{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.29890401859847227,
  "eval_steps": 500,
  "global_step": 225,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001328462304882099,
      "grad_norm": 4.078240394592285,
      "learning_rate": 0.0,
      "loss": 1.2204899787902832,
      "memory/device_reserved (GiB)": 74.0,
      "memory/max_active (GiB)": 73.31,
      "memory/max_allocated (GiB)": 73.31,
      "ppl": 3.38885,
      "step": 1,
      "tokens/total": 241280,
      "tokens/trainable": 1844
    },
    {
      "epoch": 0.002656924609764198,
      "grad_norm": 3.4430620670318604,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 1.208996057510376,
      "memory/device_reserved (GiB)": 74.01,
      "memory/max_active (GiB)": 71.94,
      "memory/max_allocated (GiB)": 71.94,
      "ppl": 3.35012,
      "step": 2,
      "tokens/total": 481728,
      "tokens/train_per_sec_per_gpu": 13.18,
      "tokens/trainable": 3740
    },
    {
      "epoch": 0.003985386914646297,
      "grad_norm": 3.8220441341400146,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 1.428513526916504,
      "memory/device_reserved (GiB)": 76.47,
      "memory/max_active (GiB)": 75.91,
      "memory/max_allocated (GiB)": 75.91,
      "ppl": 4.17249,
      "step": 3,
      "tokens/total": 699776,
      "tokens/train_per_sec_per_gpu": 13.14,
      "tokens/trainable": 5674
    },
    {
      "epoch": 0.005313849219528396,
      "grad_norm": 3.1648757457733154,
      "learning_rate": 6.818181818181818e-06,
      "loss": 1.1087837219238281,
      "memory/device_reserved (GiB)": 76.47,
      "memory/max_active (GiB)": 75.18,
      "memory/max_allocated (GiB)": 75.18,
      "ppl": 3.03067,
      "step": 4,
      "tokens/total": 961536,
      "tokens/train_per_sec_per_gpu": 10.39,
      "tokens/trainable": 7547
    },
    {
      "epoch": 0.006642311524410495,
      "grad_norm": 3.243049144744873,
      "learning_rate": 9.090909090909091e-06,
      "loss": 1.2332323789596558,
      "memory/device_reserved (GiB)": 76.47,
      "memory/max_active (GiB)": 75.5,
      "memory/max_allocated (GiB)": 75.5,
      "ppl": 3.43231,
      "step": 5,
      "tokens/total": 1220800,
      "tokens/train_per_sec_per_gpu": 10.89,
      "tokens/trainable": 9461
    },
    {
      "epoch": 0.007970773829292593,
      "grad_norm": 2.992734670639038,
      "learning_rate": 1.1363636363636365e-05,
      "loss": 1.1764527559280396,
      "memory/device_reserved (GiB)": 76.47,
      "memory/max_active (GiB)": 75.28,
      "memory/max_allocated (GiB)": 75.28,
      "ppl": 3.24285,
      "step": 6,
      "tokens/total": 1486464,
      "tokens/train_per_sec_per_gpu": 14.72,
      "tokens/trainable": 12044
    },
    {
      "epoch": 0.009299236134174693,
      "grad_norm": 7.225315093994141,
      "learning_rate": 1.3636363636363637e-05,
      "loss": 1.3602099418640137,
      "memory/device_reserved (GiB)": 76.47,
      "memory/max_active (GiB)": 74.3,
      "memory/max_allocated (GiB)": 74.3,
      "ppl": 3.89701,
      "step": 7,
      "tokens/total": 1721408,
      "tokens/train_per_sec_per_gpu": 12.8,
      "tokens/trainable": 14012
    },
    {
      "epoch": 0.010627698439056792,
      "grad_norm": 3.7612006664276123,
      "learning_rate": 1.590909090909091e-05,
      "loss": 1.3938989639282227,
      "memory/device_reserved (GiB)": 76.47,
      "memory/max_active (GiB)": 74.33,
      "memory/max_allocated (GiB)": 74.33,
      "ppl": 4.03053,
      "step": 8,
      "tokens/total": 1929152,
      "tokens/train_per_sec_per_gpu": 14.44,
      "tokens/trainable": 15860
    },
    {
      "epoch": 0.01195616074393889,
      "grad_norm": 3.4705631732940674,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 1.2233916521072388,
      "memory/device_reserved (GiB)": 76.47,
      "memory/max_active (GiB)": 72.51,
      "memory/max_allocated (GiB)": 72.51,
      "ppl": 3.3987,
      "step": 9,
      "tokens/total": 2134464,
      "tokens/train_per_sec_per_gpu": 13.58,
      "tokens/trainable": 17682
    },
    {
      "epoch": 0.01328462304882099,
      "grad_norm": 3.247605562210083,
      "learning_rate": 2.0454545454545457e-05,
      "loss": 1.082282543182373,
      "memory/device_reserved (GiB)": 76.47,
      "memory/max_active (GiB)": 72.07,
      "memory/max_allocated (GiB)": 72.07,
      "ppl": 2.95141,
      "step": 10,
      "tokens/total": 2366720,
      "tokens/train_per_sec_per_gpu": 13.6,
      "tokens/trainable": 19525
    },
    {
      "epoch": 0.014613085353703089,
      "grad_norm": 2.765240430831909,
      "learning_rate": 2.272727272727273e-05,
      "loss": 1.0681132078170776,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 76.02,
      "memory/max_allocated (GiB)": 76.02,
      "ppl": 2.90988,
      "step": 11,
      "tokens/total": 2609600,
      "tokens/train_per_sec_per_gpu": 12.68,
      "tokens/trainable": 21534
    },
    {
      "epoch": 0.015941547658585187,
      "grad_norm": 2.2525298595428467,
      "learning_rate": 2.5e-05,
      "loss": 0.8673083782196045,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 75.23,
      "memory/max_allocated (GiB)": 75.23,
      "ppl": 2.38049,
      "step": 12,
      "tokens/total": 2881216,
      "tokens/train_per_sec_per_gpu": 11.34,
      "tokens/trainable": 23688
    },
    {
      "epoch": 0.017270009963467288,
      "grad_norm": 2.2587997913360596,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 0.8689092993736267,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 75.53,
      "memory/max_allocated (GiB)": 75.53,
      "ppl": 2.38431,
      "step": 13,
      "tokens/total": 3154112,
      "tokens/train_per_sec_per_gpu": 10.46,
      "tokens/trainable": 25514
    },
    {
      "epoch": 0.018598472268349386,
      "grad_norm": 1.9934321641921997,
      "learning_rate": 2.954545454545455e-05,
      "loss": 0.7913857102394104,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 74.19,
      "memory/max_allocated (GiB)": 74.19,
      "ppl": 2.20645,
      "step": 14,
      "tokens/total": 3410304,
      "tokens/train_per_sec_per_gpu": 12.18,
      "tokens/trainable": 27547
    },
    {
      "epoch": 0.019926934573231483,
      "grad_norm": 1.800851821899414,
      "learning_rate": 3.181818181818182e-05,
      "loss": 0.8316292762756348,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 75.77,
      "memory/max_allocated (GiB)": 75.77,
      "ppl": 2.29706,
      "step": 15,
      "tokens/total": 3654464,
      "tokens/train_per_sec_per_gpu": 13.47,
      "tokens/trainable": 29471
    },
    {
      "epoch": 0.021255396878113585,
      "grad_norm": 1.455098271369934,
      "learning_rate": 3.409090909090909e-05,
      "loss": 0.6257059574127197,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 75.2,
      "memory/max_allocated (GiB)": 75.2,
      "ppl": 1.86957,
      "step": 16,
      "tokens/total": 3909248,
      "tokens/train_per_sec_per_gpu": 12.08,
      "tokens/trainable": 31464
    },
    {
      "epoch": 0.022583859182995682,
      "grad_norm": 1.604706883430481,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 0.7017816305160522,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 74.85,
      "memory/max_allocated (GiB)": 74.85,
      "ppl": 2.01734,
      "step": 17,
      "tokens/total": 4131008,
      "tokens/train_per_sec_per_gpu": 11.97,
      "tokens/trainable": 33074
    },
    {
      "epoch": 0.02391232148787778,
      "grad_norm": 0.9868711233139038,
      "learning_rate": 3.8636363636363636e-05,
      "loss": 0.5830010175704956,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 72.75,
      "memory/max_allocated (GiB)": 72.75,
      "ppl": 1.79141,
      "step": 18,
      "tokens/total": 4342272,
      "tokens/train_per_sec_per_gpu": 15.53,
      "tokens/trainable": 35185
    },
    {
      "epoch": 0.02524078379275988,
      "grad_norm": 0.8437089323997498,
      "learning_rate": 4.0909090909090915e-05,
      "loss": 0.49017488956451416,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 75.23,
      "memory/max_allocated (GiB)": 75.23,
      "ppl": 1.6326,
      "step": 19,
      "tokens/total": 4594880,
      "tokens/train_per_sec_per_gpu": 12.79,
      "tokens/trainable": 37280
    },
    {
      "epoch": 0.02656924609764198,
      "grad_norm": 0.6587421298027039,
      "learning_rate": 4.318181818181819e-05,
      "loss": 0.3691103458404541,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 75.34,
      "memory/max_allocated (GiB)": 75.34,
      "ppl": 1.44645,
      "step": 20,
      "tokens/total": 4858752,
      "tokens/train_per_sec_per_gpu": 10.83,
      "tokens/trainable": 39261
    },
    {
      "epoch": 0.027897708402524077,
      "grad_norm": 0.7513092160224915,
      "learning_rate": 4.545454545454546e-05,
      "loss": 0.4554915130138397,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 74.66,
      "memory/max_allocated (GiB)": 74.66,
      "ppl": 1.57695,
      "step": 21,
      "tokens/total": 5107648,
      "tokens/train_per_sec_per_gpu": 12.29,
      "tokens/trainable": 41174
    },
    {
      "epoch": 0.029226170707406178,
      "grad_norm": 0.5032646059989929,
      "learning_rate": 4.772727272727273e-05,
      "loss": 0.4031963348388672,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 72.94,
      "memory/max_allocated (GiB)": 72.94,
      "ppl": 1.4966,
      "step": 22,
      "tokens/total": 5310976,
      "tokens/train_per_sec_per_gpu": 16.02,
      "tokens/trainable": 43051
    },
    {
      "epoch": 0.030554633012288276,
      "grad_norm": 0.42180025577545166,
      "learning_rate": 5e-05,
      "loss": 0.44643938541412354,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 76.1,
      "memory/max_allocated (GiB)": 76.1,
      "ppl": 1.56274,
      "step": 23,
      "tokens/total": 5557504,
      "tokens/train_per_sec_per_gpu": 14.84,
      "tokens/trainable": 45232
    },
    {
      "epoch": 0.031883095317170373,
      "grad_norm": 0.5073381662368774,
      "learning_rate": 4.999976912636752e-05,
      "loss": 0.31508195400238037,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 75.94,
      "memory/max_allocated (GiB)": 75.94,
      "ppl": 1.37037,
      "step": 24,
      "tokens/total": 5795904,
      "tokens/train_per_sec_per_gpu": 10.15,
      "tokens/trainable": 46943
    },
    {
      "epoch": 0.033211557622052475,
      "grad_norm": 0.3897036910057068,
      "learning_rate": 4.999907650973428e-05,
      "loss": 0.45456668734550476,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 74.17,
      "memory/max_allocated (GiB)": 74.17,
      "ppl": 1.57549,
      "step": 25,
      "tokens/total": 6040896,
      "tokens/train_per_sec_per_gpu": 13.88,
      "tokens/trainable": 48930
    },
    {
      "epoch": 0.034540019926934576,
      "grad_norm": 0.3590036630630493,
      "learning_rate": 4.999792216289284e-05,
      "loss": 0.3161405920982361,
      "memory/device_reserved (GiB)": 77.06,
      "memory/max_active (GiB)": 75.69,
      "memory/max_allocated (GiB)": 75.69,
      "ppl": 1.37182,
      "step": 26,
      "tokens/total": 6295680,
      "tokens/train_per_sec_per_gpu": 9.24,
      "tokens/trainable": 51085
    },
    {
      "epoch": 0.03586848223181667,
      "grad_norm": 0.45661401748657227,
      "learning_rate": 4.999630610716385e-05,
      "loss": 0.462736576795578,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 76.65,
      "memory/max_allocated (GiB)": 76.65,
      "ppl": 1.58841,
      "step": 27,
      "tokens/total": 6534144,
      "tokens/train_per_sec_per_gpu": 13.52,
      "tokens/trainable": 53256
    },
    {
      "epoch": 0.03719694453669877,
      "grad_norm": 0.4172951579093933,
      "learning_rate": 4.99942283723957e-05,
      "loss": 0.45428770780563354,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 73.51,
      "memory/max_allocated (GiB)": 73.51,
      "ppl": 1.57505,
      "step": 28,
      "tokens/total": 6745152,
      "tokens/train_per_sec_per_gpu": 15.99,
      "tokens/trainable": 55353
    },
    {
