DITS_math_sft_iteration_0 / trainer_state.json
swt's picture
upload
249b2ed
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.995397008055236,
"eval_steps": 500,
"global_step": 868,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.023014959723820484,
"grad_norm": 25.436981238085195,
"learning_rate": 5.747126436781609e-08,
"loss": 0.6046,
"step": 5
},
{
"epoch": 0.04602991944764097,
"grad_norm": 20.828982286160393,
"learning_rate": 1.1494252873563217e-07,
"loss": 0.5684,
"step": 10
},
{
"epoch": 0.06904487917146145,
"grad_norm": 18.79182460742335,
"learning_rate": 1.7241379310344828e-07,
"loss": 0.5887,
"step": 15
},
{
"epoch": 0.09205983889528194,
"grad_norm": 17.689641482327932,
"learning_rate": 2.2988505747126435e-07,
"loss": 0.6554,
"step": 20
},
{
"epoch": 0.11507479861910241,
"grad_norm": 16.065937870878834,
"learning_rate": 2.873563218390804e-07,
"loss": 0.5661,
"step": 25
},
{
"epoch": 0.1380897583429229,
"grad_norm": 16.348847664761777,
"learning_rate": 3.4482758620689656e-07,
"loss": 0.6168,
"step": 30
},
{
"epoch": 0.1611047180667434,
"grad_norm": 17.719848872005795,
"learning_rate": 4.0229885057471266e-07,
"loss": 0.5178,
"step": 35
},
{
"epoch": 0.18411967779056387,
"grad_norm": 15.539337894638383,
"learning_rate": 4.597701149425287e-07,
"loss": 0.5083,
"step": 40
},
{
"epoch": 0.20713463751438435,
"grad_norm": 12.464237258707364,
"learning_rate": 5.172413793103448e-07,
"loss": 0.5685,
"step": 45
},
{
"epoch": 0.23014959723820483,
"grad_norm": 23.062114966571542,
"learning_rate": 5.747126436781608e-07,
"loss": 0.5859,
"step": 50
},
{
"epoch": 0.25316455696202533,
"grad_norm": 12.808034602784485,
"learning_rate": 6.32183908045977e-07,
"loss": 0.5568,
"step": 55
},
{
"epoch": 0.2761795166858458,
"grad_norm": 15.176748325531914,
"learning_rate": 6.896551724137931e-07,
"loss": 0.4969,
"step": 60
},
{
"epoch": 0.2991944764096663,
"grad_norm": 18.200134879195883,
"learning_rate": 7.471264367816092e-07,
"loss": 0.5486,
"step": 65
},
{
"epoch": 0.3222094361334868,
"grad_norm": 16.483146212519067,
"learning_rate": 8.045977011494253e-07,
"loss": 0.5261,
"step": 70
},
{
"epoch": 0.34522439585730724,
"grad_norm": 9.46076044473471,
"learning_rate": 8.620689655172412e-07,
"loss": 0.4849,
"step": 75
},
{
"epoch": 0.36823935558112775,
"grad_norm": 17.374061304465616,
"learning_rate": 9.195402298850574e-07,
"loss": 0.5233,
"step": 80
},
{
"epoch": 0.3912543153049482,
"grad_norm": 15.210933453312531,
"learning_rate": 9.770114942528735e-07,
"loss": 0.5727,
"step": 85
},
{
"epoch": 0.4142692750287687,
"grad_norm": 12.981120100801451,
"learning_rate": 9.999635938356475e-07,
"loss": 0.5257,
"step": 90
},
{
"epoch": 0.4372842347525892,
"grad_norm": 17.675647072729685,
"learning_rate": 9.997411309192068e-07,
"loss": 0.5459,
"step": 95
},
{
"epoch": 0.46029919447640966,
"grad_norm": 14.258404245690016,
"learning_rate": 9.993165206112856e-07,
