AmberYifan's picture
Model save
57a2c95 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 3125,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.00032,
"grad_norm": 20.05617380625413,
"learning_rate": 6.389776357827476e-08,
"loss": 1.817,
"step": 1
},
{
"epoch": 0.0016,
"grad_norm": 24.170327240578573,
"learning_rate": 3.194888178913738e-07,
"loss": 1.8628,
"step": 5
},
{
"epoch": 0.0032,
"grad_norm": 17.63201705481064,
"learning_rate": 6.389776357827476e-07,
"loss": 1.7731,
"step": 10
},
{
"epoch": 0.0048,
"grad_norm": 12.659034102299168,
"learning_rate": 9.584664536741215e-07,
"loss": 1.6969,
"step": 15
},
{
"epoch": 0.0064,
"grad_norm": 8.71204128591294,
"learning_rate": 1.2779552715654952e-06,
"loss": 1.5956,
"step": 20
},
{
"epoch": 0.008,
"grad_norm": 9.482900190556371,
"learning_rate": 1.5974440894568691e-06,
"loss": 1.5189,
"step": 25
},
{
"epoch": 0.0096,
"grad_norm": 9.082892230070819,
"learning_rate": 1.916932907348243e-06,
"loss": 1.5307,
"step": 30
},
{
"epoch": 0.0112,
"grad_norm": 8.71715912913066,
"learning_rate": 2.2364217252396165e-06,
"loss": 1.4863,
"step": 35
},
{
"epoch": 0.0128,
"grad_norm": 8.266982509072658,
"learning_rate": 2.5559105431309904e-06,
"loss": 1.4411,
"step": 40
},
{
"epoch": 0.0144,
"grad_norm": 7.43964847984838,
"learning_rate": 2.8753993610223648e-06,
"loss": 1.4772,
"step": 45
},
{
"epoch": 0.016,
"grad_norm": 8.285226877478248,
"learning_rate": 3.1948881789137383e-06,
"loss": 1.4813,
"step": 50
},
{
"epoch": 0.0176,
"grad_norm": 8.144708422975583,
"learning_rate": 3.514376996805112e-06,
"loss": 1.4048,
"step": 55
},
{
"epoch": 0.0192,
"grad_norm": 7.597036261894608,
"learning_rate": 3.833865814696486e-06,
"loss": 1.3852,
"step": 60
},
{
"epoch": 0.0208,
"grad_norm": 7.424686782409782,
"learning_rate": 4.15335463258786e-06,
"loss": 1.4124,
"step": 65
},
{
"epoch": 0.0224,
"grad_norm": 7.01187059061448,
"learning_rate": 4.472843450479233e-06,
"loss": 1.3525,
"step": 70
},
{
"epoch": 0.024,
"grad_norm": 8.203285756665721,
"learning_rate": 4.792332268370608e-06,
"loss": 1.4641,
"step": 75
},
{
"epoch": 0.0256,
"grad_norm": 8.91120559492055,
"learning_rate": 5.111821086261981e-06,
"loss": 1.4319,
"step": 80
},
{
"epoch": 0.0272,
"grad_norm": 7.95728896500941,
"learning_rate": 5.431309904153355e-06,
"loss": 1.4198,
"step": 85
},
{
"epoch": 0.0288,
"grad_norm": 8.93733979889198,
"learning_rate": 5.7507987220447296e-06,
"loss": 1.4431,
"step": 90
},
{
"epoch": 0.0304,
"grad_norm": 7.821680056160703,
"learning_rate": 6.070287539936103e-06,
"loss": 1.45,
"step": 95
},
{
"epoch": 0.032,
"grad_norm": 8.599771427970811,
"learning_rate": 6.3897763578274765e-06,
"loss": 1.4271,
"step": 100
},
{
"epoch": 0.0336,
"grad_norm": 8.069041724261744,
"learning_rate": 6.709265175718851e-06,
"loss": 1.4014,
"step": 105
},
{
"epoch": 0.0352,
"grad_norm": 8.444189736987067,
"learning_rate": 7.028753993610224e-06,
"loss": 1.4432,
"step": 110
},
{
"epoch": 0.0368,
"grad_norm": 8.440082783021184,
"learning_rate": 7.348242811501598e-06,
"loss": 1.4234,
"step": 115
},
{
"epoch": 0.0384,
"grad_norm": 9.19834025523869,
"learning_rate": 7.667731629392972e-06,
"loss": 1.4825,
"step": 120
},
{
"epoch": 0.04,
"grad_norm": 9.44008152483298,
"learning_rate": 7.987220447284347e-06,
"loss": 1.4536,
"step": 125
},
{
"epoch": 0.0416,
"grad_norm": 9.137864717981165,
"learning_rate": 8.30670926517572e-06,
"loss": 1.4832,
"step": 130
},
{
"epoch": 0.0432,
"grad_norm": 7.259296421594005,
"learning_rate": 8.626198083067093e-06,
"loss": 1.4355,
"step": 135
},
{
"epoch": 0.0448,
"grad_norm": 9.129262436644341,
"learning_rate": 8.945686900958466e-06,
"loss": 1.48,
"step": 140
},
{
"epoch": 0.0464,
"grad_norm": 7.342624679137957,
"learning_rate": 9.265175718849841e-06,
"loss": 1.4388,
"step": 145
},
{
"epoch": 0.048,
"grad_norm": 9.409291604105896,
"learning_rate": 9.584664536741216e-06,
"loss": 1.506,
"step": 150
},
{
"epoch": 0.0496,
"grad_norm": 7.195158336471212,
"learning_rate": 9.904153354632589e-06,
"loss": 1.4852,
"step": 155
},
{
"epoch": 0.0512,
"grad_norm": 7.894560573501637,
"learning_rate": 1.0223642172523962e-05,
"loss": 1.4228,
"step": 160
},
{
"epoch": 0.0528,
"grad_norm": 9.3990419995412,
"learning_rate": 1.0543130990415335e-05,
"loss": 1.4624,
"step": 165
},
{
"epoch": 0.0544,
"grad_norm": 6.764474220048631,
"learning_rate": 1.086261980830671e-05,
"loss": 1.529,
"step": 170
},
{
"epoch": 0.056,
"grad_norm": 8.762621410789437,
"learning_rate": 1.1182108626198084e-05,
"loss": 1.4889,
"step": 175
},
{
"epoch": 0.0576,
"grad_norm": 9.026712826027282,
"learning_rate": 1.1501597444089459e-05,
"loss": 1.4889,
"step": 180
},
{
"epoch": 0.0592,
"grad_norm": 10.327681032497772,
"learning_rate": 1.1821086261980832e-05,
"loss": 1.4632,
"step": 185
},
{
"epoch": 0.0608,
"grad_norm": 8.916460811078778,
"learning_rate": 1.2140575079872205e-05,
"loss": 1.4957,
"step": 190
},
{
"epoch": 0.0624,
"grad_norm": 9.570316165134905,
"learning_rate": 1.2460063897763578e-05,
"loss": 1.4974,
"step": 195
},
{
"epoch": 0.064,
"grad_norm": 7.931027623078715,
"learning_rate": 1.2779552715654953e-05,
"loss": 1.4986,
"step": 200
},
{
"epoch": 0.0656,
"grad_norm": 9.001437919455842,
"learning_rate": 1.3099041533546326e-05,
"loss": 1.4583,
"step": 205
},
{
"epoch": 0.0672,
"grad_norm": 8.357858455204125,
"learning_rate": 1.3418530351437703e-05,
"loss": 1.5086,
"step": 210
},
{
"epoch": 0.0688,
"grad_norm": 7.4570481772053325,
"learning_rate": 1.3738019169329076e-05,
"loss": 1.5331,
"step": 215
},
{
"epoch": 0.0704,
"grad_norm": 9.286290584701554,
"learning_rate": 1.4057507987220449e-05,
"loss": 1.5065,
"step": 220
},
{
"epoch": 0.072,
"grad_norm": 7.764076796357793,
"learning_rate": 1.4376996805111822e-05,
"loss": 1.489,
"step": 225
},
{
"epoch": 0.0736,
"grad_norm": 10.09652114646815,
"learning_rate": 1.4696485623003197e-05,
"loss": 1.5768,
"step": 230
},
{
"epoch": 0.0752,
"grad_norm": 8.809821488359987,
"learning_rate": 1.501597444089457e-05,
"loss": 1.5554,
"step": 235
},
{
"epoch": 0.0768,
"grad_norm": 8.262499421512615,
"learning_rate": 1.5335463258785944e-05,
"loss": 1.5407,
"step": 240
},
{
"epoch": 0.0784,
"grad_norm": 8.979598030923686,
"learning_rate": 1.5654952076677316e-05,
"loss": 1.5761,
"step": 245
},
{
"epoch": 0.08,
"grad_norm": 7.996608590588298,
"learning_rate": 1.5974440894568694e-05,
"loss": 1.5723,
"step": 250
},
{
"epoch": 0.0816,
"grad_norm": 8.28398837332719,
"learning_rate": 1.6293929712460065e-05,
"loss": 1.511,
"step": 255
},
{
"epoch": 0.0832,
"grad_norm": 10.050176837076577,
"learning_rate": 1.661341853035144e-05,
"loss": 1.62,
"step": 260
},
{
"epoch": 0.0848,
"grad_norm": 8.625153292163999,
"learning_rate": 1.693290734824281e-05,
"loss": 1.5696,
"step": 265
},
{
"epoch": 0.0864,
"grad_norm": 7.036269960407658,
"learning_rate": 1.7252396166134186e-05,
"loss": 1.6274,
"step": 270
},
{
"epoch": 0.088,
"grad_norm": 9.06080432840939,
"learning_rate": 1.757188498402556e-05,
"loss": 1.5238,
"step": 275
},
{
"epoch": 0.0896,
"grad_norm": 7.954340307347254,
"learning_rate": 1.7891373801916932e-05,
"loss": 1.6112,
"step": 280
},
{
"epoch": 0.0912,
"grad_norm": 8.326902382980357,
"learning_rate": 1.8210862619808307e-05,
"loss": 1.5639,
"step": 285
},
{
"epoch": 0.0928,
"grad_norm": 8.465352754166233,
"learning_rate": 1.8530351437699682e-05,
"loss": 1.6031,
"step": 290
},
{
"epoch": 0.0944,
"grad_norm": 7.805713490581876,
"learning_rate": 1.8849840255591057e-05,
"loss": 1.5684,
"step": 295
},
{
"epoch": 0.096,
"grad_norm": 7.8032797694553,
"learning_rate": 1.916932907348243e-05,
"loss": 1.5322,
"step": 300
},
{
"epoch": 0.0976,
"grad_norm": 7.502438683575227,
"learning_rate": 1.9488817891373803e-05,
"loss": 1.578,
"step": 305
},
{
"epoch": 0.0992,
"grad_norm": 7.42748507278892,
"learning_rate": 1.9808306709265177e-05,
"loss": 1.6214,
"step": 310
},
{
"epoch": 0.1008,
"grad_norm": 8.776137139385414,
"learning_rate": 1.9999975036876365e-05,
"loss": 1.5879,
"step": 315
},
{
"epoch": 0.1024,
"grad_norm": 7.577899989222258,
"learning_rate": 1.9999694203166786e-05,
"loss": 1.5625,
"step": 320
},
{
"epoch": 0.104,
"grad_norm": 9.304934816231782,
"learning_rate": 1.999910134063538e-05,
"loss": 1.5502,
"step": 325
},
{
"epoch": 0.1056,
"grad_norm": 8.943302127293093,
"learning_rate": 1.9998196467781738e-05,
"loss": 1.6447,
"step": 330
},
{
"epoch": 0.1072,
"grad_norm": 10.263653381131993,
"learning_rate": 1.999697961284136e-05,
"loss": 1.6365,
"step": 335
},
{
"epoch": 0.1088,
"grad_norm": 7.259185945143334,
"learning_rate": 1.9995450813784785e-05,
"loss": 1.6016,
"step": 340
},
{
"epoch": 0.1104,
"grad_norm": 9.483550631073872,
"learning_rate": 1.9993610118316417e-05,
"loss": 1.6867,
"step": 345
},
{
"epoch": 0.112,
"grad_norm": 9.069753200477997,
"learning_rate": 1.999145758387301e-05,
"loss": 1.6832,
"step": 350
},
{
"epoch": 0.1136,
"grad_norm": 9.791327883527874,
"learning_rate": 1.99889932776219e-05,
"loss": 1.6604,
"step": 355
},
{
"epoch": 0.1152,
"grad_norm": 10.288995257955278,
"learning_rate": 1.9986217276458898e-05,
"loss": 1.5407,
"step": 360
},
{
"epoch": 0.1168,
"grad_norm": 7.343516596486213,
"learning_rate": 1.9983129667005887e-05,
"loss": 1.704,
"step": 365
},
{
"epoch": 0.1184,
"grad_norm": 8.414567171107853,
"learning_rate": 1.9979730545608128e-05,
"loss": 1.6423,
"step": 370
},
{
"epoch": 0.12,
"grad_norm": 31.248624844046244,
"learning_rate": 1.9976020018331244e-05,
"loss": 1.6615,
"step": 375
},
{
"epoch": 0.1216,
"grad_norm": 7.805253897850886,
"learning_rate": 1.997199820095793e-05,
"loss": 1.7482,
"step": 380
},
{
"epoch": 0.1232,
"grad_norm": 12.941755164396426,
"learning_rate": 1.9967665218984308e-05,
"loss": 1.6302,
"step": 385
},
{
"epoch": 0.1248,
"grad_norm": 10.256908147942266,
