      "tokens/total": 733504,
      "tokens/train_per_sec_per_gpu": 77.55,
      "tokens/trainable": 434444
    },
    {
      "epoch": 1.3944265809217578,
      "grad_norm": 0.07862015813589096,
      "learning_rate": 0.0025,
      "loss": 4.405404090881348,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 19.09,
      "memory/max_allocated (GiB)": 19.09,
      "ppl": 81.89223,
      "step": 326,
      "tokens/total": 737088,
      "tokens/train_per_sec_per_gpu": 155.85,
      "tokens/trainable": 437043
    },
    {
      "epoch": 1.3987138263665595,
      "grad_norm": 0.07842207700014114,
      "learning_rate": 0.0025,
      "loss": 5.559260368347168,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 259.63073,
      "step": 327,
      "tokens/total": 738880,
      "tokens/train_per_sec_per_gpu": 24.68,
      "tokens/trainable": 437857
    },
    {
      "epoch": 1.4030010718113612,
      "grad_norm": 0.0685216560959816,
      "learning_rate": 0.0025,
      "loss": 5.114203453063965,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 166.36821,
      "step": 328,
      "tokens/total": 740928,
      "tokens/train_per_sec_per_gpu": 8.77,
      "tokens/trainable": 439031
    },
    {
      "epoch": 1.407288317256163,
      "grad_norm": 0.07267401367425919,
      "learning_rate": 0.0025,
      "loss": 4.98257303237915,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 145.84917,
      "step": 329,
      "tokens/total": 742848,
      "tokens/train_per_sec_per_gpu": 6.98,
      "tokens/trainable": 440007
    },
    {
      "epoch": 1.4115755627009645,
      "grad_norm": 0.0542726069688797,
      "learning_rate": 0.0025,
      "loss": 4.922747611999512,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 137.37956,
      "step": 330,
      "tokens/total": 744960,
      "tokens/train_per_sec_per_gpu": 109.51,
      "tokens/trainable": 441203
    },
    {
      "epoch": 1.4158628081457665,
      "grad_norm": 0.06696043908596039,
      "learning_rate": 0.0025,
      "loss": 4.691984176635742,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 109.06938,
      "step": 331,
      "tokens/total": 747136,
      "tokens/train_per_sec_per_gpu": 129.1,
      "tokens/trainable": 442561
    },
    {
      "epoch": 1.420150053590568,
      "grad_norm": 0.06947220861911774,
      "learning_rate": 0.0025,
      "loss": 4.651721000671387,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 104.76513,
      "step": 332,
      "tokens/total": 748800,
      "tokens/train_per_sec_per_gpu": 85.45,
      "tokens/trainable": 443290
    },
    {
      "epoch": 1.4244372990353698,
      "grad_norm": 0.06656062602996826,
      "learning_rate": 0.0025,
      "loss": 5.165863990783691,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 175.18875,
      "step": 333,
      "tokens/total": 751296,
      "tokens/train_per_sec_per_gpu": 130.0,
      "tokens/trainable": 444808
    },
    {
      "epoch": 1.4287245444801715,
      "grad_norm": 0.0503920316696167,
      "learning_rate": 0.0025,
      "loss": 5.2461981773376465,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.9,
      "memory/max_allocated (GiB)": 18.9,
      "ppl": 189.84314,
      "step": 334,
      "tokens/total": 755072,
      "tokens/train_per_sec_per_gpu": 87.86,
      "tokens/trainable": 447741
    },
    {
      "epoch": 1.4330117899249732,
      "grad_norm": 0.06683284789323807,
      "learning_rate": 0.0025,
      "loss": 4.544827461242676,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 94.14418,
      "step": 335,
      "tokens/total": 756544,
      "tokens/train_per_sec_per_gpu": 5.23,
      "tokens/trainable": 448356
    },
    {
      "epoch": 1.437299035369775,
      "grad_norm": 0.06916282325983047,
      "learning_rate": 0.0025,
      "loss": 5.005459308624268,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 149.22561,
      "step": 336,
      "tokens/total": 759424,
      "tokens/train_per_sec_per_gpu": 82.24,
      "tokens/trainable": 450248
    },
    {
      "epoch": 1.4415862808145765,
      "grad_norm": 0.0864240899682045,
      "learning_rate": 0.0025,
      "loss": 4.898914813995361,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.42,
      "memory/max_allocated (GiB)": 18.42,
      "ppl": 134.14413,
      "step": 337,
      "tokens/total": 761408,
      "tokens/train_per_sec_per_gpu": 25.71,
      "tokens/trainable": 451300
    },
    {
      "epoch": 1.4458735262593785,
      "grad_norm": 0.05566547438502312,
      "learning_rate": 0.0025,
      "loss": 5.002068996429443,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 148.72054,
      "step": 338,
      "tokens/total": 763904,
      "tokens/train_per_sec_per_gpu": 2.62,
      "tokens/trainable": 452916
    },
    {
      "epoch": 1.45016077170418,
      "grad_norm": 0.05582151934504509,
      "learning_rate": 0.0025,
      "loss": 5.314091682434082,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 203.17988,
      "step": 339,
      "tokens/total": 766208,
      "tokens/train_per_sec_per_gpu": 107.08,
      "tokens/trainable": 454272
    },
    {
      "epoch": 1.4544480171489818,
      "grad_norm": 0.05226564779877663,
      "learning_rate": 0.0025,
      "loss": 4.944754600524902,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.42,
      "memory/max_allocated (GiB)": 18.42,
      "ppl": 140.43638,
      "step": 340,
      "tokens/total": 768064,
      "tokens/train_per_sec_per_gpu": 112.77,
      "tokens/trainable": 455346
    },
    {
      "epoch": 1.4587352625937835,
      "grad_norm": 0.06044873222708702,
      "learning_rate": 0.0025,
      "loss": 4.841938018798828,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 126.71469,
      "step": 341,
      "tokens/total": 769664,
      "tokens/train_per_sec_per_gpu": 62.41,
      "tokens/trainable": 456009
    },
    {
      "epoch": 1.4630225080385852,
      "grad_norm": 0.04582054540514946,
      "learning_rate": 0.0025,
      "loss": 5.606667518615723,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 272.2355,
      "step": 342,
      "tokens/total": 772224,
      "tokens/train_per_sec_per_gpu": 167.3,
      "tokens/trainable": 457672
    },
    {
      "epoch": 1.467309753483387,
      "grad_norm": 0.052784983068704605,
      "learning_rate": 0.0025,
      "loss": 4.380001068115234,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 79.83812,
      "step": 343,
      "tokens/total": 774592,
      "tokens/train_per_sec_per_gpu": 64.87,
      "tokens/trainable": 459004
    },
    {
      "epoch": 1.4715969989281885,
      "grad_norm": 0.04827815666794777,
