CocoRoF's picture
Training in progress, step 2282, checkpoint
1e2c9c5 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9997672540080229,
"eval_steps": 1000,
"global_step": 2282,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0021905505127257294,
"grad_norm": 18.78125,
"learning_rate": 4.366812227074236e-07,
"loss": 24.6045,
"step": 5
},
{
"epoch": 0.004381101025451459,
"grad_norm": 9.546875,
"learning_rate": 8.733624454148472e-07,
"loss": 24.1,
"step": 10
},
{
"epoch": 0.006571651538177188,
"grad_norm": 11.9140625,
"learning_rate": 1.3100436681222709e-06,
"loss": 24.639,
"step": 15
},
{
"epoch": 0.008762202050902917,
"grad_norm": 12.515625,
"learning_rate": 1.7467248908296944e-06,
"loss": 25.1782,
"step": 20
},
{
"epoch": 0.010952752563628647,
"grad_norm": 13.0,
"learning_rate": 2.183406113537118e-06,
"loss": 24.4435,
"step": 25
},
{
"epoch": 0.013143303076354376,
"grad_norm": 9.7421875,
"learning_rate": 2.6200873362445417e-06,
"loss": 24.4044,
"step": 30
},
{
"epoch": 0.015333853589080106,
"grad_norm": 11.0859375,
"learning_rate": 3.0567685589519653e-06,
"loss": 24.7167,
"step": 35
},
{
"epoch": 0.017524404101805835,
"grad_norm": 10.90625,
"learning_rate": 3.493449781659389e-06,
"loss": 25.1666,
"step": 40
},
{
"epoch": 0.019714954614531564,
"grad_norm": 12.1171875,
"learning_rate": 3.930131004366812e-06,
"loss": 24.936,
"step": 45
},
{
"epoch": 0.021905505127257294,
"grad_norm": 23.40625,
"learning_rate": 4.366812227074236e-06,
"loss": 25.4893,
"step": 50
},
{
"epoch": 0.024096055639983023,
"grad_norm": 15.28125,
"learning_rate": 4.80349344978166e-06,
"loss": 25.2967,
"step": 55
},
{
"epoch": 0.026286606152708752,
"grad_norm": 18.9375,
"learning_rate": 5.2401746724890834e-06,
"loss": 26.1209,
"step": 60
},
{
"epoch": 0.028477156665434482,
"grad_norm": 13.75,
"learning_rate": 5.676855895196507e-06,
"loss": 26.7533,
"step": 65
},
{
"epoch": 0.03066770717816021,
"grad_norm": 12.7421875,
"learning_rate": 6.1135371179039305e-06,
"loss": 26.2389,
"step": 70
},
{
"epoch": 0.03285825769088594,
"grad_norm": 24.5,
"learning_rate": 6.550218340611354e-06,
"loss": 26.8494,
"step": 75
},
{
"epoch": 0.03504880820361167,
"grad_norm": 26.9375,
"learning_rate": 6.986899563318778e-06,
"loss": 26.7848,
"step": 80
},
{
"epoch": 0.0372393587163374,
"grad_norm": 21.078125,
"learning_rate": 7.423580786026201e-06,
"loss": 27.3516,
"step": 85
},
{
"epoch": 0.03942990922906313,
"grad_norm": 16.703125,
"learning_rate": 7.860262008733624e-06,
"loss": 27.2862,
"step": 90
},
{
"epoch": 0.04162045974178886,
"grad_norm": 39.34375,
"learning_rate": 8.296943231441049e-06,
"loss": 26.7809,
"step": 95
},
{
"epoch": 0.04381101025451459,
"grad_norm": 49.65625,
"learning_rate": 8.733624454148473e-06,
"loss": 26.86,
"step": 100
},
{
"epoch": 0.04600156076724032,
"grad_norm": 42.9375,
"learning_rate": 9.170305676855896e-06,
"loss": 27.0533,
"step": 105
},
{
"epoch": 0.048192111279966046,
"grad_norm": 43.65625,
"learning_rate": 9.60698689956332e-06,
"loss": 27.3038,
"step": 110
},
{
"epoch": 0.050382661792691776,
"grad_norm": 18.859375,
"learning_rate": 1.0043668122270742e-05,
"loss": 27.2657,
"step": 115
},
{
"epoch": 0.052573212305417505,
"grad_norm": 14.890625,
"learning_rate": 1.0480349344978167e-05,
"loss": 26.6958,
"step": 120
},
{
"epoch": 0.054763762818143234,
"grad_norm": 20.515625,
"learning_rate": 1.0917030567685592e-05,
"loss": 26.9019,
"step": 125
},
{
"epoch": 0.056954313330868964,
"grad_norm": 35.46875,
"learning_rate": 1.1353711790393014e-05,
"loss": 27.1127,
"step": 130
},
{
"epoch": 0.05914486384359469,
"grad_norm": 22.765625,
"learning_rate": 1.179039301310044e-05,
"loss": 28.2583,
"step": 135
},
{
"epoch": 0.06133541435632042,
"grad_norm": 22.46875,
"learning_rate": 1.2227074235807861e-05,
"loss": 27.8035,
"step": 140
},
{
"epoch": 0.06352596486904616,
"grad_norm": 30.421875,
"learning_rate": 1.2663755458515286e-05,
"loss": 27.3668,
"step": 145
},
{
"epoch": 0.06571651538177188,
"grad_norm": 65.625,
"learning_rate": 1.3100436681222708e-05,
"loss": 28.3746,
"step": 150
},
{
"epoch": 0.06790706589449762,
"grad_norm": 19.546875,
"learning_rate": 1.3537117903930132e-05,
"loss": 29.1302,
"step": 155
},
{
"epoch": 0.07009761640722334,
"grad_norm": 44.8125,
"learning_rate": 1.3973799126637555e-05,
"loss": 29.4078,
"step": 160
},
{
"epoch": 0.07228816691994908,
"grad_norm": 30.15625,
"learning_rate": 1.4410480349344979e-05,
"loss": 29.3431,
"step": 165
},
{
"epoch": 0.0744787174326748,
"grad_norm": 37.46875,
"learning_rate": 1.4847161572052402e-05,
"loss": 29.8867,
"step": 170
},
{
"epoch": 0.07666926794540053,
"grad_norm": 43.5625,
"learning_rate": 1.5283842794759826e-05,
"loss": 29.2998,
"step": 175
},
{
"epoch": 0.07885981845812626,
"grad_norm": 42.4375,
"learning_rate": 1.5720524017467248e-05,
"loss": 29.4236,
"step": 180
},
{
"epoch": 0.081050368970852,
"grad_norm": 57.71875,
"learning_rate": 1.6157205240174673e-05,
"loss": 29.6384,
"step": 185
},
{
"epoch": 0.08324091948357772,
"grad_norm": 60.875,
"learning_rate": 1.6593886462882098e-05,
"loss": 30.576,
"step": 190
},
{
"epoch": 0.08543146999630345,
"grad_norm": 57.6875,
"learning_rate": 1.703056768558952e-05,
"loss": 29.943,
"step": 195
},
{
"epoch": 0.08762202050902917,
"grad_norm": 39.40625,
"learning_rate": 1.7467248908296945e-05,
"loss": 29.4262,
"step": 200
},
{
"epoch": 0.08981257102175491,
"grad_norm": 27.796875,
"learning_rate": 1.7903930131004367e-05,
"loss": 29.5768,
"step": 205
},
{
"epoch": 0.09200312153448063,
"grad_norm": 76.9375,
"learning_rate": 1.8340611353711792e-05,
"loss": 29.2476,
"step": 210
},
{
"epoch": 0.09419367204720637,
"grad_norm": 46.59375,
"learning_rate": 1.8777292576419214e-05,
"loss": 29.571,
"step": 215
},
{
"epoch": 0.09638422255993209,
"grad_norm": 65.8125,
"learning_rate": 1.921397379912664e-05,
"loss": 29.4724,
"step": 220
},
{
"epoch": 0.09857477307265783,
"grad_norm": 80.875,
"learning_rate": 1.965065502183406e-05,
"loss": 30.2905,
"step": 225
},
{
"epoch": 0.10076532358538355,
"grad_norm": 70.125,
"learning_rate": 1.9990258158792012e-05,
"loss": 30.0969,
"step": 230
},
{
"epoch": 0.10295587409810929,
"grad_norm": 117.75,
"learning_rate": 1.994154895275207e-05,
"loss": 30.3242,
"step": 235
},
{
"epoch": 0.10514642461083501,
"grad_norm": 47.09375,
"learning_rate": 1.989283974671213e-05,
"loss": 30.2095,
"step": 240
},
{
"epoch": 0.10733697512356075,
"grad_norm": 104.375,
"learning_rate": 1.984413054067219e-05,
"loss": 30.7099,
"step": 245
},
{
"epoch": 0.10952752563628647,
"grad_norm": 71.375,
"learning_rate": 1.979542133463225e-05,
"loss": 29.5774,
"step": 250
},
{
"epoch": 0.1117180761490122,
"grad_norm": 85.125,
"learning_rate": 1.9746712128592305e-05,
"loss": 29.1658,
"step": 255
},
{
"epoch": 0.11390862666173793,
"grad_norm": 68.6875,
"learning_rate": 1.9698002922552364e-05,
"loss": 29.9591,
"step": 260
},
{
"epoch": 0.11609917717446366,
"grad_norm": 98.3125,
"learning_rate": 1.9649293716512423e-05,
"loss": 29.6678,
"step": 265
},
{
"epoch": 0.11828972768718939,
"grad_norm": 72.625,
"learning_rate": 1.960058451047248e-05,
"loss": 29.1617,
"step": 270
},
{
"epoch": 0.12048027819991512,
"grad_norm": 41.1875,
"learning_rate": 1.955187530443254e-05,
"loss": 28.7146,
"step": 275
},
{
"epoch": 0.12267082871264084,
"grad_norm": 40.3125,
"learning_rate": 1.9503166098392598e-05,
