{
  "best_global_step": 208,
  "best_metric": 0.305007666349411,
  "best_model_checkpoint": "tmp/out/512-96-ft-l1-r2.1_common_channel_fcmCtx8_fcmLayers2_fcmChMixingTrue_stride24_bs512_lr0.0003_89b6/checkpoint-208",
  "epoch": 104.0,
  "eval_steps": 500,
  "global_step": 208,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.6618412137031555,
      "learning_rate": 0.0002999998149449555,
      "loss": 0.3922,
      "step": 2
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.358045369386673,
      "eval_runtime": 0.9433,
      "eval_samples_per_second": 314.866,
      "eval_steps_per_second": 1.06,
      "step": 2
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.465237021446228,
      "learning_rate": 0.0002999983345073394,
      "loss": 0.3871,
      "step": 4
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.35514307022094727,
      "eval_runtime": 0.984,
      "eval_samples_per_second": 301.829,
      "eval_steps_per_second": 1.016,
      "step": 4
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.3141915500164032,
      "learning_rate": 0.00029999537364671844,
      "loss": 0.3814,
      "step": 6
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.3537431061267853,
      "eval_runtime": 1.0649,
      "eval_samples_per_second": 278.894,
      "eval_steps_per_second": 0.939,
      "step": 6
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.3491910696029663,
      "learning_rate": 0.0002999909323923152,
      "loss": 0.3733,
      "step": 8
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.3517496585845947,
      "eval_runtime": 0.9441,
      "eval_samples_per_second": 314.571,
      "eval_steps_per_second": 1.059,
      "step": 8
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.29127219319343567,
      "learning_rate": 0.000299985010787963,
      "loss": 0.3749,
      "step": 10
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.35048815608024597,
      "eval_runtime": 0.9185,
      "eval_samples_per_second": 323.353,
      "eval_steps_per_second": 1.089,
      "step": 10
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.33912065625190735,
      "learning_rate": 0.0002999776088921057,
      "loss": 0.3708,
      "step": 12
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.35033392906188965,
      "eval_runtime": 0.9493,
      "eval_samples_per_second": 312.868,
      "eval_steps_per_second": 1.053,
      "step": 12
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.311758428812027,
      "learning_rate": 0.0002999687267777971,
      "loss": 0.3685,
      "step": 14
    },
    {
      "epoch": 7.0,
      "eval_loss": 0.35117536783218384,
      "eval_runtime": 0.7975,
      "eval_samples_per_second": 372.426,
      "eval_steps_per_second": 1.254,
      "step": 14
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.30109912157058716,
      "learning_rate": 0.0002999583645327,
      "loss": 0.3639,
      "step": 16
    },
    {
      "epoch": 8.0,
      "eval_loss": 0.35066911578178406,
      "eval_runtime": 0.7655,
      "eval_samples_per_second": 388.003,
      "eval_steps_per_second": 1.306,
      "step": 16
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.3332655131816864,
      "learning_rate": 0.0002999465222590856,
      "loss": 0.3655,
      "step": 18
    },
    {
      "epoch": 9.0,
      "eval_loss": 0.350115031003952,
      "eval_runtime": 0.7681,
      "eval_samples_per_second": 386.655,
      "eval_steps_per_second": 1.302,
      "step": 18
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.36181315779685974,
      "learning_rate": 0.00029993320007383234,
      "loss": 0.3627,
      "step": 20
    },
    {
      "epoch": 10.0,
      "eval_loss": 0.35086938738822937,
      "eval_runtime": 0.7303,
      "eval_samples_per_second": 406.705,
      "eval_steps_per_second": 1.369,
      "step": 20
    },
    {
      "epoch": 11.0,
      "grad_norm": 0.3522287607192993,
      "learning_rate": 0.0002999183981084249,
      "loss": 0.3575,
      "step": 22
    },
    {
      "epoch": 11.0,
      "eval_loss": 0.3501550853252411,
      "eval_runtime": 0.6638,
      "eval_samples_per_second": 447.418,
      "eval_steps_per_second": 1.506,
      "step": 22
    },
    {
      "epoch": 12.0,
      "grad_norm": 0.48265793919563293,
      "learning_rate": 0.0002999021165089526,
      "loss": 0.356,
      "step": 24
    },
    {
      "epoch": 12.0,
      "eval_loss": 0.34953632950782776,
      "eval_runtime": 0.6847,
      "eval_samples_per_second": 433.784,
      "eval_steps_per_second": 1.461,
      "step": 24
    },
    {
      "epoch": 13.0,
      "grad_norm": 0.452486515045166,
      "learning_rate": 0.0002998843554361083,
      "loss": 0.3572,
      "step": 26
    },
    {
      "epoch": 13.0,