      "learning_rate": 0.0025,
      "loss": 4.817045211791992,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 123.59934,
      "step": 344,
      "tokens/total": 777280,
      "tokens/train_per_sec_per_gpu": 39.6,
      "tokens/trainable": 460657
    },
    {
      "epoch": 1.4758842443729905,
      "grad_norm": 0.07826294749975204,
      "learning_rate": 0.0025,
      "loss": 3.8754422664642334,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 48.20401,
      "step": 345,
      "tokens/total": 778496,
      "tokens/train_per_sec_per_gpu": 23.47,
      "tokens/trainable": 461041
    },
    {
      "epoch": 1.480171489817792,
      "grad_norm": 0.04768767207860947,
      "learning_rate": 0.0025,
      "loss": 5.038158893585205,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 154.18588,
      "step": 346,
      "tokens/total": 780992,
      "tokens/train_per_sec_per_gpu": 78.54,
      "tokens/trainable": 462567
    },
    {
      "epoch": 1.4844587352625938,
      "grad_norm": 0.2351859211921692,
      "learning_rate": 0.0025,
      "loss": 5.444571018218994,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 231.49795,
      "step": 347,
      "tokens/total": 783616,
      "tokens/train_per_sec_per_gpu": 168.35,
      "tokens/trainable": 464210
    },
    {
      "epoch": 1.4887459807073955,
      "grad_norm": 0.04593876376748085,
      "learning_rate": 0.0025,
      "loss": 5.326495170593262,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 205.71571,
      "step": 348,
      "tokens/total": 786304,
      "tokens/train_per_sec_per_gpu": 140.75,
      "tokens/trainable": 466015
    },
    {
      "epoch": 1.4930332261521972,
      "grad_norm": 0.06556063890457153,
      "learning_rate": 0.0025,
      "loss": 5.152454376220703,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 172.85522,
      "step": 349,
      "tokens/total": 788480,
      "tokens/train_per_sec_per_gpu": 224.08,
      "tokens/trainable": 467198
    },
    {
      "epoch": 1.497320471596999,
      "grad_norm": 0.06161191314458847,
      "learning_rate": 0.0025,
      "loss": 5.141845226287842,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 171.03107,
      "step": 350,
      "tokens/total": 790336,
      "tokens/train_per_sec_per_gpu": 74.32,
      "tokens/trainable": 468199
    },
    {
      "epoch": 1.5016077170418005,
      "grad_norm": 0.28069961071014404,
      "learning_rate": 0.0025,
      "loss": 5.9613165855407715,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 388.12078,
      "step": 351,
      "tokens/total": 792320,
      "tokens/train_per_sec_per_gpu": 10.1,
      "tokens/trainable": 469192
    },
    {
      "epoch": 1.5058949624866025,
      "grad_norm": 0.09382814168930054,
      "learning_rate": 0.0025,
      "loss": 5.591994762420654,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 268.27022,
      "step": 352,
      "tokens/total": 794816,
      "tokens/train_per_sec_per_gpu": 35.14,
      "tokens/trainable": 470744
    },
    {
      "epoch": 1.510182207931404,
      "grad_norm": 0.06651383638381958,
      "learning_rate": 0.0025,
      "loss": 4.758882522583008,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 116.61554,
      "step": 353,
      "tokens/total": 796800,
      "tokens/train_per_sec_per_gpu": 46.01,
      "tokens/trainable": 471734
    },
    {
      "epoch": 1.5144694533762058,
      "grad_norm": 0.07342278957366943,
      "learning_rate": 0.0025,
      "loss": 5.06705379486084,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 158.70606,
      "step": 354,
      "tokens/total": 798400,
      "tokens/train_per_sec_per_gpu": 46.96,
      "tokens/trainable": 472368
    },
    {
      "epoch": 1.5144694533762058,
      "eval_loss": 4.888035297393799,
      "eval_ppl": 132.69262,
      "eval_runtime": 17.094,
      "eval_samples_per_second": 12.168,
      "eval_steps_per_second": 12.168,
      "memory/device_reserved (GiB)": 19.64,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "step": 354
    },
    {
      "epoch": 1.5187566988210075,
      "grad_norm": 0.05180167779326439,
      "learning_rate": 0.0025,
      "loss": 4.341151714324951,
      "memory/device_reserved (GiB)": 18.71,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 76.79594,
      "step": 355,
      "tokens/total": 801024,
      "tokens/train_per_sec_per_gpu": 18.76,
      "tokens/trainable": 474170
    },
    {
      "epoch": 1.5230439442658092,
      "grad_norm": 0.05290725454688072,
      "learning_rate": 0.0025,
      "loss": 4.733070373535156,
      "memory/device_reserved (GiB)": 18.74,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 113.64396,
      "step": 356,
      "tokens/total": 803072,
      "tokens/train_per_sec_per_gpu": 10.8,
      "tokens/trainable": 475255
    },
    {
      "epoch": 1.527331189710611,
      "grad_norm": 0.08694328367710114,
      "learning_rate": 0.0025,
      "loss": 4.449097156524658,
      "memory/device_reserved (GiB)": 18.74,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 85.54967,
      "step": 357,
      "tokens/total": 804864,
      "tokens/train_per_sec_per_gpu": 36.39,
      "tokens/trainable": 476172
    },
    {
      "epoch": 1.5316184351554125,
      "grad_norm": 0.059812407940626144,
      "learning_rate": 0.0025,
      "loss": 5.0324225425720215,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 19.55,
      "memory/max_allocated (GiB)": 19.55,
      "ppl": 153.30395,
      "step": 358,
      "tokens/total": 809344,
      "tokens/train_per_sec_per_gpu": 62.57,
      "tokens/trainable": 479719
    },
    {
      "epoch": 1.5359056806002145,
      "grad_norm": 0.05654964968562126,
      "learning_rate": 0.0025,
      "loss": 5.156862735748291,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 19.12,
      "memory/max_allocated (GiB)": 19.12,
      "ppl": 173.61891,
      "step": 359,
      "tokens/total": 813440,
      "tokens/train_per_sec_per_gpu": 30.77,
      "tokens/trainable": 482873
    },
    {
      "epoch": 1.540192926045016,
      "grad_norm": 0.04540662467479706,
      "learning_rate": 0.0025,
      "loss": 4.702476501464844,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 110.21979,
      "step": 360,
      "tokens/total": 815424,
      "tokens/train_per_sec_per_gpu": 10.68,
      "tokens/trainable": 483879
    },
    {
      "epoch": 1.5444801714898178,
      "grad_norm": 0.08061626553535461,
      "learning_rate": 0.0025,
      "loss": 4.724915504455566,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 19.5,
      "memory/max_allocated (GiB)": 19.5,