"loss": 0.526,
"step": 100
},
{
"epoch": 0.48331415420023016,
"grad_norm": 14.135034352681272,
"learning_rate": 9.98689934668541e-07,
"loss": 0.537,
"step": 105
},
{
"epoch": 0.5063291139240507,
"grad_norm": 14.441236134574162,
"learning_rate": 9.978616265476252e-07,
"loss": 0.5772,
"step": 110
},
{
"epoch": 0.5293440736478712,
"grad_norm": 18.566704225296032,
"learning_rate": 9.9683193130266e-07,
"loss": 0.5539,
"step": 115
},
{
"epoch": 0.5523590333716916,
"grad_norm": 20.35165674585742,
"learning_rate": 9.956012654497072e-07,
"loss": 0.5824,
"step": 120
},
{
"epoch": 0.5753739930955121,
"grad_norm": 18.442515510965833,
"learning_rate": 9.941701267982862e-07,
"loss": 0.5599,
"step": 125
},
{
"epoch": 0.5983889528193326,
"grad_norm": 15.875900780315595,
"learning_rate": 9.925390942500064e-07,
"loss": 0.5304,
"step": 130
},
{
"epoch": 0.6214039125431531,
"grad_norm": 15.34860260658814,
"learning_rate": 9.907088275644012e-07,
"loss": 0.5846,
"step": 135
},
{
"epoch": 0.6444188722669736,
"grad_norm": 21.447791339652113,
"learning_rate": 9.8868006709205e-07,
"loss": 0.5849,
"step": 140
},
{
"epoch": 0.667433831990794,
"grad_norm": 10.808057687362574,
"learning_rate": 9.864536334751061e-07,
"loss": 0.4918,
"step": 145
},
{
"epoch": 0.6904487917146145,
"grad_norm": 12.425855367187637,
"learning_rate": 9.84030427315341e-07,
"loss": 0.5746,
"step": 150
},
{
"epoch": 0.713463751438435,
"grad_norm": 21.87717141819886,
"learning_rate": 9.814114288098486e-07,
"loss": 0.5923,
"step": 155
},
{
"epoch": 0.7364787111622555,
"grad_norm": 14.90837695531576,
"learning_rate": 9.78597697354551e-07,
"loss": 0.5658,
"step": 160
},
{
"epoch": 0.759493670886076,
"grad_norm": 13.422204897299519,
"learning_rate": 9.755903711156684e-07,
"loss": 0.5876,
"step": 165
},
{
"epoch": 0.7825086306098964,
"grad_norm": 13.807384773589732,
"learning_rate": 9.723906665693258e-07,
"loss": 0.6197,
"step": 170
},
{
"epoch": 0.8055235903337169,
"grad_norm": 17.890041232308786,
"learning_rate": 9.689998780094837e-07,
"loss": 0.6169,
"step": 175
},
{
"epoch": 0.8285385500575374,
"grad_norm": 12.171428212644345,
"learning_rate": 9.654193770243905e-07,
"loss": 0.5251,
"step": 180
},
{
"epoch": 0.8515535097813579,
"grad_norm": 14.00509518878877,
"learning_rate": 9.616506119417697e-07,
"loss": 0.5351,
"step": 185
},
{
"epoch": 0.8745684695051784,
"grad_norm": 13.855656981961861,
"learning_rate": 9.576951072429643e-07,
"loss": 0.4762,
"step": 190
},
{
"epoch": 0.8975834292289988,
"grad_norm": 12.530836070851475,
"learning_rate": 9.535544629462786e-07,
"loss": 0.5654,
"step": 195
},
{
"epoch": 0.9205983889528193,
"grad_norm": 14.94597069900537,
"learning_rate": 9.492303539597636e-07,
"loss": 0.5434,
"step": 200
},
{
"epoch": 0.9436133486766398,
"grad_norm": 12.929882120022377,
"learning_rate": 9.4472452940371e-07,
"loss": 0.5511,
"step": 205
},
{
"epoch": 0.9666283084004603,