"loss": 28.5241,
"step": 280
},
{
"epoch": 0.12486137922536658,
"grad_norm": 52.9375,
"learning_rate": 1.9454456892352657e-05,
"loss": 28.3896,
"step": 285
},
{
"epoch": 0.12705192973809232,
"grad_norm": 64.4375,
"learning_rate": 1.9405747686312716e-05,
"loss": 28.9987,
"step": 290
},
{
"epoch": 0.12924248025081803,
"grad_norm": 62.4375,
"learning_rate": 1.9357038480272775e-05,
"loss": 29.6575,
"step": 295
},
{
"epoch": 0.13143303076354376,
"grad_norm": 79.6875,
"learning_rate": 1.930832927423283e-05,
"loss": 28.972,
"step": 300
},
{
"epoch": 0.1336235812762695,
"grad_norm": 46.15625,
"learning_rate": 1.925962006819289e-05,
"loss": 29.7721,
"step": 305
},
{
"epoch": 0.13581413178899523,
"grad_norm": 50.4375,
"learning_rate": 1.921091086215295e-05,
"loss": 29.5983,
"step": 310
},
{
"epoch": 0.13800468230172094,
"grad_norm": 60.5625,
"learning_rate": 1.9162201656113005e-05,
"loss": 28.7197,
"step": 315
},
{
"epoch": 0.14019523281444668,
"grad_norm": 94.8125,
"learning_rate": 1.9113492450073065e-05,
"loss": 29.3597,
"step": 320
},
{
"epoch": 0.14238578332717242,
"grad_norm": 61.5,
"learning_rate": 1.9064783244033124e-05,
"loss": 29.0553,
"step": 325
},
{
"epoch": 0.14457633383989815,
"grad_norm": 46.34375,
"learning_rate": 1.9016074037993183e-05,
"loss": 28.7858,
"step": 330
},
{
"epoch": 0.14676688435262386,
"grad_norm": 40.84375,
"learning_rate": 1.8967364831953242e-05,
"loss": 28.9995,
"step": 335
},
{
"epoch": 0.1489574348653496,
"grad_norm": 77.0,
"learning_rate": 1.8918655625913298e-05,
"loss": 29.1814,
"step": 340
},
{
"epoch": 0.15114798537807533,
"grad_norm": 77.0625,
"learning_rate": 1.8869946419873357e-05,
"loss": 29.971,
"step": 345
},
{
"epoch": 0.15333853589080107,
"grad_norm": 78.375,
"learning_rate": 1.8821237213833417e-05,
"loss": 29.5954,
"step": 350
},
{
"epoch": 0.15552908640352678,
"grad_norm": 61.9375,
"learning_rate": 1.8772528007793472e-05,
"loss": 29.1703,
"step": 355
},
{
"epoch": 0.15771963691625251,
"grad_norm": 45.0625,
"learning_rate": 1.872381880175353e-05,
"loss": 28.6863,
"step": 360
},
{
"epoch": 0.15991018742897825,
"grad_norm": 55.0625,
"learning_rate": 1.867510959571359e-05,
"loss": 28.627,
"step": 365
},
{
"epoch": 0.162100737941704,
"grad_norm": 56.59375,
"learning_rate": 1.862640038967365e-05,
"loss": 29.2015,
"step": 370
},
{
"epoch": 0.1642912884544297,
"grad_norm": 56.5,
"learning_rate": 1.857769118363371e-05,
"loss": 29.3098,
"step": 375
},
{
"epoch": 0.16648183896715543,
"grad_norm": 68.25,
"learning_rate": 1.852898197759377e-05,
"loss": 28.7573,
"step": 380
},
{
"epoch": 0.16867238947988117,
"grad_norm": 81.25,
"learning_rate": 1.8480272771553824e-05,
"loss": 28.9687,
"step": 385
},
{
"epoch": 0.1708629399926069,
"grad_norm": 37.6875,
"learning_rate": 1.8431563565513884e-05,
"loss": 28.9996,
"step": 390
},
{
"epoch": 0.1730534905053326,
"grad_norm": 89.9375,
"learning_rate": 1.8382854359473943e-05,
"loss": 28.5982,
"step": 395
},
{
"epoch": 0.17524404101805835,
"grad_norm": 168.625,
"learning_rate": 1.8334145153434e-05,
"loss": 28.3501,
"step": 400
},
{
"epoch": 0.17743459153078409,
"grad_norm": 47.46875,
"learning_rate": 1.8285435947394058e-05,
"loss": 28.8237,
"step": 405
},
{
"epoch": 0.17962514204350982,
"grad_norm": 117.8125,
"learning_rate": 1.8236726741354117e-05,
"loss": 29.4116,
"step": 410
},
{
"epoch": 0.18181569255623553,
"grad_norm": 68.0,
"learning_rate": 1.8188017535314176e-05,
"loss": 29.0205,
"step": 415
},
{
"epoch": 0.18400624306896127,
"grad_norm": 45.125,
"learning_rate": 1.8139308329274236e-05,
"loss": 27.7405,
"step": 420
},
{
"epoch": 0.186196793581687,
"grad_norm": 188.375,
"learning_rate": 1.8090599123234295e-05,
"loss": 28.667,
"step": 425
},
{
"epoch": 0.18838734409441274,
"grad_norm": 86.25,
"learning_rate": 1.804188991719435e-05,
"loss": 29.0361,
"step": 430
},
{
"epoch": 0.19057789460713845,
"grad_norm": 69.0,
"learning_rate": 1.799318071115441e-05,
"loss": 30.2768,
"step": 435
},
{
"epoch": 0.19276844511986418,
"grad_norm": 50.1875,
"learning_rate": 1.794447150511447e-05,
"loss": 28.2305,
"step": 440
},
{
"epoch": 0.19495899563258992,
"grad_norm": 87.3125,
"learning_rate": 1.7895762299074525e-05,
"loss": 28.908,
"step": 445
},
{
"epoch": 0.19714954614531566,
"grad_norm": 58.15625,
"learning_rate": 1.7847053093034584e-05,
"loss": 28.5689,
"step": 450
},
{
"epoch": 0.19934009665804137,
"grad_norm": 70.75,
"learning_rate": 1.7798343886994643e-05,
"loss": 29.014,
"step": 455
},
{
"epoch": 0.2015306471707671,
"grad_norm": 60.875,
"learning_rate": 1.7749634680954703e-05,
"loss": 28.3204,
"step": 460
},
{
"epoch": 0.20372119768349284,
"grad_norm": 85.875,
"learning_rate": 1.7700925474914762e-05,
"loss": 28.7934,
"step": 465
},
{
"epoch": 0.20591174819621857,
"grad_norm": 39.4375,
"learning_rate": 1.7652216268874818e-05,
"loss": 28.989,
"step": 470
},
{
"epoch": 0.20810229870894428,
"grad_norm": 59.59375,
"learning_rate": 1.7603507062834877e-05,
"loss": 29.2005,
"step": 475
},
{
"epoch": 0.21029284922167002,
"grad_norm": 77.5625,
"learning_rate": 1.7554797856794936e-05,
"loss": 29.5178,
"step": 480
},
{
"epoch": 0.21248339973439576,
"grad_norm": 104.3125,
"learning_rate": 1.7506088650754992e-05,
"loss": 29.2143,
"step": 485
},
{
"epoch": 0.2146739502471215,
"grad_norm": 44.875,
"learning_rate": 1.745737944471505e-05,
"loss": 30.0929,
"step": 490
},
{
"epoch": 0.2168645007598472,
"grad_norm": 70.0625,
"learning_rate": 1.740867023867511e-05,
"loss": 29.3203,
"step": 495
},
{
"epoch": 0.21905505127257294,
"grad_norm": 53.15625,
"learning_rate": 1.735996103263517e-05,
"loss": 30.1016,
"step": 500
},
{
"epoch": 0.22124560178529867,
"grad_norm": 72.9375,
"learning_rate": 1.731125182659523e-05,
"loss": 28.0981,
"step": 505
},
{
"epoch": 0.2234361522980244,
"grad_norm": 44.34375,
"learning_rate": 1.7262542620555288e-05,
"loss": 28.6163,
"step": 510
},
{
"epoch": 0.22562670281075012,
"grad_norm": 58.5,
"learning_rate": 1.7213833414515344e-05,
"loss": 29.0698,
"step": 515
},
{
"epoch": 0.22781725332347585,
"grad_norm": 74.125,
"learning_rate": 1.7165124208475403e-05,
"loss": 28.9386,
"step": 520
},
{
"epoch": 0.2300078038362016,
"grad_norm": 35.71875,
"learning_rate": 1.7116415002435462e-05,
"loss": 28.0684,
"step": 525
},
{
"epoch": 0.23219835434892733,
"grad_norm": 62.78125,
"learning_rate": 1.7067705796395518e-05,
"loss": 28.0443,
"step": 530
},
{
"epoch": 0.23438890486165304,
"grad_norm": 80.5625,
"learning_rate": 1.7018996590355577e-05,
"loss": 29.2146,
"step": 535
},
{
"epoch": 0.23657945537437877,
"grad_norm": 102.6875,
"learning_rate": 1.6970287384315637e-05,
"loss": 29.499,
"step": 540
},
{
"epoch": 0.2387700058871045,
"grad_norm": 170.375,
"learning_rate": 1.6921578178275696e-05,
"loss": 29.1205,
"step": 545
},
{
"epoch": 0.24096055639983024,
"grad_norm": 58.875,
"learning_rate": 1.6872868972235755e-05,
"loss": 28.9004,
"step": 550
},
{
"epoch": 0.24315110691255595,
"grad_norm": 104.6875,
"learning_rate": 1.6824159766195814e-05,
"loss": 29.373,
"step": 555
},
{
"epoch": 0.2453416574252817,
"grad_norm": 74.75,
"learning_rate": 1.677545056015587e-05,
"loss": 29.4106,
"step": 560
},
{
"epoch": 0.24753220793800743,
"grad_norm": 177.875,
"learning_rate": 1.672674135411593e-05,
"loss": 28.1533,
"step": 565
},
{
"epoch": 0.24972275845073316,
"grad_norm": 154.0,
"learning_rate": 1.667803214807599e-05,