      "ppl": 112.72097,
      "step": 361,
      "tokens/total": 819520,
      "tokens/train_per_sec_per_gpu": 7.14,
      "tokens/trainable": 486988
    },
    {
      "epoch": 1.5487674169346195,
      "grad_norm": 0.047675181180238724,
      "learning_rate": 0.0025,
      "loss": 4.952810764312744,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 141.57233,
      "step": 362,
      "tokens/total": 822528,
      "tokens/train_per_sec_per_gpu": 188.99,
      "tokens/trainable": 489093
    },
    {
      "epoch": 1.5530546623794212,
      "grad_norm": 0.0705411285161972,
      "learning_rate": 0.0025,
      "loss": 4.235616683959961,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.38,
      "memory/max_allocated (GiB)": 18.38,
      "ppl": 69.10428,
      "step": 363,
      "tokens/total": 824000,
      "tokens/train_per_sec_per_gpu": 40.32,
      "tokens/trainable": 489788
    },
    {
      "epoch": 1.557341907824223,
      "grad_norm": 0.0532984621822834,
      "learning_rate": 0.0025,
      "loss": 4.592817306518555,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 98.77231,
      "step": 364,
      "tokens/total": 825984,
      "tokens/train_per_sec_per_gpu": 126.04,
      "tokens/trainable": 490775
    },
    {
      "epoch": 1.5616291532690245,
      "grad_norm": 0.05280710384249687,
      "learning_rate": 0.0025,
      "loss": 5.201740264892578,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 181.58798,
      "step": 365,
      "tokens/total": 828096,
      "tokens/train_per_sec_per_gpu": 54.56,
      "tokens/trainable": 492005
    },
    {
      "epoch": 1.5659163987138265,
      "grad_norm": 0.051445771008729935,
      "learning_rate": 0.0025,
      "loss": 4.719823360443115,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 112.14844,
      "step": 366,
      "tokens/total": 830720,
      "tokens/train_per_sec_per_gpu": 197.49,
      "tokens/trainable": 493565
    },
    {
      "epoch": 1.570203644158628,
      "grad_norm": 0.05346055328845978,
      "learning_rate": 0.0025,
      "loss": 4.971454620361328,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 144.23654,
      "step": 367,
      "tokens/total": 833216,
      "tokens/train_per_sec_per_gpu": 12.08,
      "tokens/trainable": 495164
    },
    {
      "epoch": 1.5744908896034298,
      "grad_norm": 0.0699196457862854,
      "learning_rate": 0.0025,
      "loss": 4.878042221069336,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 131.37321,
      "step": 368,
      "tokens/total": 835520,
      "tokens/train_per_sec_per_gpu": 10.86,
      "tokens/trainable": 496515
    },
    {
      "epoch": 1.5787781350482315,
      "grad_norm": 0.06464764475822449,
      "learning_rate": 0.0025,
      "loss": 4.946186542510986,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 140.63762,
      "step": 369,
      "tokens/total": 837696,
      "tokens/train_per_sec_per_gpu": 199.16,
      "tokens/trainable": 497713
    },
    {
      "epoch": 1.5830653804930332,
      "grad_norm": 0.08636437356472015,
      "learning_rate": 0.0025,
      "loss": 4.280492305755615,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 72.27601,
      "step": 370,
      "tokens/total": 839168,
      "tokens/train_per_sec_per_gpu": 71.64,
      "tokens/trainable": 498229
    },
    {
      "epoch": 1.587352625937835,
      "grad_norm": 0.047297775745391846,
      "learning_rate": 0.0025,
      "loss": 5.235410213470459,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 187.80613,
      "step": 371,
      "tokens/total": 842112,
      "tokens/train_per_sec_per_gpu": 170.73,
      "tokens/trainable": 500237
    },
    {
      "epoch": 1.5916398713826365,
      "grad_norm": 0.07992982119321823,
      "learning_rate": 0.0025,
      "loss": 4.507941246032715,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 90.73483,
      "step": 372,
      "tokens/total": 843584,
      "tokens/train_per_sec_per_gpu": 11.01,
      "tokens/trainable": 500839
    },
    {
      "epoch": 1.5959271168274385,
      "grad_norm": 0.06259352713823318,
      "learning_rate": 0.0025,
      "loss": 4.919366836547852,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 136.9159,
      "step": 373,
      "tokens/total": 845504,
      "tokens/train_per_sec_per_gpu": 41.51,
      "tokens/trainable": 501881
    },
    {
      "epoch": 1.60021436227224,
      "grad_norm": 0.08243437856435776,
      "learning_rate": 0.0025,
      "loss": 5.151687145233154,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.58,
      "memory/max_allocated (GiB)": 18.58,
      "ppl": 172.72265,
      "step": 374,
      "tokens/total": 848768,
      "tokens/train_per_sec_per_gpu": 261.87,
      "tokens/trainable": 504139
    },
    {
      "epoch": 1.6045016077170418,
      "grad_norm": 0.08473316580057144,
      "learning_rate": 0.0025,
      "loss": 5.018991470336914,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 151.25868,
      "step": 375,
      "tokens/total": 851136,
      "tokens/train_per_sec_per_gpu": 4.3,
      "tokens/trainable": 505536
    },
    {
      "epoch": 1.6087888531618435,
      "grad_norm": 0.04689257964491844,
      "learning_rate": 0.0025,
      "loss": 4.897285461425781,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 133.92574,
      "step": 376,
      "tokens/total": 853504,
      "tokens/train_per_sec_per_gpu": 29.13,
      "tokens/trainable": 507033
    },
    {
      "epoch": 1.6130760986066452,
      "grad_norm": 0.058138296008110046,
      "learning_rate": 0.0025,
      "loss": 4.437278747558594,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 84.54456,
      "step": 377,
      "tokens/total": 855040,
      "tokens/train_per_sec_per_gpu": 60.77,
      "tokens/trainable": 507780
    },
    {
      "epoch": 1.617363344051447,
      "grad_norm": 0.07955910265445709,
      "learning_rate": 0.0025,
      "loss": 4.341729640960693,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.53,
      "memory/max_allocated (GiB)": 18.53,
      "ppl": 76.84033,
      "step": 378,
      "tokens/total": 857024,
      "tokens/train_per_sec_per_gpu": 1.3,
      "tokens/trainable": 508911
    },
    {
      "epoch": 1.6216505894962485,
      "grad_norm": 0.057746682316064835,
      "learning_rate": 0.0025,
      "loss": 3.7874722480773926,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 44.14467,
      "step": 379,
      "tokens/total": 858752,
      "tokens/train_per_sec_per_gpu": 27.9,
      "tokens/trainable": 509683
    },
    {
      "epoch": 1.6259378349410505,
      "grad_norm": 0.07005178928375244,
      "learning_rate": 0.0025,