      "epoch": 0.03852540684158087,
      "grad_norm": 0.2786731421947479,
      "learning_rate": 4.99916889969639e-05,
      "loss": 0.32081466913223267,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 75.75,
      "memory/max_allocated (GiB)": 75.75,
      "ppl": 1.37825,
      "step": 29,
      "tokens/total": 6994688,
      "tokens/train_per_sec_per_gpu": 12.33,
      "tokens/trainable": 57312
    },
    {
      "epoch": 0.03985386914646297,
      "grad_norm": 0.3878122866153717,
      "learning_rate": 4.998868802777046e-05,
      "loss": 0.43424832820892334,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 76.4,
      "memory/max_allocated (GiB)": 76.4,
      "ppl": 1.5438,
      "step": 30,
      "tokens/total": 7220288,
      "tokens/train_per_sec_per_gpu": 13.43,
      "tokens/trainable": 59213
    },
    {
      "epoch": 0.04118233145134507,
      "grad_norm": 0.3197435140609741,
      "learning_rate": 4.998522552024294e-05,
      "loss": 0.40692442655563354,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 76.43,
      "memory/max_allocated (GiB)": 76.43,
      "ppl": 1.50219,
      "step": 31,
      "tokens/total": 7466432,
      "tokens/train_per_sec_per_gpu": 13.38,
      "tokens/trainable": 61213
    },
    {
      "epoch": 0.04251079375622717,
      "grad_norm": 0.34962233901023865,
      "learning_rate": 4.998130153833348e-05,
      "loss": 0.4455469250679016,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 72.26,
      "memory/max_allocated (GiB)": 72.26,
      "ppl": 1.56134,
      "step": 32,
      "tokens/total": 7682816,
      "tokens/train_per_sec_per_gpu": 14.96,
      "tokens/trainable": 63202
    },
    {
      "epoch": 0.043839256061109264,
      "grad_norm": 0.6062914133071899,
      "learning_rate": 4.9976916154517596e-05,
      "loss": 0.3847355246543884,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 71.72,
      "memory/max_allocated (GiB)": 71.72,
      "ppl": 1.46923,
      "step": 33,
      "tokens/total": 7905344,
      "tokens/train_per_sec_per_gpu": 12.46,
      "tokens/trainable": 65086
    },
    {
      "epoch": 0.045167718365991365,
      "grad_norm": 0.25035855174064636,
      "learning_rate": 4.997206944979284e-05,
      "loss": 0.281414657831192,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 76.48,
      "memory/max_allocated (GiB)": 76.48,
      "ppl": 1.325,
      "step": 34,
      "tokens/total": 8152960,
      "tokens/train_per_sec_per_gpu": 10.56,
      "tokens/trainable": 66938
    },
    {
      "epoch": 0.046496180670873466,
      "grad_norm": 0.28913062810897827,
      "learning_rate": 4.996676151367733e-05,
      "loss": 0.3254272937774658,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 72.7,
      "memory/max_allocated (GiB)": 72.7,
      "ppl": 1.38462,
      "step": 35,
      "tokens/total": 8406912,
      "tokens/train_per_sec_per_gpu": 11.75,
      "tokens/trainable": 68947
    },
    {
      "epoch": 0.04782464297575556,
      "grad_norm": 0.2699942886829376,
      "learning_rate": 4.9960992444208055e-05,
      "loss": 0.3568383455276489,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 74.66,
      "memory/max_allocated (GiB)": 74.66,
      "ppl": 1.4288,
      "step": 36,
      "tokens/total": 8657536,
      "tokens/train_per_sec_per_gpu": 12.24,
      "tokens/trainable": 70906
    },
    {
      "epoch": 0.04915310528063766,
      "grad_norm": 0.34783637523651123,
      "learning_rate": 4.99547623479391e-05,
      "loss": 0.38880521059036255,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 73.35,
      "memory/max_allocated (GiB)": 73.35,
      "ppl": 1.47522,
      "step": 37,
      "tokens/total": 8891968,
      "tokens/train_per_sec_per_gpu": 14.56,
      "tokens/trainable": 72932
    },
    {
      "epoch": 0.05048156758551976,
      "grad_norm": 0.2874269485473633,
      "learning_rate": 4.994807133993966e-05,
      "loss": 0.36390942335128784,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 73.19,
      "memory/max_allocated (GiB)": 73.19,
      "ppl": 1.43894,
      "step": 38,
      "tokens/total": 9137728,
      "tokens/train_per_sec_per_gpu": 12.32,
      "tokens/trainable": 74902
    },
    {
      "epoch": 0.05181002989040186,
      "grad_norm": 0.21942980587482452,
      "learning_rate": 4.9940919543791934e-05,
      "loss": 0.24741771817207336,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 75.69,
      "memory/max_allocated (GiB)": 75.69,
      "ppl": 1.28071,
      "step": 39,
      "tokens/total": 9373760,
      "tokens/train_per_sec_per_gpu": 12.65,
      "tokens/trainable": 76811
    },
    {
      "epoch": 0.05313849219528396,
      "grad_norm": 0.2818690836429596,
      "learning_rate": 4.9933307091588796e-05,
      "loss": 0.3861200213432312,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 73.81,
      "memory/max_allocated (GiB)": 73.81,
      "ppl": 1.47126,
      "step": 40,
      "tokens/total": 9608640,
      "tokens/train_per_sec_per_gpu": 15.3,
      "tokens/trainable": 79037
    },
    {
      "epoch": 0.05446695450016606,
      "grad_norm": 0.3083355724811554,
      "learning_rate": 4.992523412393141e-05,
      "loss": 0.3846557140350342,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 73.95,
      "memory/max_allocated (GiB)": 73.95,
      "ppl": 1.46911,
      "step": 41,
      "tokens/total": 9850688,
      "tokens/train_per_sec_per_gpu": 13.69,
      "tokens/trainable": 81194
    },
    {
      "epoch": 0.055795416805048154,
      "grad_norm": 0.27566301822662354,
      "learning_rate": 4.991670078992662e-05,
      "loss": 0.375728577375412,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 75.53,
      "memory/max_allocated (GiB)": 75.53,
      "ppl": 1.45605,
      "step": 42,
      "tokens/total": 10095872,
      "tokens/train_per_sec_per_gpu": 13.84,
      "tokens/trainable": 83205
    },
    {
      "epoch": 0.057123879109930255,
      "grad_norm": 0.2864619791507721,
      "learning_rate": 4.990770724718415e-05,
      "loss": 0.33605003356933594,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 75.53,
      "memory/max_allocated (GiB)": 75.53,
      "ppl": 1.39941,
      "step": 43,
      "tokens/total": 10312640,
      "tokens/train_per_sec_per_gpu": 14.12,
      "tokens/trainable": 85157
    },
    {
      "epoch": 0.058452341414812356,
      "grad_norm": 0.33179375529289246,
      "learning_rate": 4.9898253661813765e-05,
      "loss": 0.386963427066803,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 71.88,
      "memory/max_allocated (GiB)": 71.88,
      "ppl": 1.4725,
      "step": 44,
      "tokens/total": 10522944,
      "tokens/train_per_sec_per_gpu": 14.61,
      "tokens/trainable": 87062
    },
    {
      "epoch": 0.05978080371969445,
      "grad_norm": 0.2606225609779358,
      "learning_rate": 4.9888340208422144e-05,
      "loss": 0.2807900607585907,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 73.76,
      "memory/max_allocated (GiB)": 73.76,
      "ppl": 1.32418,
      "step": 45,
      "tokens/total": 10763456,
      "tokens/train_per_sec_per_gpu": 11.84,
      "tokens/trainable": 88876
    },
    {
      "epoch": 0.06110926602457655,
      "grad_norm": 0.29711779952049255,
      "learning_rate": 4.987796707010969e-05,
      "loss": 0.3204880952835083,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 76.21,
      "memory/max_allocated (GiB)": 76.21,
      "ppl": 1.3778,
      "step": 46,
      "tokens/total": 11004864,
      "tokens/train_per_sec_per_gpu": 11.02,
      "tokens/trainable": 90781
    },
    {
      "epoch": 0.06243772832945865,
      "grad_norm": 0.29902249574661255,
      "learning_rate": 4.986713443846714e-05,
      "loss": 0.34933191537857056,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 76.16,
      "memory/max_allocated (GiB)": 76.16,
      "ppl": 1.41812,
      "step": 47,
      "tokens/total": 11245440,
      "tokens/train_per_sec_per_gpu": 12.74,
      "tokens/trainable": 92871
    },
    {
      "epoch": 0.06376619063434075,
      "grad_norm": 0.32067298889160156,
      "learning_rate": 4.9855842513572006e-05,
      "loss": 0.4535881280899048,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 72.78,
      "memory/max_allocated (GiB)": 72.78,
      "ppl": 1.57395,
      "step": 48,
      "tokens/total": 11478272,
      "tokens/train_per_sec_per_gpu": 14.79,
      "tokens/trainable": 95022
    },
    {
      "epoch": 0.06509465293922286,
      "grad_norm": 0.36356621980667114,
      "learning_rate": 4.984409150398491e-05,
      "loss": 0.3860873579978943,
      "memory/device_reserved (GiB)": 77.39,
      "memory/max_active (GiB)": 73.32,
      "memory/max_allocated (GiB)": 73.32,
      "ppl": 1.47121,
      "step": 49,
      "tokens/total": 11701120,
      "tokens/train_per_sec_per_gpu": 15.73,
      "tokens/trainable": 97209
    },
    {
      "epoch": 0.06642311524410495,
      "grad_norm": 0.30050548911094666,
      "learning_rate": 4.983188162674571e-05,
      "loss": 0.4055514931678772,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 76.62,
      "memory/max_allocated (GiB)": 76.62,
      "ppl": 1.50013,
      "step": 50,
      "tokens/total": 11962496,
      "tokens/train_per_sec_per_gpu": 12.05,
      "tokens/trainable": 99480
    },
    {
      "epoch": 0.06775157754898704,
      "grad_norm": 0.2823496162891388,
      "learning_rate": 4.981921310736951e-05,
      "loss": 0.3524141311645508,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 71.69,
      "memory/max_allocated (GiB)": 71.69,
      "ppl": 1.4225,
      "step": 51,
      "tokens/total": 12175616,
      "tokens/train_per_sec_per_gpu": 11.42,
      "tokens/trainable": 101480
    },
    {
      "epoch": 0.06908003985386915,
      "grad_norm": 0.28376975655555725,
      "learning_rate": 4.980608617984247e-05,
      "loss": 0.34091582894325256,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 76.83,
      "memory/max_allocated (GiB)": 76.83,
      "ppl": 1.40623,
      "step": 52,
      "tokens/total": 12425856,
      "tokens/train_per_sec_per_gpu": 10.64,
      "tokens/trainable": 103474
    },
    {
      "epoch": 0.07040850215875125,
      "grad_norm": 0.3835979998111725,
      "learning_rate": 4.97925010866175e-05,
      "loss": 0.28788986802101135,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 74.28,
      "memory/max_allocated (GiB)": 74.28,
      "ppl": 1.33361,
      "step": 53,
      "tokens/total": 12673472,
      "tokens/train_per_sec_per_gpu": 10.97,
      "tokens/trainable": 105399
    },
    {
      "epoch": 0.07173696446363334,
      "grad_norm": 0.39083245396614075,
      "learning_rate": 4.97784580786098e-05,
      "loss": 0.3275277316570282,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 75.58,
      "memory/max_allocated (GiB)": 75.58,
      "ppl": 1.38753,
      "step": 54,
      "tokens/total": 12919936,
      "tokens/train_per_sec_per_gpu": 12.12,
      "tokens/trainable": 107320
    },
    {
      "epoch": 0.07306542676851545,
      "grad_norm": 0.28509923815727234,
      "learning_rate": 4.9763957415192175e-05,
      "loss": 0.288000226020813,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 71.94,
      "memory/max_allocated (GiB)": 71.94,
      "ppl": 1.33376,
      "step": 55,
      "tokens/total": 13149568,
      "tokens/train_per_sec_per_gpu": 12.46,
      "tokens/trainable": 109265
    },
    {
      "epoch": 0.07439388907339754,
      "grad_norm": 0.2659589946269989,
      "learning_rate": 4.9748999364190304e-05,
      "loss": 0.3212144374847412,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 75.94,
      "memory/max_allocated (GiB)": 75.94,
      "ppl": 1.3788,
      "step": 56,
      "tokens/total": 13412032,
      "tokens/train_per_sec_per_gpu": 11.09,
      "tokens/trainable": 111171
    },
    {
      "epoch": 0.07572235137827964,
      "grad_norm": 0.2954414486885071,
      "learning_rate": 4.973358420187776e-05,
      "loss": 0.35858094692230225,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 74.44,
      "memory/max_allocated (GiB)": 74.44,