"learning_rate": 1.996302120761605e-05,
"loss": 1.6527,
"step": 390
},
{
"epoch": 0.1264,
"grad_norm": 10.323844537641683,
"learning_rate": 1.9958066311764115e-05,
"loss": 1.6225,
"step": 395
},
{
"epoch": 0.128,
"grad_norm": 8.17544314720727,
"learning_rate": 1.9952800686040268e-05,
"loss": 1.6112,
"step": 400
},
{
"epoch": 0.1296,
"grad_norm": 9.155308331409806,
"learning_rate": 1.9947224494752236e-05,
"loss": 1.645,
"step": 405
},
{
"epoch": 0.1312,
"grad_norm": 8.093102355577809,
"learning_rate": 1.994133791189857e-05,
"loss": 1.6012,
"step": 410
},
{
"epoch": 0.1328,
"grad_norm": 8.539319454896912,
"learning_rate": 1.993514112116325e-05,
"loss": 1.6163,
"step": 415
},
{
"epoch": 0.1344,
"grad_norm": 9.376291632603513,
"learning_rate": 1.992863431590991e-05,
"loss": 1.7232,
"step": 420
},
{
"epoch": 0.136,
"grad_norm": 8.427174660902148,
"learning_rate": 1.9921817699175844e-05,
"loss": 1.6041,
"step": 425
},
{
"epoch": 0.1376,
"grad_norm": 11.225741174921657,
"learning_rate": 1.991469148366564e-05,
"loss": 1.5705,
"step": 430
},
{
"epoch": 0.1392,
"grad_norm": 7.973508937531369,
"learning_rate": 1.9907255891744562e-05,
"loss": 1.6436,
"step": 435
},
{
"epoch": 0.1408,
"grad_norm": 9.589627221263973,
"learning_rate": 1.989951115543161e-05,
"loss": 1.6753,
"step": 440
},
{
"epoch": 0.1424,
"grad_norm": 6.8530374739159345,
"learning_rate": 1.9891457516392257e-05,
"loss": 1.6441,
"step": 445
},
{
"epoch": 0.144,
"grad_norm": 8.969880080037996,
"learning_rate": 1.988309522593095e-05,
"loss": 1.632,
"step": 450
},
{
"epoch": 0.1456,
"grad_norm": 8.147836497286043,
"learning_rate": 1.9874424544983224e-05,
"loss": 1.7166,
"step": 455
},
{
"epoch": 0.1472,
"grad_norm": 10.804136842902594,
"learning_rate": 1.9865445744107593e-05,
"loss": 1.6071,
"step": 460
},
{
"epoch": 0.1488,
"grad_norm": 7.770558071875093,
"learning_rate": 1.9856159103477085e-05,
"loss": 1.7466,
"step": 465
},
{
"epoch": 0.1504,
"grad_norm": 11.053317637390126,
"learning_rate": 1.9846564912870523e-05,
"loss": 1.6221,
"step": 470
},
{
"epoch": 0.152,
"grad_norm": 8.025748690782159,
"learning_rate": 1.9836663471663454e-05,
"loss": 1.7206,
"step": 475
},
{
"epoch": 0.1536,
"grad_norm": 8.94152560502608,
"learning_rate": 1.9826455088818832e-05,
"loss": 1.6794,
"step": 480
},
{
"epoch": 0.1552,
"grad_norm": 8.723262455381919,
"learning_rate": 1.9815940082877367e-05,
"loss": 1.6909,
"step": 485
},
{
"epoch": 0.1568,
"grad_norm": 11.572838719884283,
"learning_rate": 1.980511878194758e-05,
"loss": 1.7067,
"step": 490
},
{
"epoch": 0.1584,
"grad_norm": 7.98682940495796,
"learning_rate": 1.9793991523695578e-05,
"loss": 1.7714,
"step": 495
},
{
"epoch": 0.16,
"grad_norm": 7.447864398870841,
"learning_rate": 1.9782558655334505e-05,
"loss": 1.632,
"step": 500
},
{
"epoch": 0.1616,
"grad_norm": 8.942126665322432,
"learning_rate": 1.9770820533613716e-05,
"loss": 1.6463,
"step": 505
},
{
"epoch": 0.1632,
"grad_norm": 8.629146616823142,
"learning_rate": 1.9758777524807636e-05,
"loss": 1.6367,
"step": 510
},
{
"epoch": 0.1648,
"grad_norm": 7.603172313414887,
"learning_rate": 1.9746430004704353e-05,
"loss": 1.6558,
"step": 515
},
{
"epoch": 0.1664,
"grad_norm": 11.152682711750428,
"learning_rate": 1.9733778358593852e-05,
"loss": 1.6537,
"step": 520
},
{
"epoch": 0.168,
"grad_norm": 9.887507696721732,
"learning_rate": 1.9720822981256034e-05,
"loss": 1.6715,
"step": 525
},
{
"epoch": 0.1696,
"grad_norm": 9.158423868070782,
"learning_rate": 1.970756427694837e-05,
"loss": 1.6244,
"step": 530
},
{
"epoch": 0.1712,
"grad_norm": 8.442009066532837,
"learning_rate": 1.9694002659393306e-05,
"loss": 1.7073,
"step": 535
},
{
"epoch": 0.1728,
"grad_norm": 7.50714623452273,
"learning_rate": 1.9680138551765335e-05,
"loss": 1.6241,
"step": 540
},
{
"epoch": 0.1744,
"grad_norm": 8.325913790000977,
"learning_rate": 1.9665972386677796e-05,
"loss": 1.5779,
"step": 545
},
{
"epoch": 0.176,
"grad_norm": 13.09863126345551,
"learning_rate": 1.9651504606169395e-05,
"loss": 1.6549,
"step": 550
},
{
"epoch": 0.1776,
"grad_norm": 7.5020724745885925,
"learning_rate": 1.9636735661690385e-05,
"loss": 1.683,
"step": 555
},
{
"epoch": 0.1792,
"grad_norm": 8.857991440508771,
"learning_rate": 1.9621666014088495e-05,
"loss": 1.5727,
"step": 560
},
{
"epoch": 0.1808,
"grad_norm": 10.518277008681551,
"learning_rate": 1.960629613359454e-05,
"loss": 1.5516,
"step": 565
},
{
"epoch": 0.1824,
"grad_norm": 9.763384545927902,
"learning_rate": 1.959062649980776e-05,
"loss": 1.7654,
"step": 570
},
{
"epoch": 0.184,
"grad_norm": 14.784685999297329,
"learning_rate": 1.957465760168084e-05,
"loss": 1.6173,
"step": 575
},
{
"epoch": 0.1856,
"grad_norm": 9.47338828693639,
"learning_rate": 1.9558389937504664e-05,
"loss": 1.6706,
"step": 580
},
{
"epoch": 0.1872,
"grad_norm": 8.255320111181135,
"learning_rate": 1.954182401489277e-05,
"loss": 1.6829,
"step": 585
},
{
"epoch": 0.1888,
"grad_norm": 9.748611896295628,
"learning_rate": 1.952496035076549e-05,
"loss": 1.6963,
"step": 590
},
{
"epoch": 0.1904,
"grad_norm": 10.593417080535822,
"learning_rate": 1.9507799471333842e-05,
"loss": 1.5429,
"step": 595
},
{
"epoch": 0.192,
"grad_norm": 8.495208947428331,
"learning_rate": 1.9490341912083103e-05,
"loss": 1.5996,
"step": 600
},
{
"epoch": 0.1936,
"grad_norm": 8.116738913737667,
"learning_rate": 1.947258821775609e-05,
"loss": 1.6509,
"step": 605
},
{
"epoch": 0.1952,
"grad_norm": 8.843543056646068,
"learning_rate": 1.945453894233618e-05,
"loss": 1.6621,
"step": 610
},
{
"epoch": 0.1968,
"grad_norm": 8.429840678269503,
"learning_rate": 1.9436194649030006e-05,
"loss": 1.6238,
"step": 615
},
{
"epoch": 0.1984,
"grad_norm": 7.060618638753134,
"learning_rate": 1.9417555910249905e-05,
"loss": 1.5455,
"step": 620
},
{
"epoch": 0.2,
"grad_norm": 8.735661880336002,
"learning_rate": 1.939862330759602e-05,
"loss": 1.7242,
"step": 625
},
{
"epoch": 0.2016,
"grad_norm": 7.728994210262772,
"learning_rate": 1.9379397431838194e-05,
"loss": 1.6244,
"step": 630
},
{
"epoch": 0.2032,
"grad_norm": 8.142561603441006,
"learning_rate": 1.935987888289751e-05,
"loss": 1.6033,
"step": 635
},
{
"epoch": 0.2048,
"grad_norm": 7.087122461619784,
"learning_rate": 1.9340068269827567e-05,
"loss": 1.6478,
"step": 640
},
{
"epoch": 0.2064,
"grad_norm": 8.796044551707043,
"learning_rate": 1.93199662107955e-05,
"loss": 1.622,
"step": 645
},
{
"epoch": 0.208,
"grad_norm": 7.637980213604651,
"learning_rate": 1.929957333306267e-05,
"loss": 1.6378,
"step": 650
},
{
"epoch": 0.2096,
"grad_norm": 8.176443820248227,
"learning_rate": 1.9278890272965097e-05,
"loss": 1.603,
"step": 655
},
{
"epoch": 0.2112,
"grad_norm": 8.821331818781406,
"learning_rate": 1.92579176758936e-05,
"loss": 1.6285,
"step": 660
},
{
"epoch": 0.2128,
"grad_norm": 8.751441102706378,
"learning_rate": 1.9236656196273676e-05,
"loss": 1.6783,
"step": 665
},
{
"epoch": 0.2144,
"grad_norm": 7.729648207326943,
"learning_rate": 1.9215106497545047e-05,
"loss": 1.7185,
"step": 670
},
{
"epoch": 0.216,
"grad_norm": 9.632019880573054,
"learning_rate": 1.919326925214099e-05,
"loss": 1.7044,
"step": 675
},
{
"epoch": 0.2176,
"grad_norm": 7.864267682010497,
"learning_rate": 1.9171145141467336e-05,
"loss": 1.6628,
"step": 680
},
{
"epoch": 0.2192,
"grad_norm": 7.961635669857337,
"learning_rate": 1.9148734855881218e-05,
"loss": 1.6602,
"step": 685
},
{
"epoch": 0.2208,
"grad_norm": 10.292993487639919,
"learning_rate": 1.912603909466952e-05,
"loss": 1.5731,
"step": 690
},
{
"epoch": 0.2224,
"grad_norm": 8.162330383972137,
"learning_rate": 1.9103058566027062e-05,
"loss": 1.6291,
"step": 695
},
{
"epoch": 0.224,
"grad_norm": 7.554506895298207,
"learning_rate": 1.9079793987034497e-05,
"loss": 1.6665,
"step": 700
},
{
"epoch": 0.2256,
"grad_norm": 8.135535319158393,
"learning_rate": 1.9056246083635943e-05,
"loss": 1.6404,
"step": 705
},
{
"epoch": 0.2272,
"grad_norm": 9.74804156354211,
"learning_rate": 1.9032415590616323e-05,
"loss": 1.5687,
"step": 710
},
{
"epoch": 0.2288,
"grad_norm": 8.104975985879758,
"learning_rate": 1.9008303251578445e-05,
"loss": 1.6728,
"step": 715
},
{
"epoch": 0.2304,
"grad_norm": 8.720184023804999,
"learning_rate": 1.898390981891979e-05,
"loss": 1.634,
"step": 720
},
{
"epoch": 0.232,
"grad_norm": 9.210829756244634,
"learning_rate": 1.895923605380904e-05,
"loss": 1.6422,
"step": 725
},
{
"epoch": 0.2336,
"grad_norm": 6.632734132335917,
"learning_rate": 1.8934282726162325e-05,
"loss": 1.5709,
"step": 730
},
{
"epoch": 0.2352,
"grad_norm": 7.931181213386623,
"learning_rate": 1.8909050614619197e-05,
"loss": 1.6244,
"step": 735
},
{
"epoch": 0.2368,
"grad_norm": 8.05038767649013,
"learning_rate": 1.8883540506518336e-05,
"loss": 1.6756,
"step": 740
},
{
"epoch": 0.2384,
"grad_norm": 9.138124074487578,
"learning_rate": 1.885775319787298e-05,
"loss": 1.5782,
"step": 745
},
{
"epoch": 0.24,
"grad_norm": 9.052424132753602,
"learning_rate": 1.8831689493346095e-05,
"loss": 1.6486,
"step": 750
},
{
"epoch": 0.2416,
"grad_norm": 12.11857562509831,
"learning_rate": 1.880535020622525e-05,
"loss": 1.5488,
"step": 755
},
{
"epoch": 0.2432,
"grad_norm": 8.346673787548578,
"learning_rate": 1.8778736158397244e-05,
"loss": 1.5675,
"step": 760
},
{
"epoch": 0.2448,
"grad_norm": 7.922560429887788,
"learning_rate": 1.8751848180322476e-05,
"loss": 1.5814,
"step": 765
},
{
"epoch": 0.2464,
"grad_norm": 7.448862840223704,
"learning_rate": 1.872468711100902e-05,
"loss": 1.696,
"step": 770
},
{
"epoch": 0.248,
"grad_norm": 7.381712408385049,
"learning_rate": 1.869725379798643e-05,
"loss": 1.5801,
"step": 775
},
{
"epoch": 0.2496,
"grad_norm": 7.520192699988718,
"learning_rate": 1.866954909727932e-05,
"loss": 1.5583,
"step": 780
},
{
"epoch": 0.2512,
"grad_norm": 8.458007153827033,