      "loss": 5.148179531097412,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 172.11787,
      "step": 380,
      "tokens/total": 860928,
      "tokens/train_per_sec_per_gpu": 56.8,
      "tokens/trainable": 510959
    },
    {
      "epoch": 1.630225080385852,
      "grad_norm": 0.04911843314766884,
      "learning_rate": 0.0025,
      "loss": 4.848702430725098,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 127.57475,
      "step": 381,
      "tokens/total": 862976,
      "tokens/train_per_sec_per_gpu": 93.02,
      "tokens/trainable": 512078
    },
    {
      "epoch": 1.6345123258306538,
      "grad_norm": 0.07803714275360107,
      "learning_rate": 0.0025,
      "loss": 4.658711910247803,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 105.5001,
      "step": 382,
      "tokens/total": 865088,
      "tokens/train_per_sec_per_gpu": 67.71,
      "tokens/trainable": 513209
    },
    {
      "epoch": 1.6387995712754555,
      "grad_norm": 0.039557769894599915,
      "learning_rate": 0.0025,
      "loss": 4.664157867431641,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 106.07622,
      "step": 383,
      "tokens/total": 867520,
      "tokens/train_per_sec_per_gpu": 144.15,
      "tokens/trainable": 514700
    },
    {
      "epoch": 1.6430868167202572,
      "grad_norm": 0.047754038125276566,
      "learning_rate": 0.0025,
      "loss": 4.329286575317383,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.64,
      "memory/max_allocated (GiB)": 18.64,
      "ppl": 75.89013,
      "step": 384,
      "tokens/total": 869632,
      "tokens/train_per_sec_per_gpu": 32.67,
      "tokens/trainable": 516045
    },
    {
      "epoch": 1.647374062165059,
      "grad_norm": 0.0593026764690876,
      "learning_rate": 0.0025,
      "loss": 5.5122809410095215,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.55,
      "memory/max_allocated (GiB)": 18.55,
      "ppl": 247.71551,
      "step": 385,
      "tokens/total": 872128,
      "tokens/train_per_sec_per_gpu": 6.48,
      "tokens/trainable": 517536
    },
    {
      "epoch": 1.6516613076098605,
      "grad_norm": 0.0731717124581337,
      "learning_rate": 0.0025,
      "loss": 4.586673736572266,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 98.16736,
      "step": 386,
      "tokens/total": 874240,
      "tokens/train_per_sec_per_gpu": 15.25,
      "tokens/trainable": 518738
    },
    {
      "epoch": 1.6559485530546625,
      "grad_norm": 0.049605630338191986,
      "learning_rate": 0.0025,
      "loss": 4.679656982421875,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 107.73311,
      "step": 387,
      "tokens/total": 876288,
      "tokens/train_per_sec_per_gpu": 135.43,
      "tokens/trainable": 519876
    },
    {
      "epoch": 1.660235798499464,
      "grad_norm": 0.05317911505699158,
      "learning_rate": 0.0025,
      "loss": 4.681480407714844,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.73,
      "memory/max_allocated (GiB)": 18.73,
      "ppl": 107.92973,
      "step": 388,
      "tokens/total": 879360,
      "tokens/train_per_sec_per_gpu": 368.87,
      "tokens/trainable": 521966
    },
    {
      "epoch": 1.6645230439442658,
      "grad_norm": 0.060831792652606964,
      "learning_rate": 0.0025,
      "loss": 5.22551155090332,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 185.95627,
      "step": 389,
      "tokens/total": 881664,
      "tokens/train_per_sec_per_gpu": 7.84,
      "tokens/trainable": 523262
    },
    {
      "epoch": 1.6688102893890675,
      "grad_norm": 0.10634256899356842,
      "learning_rate": 0.0025,
      "loss": 5.569226264953613,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 262.23112,
      "step": 390,
      "tokens/total": 883200,
      "tokens/train_per_sec_per_gpu": 7.0,
      "tokens/trainable": 523927
    },
    {
      "epoch": 1.6730975348338692,
      "grad_norm": 0.04912353307008743,
      "learning_rate": 0.0025,
      "loss": 4.896402359008789,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 133.80752,
      "step": 391,
      "tokens/total": 885376,
      "tokens/train_per_sec_per_gpu": 131.57,
      "tokens/trainable": 525245
    },
    {
      "epoch": 1.677384780278671,
      "grad_norm": 0.051567140966653824,
      "learning_rate": 0.0025,
      "loss": 4.537640571594238,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 93.47,
      "step": 392,
      "tokens/total": 887680,
      "tokens/train_per_sec_per_gpu": 23.66,
      "tokens/trainable": 526583
    },
    {
      "epoch": 1.6816720257234725,
      "grad_norm": 0.05488206073641777,
      "learning_rate": 0.0025,
      "loss": 4.715126037597656,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 111.62288,
      "step": 393,
      "tokens/total": 890304,
      "tokens/train_per_sec_per_gpu": 140.1,
      "tokens/trainable": 528304
    },
    {
      "epoch": 1.6859592711682745,
      "grad_norm": 0.04531604424118996,
      "learning_rate": 0.0025,
      "loss": 4.594015121459961,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 98.89069,
      "step": 394,
      "tokens/total": 892416,
      "tokens/train_per_sec_per_gpu": 63.67,
      "tokens/trainable": 529449
    },
    {
      "epoch": 1.690246516613076,
      "grad_norm": 0.054829467087984085,
      "learning_rate": 0.0025,
      "loss": 4.549041271209717,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 94.54172,
      "step": 395,
      "tokens/total": 894144,
      "tokens/train_per_sec_per_gpu": 30.68,
      "tokens/trainable": 530299
    },
    {
      "epoch": 1.694533762057878,
      "grad_norm": 0.053975410759449005,
      "learning_rate": 0.0025,
      "loss": 4.356207370758057,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.59,
      "memory/max_allocated (GiB)": 18.59,
      "ppl": 77.9609,
      "step": 396,
      "tokens/total": 896192,
      "tokens/train_per_sec_per_gpu": 23.37,
      "tokens/trainable": 531383
    },
    {
      "epoch": 1.6988210075026795,
      "grad_norm": 0.06466397643089294,
      "learning_rate": 0.0025,
      "loss": 4.832691192626953,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 125.54838,
      "step": 397,
      "tokens/total": 898368,
      "tokens/train_per_sec_per_gpu": 28.95,
      "tokens/trainable": 532533
    },
    {
      "epoch": 1.7031082529474812,
      "grad_norm": 0.0506359338760376,
      "learning_rate": 0.0025,
      "loss": 4.359593868255615,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 78.22536,
      "step": 398,
      "tokens/total": 900736,
      "tokens/train_per_sec_per_gpu": 7.39,