      "ppl": 1.4313,
      "step": 57,
      "tokens/total": 13670784,
      "tokens/train_per_sec_per_gpu": 13.99,
      "tokens/trainable": 113207
    },
    {
      "epoch": 0.07705081368316175,
      "grad_norm": 0.29742714762687683,
      "learning_rate": 4.971771221297088e-05,
      "loss": 0.3568863570690155,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 74.6,
      "memory/max_allocated (GiB)": 74.6,
      "ppl": 1.42887,
      "step": 58,
      "tokens/total": 13901888,
      "tokens/train_per_sec_per_gpu": 13.24,
      "tokens/trainable": 115221
    },
    {
      "epoch": 0.07837927598804384,
      "grad_norm": 0.5640520453453064,
      "learning_rate": 4.970138369062359e-05,
      "loss": 0.28157147765159607,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 76.02,
      "memory/max_allocated (GiB)": 76.02,
      "ppl": 1.32521,
      "step": 59,
      "tokens/total": 14140224,
      "tokens/train_per_sec_per_gpu": 11.3,
      "tokens/trainable": 117086
    },
    {
      "epoch": 0.07970773829292593,
      "grad_norm": 0.28851282596588135,
      "learning_rate": 4.968459893642189e-05,
      "loss": 0.34128955006599426,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 74.6,
      "memory/max_allocated (GiB)": 74.6,
      "ppl": 1.40676,
      "step": 60,
      "tokens/total": 14384256,
      "tokens/train_per_sec_per_gpu": 14.41,
      "tokens/trainable": 119322
    },
    {
      "epoch": 0.08103620059780804,
      "grad_norm": 0.3004905879497528,
      "learning_rate": 4.966735826037836e-05,
      "loss": 0.34793412685394287,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 75.72,
      "memory/max_allocated (GiB)": 75.72,
      "ppl": 1.41614,
      "step": 61,
      "tokens/total": 14647168,
      "tokens/train_per_sec_per_gpu": 11.17,
      "tokens/trainable": 121273
    },
    {
      "epoch": 0.08236466290269014,
      "grad_norm": 0.656767725944519,
      "learning_rate": 4.9649661980926406e-05,
      "loss": 0.3032311797142029,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 75.26,
      "memory/max_allocated (GiB)": 75.26,
      "ppl": 1.35423,
      "step": 62,
      "tokens/total": 14905792,
      "tokens/train_per_sec_per_gpu": 13.11,
      "tokens/trainable": 123323
    },
    {
      "epoch": 0.08369312520757223,
      "grad_norm": 0.3169984221458435,
      "learning_rate": 4.963151042491437e-05,
      "loss": 0.365967720746994,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 75.28,
      "memory/max_allocated (GiB)": 75.28,
      "ppl": 1.44191,
      "step": 63,
      "tokens/total": 15153664,
      "tokens/train_per_sec_per_gpu": 10.93,
      "tokens/trainable": 125382
    },
    {
      "epoch": 0.08502158751245434,
      "grad_norm": 0.33295902609825134,
      "learning_rate": 4.9612903927599505e-05,
      "loss": 0.3628847599029541,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 72.29,
      "memory/max_allocated (GiB)": 72.29,
      "ppl": 1.43747,
      "step": 64,
      "tokens/total": 15367552,
      "tokens/train_per_sec_per_gpu": 15.24,
      "tokens/trainable": 127310
    },
    {
      "epoch": 0.08635004981733643,
      "grad_norm": 0.41118544340133667,
      "learning_rate": 4.959384283264179e-05,
      "loss": 0.41343849897384644,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 72.56,
      "memory/max_allocated (GiB)": 72.56,
      "ppl": 1.51201,
      "step": 65,
      "tokens/total": 15590848,
      "tokens/train_per_sec_per_gpu": 14.13,
      "tokens/trainable": 129279
    },
    {
      "epoch": 0.08767851212221853,
      "grad_norm": 0.29445692896842957,
      "learning_rate": 4.957432749209755e-05,
      "loss": 0.34112682938575745,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 73.44,
      "memory/max_allocated (GiB)": 73.44,
      "ppl": 1.40653,
      "step": 66,
      "tokens/total": 15858432,
      "tokens/train_per_sec_per_gpu": 11.05,
      "tokens/trainable": 131138
    },
    {
      "epoch": 0.08900697442710064,
      "grad_norm": 0.3269948661327362,
      "learning_rate": 4.955435826641299e-05,
      "loss": 0.3915187120437622,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 74.88,
      "memory/max_allocated (GiB)": 74.88,
      "ppl": 1.47923,
      "step": 67,
      "tokens/total": 16085504,
      "tokens/train_per_sec_per_gpu": 17.46,
      "tokens/trainable": 133359
    },
    {
      "epoch": 0.09033543673198273,
      "grad_norm": 0.30889901518821716,
      "learning_rate": 4.953393552441753e-05,
      "loss": 0.2897987961769104,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 74.74,
      "memory/max_allocated (GiB)": 74.74,
      "ppl": 1.33616,
      "step": 68,
      "tokens/total": 16326272,
      "tokens/train_per_sec_per_gpu": 12.32,
      "tokens/trainable": 135250
    },
    {
      "epoch": 0.09166389903686482,
      "grad_norm": 0.3052501082420349,
      "learning_rate": 4.951305964331698e-05,
      "loss": 0.2914356589317322,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 75.96,
      "memory/max_allocated (GiB)": 75.96,
      "ppl": 1.33835,
      "step": 69,
      "tokens/total": 16568384,
      "tokens/train_per_sec_per_gpu": 12.94,
      "tokens/trainable": 137383
    },
    {
      "epoch": 0.09299236134174693,
      "grad_norm": 0.2697888910770416,
      "learning_rate": 4.9491731008686584e-05,
      "loss": 0.2577957212924957,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 75.47,
      "memory/max_allocated (GiB)": 75.47,
      "ppl": 1.29407,
      "step": 70,
      "tokens/total": 16793536,
      "tokens/train_per_sec_per_gpu": 13.0,
      "tokens/trainable": 139172
    },
    {
      "epoch": 0.09432082364662903,
      "grad_norm": 0.29883286356925964,
      "learning_rate": 4.9469950014463876e-05,
      "loss": 0.30956798791885376,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 74.82,
      "memory/max_allocated (GiB)": 74.82,
      "ppl": 1.36284,
      "step": 71,
      "tokens/total": 17018816,
      "tokens/train_per_sec_per_gpu": 13.42,
      "tokens/trainable": 141287
    },
    {
      "epoch": 0.09564928595151112,
      "grad_norm": 0.26738041639328003,
      "learning_rate": 4.944771706294145e-05,
      "loss": 0.306536465883255,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 74.39,
      "memory/max_allocated (GiB)": 74.39,
      "ppl": 1.35871,
      "step": 72,
      "tokens/total": 17269248,
      "tokens/train_per_sec_per_gpu": 11.76,
      "tokens/trainable": 143236
    },
    {
      "epoch": 0.09697774825639323,
      "grad_norm": 0.3274018168449402,
      "learning_rate": 4.9425032564759485e-05,
      "loss": 0.34317129850387573,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 76.24,
      "memory/max_allocated (GiB)": 76.24,
      "ppl": 1.40941,
      "step": 73,
      "tokens/total": 17508992,
      "tokens/train_per_sec_per_gpu": 14.64,
      "tokens/trainable": 145571
    },
    {
      "epoch": 0.09830621056127532,
      "grad_norm": 0.27942293882369995,
      "learning_rate": 4.9401896938898185e-05,
      "loss": 0.32496753334999084,
      "memory/device_reserved (GiB)": 77.59,
      "memory/max_active (GiB)": 74.9,
      "memory/max_allocated (GiB)": 74.9,
      "ppl": 1.38399,
      "step": 74,
      "tokens/total": 17736448,
      "tokens/train_per_sec_per_gpu": 13.25,
      "tokens/trainable": 147614
    },
    {
      "epoch": 0.09963467286615742,
      "grad_norm": 0.20825843513011932,
      "learning_rate": 4.937831061267001e-05,
      "loss": 0.22000554203987122,
      "memory/device_reserved (GiB)": 77.79,
      "memory/max_active (GiB)": 76.86,
      "memory/max_allocated (GiB)": 76.86,
      "ppl": 1.24608,
      "step": 75,
      "tokens/total": 18018752,
      "tokens/train_per_sec_per_gpu": 11.4,
      "tokens/trainable": 149818
    },
    {
      "epoch": 0.10096313517103953,
      "grad_norm": 0.28913629055023193,
      "learning_rate": 4.935427402171185e-05,
      "loss": 0.34746575355529785,
      "memory/device_reserved (GiB)": 77.79,
      "memory/max_active (GiB)": 75.58,
      "memory/max_allocated (GiB)": 75.58,
      "ppl": 1.41548,
      "step": 76,
      "tokens/total": 18233856,
      "tokens/train_per_sec_per_gpu": 9.96,
      "tokens/trainable": 151707
    },
    {
      "epoch": 0.10229159747592162,
      "grad_norm": 0.25012022256851196,
      "learning_rate": 4.9329787609976893e-05,
      "loss": 0.28436708450317383,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 76.32,
      "memory/max_allocated (GiB)": 76.32,
      "ppl": 1.32892,
      "step": 77,
      "tokens/total": 18452672,
      "tokens/train_per_sec_per_gpu": 12.01,
      "tokens/trainable": 153598
    },
    {
      "epoch": 0.10362005978080371,
      "grad_norm": 0.49168774485588074,
      "learning_rate": 4.9304851829726494e-05,
      "loss": 0.28677767515182495,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 75.34,
      "memory/max_allocated (GiB)": 75.34,
      "ppl": 1.33213,
      "step": 78,
      "tokens/total": 18695552,
      "tokens/train_per_sec_per_gpu": 12.31,
      "tokens/trainable": 155561
    },
    {
      "epoch": 0.10494852208568582,
      "grad_norm": 0.5867858529090881,
      "learning_rate": 4.927946714152178e-05,
      "loss": 0.38911110162734985,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 74.52,
      "memory/max_allocated (GiB)": 74.52,
      "ppl": 1.47567,
      "step": 79,
      "tokens/total": 18907712,
      "tokens/train_per_sec_per_gpu": 15.22,
      "tokens/trainable": 157505
    },
    {
      "epoch": 0.10627698439056792,
      "grad_norm": 0.27808260917663574,
      "learning_rate": 4.925363401421517e-05,
      "loss": 0.3041893541812897,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 76.4,
      "memory/max_allocated (GiB)": 76.4,
      "ppl": 1.35553,
      "step": 80,
      "tokens/total": 19134144,
      "tokens/train_per_sec_per_gpu": 12.57,
      "tokens/trainable": 159379
    },
    {
      "epoch": 0.10760544669545001,
      "grad_norm": 0.2894990146160126,
      "learning_rate": 4.92273529249417e-05,
      "loss": 0.3562927842140198,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 72.18,
      "memory/max_allocated (GiB)": 72.18,
      "ppl": 1.42803,
      "step": 81,
      "tokens/total": 19376000,
      "tokens/train_per_sec_per_gpu": 12.75,
      "tokens/trainable": 161333
    },
    {
      "epoch": 0.10893390900033212,
      "grad_norm": 0.28392910957336426,
      "learning_rate": 4.9200624359110206e-05,
      "loss": 0.32886797189712524,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 74.09,
      "memory/max_allocated (GiB)": 74.09,
      "ppl": 1.38939,
      "step": 82,
      "tokens/total": 19613056,
      "tokens/train_per_sec_per_gpu": 16.55,
      "tokens/trainable": 163537
    },
    {
      "epoch": 0.11026237130521421,
      "grad_norm": 0.272851824760437,
      "learning_rate": 4.9173448810394385e-05,
      "loss": 0.25960707664489746,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 75.37,
      "memory/max_allocated (GiB)": 75.37,
      "ppl": 1.29642,
      "step": 83,
      "tokens/total": 19840960,
      "tokens/train_per_sec_per_gpu": 10.34,
      "tokens/trainable": 165415
    },
    {
      "epoch": 0.11159083361009631,
      "grad_norm": 0.30733823776245117,
      "learning_rate": 4.9145826780723634e-05,
      "loss": 0.29021644592285156,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 74.69,
      "memory/max_allocated (GiB)": 74.69,
      "ppl": 1.33672,
      "step": 84,
      "tokens/total": 20084608,
      "tokens/train_per_sec_per_gpu": 10.81,
      "tokens/trainable": 167156
    },
    {
      "epoch": 0.11291929591497842,
      "grad_norm": 0.24903596937656403,
      "learning_rate": 4.911775878027384e-05,
      "loss": 0.25815725326538086,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 73.24,
      "memory/max_allocated (GiB)": 73.24,
      "ppl": 1.29454,
      "step": 85,
      "tokens/total": 20300224,
      "tokens/train_per_sec_per_gpu": 13.75,