      "eval_loss": 0.34998443722724915,
      "eval_runtime": 0.6957,
      "eval_samples_per_second": 426.915,
      "eval_steps_per_second": 1.437,
      "step": 26
    },
    {
      "epoch": 14.0,
      "grad_norm": 0.5479589700698853,
      "learning_rate": 0.0002998651150651866,
      "loss": 0.3602,
      "step": 28
    },
    {
      "epoch": 14.0,
      "eval_loss": 0.3498837649822235,
      "eval_runtime": 0.7349,
      "eval_samples_per_second": 404.119,
      "eval_steps_per_second": 1.361,
      "step": 28
    },
    {
      "epoch": 15.0,
      "grad_norm": 0.43432193994522095,
      "learning_rate": 0.00029984439558608224,
      "loss": 0.3552,
      "step": 30
    },
    {
      "epoch": 15.0,
      "eval_loss": 0.34781429171562195,
      "eval_runtime": 0.6365,
      "eval_samples_per_second": 466.635,
      "eval_steps_per_second": 1.571,
      "step": 30
    },
    {
      "epoch": 16.0,
      "grad_norm": 0.35709846019744873,
      "learning_rate": 0.00029982219720328803,
      "loss": 0.3442,
      "step": 32
    },
    {
      "epoch": 16.0,
      "eval_loss": 0.34714677929878235,
      "eval_runtime": 0.7389,
      "eval_samples_per_second": 401.956,
      "eval_steps_per_second": 1.353,
      "step": 32
    },
    {
      "epoch": 17.0,
      "grad_norm": 0.3510877192020416,
      "learning_rate": 0.00029979852013589306,
      "loss": 0.3466,
      "step": 34
    },
    {
      "epoch": 17.0,
      "eval_loss": 0.3476266860961914,
      "eval_runtime": 0.582,
      "eval_samples_per_second": 510.287,
      "eval_steps_per_second": 1.718,
      "step": 34
    },
    {
      "epoch": 18.0,
      "grad_norm": 0.6324598789215088,
      "learning_rate": 0.0002997733646175805,
      "loss": 0.3495,
      "step": 36
    },
    {
      "epoch": 18.0,
      "eval_loss": 0.34712979197502136,
      "eval_runtime": 0.6306,
      "eval_samples_per_second": 470.991,
      "eval_steps_per_second": 1.586,
      "step": 36
    },
    {
      "epoch": 19.0,
      "grad_norm": 0.3290903866291046,
      "learning_rate": 0.00029974673089662506,
      "loss": 0.343,
      "step": 38
    },
    {
      "epoch": 19.0,
      "eval_loss": 0.3468128442764282,
      "eval_runtime": 0.6393,
      "eval_samples_per_second": 464.534,
      "eval_steps_per_second": 1.564,
      "step": 38
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.3430045545101166,
      "learning_rate": 0.00029971861923589095,
      "loss": 0.342,
      "step": 40
    },
    {
      "epoch": 20.0,
      "eval_loss": 0.3459309935569763,
      "eval_runtime": 0.7921,
      "eval_samples_per_second": 374.931,
      "eval_steps_per_second": 1.262,
      "step": 40
    },
    {
      "epoch": 21.0,
      "grad_norm": 0.39738816022872925,
      "learning_rate": 0.0002996890299128288,
      "loss": 0.3444,
      "step": 42
    },
    {
      "epoch": 21.0,
      "eval_loss": 0.3452853560447693,
      "eval_runtime": 0.7918,
      "eval_samples_per_second": 375.108,
      "eval_steps_per_second": 1.263,
      "step": 42
    },
    {
      "epoch": 22.0,
      "grad_norm": 0.3685533106327057,
      "learning_rate": 0.00029965796321947336,
      "loss": 0.3368,
      "step": 44
    },
    {
      "epoch": 22.0,
      "eval_loss": 0.34537574648857117,
      "eval_runtime": 0.715,
      "eval_samples_per_second": 415.38,
      "eval_steps_per_second": 1.399,
      "step": 44
    },
    {
      "epoch": 23.0,
      "grad_norm": 0.41565972566604614,
      "learning_rate": 0.00029962541946244024,
      "loss": 0.3337,
      "step": 46
    },
    {
      "epoch": 23.0,
      "eval_loss": 0.34441086649894714,
      "eval_runtime": 0.5609,
      "eval_samples_per_second": 529.532,
      "eval_steps_per_second": 1.783,
      "step": 46
    },
    {
      "epoch": 24.0,
      "grad_norm": 0.36327725648880005,
      "learning_rate": 0.00029959139896292323,
      "loss": 0.3342,
      "step": 48
    },
    {
      "epoch": 24.0,
      "eval_loss": 0.34298038482666016,
      "eval_runtime": 0.672,
      "eval_samples_per_second": 441.965,
      "eval_steps_per_second": 1.488,
      "step": 48
    },
    {
      "epoch": 25.0,
      "grad_norm": 0.4043494164943695,
      "learning_rate": 0.0002995559020566911,
      "loss": 0.3414,
      "step": 50
    },
    {
      "epoch": 25.0,
      "eval_loss": 0.3428707420825958,
      "eval_runtime": 0.6633,
      "eval_samples_per_second": 447.772,
      "eval_steps_per_second": 1.508,
      "step": 50
    },
    {
      "epoch": 26.0,
      "grad_norm": 0.40012046694755554,
      "learning_rate": 0.0002995189290940837,
      "loss": 0.3387,
      "step": 52
    },
    {
      "epoch": 26.0,
      "eval_loss": 0.3441244959831238,
      "eval_runtime": 0.7445,
      "eval_samples_per_second": 398.927,