"grad_norm": 11.785693957022563,
"learning_rate": 9.400388119031211e-07,
"loss": 0.5743,
"step": 210
},
{
"epoch": 0.9896432681242808,
"grad_norm": 11.4673892135634,
"learning_rate": 9.351750968504539e-07,
"loss": 0.5289,
"step": 215
},
{
"epoch": 0.998849252013809,
"eval_loss": 0.5940150022506714,
"eval_runtime": 19.3038,
"eval_samples_per_second": 7.356,
"eval_steps_per_second": 0.932,
"step": 217
},
{
"epoch": 1.0126582278481013,
"grad_norm": 9.979009329427443,
"learning_rate": 9.301353516389246e-07,
"loss": 0.493,
"step": 220
},
{
"epoch": 1.0356731875719218,
"grad_norm": 9.627235780909444,
"learning_rate": 9.249216148666895e-07,
"loss": 0.4392,
"step": 225
},
{
"epoch": 1.0586881472957423,
"grad_norm": 10.663799414852592,
"learning_rate": 9.195359955122243e-07,
"loss": 0.4452,
"step": 230
},
{
"epoch": 1.0817031070195626,
"grad_norm": 9.923849389538619,
"learning_rate": 9.139806720812324e-07,
"loss": 0.4312,
"step": 235
},
{
"epoch": 1.1047180667433831,
"grad_norm": 13.08615355461209,
"learning_rate": 9.082578917254309e-07,
"loss": 0.4213,
"step": 240
},
{
"epoch": 1.1277330264672036,
"grad_norm": 14.694279179845232,
"learning_rate": 9.023699693335678e-07,
"loss": 0.4362,
"step": 245
},
{
"epoch": 1.1507479861910241,
"grad_norm": 13.373390196527911,
"learning_rate": 8.963192865950402e-07,
"loss": 0.4383,
"step": 250
},
{
"epoch": 1.1737629459148446,
"grad_norm": 15.92838239960798,
"learning_rate": 8.901082910364906e-07,
"loss": 0.4091,
"step": 255
},
{
"epoch": 1.1967779056386652,
"grad_norm": 11.132023780497018,
"learning_rate": 8.83739495031772e-07,
"loss": 0.4542,
"step": 260
},
{
"epoch": 1.2197928653624857,
"grad_norm": 12.973494941543084,
"learning_rate": 8.772154747856825e-07,
"loss": 0.464,
"step": 265
},
{
"epoch": 1.2428078250863062,
"grad_norm": 10.383261084427103,
"learning_rate": 8.705388692918792e-07,
"loss": 0.4561,
"step": 270
},
{
"epoch": 1.2658227848101267,
"grad_norm": 12.748442378817272,
"learning_rate": 8.637123792653946e-07,
"loss": 0.4947,
"step": 275
},
{
"epoch": 1.288837744533947,
"grad_norm": 13.222343698654353,
"learning_rate": 8.567387660501851e-07,
"loss": 0.468,
"step": 280
},
{
"epoch": 1.3118527042577677,
"grad_norm": 13.75340991399124,
"learning_rate": 8.49620850502157e-07,
"loss": 0.4949,
"step": 285
},
{
"epoch": 1.334867663981588,
"grad_norm": 13.111144430939557,
"learning_rate": 8.423615118481175e-07,
"loss": 0.4303,
"step": 290
},
{
"epoch": 1.3578826237054085,
"grad_norm": 12.913854941381981,
"learning_rate": 8.349636865211156e-07,
"loss": 0.4768,
"step": 295
},
{
"epoch": 1.380897583429229,
"grad_norm": 11.088379869831822,
"learning_rate": 8.274303669726426e-07,
"loss": 0.3887,
"step": 300
},
{
"epoch": 1.4039125431530495,
"grad_norm": 14.136934996189732,
"learning_rate": 8.197646004621728e-07,
"loss": 0.4006,
"step": 305
},
{
"epoch": 1.42692750287687,
"grad_norm": 13.472677332352156,
"learning_rate": 8.119694878245342e-07,