      "tokens/trainable": 169157
    },
    {
      "epoch": 0.11424775821986051,
      "grad_norm": 0.2804400324821472,
      "learning_rate": 4.9089245327457876e-05,
      "loss": 0.3358738124370575,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 73.6,
      "memory/max_allocated (GiB)": 73.6,
      "ppl": 1.39916,
      "step": 86,
      "tokens/total": 20541056,
      "tokens/train_per_sec_per_gpu": 13.75,
      "tokens/trainable": 171164
    },
    {
      "epoch": 0.1155762205247426,
      "grad_norm": 0.3141542375087738,
      "learning_rate": 4.906028694891612e-05,
      "loss": 0.3537227511405945,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 76.13,
      "memory/max_allocated (GiB)": 76.13,
      "ppl": 1.42436,
      "step": 87,
      "tokens/total": 20785216,
      "tokens/train_per_sec_per_gpu": 13.88,
      "tokens/trainable": 173239
    },
    {
      "epoch": 0.11690468282962471,
      "grad_norm": 0.2685106098651886,
      "learning_rate": 4.903088417950664e-05,
      "loss": 0.27877840399742126,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 72.67,
      "memory/max_allocated (GiB)": 72.67,
      "ppl": 1.32151,
      "step": 88,
      "tokens/total": 21016768,
      "tokens/train_per_sec_per_gpu": 12.39,
      "tokens/trainable": 175096
    },
    {
      "epoch": 0.1182331451345068,
      "grad_norm": 0.3396528661251068,
      "learning_rate": 4.900103756229538e-05,
      "loss": 0.31499725580215454,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 74.71,
      "memory/max_allocated (GiB)": 74.71,
      "ppl": 1.37026,
      "step": 89,
      "tokens/total": 21246976,
      "tokens/train_per_sec_per_gpu": 13.24,
      "tokens/trainable": 177111
    },
    {
      "epoch": 0.1195616074393889,
      "grad_norm": 0.2889237403869629,
      "learning_rate": 4.897074764854608e-05,
      "loss": 0.3066895306110382,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 74.09,
      "memory/max_allocated (GiB)": 74.09,
      "ppl": 1.35892,
      "step": 90,
      "tokens/total": 21491072,
      "tokens/train_per_sec_per_gpu": 12.02,
      "tokens/trainable": 179077
    },
    {
      "epoch": 0.12089006974427101,
      "grad_norm": 0.2633538842201233,
      "learning_rate": 4.894001499771015e-05,
      "loss": 0.3202817440032959,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 75.86,
      "memory/max_allocated (GiB)": 75.86,
      "ppl": 1.37752,
      "step": 91,
      "tokens/total": 21734208,
      "tokens/train_per_sec_per_gpu": 13.09,
      "tokens/trainable": 181123
    },
    {
      "epoch": 0.1222185320491531,
      "grad_norm": 0.28408122062683105,
      "learning_rate": 4.890884017741628e-05,
      "loss": 0.30145886540412903,
      "memory/device_reserved (GiB)": 77.2,
      "memory/max_active (GiB)": 75.53,
      "memory/max_allocated (GiB)": 75.53,
      "ppl": 1.35183,
      "step": 92,
      "tokens/total": 21955776,
      "tokens/train_per_sec_per_gpu": 15.96,
      "tokens/trainable": 183236
    },
    {
      "epoch": 0.12354699435403521,
      "grad_norm": 0.2829677164554596,
      "learning_rate": 4.887722376345999e-05,
      "loss": 0.386635959148407,
      "memory/device_reserved (GiB)": 77.55,
      "memory/max_active (GiB)": 76.84,
      "memory/max_allocated (GiB)": 76.84,
      "ppl": 1.47202,
      "step": 93,
      "tokens/total": 22220800,
      "tokens/train_per_sec_per_gpu": 13.65,
      "tokens/trainable": 185470
    },
    {
      "epoch": 0.1248754566589173,
      "grad_norm": 0.2762235701084137,
      "learning_rate": 4.884516633979299e-05,
      "loss": 0.2841187119483948,
      "memory/device_reserved (GiB)": 77.55,
      "memory/max_active (GiB)": 76.16,
      "memory/max_allocated (GiB)": 76.16,
      "ppl": 1.32859,
      "step": 94,
      "tokens/total": 22472000,
      "tokens/train_per_sec_per_gpu": 12.02,
      "tokens/trainable": 187632
    },
    {
      "epoch": 0.1262039189637994,
      "grad_norm": 0.27298712730407715,
      "learning_rate": 4.881266849851239e-05,
      "loss": 0.31646984815597534,
      "memory/device_reserved (GiB)": 77.55,
      "memory/max_active (GiB)": 72.1,
      "memory/max_allocated (GiB)": 72.1,
      "ppl": 1.37227,
      "step": 95,
      "tokens/total": 22724928,
      "tokens/train_per_sec_per_gpu": 10.78,
      "tokens/trainable": 189511
    },
    {
      "epoch": 0.1275323812686815,
      "grad_norm": 0.30283185839653015,
      "learning_rate": 4.877973083984976e-05,
      "loss": 0.32183483242988586,
      "memory/device_reserved (GiB)": 77.55,
      "memory/max_active (GiB)": 73.43,
      "memory/max_allocated (GiB)": 73.43,
      "ppl": 1.37966,
      "step": 96,
      "tokens/total": 22930688,
      "tokens/train_per_sec_per_gpu": 17.58,
      "tokens/trainable": 191527
    },
    {
      "epoch": 0.1288608435735636,
      "grad_norm": 0.39260604977607727,
      "learning_rate": 4.874635397216005e-05,
      "loss": 0.30737465620040894,
      "memory/device_reserved (GiB)": 77.55,
      "memory/max_active (GiB)": 72.92,
      "memory/max_allocated (GiB)": 72.92,
      "ppl": 1.35985,
      "step": 97,
      "tokens/total": 23159808,
      "tokens/train_per_sec_per_gpu": 13.92,
      "tokens/trainable": 193452
    },
    {
      "epoch": 0.1301893058784457,
      "grad_norm": 0.3084384799003601,
      "learning_rate": 4.8712538511910354e-05,
      "loss": 0.2816072404384613,
      "memory/device_reserved (GiB)": 77.55,
      "memory/max_active (GiB)": 76.51,
      "memory/max_allocated (GiB)": 76.51,
      "ppl": 1.32526,
      "step": 98,
      "tokens/total": 23417536,
      "tokens/train_per_sec_per_gpu": 11.32,
      "tokens/trainable": 195481
    },
    {
      "epoch": 0.1315177681833278,
      "grad_norm": 0.3135785460472107,
      "learning_rate": 4.867828508366853e-05,
      "loss": 0.3494204878807068,
      "memory/device_reserved (GiB)": 77.55,
      "memory/max_active (GiB)": 73.65,
      "memory/max_allocated (GiB)": 73.65,
      "ppl": 1.41825,
      "step": 99,
      "tokens/total": 23646592,
      "tokens/train_per_sec_per_gpu": 15.39,
      "tokens/trainable": 197676
    },
    {
      "epoch": 0.1328462304882099,
      "grad_norm": 0.40487584471702576,
      "learning_rate": 4.864359432009164e-05,
      "loss": 0.36197125911712646,
      "memory/device_reserved (GiB)": 77.55,
      "memory/max_active (GiB)": 76.02,
      "memory/max_allocated (GiB)": 76.02,
      "ppl": 1.43616,
      "step": 100,
      "tokens/total": 23892672,
      "tokens/train_per_sec_per_gpu": 11.81,
      "tokens/trainable": 199656
    },
    {
      "epoch": 0.134174692793092,
      "grad_norm": 0.31580036878585815,
      "learning_rate": 4.860846686191431e-05,
      "loss": 0.31181108951568604,
      "memory/device_reserved (GiB)": 77.55,
      "memory/max_active (GiB)": 74.41,
      "memory/max_allocated (GiB)": 74.41,
      "ppl": 1.3659,
      "step": 101,
      "tokens/total": 24151808,
      "tokens/train_per_sec_per_gpu": 9.4,
      "tokens/trainable": 201706
    },
    {
      "epoch": 0.1355031550979741,
      "grad_norm": 0.32332488894462585,
      "learning_rate": 4.857290335793683e-05,
      "loss": 0.3490774631500244,
      "memory/device_reserved (GiB)": 76.51,
      "memory/max_active (GiB)": 75.5,
      "memory/max_allocated (GiB)": 75.5,
      "ppl": 1.41776,
      "step": 102,
      "tokens/total": 24394816,
      "tokens/train_per_sec_per_gpu": 13.6,
      "tokens/trainable": 203823
    },
    {
      "epoch": 0.13683161740285618,
      "grad_norm": 0.3062637746334076,
      "learning_rate": 4.853690446501323e-05,
      "loss": 0.34347471594810486,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 76.86,
      "memory/max_allocated (GiB)": 76.86,
      "ppl": 1.40984,
      "step": 103,
      "tokens/total": 24640512,
      "tokens/train_per_sec_per_gpu": 12.27,
      "tokens/trainable": 205892
    },
    {
      "epoch": 0.1381600797077383,
      "grad_norm": 0.31956177949905396,
      "learning_rate": 4.850047084803914e-05,
      "loss": 0.3540318012237549,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 74.41,
      "memory/max_allocated (GiB)": 74.41,
      "ppl": 1.4248,
      "step": 104,
      "tokens/total": 24887680,
      "tokens/train_per_sec_per_gpu": 14.75,
      "tokens/trainable": 208306
    },
    {
      "epoch": 0.1394885420126204,
      "grad_norm": 0.2798771560192108,
      "learning_rate": 4.846360317993947e-05,
      "loss": 0.3245096206665039,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 75.31,
      "memory/max_allocated (GiB)": 75.31,
      "ppl": 1.38335,
      "step": 105,
      "tokens/total": 25134848,
      "tokens/train_per_sec_per_gpu": 12.61,
      "tokens/trainable": 210329
    },
    {
      "epoch": 0.1408170043175025,
      "grad_norm": 0.3507809042930603,
      "learning_rate": 4.842630214165601e-05,
      "loss": 0.38814184069633484,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 72.4,
      "memory/max_allocated (GiB)": 72.4,
      "ppl": 1.47424,
      "step": 106,
      "tokens/total": 25371904,
      "tokens/train_per_sec_per_gpu": 14.97,
      "tokens/trainable": 212511
    },
    {
      "epoch": 0.1421454666223846,
      "grad_norm": 0.26299262046813965,
      "learning_rate": 4.838856842213486e-05,
      "loss": 0.2627770006656647,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 73.71,
      "memory/max_allocated (GiB)": 73.71,
      "ppl": 1.30054,
      "step": 107,
      "tokens/total": 25620672,
      "tokens/train_per_sec_per_gpu": 11.23,
      "tokens/trainable": 214498
    },
    {
      "epoch": 0.14347392892726668,
      "grad_norm": 0.2580498158931732,
      "learning_rate": 4.83504027183137e-05,
      "loss": 0.2704671323299408,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 75.56,
      "memory/max_allocated (GiB)": 75.56,
      "ppl": 1.31058,
      "step": 108,
      "tokens/total": 25867840,
      "tokens/train_per_sec_per_gpu": 11.28,
      "tokens/trainable": 216425
    },
    {
      "epoch": 0.14480239123214877,
      "grad_norm": 0.29200857877731323,
      "learning_rate": 4.8311805735108894e-05,
      "loss": 0.33830636739730835,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 76.02,
      "memory/max_allocated (GiB)": 76.02,
      "ppl": 1.40257,
      "step": 109,
      "tokens/total": 26134080,
      "tokens/train_per_sec_per_gpu": 10.53,
      "tokens/trainable": 218302
    },
    {
      "epoch": 0.1461308535370309,
      "grad_norm": 0.2611442804336548,
      "learning_rate": 4.827277818540251e-05,
      "loss": 0.2801956534385681,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 75.74,
      "memory/max_allocated (GiB)": 75.74,
      "ppl": 1.32339,
      "step": 110,
      "tokens/total": 26382912,
      "tokens/train_per_sec_per_gpu": 12.55,
      "tokens/trainable": 220213
    },
    {
      "epoch": 0.147459315841913,
      "grad_norm": 0.31871896982192993,
      "learning_rate": 4.8233320790029106e-05,
      "loss": 0.295947402715683,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 73.9,
      "memory/max_allocated (GiB)": 73.9,
      "ppl": 1.3444,
      "step": 111,
      "tokens/total": 26601472,
      "tokens/train_per_sec_per_gpu": 12.96,
      "tokens/trainable": 222033
    },
    {
      "epoch": 0.14878777814679509,
      "grad_norm": 0.2990409731864929,
      "learning_rate": 4.819343427776247e-05,
      "loss": 0.30516964197158813,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 73.84,
      "memory/max_allocated (GiB)": 73.84,
      "ppl": 1.35686,
      "step": 112,
      "tokens/total": 26833216,
      "tokens/train_per_sec_per_gpu": 12.57,
      "tokens/trainable": 224044
    },
    {
      "epoch": 0.15011624045167718,
      "grad_norm": 0.33460795879364014,
      "learning_rate": 4.815311938530211e-05,
      "loss": 0.3858802914619446,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 76.4,
      "memory/max_allocated (GiB)": 76.4,
      "ppl": 1.47091,
      "step": 113,
      "tokens/total": 27077120,