"learning_rate": 1.864157387338064e-05,
"loss": 1.5893,
"step": 785
},
{
"epoch": 0.2528,
"grad_norm": 11.023935513739714,
"learning_rate": 1.86133289992247e-05,
"loss": 1.6194,
"step": 790
},
{
"epoch": 0.2544,
"grad_norm": 7.279897256975987,
"learning_rate": 1.8584815356159932e-05,
"loss": 1.6186,
"step": 795
},
{
"epoch": 0.256,
"grad_norm": 7.807795169814656,
"learning_rate": 1.8556033833921386e-05,
"loss": 1.7446,
"step": 800
},
{
"epoch": 0.2576,
"grad_norm": 7.782885718886869,
"learning_rate": 1.8526985330602973e-05,
"loss": 1.6365,
"step": 805
},
{
"epoch": 0.2592,
"grad_norm": 7.881283617531603,
"learning_rate": 1.8497670752629437e-05,
"loss": 1.7161,
"step": 810
},
{
"epoch": 0.2608,
"grad_norm": 7.32964730840398,
"learning_rate": 1.8468091014728076e-05,
"loss": 1.6361,
"step": 815
},
{
"epoch": 0.2624,
"grad_norm": 7.4867507136615075,
"learning_rate": 1.843824703990019e-05,
"loss": 1.7187,
"step": 820
},
{
"epoch": 0.264,
"grad_norm": 7.878583656595656,
"learning_rate": 1.840813975939229e-05,
"loss": 1.6064,
"step": 825
},
{
"epoch": 0.2656,
"grad_norm": 7.641254628614174,
"learning_rate": 1.8377770112667024e-05,
"loss": 1.6189,
"step": 830
},
{
"epoch": 0.2672,
"grad_norm": 7.869146750924817,
"learning_rate": 1.8347139047373885e-05,
"loss": 1.6925,
"step": 835
},
{
"epoch": 0.2688,
"grad_norm": 7.488977024706195,
"learning_rate": 1.8316247519319625e-05,
"loss": 1.6646,
"step": 840
},
{
"epoch": 0.2704,
"grad_norm": 7.519265775038011,
"learning_rate": 1.8285096492438424e-05,
"loss": 1.6267,
"step": 845
},
{
"epoch": 0.272,
"grad_norm": 9.052319258879985,
"learning_rate": 1.825368693876183e-05,
"loss": 1.6113,
"step": 850
},
{
"epoch": 0.2736,
"grad_norm": 7.425912485447768,
"learning_rate": 1.8222019838388422e-05,
"loss": 1.6341,
"step": 855
},
{
"epoch": 0.2752,
"grad_norm": 7.616489466474544,
"learning_rate": 1.8190096179453213e-05,
"loss": 1.6195,
"step": 860
},
{
"epoch": 0.2768,
"grad_norm": 8.57212898020857,
"learning_rate": 1.8157916958096837e-05,
"loss": 1.5484,
"step": 865
},
{
"epoch": 0.2784,
"grad_norm": 8.058170481678996,
"learning_rate": 1.8125483178434448e-05,
"loss": 1.6392,
"step": 870
},
{
"epoch": 0.28,
"grad_norm": 7.759361204922903,
"learning_rate": 1.8092795852524404e-05,
"loss": 1.572,
"step": 875
},
{
"epoch": 0.2816,
"grad_norm": 7.814935898355,
"learning_rate": 1.8059856000336675e-05,
"loss": 1.5755,
"step": 880
},
{
"epoch": 0.2832,
"grad_norm": 8.862683471146346,
"learning_rate": 1.8026664649721016e-05,
"loss": 1.5711,
"step": 885
},
{
"epoch": 0.2848,
"grad_norm": 9.253730250566822,
"learning_rate": 1.7993222836374904e-05,
"loss": 1.6258,
"step": 890
},
{
"epoch": 0.2864,
"grad_norm": 8.877026192983683,
"learning_rate": 1.795953160381121e-05,
"loss": 1.6478,
"step": 895
},
{
"epoch": 0.288,
"grad_norm": 9.101400243172016,
"learning_rate": 1.792559200332564e-05,
"loss": 1.6171,
"step": 900
},
{
"epoch": 0.2896,
"grad_norm": 7.136286110367266,
"learning_rate": 1.789140509396394e-05,
"loss": 1.5754,
"step": 905
},
{
"epoch": 0.2912,
"grad_norm": 7.912268476013011,
"learning_rate": 1.7856971942488826e-05,
"loss": 1.5975,
"step": 910
},
{
"epoch": 0.2928,
"grad_norm": 7.20611974657277,
"learning_rate": 1.7822293623346736e-05,
"loss": 1.658,
"step": 915
},
{
"epoch": 0.2944,
"grad_norm": 10.223106004235314,
"learning_rate": 1.7787371218634263e-05,
"loss": 1.6974,
"step": 920
},
{
"epoch": 0.296,
"grad_norm": 7.6171324527749,
"learning_rate": 1.77522058180644e-05,
"loss": 1.692,
"step": 925
},
{
"epoch": 0.2976,
"grad_norm": 9.169763628140181,
"learning_rate": 1.7716798518932564e-05,
"loss": 1.623,
"step": 930
},
{
"epoch": 0.2992,
"grad_norm": 7.322606018864152,
"learning_rate": 1.7681150426082322e-05,
"loss": 1.6117,
"step": 935
},
{
"epoch": 0.3008,
"grad_norm": 7.081332396019932,
"learning_rate": 1.7645262651870926e-05,
"loss": 1.7026,
"step": 940
},
{
"epoch": 0.3024,
"grad_norm": 7.141747828380935,
"learning_rate": 1.7609136316134616e-05,
"loss": 1.5313,
"step": 945
},
{
"epoch": 0.304,
"grad_norm": 8.73987298901696,
"learning_rate": 1.7572772546153657e-05,
"loss": 1.6529,
"step": 950
},
{
"epoch": 0.3056,
"grad_norm": 8.397556088140782,
"learning_rate": 1.7536172476617183e-05,
"loss": 1.6707,
"step": 955
},
{
"epoch": 0.3072,
"grad_norm": 7.447729626374141,
"learning_rate": 1.749933724958777e-05,
"loss": 1.6343,
"step": 960
},
{
"epoch": 0.3088,
"grad_norm": 6.907736776247304,
"learning_rate": 1.746226801446582e-05,
"loss": 1.6028,
"step": 965
},
{
"epoch": 0.3104,
"grad_norm": 8.374764423916597,
"learning_rate": 1.742496592795368e-05,
"loss": 1.5854,
"step": 970
},
{
"epoch": 0.312,
"grad_norm": 6.656458694402831,
"learning_rate": 1.738743215401955e-05,
"loss": 1.5937,
"step": 975
},
{
"epoch": 0.3136,
"grad_norm": 8.119235048754572,
"learning_rate": 1.7349667863861175e-05,
"loss": 1.6238,
"step": 980
},
{
"epoch": 0.3152,
"grad_norm": 10.758877616976827,
"learning_rate": 1.7311674235869285e-05,
"loss": 1.6329,
"step": 985
},
{
"epoch": 0.3168,
"grad_norm": 6.855294162792502,
"learning_rate": 1.7273452455590835e-05,
"loss": 1.5509,
"step": 990
},
{
"epoch": 0.3184,
"grad_norm": 7.436276047950917,
"learning_rate": 1.7235003715691996e-05,
"loss": 1.6302,
"step": 995
},
{
"epoch": 0.32,
"grad_norm": 9.112788481739514,
"learning_rate": 1.7196329215920963e-05,
"loss": 1.5555,
"step": 1000
},
{
"epoch": 0.3216,
"grad_norm": 7.707360400350489,
"learning_rate": 1.71574301630705e-05,
"loss": 1.5711,
"step": 1005
},
{
"epoch": 0.3232,
"grad_norm": 8.985392901674105,
"learning_rate": 1.711830777094028e-05,
"loss": 1.5719,
"step": 1010
},
{
"epoch": 0.3248,
"grad_norm": 7.854703662416667,
"learning_rate": 1.707896326029903e-05,
"loss": 1.687,
"step": 1015
},
{
"epoch": 0.3264,
"grad_norm": 7.991591880545251,
"learning_rate": 1.7039397858846428e-05,
"loss": 1.5265,
"step": 1020
},
{
"epoch": 0.328,
"grad_norm": 8.38735368829394,
"learning_rate": 1.6999612801174782e-05,
"loss": 1.5071,
"step": 1025
},
{
"epoch": 0.3296,
"grad_norm": 7.088683318490433,
"learning_rate": 1.6959609328730526e-05,
"loss": 1.5594,
"step": 1030
},
{
"epoch": 0.3312,
"grad_norm": 7.732979117471935,
"learning_rate": 1.6919388689775463e-05,
"loss": 1.5789,
"step": 1035
},
{
"epoch": 0.3328,
"grad_norm": 7.480645400348206,
"learning_rate": 1.6878952139347834e-05,
"loss": 1.5923,
"step": 1040
},
{
"epoch": 0.3344,
"grad_norm": 7.858191371747195,
"learning_rate": 1.6838300939223144e-05,
"loss": 1.5487,
"step": 1045
},
{
"epoch": 0.336,
"grad_norm": 8.85580923674649,
"learning_rate": 1.679743635787479e-05,
"loss": 1.5857,
"step": 1050
},
{
"epoch": 0.3376,
"grad_norm": 7.901857553601464,
"learning_rate": 1.6756359670434478e-05,
"loss": 1.593,
"step": 1055
},
{
"epoch": 0.3392,
"grad_norm": 9.026231263712882,
"learning_rate": 1.6715072158652444e-05,
"loss": 1.6723,
"step": 1060
},
{
"epoch": 0.3408,
"grad_norm": 13.504122295950328,
"learning_rate": 1.6673575110857457e-05,
"loss": 1.6176,
"step": 1065
},
{
"epoch": 0.3424,
"grad_norm": 10.243586542935969,
"learning_rate": 1.6631869821916602e-05,
"loss": 1.5779,
"step": 1070
},
{
"epoch": 0.344,
"grad_norm": 7.489622112226259,
"learning_rate": 1.6589957593194887e-05,
"loss": 1.6419,
"step": 1075
},
{
"epoch": 0.3456,
"grad_norm": 7.573818239927097,
"learning_rate": 1.6547839732514646e-05,
"loss": 1.5614,
"step": 1080
},
{
"epoch": 0.3472,
"grad_norm": 10.0936611445675,
"learning_rate": 1.650551755411471e-05,
"loss": 1.5626,
"step": 1085
},
{
"epoch": 0.3488,
"grad_norm": 7.874542575121299,
"learning_rate": 1.646299237860941e-05,
"loss": 1.6234,
"step": 1090
},
{
"epoch": 0.3504,
"grad_norm": 8.356085115655546,
"learning_rate": 1.6420265532947364e-05,
"loss": 1.4703,
"step": 1095
},
{
"epoch": 0.352,
"grad_norm": 9.024857376590765,
"learning_rate": 1.6377338350370077e-05,
"loss": 1.5622,
"step": 1100
},
{
"epoch": 0.3536,
"grad_norm": 7.46614009546006,
"learning_rate": 1.6334212170370323e-05,
"loss": 1.6042,
"step": 1105
},
{
"epoch": 0.3552,
"grad_norm": 8.565479511807924,
"learning_rate": 1.6290888338650373e-05,
"loss": 1.4699,
"step": 1110
},
{
"epoch": 0.3568,
"grad_norm": 7.2168157043192025,
"learning_rate": 1.624736820707998e-05,
"loss": 1.5334,
"step": 1115
},
{
"epoch": 0.3584,
"grad_norm": 6.584169033723296,
"learning_rate": 1.6203653133654213e-05,
"loss": 1.6832,
"step": 1120
},
{
"epoch": 0.36,
"grad_norm": 6.225227952156274,
"learning_rate": 1.615974448245107e-05,
"loss": 1.5323,
"step": 1125
},
{
"epoch": 0.3616,
"grad_norm": 10.601001964167809,
"learning_rate": 1.6115643623588915e-05,
"loss": 1.6198,
"step": 1130
},
{
"epoch": 0.3632,
"grad_norm": 10.795794042349174,
"learning_rate": 1.6071351933183736e-05,
"loss": 1.5853,
"step": 1135
},
{
"epoch": 0.3648,
"grad_norm": 7.346429916746835,
"learning_rate": 1.602687079330619e-05,
"loss": 1.5795,
"step": 1140
},
{
"epoch": 0.3664,
"grad_norm": 6.659543911557079,
"learning_rate": 1.5982201591938496e-05,
"loss": 1.5804,
"step": 1145
},
{
"epoch": 0.368,
"grad_norm": 7.221758133107811,
"learning_rate": 1.5937345722931098e-05,
"loss": 1.6121,
"step": 1150
},
{
"epoch": 0.3696,
"grad_norm": 9.731811312205782,
"learning_rate": 1.5892304585959193e-05,
"loss": 1.6606,
"step": 1155
},
{
"epoch": 0.3712,
"grad_norm": 8.589776786846603,
"learning_rate": 1.5847079586479052e-05,
"loss": 1.6185,
"step": 1160
},
{
"epoch": 0.3728,
"grad_norm": 7.386876191799319,
"learning_rate": 1.580167213568416e-05,
"loss": 1.556,
"step": 1165
},
{
"epoch": 0.3744,
"grad_norm": 8.140967833939259,
"learning_rate": 1.575608365046118e-05,
"loss": 1.5469,
"step": 1170
},
{
"epoch": 0.376,
"grad_norm": 12.874290112872854,
"learning_rate": 1.571031555334575e-05,
"loss": 1.5306,
"step": 1175
},
{
"epoch": 0.3776,
"grad_norm": 7.6631355790039395,