"loss": 0.4591,
"step": 310
},
{
"epoch": 1.4499424626006905,
"grad_norm": 16.812648158386022,
"learning_rate": 8.040481822156082e-07,
"loss": 0.4317,
"step": 315
},
{
"epoch": 1.472957422324511,
"grad_norm": 12.431554542835052,
"learning_rate": 7.96003887836864e-07,
"loss": 0.495,
"step": 320
},
{
"epoch": 1.4959723820483315,
"grad_norm": 11.033195564551397,
"learning_rate": 7.878398586392461e-07,
"loss": 0.4493,
"step": 325
},
{
"epoch": 1.518987341772152,
"grad_norm": 14.98570304225286,
"learning_rate": 7.795593970069373e-07,
"loss": 0.423,
"step": 330
},
{
"epoch": 1.5420023014959723,
"grad_norm": 15.240270393540726,
"learning_rate": 7.711658524215305e-07,
"loss": 0.427,
"step": 335
},
{
"epoch": 1.565017261219793,
"grad_norm": 13.308584149699858,
"learning_rate": 7.626626201071493e-07,
"loss": 0.4334,
"step": 340
},
{
"epoch": 1.5880322209436133,
"grad_norm": 11.918078822598936,
"learning_rate": 7.540531396570655e-07,
"loss": 0.4526,
"step": 345
},
{
"epoch": 1.6110471806674338,
"grad_norm": 16.581675862528876,
"learning_rate": 7.453408936423687e-07,
"loss": 0.5097,
"step": 350
},
{
"epoch": 1.6340621403912543,
"grad_norm": 15.37481667798687,
"learning_rate": 7.365294062032528e-07,
"loss": 0.432,
"step": 355
},
{
"epoch": 1.6570771001150748,
"grad_norm": 14.096321138322955,
"learning_rate": 7.27622241623485e-07,
"loss": 0.4631,
"step": 360
},
{
"epoch": 1.6800920598388953,
"grad_norm": 11.953450213331397,
"learning_rate": 7.18623002888639e-07,
"loss": 0.4227,
"step": 365
},
{
"epoch": 1.7031070195627158,
"grad_norm": 15.401901846965378,
"learning_rate": 7.095353302286721e-07,
"loss": 0.4424,
"step": 370
},
{
"epoch": 1.7261219792865363,
"grad_norm": 12.52624816605781,
"learning_rate": 7.003628996454361e-07,
"loss": 0.4238,
"step": 375
},
{
"epoch": 1.7491369390103566,
"grad_norm": 13.884043525893127,
"learning_rate": 6.911094214257204e-07,
"loss": 0.4881,
"step": 380
},
{
"epoch": 1.7721518987341773,
"grad_norm": 12.763240064118046,
"learning_rate": 6.817786386404237e-07,
"loss": 0.4292,
"step": 385
},
{
"epoch": 1.7951668584579976,
"grad_norm": 14.214477794634833,
"learning_rate": 6.723743256304676e-07,
"loss": 0.3902,
"step": 390
},
{
"epoch": 1.8181818181818183,
"grad_norm": 17.56505032673697,
"learning_rate": 6.629002864800588e-07,
"loss": 0.4949,
"step": 395
},
{
"epoch": 1.8411967779056386,
"grad_norm": 13.349528895489893,
"learning_rate": 6.533603534779215e-07,
"loss": 0.4711,
"step": 400
},
{
"epoch": 1.8642117376294591,
"grad_norm": 12.236389004746698,
"learning_rate": 6.437583855671204e-07,
"loss": 0.4669,
"step": 405
},
{
"epoch": 1.8872266973532796,
"grad_norm": 15.00019380907836,
"learning_rate": 6.340982667841021e-07,
"loss": 0.412,
"step": 410
},
{
"epoch": 1.9102416570771001,
"grad_norm": 14.956008054688557,
"learning_rate": 6.243839046875853e-07,
"loss": 0.4395,
"step": 415
},
{
"epoch": 1.9332566168009206,