      "tokens/train_per_sec_per_gpu": 12.46,
      "tokens/trainable": 226182
    },
    {
      "epoch": 0.15144470275655927,
      "grad_norm": 0.2794348895549774,
      "learning_rate": 4.8112376857259696e-05,
      "loss": 0.2724725008010864,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 73.79,
      "memory/max_allocated (GiB)": 73.79,
      "ppl": 1.31321,
      "step": 114,
      "tokens/total": 27304640,
      "tokens/train_per_sec_per_gpu": 14.11,
      "tokens/trainable": 227959
    },
    {
      "epoch": 0.15277316506144137,
      "grad_norm": 0.29718488454818726,
      "learning_rate": 4.8071207446145244e-05,
      "loss": 0.2983158528804779,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 73.05,
      "memory/max_allocated (GiB)": 73.05,
      "ppl": 1.34759,
      "step": 115,
      "tokens/total": 27544512,
      "tokens/train_per_sec_per_gpu": 12.63,
      "tokens/trainable": 230054
    },
    {
      "epoch": 0.1541016273663235,
      "grad_norm": 0.25398096442222595,
      "learning_rate": 4.802961191235329e-05,
      "loss": 0.25493133068084717,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 74.14,
      "memory/max_allocated (GiB)": 74.14,
      "ppl": 1.29037,
      "step": 116,
      "tokens/total": 27787584,
      "tokens/train_per_sec_per_gpu": 11.73,
      "tokens/trainable": 231975
    },
    {
      "epoch": 0.15543008967120558,
      "grad_norm": 0.5715415477752686,
      "learning_rate": 4.798759102414878e-05,
      "loss": 0.327411413192749,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 74.66,
      "memory/max_allocated (GiB)": 74.66,
      "ppl": 1.38737,
      "step": 117,
      "tokens/total": 28038336,
      "tokens/train_per_sec_per_gpu": 13.38,
      "tokens/trainable": 233982
    },
    {
      "epoch": 0.15675855197608768,
      "grad_norm": 0.3132580518722534,
      "learning_rate": 4.794514555765293e-05,
      "loss": 0.334320604801178,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 71.85,
      "memory/max_allocated (GiB)": 71.85,
      "ppl": 1.39699,
      "step": 118,
      "tokens/total": 28277568,
      "tokens/train_per_sec_per_gpu": 13.47,
      "tokens/trainable": 235995
    },
    {
      "epoch": 0.15808701428096977,
      "grad_norm": 0.3772391378879547,
      "learning_rate": 4.790227629682885e-05,
      "loss": 0.3174671530723572,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 75.45,
      "memory/max_allocated (GiB)": 75.45,
      "ppl": 1.37364,
      "step": 119,
      "tokens/total": 28516096,
      "tokens/train_per_sec_per_gpu": 13.49,
      "tokens/trainable": 238010
    },
    {
      "epoch": 0.15941547658585187,
      "grad_norm": 0.39647936820983887,
      "learning_rate": 4.785898403346711e-05,
      "loss": 0.36018556356430054,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 72.56,
      "memory/max_allocated (GiB)": 72.56,
      "ppl": 1.4336,
      "step": 120,
      "tokens/total": 28744448,
      "tokens/train_per_sec_per_gpu": 14.59,
      "tokens/trainable": 240127
    },
    {
      "epoch": 0.16074393889073396,
      "grad_norm": 0.24146312475204468,
      "learning_rate": 4.7815269567171084e-05,
      "loss": 0.24168699979782104,
      "memory/device_reserved (GiB)": 77.73,
      "memory/max_active (GiB)": 76.84,
      "memory/max_allocated (GiB)": 76.84,
      "ppl": 1.2734,
      "step": 121,
      "tokens/total": 29015616,
      "tokens/train_per_sec_per_gpu": 10.19,
      "tokens/trainable": 242077
    },
    {
      "epoch": 0.16207240119561608,
      "grad_norm": 0.2935802936553955,
      "learning_rate": 4.777113370534216e-05,
      "loss": 0.3234092593193054,
      "memory/device_reserved (GiB)": 77.73,
      "memory/max_active (GiB)": 71.77,
      "memory/max_allocated (GiB)": 71.77,
      "ppl": 1.38183,
      "step": 122,
      "tokens/total": 29278976,
      "tokens/train_per_sec_per_gpu": 13.4,
      "tokens/trainable": 244412
    },
    {
      "epoch": 0.16340086350049818,
      "grad_norm": 0.2624473571777344,
      "learning_rate": 4.7726577263164905e-05,
      "loss": 0.2843772768974304,
      "memory/device_reserved (GiB)": 77.73,
      "memory/max_active (GiB)": 71.2,
      "memory/max_allocated (GiB)": 71.2,
      "ppl": 1.32893,
      "step": 123,
      "tokens/total": 29514048,
      "tokens/train_per_sec_per_gpu": 11.68,
      "tokens/trainable": 246305
    },
    {
      "epoch": 0.16472932580538027,
      "grad_norm": 0.2984730303287506,
      "learning_rate": 4.768160106359191e-05,
      "loss": 0.3387468457221985,
      "memory/device_reserved (GiB)": 77.73,
      "memory/max_active (GiB)": 74.79,
      "memory/max_allocated (GiB)": 74.79,
      "ppl": 1.40319,
      "step": 124,
      "tokens/total": 29758656,
      "tokens/train_per_sec_per_gpu": 14.46,
      "tokens/trainable": 248446
    },
    {
      "epoch": 0.16605778811026237,
      "grad_norm": 0.27655282616615295,
      "learning_rate": 4.763620593732867e-05,
      "loss": 0.3312479853630066,
      "memory/device_reserved (GiB)": 77.73,
      "memory/max_active (GiB)": 76.29,
      "memory/max_allocated (GiB)": 76.29,
      "ppl": 1.39271,
      "step": 125,
      "tokens/total": 30048000,
      "tokens/train_per_sec_per_gpu": 10.94,
      "tokens/trainable": 250450
    },
    {
      "epoch": 0.16738625041514446,
      "grad_norm": 0.3455372452735901,
      "learning_rate": 4.75903927228182e-05,
      "loss": 0.3507857918739319,
      "memory/device_reserved (GiB)": 77.73,
      "memory/max_active (GiB)": 72.86,
      "memory/max_allocated (GiB)": 72.86,
      "ppl": 1.42018,
      "step": 126,
      "tokens/total": 30278784,
      "tokens/train_per_sec_per_gpu": 9.1,
      "tokens/trainable": 252279
    },
    {
      "epoch": 0.16871471272002658,
      "grad_norm": 0.2768387496471405,
      "learning_rate": 4.7544162266225546e-05,
      "loss": 0.3144207000732422,
      "memory/device_reserved (GiB)": 75.11,
      "memory/max_active (GiB)": 74.47,
      "memory/max_allocated (GiB)": 74.47,
      "ppl": 1.36947,
      "step": 127,
      "tokens/total": 30532800,
      "tokens/train_per_sec_per_gpu": 13.95,
      "tokens/trainable": 254541
    },
    {
      "epoch": 0.17004317502490868,
      "grad_norm": 0.2863609790802002,
      "learning_rate": 4.7497515421422207e-05,
      "loss": 0.3234996199607849,
      "memory/device_reserved (GiB)": 77.65,
      "memory/max_active (GiB)": 76.75,
      "memory/max_allocated (GiB)": 76.75,
      "ppl": 1.38196,
      "step": 128,
      "tokens/total": 30798336,
      "tokens/train_per_sec_per_gpu": 12.97,
      "tokens/trainable": 256801
    },
    {
      "epoch": 0.17137163732979077,
      "grad_norm": 0.32456719875335693,
      "learning_rate": 4.745045304997029e-05,
      "loss": 0.3574495315551758,
      "memory/device_reserved (GiB)": 77.65,
      "memory/max_active (GiB)": 74.69,
      "memory/max_allocated (GiB)": 74.69,
      "ppl": 1.42968,
      "step": 129,
      "tokens/total": 31032448,
      "tokens/train_per_sec_per_gpu": 11.0,
      "tokens/trainable": 258757
    },
    {
      "epoch": 0.17270009963467287,
      "grad_norm": 0.31199607253074646,
      "learning_rate": 4.7402976021106646e-05,
      "loss": 0.4081224501132965,
      "memory/device_reserved (GiB)": 77.65,
      "memory/max_active (GiB)": 74.93,
      "memory/max_allocated (GiB)": 74.93,
      "ppl": 1.50399,
      "step": 130,
      "tokens/total": 31266880,
      "tokens/train_per_sec_per_gpu": 13.45,
      "tokens/trainable": 260916
    },
    {
      "epoch": 0.17402856193955496,
      "grad_norm": 0.281172513961792,
      "learning_rate": 4.735508521172681e-05,
      "loss": 0.288382887840271,
      "memory/device_reserved (GiB)": 77.65,
      "memory/max_active (GiB)": 74.41,
      "memory/max_allocated (GiB)": 74.41,
      "ppl": 1.33427,
      "step": 131,
      "tokens/total": 31526016,
      "tokens/train_per_sec_per_gpu": 12.06,
      "tokens/trainable": 263039
    },
    {
      "epoch": 0.17535702424443705,
      "grad_norm": 0.32762470841407776,
      "learning_rate": 4.730678150636879e-05,
      "loss": 0.334347128868103,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 76.81,
      "memory/max_allocated (GiB)": 76.81,
      "ppl": 1.39703,
      "step": 132,
      "tokens/total": 31792512,
      "tokens/train_per_sec_per_gpu": 11.84,
      "tokens/trainable": 265042
    },
    {
      "epoch": 0.17668548654931918,
      "grad_norm": 0.2732056975364685,
      "learning_rate": 4.7258065797196746e-05,
      "loss": 0.2946048378944397,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 75.15,
      "memory/max_allocated (GiB)": 75.15,
      "ppl": 1.3426,
      "step": 133,
      "tokens/total": 32026880,
      "tokens/train_per_sec_per_gpu": 13.67,
      "tokens/trainable": 267150
    },
    {
      "epoch": 0.17801394885420127,
      "grad_norm": 0.25782883167266846,
      "learning_rate": 4.720893898398448e-05,
      "loss": 0.2663328945636749,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 75.94,
      "memory/max_allocated (GiB)": 75.94,
      "ppl": 1.30517,
      "step": 134,
      "tokens/total": 32308160,
      "tokens/train_per_sec_per_gpu": 11.08,
      "tokens/trainable": 269279
    },
    {
      "epoch": 0.17934241115908336,
      "grad_norm": 0.342846542596817,
      "learning_rate": 4.715940197409887e-05,
      "loss": 0.2981494069099426,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 70.77,
      "memory/max_allocated (GiB)": 70.77,
      "ppl": 1.34736,
      "step": 135,
      "tokens/total": 32536512,
      "tokens/train_per_sec_per_gpu": 10.64,
      "tokens/trainable": 271049
    },
    {
      "epoch": 0.18067087346396546,
      "grad_norm": 0.3146810829639435,
      "learning_rate": 4.710945568248306e-05,
      "loss": 0.3311242461204529,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 76.37,
      "memory/max_allocated (GiB)": 76.37,
      "ppl": 1.39253,
      "step": 136,
      "tokens/total": 32801920,
      "tokens/train_per_sec_per_gpu": 11.55,
      "tokens/trainable": 272970
    },
    {
      "epoch": 0.18199933576884755,
      "grad_norm": 0.358317106962204,
      "learning_rate": 4.705910103163961e-05,
      "loss": 0.28497856855392456,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 73.76,
      "memory/max_allocated (GiB)": 73.76,
      "ppl": 1.32973,
      "step": 137,
      "tokens/total": 33045184,
      "tokens/train_per_sec_per_gpu": 13.69,
      "tokens/trainable": 275029
    },
    {
      "epoch": 0.18332779807372965,
      "grad_norm": 0.27245497703552246,
      "learning_rate": 4.7008338951613394e-05,
      "loss": 0.28498828411102295,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 75.31,
      "memory/max_allocated (GiB)": 75.31,
      "ppl": 1.32975,
      "step": 138,
      "tokens/total": 33309824,
      "tokens/train_per_sec_per_gpu": 13.06,
      "tokens/trainable": 277152
    },
    {
      "epoch": 0.18465626037861177,
      "grad_norm": 0.29116177558898926,
      "learning_rate": 4.695717037997448e-05,
      "loss": 0.275570273399353,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 74.06,
      "memory/max_allocated (GiB)": 74.06,
      "ppl": 1.31728,
      "step": 139,
      "tokens/total": 33522560,
      "tokens/train_per_sec_per_gpu": 13.63,
      "tokens/trainable": 278902
    },
    {
      "epoch": 0.18598472268349386,
      "grad_norm": 0.2572404742240906,
      "learning_rate": 4.69055962618008e-05,
      "loss": 0.2584611177444458,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 75.82,
      "memory/max_allocated (GiB)": 75.82,
      "ppl": 1.29494,
      "step": 140,
      "tokens/total": 33747968,
      "tokens/train_per_sec_per_gpu": 14.16,
      "tokens/trainable": 280769
    },
    {
      "epoch": 0.18731318498837596,
      "grad_norm": 0.2903401255607605,
      "learning_rate": 4.685361754966065e-05,
      "loss": 0.3318089246749878,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 75.07,
      "memory/max_allocated (GiB)": 75.07,
      "ppl": 1.39349,