"learning_rate": 1.566436927247808e-05,
"loss": 1.5181,
"step": 1180
},
{
"epoch": 0.3792,
"grad_norm": 6.840987993796332,
"learning_rate": 1.5618246241558402e-05,
"loss": 1.6786,
"step": 1185
},
{
"epoch": 0.3808,
"grad_norm": 10.867647495390518,
"learning_rate": 1.5571947899802227e-05,
"loss": 1.6774,
"step": 1190
},
{
"epoch": 0.3824,
"grad_norm": 7.370178174067006,
"learning_rate": 1.5525475691895438e-05,
"loss": 1.5014,
"step": 1195
},
{
"epoch": 0.384,
"grad_norm": 7.247801379833239,
"learning_rate": 1.5478831067949203e-05,
"loss": 1.5683,
"step": 1200
},
{
"epoch": 0.3856,
"grad_norm": 7.548129806153257,
"learning_rate": 1.5432015483454736e-05,
"loss": 1.5433,
"step": 1205
},
{
"epoch": 0.3872,
"grad_norm": 7.403487909131467,
"learning_rate": 1.5385030399237878e-05,
"loss": 1.617,
"step": 1210
},
{
"epoch": 0.3888,
"grad_norm": 7.085555289389217,
"learning_rate": 1.533787728141351e-05,
"loss": 1.5484,
"step": 1215
},
{
"epoch": 0.3904,
"grad_norm": 6.688525472576119,
"learning_rate": 1.5290557601339807e-05,
"loss": 1.603,
"step": 1220
},
{
"epoch": 0.392,
"grad_norm": 7.062961511805713,
"learning_rate": 1.5243072835572319e-05,
"loss": 1.6592,
"step": 1225
},
{
"epoch": 0.3936,
"grad_norm": 7.338639370748799,
"learning_rate": 1.5195424465817911e-05,
"loss": 1.5468,
"step": 1230
},
{
"epoch": 0.3952,
"grad_norm": 7.404342838798467,
"learning_rate": 1.5147613978888514e-05,
"loss": 1.4803,
"step": 1235
},
{
"epoch": 0.3968,
"grad_norm": 7.815793838467086,
"learning_rate": 1.5099642866654747e-05,
"loss": 1.6002,
"step": 1240
},
{
"epoch": 0.3984,
"grad_norm": 6.983719434750113,
"learning_rate": 1.505151262599934e-05,
"loss": 1.4679,
"step": 1245
},
{
"epoch": 0.4,
"grad_norm": 7.689132853147396,
"learning_rate": 1.5003224758770447e-05,
"loss": 1.5426,
"step": 1250
},
{
"epoch": 0.4016,
"grad_norm": 7.99779675336406,
"learning_rate": 1.4954780771734783e-05,
"loss": 1.6113,
"step": 1255
},
{
"epoch": 0.4032,
"grad_norm": 7.956111242399946,
"learning_rate": 1.4906182176530588e-05,
"loss": 1.6137,
"step": 1260
},
{
"epoch": 0.4048,
"grad_norm": 8.34462398939598,
"learning_rate": 1.4857430489620476e-05,
"loss": 1.6686,
"step": 1265
},
{
"epoch": 0.4064,
"grad_norm": 6.87092130604808,
"learning_rate": 1.4808527232244113e-05,
"loss": 1.5328,
"step": 1270
},
{
"epoch": 0.408,
"grad_norm": 7.069072459949788,
"learning_rate": 1.4759473930370738e-05,
"loss": 1.5685,
"step": 1275
},
{
"epoch": 0.4096,
"grad_norm": 9.020054344062027,
"learning_rate": 1.4710272114651555e-05,
"loss": 1.5414,
"step": 1280
},
{
"epoch": 0.4112,
"grad_norm": 9.445393180818964,
"learning_rate": 1.4660923320371974e-05,
"loss": 1.5297,
"step": 1285
},
{
"epoch": 0.4128,
"grad_norm": 7.558351219070578,
"learning_rate": 1.4611429087403695e-05,
"loss": 1.5524,
"step": 1290
},
{
"epoch": 0.4144,
"grad_norm": 7.092500501081219,
"learning_rate": 1.456179096015667e-05,
"loss": 1.686,
"step": 1295
},
{
"epoch": 0.416,
"grad_norm": 7.207957772603401,
"learning_rate": 1.4512010487530899e-05,
"loss": 1.5716,
"step": 1300
},
{
"epoch": 0.4176,
"grad_norm": 7.34637827560702,
"learning_rate": 1.4462089222868099e-05,
"loss": 1.469,
"step": 1305
},
{
"epoch": 0.4192,
"grad_norm": 9.334037108256902,
"learning_rate": 1.4412028723903251e-05,
"loss": 1.5605,
"step": 1310
},
{
"epoch": 0.4208,
"grad_norm": 7.693732678205702,
"learning_rate": 1.4361830552715973e-05,
"loss": 1.6241,
"step": 1315
},
{
"epoch": 0.4224,
"grad_norm": 8.129038938122736,
"learning_rate": 1.4311496275681785e-05,
"loss": 1.4064,
"step": 1320
},
{
"epoch": 0.424,
"grad_norm": 6.237348905953311,
"learning_rate": 1.4261027463423232e-05,
"loss": 1.4584,
"step": 1325
},
{
"epoch": 0.4256,
"grad_norm": 6.031996271045834,
"learning_rate": 1.4210425690760876e-05,
"loss": 1.5006,
"step": 1330
},
{
"epoch": 0.4272,
"grad_norm": 7.539819121652781,
"learning_rate": 1.4159692536664147e-05,
"loss": 1.5603,
"step": 1335
},
{
"epoch": 0.4288,
"grad_norm": 7.523103920232781,
"learning_rate": 1.410882958420209e-05,
"loss": 1.5839,
"step": 1340
},
{
"epoch": 0.4304,
"grad_norm": 7.666216752529669,
"learning_rate": 1.405783842049395e-05,
"loss": 1.5192,
"step": 1345
},
{
"epoch": 0.432,
"grad_norm": 7.32146225643351,
"learning_rate": 1.4006720636659656e-05,
"loss": 1.5088,
"step": 1350
},
{
"epoch": 0.4336,
"grad_norm": 8.854549575824441,
"learning_rate": 1.3955477827770174e-05,
"loss": 1.5657,
"step": 1355
},
{
"epoch": 0.4352,
"grad_norm": 8.471943810603415,
"learning_rate": 1.3904111592797724e-05,
"loss": 1.4975,
"step": 1360
},
{
"epoch": 0.4368,
"grad_norm": 7.881715416808106,
"learning_rate": 1.3852623534565901e-05,
"loss": 1.4941,
"step": 1365
},
{
"epoch": 0.4384,
"grad_norm": 7.441858474366784,
"learning_rate": 1.3801015259699648e-05,
"loss": 1.5424,
"step": 1370
},
{
"epoch": 0.44,
"grad_norm": 8.277844600049658,
"learning_rate": 1.3749288378575133e-05,
"loss": 1.5154,
"step": 1375
},
{
"epoch": 0.4416,
"grad_norm": 7.1055378423472835,
"learning_rate": 1.3697444505269489e-05,
"loss": 1.5948,
"step": 1380
},
{
"epoch": 0.4432,
"grad_norm": 7.146958460316305,
"learning_rate": 1.3645485257510456e-05,
"loss": 1.5807,
"step": 1385
},
{
"epoch": 0.4448,
"grad_norm": 6.5302083293339965,
"learning_rate": 1.3593412256625898e-05,
"loss": 1.4385,
"step": 1390
},
{
"epoch": 0.4464,
"grad_norm": 7.24352130131181,
"learning_rate": 1.3541227127493218e-05,
"loss": 1.4763,
"step": 1395
},
{
"epoch": 0.448,
"grad_norm": 7.695971362233706,
"learning_rate": 1.348893149848865e-05,
"loss": 1.4818,
"step": 1400
},
{
"epoch": 0.4496,
"grad_norm": 7.454125385613667,
"learning_rate": 1.3436527001436437e-05,
"loss": 1.4103,
"step": 1405
},
{
"epoch": 0.4512,
"grad_norm": 7.82334723806768,
"learning_rate": 1.3384015271557938e-05,
"loss": 1.4473,
"step": 1410
},
{
"epoch": 0.4528,
"grad_norm": 7.813760357009132,
"learning_rate": 1.3331397947420578e-05,
"loss": 1.4581,
"step": 1415
},
{
"epoch": 0.4544,
"grad_norm": 6.911969016129946,
"learning_rate": 1.3278676670886728e-05,
"loss": 1.4735,
"step": 1420
},
{
"epoch": 0.456,
"grad_norm": 7.885416119688343,
"learning_rate": 1.3225853087062481e-05,
"loss": 1.4771,
"step": 1425
},
{
"epoch": 0.4576,
"grad_norm": 7.171521274701965,
"learning_rate": 1.3172928844246297e-05,
"loss": 1.4909,
"step": 1430
},
{
"epoch": 0.4592,
"grad_norm": 6.038998548284215,
"learning_rate": 1.3119905593877593e-05,
"loss": 1.4862,
"step": 1435
},
{
"epoch": 0.4608,
"grad_norm": 7.850277653518824,
"learning_rate": 1.3066784990485202e-05,
"loss": 1.4361,
"step": 1440
},
{
"epoch": 0.4624,
"grad_norm": 8.240535964327846,
"learning_rate": 1.3013568691635733e-05,
"loss": 1.4437,
"step": 1445
},
{
"epoch": 0.464,
"grad_norm": 6.764355409884025,
"learning_rate": 1.2960258357881875e-05,
"loss": 1.5192,
"step": 1450
},
{
"epoch": 0.4656,
"grad_norm": 7.159838600132645,
"learning_rate": 1.2906855652710557e-05,
"loss": 1.5827,
"step": 1455
},
{
"epoch": 0.4672,
"grad_norm": 8.306859326063204,
"learning_rate": 1.2853362242491054e-05,
"loss": 1.5463,
"step": 1460
},
{
"epoch": 0.4688,
"grad_norm": 7.019302747818116,
"learning_rate": 1.279977979642299e-05,
"loss": 1.4193,
"step": 1465
},
{
"epoch": 0.4704,
"grad_norm": 7.622978792348708,
"learning_rate": 1.2746109986484236e-05,
"loss": 1.422,
"step": 1470
},
{
"epoch": 0.472,
"grad_norm": 6.56284738301426,
"learning_rate": 1.2692354487378768e-05,
"loss": 1.5312,
"step": 1475
},
{
"epoch": 0.4736,
"grad_norm": 6.387578487452188,
"learning_rate": 1.2638514976484384e-05,
"loss": 1.4795,
"step": 1480
},
{
"epoch": 0.4752,
"grad_norm": 6.050045115060615,
"learning_rate": 1.2584593133800374e-05,
"loss": 1.4694,
"step": 1485
},
{
"epoch": 0.4768,
"grad_norm": 7.023845599021495,
"learning_rate": 1.2530590641895089e-05,
"loss": 1.5678,
"step": 1490
},
{
"epoch": 0.4784,
"grad_norm": 6.600644496789812,
"learning_rate": 1.2476509185853456e-05,
"loss": 1.5587,
"step": 1495
},
{
"epoch": 0.48,
"grad_norm": 6.483920132677383,
"learning_rate": 1.242235045322438e-05,
"loss": 1.4022,
"step": 1500
},
{
"epoch": 0.4816,
"grad_norm": 7.393698504360439,
"learning_rate": 1.2368116133968091e-05,
"loss": 1.4542,
"step": 1505
},
{
"epoch": 0.4832,
"grad_norm": 6.59515492702019,
"learning_rate": 1.2313807920403419e-05,
"loss": 1.4563,
"step": 1510
},
{
"epoch": 0.4848,
"grad_norm": 11.46310019096542,
"learning_rate": 1.2259427507154964e-05,
"loss": 1.4436,
"step": 1515
},
{
"epoch": 0.4864,
"grad_norm": 7.492782202626292,
"learning_rate": 1.2204976591100253e-05,
"loss": 1.5193,
"step": 1520
},
{
"epoch": 0.488,
"grad_norm": 7.7736261505421025,
"learning_rate": 1.2150456871316758e-05,
"loss": 1.4168,
"step": 1525
},
{
"epoch": 0.4896,
"grad_norm": 7.9354251057792515,
"learning_rate": 1.2095870049028898e-05,
"loss": 1.4414,
"step": 1530
},
{
"epoch": 0.4912,
"grad_norm": 16.059395970692933,
"learning_rate": 1.2041217827554939e-05,
"loss": 1.4695,
"step": 1535
},
{
"epoch": 0.4928,
"grad_norm": 6.808900956376759,
"learning_rate": 1.1986501912253863e-05,
"loss": 1.4531,
"step": 1540
},
{
"epoch": 0.4944,
"grad_norm": 7.736857454242817,
"learning_rate": 1.1931724010472135e-05,
"loss": 1.4924,
"step": 1545
},
{
"epoch": 0.496,
"grad_norm": 8.3359924924688,
"learning_rate": 1.1876885831490442e-05,
"loss": 1.555,
"step": 1550
},
{
"epoch": 0.4976,
"grad_norm": 7.120202493727813,
"learning_rate": 1.1821989086470349e-05,
"loss": 1.4645,
"step": 1555
},
{
"epoch": 0.4992,
"grad_norm": 7.599913279445224,
"learning_rate": 1.1767035488400903e-05,
"loss": 1.4863,
"step": 1560
},
{
"epoch": 0.5008,
"grad_norm": 7.125989534788617,
"learning_rate": 1.1712026752045189e-05,
"loss": 1.4652,
"step": 1565
},
{
"epoch": 0.5024,
"grad_norm": 8.088722450627488,
"learning_rate": 1.1656964593886819e-05,
"loss": 1.3223,