      "eval_steps_per_second": 1.343,
      "step": 52
    },
    {
      "epoch": 27.0,
      "grad_norm": 0.4684393107891083,
      "learning_rate": 0.0002994804804400094,
      "loss": 0.3343,
      "step": 54
    },
    {
      "epoch": 27.0,
      "eval_loss": 0.3433794677257538,
      "eval_runtime": 0.681,
      "eval_samples_per_second": 436.153,
      "eval_steps_per_second": 1.469,
      "step": 54
    },
    {
      "epoch": 28.0,
      "grad_norm": 0.4722350239753723,
      "learning_rate": 0.00029944055647394087,
      "loss": 0.3366,
      "step": 56
    },
    {
      "epoch": 28.0,
      "eval_loss": 0.3417203724384308,
      "eval_runtime": 0.6481,
      "eval_samples_per_second": 458.289,
      "eval_steps_per_second": 1.543,
      "step": 56
    },
    {
      "epoch": 29.0,
      "grad_norm": 0.5819035768508911,
      "learning_rate": 0.0002993991575899116,
      "loss": 0.3338,
      "step": 58
    },
    {
      "epoch": 29.0,
      "eval_loss": 0.3421356976032257,
      "eval_runtime": 0.8299,
      "eval_samples_per_second": 357.891,
      "eval_steps_per_second": 1.205,
      "step": 58
    },
    {
      "epoch": 30.0,
      "grad_norm": 0.4985131621360779,
      "learning_rate": 0.0002993562841965118,
      "loss": 0.3314,
      "step": 60
    },
    {
      "epoch": 30.0,
      "eval_loss": 0.34163638949394226,
      "eval_runtime": 0.6224,
      "eval_samples_per_second": 477.149,
      "eval_steps_per_second": 1.607,
      "step": 60
    },
    {
      "epoch": 31.0,
      "grad_norm": 0.5123302340507507,
      "learning_rate": 0.00029931193671688453,
      "loss": 0.33,
      "step": 62
    },
    {
      "epoch": 31.0,
      "eval_loss": 0.33960819244384766,
      "eval_runtime": 0.7702,
      "eval_samples_per_second": 385.59,
      "eval_steps_per_second": 1.298,
      "step": 62
    },
    {
      "epoch": 32.0,
      "grad_norm": 0.34553447365760803,
      "learning_rate": 0.0002992661155887215,
      "loss": 0.3253,
      "step": 64
    },
    {
      "epoch": 32.0,
      "eval_loss": 0.33889296650886536,
      "eval_runtime": 0.6787,
      "eval_samples_per_second": 437.601,
      "eval_steps_per_second": 1.473,
      "step": 64
    },
    {
      "epoch": 33.0,
      "grad_norm": 0.4334942698478699,
      "learning_rate": 0.0002992188212642587,
      "loss": 0.3221,
      "step": 66
    },
    {
      "epoch": 33.0,
      "eval_loss": 0.3390357494354248,
      "eval_runtime": 0.7883,
      "eval_samples_per_second": 376.773,
      "eval_steps_per_second": 1.269,
      "step": 66
    },
    {
      "epoch": 34.0,
      "grad_norm": 0.5000684261322021,
      "learning_rate": 0.0002991700542102722,
      "loss": 0.3228,
      "step": 68
    },
    {
      "epoch": 34.0,
      "eval_loss": 0.3387429118156433,
      "eval_runtime": 0.7066,
      "eval_samples_per_second": 420.345,
      "eval_steps_per_second": 1.415,
      "step": 68
    },
    {
      "epoch": 35.0,
      "grad_norm": 0.5403483510017395,
      "learning_rate": 0.000299119814908073,
      "loss": 0.3246,
      "step": 70
    },
    {
      "epoch": 35.0,
      "eval_loss": 0.337623655796051,
      "eval_runtime": 0.5666,
      "eval_samples_per_second": 524.176,
      "eval_steps_per_second": 1.765,
      "step": 70
    },
    {
      "epoch": 36.0,
      "grad_norm": 0.3801553547382355,
      "learning_rate": 0.00029906810385350283,
      "loss": 0.3215,
      "step": 72
    },
    {
      "epoch": 36.0,
      "eval_loss": 0.3370514214038849,
      "eval_runtime": 0.7169,
      "eval_samples_per_second": 414.265,
      "eval_steps_per_second": 1.395,
      "step": 72
    },
    {
      "epoch": 37.0,
      "grad_norm": 0.6648715138435364,
      "learning_rate": 0.00029901492155692876,
      "loss": 0.3205,
      "step": 74
    },
    {
      "epoch": 37.0,
      "eval_loss": 0.3367176651954651,
      "eval_runtime": 0.6219,
      "eval_samples_per_second": 477.599,
      "eval_steps_per_second": 1.608,
      "step": 74
    },
    {
      "epoch": 38.0,
      "grad_norm": 0.38467201590538025,
      "learning_rate": 0.0002989602685432388,
      "loss": 0.3175,
      "step": 76
    },
    {
      "epoch": 38.0,
      "eval_loss": 0.3385818302631378,
      "eval_runtime": 0.7477,
      "eval_samples_per_second": 397.199,
      "eval_steps_per_second": 1.337,
      "step": 76
    },
    {
      "epoch": 39.0,
      "grad_norm": 0.5086773633956909,
      "learning_rate": 0.00029890414535183583,
      "loss": 0.3188,
      "step": 78
    },
    {
      "epoch": 39.0,
      "eval_loss": 0.33492594957351685,
      "eval_runtime": 0.7572,
      "eval_samples_per_second": 392.254,
      "eval_steps_per_second": 1.321,
      "step": 78
    },
    {
      "epoch": 40.0,