"loss": 28.1538,
"step": 570
},
{
"epoch": 0.25191330896345887,
"grad_norm": 47.59375,
"learning_rate": 1.6629322942036044e-05,
"loss": 28.6026,
"step": 575
},
{
"epoch": 0.25410385947618463,
"grad_norm": 60.5625,
"learning_rate": 1.6580613735996104e-05,
"loss": 28.6633,
"step": 580
},
{
"epoch": 0.25629440998891034,
"grad_norm": 103.375,
"learning_rate": 1.6531904529956163e-05,
"loss": 27.9769,
"step": 585
},
{
"epoch": 0.25848496050163605,
"grad_norm": 72.625,
"learning_rate": 1.6483195323916222e-05,
"loss": 28.8252,
"step": 590
},
{
"epoch": 0.2606755110143618,
"grad_norm": 289.25,
"learning_rate": 1.643448611787628e-05,
"loss": 29.3678,
"step": 595
},
{
"epoch": 0.2628660615270875,
"grad_norm": 63.625,
"learning_rate": 1.6385776911836337e-05,
"loss": 28.1857,
"step": 600
},
{
"epoch": 0.26505661203981323,
"grad_norm": 65.8125,
"learning_rate": 1.6337067705796396e-05,
"loss": 29.0309,
"step": 605
},
{
"epoch": 0.267247162552539,
"grad_norm": 302.25,
"learning_rate": 1.6288358499756456e-05,
"loss": 28.1751,
"step": 610
},
{
"epoch": 0.2694377130652647,
"grad_norm": 53.21875,
"learning_rate": 1.623964929371651e-05,
"loss": 28.5572,
"step": 615
},
{
"epoch": 0.27162826357799047,
"grad_norm": 74.125,
"learning_rate": 1.619094008767657e-05,
"loss": 28.3836,
"step": 620
},
{
"epoch": 0.2738188140907162,
"grad_norm": 79.875,
"learning_rate": 1.614223088163663e-05,
"loss": 28.5693,
"step": 625
},
{
"epoch": 0.2760093646034419,
"grad_norm": 67.625,
"learning_rate": 1.609352167559669e-05,
"loss": 28.4494,
"step": 630
},
{
"epoch": 0.27819991511616765,
"grad_norm": 42.09375,
"learning_rate": 1.604481246955675e-05,
"loss": 28.2104,
"step": 635
},
{
"epoch": 0.28039046562889336,
"grad_norm": 64.125,
"learning_rate": 1.5996103263516808e-05,
"loss": 28.4881,
"step": 640
},
{
"epoch": 0.28258101614161907,
"grad_norm": 66.25,
"learning_rate": 1.5947394057476863e-05,
"loss": 28.476,
"step": 645
},
{
"epoch": 0.28477156665434483,
"grad_norm": 51.4375,
"learning_rate": 1.5898684851436923e-05,
"loss": 28.3387,
"step": 650
},
{
"epoch": 0.28696211716707054,
"grad_norm": 124.1875,
"learning_rate": 1.5849975645396982e-05,
"loss": 28.5756,
"step": 655
},
{
"epoch": 0.2891526676797963,
"grad_norm": 44.90625,
"learning_rate": 1.5801266439357038e-05,
"loss": 28.3304,
"step": 660
},
{
"epoch": 0.291343218192522,
"grad_norm": 52.4375,
"learning_rate": 1.5752557233317097e-05,
"loss": 28.2227,
"step": 665
},
{
"epoch": 0.2935337687052477,
"grad_norm": 177.625,
"learning_rate": 1.5703848027277156e-05,
"loss": 27.6904,
"step": 670
},
{
"epoch": 0.2957243192179735,
"grad_norm": 53.5625,
"learning_rate": 1.5655138821237215e-05,
"loss": 28.4781,
"step": 675
},
{
"epoch": 0.2979148697306992,
"grad_norm": 88.75,
"learning_rate": 1.5606429615197275e-05,
"loss": 28.3335,
"step": 680
},
{
"epoch": 0.3001054202434249,
"grad_norm": 48.5,
"learning_rate": 1.5557720409157334e-05,
"loss": 27.4771,
"step": 685
},
{
"epoch": 0.30229597075615067,
"grad_norm": 140.375,
"learning_rate": 1.550901120311739e-05,
"loss": 27.8058,
"step": 690
},
{
"epoch": 0.3044865212688764,
"grad_norm": 61.96875,
"learning_rate": 1.546030199707745e-05,
"loss": 27.8123,
"step": 695
},
{
"epoch": 0.30667707178160214,
"grad_norm": 49.125,
"learning_rate": 1.5411592791037505e-05,
"loss": 27.3922,
"step": 700
},
{
"epoch": 0.30886762229432785,
"grad_norm": 39.28125,
"learning_rate": 1.5362883584997564e-05,
"loss": 28.6846,
"step": 705
},
{
"epoch": 0.31105817280705356,
"grad_norm": 61.0,
"learning_rate": 1.5314174378957623e-05,
"loss": 27.2469,
"step": 710
},
{
"epoch": 0.3132487233197793,
"grad_norm": 62.125,
"learning_rate": 1.5265465172917682e-05,
"loss": 28.4207,
"step": 715
},
{
"epoch": 0.31543927383250503,
"grad_norm": 74.5625,
"learning_rate": 1.5216755966877742e-05,
"loss": 28.4193,
"step": 720
},
{
"epoch": 0.31762982434523074,
"grad_norm": 54.71875,
"learning_rate": 1.51680467608378e-05,
"loss": 28.2975,
"step": 725
},
{
"epoch": 0.3198203748579565,
"grad_norm": 96.5,
"learning_rate": 1.5119337554797857e-05,
"loss": 27.9856,
"step": 730
},
{
"epoch": 0.3220109253706822,
"grad_norm": 52.09375,
"learning_rate": 1.5070628348757916e-05,
"loss": 27.6961,
"step": 735
},
{
"epoch": 0.324201475883408,
"grad_norm": 72.75,
"learning_rate": 1.5021919142717975e-05,
"loss": 27.8061,
"step": 740
},
{
"epoch": 0.3263920263961337,
"grad_norm": 81.0625,
"learning_rate": 1.4973209936678033e-05,
"loss": 28.0795,
"step": 745
},
{
"epoch": 0.3285825769088594,
"grad_norm": 90.4375,
"learning_rate": 1.4924500730638092e-05,
"loss": 27.7571,
"step": 750
},
{
"epoch": 0.33077312742158516,
"grad_norm": 57.65625,
"learning_rate": 1.4875791524598151e-05,
"loss": 27.1296,
"step": 755
},
{
"epoch": 0.33296367793431086,
"grad_norm": 91.3125,
"learning_rate": 1.4827082318558209e-05,
"loss": 28.0927,
"step": 760
},
{
"epoch": 0.3351542284470366,
"grad_norm": 52.34375,
"learning_rate": 1.4778373112518268e-05,
"loss": 27.571,
"step": 765
},
{
"epoch": 0.33734477895976234,
"grad_norm": 69.875,
"learning_rate": 1.4729663906478327e-05,
"loss": 27.6735,
"step": 770
},
{
"epoch": 0.33953532947248805,
"grad_norm": 213.5,
"learning_rate": 1.4680954700438383e-05,
"loss": 27.7685,
"step": 775
},
{
"epoch": 0.3417258799852138,
"grad_norm": 53.3125,
"learning_rate": 1.4632245494398442e-05,
"loss": 27.675,
"step": 780
},
{
"epoch": 0.3439164304979395,
"grad_norm": 72.6875,
"learning_rate": 1.4583536288358501e-05,
"loss": 27.887,
"step": 785
},
{
"epoch": 0.3461069810106652,
"grad_norm": 70.375,
"learning_rate": 1.4534827082318559e-05,
"loss": 26.8192,
"step": 790
},
{
"epoch": 0.348297531523391,
"grad_norm": 91.375,
"learning_rate": 1.4486117876278618e-05,
"loss": 27.3069,
"step": 795
},
{
"epoch": 0.3504880820361167,
"grad_norm": 46.5,
"learning_rate": 1.4437408670238677e-05,
"loss": 27.452,
"step": 800
},
{
"epoch": 0.3526786325488424,
"grad_norm": 69.0625,
"learning_rate": 1.4388699464198735e-05,
"loss": 27.5778,
"step": 805
},
{
"epoch": 0.35486918306156817,
"grad_norm": 119.5625,
"learning_rate": 1.4339990258158794e-05,
"loss": 27.2898,
"step": 810
},
{
"epoch": 0.3570597335742939,
"grad_norm": 69.9375,
"learning_rate": 1.429128105211885e-05,
"loss": 28.0446,
"step": 815
},
{
"epoch": 0.35925028408701964,
"grad_norm": 51.15625,
"learning_rate": 1.4242571846078909e-05,
"loss": 27.8762,
"step": 820
},
{
"epoch": 0.36144083459974535,
"grad_norm": 44.78125,
"learning_rate": 1.4193862640038968e-05,
"loss": 27.4756,
"step": 825
},
{
"epoch": 0.36363138511247106,
"grad_norm": 55.28125,
"learning_rate": 1.4145153433999026e-05,
"loss": 27.2165,
"step": 830
},
{
"epoch": 0.3658219356251968,
"grad_norm": 73.5625,
"learning_rate": 1.4096444227959085e-05,
"loss": 27.3859,
"step": 835
},
{
"epoch": 0.36801248613792253,
"grad_norm": 64.875,
"learning_rate": 1.4047735021919144e-05,
"loss": 27.2625,
"step": 840
},
{
"epoch": 0.37020303665064824,
"grad_norm": 54.1875,
"learning_rate": 1.3999025815879202e-05,
"loss": 27.9675,
"step": 845
},
{
"epoch": 0.372393587163374,
"grad_norm": 57.40625,
"learning_rate": 1.3950316609839261e-05,
"loss": 28.0232,
"step": 850
},
{
"epoch": 0.3745841376760997,
"grad_norm": 97.625,
"learning_rate": 1.390160740379932e-05,
"loss": 27.9414,
"step": 855
},
{
"epoch": 0.3767746881888255,
"grad_norm": 67.6875,
"learning_rate": 1.3852898197759376e-05,