"step": 1570
},
{
"epoch": 0.504,
"grad_norm": 7.516194041566506,
"learning_rate": 1.1601850732076361e-05,
"loss": 1.5801,
"step": 1575
},
{
"epoch": 0.5056,
"grad_norm": 7.239851442002106,
"learning_rate": 1.1546686886377745e-05,
"loss": 1.4731,
"step": 1580
},
{
"epoch": 0.5072,
"grad_norm": 7.520103127972609,
"learning_rate": 1.1491474778114588e-05,
"loss": 1.3683,
"step": 1585
},
{
"epoch": 0.5088,
"grad_norm": 9.131577917600616,
"learning_rate": 1.143621613011648e-05,
"loss": 1.562,
"step": 1590
},
{
"epoch": 0.5104,
"grad_norm": 7.486194993220994,
"learning_rate": 1.1380912666665234e-05,
"loss": 1.4111,
"step": 1595
},
{
"epoch": 0.512,
"grad_norm": 6.496362336710197,
"learning_rate": 1.1325566113441074e-05,
"loss": 1.5706,
"step": 1600
},
{
"epoch": 0.5136,
"grad_norm": 6.823989380990125,
"learning_rate": 1.1270178197468788e-05,
"loss": 1.4575,
"step": 1605
},
{
"epoch": 0.5152,
"grad_norm": 7.5663657895225125,
"learning_rate": 1.121475064706385e-05,
"loss": 1.3982,
"step": 1610
},
{
"epoch": 0.5168,
"grad_norm": 6.5005524818168965,
"learning_rate": 1.1159285191778473e-05,
"loss": 1.4888,
"step": 1615
},
{
"epoch": 0.5184,
"grad_norm": 7.387470263764186,
"learning_rate": 1.1103783562347642e-05,
"loss": 1.359,
"step": 1620
},
{
"epoch": 0.52,
"grad_norm": 7.505035445167394,
"learning_rate": 1.1048247490635133e-05,
"loss": 1.4775,
"step": 1625
},
{
"epoch": 0.5216,
"grad_norm": 6.645939875937268,
"learning_rate": 1.099267870957943e-05,
"loss": 1.5036,
"step": 1630
},
{
"epoch": 0.5232,
"grad_norm": 6.913899992846711,
"learning_rate": 1.0937078953139691e-05,
"loss": 1.4273,
"step": 1635
},
{
"epoch": 0.5248,
"grad_norm": 7.88033348199259,
"learning_rate": 1.0881449956241616e-05,
"loss": 1.2594,
"step": 1640
},
{
"epoch": 0.5264,
"grad_norm": 7.305128083335553,
"learning_rate": 1.0825793454723325e-05,
"loss": 1.4711,
"step": 1645
},
{
"epoch": 0.528,
"grad_norm": 8.41830449904285,
"learning_rate": 1.0770111185281182e-05,
"loss": 1.4567,
"step": 1650
},
{
"epoch": 0.5296,
"grad_norm": 8.793536036546655,
"learning_rate": 1.071440488541562e-05,
"loss": 1.4354,
"step": 1655
},
{
"epoch": 0.5312,
"grad_norm": 7.225955221122391,
"learning_rate": 1.0658676293376894e-05,
"loss": 1.4268,
"step": 1660
},
{
"epoch": 0.5328,
"grad_norm": 7.296610012317386,
"learning_rate": 1.0602927148110882e-05,
"loss": 1.355,
"step": 1665
},
{
"epoch": 0.5344,
"grad_norm": 7.368094614876765,
"learning_rate": 1.0547159189204788e-05,
"loss": 1.3505,
"step": 1670
},
{
"epoch": 0.536,
"grad_norm": 6.381999606829369,
"learning_rate": 1.0491374156832875e-05,
"loss": 1.3813,
"step": 1675
},
{
"epoch": 0.5376,
"grad_norm": 7.047380620438761,
"learning_rate": 1.043557379170217e-05,
"loss": 1.4362,
"step": 1680
},
{
"epoch": 0.5392,
"grad_norm": 6.040591759821475,
"learning_rate": 1.0379759834998133e-05,
"loss": 1.4112,
"step": 1685
},
{
"epoch": 0.5408,
"grad_norm": 6.915219064223463,
"learning_rate": 1.0323934028330337e-05,
"loss": 1.5057,
"step": 1690
},
{
"epoch": 0.5424,
"grad_norm": 7.271768400559689,
"learning_rate": 1.0268098113678124e-05,
"loss": 1.4705,
"step": 1695
},
{
"epoch": 0.544,
"grad_norm": 6.053160359266474,
"learning_rate": 1.0212253833336237e-05,
"loss": 1.477,
"step": 1700
},
{
"epoch": 0.5456,
"grad_norm": 6.534057764124044,
"learning_rate": 1.015640292986046e-05,
"loss": 1.4277,
"step": 1705
},
{
"epoch": 0.5472,
"grad_norm": 8.2582991694942,
"learning_rate": 1.0100547146013252e-05,
"loss": 1.4827,
"step": 1710
},
{
"epoch": 0.5488,
"grad_norm": 6.985572611193969,
"learning_rate": 1.0044688224709346e-05,
"loss": 1.3615,
"step": 1715
},
{
"epoch": 0.5504,
"grad_norm": 6.902453232669348,
"learning_rate": 9.988827908961392e-06,
"loss": 1.4461,
"step": 1720
},
{
"epoch": 0.552,
"grad_norm": 6.996523125729606,
"learning_rate": 9.932967941825539e-06,
"loss": 1.3792,
"step": 1725
},
{
"epoch": 0.5536,
"grad_norm": 6.402198205444748,
"learning_rate": 9.87711006634706e-06,
"loss": 1.5359,
"step": 1730
},
{
"epoch": 0.5552,
"grad_norm": 6.398796292025464,
"learning_rate": 9.821256025505964e-06,
"loss": 1.3898,
"step": 1735
},
{
"epoch": 0.5568,
"grad_norm": 6.940842238375595,
"learning_rate": 9.765407562162606e-06,
"loss": 1.4426,
"step": 1740
},
{
"epoch": 0.5584,
"grad_norm": 8.063938708721054,
"learning_rate": 9.709566419003292e-06,
"loss": 1.4324,
"step": 1745
},
{
"epoch": 0.56,
"grad_norm": 7.113579365707868,
"learning_rate": 9.653734338485924e-06,
"loss": 1.3696,
"step": 1750
},
{
"epoch": 0.5616,
"grad_norm": 9.305625046247098,
"learning_rate": 9.597913062785603e-06,
"loss": 1.3767,
"step": 1755
},
{
"epoch": 0.5632,
"grad_norm": 7.713589183858762,
"learning_rate": 9.54210433374028e-06,
"loss": 1.3993,
"step": 1760
},
{
"epoch": 0.5648,
"grad_norm": 6.882609362707895,
"learning_rate": 9.486309892796413e-06,
"loss": 1.2881,
"step": 1765
},
{
"epoch": 0.5664,
"grad_norm": 7.382583057213062,
"learning_rate": 9.430531480954605e-06,
"loss": 1.3868,
"step": 1770
},
{
"epoch": 0.568,
"grad_norm": 7.060465219449526,
"learning_rate": 9.374770838715289e-06,
"loss": 1.3008,
"step": 1775
},
{
"epoch": 0.5696,
"grad_norm": 6.548055523692506,
"learning_rate": 9.319029706024428e-06,
"loss": 1.4179,
"step": 1780
},
{
"epoch": 0.5712,
"grad_norm": 7.296597392978876,
"learning_rate": 9.2633098222192e-06,
"loss": 1.3648,
"step": 1785
},
{
"epoch": 0.5728,
"grad_norm": 7.6566983852007695,
"learning_rate": 9.20761292597375e-06,
"loss": 1.48,
"step": 1790
},
{
"epoch": 0.5744,
"grad_norm": 6.346223117268573,
"learning_rate": 9.151940755244912e-06,
"loss": 1.4082,
"step": 1795
},
{
"epoch": 0.576,
"grad_norm": 7.11841923298027,
"learning_rate": 9.096295047217988e-06,
"loss": 1.3294,
"step": 1800
},
{
"epoch": 0.5776,
"grad_norm": 8.50296732146637,
"learning_rate": 9.040677538252555e-06,
"loss": 1.4083,
"step": 1805
},
{
"epoch": 0.5792,
"grad_norm": 7.014419702757138,
"learning_rate": 8.985089963828262e-06,
"loss": 1.4773,
"step": 1810
},
{
"epoch": 0.5808,
"grad_norm": 7.176240770998421,
"learning_rate": 8.929534058490682e-06,
"loss": 1.2781,
"step": 1815
},
{
"epoch": 0.5824,
"grad_norm": 7.803766989451665,
"learning_rate": 8.8740115557972e-06,
"loss": 1.4017,
"step": 1820
},
{
"epoch": 0.584,
"grad_norm": 7.779698071769793,
"learning_rate": 8.8185241882629e-06,
"loss": 1.4537,
"step": 1825
},
{
"epoch": 0.5856,
"grad_norm": 6.415990926093912,
"learning_rate": 8.763073687306523e-06,
"loss": 1.2469,
"step": 1830
},
{
"epoch": 0.5872,
"grad_norm": 8.66418584105546,
"learning_rate": 8.707661783196432e-06,
"loss": 1.352,
"step": 1835
},
{
"epoch": 0.5888,
"grad_norm": 7.910777048810527,
"learning_rate": 8.652290204996613e-06,
"loss": 1.4686,
"step": 1840
},
{
"epoch": 0.5904,
"grad_norm": 6.3176543086259995,
"learning_rate": 8.59696068051273e-06,
"loss": 1.4047,
"step": 1845
},
{
"epoch": 0.592,
"grad_norm": 7.512370939346078,
"learning_rate": 8.541674936238219e-06,
"loss": 1.422,
"step": 1850
},
{
"epoch": 0.5936,
"grad_norm": 7.8096285750182215,
"learning_rate": 8.486434697300394e-06,
"loss": 1.4087,
"step": 1855
},
{
"epoch": 0.5952,
"grad_norm": 6.178330789312308,
"learning_rate": 8.431241687406631e-06,
"loss": 1.3726,
"step": 1860
},
{
"epoch": 0.5968,
"grad_norm": 6.624672273010726,
"learning_rate": 8.376097628790586e-06,
"loss": 1.3789,
"step": 1865
},
{
"epoch": 0.5984,
"grad_norm": 7.3241310271547295,
"learning_rate": 8.321004242158439e-06,
"loss": 1.3609,
"step": 1870
},
{
"epoch": 0.6,
"grad_norm": 6.04309140021108,
"learning_rate": 8.265963246635212e-06,
"loss": 1.288,
"step": 1875
},
{
"epoch": 0.6016,
"grad_norm": 6.865692791476556,
"learning_rate": 8.210976359711124e-06,
"loss": 1.3762,
"step": 1880
},
{
"epoch": 0.6032,
"grad_norm": 6.771762456175988,
"learning_rate": 8.156045297187994e-06,
"loss": 1.2717,
"step": 1885
},
{
"epoch": 0.6048,
"grad_norm": 6.154571636328125,
"learning_rate": 8.101171773125716e-06,
"loss": 1.3669,
"step": 1890
},
{
"epoch": 0.6064,
"grad_norm": 6.572968900317892,
"learning_rate": 8.046357499788757e-06,
"loss": 1.38,
"step": 1895
},
{
"epoch": 0.608,
"grad_norm": 8.531193639514733,
"learning_rate": 7.991604187592732e-06,
"loss": 1.5369,
"step": 1900
},
{
"epoch": 0.6096,
"grad_norm": 6.698720683326016,
"learning_rate": 7.93691354505103e-06,
"loss": 1.4231,
"step": 1905
},
{
"epoch": 0.6112,
"grad_norm": 7.521329845585438,
"learning_rate": 7.882287278721523e-06,
"loss": 1.4031,
"step": 1910
},
{
"epoch": 0.6128,
"grad_norm": 7.6211310302707505,
"learning_rate": 7.82772709315328e-06,
"loss": 1.3337,
"step": 1915
},
{
"epoch": 0.6144,
"grad_norm": 6.018003479238453,
"learning_rate": 7.77323469083341e-06,
"loss": 1.3658,
"step": 1920
},
{
"epoch": 0.616,
"grad_norm": 6.464569749783369,
"learning_rate": 7.718811772133918e-06,
"loss": 1.2912,
"step": 1925
},
{
"epoch": 0.6176,
"grad_norm": 6.514713737748858,
"learning_rate": 7.664460035258651e-06,
"loss": 1.5244,
"step": 1930
},
{
"epoch": 0.6192,
"grad_norm": 6.205352658391803,
"learning_rate": 7.610181176190318e-06,
"loss": 1.2526,
"step": 1935
},
{
"epoch": 0.6208,
"grad_norm": 7.014357158748993,
"learning_rate": 7.555976888637556e-06,
"loss": 1.3358,
"step": 1940
},
{
"epoch": 0.6224,
"grad_norm": 7.288009471718571,
"learning_rate": 7.501848863982082e-06,
"loss": 1.419,
"step": 1945
},
{
"epoch": 0.624,
"grad_norm": 7.468562712854107,
"learning_rate": 7.447798791225925e-06,
"loss": 1.2961,
"step": 1950
},
{
"epoch": 0.6256,
"grad_norm": 7.203677073348877,
"learning_rate": 7.393828356938709e-06,
"loss": 1.3685,
"step": 1955
},
{
"epoch": 0.6272,
"grad_norm": 7.047746758873238,
"learning_rate": 7.3399392452050385e-06,
"loss": 1.4267,
"step": 1960
},
{
"epoch": 0.6288,
"grad_norm": 7.352486974428156,
"learning_rate": 7.286133137571938e-06,