      "tokens/trainable": 533957
    },
    {
      "epoch": 1.707395498392283,
      "grad_norm": 0.06138148903846741,
      "learning_rate": 0.0025,
      "loss": 4.625148296356201,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 102.0179,
      "step": 399,
      "tokens/total": 902592,
      "tokens/train_per_sec_per_gpu": 20.91,
      "tokens/trainable": 534928
    },
    {
      "epoch": 1.7116827438370845,
      "grad_norm": 0.047848962247371674,
      "learning_rate": 0.0025,
      "loss": 4.433683395385742,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 84.24114,
      "step": 400,
      "tokens/total": 904448,
      "tokens/train_per_sec_per_gpu": 18.74,
      "tokens/trainable": 535890
    },
    {
      "epoch": 1.7159699892818865,
      "grad_norm": 0.06819909065961838,
      "learning_rate": 0.0025,
      "loss": 4.891788482666016,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 133.19157,
      "step": 401,
      "tokens/total": 906048,
      "tokens/train_per_sec_per_gpu": 52.8,
      "tokens/trainable": 536499
    },
    {
      "epoch": 1.720257234726688,
      "grad_norm": 0.04256964474916458,
      "learning_rate": 0.0025,
      "loss": 4.25316858291626,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 70.3279,
      "step": 402,
      "tokens/total": 907968,
      "tokens/train_per_sec_per_gpu": 121.28,
      "tokens/trainable": 537519
    },
    {
      "epoch": 1.72454448017149,
      "grad_norm": 0.04542100802063942,
      "learning_rate": 0.0025,
      "loss": 5.214837551116943,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.52,
      "memory/max_allocated (GiB)": 18.52,
      "ppl": 183.98193,
      "step": 403,
      "tokens/total": 910848,
      "tokens/train_per_sec_per_gpu": 233.51,
      "tokens/trainable": 539428
    },
    {
      "epoch": 1.7288317256162915,
      "grad_norm": 0.04585760459303856,
      "learning_rate": 0.0025,
      "loss": 4.585484027862549,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.53,
      "memory/max_allocated (GiB)": 18.53,
      "ppl": 98.05063,
      "step": 404,
      "tokens/total": 913536,
      "tokens/train_per_sec_per_gpu": 183.23,
      "tokens/trainable": 541209
    },
    {
      "epoch": 1.7331189710610932,
      "grad_norm": 0.0482996366918087,
      "learning_rate": 0.0025,
      "loss": 4.8759870529174805,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.63,
      "memory/max_allocated (GiB)": 18.63,
      "ppl": 131.1035,
      "step": 405,
      "tokens/total": 916032,
      "tokens/train_per_sec_per_gpu": 65.34,
      "tokens/trainable": 542856
    },
    {
      "epoch": 1.737406216505895,
      "grad_norm": 0.04029145836830139,
      "learning_rate": 0.0025,
      "loss": 4.391783714294434,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.4,
      "memory/max_allocated (GiB)": 18.4,
      "ppl": 80.78439,
      "step": 406,
      "tokens/total": 917824,
      "tokens/train_per_sec_per_gpu": 6.99,
      "tokens/trainable": 543743
    },
    {
      "epoch": 1.7416934619506965,
      "grad_norm": 0.03549795225262642,
      "learning_rate": 0.0025,
      "loss": 4.487819671630859,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 88.92734,
      "step": 407,
      "tokens/total": 920192,
      "tokens/train_per_sec_per_gpu": 142.0,
      "tokens/trainable": 545102
    },
    {
      "epoch": 1.7459807073954985,
      "grad_norm": 0.05987889692187309,
      "learning_rate": 0.0025,
      "loss": 4.304838180541992,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 74.05723,
      "step": 408,
      "tokens/total": 921920,
      "tokens/train_per_sec_per_gpu": 80.61,
      "tokens/trainable": 545864
    },
    {
      "epoch": 1.7502679528403,
      "grad_norm": 0.03903573006391525,
      "learning_rate": 0.0025,
      "loss": 4.431785583496094,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 84.08142,
      "step": 409,
      "tokens/total": 924160,
      "tokens/train_per_sec_per_gpu": 56.37,
      "tokens/trainable": 547197
    },
    {
      "epoch": 1.754555198285102,
      "grad_norm": 0.08176471292972565,
      "learning_rate": 0.0025,
      "loss": 5.05267333984375,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 156.44012,
      "step": 410,
      "tokens/total": 926400,
      "tokens/train_per_sec_per_gpu": 78.56,
      "tokens/trainable": 548428
    },
    {
      "epoch": 1.7588424437299035,
      "grad_norm": 0.06569211184978485,
      "learning_rate": 0.0025,
      "loss": 4.557641506195068,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 95.35831,
      "step": 411,
      "tokens/total": 928256,
      "tokens/train_per_sec_per_gpu": 69.11,
      "tokens/trainable": 549334
    },
    {
      "epoch": 1.7631296891747053,
      "grad_norm": 0.0812261626124382,
      "learning_rate": 0.0025,
      "loss": 4.661388397216797,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.72,
      "memory/max_allocated (GiB)": 18.72,
      "ppl": 105.78285,
      "step": 412,
      "tokens/total": 930432,
      "tokens/train_per_sec_per_gpu": 30.49,
      "tokens/trainable": 550597
    },
    {
      "epoch": 1.767416934619507,
      "grad_norm": 0.06816331297159195,
      "learning_rate": 0.0025,
      "loss": 4.940434455871582,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 139.83099,
      "step": 413,
      "tokens/total": 932672,
      "tokens/train_per_sec_per_gpu": 46.78,
      "tokens/trainable": 551971
    },
    {
      "epoch": 1.767416934619507,
      "eval_loss": 4.7477898597717285,
      "eval_ppl": 115.32911,
      "eval_runtime": 16.9681,
      "eval_samples_per_second": 12.258,
      "eval_steps_per_second": 12.258,
      "memory/device_reserved (GiB)": 19.92,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "step": 413
    },
    {
      "epoch": 1.7717041800643085,
      "grad_norm": 0.04632922261953354,
      "learning_rate": 0.0025,
      "loss": 4.674541473388672,
      "memory/device_reserved (GiB)": 18.75,
      "memory/max_active (GiB)": 18.53,
      "memory/max_allocated (GiB)": 18.53,
      "ppl": 107.18341,
      "step": 414,
      "tokens/total": 934720,
      "tokens/train_per_sec_per_gpu": 21.87,
      "tokens/trainable": 553178
    },
    {
      "epoch": 1.7759914255091105,
      "grad_norm": 0.0418037474155426,
      "learning_rate": 0.0025,
      "loss": 4.476470947265625,
      "memory/device_reserved (GiB)": 18.75,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 87.92384,
      "step": 415,
      "tokens/total": 936768,
      "tokens/train_per_sec_per_gpu": 44.4,
      "tokens/trainable": 554330