      "grad_norm": 0.4003610610961914,
      "learning_rate": 0.00029884655253663344,
      "loss": 0.3197,
      "step": 80
    },
    {
      "epoch": 40.0,
      "eval_loss": 0.33226659893989563,
      "eval_runtime": 0.738,
      "eval_samples_per_second": 402.452,
      "eval_steps_per_second": 1.355,
      "step": 80
    },
    {
      "epoch": 41.0,
      "grad_norm": 0.4991908371448517,
      "learning_rate": 0.00029878749066604925,
      "loss": 0.313,
      "step": 82
    },
    {
      "epoch": 41.0,
      "eval_loss": 0.331875741481781,
      "eval_runtime": 0.7609,
      "eval_samples_per_second": 390.343,
      "eval_steps_per_second": 1.314,
      "step": 82
    },
    {
      "epoch": 42.0,
      "grad_norm": 0.450600802898407,
      "learning_rate": 0.0002987269603230001,
      "loss": 0.3142,
      "step": 84
    },
    {
      "epoch": 42.0,
      "eval_loss": 0.33316686749458313,
      "eval_runtime": 0.6259,
      "eval_samples_per_second": 474.552,
      "eval_steps_per_second": 1.598,
      "step": 84
    },
    {
      "epoch": 43.0,
      "grad_norm": 0.7004669308662415,
      "learning_rate": 0.00029866496210489605,
      "loss": 0.3112,
      "step": 86
    },
    {
      "epoch": 43.0,
      "eval_loss": 0.3317732810974121,
      "eval_runtime": 0.7352,
      "eval_samples_per_second": 403.996,
      "eval_steps_per_second": 1.36,
      "step": 86
    },
    {
      "epoch": 44.0,
      "grad_norm": 0.4675546884536743,
      "learning_rate": 0.0002986014966236345,
      "loss": 0.3118,
      "step": 88
    },
    {
      "epoch": 44.0,
      "eval_loss": 0.32955238223075867,
      "eval_runtime": 0.8122,
      "eval_samples_per_second": 365.683,
      "eval_steps_per_second": 1.231,
      "step": 88
    },
    {
      "epoch": 45.0,
      "grad_norm": 0.5740877389907837,
      "learning_rate": 0.00029853656450559414,
      "loss": 0.3083,
      "step": 90
    },
    {
      "epoch": 45.0,
      "eval_loss": 0.32967710494995117,
      "eval_runtime": 0.7609,
      "eval_samples_per_second": 390.346,
      "eval_steps_per_second": 1.314,
      "step": 90
    },
    {
      "epoch": 46.0,
      "grad_norm": 0.5421927571296692,
      "learning_rate": 0.00029847016639162867,
      "loss": 0.3125,
      "step": 92
    },
    {
      "epoch": 46.0,
      "eval_loss": 0.33168312907218933,
      "eval_runtime": 0.6488,
      "eval_samples_per_second": 457.788,
      "eval_steps_per_second": 1.541,
      "step": 92
    },
    {
      "epoch": 47.0,
      "grad_norm": 0.6857365369796753,
      "learning_rate": 0.0002984023029370609,
      "loss": 0.3035,
      "step": 94
    },
    {
      "epoch": 47.0,
      "eval_loss": 0.3298337161540985,
      "eval_runtime": 0.6488,
      "eval_samples_per_second": 457.751,
      "eval_steps_per_second": 1.541,
      "step": 94
    },
    {
      "epoch": 48.0,
      "grad_norm": 0.8388419151306152,
      "learning_rate": 0.0002983329748116755,
      "loss": 0.3031,
      "step": 96
    },
    {
      "epoch": 48.0,
      "eval_loss": 0.32884910702705383,
      "eval_runtime": 0.6776,
      "eval_samples_per_second": 438.283,
      "eval_steps_per_second": 1.476,
      "step": 96
    },
    {
      "epoch": 49.0,
      "grad_norm": 0.46255478262901306,
      "learning_rate": 0.00029826218269971314,
      "loss": 0.305,
      "step": 98
    },
    {
      "epoch": 49.0,
      "eval_loss": 0.3276749551296234,
      "eval_runtime": 0.733,
      "eval_samples_per_second": 405.188,
      "eval_steps_per_second": 1.364,
      "step": 98
    },
    {
      "epoch": 50.0,
      "grad_norm": 0.44940125942230225,
      "learning_rate": 0.0002981899272998634,
      "loss": 0.302,
      "step": 100
    },
    {
      "epoch": 50.0,
      "eval_loss": 0.3282870352268219,
      "eval_runtime": 0.7162,
      "eval_samples_per_second": 414.71,
      "eval_steps_per_second": 1.396,
      "step": 100
    },
    {
      "epoch": 51.0,
      "grad_norm": 0.5168047547340393,
      "learning_rate": 0.0002981162093252579,
      "loss": 0.2989,
      "step": 102
    },
    {
      "epoch": 51.0,
      "eval_loss": 0.33124735951423645,
      "eval_runtime": 0.629,
      "eval_samples_per_second": 472.143,
      "eval_steps_per_second": 1.59,
      "step": 102
    },
    {
      "epoch": 52.0,
      "grad_norm": 0.5312138795852661,
      "learning_rate": 0.00029804102950346334,
      "loss": 0.2957,
      "step": 104
    },
    {
      "epoch": 52.0,
      "eval_loss": 0.3266555964946747,
      "eval_runtime": 0.6431,
      "eval_samples_per_second": 461.793,
      "eval_steps_per_second": 1.555,
      "step": 104
    },
    {
      "epoch": 53.0,
      "grad_norm": 0.5349692702293396,
      "learning_rate": 0.0002979643885764741,