"loss": 1.284,
"step": 1965
},
{
"epoch": 0.6304,
"grad_norm": 6.913404968326128,
"learning_rate": 7.2324117129963815e-06,
"loss": 1.3034,
"step": 1970
},
{
"epoch": 0.632,
"grad_norm": 7.129552158427298,
"learning_rate": 7.178776647792918e-06,
"loss": 1.3451,
"step": 1975
},
{
"epoch": 0.6336,
"grad_norm": 7.337355717351341,
"learning_rate": 7.125229615581346e-06,
"loss": 1.2403,
"step": 1980
},
{
"epoch": 0.6352,
"grad_norm": 7.212501447309288,
"learning_rate": 7.071772287234497e-06,
"loss": 1.2539,
"step": 1985
},
{
"epoch": 0.6368,
"grad_norm": 7.268649220162584,
"learning_rate": 7.018406330826096e-06,
"loss": 1.3539,
"step": 1990
},
{
"epoch": 0.6384,
"grad_norm": 7.2645693935162905,
"learning_rate": 6.96513341157872e-06,
"loss": 1.3748,
"step": 1995
},
{
"epoch": 0.64,
"grad_norm": 8.342662156660523,
"learning_rate": 6.911955191811819e-06,
"loss": 1.3324,
"step": 2000
},
{
"epoch": 0.6416,
"grad_norm": 7.283104811608162,
"learning_rate": 6.858873330889868e-06,
"loss": 1.3388,
"step": 2005
},
{
"epoch": 0.6432,
"grad_norm": 6.514509652191098,
"learning_rate": 6.8058894851705655e-06,
"loss": 1.3698,
"step": 2010
},
{
"epoch": 0.6448,
"grad_norm": 7.590737838567425,
"learning_rate": 6.7530053079531664e-06,
"loss": 1.3399,
"step": 2015
},
{
"epoch": 0.6464,
"grad_norm": 10.218109612754096,
"learning_rate": 6.700222449426885e-06,
"loss": 1.3822,
"step": 2020
},
{
"epoch": 0.648,
"grad_norm": 7.9917404827520055,
"learning_rate": 6.6475425566194006e-06,
"loss": 1.406,
"step": 2025
},
{
"epoch": 0.6496,
"grad_norm": 5.976353190616539,
"learning_rate": 6.59496727334547e-06,
"loss": 1.3219,
"step": 2030
},
{
"epoch": 0.6512,
"grad_norm": 8.101871551936691,
"learning_rate": 6.5424982401556305e-06,
"loss": 1.3451,
"step": 2035
},
{
"epoch": 0.6528,
"grad_norm": 6.374533499924878,
"learning_rate": 6.490137094285008e-06,
"loss": 1.3402,
"step": 2040
},
{
"epoch": 0.6544,
"grad_norm": 6.459403952059301,
"learning_rate": 6.437885469602235e-06,
"loss": 1.3804,
"step": 2045
},
{
"epoch": 0.656,
"grad_norm": 7.0688192727901935,
"learning_rate": 6.385744996558456e-06,
"loss": 1.3406,
"step": 2050
},
{
"epoch": 0.6576,
"grad_norm": 6.7313792745199486,
"learning_rate": 6.333717302136457e-06,
"loss": 1.3536,
"step": 2055
},
{
"epoch": 0.6592,
"grad_norm": 6.882985114725421,
"learning_rate": 6.28180400979991e-06,
"loss": 1.2695,
"step": 2060
},
{
"epoch": 0.6608,
"grad_norm": 6.533077979959867,
"learning_rate": 6.230006739442692e-06,
"loss": 1.2323,
"step": 2065
},
{
"epoch": 0.6624,
"grad_norm": 6.540348895029834,
"learning_rate": 6.178327107338353e-06,
"loss": 1.4556,
"step": 2070
},
{
"epoch": 0.664,
"grad_norm": 8.070984765040714,
"learning_rate": 6.1267667260896755e-06,
"loss": 1.3297,
"step": 2075
},
{
"epoch": 0.6656,
"grad_norm": 10.638827242283833,
"learning_rate": 6.075327204578363e-06,
"loss": 1.3589,
"step": 2080
},
{
"epoch": 0.6672,
"grad_norm": 6.9949360978136434,
"learning_rate": 6.024010147914826e-06,
"loss": 1.2704,
"step": 2085
},
{
"epoch": 0.6688,
"grad_norm": 7.6982660712041255,
"learning_rate": 5.972817157388106e-06,
"loss": 1.3201,
"step": 2090
},
{
"epoch": 0.6704,
"grad_norm": 7.526592127466844,
"learning_rate": 5.921749830415905e-06,
"loss": 1.3338,
"step": 2095
},
{
"epoch": 0.672,
"grad_norm": 7.590856683299185,
"learning_rate": 5.870809760494734e-06,
"loss": 1.2958,
"step": 2100
},
{
"epoch": 0.6736,
"grad_norm": 6.9630835241832125,
"learning_rate": 5.819998537150203e-06,
"loss": 1.2639,
"step": 2105
},
{
"epoch": 0.6752,
"grad_norm": 6.253694167385759,
"learning_rate": 5.769317745887413e-06,
"loss": 1.324,
"step": 2110
},
{
"epoch": 0.6768,
"grad_norm": 6.682246580824873,
"learning_rate": 5.718768968141482e-06,
"loss": 1.33,
"step": 2115
},
{
"epoch": 0.6784,
"grad_norm": 7.24085307136985,
"learning_rate": 5.668353781228193e-06,
"loss": 1.3596,
"step": 2120
},
{
"epoch": 0.68,
"grad_norm": 6.889515698119121,
"learning_rate": 5.618073758294802e-06,
"loss": 1.244,
"step": 2125
},
{
"epoch": 0.6816,
"grad_norm": 7.707218870893382,
"learning_rate": 5.567930468270911e-06,
"loss": 1.3282,
"step": 2130
},
{
"epoch": 0.6832,
"grad_norm": 7.803800032255029,
"learning_rate": 5.517925475819539e-06,
"loss": 1.34,
"step": 2135
},
{
"epoch": 0.6848,
"grad_norm": 7.2010746561843755,
"learning_rate": 5.468060341288286e-06,
"loss": 1.2944,
"step": 2140
},
{
"epoch": 0.6864,
"grad_norm": 6.695133409250474,
"learning_rate": 5.418336620660658e-06,
"loss": 1.2467,
"step": 2145
},
{
"epoch": 0.688,
"grad_norm": 6.811091226395264,
"learning_rate": 5.36875586550749e-06,
"loss": 1.2451,
"step": 2150
},
{
"epoch": 0.6896,
"grad_norm": 7.299012057736775,
"learning_rate": 5.319319622938563e-06,
"loss": 1.2175,
"step": 2155
},
{
"epoch": 0.6912,
"grad_norm": 6.523483550221802,
"learning_rate": 5.270029435554295e-06,
"loss": 1.3653,
"step": 2160
},
{
"epoch": 0.6928,
"grad_norm": 7.006983737221144,
"learning_rate": 5.22088684139763e-06,
"loss": 1.3037,
"step": 2165
},
{
"epoch": 0.6944,
"grad_norm": 9.658145298208773,
"learning_rate": 5.171893373906036e-06,
"loss": 1.311,
"step": 2170
},
{
"epoch": 0.696,
"grad_norm": 6.9236063967034145,
"learning_rate": 5.1230505618636575e-06,
"loss": 1.3054,
"step": 2175
},
{
"epoch": 0.6976,
"grad_norm": 6.910379366692173,
"learning_rate": 5.074359929353604e-06,
"loss": 1.1699,
"step": 2180
},
{
"epoch": 0.6992,
"grad_norm": 7.780823172347376,
"learning_rate": 5.025822995710414e-06,
"loss": 1.2552,
"step": 2185
},
{
"epoch": 0.7008,
"grad_norm": 6.782111700218943,
"learning_rate": 4.977441275472622e-06,
"loss": 1.4024,
"step": 2190
},
{
"epoch": 0.7024,
"grad_norm": 7.01625141674153,
"learning_rate": 4.929216278335508e-06,
"loss": 1.3484,
"step": 2195
},
{
"epoch": 0.704,
"grad_norm": 6.611844691268078,
"learning_rate": 4.881149509103993e-06,
"loss": 1.3066,
"step": 2200
},
{
"epoch": 0.7056,
"grad_norm": 6.399244868619568,
"learning_rate": 4.833242467645677e-06,
"loss": 1.4538,
"step": 2205
},
{
"epoch": 0.7072,
"grad_norm": 7.977931392416363,
"learning_rate": 4.785496648844049e-06,
"loss": 1.219,
"step": 2210
},
{
"epoch": 0.7088,
"grad_norm": 7.034256666397906,
"learning_rate": 4.737913542551824e-06,
"loss": 1.319,
"step": 2215
},
{
"epoch": 0.7104,
"grad_norm": 7.5123049812985565,
"learning_rate": 4.690494633544466e-06,
"loss": 1.3271,
"step": 2220
},
{
"epoch": 0.712,
"grad_norm": 6.515957790093329,
"learning_rate": 4.643241401473849e-06,
"loss": 1.3427,
"step": 2225
},
{
"epoch": 0.7136,
"grad_norm": 6.839639945427634,
"learning_rate": 4.596155320822103e-06,
"loss": 1.2736,
"step": 2230
},
{
"epoch": 0.7152,
"grad_norm": 7.689485372330819,
"learning_rate": 4.549237860855578e-06,
"loss": 1.278,
"step": 2235
},
{
"epoch": 0.7168,
"grad_norm": 7.447174910713708,
"learning_rate": 4.502490485579024e-06,
"loss": 1.2915,
"step": 2240
},
{
"epoch": 0.7184,
"grad_norm": 7.345469812975139,
"learning_rate": 4.455914653689889e-06,
"loss": 1.2697,
"step": 2245
},
{
"epoch": 0.72,
"grad_norm": 6.852760914424491,
"learning_rate": 4.409511818532809e-06,
"loss": 1.3326,
"step": 2250
},
{
"epoch": 0.7216,
"grad_norm": 9.60045018569782,
"learning_rate": 4.363283428054262e-06,
"loss": 1.2859,
"step": 2255
},
{
"epoch": 0.7232,
"grad_norm": 7.195814960386886,
"learning_rate": 4.317230924757379e-06,
"loss": 1.2189,
"step": 2260
},
{
"epoch": 0.7248,
"grad_norm": 6.838988916881499,
"learning_rate": 4.271355745656934e-06,
"loss": 1.2248,
"step": 2265
},
{
"epoch": 0.7264,
"grad_norm": 7.060258323380059,
"learning_rate": 4.2256593222345185e-06,
"loss": 1.3148,
"step": 2270
},
{
"epoch": 0.728,
"grad_norm": 7.168253685594636,
"learning_rate": 4.1801430803938496e-06,
"loss": 1.2706,
"step": 2275
},
{
"epoch": 0.7296,
"grad_norm": 8.110296241359842,
"learning_rate": 4.1348084404162895e-06,
"loss": 1.22,
"step": 2280
},
{
"epoch": 0.7312,
"grad_norm": 6.146120619441478,
"learning_rate": 4.089656816916525e-06,
"loss": 1.2596,
"step": 2285
},
{
"epoch": 0.7328,
"grad_norm": 7.51200746635356,
"learning_rate": 4.0446896187984275e-06,
"loss": 1.2051,
"step": 2290
},
{
"epoch": 0.7344,
"grad_norm": 7.00889844522249,
"learning_rate": 3.999908249211096e-06,
"loss": 1.3089,
"step": 2295
},
{
"epoch": 0.736,
"grad_norm": 7.433128694008251,
"learning_rate": 3.955314105505056e-06,
"loss": 1.1858,
"step": 2300
},
{
"epoch": 0.7376,
"grad_norm": 7.682196556731105,
"learning_rate": 3.910908579188672e-06,
"loss": 1.2912,
"step": 2305
},
{
"epoch": 0.7392,
"grad_norm": 8.377582759029309,
"learning_rate": 3.866693055884723e-06,
"loss": 1.2959,
"step": 2310
},
{
"epoch": 0.7408,
"grad_norm": 7.525533674291028,
"learning_rate": 3.8226689152871576e-06,
"loss": 1.3002,
"step": 2315
},
{
"epoch": 0.7424,
"grad_norm": 6.5405611376318085,
"learning_rate": 3.7788375311180624e-06,
"loss": 1.1595,
"step": 2320
},
{
"epoch": 0.744,
"grad_norm": 6.351041058258562,
"learning_rate": 3.735200271084779e-06,
"loss": 1.2756,
"step": 2325
},
{
"epoch": 0.7456,
"grad_norm": 6.945497138509825,
"learning_rate": 3.691758496837228e-06,
"loss": 1.3431,
"step": 2330
},
{
"epoch": 0.7472,
"grad_norm": 7.80452005536971,
"learning_rate": 3.6485135639254234e-06,
"loss": 1.1743,
"step": 2335
},
{
"epoch": 0.7488,
"grad_norm": 6.904075689429772,
"learning_rate": 3.6054668217571774e-06,
"loss": 1.2647,
"step": 2340
},
{
"epoch": 0.7504,
"grad_norm": 5.826075155016658,
"learning_rate": 3.5626196135559898e-06,
"loss": 1.4307,
"step": 2345
},
{
"epoch": 0.752,
"grad_norm": 7.4800929811344234,
"learning_rate": 3.5199732763191317e-06,
"loss": 1.3035,
"step": 2350
},
{
"epoch": 0.7536,
"grad_norm": 6.592404311501557,
"learning_rate": 3.4775291407759393e-06,
"loss": 1.3101,
"step": 2355
},
{
"epoch": 0.7552,
"grad_norm": 6.0603172825038625,
"learning_rate": 3.435288531346269e-06,