      "step": 141,
      "tokens/total": 33982976,
      "tokens/train_per_sec_per_gpu": 16.53,
      "tokens/trainable": 282960
    },
    {
      "epoch": 0.18864164729325805,
      "grad_norm": 0.28281405568122864,
      "learning_rate": 4.6801235203595195e-05,
      "loss": 0.26338034868240356,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 74.9,
      "memory/max_allocated (GiB)": 74.9,
      "ppl": 1.30132,
      "step": 142,
      "tokens/total": 34235328,
      "tokens/train_per_sec_per_gpu": 12.81,
      "tokens/trainable": 285014
    },
    {
      "epoch": 0.18997010959814015,
      "grad_norm": 0.35500800609588623,
      "learning_rate": 4.67484501911006e-05,
      "loss": 0.40566298365592957,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 74.06,
      "memory/max_allocated (GiB)": 74.06,
      "ppl": 1.5003,
      "step": 143,
      "tokens/total": 34485952,
      "tokens/train_per_sec_per_gpu": 13.65,
      "tokens/trainable": 287043
    },
    {
      "epoch": 0.19129857190302224,
      "grad_norm": 0.2855430245399475,
      "learning_rate": 4.669526348711028e-05,
      "loss": 0.2761286497116089,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 73.19,
      "memory/max_allocated (GiB)": 73.19,
      "ppl": 1.31802,
      "step": 144,
      "tokens/total": 34707328,
      "tokens/train_per_sec_per_gpu": 16.14,
      "tokens/trainable": 289319
    },
    {
      "epoch": 0.19262703420790436,
      "grad_norm": 0.26699694991111755,
      "learning_rate": 4.664167607397686e-05,
      "loss": 0.23146027326583862,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 74.28,
      "memory/max_allocated (GiB)": 74.28,
      "ppl": 1.26044,
      "step": 145,
      "tokens/total": 34971200,
      "tokens/train_per_sec_per_gpu": 10.74,
      "tokens/trainable": 291326
    },
    {
      "epoch": 0.19395549651278646,
      "grad_norm": 0.32218310236930847,
      "learning_rate": 4.658768894145397e-05,
      "loss": 0.32315659523010254,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 76.67,
      "memory/max_allocated (GiB)": 76.67,
      "ppl": 1.38148,
      "step": 146,
      "tokens/total": 35194240,
      "tokens/train_per_sec_per_gpu": 15.46,
      "tokens/trainable": 293573
    },
    {
      "epoch": 0.19528395881766855,
      "grad_norm": 0.3315838575363159,
      "learning_rate": 4.653330308667805e-05,
      "loss": 0.2897493243217468,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 73.57,
      "memory/max_allocated (GiB)": 73.57,
      "ppl": 1.33609,
      "step": 147,
      "tokens/total": 35438912,
      "tokens/train_per_sec_per_gpu": 13.66,
      "tokens/trainable": 295640
    },
    {
      "epoch": 0.19661242112255065,
      "grad_norm": 0.28656816482543945,
      "learning_rate": 4.64785195141499e-05,
      "loss": 0.3137304186820984,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 73.57,
      "memory/max_allocated (GiB)": 73.57,
      "ppl": 1.36852,
      "step": 148,
      "tokens/total": 35660288,
      "tokens/train_per_sec_per_gpu": 14.06,
      "tokens/trainable": 297574
    },
    {
      "epoch": 0.19794088342743274,
      "grad_norm": 0.29270216822624207,
      "learning_rate": 4.64233392357161e-05,
      "loss": 0.2963527739048004,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 75.34,
      "memory/max_allocated (GiB)": 75.34,
      "ppl": 1.34494,
      "step": 149,
      "tokens/total": 35924928,
      "tokens/train_per_sec_per_gpu": 11.77,
      "tokens/trainable": 299725
    },
    {
      "epoch": 0.19926934573231483,
      "grad_norm": 0.3433849811553955,
      "learning_rate": 4.636776327055035e-05,
      "loss": 0.317320317029953,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 75.83,
      "memory/max_allocated (GiB)": 75.83,
      "ppl": 1.37344,
      "step": 150,
      "tokens/total": 36142656,
      "tokens/train_per_sec_per_gpu": 12.91,
      "tokens/trainable": 301695
    },
    {
      "epoch": 0.20059780803719696,
      "grad_norm": 0.31213390827178955,
      "learning_rate": 4.6311792645134666e-05,
      "loss": 0.32360923290252686,
      "memory/device_reserved (GiB)": 77.77,
      "memory/max_active (GiB)": 76.75,
      "memory/max_allocated (GiB)": 76.75,
      "ppl": 1.38211,
      "step": 151,
      "tokens/total": 36373632,
      "tokens/train_per_sec_per_gpu": 9.87,
      "tokens/trainable": 303672
    },
    {
      "epoch": 0.20192627034207905,
      "grad_norm": 0.3097369074821472,
      "learning_rate": 4.625542839324036e-05,
      "loss": 0.3049156963825226,
      "memory/device_reserved (GiB)": 77.43,
      "memory/max_active (GiB)": 74.93,
      "memory/max_allocated (GiB)": 74.93,
      "ppl": 1.35651,
      "step": 152,
      "tokens/total": 36624000,
      "tokens/train_per_sec_per_gpu": 13.25,
      "tokens/trainable": 305735
    },
    {
      "epoch": 0.20325473264696114,
      "grad_norm": 0.30076032876968384,
      "learning_rate": 4.6198671555909e-05,
      "loss": 0.32292836904525757,
      "memory/device_reserved (GiB)": 77.43,
      "memory/max_active (GiB)": 74.6,
      "memory/max_allocated (GiB)": 74.6,
      "ppl": 1.38117,
      "step": 153,
      "tokens/total": 36853696,
      "tokens/train_per_sec_per_gpu": 13.61,
      "tokens/trainable": 307623
    },
    {
      "epoch": 0.20458319495184324,
      "grad_norm": 0.3042234182357788,
      "learning_rate": 4.6141523181433165e-05,
      "loss": 0.34925395250320435,
      "memory/device_reserved (GiB)": 77.43,
      "memory/max_active (GiB)": 75.88,
      "memory/max_allocated (GiB)": 75.88,
      "ppl": 1.41801,
      "step": 154,
      "tokens/total": 37093312,
      "tokens/train_per_sec_per_gpu": 14.79,
      "tokens/trainable": 309875
    },
    {
      "epoch": 0.20591165725672533,
      "grad_norm": 0.25751274824142456,
      "learning_rate": 4.608398432533708e-05,
      "loss": 0.22095003724098206,
      "memory/device_reserved (GiB)": 77.43,
      "memory/max_active (GiB)": 75.39,
      "memory/max_allocated (GiB)": 75.39,
      "ppl": 1.24726,
      "step": 155,
      "tokens/total": 37312896,
      "tokens/train_per_sec_per_gpu": 11.13,
      "tokens/trainable": 311609
    },
    {
      "epoch": 0.20724011956160743,
      "grad_norm": 0.2809881567955017,
      "learning_rate": 4.602605605035713e-05,
      "loss": 0.3040582835674286,
      "memory/device_reserved (GiB)": 77.43,
      "memory/max_active (GiB)": 73.62,
      "memory/max_allocated (GiB)": 73.62,
      "ppl": 1.35535,
      "step": 156,
      "tokens/total": 37581184,
      "tokens/train_per_sec_per_gpu": 11.32,
      "tokens/trainable": 313404
    },
    {
      "epoch": 0.20856858186648955,
      "grad_norm": 0.2722752094268799,
      "learning_rate": 4.5967739426422193e-05,
      "loss": 0.24645733833312988,
      "memory/device_reserved (GiB)": 77.43,
      "memory/max_active (GiB)": 74.28,
      "memory/max_allocated (GiB)": 74.28,
      "ppl": 1.27948,
      "step": 157,
      "tokens/total": 37820352,
      "tokens/train_per_sec_per_gpu": 11.93,
      "tokens/trainable": 315316
    },
    {
      "epoch": 0.20989704417137164,
      "grad_norm": 0.2881944477558136,
      "learning_rate": 4.590903553063396e-05,
      "loss": 0.29333794116973877,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 76.67,
      "memory/max_allocated (GiB)": 76.67,
      "ppl": 1.3409,
      "step": 158,
      "tokens/total": 38062400,
      "tokens/train_per_sec_per_gpu": 13.87,
      "tokens/trainable": 317431
    },
    {
      "epoch": 0.21122550647625374,
      "grad_norm": 0.3498353064060211,
      "learning_rate": 4.584994544724695e-05,
      "loss": 0.2311086356639862,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 72.75,
      "memory/max_allocated (GiB)": 72.75,
      "ppl": 1.26,
      "step": 159,
      "tokens/total": 38270528,
      "tokens/train_per_sec_per_gpu": 14.7,
      "tokens/trainable": 319251
    },
    {
      "epoch": 0.21255396878113583,
      "grad_norm": 0.3815538287162781,
      "learning_rate": 4.5790470267648546e-05,
      "loss": 0.3265290856361389,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 74.9,
      "memory/max_allocated (GiB)": 74.9,
      "ppl": 1.38615,
      "step": 160,
      "tokens/total": 38513600,
      "tokens/train_per_sec_per_gpu": 13.21,
      "tokens/trainable": 321202
    },
    {
      "epoch": 0.21388243108601793,
      "grad_norm": 0.2811428904533386,
      "learning_rate": 4.573061109033879e-05,
      "loss": 0.28701144456863403,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 75.56,
      "memory/max_allocated (GiB)": 75.56,
      "ppl": 1.33244,
      "step": 161,
      "tokens/total": 38774720,
      "tokens/train_per_sec_per_gpu": 12.19,
      "tokens/trainable": 323209
    },
    {
      "epoch": 0.21521089339090002,
      "grad_norm": 0.28593000769615173,
      "learning_rate": 4.5670369020910165e-05,
      "loss": 0.25745725631713867,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 76.24,
      "memory/max_allocated (GiB)": 76.24,
      "ppl": 1.29364,
      "step": 162,
      "tokens/total": 39011968,
      "tokens/train_per_sec_per_gpu": 13.34,
      "tokens/trainable": 325118
    },
    {
      "epoch": 0.21653935569578214,
      "grad_norm": 0.2986576557159424,
      "learning_rate": 4.560974517202709e-05,
      "loss": 0.2792147994041443,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 76.26,
      "memory/max_allocated (GiB)": 76.26,
      "ppl": 1.32209,
      "step": 163,
      "tokens/total": 39264832,
      "tokens/train_per_sec_per_gpu": 10.73,
      "tokens/trainable": 326998
    },
    {
      "epoch": 0.21786781800066424,
      "grad_norm": 0.2871696352958679,
      "learning_rate": 4.554874066340541e-05,
      "loss": 0.288088321685791,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 75.42,
      "memory/max_allocated (GiB)": 75.42,
      "ppl": 1.33388,
      "step": 164,
      "tokens/total": 39501440,
      "tokens/train_per_sec_per_gpu": 13.11,
      "tokens/trainable": 329049
    },
    {
      "epoch": 0.21919628030554633,
      "grad_norm": 0.37666866183280945,
      "learning_rate": 4.548735662179174e-05,
      "loss": 0.34987354278564453,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 76.29,
      "memory/max_allocated (GiB)": 76.29,
      "ppl": 1.41889,
      "step": 165,
      "tokens/total": 39751360,
      "tokens/train_per_sec_per_gpu": 12.12,
      "tokens/trainable": 331007
    },
    {
      "epoch": 0.22052474261042843,
      "grad_norm": 0.30536556243896484,
      "learning_rate": 4.542559418094261e-05,
      "loss": 0.27000105381011963,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 71.86,
      "memory/max_allocated (GiB)": 71.86,
      "ppl": 1.30997,
      "step": 166,
      "tokens/total": 39971904,
      "tokens/train_per_sec_per_gpu": 14.19,
      "tokens/trainable": 333019
    },
    {
      "epoch": 0.22185320491531052,
      "grad_norm": 0.28819429874420166,
      "learning_rate": 4.536345448160355e-05,
      "loss": 0.26946187019348145,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 75.37,
      "memory/max_allocated (GiB)": 75.37,
      "ppl": 1.30926,
      "step": 167,
      "tokens/total": 40189632,
      "tokens/train_per_sec_per_gpu": 13.38,
      "tokens/trainable": 334772
    },
    {
      "epoch": 0.22318166722019261,
      "grad_norm": 0.2997879981994629,
      "learning_rate": 4.5300938671487994e-05,
      "loss": 0.3346363306045532,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 74.55,
      "memory/max_allocated (GiB)": 74.55,
      "ppl": 1.39743,
      "step": 168,
      "tokens/total": 40410560,
      "tokens/train_per_sec_per_gpu": 14.8,
      "tokens/trainable": 336836
    },
    {
      "epoch": 0.22451012952507474,
      "grad_norm": 0.347880482673645,
      "learning_rate": 4.5238047905256133e-05,
      "loss": 0.29474785923957825,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 74.17,
      "memory/max_allocated (GiB)": 74.17,
      "ppl": 1.34279,
      "step": 169,