      "loss": 0.2975,
      "step": 106
    },
    {
      "epoch": 53.0,
      "eval_loss": 0.3233616352081299,
      "eval_runtime": 0.8057,
      "eval_samples_per_second": 368.612,
      "eval_steps_per_second": 1.241,
      "step": 106
    },
    {
      "epoch": 54.0,
      "grad_norm": 0.36133870482444763,
      "learning_rate": 0.00029788628730070533,
      "loss": 0.2957,
      "step": 108
    },
    {
      "epoch": 54.0,
      "eval_loss": 0.323390930891037,
      "eval_runtime": 0.6815,
      "eval_samples_per_second": 435.829,
      "eval_steps_per_second": 1.467,
      "step": 108
    },
    {
      "epoch": 55.0,
      "grad_norm": 0.4435558021068573,
      "learning_rate": 0.00029780672644698494,
      "loss": 0.2956,
      "step": 110
    },
    {
      "epoch": 55.0,
      "eval_loss": 0.3230491578578949,
      "eval_runtime": 0.6254,
      "eval_samples_per_second": 474.93,
      "eval_steps_per_second": 1.599,
      "step": 110
    },
    {
      "epoch": 56.0,
      "grad_norm": 0.4963320791721344,
      "learning_rate": 0.0002977257068005465,
      "loss": 0.2927,
      "step": 112
    },
    {
      "epoch": 56.0,
      "eval_loss": 0.321744829416275,
      "eval_runtime": 0.7031,
      "eval_samples_per_second": 422.43,
      "eval_steps_per_second": 1.422,
      "step": 112
    },
    {
      "epoch": 57.0,
      "grad_norm": 0.47572246193885803,
      "learning_rate": 0.0002976432291610213,
      "loss": 0.2945,
      "step": 114
    },
    {
      "epoch": 57.0,
      "eval_loss": 0.32216939330101013,
      "eval_runtime": 0.77,
      "eval_samples_per_second": 385.727,
      "eval_steps_per_second": 1.299,
      "step": 114
    },
    {
      "epoch": 58.0,
      "grad_norm": 0.4532373547554016,
      "learning_rate": 0.00029755929434243023,
      "loss": 0.2912,
      "step": 116
    },
    {
      "epoch": 58.0,
      "eval_loss": 0.32353639602661133,
      "eval_runtime": 0.7486,
      "eval_samples_per_second": 396.744,
      "eval_steps_per_second": 1.336,
      "step": 116
    },
    {
      "epoch": 59.0,
      "grad_norm": 0.37667471170425415,
      "learning_rate": 0.00029747390317317603,
      "loss": 0.2892,
      "step": 118
    },
    {
      "epoch": 59.0,
      "eval_loss": 0.3225230276584625,
      "eval_runtime": 0.6999,
      "eval_samples_per_second": 424.335,
      "eval_steps_per_second": 1.429,
      "step": 118
    },
    {
      "epoch": 60.0,
      "grad_norm": 0.48530539870262146,
      "learning_rate": 0.0002973870564960352,
      "loss": 0.2874,
      "step": 120
    },
    {
      "epoch": 60.0,
      "eval_loss": 0.32104259729385376,
      "eval_runtime": 0.6522,
      "eval_samples_per_second": 455.417,
      "eval_steps_per_second": 1.533,
      "step": 120
    },
    {
      "epoch": 61.0,
      "grad_norm": 0.46407485008239746,
      "learning_rate": 0.00029729875516814935,
      "loss": 0.2836,
      "step": 122
    },
    {
      "epoch": 61.0,
      "eval_loss": 0.32013222575187683,
      "eval_runtime": 0.8233,
      "eval_samples_per_second": 360.757,
      "eval_steps_per_second": 1.215,
      "step": 122
    },
    {
      "epoch": 62.0,
      "grad_norm": 0.6122850179672241,
      "learning_rate": 0.0002972090000610169,
      "loss": 0.2894,
      "step": 124
    },
    {
      "epoch": 62.0,
      "eval_loss": 0.3198564350605011,
      "eval_runtime": 0.6743,
      "eval_samples_per_second": 440.478,
      "eval_steps_per_second": 1.483,
      "step": 124
    },
    {
      "epoch": 63.0,
      "grad_norm": 0.5037839412689209,
      "learning_rate": 0.0002971177920604845,
      "loss": 0.2891,
      "step": 126
    },
    {
      "epoch": 63.0,
      "eval_loss": 0.31861403584480286,
      "eval_runtime": 0.6524,
      "eval_samples_per_second": 455.235,
      "eval_steps_per_second": 1.533,
      "step": 126
    },
    {
      "epoch": 64.0,
      "grad_norm": 0.4292280972003937,
      "learning_rate": 0.00029702513206673827,
      "loss": 0.2895,
      "step": 128
    },
    {
      "epoch": 64.0,
      "eval_loss": 0.32153990864753723,
      "eval_runtime": 0.6418,
      "eval_samples_per_second": 462.761,
      "eval_steps_per_second": 1.558,
      "step": 128
    },
    {
      "epoch": 65.0,
      "grad_norm": 0.4504665732383728,
      "learning_rate": 0.000296931020994295,
      "loss": 0.2829,
      "step": 130
    },
    {
      "epoch": 65.0,
      "eval_loss": 0.3232710063457489,
      "eval_runtime": 0.7745,
      "eval_samples_per_second": 383.487,
      "eval_steps_per_second": 1.291,
      "step": 130
    },
    {
      "epoch": 66.0,
      "grad_norm": 0.6219745874404907,
      "learning_rate": 0.000296835459771993,
      "loss": 0.2859,
      "step": 132
    },
    {