"grad_norm": 12.692934607546489,
"learning_rate": 6.146192287779377e-07,
"loss": 0.4188,
"step": 420
},
{
"epoch": 1.956271576524741,
"grad_norm": 13.129064291695105,
"learning_rate": 6.048081889076766e-07,
"loss": 0.4306,
"step": 425
},
{
"epoch": 1.9792865362485617,
"grad_norm": 14.905784082915199,
"learning_rate": 5.949547536837359e-07,
"loss": 0.4472,
"step": 430
},
{
"epoch": 1.997698504027618,
"eval_loss": 0.6037406325340271,
"eval_runtime": 19.2686,
"eval_samples_per_second": 7.369,
"eval_steps_per_second": 0.934,
"step": 434
},
{
"epoch": 2.002301495972382,
"grad_norm": 10.451580707887741,
"learning_rate": 5.85062908862149e-07,
"loss": 0.4338,
"step": 435
},
{
"epoch": 2.0253164556962027,
"grad_norm": 10.700383234622784,
"learning_rate": 5.751366557357933e-07,
"loss": 0.3787,
"step": 440
},
{
"epoch": 2.048331415420023,
"grad_norm": 11.08124701052113,
"learning_rate": 5.651800095158501e-07,
"loss": 0.3198,
"step": 445
},
{
"epoch": 2.0713463751438437,
"grad_norm": 8.676543638522384,
"learning_rate": 5.551969977076349e-07,
"loss": 0.3124,
"step": 450
},
{
"epoch": 2.094361334867664,
"grad_norm": 11.853074754996486,
"learning_rate": 5.451916584814551e-07,
"loss": 0.3579,
"step": 455
},
{
"epoch": 2.1173762945914847,
"grad_norm": 16.53049309392707,
"learning_rate": 5.351680390391524e-07,
"loss": 0.3046,
"step": 460
},
{
"epoch": 2.140391254315305,
"grad_norm": 11.731708343253732,
"learning_rate": 5.25130193976993e-07,
"loss": 0.3058,
"step": 465
},
{
"epoch": 2.1634062140391253,
"grad_norm": 11.499082702306787,
"learning_rate": 5.150821836455659e-07,
"loss": 0.3376,
"step": 470
},
{
"epoch": 2.186421173762946,
"grad_norm": 12.34107746817991,
"learning_rate": 5.05028072507355e-07,
"loss": 0.3237,
"step": 475
},
{
"epoch": 2.2094361334867663,
"grad_norm": 17.370024261115674,
"learning_rate": 4.949719274926452e-07,
"loss": 0.2975,
"step": 480
},
{
"epoch": 2.232451093210587,
"grad_norm": 9.197724681633677,
"learning_rate": 4.84917816354434e-07,
"loss": 0.2967,
"step": 485
},
{
"epoch": 2.2554660529344073,
"grad_norm": 12.648733804765213,
"learning_rate": 4.748698060230071e-07,
"loss": 0.3198,
"step": 490
},
{
"epoch": 2.278481012658228,
"grad_norm": 13.903180310283751,
"learning_rate": 4.648319609608476e-07,
"loss": 0.3128,
"step": 495
},
{
"epoch": 2.3014959723820483,
"grad_norm": 15.830064172391179,
"learning_rate": 4.548083415185448e-07,
"loss": 0.311,
"step": 500
},
{
"epoch": 2.324510932105869,
"grad_norm": 14.674255723714507,
"learning_rate": 4.4480300229236517e-07,
"loss": 0.3207,
"step": 505
},
{
"epoch": 2.3475258918296893,
"grad_norm": 12.895709713108205,
"learning_rate": 4.3481999048415e-07,
"loss": 0.3417,
"step": 510
},
{
"epoch": 2.3705408515535096,
"grad_norm": 8.929819185065151,
"learning_rate": 4.248633442642067e-07,
"loss": 0.3177,
"step": 515
},
{
"epoch": 2.3935558112773303,
"grad_norm": 13.644125606933226,