      "tokens/total": 40637568,
      "tokens/train_per_sec_per_gpu": 12.59,
      "tokens/trainable": 338683
    },
    {
      "epoch": 0.22583859182995683,
      "grad_norm": 0.3111642003059387,
      "learning_rate": 4.517478334449353e-05,
      "loss": 0.29106810688972473,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 74.85,
      "memory/max_allocated (GiB)": 74.85,
      "ppl": 1.33786,
      "step": 170,
      "tokens/total": 40868480,
      "tokens/train_per_sec_per_gpu": 14.64,
      "tokens/trainable": 340786
    },
    {
      "epoch": 0.22716705413483892,
      "grad_norm": 0.31952518224716187,
      "learning_rate": 4.51111461576897e-05,
      "loss": 0.27892524003982544,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 73.84,
      "memory/max_allocated (GiB)": 73.84,
      "ppl": 1.32171,
      "step": 171,
      "tokens/total": 41112064,
      "tokens/train_per_sec_per_gpu": 13.66,
      "tokens/trainable": 342770
    },
    {
      "epoch": 0.22849551643972102,
      "grad_norm": 0.3235322833061218,
      "learning_rate": 4.5047137520216525e-05,
      "loss": 0.2796081304550171,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 75.99,
      "memory/max_allocated (GiB)": 75.99,
      "ppl": 1.32261,
      "step": 172,
      "tokens/total": 41374592,
      "tokens/train_per_sec_per_gpu": 11.71,
      "tokens/trainable": 344707
    },
    {
      "epoch": 0.2298239787446031,
      "grad_norm": 0.2968906760215759,
      "learning_rate": 4.4982758614306536e-05,
      "loss": 0.298882395029068,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 75.04,
      "memory/max_allocated (GiB)": 75.04,
      "ppl": 1.34835,
      "step": 173,
      "tokens/total": 41624000,
      "tokens/train_per_sec_per_gpu": 14.44,
      "tokens/trainable": 346771
    },
    {
      "epoch": 0.2311524410494852,
      "grad_norm": 0.3555050492286682,
      "learning_rate": 4.491801062903108e-05,
      "loss": 0.3100724220275879,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 74.28,
      "memory/max_allocated (GiB)": 74.28,
      "ppl": 1.36352,
      "step": 174,
      "tokens/total": 41874048,
      "tokens/train_per_sec_per_gpu": 13.33,
      "tokens/trainable": 348815
    },
    {
      "epoch": 0.23248090335436733,
      "grad_norm": 0.30196645855903625,
      "learning_rate": 4.4852894760278365e-05,
      "loss": 0.29532450437545776,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 71.06,
      "memory/max_allocated (GiB)": 71.06,
      "ppl": 1.34356,
      "step": 175,
      "tokens/total": 42112192,
      "tokens/train_per_sec_per_gpu": 11.1,
      "tokens/trainable": 350521
    },
    {
      "epoch": 0.23380936565924942,
      "grad_norm": 0.43994855880737305,
      "learning_rate": 4.478741221073136e-05,
      "loss": 0.2932076156139374,
      "memory/device_reserved (GiB)": 77.51,
      "memory/max_active (GiB)": 73.82,
      "memory/max_allocated (GiB)": 73.82,
      "ppl": 1.34072,
      "step": 176,
      "tokens/total": 42375168,
      "tokens/train_per_sec_per_gpu": 8.54,
      "tokens/trainable": 352486
    },
    {
      "epoch": 0.23513782796413152,
      "grad_norm": 0.34153908491134644,
      "learning_rate": 4.4721564189845586e-05,
      "loss": 0.31959718465805054,
      "memory/device_reserved (GiB)": 74.4,
      "memory/max_active (GiB)": 73.43,
      "memory/max_allocated (GiB)": 73.43,
      "ppl": 1.37657,
      "step": 177,
      "tokens/total": 42638528,
      "tokens/train_per_sec_per_gpu": 12.91,
      "tokens/trainable": 354742
    },
    {
      "epoch": 0.2364662902690136,
      "grad_norm": 0.33175817131996155,
      "learning_rate": 4.465535191382679e-05,
      "loss": 0.3373476564884186,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 76.51,
      "memory/max_allocated (GiB)": 76.51,
      "ppl": 1.40123,
      "step": 178,
      "tokens/total": 42876288,
      "tokens/train_per_sec_per_gpu": 10.26,
      "tokens/trainable": 356530
    },
    {
      "epoch": 0.2377947525738957,
      "grad_norm": 0.28470543026924133,
      "learning_rate": 4.4588776605608476e-05,
      "loss": 0.3152643144130707,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 73.92,
      "memory/max_allocated (GiB)": 73.92,
      "ppl": 1.37062,
      "step": 179,
      "tokens/total": 43082304,
      "tokens/train_per_sec_per_gpu": 16.86,
      "tokens/trainable": 358735
    },
    {
      "epoch": 0.2391232148787778,
      "grad_norm": 0.2639199495315552,
      "learning_rate": 4.4521839494829287e-05,
      "loss": 0.3099166750907898,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 71.91,
      "memory/max_allocated (GiB)": 71.91,
      "ppl": 1.36331,
      "step": 180,
      "tokens/total": 43306368,
      "tokens/train_per_sec_per_gpu": 14.48,
      "tokens/trainable": 360825
    },
    {
      "epoch": 0.24045167718365992,
      "grad_norm": 0.36245235800743103,
      "learning_rate": 4.445454181781035e-05,
      "loss": 0.32269325852394104,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 71.64,
      "memory/max_allocated (GiB)": 71.64,
      "ppl": 1.38084,
      "step": 181,
      "tokens/total": 43538880,
      "tokens/train_per_sec_per_gpu": 15.62,
      "tokens/trainable": 362988
    },
    {
      "epoch": 0.24178013948854202,
      "grad_norm": 0.2659633159637451,
      "learning_rate": 4.4386884817532385e-05,
      "loss": 0.25600507855415344,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 75.26,
      "memory/max_allocated (GiB)": 75.26,
      "ppl": 1.29176,
      "step": 182,
      "tokens/total": 43779072,
      "tokens/train_per_sec_per_gpu": 15.72,
      "tokens/trainable": 365305
    },
    {
      "epoch": 0.2431086017934241,
      "grad_norm": 0.27374452352523804,
      "learning_rate": 4.4318869743612795e-05,
      "loss": 0.27148744463920593,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 76.07,
      "memory/max_allocated (GiB)": 76.07,
      "ppl": 1.31191,
      "step": 183,
      "tokens/total": 44008576,
      "tokens/train_per_sec_per_gpu": 11.05,
      "tokens/trainable": 367095
    },
    {
      "epoch": 0.2444370640983062,
      "grad_norm": 1.360638976097107,
      "learning_rate": 4.4250497852282543e-05,
      "loss": 0.2803201973438263,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 72.53,
      "memory/max_allocated (GiB)": 72.53,
      "ppl": 1.32355,
      "step": 184,
      "tokens/total": 44240256,
      "tokens/train_per_sec_per_gpu": 14.35,
      "tokens/trainable": 369494
    },
    {
      "epoch": 0.2457655264031883,
      "grad_norm": 0.3048482835292816,
      "learning_rate": 4.4181770406363e-05,
      "loss": 0.3505616784095764,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 72.54,
      "memory/max_allocated (GiB)": 72.54,
      "ppl": 1.41986,
      "step": 185,
      "tokens/total": 44488832,
      "tokens/train_per_sec_per_gpu": 12.24,
      "tokens/trainable": 371547
    },
    {
      "epoch": 0.24709398870807042,
      "grad_norm": 0.31528833508491516,
      "learning_rate": 4.411268867524256e-05,
      "loss": 0.30864253640174866,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 73.16,
      "memory/max_allocated (GiB)": 73.16,
      "ppl": 1.36158,
      "step": 186,
      "tokens/total": 44722240,
      "tokens/train_per_sec_per_gpu": 13.16,
      "tokens/trainable": 373550
    },
    {
      "epoch": 0.24842245101295252,
      "grad_norm": 0.2920733392238617,
      "learning_rate": 4.404325393485324e-05,
      "loss": 0.31081369519233704,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 74.88,
      "memory/max_allocated (GiB)": 74.88,
      "ppl": 1.36453,
      "step": 187,
      "tokens/total": 44962176,
      "tokens/train_per_sec_per_gpu": 13.27,
      "tokens/trainable": 375701
    },
    {
      "epoch": 0.2497509133178346,
      "grad_norm": 0.2978683114051819,
      "learning_rate": 4.39734674676471e-05,
      "loss": 0.23271390795707703,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 72.1,
      "memory/max_allocated (GiB)": 72.1,
      "ppl": 1.26202,
      "step": 188,
      "tokens/total": 45165184,
      "tokens/train_per_sec_per_gpu": 14.17,
      "tokens/trainable": 377472
    },
    {
      "epoch": 0.2510793756227167,
      "grad_norm": 0.2642655670642853,
      "learning_rate": 4.390333056257255e-05,
      "loss": 0.2638115882873535,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 75.31,
      "memory/max_allocated (GiB)": 75.31,
      "ppl": 1.30188,
      "step": 189,
      "tokens/total": 45422144,
      "tokens/train_per_sec_per_gpu": 10.56,
      "tokens/trainable": 379359
    },
    {
      "epoch": 0.2524078379275988,
      "grad_norm": 0.2818956673145294,
      "learning_rate": 4.3832844515050566e-05,
      "loss": 0.2970280647277832,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 75.26,
      "memory/max_allocated (GiB)": 75.26,
      "ppl": 1.34585,
      "step": 190,
      "tokens/total": 45653952,
      "tokens/train_per_sec_per_gpu": 12.61,
      "tokens/trainable": 381373
    },
    {
      "epoch": 0.2537363002324809,
      "grad_norm": 0.2860153317451477,
      "learning_rate": 4.376201062695073e-05,
      "loss": 0.2925737202167511,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 73.92,
      "memory/max_allocated (GiB)": 73.92,
      "ppl": 1.33987,
      "step": 191,
      "tokens/total": 45908032,
      "tokens/train_per_sec_per_gpu": 12.55,
      "tokens/trainable": 383361
    },
    {
      "epoch": 0.255064762537363,
      "grad_norm": 0.3126516044139862,
      "learning_rate": 4.369083020656719e-05,
      "loss": 0.28758862614631653,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 72.4,
      "memory/max_allocated (GiB)": 72.4,
      "ppl": 1.33321,
      "step": 192,
      "tokens/total": 46158464,
      "tokens/train_per_sec_per_gpu": 10.96,
      "tokens/trainable": 385343
    },
    {
      "epoch": 0.2563932248422451,
      "grad_norm": 0.3115479052066803,
      "learning_rate": 4.361930456859455e-05,
      "loss": 0.3113458454608917,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 72.81,
      "memory/max_allocated (GiB)": 72.81,
      "ppl": 1.36526,
      "step": 193,
      "tokens/total": 46396736,
      "tokens/train_per_sec_per_gpu": 13.81,
      "tokens/trainable": 387407
    },
    {
      "epoch": 0.2577216871471272,
      "grad_norm": 0.2858979403972626,
      "learning_rate": 4.35474350341035e-05,
      "loss": 0.3073645830154419,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 75.39,
      "memory/max_allocated (GiB)": 75.39,
      "ppl": 1.35984,
      "step": 194,
      "tokens/total": 46657664,
      "tokens/train_per_sec_per_gpu": 12.35,
      "tokens/trainable": 389442
    },
    {
      "epoch": 0.2590501494520093,
      "grad_norm": 0.28876593708992004,
      "learning_rate": 4.347522293051648e-05,
      "loss": 0.3285921812057495,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 73.54,
      "memory/max_allocated (GiB)": 73.54,
      "ppl": 1.38901,
      "step": 195,
      "tokens/total": 46902976,
      "tokens/train_per_sec_per_gpu": 12.02,
      "tokens/trainable": 391494
    },
    {
      "epoch": 0.2603786117568914,
      "grad_norm": 0.31868475675582886,
      "learning_rate": 4.340266959158315e-05,
      "loss": 0.35973045229911804,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 72.67,
      "memory/max_allocated (GiB)": 72.67,
      "ppl": 1.43294,
      "step": 196,
      "tokens/total": 47111552,
      "tokens/train_per_sec_per_gpu": 12.57,
      "tokens/trainable": 393605
    },
    {
      "epoch": 0.2617070740617735,
      "grad_norm": 0.2687374949455261,
      "learning_rate": 4.332977635735575e-05,
      "loss": 0.22261589765548706,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 72.21,
      "memory/max_allocated (GiB)": 72.21,
      "ppl": 1.24934,
      "step": 197,
      "tokens/total": 47342400,
      "tokens/train_per_sec_per_gpu": 13.55,
      "tokens/trainable": 395583
    },
    {
      "epoch": 0.2630355363666556,