"loss": 1.1216,
"step": 2360
},
{
"epoch": 0.7568,
"grad_norm": 7.862732877894627,
"learning_rate": 3.3932527660991877e-06,
"loss": 1.2707,
"step": 2365
},
{
"epoch": 0.7584,
"grad_norm": 6.934609447705353,
"learning_rate": 3.351423156711836e-06,
"loss": 1.1406,
"step": 2370
},
{
"epoch": 0.76,
"grad_norm": 6.676258909833237,
"learning_rate": 3.309801008428498e-06,
"loss": 1.1754,
"step": 2375
},
{
"epoch": 0.7616,
"grad_norm": 7.059333862950036,
"learning_rate": 3.268387620019885e-06,
"loss": 1.214,
"step": 2380
},
{
"epoch": 0.7632,
"grad_norm": 6.4985460219892115,
"learning_rate": 3.2271842837425917e-06,
"loss": 1.2429,
"step": 2385
},
{
"epoch": 0.7648,
"grad_norm": 6.318472678199636,
"learning_rate": 3.1861922852987794e-06,
"loss": 1.2567,
"step": 2390
},
{
"epoch": 0.7664,
"grad_norm": 8.106856700967048,
"learning_rate": 3.1454129037960614e-06,
"loss": 1.2827,
"step": 2395
},
{
"epoch": 0.768,
"grad_norm": 6.544814378582256,
"learning_rate": 3.1048474117075834e-06,
"loss": 1.2408,
"step": 2400
},
{
"epoch": 0.7696,
"grad_norm": 8.187702879843156,
"learning_rate": 3.0644970748323253e-06,
"loss": 1.3093,
"step": 2405
},
{
"epoch": 0.7712,
"grad_norm": 7.060252319548262,
"learning_rate": 3.0243631522556027e-06,
"loss": 1.2043,
"step": 2410
},
{
"epoch": 0.7728,
"grad_norm": 7.212166473447804,
"learning_rate": 2.984446896309764e-06,
"loss": 1.29,
"step": 2415
},
{
"epoch": 0.7744,
"grad_norm": 6.205766494665441,
"learning_rate": 2.94474955253513e-06,
"loss": 1.2984,
"step": 2420
},
{
"epoch": 0.776,
"grad_norm": 8.705823518308824,
"learning_rate": 2.9052723596411194e-06,
"loss": 1.272,
"step": 2425
},
{
"epoch": 0.7776,
"grad_norm": 6.193262549894205,
"learning_rate": 2.866016549467602e-06,
"loss": 1.2184,
"step": 2430
},
{
"epoch": 0.7792,
"grad_norm": 7.425984999552959,
"learning_rate": 2.82698334694645e-06,
"loss": 1.2093,
"step": 2435
},
{
"epoch": 0.7808,
"grad_norm": 6.720543568523316,
"learning_rate": 2.7881739700633382e-06,
"loss": 1.2015,
"step": 2440
},
{
"epoch": 0.7824,
"grad_norm": 7.41759595687626,
"learning_rate": 2.749589629819708e-06,
"loss": 1.1781,
"step": 2445
},
{
"epoch": 0.784,
"grad_norm": 6.10376266829222,
"learning_rate": 2.7112315301949986e-06,
"loss": 1.2669,
"step": 2450
},
{
"epoch": 0.7856,
"grad_norm": 7.211278115666269,
"learning_rate": 2.6731008681090763e-06,
"loss": 1.2374,
"step": 2455
},
{
"epoch": 0.7872,
"grad_norm": 6.323650227039773,
"learning_rate": 2.6351988333848787e-06,
"loss": 1.2188,
"step": 2460
},
{
"epoch": 0.7888,
"grad_norm": 8.023736585244432,
"learning_rate": 2.5975266087113015e-06,
"loss": 1.1623,
"step": 2465
},
{
"epoch": 0.7904,
"grad_norm": 6.611440556841849,
"learning_rate": 2.5600853696062766e-06,
"loss": 1.2767,
"step": 2470
},
{
"epoch": 0.792,
"grad_norm": 6.563637825114816,
"learning_rate": 2.5228762843801047e-06,
"loss": 1.255,
"step": 2475
},
{
"epoch": 0.7936,
"grad_norm": 7.2147131416231325,
"learning_rate": 2.485900514098991e-06,
"loss": 1.1983,
"step": 2480
},
{
"epoch": 0.7952,
"grad_norm": 6.707678159171325,
"learning_rate": 2.4491592125488206e-06,
"loss": 1.0984,
"step": 2485
},
{
"epoch": 0.7968,
"grad_norm": 6.811597254056078,
"learning_rate": 2.4126535261991577e-06,
"loss": 1.256,
"step": 2490
},
{
"epoch": 0.7984,
"grad_norm": 7.126090842225368,
"learning_rate": 2.3763845941674703e-06,
"loss": 1.0681,
"step": 2495
},
{
"epoch": 0.8,
"grad_norm": 7.3532700559553215,
"learning_rate": 2.340353548183575e-06,
"loss": 1.318,
"step": 2500
},
{
"epoch": 0.8016,
"grad_norm": 7.106651921925259,
"learning_rate": 2.3045615125543353e-06,
"loss": 1.1499,
"step": 2505
},
{
"epoch": 0.8032,
"grad_norm": 6.895206052430061,
"learning_rate": 2.2690096041285757e-06,
"loss": 1.2491,
"step": 2510
},
{
"epoch": 0.8048,
"grad_norm": 7.077398116538314,
"learning_rate": 2.2336989322622306e-06,
"loss": 1.1645,
"step": 2515
},
{
"epoch": 0.8064,
"grad_norm": 7.163449736184592,
"learning_rate": 2.198630598783723e-06,
"loss": 1.2058,
"step": 2520
},
{
"epoch": 0.808,
"grad_norm": 7.411419676699458,
"learning_rate": 2.1638056979596012e-06,
"loss": 1.2915,
"step": 2525
},
{
"epoch": 0.8096,
"grad_norm": 6.83424681277287,
"learning_rate": 2.1292253164603673e-06,
"loss": 1.236,
"step": 2530
},
{
"epoch": 0.8112,
"grad_norm": 7.517941880919823,
"learning_rate": 2.094890533326589e-06,
"loss": 1.215,
"step": 2535
},
{
"epoch": 0.8128,
"grad_norm": 7.707098663070382,
"learning_rate": 2.0608024199352216e-06,
"loss": 1.2385,
"step": 2540
},
{
"epoch": 0.8144,
"grad_norm": 7.918662426971106,
"learning_rate": 2.026962039966176e-06,
"loss": 1.1269,
"step": 2545
},
{
"epoch": 0.816,
"grad_norm": 8.318700772754628,
"learning_rate": 1.9933704493691354e-06,
"loss": 1.1145,
"step": 2550
},
{
"epoch": 0.8176,
"grad_norm": 7.58106431146203,
"learning_rate": 1.960028696330596e-06,
"loss": 1.2279,
"step": 2555
},
{
"epoch": 0.8192,
"grad_norm": 7.987545520077578,
"learning_rate": 1.926937821241164e-06,
"loss": 1.1885,
"step": 2560
},
{
"epoch": 0.8208,
"grad_norm": 7.251449924956015,
"learning_rate": 1.8940988566630903e-06,
"loss": 1.2175,
"step": 2565
},
{
"epoch": 0.8224,
"grad_norm": 6.84899161592509,
"learning_rate": 1.861512827298051e-06,
"loss": 1.0726,
"step": 2570
},
{
"epoch": 0.824,
"grad_norm": 7.124772325379492,
"learning_rate": 1.8291807499551772e-06,
"loss": 1.238,
"step": 2575
},
{
"epoch": 0.8256,
"grad_norm": 7.245848004677167,
"learning_rate": 1.7971036335193249e-06,
"loss": 1.1557,
"step": 2580
},
{
"epoch": 0.8272,
"grad_norm": 8.125243585964292,
"learning_rate": 1.7652824789195811e-06,
"loss": 1.2371,
"step": 2585
},
{
"epoch": 0.8288,
"grad_norm": 7.1050108164531975,
"learning_rate": 1.73371827909805e-06,
"loss": 1.2496,
"step": 2590
},
{
"epoch": 0.8304,
"grad_norm": 6.936091197387642,
"learning_rate": 1.7024120189788573e-06,
"loss": 1.136,
"step": 2595
},
{
"epoch": 0.832,
"grad_norm": 7.21210256556138,
"learning_rate": 1.6713646754374225e-06,
"loss": 1.1357,
"step": 2600
},
{
"epoch": 0.8336,
"grad_norm": 6.3383661134472264,
"learning_rate": 1.6405772172699696e-06,
"loss": 1.1153,
"step": 2605
},
{
"epoch": 0.8352,
"grad_norm": 8.718135834059972,
"learning_rate": 1.6100506051633136e-06,
"loss": 1.1553,
"step": 2610
},
{
"epoch": 0.8368,
"grad_norm": 8.95033808717262,
"learning_rate": 1.5797857916648596e-06,
"loss": 1.2361,
"step": 2615
},
{
"epoch": 0.8384,
"grad_norm": 6.673199959521796,
"learning_rate": 1.5497837211528965e-06,
"loss": 1.185,
"step": 2620
},
{
"epoch": 0.84,
"grad_norm": 6.696629258511363,
"learning_rate": 1.5200453298071238e-06,
"loss": 1.2785,
"step": 2625
},
{
"epoch": 0.8416,
"grad_norm": 7.572117970721777,
"learning_rate": 1.4905715455794379e-06,
"loss": 1.2972,
"step": 2630
},
{
"epoch": 0.8432,
"grad_norm": 9.096173969266975,
"learning_rate": 1.461363288164983e-06,
"loss": 1.262,
"step": 2635
},
{
"epoch": 0.8448,
"grad_norm": 7.743393031922831,
"learning_rate": 1.432421468973444e-06,
"loss": 1.1165,
"step": 2640
},
{
"epoch": 0.8464,
"grad_norm": 7.2853153371541675,
"learning_rate": 1.4037469911006096e-06,
"loss": 1.2781,
"step": 2645
},
{
"epoch": 0.848,
"grad_norm": 7.4707067773918965,
"learning_rate": 1.3753407493001968e-06,
"loss": 1.2033,
"step": 2650
},
{
"epoch": 0.8496,
"grad_norm": 6.265617058277776,
"learning_rate": 1.3472036299559255e-06,
"loss": 1.1115,
"step": 2655
},
{
"epoch": 0.8512,
"grad_norm": 6.784902517223355,
"learning_rate": 1.3193365110538647e-06,
"loss": 1.218,
"step": 2660
},
{
"epoch": 0.8528,
"grad_norm": 6.719604118534906,
"learning_rate": 1.2917402621550369e-06,
"loss": 1.0628,
"step": 2665
},
{
"epoch": 0.8544,
"grad_norm": 9.25094531147658,
"learning_rate": 1.2644157443682737e-06,
"loss": 1.2604,
"step": 2670
},
{
"epoch": 0.856,
"grad_norm": 7.5401988168853125,
"learning_rate": 1.23736381032336e-06,
"loss": 1.2161,
"step": 2675
},
{
"epoch": 0.8576,
"grad_norm": 5.8986225646917845,
"learning_rate": 1.2105853041444172e-06,
"loss": 1.1727,
"step": 2680
},
{
"epoch": 0.8592,
"grad_norm": 7.05150428234311,
"learning_rate": 1.184081061423572e-06,
"loss": 1.2429,
"step": 2685
},
{
"epoch": 0.8608,
"grad_norm": 7.8831648202761055,
"learning_rate": 1.157851909194876e-06,
"loss": 1.1535,
"step": 2690
},
{
"epoch": 0.8624,
"grad_norm": 6.764706659905809,
"learning_rate": 1.1318986659085062e-06,
"loss": 1.2213,
"step": 2695
},
{
"epoch": 0.864,
"grad_norm": 7.091883009667597,
"learning_rate": 1.10622214140522e-06,
"loss": 1.2937,
"step": 2700
},
{
"epoch": 0.8656,
"grad_norm": 8.205435093308155,
"learning_rate": 1.080823136891086e-06,
"loss": 1.0753,
"step": 2705
},
{
"epoch": 0.8672,
"grad_norm": 7.707709432086696,
"learning_rate": 1.0557024449124854e-06,
"loss": 1.22,
"step": 2710
},
{
"epoch": 0.8688,
"grad_norm": 7.701900190738418,
"learning_rate": 1.0308608493313776e-06,
"loss": 1.2737,
"step": 2715
},
{
"epoch": 0.8704,
"grad_norm": 7.957037822325651,
"learning_rate": 1.0062991253008525e-06,
"loss": 1.1962,
"step": 2720
},
{
"epoch": 0.872,
"grad_norm": 7.098180709287834,
"learning_rate": 9.820180392409252e-07,
"loss": 1.2093,
"step": 2725
},
{
"epoch": 0.8736,
"grad_norm": 7.610067003127613,
"learning_rate": 9.580183488146323e-07,
"loss": 1.2104,
"step": 2730
},
{
"epoch": 0.8752,
"grad_norm": 7.399913871553443,
"learning_rate": 9.343008029043876e-07,
"loss": 1.2166,
"step": 2735
},
{
"epoch": 0.8768,
"grad_norm": 7.845545343933413,
"learning_rate": 9.108661415886111e-07,
"loss": 1.224,
"step": 2740
},
{
"epoch": 0.8784,
"grad_norm": 8.250129826284104,
"learning_rate": 8.87715096118642e-07,
"loss": 1.2828,
"step": 2745
},
{
"epoch": 0.88,
"grad_norm": 7.754702962013714,
"learning_rate": 8.64848388895917e-07,
"loss": 1.2148,
"step": 2750
},
{
"epoch": 0.8816,
"grad_norm": 7.13532064965064,