      "epoch": 66.0,
      "eval_loss": 0.3181236982345581,
      "eval_runtime": 0.7429,
      "eval_samples_per_second": 399.762,
      "eval_steps_per_second": 1.346,
      "step": 132
    },
    {
      "epoch": 67.0,
      "grad_norm": 0.37324589490890503,
      "learning_rate": 0.0002967384493429829,
      "loss": 0.2798,
      "step": 134
    },
    {
      "epoch": 67.0,
      "eval_loss": 0.31664231419563293,
      "eval_runtime": 0.6029,
      "eval_samples_per_second": 492.583,
      "eval_steps_per_second": 1.659,
      "step": 134
    },
    {
      "epoch": 68.0,
      "grad_norm": 0.48705121874809265,
      "learning_rate": 0.0002966399906647185,
      "loss": 0.281,
      "step": 136
    },
    {
      "epoch": 68.0,
      "eval_loss": 0.3166642189025879,
      "eval_runtime": 0.7131,
      "eval_samples_per_second": 416.468,
      "eval_steps_per_second": 1.402,
      "step": 136
    },
    {
      "epoch": 69.0,
      "grad_norm": 0.4066949188709259,
      "learning_rate": 0.0002965400847089472,
      "loss": 0.2822,
      "step": 138
    },
    {
      "epoch": 69.0,
      "eval_loss": 0.3151547312736511,
      "eval_runtime": 0.7062,
      "eval_samples_per_second": 420.548,
      "eval_steps_per_second": 1.416,
      "step": 138
    },
    {
      "epoch": 70.0,
      "grad_norm": 0.3984168469905853,
      "learning_rate": 0.00029643873246170045,
      "loss": 0.2801,
      "step": 140
    },
    {
      "epoch": 70.0,
      "eval_loss": 0.314954549074173,
      "eval_runtime": 0.6916,
      "eval_samples_per_second": 429.454,
      "eval_steps_per_second": 1.446,
      "step": 140
    },
    {
      "epoch": 71.0,
      "grad_norm": 0.4139808118343353,
      "learning_rate": 0.000296335934923284,
      "loss": 0.2815,
      "step": 142
    },
    {
      "epoch": 71.0,
      "eval_loss": 0.31498512625694275,
      "eval_runtime": 0.6363,
      "eval_samples_per_second": 466.781,
      "eval_steps_per_second": 1.572,
      "step": 142
    },
    {
      "epoch": 72.0,
      "grad_norm": 0.4868161678314209,
      "learning_rate": 0.0002962316931082681,
      "loss": 0.2781,
      "step": 144
    },
    {
      "epoch": 72.0,
      "eval_loss": 0.31632447242736816,
      "eval_runtime": 0.7076,
      "eval_samples_per_second": 419.738,
      "eval_steps_per_second": 1.413,
      "step": 144
    },
    {
      "epoch": 73.0,
      "grad_norm": 0.41822776198387146,
      "learning_rate": 0.0002961260080454773,
      "loss": 0.2771,
      "step": 146
    },
    {
      "epoch": 73.0,
      "eval_loss": 0.31577062606811523,
      "eval_runtime": 0.6483,
      "eval_samples_per_second": 458.136,
      "eval_steps_per_second": 1.543,
      "step": 146
    },
    {
      "epoch": 74.0,
      "grad_norm": 0.5603657960891724,
      "learning_rate": 0.0002960188807779805,
      "loss": 0.2732,
      "step": 148
    },
    {
      "epoch": 74.0,
      "eval_loss": 0.31435203552246094,
      "eval_runtime": 0.691,
      "eval_samples_per_second": 429.801,
      "eval_steps_per_second": 1.447,
      "step": 148
    },
    {
      "epoch": 75.0,
      "grad_norm": 0.5573254227638245,
      "learning_rate": 0.0002959103123630807,
      "loss": 0.2766,
      "step": 150
    },
    {
      "epoch": 75.0,
      "eval_loss": 0.3145710527896881,
      "eval_runtime": 0.5527,
      "eval_samples_per_second": 537.332,
      "eval_steps_per_second": 1.809,
      "step": 150
    },
    {
      "epoch": 76.0,
      "grad_norm": 0.4642890393733978,
      "learning_rate": 0.0002958003038723042,
      "loss": 0.2798,
      "step": 152
    },
    {
      "epoch": 76.0,
      "eval_loss": 0.3158363401889801,
      "eval_runtime": 0.6733,
      "eval_samples_per_second": 441.12,
      "eval_steps_per_second": 1.485,
      "step": 152
    },
    {
      "epoch": 77.0,
      "grad_norm": 0.4938775300979614,
      "learning_rate": 0.00029568885639139053,
      "loss": 0.2753,
      "step": 154
    },
    {
      "epoch": 77.0,
      "eval_loss": 0.31477540731430054,
      "eval_runtime": 0.7417,
      "eval_samples_per_second": 400.422,
      "eval_steps_per_second": 1.348,
      "step": 154
    },
    {
      "epoch": 78.0,
      "grad_norm": 0.39144381880760193,
      "learning_rate": 0.00029557597102028123,
      "loss": 0.2732,
      "step": 156
    },
    {
      "epoch": 78.0,
      "eval_loss": 0.31341299414634705,
      "eval_runtime": 0.7854,
      "eval_samples_per_second": 378.137,
      "eval_steps_per_second": 1.273,
      "step": 156
    },
    {
      "epoch": 79.0,
      "grad_norm": 0.4929453432559967,
      "learning_rate": 0.00029546164887310933,
      "loss": 0.2711,
      "step": 158
    },
    {
      "epoch": 79.0,
      "eval_loss": 0.31385812163352966,