"learning_rate": 4.1493709113785087e-07,
"loss": 0.3031,
"step": 520
},
{
"epoch": 2.4165707710011506,
"grad_norm": 11.673758922253713,
"learning_rate": 4.050452463162642e-07,
"loss": 0.3104,
"step": 525
},
{
"epoch": 2.4395857307249713,
"grad_norm": 12.820622173176774,
"learning_rate": 3.9519181109232345e-07,
"loss": 0.3063,
"step": 530
},
{
"epoch": 2.4626006904487916,
"grad_norm": 18.02707809486906,
"learning_rate": 3.853807712220622e-07,
"loss": 0.3025,
"step": 535
},
{
"epoch": 2.4856156501726123,
"grad_norm": 11.49239340519658,
"learning_rate": 3.756160953124149e-07,
"loss": 0.307,
"step": 540
},
{
"epoch": 2.5086306098964326,
"grad_norm": 13.193325188422616,
"learning_rate": 3.65901733215898e-07,
"loss": 0.317,
"step": 545
},
{
"epoch": 2.5316455696202533,
"grad_norm": 12.852961746212529,
"learning_rate": 3.5624161443287947e-07,
"loss": 0.3089,
"step": 550
},
{
"epoch": 2.5546605293440736,
"grad_norm": 15.085048095240678,
"learning_rate": 3.4663964652207836e-07,
"loss": 0.3162,
"step": 555
},
{
"epoch": 2.577675489067894,
"grad_norm": 15.098026200789873,
"learning_rate": 3.3709971351994126e-07,
"loss": 0.3326,
"step": 560
},
{
"epoch": 2.6006904487917146,
"grad_norm": 13.107735056669913,
"learning_rate": 3.276256743695324e-07,
"loss": 0.3041,
"step": 565
},
{
"epoch": 2.6237054085155354,
"grad_norm": 13.701942344015102,
"learning_rate": 3.1822136135957635e-07,
"loss": 0.3054,
"step": 570
},
{
"epoch": 2.6467203682393556,
"grad_norm": 12.49222926227634,
"learning_rate": 3.088905785742797e-07,
"loss": 0.3032,
"step": 575
},
{
"epoch": 2.669735327963176,
"grad_norm": 19.83941830585977,
"learning_rate": 2.9963710035456393e-07,
"loss": 0.3066,
"step": 580
},
{
"epoch": 2.6927502876869966,
"grad_norm": 12.875796703207728,
"learning_rate": 2.904646697713278e-07,
"loss": 0.2887,
"step": 585
},
{
"epoch": 2.715765247410817,
"grad_norm": 13.779139491323368,
"learning_rate": 2.813769971113608e-07,
"loss": 0.3195,
"step": 590
},
{
"epoch": 2.7387802071346377,
"grad_norm": 15.357477574862349,
"learning_rate": 2.72377758376515e-07,
"loss": 0.3369,
"step": 595
},
{
"epoch": 2.761795166858458,
"grad_norm": 14.296503558288,
"learning_rate": 2.634705937967471e-07,
"loss": 0.2943,
"step": 600
},
{
"epoch": 2.7848101265822782,
"grad_norm": 12.061535257713842,
"learning_rate": 2.546591063576312e-07,
"loss": 0.284,
"step": 605
},
{
"epoch": 2.807825086306099,
"grad_norm": 16.839618374963486,
"learning_rate": 2.459468603429345e-07,
"loss": 0.3,
"step": 610
},
{
"epoch": 2.8308400460299197,
"grad_norm": 14.189090458195743,
"learning_rate": 2.3733737989285068e-07,
"loss": 0.302,
"step": 615
},
{
"epoch": 2.85385500575374,
"grad_norm": 13.32887690082132,
"learning_rate": 2.2883414757846948e-07,
"loss": 0.3137,
"step": 620
},
{
"epoch": 2.8768699654775602,
"grad_norm": 12.574497648587789,
"learning_rate": 2.2044060299306267e-07,
"loss": 0.2839,
"step": 625
},
{