"learning_rate": 8.42266733449425e-07,
"loss": 1.1721,
"step": 2755
},
{
"epoch": 0.8832,
"grad_norm": 7.085924370440946,
"learning_rate": 8.199708344134493e-07,
"loss": 1.1104,
"step": 2760
},
{
"epoch": 0.8848,
"grad_norm": 7.73698788059968,
"learning_rate": 7.979613875055736e-07,
"loss": 1.2089,
"step": 2765
},
{
"epoch": 0.8864,
"grad_norm": 7.791514673450203,
"learning_rate": 7.76239079504979e-07,
"loss": 1.2016,
"step": 2770
},
{
"epoch": 0.888,
"grad_norm": 6.7303438186194855,
"learning_rate": 7.548045882310084e-07,
"loss": 1.0739,
"step": 2775
},
{
"epoch": 0.8896,
"grad_norm": 6.776201230293054,
"learning_rate": 7.336585825220244e-07,
"loss": 1.1817,
"step": 2780
},
{
"epoch": 0.8912,
"grad_norm": 7.513647159391496,
"learning_rate": 7.128017222145267e-07,
"loss": 1.1978,
"step": 2785
},
{
"epoch": 0.8928,
"grad_norm": 7.203824082806031,
"learning_rate": 6.922346581225725e-07,
"loss": 1.1137,
"step": 2790
},
{
"epoch": 0.8944,
"grad_norm": 7.674190673017307,
"learning_rate": 6.719580320174657e-07,
"loss": 1.1589,
"step": 2795
},
{
"epoch": 0.896,
"grad_norm": 7.427895020925505,
"learning_rate": 6.519724766077262e-07,
"loss": 1.1643,
"step": 2800
},
{
"epoch": 0.8976,
"grad_norm": 7.522555453911278,
"learning_rate": 6.322786155193594e-07,
"loss": 1.1759,
"step": 2805
},
{
"epoch": 0.8992,
"grad_norm": 7.160309775741213,
"learning_rate": 6.128770632763825e-07,
"loss": 1.1401,
"step": 2810
},
{
"epoch": 0.9008,
"grad_norm": 7.064709958323231,
"learning_rate": 5.937684252816578e-07,
"loss": 1.2335,
"step": 2815
},
{
"epoch": 0.9024,
"grad_norm": 7.336329439580518,
"learning_rate": 5.749532977979977e-07,
"loss": 1.1347,
"step": 2820
},
{
"epoch": 0.904,
"grad_norm": 7.690187518015022,
"learning_rate": 5.564322679295619e-07,
"loss": 1.1116,
"step": 2825
},
{
"epoch": 0.9056,
"grad_norm": 7.108832953401795,
"learning_rate": 5.382059136035389e-07,
"loss": 1.1324,
"step": 2830
},
{
"epoch": 0.9072,
"grad_norm": 7.578023550663353,
"learning_rate": 5.202748035521021e-07,
"loss": 1.1919,
"step": 2835
},
{
"epoch": 0.9088,
"grad_norm": 6.903932907016643,
"learning_rate": 5.026394972946813e-07,
"loss": 1.1659,
"step": 2840
},
{
"epoch": 0.9104,
"grad_norm": 6.079707510002676,
"learning_rate": 4.85300545120484e-07,
"loss": 1.2324,
"step": 2845
},
{
"epoch": 0.912,
"grad_norm": 7.452754847950774,
"learning_rate": 4.6825848807133813e-07,
"loss": 1.2998,
"step": 2850
},
{
"epoch": 0.9136,
"grad_norm": 6.122150122845729,
"learning_rate": 4.515138579248035e-07,
"loss": 1.2366,
"step": 2855
},
{
"epoch": 0.9152,
"grad_norm": 8.029542518584094,
"learning_rate": 4.350671771775772e-07,
"loss": 1.1056,
"step": 2860
},
{
"epoch": 0.9168,
"grad_norm": 5.867322954044555,
"learning_rate": 4.189189590291975e-07,
"loss": 1.2582,
"step": 2865
},
{
"epoch": 0.9184,
"grad_norm": 8.148008523277348,
"learning_rate": 4.030697073660217e-07,
"loss": 1.2402,
"step": 2870
},
{
"epoch": 0.92,
"grad_norm": 6.608369807826164,
"learning_rate": 3.875199167455035e-07,
"loss": 1.2935,
"step": 2875
},
{
"epoch": 0.9216,
"grad_norm": 6.225848002947905,
"learning_rate": 3.7227007238076596e-07,
"loss": 1.1169,
"step": 2880
},
{
"epoch": 0.9232,
"grad_norm": 7.4677935130036435,
"learning_rate": 3.573206501254556e-07,
"loss": 1.2367,
"step": 2885
},
{
"epoch": 0.9248,
"grad_norm": 6.610580655839769,
"learning_rate": 3.4267211645890306e-07,
"loss": 1.2118,
"step": 2890
},
{
"epoch": 0.9264,
"grad_norm": 5.9140036705595564,
"learning_rate": 3.283249284715528e-07,
"loss": 1.1834,
"step": 2895
},
{
"epoch": 0.928,
"grad_norm": 6.959822912506896,
"learning_rate": 3.1427953385071207e-07,
"loss": 1.2302,
"step": 2900
},
{
"epoch": 0.9296,
"grad_norm": 7.043289170717119,
"learning_rate": 3.005363708665765e-07,
"loss": 1.0955,
"step": 2905
},
{
"epoch": 0.9312,
"grad_norm": 6.32656303603902,
"learning_rate": 2.870958683585545e-07,
"loss": 1.1267,
"step": 2910
},
{
"epoch": 0.9328,
"grad_norm": 6.352958182641917,
"learning_rate": 2.7395844572188915e-07,
"loss": 1.1378,
"step": 2915
},
{
"epoch": 0.9344,
"grad_norm": 7.408074075398155,
"learning_rate": 2.6112451289456495e-07,
"loss": 1.159,
"step": 2920
},
{
"epoch": 0.936,
"grad_norm": 6.867410397272755,
"learning_rate": 2.4859447034452424e-07,
"loss": 1.1076,
"step": 2925
},
{
"epoch": 0.9376,
"grad_norm": 7.248191222610278,
"learning_rate": 2.3636870905716424e-07,
"loss": 1.0558,
"step": 2930
},
{
"epoch": 0.9392,
"grad_norm": 6.2060869333572946,
"learning_rate": 2.2444761052313857e-07,
"loss": 1.0717,
"step": 2935
},
{
"epoch": 0.9408,
"grad_norm": 7.123428377796361,
"learning_rate": 2.1283154672645522e-07,
"loss": 1.2433,
"step": 2940
},
{
"epoch": 0.9424,
"grad_norm": 7.587226030653606,
"learning_rate": 2.015208801328694e-07,
"loss": 1.137,
"step": 2945
},
{
"epoch": 0.944,
"grad_norm": 8.023362652127654,
"learning_rate": 1.905159636785714e-07,
"loss": 1.1554,
"step": 2950
},
{
"epoch": 0.9456,
"grad_norm": 7.1454832029115485,
"learning_rate": 1.79817140759172e-07,
"loss": 1.1245,
"step": 2955
},
{
"epoch": 0.9472,
"grad_norm": 6.22220394431011,
"learning_rate": 1.6942474521899232e-07,
"loss": 1.1412,
"step": 2960
},
{
"epoch": 0.9488,
"grad_norm": 6.91072459977162,
"learning_rate": 1.5933910134064202e-07,
"loss": 1.1616,
"step": 2965
},
{
"epoch": 0.9504,
"grad_norm": 7.785687521486665,
"learning_rate": 1.4956052383490295e-07,
"loss": 1.2241,
"step": 2970
},
{
"epoch": 0.952,
"grad_norm": 6.558516061888123,
"learning_rate": 1.4008931783090707e-07,
"loss": 1.1245,
"step": 2975
},
{
"epoch": 0.9536,
"grad_norm": 7.69748620578705,
"learning_rate": 1.309257788666174e-07,
"loss": 1.0904,
"step": 2980
},
{
"epoch": 0.9552,
"grad_norm": 7.747365211882861,
"learning_rate": 1.220701928796042e-07,
"loss": 1.1514,
"step": 2985
},
{
"epoch": 0.9568,
"grad_norm": 6.611064880165885,
"learning_rate": 1.1352283619812443e-07,
"loss": 1.0847,
"step": 2990
},
{
"epoch": 0.9584,
"grad_norm": 7.002989095386839,
"learning_rate": 1.0528397553249636e-07,
"loss": 1.1879,
"step": 2995
},
{
"epoch": 0.96,
"grad_norm": 7.0152838160784805,
"learning_rate": 9.73538679667807e-08,
"loss": 1.2758,
"step": 3000
},
{
"epoch": 0.9616,
"grad_norm": 7.094274121519631,
"learning_rate": 8.97327609507559e-08,
"loss": 1.1738,
"step": 3005
},
{
"epoch": 0.9632,
"grad_norm": 8.216237482299412,
"learning_rate": 8.242089229219984e-08,
"loss": 1.1826,
"step": 3010
},
{
"epoch": 0.9648,
"grad_norm": 7.697329542588331,
"learning_rate": 7.541849014946479e-08,
"loss": 1.1891,
"step": 3015
},
{
"epoch": 0.9664,
"grad_norm": 5.975230690605909,
"learning_rate": 6.872577302436179e-08,
"loss": 1.2118,
"step": 3020
},
{
"epoch": 0.968,
"grad_norm": 6.8498253272657434,
"learning_rate": 6.234294975534183e-08,
"loss": 1.0708,
"step": 3025
},
{
"epoch": 0.9696,
"grad_norm": 7.431417632188548,
"learning_rate": 5.6270219510975445e-08,
"loss": 1.2071,
"step": 3030
},
{
"epoch": 0.9712,
"grad_norm": 7.544413890406052,
"learning_rate": 5.050777178374544e-08,
"loss": 1.1812,
"step": 3035
},
{
"epoch": 0.9728,
"grad_norm": 7.824359941406518,
"learning_rate": 4.505578638412722e-08,
"loss": 1.1669,
"step": 3040
},
{
"epoch": 0.9744,
"grad_norm": 7.681706989711725,
"learning_rate": 3.9914433434982135e-08,
"loss": 1.1548,
"step": 3045
},
{
"epoch": 0.976,
"grad_norm": 6.900708976315526,
"learning_rate": 3.508387336624619e-08,
"loss": 1.0863,
"step": 3050
},
{
"epoch": 0.9776,
"grad_norm": 8.898326692320722,
"learning_rate": 3.056425690992404e-08,
"loss": 1.2441,
"step": 3055
},
{
"epoch": 0.9792,
"grad_norm": 6.965073943785161,
"learning_rate": 2.6355725095389416e-08,
"loss": 1.2815,
"step": 3060
},
{
"epoch": 0.9808,
"grad_norm": 8.318803563609041,
"learning_rate": 2.2458409244979772e-08,
"loss": 1.266,
"step": 3065
},
{
"epoch": 0.9824,
"grad_norm": 7.233363975555559,
"learning_rate": 1.8872430969901766e-08,
"loss": 1.227,
"step": 3070
},
{
"epoch": 0.984,
"grad_norm": 7.484219660419537,
"learning_rate": 1.559790216643542e-08,
"loss": 1.1119,
"step": 3075
},
{
"epoch": 0.9856,
"grad_norm": 6.4306561615634426,
"learning_rate": 1.2634925012440235e-08,
"loss": 1.1649,
"step": 3080
},
{
"epoch": 0.9872,
"grad_norm": 7.328910487014067,
"learning_rate": 9.983591964171091e-09,
"loss": 1.2878,
"step": 3085
},
{
"epoch": 0.9888,
"grad_norm": 7.278976626995467,
"learning_rate": 7.643985753390537e-09,
"loss": 1.0835,
"step": 3090
},
{
"epoch": 0.9904,
"grad_norm": 7.408823192054949,
"learning_rate": 5.616179384788645e-09,
"loss": 1.2181,
"step": 3095
},
{
"epoch": 0.992,
"grad_norm": 7.7605788783327485,
"learning_rate": 3.900236133703717e-09,
"loss": 1.1201,
"step": 3100
},
{
"epoch": 0.9936,
"grad_norm": 7.585618897897224,
"learning_rate": 2.496209544147199e-09,
"loss": 1.2209,
"step": 3105
},
{
"epoch": 0.9952,
"grad_norm": 7.2343937302933,
"learning_rate": 1.4041434271350184e-09,
"loss": 1.1871,
"step": 3110
},
{
"epoch": 0.9968,
"grad_norm": 7.2380371709860265,
"learning_rate": 6.240718593208961e-10,
"loss": 1.0477,
"step": 3115
},
{
"epoch": 0.9984,
"grad_norm": 7.4068514185411995,
"learning_rate": 1.5601918192942322e-10,
"loss": 1.1603,
"step": 3120
},
{
"epoch": 1.0,
"grad_norm": 7.2891019708519105,
"learning_rate": 0.0,
"loss": 1.1876,
"step": 3125
},
{
"epoch": 1.0,
"eval_loss": 1.1837999820709229,
"eval_runtime": 37.7062,
"eval_samples_per_second": 13.26,
"eval_steps_per_second": 0.849,
"step": 3125
},
{
"epoch": 1.0,
"step": 3125,
"total_flos": 40105673687040.0,
"train_loss": 1.4248850550079346,
"train_runtime": 11739.5291,
"train_samples_per_second": 4.259,
"train_steps_per_second": 0.266
}
],
"logging_steps": 5,
"max_steps": 3125,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 40105673687040.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}