      "eval_runtime": 0.6657,
      "eval_samples_per_second": 446.114,
      "eval_steps_per_second": 1.502,
      "step": 158
    },
    {
      "epoch": 80.0,
      "grad_norm": 0.40074649453163147,
      "learning_rate": 0.0002953458910781883,
      "loss": 0.2677,
      "step": 160
    },
    {
      "epoch": 80.0,
      "eval_loss": 0.3135402500629425,
      "eval_runtime": 0.6596,
      "eval_samples_per_second": 450.303,
      "eval_steps_per_second": 1.516,
      "step": 160
    },
    {
      "epoch": 81.0,
      "grad_norm": 0.41622215509414673,
      "learning_rate": 0.0002952286987780008,
      "loss": 0.2668,
      "step": 162
    },
    {
      "epoch": 81.0,
      "eval_loss": 0.31341665983200073,
      "eval_runtime": 0.5644,
      "eval_samples_per_second": 526.236,
      "eval_steps_per_second": 1.772,
      "step": 162
    },
    {
      "epoch": 82.0,
      "grad_norm": 0.5020734667778015,
      "learning_rate": 0.0002951100731291876,
      "loss": 0.2734,
      "step": 164
    },
    {
      "epoch": 82.0,
      "eval_loss": 0.3154440224170685,
      "eval_runtime": 0.7427,
      "eval_samples_per_second": 399.916,
      "eval_steps_per_second": 1.347,
      "step": 164
    },
    {
      "epoch": 83.0,
      "grad_norm": 0.7120861411094666,
      "learning_rate": 0.0002949900153025359,
      "loss": 0.2712,
      "step": 166
    },
    {
      "epoch": 83.0,
      "eval_loss": 0.3155328929424286,
      "eval_runtime": 0.734,
      "eval_samples_per_second": 404.611,
      "eval_steps_per_second": 1.362,
      "step": 166
    },
    {
      "epoch": 84.0,
      "grad_norm": 0.5501344799995422,
      "learning_rate": 0.00029486852648296806,
      "loss": 0.2705,
      "step": 168
    },
    {
      "epoch": 84.0,
      "eval_loss": 0.31369757652282715,
      "eval_runtime": 0.6073,
      "eval_samples_per_second": 489.085,
      "eval_steps_per_second": 1.647,
      "step": 168
    },
    {
      "epoch": 85.0,
      "grad_norm": 0.3931790292263031,
      "learning_rate": 0.00029474560786952957,
      "loss": 0.273,
      "step": 170
    },
    {
      "epoch": 85.0,
      "eval_loss": 0.3132490813732147,
      "eval_runtime": 0.6683,
      "eval_samples_per_second": 444.398,
      "eval_steps_per_second": 1.496,
      "step": 170
    },
    {
      "epoch": 86.0,
      "grad_norm": 0.4931921064853668,
      "learning_rate": 0.00029462126067537756,
      "loss": 0.2713,
      "step": 172
    },
    {
      "epoch": 86.0,
      "eval_loss": 0.3119468688964844,
      "eval_runtime": 0.5093,
      "eval_samples_per_second": 583.176,
      "eval_steps_per_second": 1.964,
      "step": 172
    },
    {
      "epoch": 87.0,
      "grad_norm": 0.5810508131980896,
      "learning_rate": 0.00029449548612776866,
      "loss": 0.2665,
      "step": 174
    },
    {
      "epoch": 87.0,
      "eval_loss": 0.31254178285598755,
      "eval_runtime": 0.7282,
      "eval_samples_per_second": 407.876,
      "eval_steps_per_second": 1.373,
      "step": 174
    },
    {
      "epoch": 88.0,
      "grad_norm": 0.6227275729179382,
      "learning_rate": 0.00029436828546804686,
      "loss": 0.2719,
      "step": 176
    },
    {
      "epoch": 88.0,
      "eval_loss": 0.3148057162761688,
      "eval_runtime": 0.7497,
      "eval_samples_per_second": 396.169,
      "eval_steps_per_second": 1.334,
      "step": 176
    },
    {
      "epoch": 89.0,
      "grad_norm": 0.6895466446876526,
      "learning_rate": 0.0002942396599516314,
      "loss": 0.2666,
      "step": 178
    },
    {
      "epoch": 89.0,
      "eval_loss": 0.31230518221855164,
      "eval_runtime": 0.6871,
      "eval_samples_per_second": 432.239,
      "eval_steps_per_second": 1.455,
      "step": 178
    },
    {
      "epoch": 90.0,
      "grad_norm": 0.6018730401992798,
      "learning_rate": 0.0002941096108480041,
      "loss": 0.2641,
      "step": 180
    },
    {
      "epoch": 90.0,
      "eval_loss": 0.3139157295227051,
      "eval_runtime": 0.6814,
      "eval_samples_per_second": 435.86,
      "eval_steps_per_second": 1.468,
      "step": 180
    },
    {
      "epoch": 91.0,
      "grad_norm": 0.5773816704750061,
      "learning_rate": 0.0002939781394406971,
      "loss": 0.2679,
      "step": 182
    },
    {
      "epoch": 91.0,
      "eval_loss": 0.314519464969635,
      "eval_runtime": 0.6242,
      "eval_samples_per_second": 475.778,
      "eval_steps_per_second": 1.602,
      "step": 182
    },
    {
      "epoch": 92.0,
      "grad_norm": 0.430232971906662,
      "learning_rate": 0.00029384524702728013,
      "loss": 0.2636,
      "step": 184
    },
    {
      "epoch": 92.0,
      "eval_loss": 0.3126896917819977,
      "eval_runtime": 0.686,
      "eval_samples_per_second": 432.948,