      "grad_norm": 0.31248337030410767,
      "learning_rate": 4.325654457416431e-05,
      "loss": 0.38228270411491394,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 71.8,
      "memory/max_allocated (GiB)": 71.8,
      "ppl": 1.46563,
      "step": 198,
      "tokens/total": 47562368,
      "tokens/train_per_sec_per_gpu": 14.3,
      "tokens/trainable": 397573
    },
    {
      "epoch": 0.2643639986715377,
      "grad_norm": 0.3104705810546875,
      "learning_rate": 4.318297559459189e-05,
      "loss": 0.2662810981273651,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 75.5,
      "memory/max_allocated (GiB)": 75.5,
      "ppl": 1.3051,
      "step": 199,
      "tokens/total": 47807936,
      "tokens/train_per_sec_per_gpu": 12.86,
      "tokens/trainable": 399696
    },
    {
      "epoch": 0.2656924609764198,
      "grad_norm": 0.259775310754776,
      "learning_rate": 4.310907077744948e-05,
      "loss": 0.23782427608966827,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 76.51,
      "memory/max_allocated (GiB)": 76.51,
      "ppl": 1.26849,
      "step": 200,
      "tokens/total": 48060864,
      "tokens/train_per_sec_per_gpu": 11.64,
      "tokens/trainable": 401489
    },
    {
      "epoch": 0.2670209232813019,
      "grad_norm": 0.27360811829566956,
      "learning_rate": 4.303483148775096e-05,
      "loss": 0.3060590624809265,
      "memory/device_reserved (GiB)": 77.53,
      "memory/max_active (GiB)": 74.41,
      "memory/max_allocated (GiB)": 74.41,
      "ppl": 1.35806,
      "step": 201,
      "tokens/total": 48323904,
      "tokens/train_per_sec_per_gpu": 10.02,
      "tokens/trainable": 403580
    },
    {
      "epoch": 0.268349385586184,
      "grad_norm": 0.31186768412590027,
      "learning_rate": 4.296025909668789e-05,
      "loss": 0.3071833848953247,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 76.86,
      "memory/max_allocated (GiB)": 76.86,
      "ppl": 1.35959,
      "step": 202,
      "tokens/total": 48563456,
      "tokens/train_per_sec_per_gpu": 11.94,
      "tokens/trainable": 405581
    },
    {
      "epoch": 0.2696778478910661,
      "grad_norm": 0.3315155506134033,
      "learning_rate": 4.28853549816042e-05,
      "loss": 0.2834095358848572,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 71.58,
      "memory/max_allocated (GiB)": 71.58,
      "ppl": 1.32765,
      "step": 203,
      "tokens/total": 48813440,
      "tokens/train_per_sec_per_gpu": 12.93,
      "tokens/trainable": 407764
    },
    {
      "epoch": 0.2710063101959482,
      "grad_norm": 0.2930338978767395,
      "learning_rate": 4.2810120525970665e-05,
      "loss": 0.31087392568588257,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 74.22,
      "memory/max_allocated (GiB)": 74.22,
      "ppl": 1.36462,
      "step": 204,
      "tokens/total": 49066304,
      "tokens/train_per_sec_per_gpu": 13.43,
      "tokens/trainable": 409824
    },
    {
      "epoch": 0.27233477250083027,
      "grad_norm": 0.32186028361320496,
      "learning_rate": 4.2734557119359476e-05,
      "loss": 0.30415642261505127,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 72.64,
      "memory/max_allocated (GiB)": 72.64,
      "ppl": 1.35548,
      "step": 205,
      "tokens/total": 49293568,
      "tokens/train_per_sec_per_gpu": 12.27,
      "tokens/trainable": 411722
    },
    {
      "epoch": 0.27366323480571236,
      "grad_norm": 0.47513481974601746,
      "learning_rate": 4.265866615741847e-05,
      "loss": 0.3157661259174347,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 71.42,
      "memory/max_allocated (GiB)": 71.42,
      "ppl": 1.37131,
      "step": 206,
      "tokens/total": 49512768,
      "tokens/train_per_sec_per_gpu": 13.7,
      "tokens/trainable": 413768
    },
    {
      "epoch": 0.2749916971105945,
      "grad_norm": 0.3138316869735718,
      "learning_rate": 4.258244904184542e-05,
      "loss": 0.3363015651702881,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 75.04,
      "memory/max_allocated (GiB)": 75.04,
      "ppl": 1.39976,
      "step": 207,
      "tokens/total": 49771392,
      "tokens/train_per_sec_per_gpu": 12.13,
      "tokens/trainable": 415765
    },
    {
      "epoch": 0.2763201594154766,
      "grad_norm": 0.34376972913742065,
      "learning_rate": 4.250590718036211e-05,
      "loss": 0.3455408811569214,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 72.07,
      "memory/max_allocated (GiB)": 72.07,
      "ppl": 1.41275,
      "step": 208,
      "tokens/total": 50012288,
      "tokens/train_per_sec_per_gpu": 12.23,
      "tokens/trainable": 417719
    },
    {
      "epoch": 0.2776486217203587,
      "grad_norm": 0.34443584084510803,
      "learning_rate": 4.242904198668835e-05,
      "loss": 0.3519732356071472,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 74.52,
      "memory/max_allocated (GiB)": 74.52,
      "ppl": 1.42187,
      "step": 209,
      "tokens/total": 50252032,
      "tokens/train_per_sec_per_gpu": 13.41,
      "tokens/trainable": 419785
    },
    {
      "epoch": 0.2789770840252408,
      "grad_norm": 0.30584385991096497,
      "learning_rate": 4.235185488051585e-05,
      "loss": 0.33031266927719116,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 72.78,
      "memory/max_allocated (GiB)": 72.78,
      "ppl": 1.3914,
      "step": 210,
      "tokens/total": 50501056,
      "tokens/train_per_sec_per_gpu": 14.24,
      "tokens/trainable": 421997
    },
    {
      "epoch": 0.2803055463301229,
      "grad_norm": 0.26669830083847046,
      "learning_rate": 4.227434728748203e-05,
      "loss": 0.26533663272857666,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 74.3,
      "memory/max_allocated (GiB)": 74.3,
      "ppl": 1.30387,
      "step": 211,
      "tokens/total": 50736384,
      "tokens/train_per_sec_per_gpu": 12.02,
      "tokens/trainable": 423988
    },
    {
      "epoch": 0.281634008635005,
      "grad_norm": 0.27942854166030884,
      "learning_rate": 4.219652063914364e-05,
      "loss": 0.2868190109729767,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 73.57,
      "memory/max_allocated (GiB)": 73.57,
      "ppl": 1.33218,
      "step": 212,
      "tokens/total": 50987072,
      "tokens/train_per_sec_per_gpu": 12.86,
      "tokens/trainable": 426032
    },
    {
      "epoch": 0.2829624709398871,
      "grad_norm": 0.3654444217681885,
      "learning_rate": 4.211837637295037e-05,
      "loss": 0.40046119689941406,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 75.58,
      "memory/max_allocated (GiB)": 75.58,
      "ppl": 1.49251,
      "step": 213,
      "tokens/total": 51247232,
      "tokens/train_per_sec_per_gpu": 13.16,
      "tokens/trainable": 428093
    },
    {
      "epoch": 0.2842909332447692,
      "grad_norm": 0.2786194980144501,
      "learning_rate": 4.203991593221826e-05,
      "loss": 0.29233628511428833,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 74.06,
      "memory/max_allocated (GiB)": 74.06,
      "ppl": 1.33955,
      "step": 214,
      "tokens/total": 51507072,
      "tokens/train_per_sec_per_gpu": 13.94,
      "tokens/trainable": 430322
    },
    {
      "epoch": 0.28561939554965127,
      "grad_norm": 0.34399065375328064,
      "learning_rate": 4.1961140766103086e-05,
      "loss": 0.24060076475143433,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 75.56,
      "memory/max_allocated (GiB)": 75.56,
      "ppl": 1.27201,
      "step": 215,
      "tokens/total": 51770304,
      "tokens/train_per_sec_per_gpu": 10.43,
      "tokens/trainable": 432220
    },
    {
      "epoch": 0.28694785785453336,
      "grad_norm": 0.3530604839324951,
      "learning_rate": 4.1882052329573525e-05,
      "loss": 0.303854763507843,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 72.37,
      "memory/max_allocated (GiB)": 72.37,
      "ppl": 1.35507,
      "step": 216,
      "tokens/total": 52006144,
      "tokens/train_per_sec_per_gpu": 12.51,
      "tokens/trainable": 434209
    },
    {
      "epoch": 0.28827632015941546,
      "grad_norm": 0.2763025462627411,
      "learning_rate": 4.180265208338435e-05,
      "loss": 0.272201806306839,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 73.51,
      "memory/max_allocated (GiB)": 73.51,
      "ppl": 1.31285,
      "step": 217,
      "tokens/total": 52268864,
      "tokens/train_per_sec_per_gpu": 10.91,
      "tokens/trainable": 436165
    },
    {
      "epoch": 0.28960478246429755,
      "grad_norm": 0.3117513060569763,
      "learning_rate": 4.1722941494049434e-05,
      "loss": 0.31475740671157837,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 74.71,
      "memory/max_allocated (GiB)": 74.71,
      "ppl": 1.36993,
      "step": 218,
      "tokens/total": 52521920,
      "tokens/train_per_sec_per_gpu": 12.28,
      "tokens/trainable": 438058
    },
    {
      "epoch": 0.2909332447691797,
      "grad_norm": 0.2993617057800293,
      "learning_rate": 4.1642922033814634e-05,
      "loss": 0.2623658776283264,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 72.15,
      "memory/max_allocated (GiB)": 72.15,
      "ppl": 1.3,
      "step": 219,
      "tokens/total": 52766336,
      "tokens/train_per_sec_per_gpu": 13.59,
      "tokens/trainable": 440175
    },
    {
      "epoch": 0.2922617070740618,
      "grad_norm": 0.29368656873703003,
      "learning_rate": 4.1562595180630623e-05,
      "loss": 0.28216126561164856,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 73.13,
      "memory/max_allocated (GiB)": 73.13,
      "ppl": 1.32599,
      "step": 220,
      "tokens/total": 53002816,
      "tokens/train_per_sec_per_gpu": 11.78,
      "tokens/trainable": 442077
    },
    {
      "epoch": 0.2935901693789439,
      "grad_norm": 0.35555794835090637,
      "learning_rate": 4.14819624181256e-05,
      "loss": 0.33095845580101013,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 76.32,
      "memory/max_allocated (GiB)": 76.32,
      "ppl": 1.3923,
      "step": 221,
      "tokens/total": 53271936,
      "tokens/train_per_sec_per_gpu": 12.62,
      "tokens/trainable": 444197
    },
    {
      "epoch": 0.294918631683826,
      "grad_norm": 0.2814570665359497,
      "learning_rate": 4.140102523557786e-05,
      "loss": 0.28345173597335815,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 73.49,
      "memory/max_allocated (GiB)": 73.49,
      "ppl": 1.3277,
      "step": 222,
      "tokens/total": 53528320,
      "tokens/train_per_sec_per_gpu": 13.0,
      "tokens/trainable": 446301
    },
    {
      "epoch": 0.2962470939887081,
      "grad_norm": 0.37089794874191284,
      "learning_rate": 4.131978512788832e-05,
      "loss": 0.25006091594696045,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 76.1,
      "memory/max_allocated (GiB)": 76.1,
      "ppl": 1.2841,
      "step": 223,
      "tokens/total": 53759936,
      "tokens/train_per_sec_per_gpu": 11.94,
      "tokens/trainable": 448283
    },
    {
      "epoch": 0.29757555629359017,
      "grad_norm": 0.3063358664512634,
      "learning_rate": 4.123824359555286e-05,
      "loss": 0.32116472721099854,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 74.63,
      "memory/max_allocated (GiB)": 74.63,
      "ppl": 1.37873,
      "step": 224,
      "tokens/total": 54002880,
      "tokens/train_per_sec_per_gpu": 12.51,
      "tokens/trainable": 450133
    },
    {
      "epoch": 0.29890401859847227,
      "grad_norm": 0.396696001291275,
      "learning_rate": 4.115640214463469e-05,
      "loss": 0.3590637147426605,
      "memory/device_reserved (GiB)": 77.63,
      "memory/max_active (GiB)": 73.62,
      "memory/max_allocated (GiB)": 73.62,
      "ppl": 1.43199,
      "step": 225,
      "tokens/total": 54252032,
      "tokens/train_per_sec_per_gpu": 16.38,
      "tokens/trainable": 452632
    }
  ],
  "logging_steps": 1,
  "max_steps": 753,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 25,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0046436664399626e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}