"loss": 27.572,
"step": 860
},
{
"epoch": 0.3789652387015512,
"grad_norm": 55.03125,
"learning_rate": 1.3804188991719435e-05,
"loss": 28.1618,
"step": 865
},
{
"epoch": 0.3811557892142769,
"grad_norm": 180.625,
"learning_rate": 1.3755479785679495e-05,
"loss": 27.3198,
"step": 870
},
{
"epoch": 0.38334633972700266,
"grad_norm": 73.625,
"learning_rate": 1.3706770579639552e-05,
"loss": 27.5609,
"step": 875
},
{
"epoch": 0.38553689023972837,
"grad_norm": 114.4375,
"learning_rate": 1.3658061373599611e-05,
"loss": 27.3752,
"step": 880
},
{
"epoch": 0.3877274407524541,
"grad_norm": 48.28125,
"learning_rate": 1.360935216755967e-05,
"loss": 27.7674,
"step": 885
},
{
"epoch": 0.38991799126517984,
"grad_norm": 126.1875,
"learning_rate": 1.3560642961519728e-05,
"loss": 27.0909,
"step": 890
},
{
"epoch": 0.39210854177790555,
"grad_norm": 63.125,
"learning_rate": 1.3511933755479787e-05,
"loss": 27.4967,
"step": 895
},
{
"epoch": 0.3942990922906313,
"grad_norm": 60.34375,
"learning_rate": 1.3463224549439847e-05,
"loss": 26.7753,
"step": 900
},
{
"epoch": 0.396489642803357,
"grad_norm": 59.375,
"learning_rate": 1.3414515343399902e-05,
"loss": 26.6501,
"step": 905
},
{
"epoch": 0.39868019331608273,
"grad_norm": 219.375,
"learning_rate": 1.3365806137359962e-05,
"loss": 27.3259,
"step": 910
},
{
"epoch": 0.4008707438288085,
"grad_norm": 59.25,
"learning_rate": 1.331709693132002e-05,
"loss": 27.6461,
"step": 915
},
{
"epoch": 0.4030612943415342,
"grad_norm": 54.625,
"learning_rate": 1.3268387725280078e-05,
"loss": 26.9045,
"step": 920
},
{
"epoch": 0.4052518448542599,
"grad_norm": 65.875,
"learning_rate": 1.3219678519240138e-05,
"loss": 26.6124,
"step": 925
},
{
"epoch": 0.4074423953669857,
"grad_norm": 56.125,
"learning_rate": 1.3170969313200197e-05,
"loss": 27.6301,
"step": 930
},
{
"epoch": 0.4096329458797114,
"grad_norm": 59.6875,
"learning_rate": 1.3122260107160254e-05,
"loss": 27.8693,
"step": 935
},
{
"epoch": 0.41182349639243715,
"grad_norm": 47.71875,
"learning_rate": 1.3073550901120314e-05,
"loss": 26.9383,
"step": 940
},
{
"epoch": 0.41401404690516286,
"grad_norm": 130.75,
"learning_rate": 1.302484169508037e-05,
"loss": 27.202,
"step": 945
},
{
"epoch": 0.41620459741788857,
"grad_norm": 48.4375,
"learning_rate": 1.2976132489040429e-05,
"loss": 27.2644,
"step": 950
},
{
"epoch": 0.41839514793061433,
"grad_norm": 57.84375,
"learning_rate": 1.2927423283000488e-05,
"loss": 26.7067,
"step": 955
},
{
"epoch": 0.42058569844334004,
"grad_norm": 65.0625,
"learning_rate": 1.2878714076960545e-05,
"loss": 27.7302,
"step": 960
},
{
"epoch": 0.42277624895606575,
"grad_norm": 58.1875,
"learning_rate": 1.2830004870920605e-05,
"loss": 27.5106,
"step": 965
},
{
"epoch": 0.4249667994687915,
"grad_norm": 105.4375,
"learning_rate": 1.2781295664880664e-05,
"loss": 27.0035,
"step": 970
},
{
"epoch": 0.4271573499815172,
"grad_norm": 59.28125,
"learning_rate": 1.2732586458840721e-05,
"loss": 26.4531,
"step": 975
},
{
"epoch": 0.429347900494243,
"grad_norm": 187.625,
"learning_rate": 1.268387725280078e-05,
"loss": 27.0814,
"step": 980
},
{
"epoch": 0.4315384510069687,
"grad_norm": 80.625,
"learning_rate": 1.263516804676084e-05,
"loss": 27.4558,
"step": 985
},
{
"epoch": 0.4337290015196944,
"grad_norm": 97.5,
"learning_rate": 1.2586458840720897e-05,
"loss": 27.2832,
"step": 990
},
{
"epoch": 0.43591955203242017,
"grad_norm": 80.8125,
"learning_rate": 1.2537749634680957e-05,
"loss": 27.6556,
"step": 995
},
{
"epoch": 0.4381101025451459,
"grad_norm": 64.9375,
"learning_rate": 1.2489040428641016e-05,
"loss": 26.8605,
"step": 1000
},
{
"epoch": 0.4381101025451459,
"eval_loss": NaN,
"eval_runtime": 245.1855,
"eval_samples_per_second": 1003.457,
"eval_steps_per_second": 31.36,
"step": 1000
},
{
"epoch": 0.4403006530578716,
"grad_norm": 48.5625,
"learning_rate": 1.2440331222601072e-05,
"loss": 26.861,
"step": 1005
},
{
"epoch": 0.44249120357059735,
"grad_norm": 61.0625,
"learning_rate": 1.2391622016561131e-05,
"loss": 26.9305,
"step": 1010
},
{
"epoch": 0.44468175408332306,
"grad_norm": 30.296875,
"learning_rate": 1.234291281052119e-05,
"loss": 27.4806,
"step": 1015
},
{
"epoch": 0.4468723045960488,
"grad_norm": 53.3125,
"learning_rate": 1.2294203604481248e-05,
"loss": 26.5761,
"step": 1020
},
{
"epoch": 0.44906285510877453,
"grad_norm": 45.53125,
"learning_rate": 1.2245494398441307e-05,
"loss": 26.5082,
"step": 1025
},
{
"epoch": 0.45125340562150024,
"grad_norm": 39.9375,
"learning_rate": 1.2196785192401366e-05,
"loss": 27.5339,
"step": 1030
},
{
"epoch": 0.453443956134226,
"grad_norm": 53.3125,
"learning_rate": 1.2148075986361424e-05,
"loss": 27.3613,
"step": 1035
},
{
"epoch": 0.4556345066469517,
"grad_norm": 60.0,
"learning_rate": 1.2099366780321483e-05,
"loss": 26.3353,
"step": 1040
},
{
"epoch": 0.4578250571596774,
"grad_norm": 52.84375,
"learning_rate": 1.2050657574281542e-05,
"loss": 26.9829,
"step": 1045
},
{
"epoch": 0.4600156076724032,
"grad_norm": 377.5,
"learning_rate": 1.2001948368241598e-05,
"loss": 26.4438,
"step": 1050
},
{
"epoch": 0.4622061581851289,
"grad_norm": 61.65625,
"learning_rate": 1.1953239162201657e-05,
"loss": 27.1729,
"step": 1055
},
{
"epoch": 0.46439670869785465,
"grad_norm": 33.28125,
"learning_rate": 1.1904529956161715e-05,
"loss": 26.2734,
"step": 1060
},
{
"epoch": 0.46658725921058036,
"grad_norm": 62.375,
"learning_rate": 1.1855820750121774e-05,
"loss": 27.0024,
"step": 1065
},
{
"epoch": 0.46877780972330607,
"grad_norm": 140.125,
"learning_rate": 1.1807111544081833e-05,
"loss": 26.6238,
"step": 1070
},
{
"epoch": 0.47096836023603184,
"grad_norm": 31.578125,
"learning_rate": 1.175840233804189e-05,
"loss": 26.7271,
"step": 1075
},
{
"epoch": 0.47315891074875754,
"grad_norm": 249.375,
"learning_rate": 1.170969313200195e-05,
"loss": 26.5206,
"step": 1080
},
{
"epoch": 0.47534946126148325,
"grad_norm": 62.15625,
"learning_rate": 1.1660983925962009e-05,
"loss": 26.3579,
"step": 1085
},
{
"epoch": 0.477540011774209,
"grad_norm": 27.203125,
"learning_rate": 1.1612274719922065e-05,
"loss": 27.2503,
"step": 1090
},
{
"epoch": 0.4797305622869347,
"grad_norm": 50.1875,
"learning_rate": 1.1563565513882124e-05,
"loss": 27.0137,
"step": 1095
},
{
"epoch": 0.4819211127996605,
"grad_norm": 44.09375,
"learning_rate": 1.1514856307842183e-05,
"loss": 26.7031,
"step": 1100
},
{
"epoch": 0.4841116633123862,
"grad_norm": 76.0625,
"learning_rate": 1.1466147101802241e-05,
"loss": 26.4805,
"step": 1105
},
{
"epoch": 0.4863022138251119,
"grad_norm": 43.4375,
"learning_rate": 1.14174378957623e-05,
"loss": 26.4291,
"step": 1110
},
{
"epoch": 0.48849276433783767,
"grad_norm": 47.15625,
"learning_rate": 1.136872868972236e-05,
"loss": 27.9976,
"step": 1115
},
{
"epoch": 0.4906833148505634,
"grad_norm": 119.0,
"learning_rate": 1.1320019483682417e-05,
"loss": 26.0667,
"step": 1120
},
{
"epoch": 0.4928738653632891,
"grad_norm": 130.5,
"learning_rate": 1.1271310277642476e-05,
"loss": 26.5342,
"step": 1125
},
{
"epoch": 0.49506441587601485,
"grad_norm": 30.65625,
"learning_rate": 1.1222601071602535e-05,
"loss": 25.9197,
"step": 1130
},
{
"epoch": 0.49725496638874056,
"grad_norm": 40.03125,
"learning_rate": 1.1173891865562591e-05,
"loss": 26.8289,
"step": 1135
},
{
"epoch": 0.4994455169014663,
"grad_norm": 70.25,
"learning_rate": 1.112518265952265e-05,
"loss": 26.8773,
"step": 1140
},
{
"epoch": 0.501636067414192,