    },
    {
      "epoch": 1.780278670953912,
      "grad_norm": 0.0720926821231842,
      "learning_rate": 0.0025,
      "loss": 4.901095390319824,
      "memory/device_reserved (GiB)": 18.75,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 134.43696,
      "step": 416,
      "tokens/total": 938752,
      "tokens/train_per_sec_per_gpu": 58.68,
      "tokens/trainable": 555375
    },
    {
      "epoch": 1.784565916398714,
      "grad_norm": 0.0682898610830307,
      "learning_rate": 0.0025,
      "loss": 4.806495189666748,
      "memory/device_reserved (GiB)": 18.75,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 122.30222,
      "step": 417,
      "tokens/total": 940736,
      "tokens/train_per_sec_per_gpu": 60.27,
      "tokens/trainable": 556410
    },
    {
      "epoch": 1.7888531618435155,
      "grad_norm": 0.05859844386577606,
      "learning_rate": 0.0025,
      "loss": 4.619528293609619,
      "memory/device_reserved (GiB)": 18.75,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 101.44617,
      "step": 418,
      "tokens/total": 942720,
      "tokens/train_per_sec_per_gpu": 119.79,
      "tokens/trainable": 557481
    },
    {
      "epoch": 1.7931404072883173,
      "grad_norm": 0.0587584562599659,
      "learning_rate": 0.0025,
      "loss": 4.6873064041137695,
      "memory/device_reserved (GiB)": 18.75,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 108.56037,
      "step": 419,
      "tokens/total": 944576,
      "tokens/train_per_sec_per_gpu": 103.57,
      "tokens/trainable": 558411
    },
    {
      "epoch": 1.797427652733119,
      "grad_norm": 0.048332419246435165,
      "learning_rate": 0.0025,
      "loss": 4.776644706726074,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 19.42,
      "memory/max_allocated (GiB)": 19.42,
      "ppl": 118.70539,
      "step": 420,
      "tokens/total": 948160,
      "tokens/train_per_sec_per_gpu": 18.63,
      "tokens/trainable": 561055
    },
    {
      "epoch": 1.8017148981779205,
      "grad_norm": 0.045766137540340424,
      "learning_rate": 0.0025,
      "loss": 4.8820977210998535,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 131.90708,
      "step": 421,
      "tokens/total": 950784,
      "tokens/train_per_sec_per_gpu": 58.05,
      "tokens/trainable": 562707
    },
    {
      "epoch": 1.8060021436227225,
      "grad_norm": 0.07353589683771133,
      "learning_rate": 0.0025,
      "loss": 4.372645854949951,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 79.25305,
      "step": 422,
      "tokens/total": 952384,
      "tokens/train_per_sec_per_gpu": 94.96,
      "tokens/trainable": 563370
    },
    {
      "epoch": 1.810289389067524,
      "grad_norm": 0.057513438165187836,
      "learning_rate": 0.0025,
      "loss": 4.444611549377441,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 85.16679,
      "step": 423,
      "tokens/total": 954560,
      "tokens/train_per_sec_per_gpu": 14.88,
      "tokens/trainable": 564680
    },
    {
      "epoch": 1.814576634512326,
      "grad_norm": 0.054809462279081345,
      "learning_rate": 0.0025,
      "loss": 4.237483501434326,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 69.23341,
      "step": 424,
      "tokens/total": 957696,
      "tokens/train_per_sec_per_gpu": 52.44,
      "tokens/trainable": 566858
    },
    {
      "epoch": 1.8188638799571275,
      "grad_norm": 0.042437877506017685,
      "learning_rate": 0.0025,
      "loss": 4.65795373916626,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 105.42014,
      "step": 425,
      "tokens/total": 960064,
      "tokens/train_per_sec_per_gpu": 58.49,
      "tokens/trainable": 568277
    },
    {
      "epoch": 1.8231511254019293,
      "grad_norm": 0.0690232664346695,
      "learning_rate": 0.0025,
      "loss": 4.57647180557251,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 19.1,
      "memory/max_allocated (GiB)": 19.1,
      "ppl": 97.17095,
      "step": 426,
      "tokens/total": 963264,
      "tokens/train_per_sec_per_gpu": 604.98,
      "tokens/trainable": 570607
    },
    {
      "epoch": 1.827438370846731,
      "grad_norm": 0.05784786492586136,
      "learning_rate": 0.0025,
      "loss": 5.262682914733887,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 192.9986,
      "step": 427,
      "tokens/total": 965824,
      "tokens/train_per_sec_per_gpu": 120.41,
      "tokens/trainable": 572251
    },
    {
      "epoch": 1.8317256162915327,
      "grad_norm": 0.07479379326105118,
      "learning_rate": 0.0025,
      "loss": 5.224900722503662,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 185.84272,
      "step": 428,
      "tokens/total": 968064,
      "tokens/train_per_sec_per_gpu": 7.22,
      "tokens/trainable": 573467
    },
    {
      "epoch": 1.8360128617363345,
      "grad_norm": 0.048712894320487976,
      "learning_rate": 0.0025,
      "loss": 5.22140645980835,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 185.19447,
      "step": 429,
      "tokens/total": 970432,
      "tokens/train_per_sec_per_gpu": 17.21,
      "tokens/trainable": 574765
    },
    {
      "epoch": 1.840300107181136,
      "grad_norm": 0.0432349257171154,
      "learning_rate": 0.0025,
      "loss": 4.933577060699463,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 138.87539,
      "step": 430,
      "tokens/total": 972928,
      "tokens/train_per_sec_per_gpu": 126.94,
      "tokens/trainable": 576351
    },
    {
      "epoch": 1.844587352625938,
      "grad_norm": 0.035638924688100815,
      "learning_rate": 0.0025,
      "loss": 5.010141372680664,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.5,
      "memory/max_allocated (GiB)": 18.5,
      "ppl": 149.92593,
      "step": 431,
      "tokens/total": 975616,
      "tokens/train_per_sec_per_gpu": 55.33,
      "tokens/trainable": 578016
    },
    {
      "epoch": 1.8488745980707395,
      "grad_norm": 0.05865227058529854,
      "learning_rate": 0.0025,
      "loss": 4.759288787841797,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.47,
      "memory/max_allocated (GiB)": 18.47,
      "ppl": 116.66292,
      "step": 432,
      "tokens/total": 977344,
      "tokens/train_per_sec_per_gpu": 17.53,
      "tokens/trainable": 578907
    },
    {
      "epoch": 1.8531618435155413,
      "grad_norm": 0.037786636501550674,
      "learning_rate": 0.0025,
      "loss": 4.746092796325684,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.56,
      "memory/max_allocated (GiB)": 18.56,
      "ppl": 115.13355,
      "step": 433,
      "tokens/total": 980800,