      "eval_steps_per_second": 1.458,
      "step": 184
    },
    {
      "epoch": 93.0,
      "grad_norm": 0.5966992974281311,
      "learning_rate": 0.0002937109349193477,
      "loss": 0.2662,
      "step": 186
    },
    {
      "epoch": 93.0,
      "eval_loss": 0.3127867877483368,
      "eval_runtime": 0.704,
      "eval_samples_per_second": 421.861,
      "eval_steps_per_second": 1.42,
      "step": 186
    },
    {
      "epoch": 94.0,
      "grad_norm": 0.5436354875564575,
      "learning_rate": 0.000293575204442506,
      "loss": 0.2625,
      "step": 188
    },
    {
      "epoch": 94.0,
      "eval_loss": 0.3121253550052643,
      "eval_runtime": 0.6839,
      "eval_samples_per_second": 434.267,
      "eval_steps_per_second": 1.462,
      "step": 188
    },
    {
      "epoch": 95.0,
      "grad_norm": 0.6525291800498962,
      "learning_rate": 0.00029343805693636017,
      "loss": 0.2643,
      "step": 190
    },
    {
      "epoch": 95.0,
      "eval_loss": 0.30872872471809387,
      "eval_runtime": 0.7643,
      "eval_samples_per_second": 388.597,
      "eval_steps_per_second": 1.308,
      "step": 190
    },
    {
      "epoch": 96.0,
      "grad_norm": 0.41923660039901733,
      "learning_rate": 0.0002932994937545007,
      "loss": 0.2605,
      "step": 192
    },
    {
      "epoch": 96.0,
      "eval_loss": 0.30773305892944336,
      "eval_runtime": 0.621,
      "eval_samples_per_second": 478.226,
      "eval_steps_per_second": 1.61,
      "step": 192
    },
    {
      "epoch": 97.0,
      "grad_norm": 0.49382027983665466,
      "learning_rate": 0.0002931595162644901,
      "loss": 0.2613,
      "step": 194
    },
    {
      "epoch": 97.0,
      "eval_loss": 0.3081483542919159,
      "eval_runtime": 0.5459,
      "eval_samples_per_second": 544.096,
      "eval_steps_per_second": 1.832,
      "step": 194
    },
    {
      "epoch": 98.0,
      "grad_norm": 0.3870932459831238,
      "learning_rate": 0.0002930181258478499,
      "loss": 0.2655,
      "step": 196
    },
    {
      "epoch": 98.0,
      "eval_loss": 0.3087702691555023,
      "eval_runtime": 0.7521,
      "eval_samples_per_second": 394.875,
      "eval_steps_per_second": 1.33,
      "step": 196
    },
    {
      "epoch": 99.0,
      "grad_norm": 0.3989209532737732,
      "learning_rate": 0.00029287532390004633,
      "loss": 0.2605,
      "step": 198
    },
    {
      "epoch": 99.0,
      "eval_loss": 0.30880430340766907,
      "eval_runtime": 0.6471,
      "eval_samples_per_second": 458.95,
      "eval_steps_per_second": 1.545,
      "step": 198
    },
    {
      "epoch": 100.0,
      "grad_norm": 0.6288752555847168,
      "learning_rate": 0.00029273111183047697,
      "loss": 0.2624,
      "step": 200
    },
    {
      "epoch": 100.0,
      "eval_loss": 0.30914193391799927,
      "eval_runtime": 0.7927,
      "eval_samples_per_second": 374.659,
      "eval_steps_per_second": 1.261,
      "step": 200
    },
    {
      "epoch": 101.0,
      "grad_norm": 0.7080681324005127,
      "learning_rate": 0.0002925854910624568,
      "loss": 0.2532,
      "step": 202
    },
    {
      "epoch": 101.0,
      "eval_loss": 0.3055001199245453,
      "eval_runtime": 0.6808,
      "eval_samples_per_second": 436.251,
      "eval_steps_per_second": 1.469,
      "step": 202
    },
    {
      "epoch": 102.0,
      "grad_norm": 0.6294286251068115,
      "learning_rate": 0.00029243846303320386,
      "loss": 0.2603,
      "step": 204
    },
    {
      "epoch": 102.0,
      "eval_loss": 0.3056044280529022,
      "eval_runtime": 0.7558,
      "eval_samples_per_second": 392.961,
      "eval_steps_per_second": 1.323,
      "step": 204
    },
    {
      "epoch": 103.0,
      "grad_norm": 0.6036813259124756,
      "learning_rate": 0.0002922900291938255,
      "loss": 0.2576,
      "step": 206
    },
    {
      "epoch": 103.0,
      "eval_loss": 0.30621686577796936,
      "eval_runtime": 0.7737,
      "eval_samples_per_second": 383.867,
      "eval_steps_per_second": 1.292,
      "step": 206
    },
    {
      "epoch": 104.0,
      "grad_norm": 0.44163164496421814,
      "learning_rate": 0.00029214019100930384,
      "loss": 0.2532,
      "step": 208
    },
    {
      "epoch": 104.0,
      "eval_loss": 0.305007666349411,
      "eval_runtime": 0.7019,
      "eval_samples_per_second": 423.12,
      "eval_steps_per_second": 1.425,
      "step": 208
    }
  ],
  "logging_steps": 500,
  "max_steps": 2000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1000,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 10,
        "early_stopping_threshold": 1e-05
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1925114350927872.0,
  "train_batch_size": 512,
  "trial_name": null,
  "trial_params": null
}