"grad_norm": 78.75,
"learning_rate": 1.107647345348271e-05,
"loss": 26.8103,
"step": 1145
},
{
"epoch": 0.5038266179269177,
"grad_norm": 45.5,
"learning_rate": 1.1027764247442767e-05,
"loss": 26.947,
"step": 1150
},
{
"epoch": 0.5060171684396435,
"grad_norm": 108.375,
"learning_rate": 1.0979055041402826e-05,
"loss": 26.483,
"step": 1155
},
{
"epoch": 0.5082077189523693,
"grad_norm": 97.125,
"learning_rate": 1.0930345835362886e-05,
"loss": 26.3236,
"step": 1160
},
{
"epoch": 0.5103982694650949,
"grad_norm": 68.5,
"learning_rate": 1.0881636629322943e-05,
"loss": 26.8906,
"step": 1165
},
{
"epoch": 0.5125888199778207,
"grad_norm": 70.9375,
"learning_rate": 1.0832927423283002e-05,
"loss": 26.6497,
"step": 1170
},
{
"epoch": 0.5147793704905465,
"grad_norm": 129.625,
"learning_rate": 1.0784218217243058e-05,
"loss": 25.7505,
"step": 1175
},
{
"epoch": 0.5169699210032721,
"grad_norm": 38.875,
"learning_rate": 1.0735509011203117e-05,
"loss": 26.3048,
"step": 1180
},
{
"epoch": 0.5191604715159979,
"grad_norm": 60.6875,
"learning_rate": 1.0686799805163177e-05,
"loss": 25.8012,
"step": 1185
},
{
"epoch": 0.5213510220287236,
"grad_norm": 94.5,
"learning_rate": 1.0638090599123234e-05,
"loss": 26.6164,
"step": 1190
},
{
"epoch": 0.5235415725414493,
"grad_norm": 153.0,
"learning_rate": 1.0589381393083293e-05,
"loss": 26.4983,
"step": 1195
},
{
"epoch": 0.525732123054175,
"grad_norm": 32.09375,
"learning_rate": 1.0540672187043353e-05,
"loss": 26.398,
"step": 1200
},
{
"epoch": 0.5279226735669008,
"grad_norm": 46.09375,
"learning_rate": 1.049196298100341e-05,
"loss": 26.5191,
"step": 1205
},
{
"epoch": 0.5301132240796265,
"grad_norm": 37.65625,
"learning_rate": 1.044325377496347e-05,
"loss": 26.5658,
"step": 1210
},
{
"epoch": 0.5323037745923522,
"grad_norm": 87.75,
"learning_rate": 1.0394544568923529e-05,
"loss": 26.4589,
"step": 1215
},
{
"epoch": 0.534494325105078,
"grad_norm": 56.21875,
"learning_rate": 1.0345835362883584e-05,
"loss": 26.3349,
"step": 1220
},
{
"epoch": 0.5366848756178036,
"grad_norm": 83.1875,
"learning_rate": 1.0297126156843644e-05,
"loss": 26.8146,
"step": 1225
},
{
"epoch": 0.5388754261305294,
"grad_norm": 53.3125,
"learning_rate": 1.0248416950803703e-05,
"loss": 26.5554,
"step": 1230
},
{
"epoch": 0.5410659766432552,
"grad_norm": 57.25,
"learning_rate": 1.019970774476376e-05,
"loss": 26.8582,
"step": 1235
},
{
"epoch": 0.5432565271559809,
"grad_norm": 46.15625,
"learning_rate": 1.015099853872382e-05,
"loss": 26.3294,
"step": 1240
},
{
"epoch": 0.5454470776687066,
"grad_norm": 107.75,
"learning_rate": 1.0102289332683879e-05,
"loss": 26.2168,
"step": 1245
},
{
"epoch": 0.5476376281814324,
"grad_norm": 88.4375,
"learning_rate": 1.0053580126643936e-05,
"loss": 26.1914,
"step": 1250
},
{
"epoch": 0.5498281786941581,
"grad_norm": 41.53125,
"learning_rate": 1.0004870920603996e-05,
"loss": 25.9535,
"step": 1255
},
{
"epoch": 0.5520187292068838,
"grad_norm": 79.4375,
"learning_rate": 9.956161714564053e-06,
"loss": 25.7015,
"step": 1260
},
{
"epoch": 0.5542092797196095,
"grad_norm": 51.65625,
"learning_rate": 9.90745250852411e-06,
"loss": 25.5487,
"step": 1265
},
{
"epoch": 0.5563998302323353,
"grad_norm": 40.15625,
"learning_rate": 9.85874330248417e-06,
"loss": 25.0215,
"step": 1270
},
{
"epoch": 0.558590380745061,
"grad_norm": 88.1875,
"learning_rate": 9.810034096444229e-06,
"loss": 25.8621,
"step": 1275
},
{
"epoch": 0.5607809312577867,
"grad_norm": 55.21875,
"learning_rate": 9.761324890404287e-06,
"loss": 26.0285,
"step": 1280
},
{
"epoch": 0.5629714817705125,
"grad_norm": 36.21875,
"learning_rate": 9.712615684364346e-06,
"loss": 25.7962,
"step": 1285
},
{
"epoch": 0.5651620322832381,
"grad_norm": 26.390625,
"learning_rate": 9.663906478324403e-06,
"loss": 25.767,
"step": 1290
},
{
"epoch": 0.5673525827959639,
"grad_norm": 35.34375,
"learning_rate": 9.615197272284463e-06,
"loss": 26.0008,
"step": 1295
},
{
"epoch": 0.5695431333086897,
"grad_norm": 57.9375,
"learning_rate": 9.566488066244522e-06,
"loss": 25.369,
"step": 1300
},
{
"epoch": 0.5717336838214153,
"grad_norm": 44.78125,
"learning_rate": 9.51777886020458e-06,
"loss": 25.3812,
"step": 1305
},
{
"epoch": 0.5739242343341411,
"grad_norm": 47.28125,
"learning_rate": 9.469069654164637e-06,
"loss": 25.925,
"step": 1310
},
{
"epoch": 0.5761147848468668,
"grad_norm": 30.828125,
"learning_rate": 9.420360448124696e-06,
"loss": 25.9225,
"step": 1315
},
{
"epoch": 0.5783053353595926,
"grad_norm": 50.5,
"learning_rate": 9.371651242084755e-06,
"loss": 25.9306,
"step": 1320
},
{
"epoch": 0.5804958858723183,
"grad_norm": 61.0,
"learning_rate": 9.322942036044813e-06,
"loss": 26.6956,
"step": 1325
},
{
"epoch": 0.582686436385044,
"grad_norm": 42.90625,
"learning_rate": 9.27423283000487e-06,
"loss": 26.0609,
"step": 1330
},
{
"epoch": 0.5848769868977698,
"grad_norm": 61.9375,
"learning_rate": 9.22552362396493e-06,
"loss": 26.5326,
"step": 1335
},
{
"epoch": 0.5870675374104954,
"grad_norm": 58.21875,
"learning_rate": 9.176814417924989e-06,
"loss": 26.1951,
"step": 1340
},
{
"epoch": 0.5892580879232212,
"grad_norm": 46.90625,
"learning_rate": 9.128105211885046e-06,
"loss": 26.3855,
"step": 1345
},
{
"epoch": 0.591448638435947,
"grad_norm": 40.5,
"learning_rate": 9.079396005845106e-06,
"loss": 26.3907,
"step": 1350
},
{
"epoch": 0.5936391889486726,
"grad_norm": 33.34375,
"learning_rate": 9.030686799805163e-06,
"loss": 25.9174,
"step": 1355
},
{
"epoch": 0.5958297394613984,
"grad_norm": 169.5,
"learning_rate": 8.981977593765222e-06,
"loss": 25.799,
"step": 1360
},
{
"epoch": 0.5980202899741242,
"grad_norm": 75.25,
"learning_rate": 8.933268387725282e-06,
"loss": 25.4771,
"step": 1365
},
{
"epoch": 0.6002108404868498,
"grad_norm": 76.6875,
"learning_rate": 8.884559181685339e-06,
"loss": 25.7814,
"step": 1370
},
{
"epoch": 0.6024013909995756,
"grad_norm": 41.96875,
"learning_rate": 8.835849975645398e-06,
"loss": 25.9839,
"step": 1375
},
{
"epoch": 0.6045919415123013,
"grad_norm": 60.59375,
"learning_rate": 8.787140769605456e-06,
"loss": 25.6093,
"step": 1380
},
{
"epoch": 0.606782492025027,
"grad_norm": 58.46875,
"learning_rate": 8.738431563565515e-06,
"loss": 26.7859,
"step": 1385
},
{
"epoch": 0.6089730425377528,
"grad_norm": 98.4375,
"learning_rate": 8.689722357525573e-06,
"loss": 25.9515,
"step": 1390
},
{
"epoch": 0.6111635930504785,
"grad_norm": 58.28125,
"learning_rate": 8.641013151485632e-06,
"loss": 26.1697,
"step": 1395
},
{
"epoch": 0.6133541435632043,
"grad_norm": 53.59375,
"learning_rate": 8.592303945445691e-06,
"loss": 25.9631,
"step": 1400
},
{
"epoch": 0.6155446940759299,
"grad_norm": 39.28125,
"learning_rate": 8.543594739405749e-06,
"loss": 25.1179,
"step": 1405
},
{
"epoch": 0.6177352445886557,
"grad_norm": 55.3125,
"learning_rate": 8.494885533365806e-06,
"loss": 25.3475,
"step": 1410
},
{
"epoch": 0.6199257951013815,
"grad_norm": 63.9375,
"learning_rate": 8.446176327325865e-06,
"loss": 26.4752,
"step": 1415
},
{
"epoch": 0.6221163456141071,
"grad_norm": 42.125,
"learning_rate": 8.397467121285925e-06,
"loss": 26.7149,
"step": 1420
},
{
"epoch": 0.6243068961268329,
"grad_norm": 47.53125,
"learning_rate": 8.348757915245982e-06,
"loss": 25.6236,
"step": 1425
},
{
"epoch": 0.6264974466395586,
"grad_norm": 37.0625,
"learning_rate": 8.300048709206041e-06,
"loss": 26.059,
"step": 1430
},
{
"epoch": 0.6286879971522843,