      "tokens/train_per_sec_per_gpu": 37.5,
      "tokens/trainable": 581317
    },
    {
      "epoch": 1.857449088960343,
      "grad_norm": 0.036064039915800095,
      "learning_rate": 0.0025,
      "loss": 4.497668743133545,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 89.80752,
      "step": 434,
      "tokens/total": 982976,
      "tokens/train_per_sec_per_gpu": 41.65,
      "tokens/trainable": 582503
    },
    {
      "epoch": 1.8617363344051447,
      "grad_norm": 0.059433262795209885,
      "learning_rate": 0.0025,
      "loss": 5.0172624588012695,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.55,
      "memory/max_allocated (GiB)": 18.55,
      "ppl": 150.99738,
      "step": 435,
      "tokens/total": 985600,
      "tokens/train_per_sec_per_gpu": 18.07,
      "tokens/trainable": 584121
    },
    {
      "epoch": 1.8660235798499465,
      "grad_norm": 0.05385487526655197,
      "learning_rate": 0.0025,
      "loss": 5.047094821929932,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.71,
      "memory/max_allocated (GiB)": 18.71,
      "ppl": 155.56985,
      "step": 436,
      "tokens/total": 988352,
      "tokens/train_per_sec_per_gpu": 21.47,
      "tokens/trainable": 585931
    },
    {
      "epoch": 1.870310825294748,
      "grad_norm": 0.07723033428192139,
      "learning_rate": 0.0025,
      "loss": 5.156320571899414,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.53,
      "memory/max_allocated (GiB)": 18.53,
      "ppl": 173.52481,
      "step": 437,
      "tokens/total": 991104,
      "tokens/train_per_sec_per_gpu": 210.83,
      "tokens/trainable": 587715
    },
    {
      "epoch": 1.87459807073955,
      "grad_norm": 0.054785728454589844,
      "learning_rate": 0.0025,
      "loss": 4.405440330505371,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.36,
      "memory/max_allocated (GiB)": 18.36,
      "ppl": 81.8952,
      "step": 438,
      "tokens/total": 992448,
      "tokens/train_per_sec_per_gpu": 34.58,
      "tokens/trainable": 588195
    },
    {
      "epoch": 1.8788853161843515,
      "grad_norm": 0.044970739632844925,
      "learning_rate": 0.0025,
      "loss": 4.495140075683594,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 89.58072,
      "step": 439,
      "tokens/total": 995008,
      "tokens/train_per_sec_per_gpu": 10.07,
      "tokens/trainable": 589904
    },
    {
      "epoch": 1.8831725616291533,
      "grad_norm": 0.05420251190662384,
      "learning_rate": 0.0025,
      "loss": 4.917167663574219,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.63,
      "memory/max_allocated (GiB)": 18.63,
      "ppl": 136.61512,
      "step": 440,
      "tokens/total": 997568,
      "tokens/train_per_sec_per_gpu": 58.61,
      "tokens/trainable": 591429
    },
    {
      "epoch": 1.887459807073955,
      "grad_norm": 0.06885336339473724,
      "learning_rate": 0.0025,
      "loss": 4.854959011077881,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.64,
      "memory/max_allocated (GiB)": 18.64,
      "ppl": 128.37543,
      "step": 441,
      "tokens/total": 1000000,
      "tokens/train_per_sec_per_gpu": 2.19,
      "tokens/trainable": 592888
    },
    {
      "epoch": 1.8917470525187567,
      "grad_norm": 0.04857528582215309,
      "learning_rate": 0.0025,
      "loss": 4.567473888397217,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.67,
      "memory/max_allocated (GiB)": 18.67,
      "ppl": 96.30054,
      "step": 442,
      "tokens/total": 1002496,
      "tokens/train_per_sec_per_gpu": 28.11,
      "tokens/trainable": 594412
    },
    {
      "epoch": 1.8960342979635585,
      "grad_norm": 0.05679011344909668,
      "learning_rate": 0.0025,
      "loss": 4.383760452270508,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 80.13883,
      "step": 443,
      "tokens/total": 1004032,
      "tokens/train_per_sec_per_gpu": 7.89,
      "tokens/trainable": 595042
    },
    {
      "epoch": 1.90032154340836,
      "grad_norm": 0.08521363139152527,
      "learning_rate": 0.0025,
      "loss": 5.441523551940918,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 230.79354,
      "step": 444,
      "tokens/total": 1005824,
      "tokens/train_per_sec_per_gpu": 182.26,
      "tokens/trainable": 595958
    },
    {
      "epoch": 1.904608788853162,
      "grad_norm": 0.04015873000025749,
      "learning_rate": 0.0025,
      "loss": 4.99971866607666,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.57,
      "memory/max_allocated (GiB)": 18.57,
      "ppl": 148.37141,
      "step": 445,
      "tokens/total": 1008640,
      "tokens/train_per_sec_per_gpu": 16.68,
      "tokens/trainable": 597780
    },
    {
      "epoch": 1.9088960342979635,
      "grad_norm": 0.05560390651226044,
      "learning_rate": 0.0025,
      "loss": 4.3576273918151855,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 78.07168,
      "step": 446,
      "tokens/total": 1010752,
      "tokens/train_per_sec_per_gpu": 100.98,
      "tokens/trainable": 598947
    },
    {
      "epoch": 1.9131832797427653,
      "grad_norm": 0.07791434973478317,
      "learning_rate": 0.0025,
      "loss": 4.918298721313477,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 136.76973,
      "step": 447,
      "tokens/total": 1012352,
      "tokens/train_per_sec_per_gpu": 78.47,
      "tokens/trainable": 599685
    },
    {
      "epoch": 1.917470525187567,
      "grad_norm": 0.05341208353638649,
      "learning_rate": 0.0025,
      "loss": 5.289507865905762,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.49,
      "memory/max_allocated (GiB)": 18.49,
      "ppl": 198.24584,
      "step": 448,
      "tokens/total": 1014592,
      "tokens/train_per_sec_per_gpu": 6.23,
      "tokens/trainable": 601074
    },
    {
      "epoch": 1.9217577706323687,
      "grad_norm": 0.06588708609342575,
      "learning_rate": 0.0025,
      "loss": 4.922712802886963,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 137.37478,
      "step": 449,
      "tokens/total": 1016832,
      "tokens/train_per_sec_per_gpu": 6.2,
      "tokens/trainable": 602419
    },
    {
      "epoch": 1.9260450160771705,
      "grad_norm": 0.08113836497068405,
      "learning_rate": 0.0025,
      "loss": 4.302712440490723,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 73.89997,
      "step": 450,
      "tokens/total": 1018880,
      "tokens/train_per_sec_per_gpu": 54.48,
      "tokens/trainable": 603525
    },
    {
      "epoch": 1.930332261521972,
      "grad_norm": 0.038718972355127335,
      "learning_rate": 0.0025,