"grad_norm": 59.4375,
"learning_rate": 8.251339503166099e-06,
"loss": 26.4223,
"step": 1435
},
{
"epoch": 0.6308785476650101,
"grad_norm": 45.5,
"learning_rate": 8.202630297126158e-06,
"loss": 25.2686,
"step": 1440
},
{
"epoch": 0.6330690981777358,
"grad_norm": 97.625,
"learning_rate": 8.153921091086216e-06,
"loss": 25.5556,
"step": 1445
},
{
"epoch": 0.6352596486904615,
"grad_norm": 65.5,
"learning_rate": 8.105211885046275e-06,
"loss": 25.4199,
"step": 1450
},
{
"epoch": 0.6374501992031872,
"grad_norm": 59.5,
"learning_rate": 8.056502679006332e-06,
"loss": 26.1299,
"step": 1455
},
{
"epoch": 0.639640749715913,
"grad_norm": 66.1875,
"learning_rate": 8.007793472966392e-06,
"loss": 25.1407,
"step": 1460
},
{
"epoch": 0.6418313002286387,
"grad_norm": 41.46875,
"learning_rate": 7.95908426692645e-06,
"loss": 25.0766,
"step": 1465
},
{
"epoch": 0.6440218507413644,
"grad_norm": 55.6875,
"learning_rate": 7.910375060886508e-06,
"loss": 25.9446,
"step": 1470
},
{
"epoch": 0.6462124012540902,
"grad_norm": 43.40625,
"learning_rate": 7.861665854846566e-06,
"loss": 25.4919,
"step": 1475
},
{
"epoch": 0.648402951766816,
"grad_norm": 60.90625,
"learning_rate": 7.812956648806625e-06,
"loss": 25.3909,
"step": 1480
},
{
"epoch": 0.6505935022795416,
"grad_norm": 30.421875,
"learning_rate": 7.764247442766684e-06,
"loss": 25.5321,
"step": 1485
},
{
"epoch": 0.6527840527922674,
"grad_norm": 41.5,
"learning_rate": 7.715538236726742e-06,
"loss": 25.4873,
"step": 1490
},
{
"epoch": 0.6549746033049931,
"grad_norm": 175.125,
"learning_rate": 7.666829030686801e-06,
"loss": 24.9937,
"step": 1495
},
{
"epoch": 0.6571651538177188,
"grad_norm": 60.9375,
"learning_rate": 7.6181198246468595e-06,
"loss": 25.4684,
"step": 1500
},
{
"epoch": 0.6593557043304445,
"grad_norm": 30.8125,
"learning_rate": 7.569410618606917e-06,
"loss": 25.0593,
"step": 1505
},
{
"epoch": 0.6615462548431703,
"grad_norm": 58.96875,
"learning_rate": 7.520701412566975e-06,
"loss": 24.7999,
"step": 1510
},
{
"epoch": 0.663736805355896,
"grad_norm": 24.328125,
"learning_rate": 7.471992206527035e-06,
"loss": 25.6823,
"step": 1515
},
{
"epoch": 0.6659273558686217,
"grad_norm": 100.6875,
"learning_rate": 7.423283000487093e-06,
"loss": 24.635,
"step": 1520
},
{
"epoch": 0.6681179063813475,
"grad_norm": 37.9375,
"learning_rate": 7.3745737944471505e-06,
"loss": 25.5355,
"step": 1525
},
{
"epoch": 0.6703084568940731,
"grad_norm": 37.65625,
"learning_rate": 7.32586458840721e-06,
"loss": 25.9512,
"step": 1530
},
{
"epoch": 0.6724990074067989,
"grad_norm": 35.625,
"learning_rate": 7.277155382367268e-06,
"loss": 25.8102,
"step": 1535
},
{
"epoch": 0.6746895579195247,
"grad_norm": 38.875,
"learning_rate": 7.2284461763273265e-06,
"loss": 25.373,
"step": 1540
},
{
"epoch": 0.6768801084322503,
"grad_norm": 31.3125,
"learning_rate": 7.179736970287386e-06,
"loss": 24.6079,
"step": 1545
},
{
"epoch": 0.6790706589449761,
"grad_norm": 54.75,
"learning_rate": 7.131027764247443e-06,
"loss": 25.265,
"step": 1550
},
{
"epoch": 0.6812612094577019,
"grad_norm": 44.09375,
"learning_rate": 7.082318558207502e-06,
"loss": 25.7861,
"step": 1555
},
{
"epoch": 0.6834517599704276,
"grad_norm": 24.5,
"learning_rate": 7.03360935216756e-06,
"loss": 25.4685,
"step": 1560
},
{
"epoch": 0.6856423104831533,
"grad_norm": 51.59375,
"learning_rate": 6.984900146127619e-06,
"loss": 25.6348,
"step": 1565
},
{
"epoch": 0.687832860995879,
"grad_norm": 31.6875,
"learning_rate": 6.936190940087677e-06,
"loss": 25.9039,
"step": 1570
},
{
"epoch": 0.6900234115086048,
"grad_norm": 33.53125,
"learning_rate": 6.887481734047735e-06,
"loss": 25.5847,
"step": 1575
},
{
"epoch": 0.6922139620213305,
"grad_norm": 41.4375,
"learning_rate": 6.838772528007794e-06,
"loss": 24.8405,
"step": 1580
},
{
"epoch": 0.6944045125340562,
"grad_norm": 43.34375,
"learning_rate": 6.790063321967853e-06,
"loss": 25.3519,
"step": 1585
},
{
"epoch": 0.696595063046782,
"grad_norm": 72.75,
"learning_rate": 6.74135411592791e-06,
"loss": 25.3445,
"step": 1590
},
{
"epoch": 0.6987856135595076,
"grad_norm": 21.6875,
"learning_rate": 6.6926449098879695e-06,
"loss": 25.4516,
"step": 1595
},
{
"epoch": 0.7009761640722334,
"grad_norm": 54.34375,
"learning_rate": 6.643935703848028e-06,
"loss": 25.2707,
"step": 1600
},
{
"epoch": 0.7031667145849592,
"grad_norm": 35.78125,
"learning_rate": 6.595226497808086e-06,
"loss": 25.5174,
"step": 1605
},
{
"epoch": 0.7053572650976848,
"grad_norm": 34.6875,
"learning_rate": 6.5465172917681454e-06,
"loss": 25.4142,
"step": 1610
},
{
"epoch": 0.7075478156104106,
"grad_norm": 64.5625,
"learning_rate": 6.497808085728203e-06,
"loss": 25.1589,
"step": 1615
},
{
"epoch": 0.7097383661231363,
"grad_norm": 49.84375,
"learning_rate": 6.449098879688261e-06,
"loss": 24.5071,
"step": 1620
},
{
"epoch": 0.711928916635862,
"grad_norm": 34.03125,
"learning_rate": 6.40038967364832e-06,
"loss": 25.2763,
"step": 1625
},
{
"epoch": 0.7141194671485878,
"grad_norm": 31.3125,
"learning_rate": 6.351680467608379e-06,
"loss": 24.6877,
"step": 1630
},
{
"epoch": 0.7163100176613135,
"grad_norm": 39.1875,
"learning_rate": 6.3029712615684365e-06,
"loss": 25.116,
"step": 1635
},
{
"epoch": 0.7185005681740393,
"grad_norm": 41.15625,
"learning_rate": 6.254262055528495e-06,
"loss": 24.8634,
"step": 1640
},
{
"epoch": 0.7206911186867649,
"grad_norm": 31.109375,
"learning_rate": 6.205552849488554e-06,
"loss": 24.5188,
"step": 1645
},
{
"epoch": 0.7228816691994907,
"grad_norm": 33.1875,
"learning_rate": 6.1568436434486125e-06,
"loss": 25.0722,
"step": 1650
},
{
"epoch": 0.7250722197122165,
"grad_norm": 51.1875,
"learning_rate": 6.10813443740867e-06,
"loss": 25.156,
"step": 1655
},
{
"epoch": 0.7272627702249421,
"grad_norm": 35.25,
"learning_rate": 6.059425231368729e-06,
"loss": 24.2915,
"step": 1660
},
{
"epoch": 0.7294533207376679,
"grad_norm": 50.1875,
"learning_rate": 6.010716025328788e-06,
"loss": 24.7464,
"step": 1665
},
{
"epoch": 0.7316438712503937,
"grad_norm": 48.46875,
"learning_rate": 5.962006819288846e-06,
"loss": 24.5207,
"step": 1670
},
{
"epoch": 0.7338344217631193,
"grad_norm": 26.859375,
"learning_rate": 5.913297613248905e-06,
"loss": 24.295,
"step": 1675
},
{
"epoch": 0.7360249722758451,
"grad_norm": 127.375,
"learning_rate": 5.8645884072089636e-06,
"loss": 24.5732,
"step": 1680
},
{
"epoch": 0.7382155227885708,
"grad_norm": 34.5625,
"learning_rate": 5.815879201169021e-06,
"loss": 24.7841,
"step": 1685
},
{
"epoch": 0.7404060733012965,
"grad_norm": 42.84375,
"learning_rate": 5.7671699951290795e-06,
"loss": 24.7729,
"step": 1690
},
{
"epoch": 0.7425966238140222,
"grad_norm": 29.46875,
"learning_rate": 5.718460789089139e-06,
"loss": 24.6547,
"step": 1695
},
{
"epoch": 0.744787174326748,
"grad_norm": 31.328125,
"learning_rate": 5.669751583049197e-06,
"loss": 24.9155,
"step": 1700
},
{
"epoch": 0.7469777248394737,
"grad_norm": 25.96875,
"learning_rate": 5.621042377009255e-06,
"loss": 24.943,
"step": 1705
},
{
"epoch": 0.7491682753521994,
"grad_norm": 23.265625,
"learning_rate": 5.572333170969314e-06,
"loss": 24.2714,
"step": 1710
},
{
"epoch": 0.7513588258649252,
"grad_norm": 32.40625,
"learning_rate": 5.523623964929372e-06,
"loss": 24.6501,
"step": 1715
},
{
"epoch": 0.753549376377651,
"grad_norm": 33.21875,
"learning_rate": 5.474914758889431e-06,
"loss": 24.2942,
"step": 1720
},
{
"epoch": 0.7557399268903766,
"grad_norm": 30.796875,
"learning_rate": 5.42620555284949e-06,
"loss": 24.7289,
"step": 1725
},
{
"epoch": 0.7579304774031024,
"grad_norm": 35.625,
"learning_rate": 5.377496346809547e-06,
"loss": 24.3356,
"step": 1730
},
{
"epoch": 0.7601210279158281,
"grad_norm": 21.140625,
"learning_rate": 5.328787140769606e-06,
"loss": 24.5246,
"step": 1735
},
{
"epoch": 0.7623115784285538,
"grad_norm": 16.625,
"learning_rate": 5.280077934729664e-06,
"loss": 24.1494,
"step": 1740
},
{
"epoch": 0.7645021289412796,
"grad_norm": 25.28125,
"learning_rate": 5.231368728689723e-06,
"loss": 24.7821,
"step": 1745
},
{
"epoch": 0.7666926794540053,
"grad_norm": 49.96875,
"learning_rate": 5.182659522649781e-06,
"loss": 24.9474,
"step": 1750
},
{
"epoch": 0.768883229966731,
"grad_norm": 23.359375,
"learning_rate": 5.133950316609839e-06,
"loss": 24.7685,
"step": 1755
},
{
"epoch": 0.7710737804794567,
"grad_norm": 25.90625,
"learning_rate": 5.0852411105698985e-06,
"loss": 24.6903,
"step": 1760
},
{
"epoch": 0.7732643309921825,
"grad_norm": 310.25,
"learning_rate": 5.036531904529957e-06,
"loss": 24.0739,
"step": 1765
},
{
"epoch": 0.7754548815049082,
"grad_norm": 39.40625,
"learning_rate": 4.987822698490015e-06,
"loss": 25.2116,
"step": 1770
},
{
"epoch": 0.7776454320176339,
"grad_norm": 39.625,
"learning_rate": 4.939113492450074e-06,
"loss": 24.6392,
"step": 1775
},
{
"epoch": 0.7798359825303597,
"grad_norm": 37.15625,
"learning_rate": 4.890404286410132e-06,
"loss": 24.367,
"step": 1780
},
{
"epoch": 0.7820265330430853,
"grad_norm": 45.34375,
"learning_rate": 4.84169508037019e-06,
"loss": 24.2522,
"step": 1785
},
{
"epoch": 0.7842170835558111,
"grad_norm": 30.046875,
"learning_rate": 4.792985874330249e-06,
"loss": 24.7172,
"step": 1790
},
{
"epoch": 0.7864076340685369,
"grad_norm": 38.3125,
"learning_rate": 4.744276668290307e-06,
"loss": 24.4997,
"step": 1795
},
{
"epoch": 0.7885981845812626,
"grad_norm": 36.0625,
"learning_rate": 4.6955674622503655e-06,
"loss": 24.4261,
"step": 1800
},
{
"epoch": 0.7907887350939883,
"grad_norm": 30.09375,
"learning_rate": 4.646858256210424e-06,
"loss": 23.8328,
"step": 1805
},
{
"epoch": 0.792979285606714,
"grad_norm": 57.78125,
"learning_rate": 4.598149050170483e-06,
"loss": 24.3202,
"step": 1810
},
{
"epoch": 0.7951698361194398,
"grad_norm": 37.28125,
"learning_rate": 4.549439844130541e-06,
"loss": 25.0197,
"step": 1815
},
{
"epoch": 0.7973603866321655,
"grad_norm": 43.90625,
"learning_rate": 4.5007306380906e-06,
"loss": 24.1631,
"step": 1820
},
{
"epoch": 0.7995509371448912,
"grad_norm": 39.5625,
"learning_rate": 4.452021432050657e-06,
"loss": 24.1568,
"step": 1825
},
{
"epoch": 0.801741487657617,
"grad_norm": 30.6875,
"learning_rate": 4.403312226010717e-06,
"loss": 24.2651,
"step": 1830
},
{
"epoch": 0.8039320381703426,
"grad_norm": 31.03125,
"learning_rate": 4.354603019970775e-06,
"loss": 24.0601,
"step": 1835
},
{
"epoch": 0.8061225886830684,
"grad_norm": 39.84375,
"learning_rate": 4.305893813930833e-06,
"loss": 23.9148,
"step": 1840
},
{
"epoch": 0.8083131391957942,
"grad_norm": 35.53125,
"learning_rate": 4.257184607890892e-06,
"loss": 23.9099,
"step": 1845
},
{
"epoch": 0.8105036897085198,
"grad_norm": 25.609375,
"learning_rate": 4.20847540185095e-06,
"loss": 24.0205,
"step": 1850
},
{
"epoch": 0.8126942402212456,
"grad_norm": 41.125,
"learning_rate": 4.1597661958110085e-06,
"loss": 24.6065,
"step": 1855
},
{
"epoch": 0.8148847907339714,
"grad_norm": 60.21875,
"learning_rate": 4.111056989771067e-06,
"loss": 24.6779,
"step": 1860
},
{
"epoch": 0.817075341246697,
"grad_norm": 26.515625,
"learning_rate": 4.062347783731125e-06,
"loss": 24.5723,
"step": 1865
},
{
"epoch": 0.8192658917594228,
"grad_norm": 52.0625,
"learning_rate": 4.0136385776911845e-06,
"loss": 24.7251,
"step": 1870
},
{
"epoch": 0.8214564422721485,
"grad_norm": 52.40625,
"learning_rate": 3.964929371651242e-06,
"loss": 23.8427,
"step": 1875
},
{
"epoch": 0.8236469927848743,
"grad_norm": 29.109375,
"learning_rate": 3.916220165611301e-06,
"loss": 23.9856,
"step": 1880
},
{
"epoch": 0.8258375432976,
"grad_norm": 53.8125,
"learning_rate": 3.86751095957136e-06,
"loss": 24.4032,
"step": 1885
},
{
"epoch": 0.8280280938103257,
"grad_norm": 33.5,
"learning_rate": 3.818801753531418e-06,
"loss": 23.6485,
"step": 1890
},
{
"epoch": 0.8302186443230515,
"grad_norm": 79.3125,
"learning_rate": 3.7700925474914763e-06,
"loss": 24.0104,
"step": 1895
},
{
"epoch": 0.8324091948357771,
"grad_norm": 46.21875,
"learning_rate": 3.7213833414515347e-06,
"loss": 24.379,
"step": 1900
},
{
"epoch": 0.8345997453485029,
"grad_norm": 24.046875,
"learning_rate": 3.672674135411593e-06,
"loss": 23.9105,
"step": 1905
},
{
"epoch": 0.8367902958612287,
"grad_norm": 26.84375,
"learning_rate": 3.6239649293716515e-06,
"loss": 23.4966,
"step": 1910
},
{
"epoch": 0.8389808463739543,
"grad_norm": 32.28125,
"learning_rate": 3.57525572333171e-06,
"loss": 24.439,
"step": 1915
},
{
"epoch": 0.8411713968866801,
"grad_norm": 24.9375,
"learning_rate": 3.5265465172917687e-06,
"loss": 24.5451,
"step": 1920
},
{
"epoch": 0.8433619473994058,
"grad_norm": 35.1875,
"learning_rate": 3.477837311251827e-06,
"loss": 24.2255,
"step": 1925
},
{
"epoch": 0.8455524979121315,
"grad_norm": 101.5,
"learning_rate": 3.4291281052118854e-06,
"loss": 23.8204,
"step": 1930
},
{
"epoch": 0.8477430484248573,
"grad_norm": 43.90625,
"learning_rate": 3.3804188991719438e-06,
"loss": 23.8947,
"step": 1935
},
{
"epoch": 0.849933598937583,
"grad_norm": 60.625,
"learning_rate": 3.331709693132002e-06,
"loss": 23.6479,
"step": 1940
},
{
"epoch": 0.8521241494503087,
"grad_norm": 19.59375,
"learning_rate": 3.2830004870920605e-06,
"loss": 24.7873,
"step": 1945
},
{
"epoch": 0.8543146999630344,
"grad_norm": 52.5625,
"learning_rate": 3.2342912810521193e-06,
"loss": 23.4077,
"step": 1950
},
{
"epoch": 0.8565052504757602,
"grad_norm": 30.90625,
"learning_rate": 3.1855820750121773e-06,
"loss": 23.5133,
"step": 1955
},
{
"epoch": 0.858695800988486,
"grad_norm": 56.34375,
"learning_rate": 3.136872868972236e-06,
"loss": 23.5646,
"step": 1960
},
{
"epoch": 0.8608863515012116,
"grad_norm": 338.0,
"learning_rate": 3.088163662932294e-06,
"loss": 24.1954,
"step": 1965
},
{
"epoch": 0.8630769020139374,
"grad_norm": 52.21875,
"learning_rate": 3.039454456892353e-06,
"loss": 24.0436,
"step": 1970
},
{
"epoch": 0.8652674525266632,
"grad_norm": 50.96875,
"learning_rate": 2.9907452508524117e-06,
"loss": 23.8306,
"step": 1975
},
{
"epoch": 0.8674580030393888,
"grad_norm": 27.390625,
"learning_rate": 2.9420360448124696e-06,
"loss": 24.0851,
"step": 1980
},
{
"epoch": 0.8696485535521146,
"grad_norm": 23.109375,
"learning_rate": 2.8933268387725284e-06,
"loss": 24.2154,
"step": 1985
},
{
"epoch": 0.8718391040648403,
"grad_norm": 17.21875,
"learning_rate": 2.8446176327325868e-06,
"loss": 23.9251,
"step": 1990
},
{
"epoch": 0.874029654577566,
"grad_norm": 16.265625,
"learning_rate": 2.795908426692645e-06,
"loss": 23.6277,
"step": 1995
},
{
"epoch": 0.8762202050902917,
"grad_norm": 27.78125,
"learning_rate": 2.7471992206527035e-06,
"loss": 23.5734,
"step": 2000
},
{
"epoch": 0.8762202050902917,
"eval_loss": NaN,
"eval_runtime": 248.7839,
"eval_samples_per_second": 988.943,
"eval_steps_per_second": 30.906,
"step": 2000
},
{
"epoch": 0.8784107556030175,
"grad_norm": 29.203125,