      "loss": 5.016265869140625,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 150.84697,
      "step": 451,
      "tokens/total": 1021440,
      "tokens/train_per_sec_per_gpu": 27.21,
      "tokens/trainable": 605190
    },
    {
      "epoch": 1.934619506966774,
      "grad_norm": 0.06918424367904663,
      "learning_rate": 0.0025,
      "loss": 4.20094633102417,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.39,
      "memory/max_allocated (GiB)": 18.39,
      "ppl": 66.74947,
      "step": 452,
      "tokens/total": 1022976,
      "tokens/train_per_sec_per_gpu": 59.21,
      "tokens/trainable": 605785
    },
    {
      "epoch": 1.9389067524115755,
      "grad_norm": 0.05727904662489891,
      "learning_rate": 0.0025,
      "loss": 4.495724201202393,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 89.63306,
      "step": 453,
      "tokens/total": 1024832,
      "tokens/train_per_sec_per_gpu": 15.8,
      "tokens/trainable": 606673
    },
    {
      "epoch": 1.9431939978563773,
      "grad_norm": 0.050397999584674835,
      "learning_rate": 0.0025,
      "loss": 4.682392120361328,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.45,
      "memory/max_allocated (GiB)": 18.45,
      "ppl": 108.02818,
      "step": 454,
      "tokens/total": 1026880,
      "tokens/train_per_sec_per_gpu": 14.36,
      "tokens/trainable": 607794
    },
    {
      "epoch": 1.947481243301179,
      "grad_norm": 0.05501880869269371,
      "learning_rate": 0.0025,
      "loss": 4.579135894775391,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.94,
      "memory/max_allocated (GiB)": 18.94,
      "ppl": 97.43017,
      "step": 455,
      "tokens/total": 1030144,
      "tokens/train_per_sec_per_gpu": 522.12,
      "tokens/trainable": 610019
    },
    {
      "epoch": 1.9517684887459807,
      "grad_norm": 0.04997771605849266,
      "learning_rate": 0.0025,
      "loss": 4.384857177734375,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 80.22676,
      "step": 456,
      "tokens/total": 1032128,
      "tokens/train_per_sec_per_gpu": 14.95,
      "tokens/trainable": 611215
    },
    {
      "epoch": 1.9560557341907825,
      "grad_norm": 0.036863308399915695,
      "learning_rate": 0.0025,
      "loss": 4.938703536987305,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 139.58916,
      "step": 457,
      "tokens/total": 1034240,
      "tokens/train_per_sec_per_gpu": 39.48,
      "tokens/trainable": 612358
    },
    {
      "epoch": 1.960342979635584,
      "grad_norm": 0.0519835501909256,
      "learning_rate": 0.0025,
      "loss": 4.352743148803711,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 77.69129,
      "step": 458,
      "tokens/total": 1035712,
      "tokens/train_per_sec_per_gpu": 86.87,
      "tokens/trainable": 612911
    },
    {
      "epoch": 1.964630225080386,
      "grad_norm": 0.06522325426340103,
      "learning_rate": 0.0025,
      "loss": 4.363661766052246,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.44,
      "memory/max_allocated (GiB)": 18.44,
      "ppl": 78.54422,
      "step": 459,
      "tokens/total": 1037376,
      "tokens/train_per_sec_per_gpu": 48.59,
      "tokens/trainable": 613674
    },
    {
      "epoch": 1.9689174705251875,
      "grad_norm": 0.04710303246974945,
      "learning_rate": 0.0025,
      "loss": 4.830078601837158,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.51,
      "memory/max_allocated (GiB)": 18.51,
      "ppl": 125.2208,
      "step": 460,
      "tokens/total": 1039296,
      "tokens/train_per_sec_per_gpu": 21.74,
      "tokens/trainable": 614688
    },
    {
      "epoch": 1.9732047159699893,
      "grad_norm": 0.05254080519080162,
      "learning_rate": 0.0025,
      "loss": 4.752572536468506,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 115.88201,
      "step": 461,
      "tokens/total": 1041280,
      "tokens/train_per_sec_per_gpu": 28.76,
      "tokens/trainable": 615727
    },
    {
      "epoch": 1.977491961414791,
      "grad_norm": 0.04451625421643257,
      "learning_rate": 0.0025,
      "loss": 4.789048671722412,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.77,
      "memory/max_allocated (GiB)": 18.77,
      "ppl": 120.18698,
      "step": 462,
      "tokens/total": 1044224,
      "tokens/train_per_sec_per_gpu": 95.08,
      "tokens/trainable": 617841
    },
    {
      "epoch": 1.9817792068595927,
      "grad_norm": 0.07913687825202942,
      "learning_rate": 0.0025,
      "loss": 4.311519622802734,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.43,
      "memory/max_allocated (GiB)": 18.43,
      "ppl": 74.5537,
      "step": 463,
      "tokens/total": 1046208,
      "tokens/train_per_sec_per_gpu": 8.84,
      "tokens/trainable": 618874
    },
    {
      "epoch": 1.9860664523043945,
      "grad_norm": 0.047682974487543106,
      "learning_rate": 0.0025,
      "loss": 5.018848419189453,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.48,
      "memory/max_allocated (GiB)": 18.48,
      "ppl": 151.23704,
      "step": 464,
      "tokens/total": 1048704,
      "tokens/train_per_sec_per_gpu": 10.86,
      "tokens/trainable": 620535
    },
    {
      "epoch": 1.990353697749196,
      "grad_norm": 0.06579267233610153,
      "learning_rate": 0.0025,
      "loss": 4.598994255065918,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.41,
      "memory/max_allocated (GiB)": 18.41,
      "ppl": 99.38431,
      "step": 465,
      "tokens/total": 1050496,
      "tokens/train_per_sec_per_gpu": 18.04,
      "tokens/trainable": 621414
    },
    {
      "epoch": 1.994640943193998,
      "grad_norm": 0.049845289438962936,
      "learning_rate": 0.0025,
      "loss": 4.27248477935791,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.46,
      "memory/max_allocated (GiB)": 18.46,
      "ppl": 71.69957,
      "step": 466,
      "tokens/total": 1052480,
      "tokens/train_per_sec_per_gpu": 112.51,
      "tokens/trainable": 622451
    },
    {
      "epoch": 1.9989281886387995,
      "grad_norm": 0.04582377150654793,
      "learning_rate": 0.0025,
      "loss": 4.850298881530762,
      "memory/device_reserved (GiB)": 19.85,
      "memory/max_active (GiB)": 18.53,
      "memory/max_allocated (GiB)": 18.53,
      "ppl": 127.77857,
      "step": 467,
      "tokens/total": 1054528,
      "tokens/train_per_sec_per_gpu": 210.97,
      "tokens/trainable": 623603
    }
  ],
  "logging_steps": 1,
  "max_steps": 467,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 234,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.347829287307059e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}