"learning_rate": 2.698490014612762e-06,
"loss": 24.1809,
"step": 2005
},
{
"epoch": 0.8806013061157432,
"grad_norm": 28.46875,
"learning_rate": 2.6497808085728203e-06,
"loss": 23.4948,
"step": 2010
},
{
"epoch": 0.8827918566284689,
"grad_norm": 16.171875,
"learning_rate": 2.601071602532879e-06,
"loss": 23.6219,
"step": 2015
},
{
"epoch": 0.8849824071411947,
"grad_norm": 31.9375,
"learning_rate": 2.5523623964929375e-06,
"loss": 22.9886,
"step": 2020
},
{
"epoch": 0.8871729576539203,
"grad_norm": 13.4453125,
"learning_rate": 2.503653190452996e-06,
"loss": 23.6685,
"step": 2025
},
{
"epoch": 0.8893635081666461,
"grad_norm": 27.390625,
"learning_rate": 2.4549439844130542e-06,
"loss": 23.4918,
"step": 2030
},
{
"epoch": 0.8915540586793719,
"grad_norm": 53.375,
"learning_rate": 2.4062347783731126e-06,
"loss": 23.38,
"step": 2035
},
{
"epoch": 0.8937446091920976,
"grad_norm": 18.734375,
"learning_rate": 2.357525572333171e-06,
"loss": 23.5452,
"step": 2040
},
{
"epoch": 0.8959351597048233,
"grad_norm": 25.671875,
"learning_rate": 2.3088163662932294e-06,
"loss": 23.5461,
"step": 2045
},
{
"epoch": 0.8981257102175491,
"grad_norm": 28.953125,
"learning_rate": 2.260107160253288e-06,
"loss": 23.2701,
"step": 2050
},
{
"epoch": 0.9003162607302748,
"grad_norm": 16.4375,
"learning_rate": 2.2113979542133465e-06,
"loss": 23.5859,
"step": 2055
},
{
"epoch": 0.9025068112430005,
"grad_norm": 20.59375,
"learning_rate": 2.162688748173405e-06,
"loss": 22.9554,
"step": 2060
},
{
"epoch": 0.9046973617557262,
"grad_norm": 31.21875,
"learning_rate": 2.1139795421334633e-06,
"loss": 23.3924,
"step": 2065
},
{
"epoch": 0.906887912268452,
"grad_norm": 29.359375,
"learning_rate": 2.065270336093522e-06,
"loss": 23.0099,
"step": 2070
},
{
"epoch": 0.9090784627811777,
"grad_norm": 22.734375,
"learning_rate": 2.0165611300535805e-06,
"loss": 23.4626,
"step": 2075
},
{
"epoch": 0.9112690132939034,
"grad_norm": 59.6875,
"learning_rate": 1.967851924013639e-06,
"loss": 22.6226,
"step": 2080
},
{
"epoch": 0.9134595638066292,
"grad_norm": 37.71875,
"learning_rate": 1.9191427179736972e-06,
"loss": 22.4083,
"step": 2085
},
{
"epoch": 0.9156501143193548,
"grad_norm": 23.28125,
"learning_rate": 1.8704335119337556e-06,
"loss": 23.4292,
"step": 2090
},
{
"epoch": 0.9178406648320806,
"grad_norm": 31.0625,
"learning_rate": 1.8217243058938142e-06,
"loss": 23.3258,
"step": 2095
},
{
"epoch": 0.9200312153448064,
"grad_norm": 16.359375,
"learning_rate": 1.7730150998538726e-06,
"loss": 23.5227,
"step": 2100
},
{
"epoch": 0.922221765857532,
"grad_norm": 63.59375,
"learning_rate": 1.724305893813931e-06,
"loss": 23.928,
"step": 2105
},
{
"epoch": 0.9244123163702578,
"grad_norm": 30.21875,
"learning_rate": 1.6755966877739893e-06,
"loss": 22.57,
"step": 2110
},
{
"epoch": 0.9266028668829835,
"grad_norm": 28.078125,
"learning_rate": 1.626887481734048e-06,
"loss": 23.4965,
"step": 2115
},
{
"epoch": 0.9287934173957093,
"grad_norm": 21.421875,
"learning_rate": 1.5781782756941063e-06,
"loss": 23.3915,
"step": 2120
},
{
"epoch": 0.930983967908435,
"grad_norm": 14.703125,
"learning_rate": 1.5294690696541647e-06,
"loss": 23.4732,
"step": 2125
},
{
"epoch": 0.9331745184211607,
"grad_norm": 21.5625,
"learning_rate": 1.480759863614223e-06,
"loss": 23.06,
"step": 2130
},
{
"epoch": 0.9353650689338865,
"grad_norm": 26.3125,
"learning_rate": 1.4320506575742814e-06,
"loss": 22.6973,
"step": 2135
},
{
"epoch": 0.9375556194466121,
"grad_norm": 19.078125,
"learning_rate": 1.3833414515343402e-06,
"loss": 22.8054,
"step": 2140
},
{
"epoch": 0.9397461699593379,
"grad_norm": 30.84375,
"learning_rate": 1.3346322454943986e-06,
"loss": 23.4105,
"step": 2145
},
{
"epoch": 0.9419367204720637,
"grad_norm": 62.21875,
"learning_rate": 1.285923039454457e-06,
"loss": 22.8058,
"step": 2150
},
{
"epoch": 0.9441272709847893,
"grad_norm": 15.203125,
"learning_rate": 1.2372138334145156e-06,
"loss": 23.0896,
"step": 2155
},
{
"epoch": 0.9463178214975151,
"grad_norm": 27.8125,
"learning_rate": 1.188504627374574e-06,
"loss": 23.386,
"step": 2160
},
{
"epoch": 0.9485083720102409,
"grad_norm": 17.546875,
"learning_rate": 1.1397954213346323e-06,
"loss": 23.0428,
"step": 2165
},
{
"epoch": 0.9506989225229665,
"grad_norm": 20.015625,
"learning_rate": 1.0910862152946907e-06,
"loss": 23.0953,
"step": 2170
},
{
"epoch": 0.9528894730356923,
"grad_norm": 24.59375,
"learning_rate": 1.0423770092547493e-06,
"loss": 22.9106,
"step": 2175
},
{
"epoch": 0.955080023548418,
"grad_norm": 27.9375,
"learning_rate": 9.936678032148077e-07,
"loss": 22.686,
"step": 2180
},
{
"epoch": 0.9572705740611437,
"grad_norm": 22.078125,
"learning_rate": 9.44958597174866e-07,
"loss": 23.1452,
"step": 2185
},
{
"epoch": 0.9594611245738695,
"grad_norm": 20.65625,
"learning_rate": 8.962493911349246e-07,
"loss": 22.4537,
"step": 2190
},
{
"epoch": 0.9616516750865952,
"grad_norm": 114.75,
"learning_rate": 8.47540185094983e-07,
"loss": 23.5824,
"step": 2195
},
{
"epoch": 0.963842225599321,
"grad_norm": 21.65625,
"learning_rate": 7.988309790550415e-07,
"loss": 23.2908,
"step": 2200
},
{
"epoch": 0.9660327761120466,
"grad_norm": 14.1875,
"learning_rate": 7.501217730150999e-07,
"loss": 22.8018,
"step": 2205
},
{
"epoch": 0.9682233266247724,
"grad_norm": 15.3125,
"learning_rate": 7.014125669751585e-07,
"loss": 22.5503,
"step": 2210
},
{
"epoch": 0.9704138771374982,
"grad_norm": 33.15625,
"learning_rate": 6.527033609352168e-07,
"loss": 22.7779,
"step": 2215
},
{
"epoch": 0.9726044276502238,
"grad_norm": 14.2890625,
"learning_rate": 6.039941548952752e-07,
"loss": 22.3891,
"step": 2220
},
{
"epoch": 0.9747949781629496,
"grad_norm": 61.40625,
"learning_rate": 5.552849488553337e-07,
"loss": 22.7335,
"step": 2225
},
{
"epoch": 0.9769855286756753,
"grad_norm": 48.46875,
"learning_rate": 5.065757428153922e-07,
"loss": 23.5739,
"step": 2230
},
{
"epoch": 0.979176079188401,
"grad_norm": 15.5703125,
"learning_rate": 4.578665367754506e-07,
"loss": 22.5284,
"step": 2235
},
{
"epoch": 0.9813666297011268,
"grad_norm": 21.296875,
"learning_rate": 4.091573307355091e-07,
"loss": 23.336,
"step": 2240
},
{
"epoch": 0.9835571802138525,
"grad_norm": 22.671875,
"learning_rate": 3.6044812469556747e-07,
"loss": 22.2249,
"step": 2245
},
{
"epoch": 0.9857477307265782,
"grad_norm": 31.71875,
"learning_rate": 3.1173891865562595e-07,
"loss": 22.8697,
"step": 2250
},
{
"epoch": 0.9879382812393039,
"grad_norm": 17.65625,
"learning_rate": 2.630297126156844e-07,
"loss": 22.5376,
"step": 2255
},
{
"epoch": 0.9901288317520297,
"grad_norm": 25.34375,
"learning_rate": 2.1432050657574284e-07,
"loss": 22.7322,
"step": 2260
},
{
"epoch": 0.9923193822647554,
"grad_norm": 57.375,
"learning_rate": 1.6561130053580127e-07,
"loss": 22.5554,
"step": 2265
},
{
"epoch": 0.9945099327774811,
"grad_norm": 26.25,
"learning_rate": 1.1690209449585972e-07,
"loss": 22.8525,
"step": 2270
},
{
"epoch": 0.9967004832902069,
"grad_norm": 32.375,
"learning_rate": 6.819288845591817e-08,
"loss": 22.8014,
"step": 2275
},
{
"epoch": 0.9988910338029326,
"grad_norm": 21.90625,
"learning_rate": 1.9483682415976622e-08,
"loss": 22.4073,
"step": 2280
}
],
"logging_steps": 5,
"max_steps": 2282,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 1000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1.9772879918220706e+19,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}