"epoch": 2.899884925201381,
"grad_norm": 15.048298566531198,
"learning_rate": 2.12160141360754e-07,
"loss": 0.2973,
"step": 630
},
{
"epoch": 2.9228998849252013,
"grad_norm": 11.951339536247458,
"learning_rate": 2.03996112163136e-07,
"loss": 0.3097,
"step": 635
},
{
"epoch": 2.945914844649022,
"grad_norm": 15.69565784514616,
"learning_rate": 1.9595181778439174e-07,
"loss": 0.2962,
"step": 640
},
{
"epoch": 2.9689298043728423,
"grad_norm": 12.999365393476868,
"learning_rate": 1.8803051217546584e-07,
"loss": 0.3009,
"step": 645
},
{
"epoch": 2.991944764096663,
"grad_norm": 14.935058458987008,
"learning_rate": 1.8023539953782735e-07,
"loss": 0.2872,
"step": 650
},
{
"epoch": 2.996547756041427,
"eval_loss": 0.6948590278625488,
"eval_runtime": 19.2791,
"eval_samples_per_second": 7.365,
"eval_steps_per_second": 0.934,
"step": 651
},
{
"epoch": 3.0149597238204833,
"grad_norm": 13.075885224885349,
"learning_rate": 1.725696330273575e-07,
"loss": 0.2512,
"step": 655
},
{
"epoch": 3.037974683544304,
"grad_norm": 9.704290953271537,
"learning_rate": 1.6503631347888436e-07,
"loss": 0.2256,
"step": 660
},
{
"epoch": 3.0609896432681243,
"grad_norm": 12.84818126176939,
"learning_rate": 1.5763848815188258e-07,
"loss": 0.2491,
"step": 665
},
{
"epoch": 3.0840046029919446,
"grad_norm": 11.493489134477986,
"learning_rate": 1.5037914949784296e-07,
"loss": 0.2386,
"step": 670
},
{
"epoch": 3.1070195627157653,
"grad_norm": 12.705781953882676,
"learning_rate": 1.432612339498148e-07,
"loss": 0.2488,
"step": 675
},
{
"epoch": 3.1300345224395856,
"grad_norm": 13.901431552375566,
"learning_rate": 1.3628762073460548e-07,
"loss": 0.2305,
"step": 680
},
{
"epoch": 3.1530494821634063,
"grad_norm": 14.715304886124786,
"learning_rate": 1.294611307081207e-07,
"loss": 0.2386,
"step": 685
},
{
"epoch": 3.1760644418872266,
"grad_norm": 10.222299037925385,
"learning_rate": 1.2278452521431744e-07,
"loss": 0.2339,
"step": 690
},
{
"epoch": 3.1990794016110473,
"grad_norm": 10.930465595389542,
"learning_rate": 1.1626050496822793e-07,
"loss": 0.2413,
"step": 695
},
{
"epoch": 3.2220943613348676,
"grad_norm": 12.451641988411806,
"learning_rate": 1.0989170896350947e-07,
"loss": 0.2284,
"step": 700
},
{
"epoch": 3.2451093210586883,
"grad_norm": 9.137656540405636,
"learning_rate": 1.0368071340495976e-07,
"loss": 0.2395,
"step": 705
},
{
"epoch": 3.2681242807825086,
"grad_norm": 11.671230840455808,
"learning_rate": 9.76300306664321e-08,
"loss": 0.253,
"step": 710
},
{
"epoch": 3.291139240506329,
"grad_norm": 17.933280474807955,
"learning_rate": 9.174210827456913e-08,
"loss": 0.2585,
"step": 715
},
{
"epoch": 3.3141542002301496,
"grad_norm": 12.982890796403161,
"learning_rate": 8.601932791876754e-08,
"loss": 0.2269,
"step": 720
},
{
"epoch": 3.33716915995397,
"grad_norm": 8.384602246690656,
"learning_rate": 8.046400448777574e-08,
"loss": 0.2253,
"step": 725
},
{
"epoch": 3.3601841196777906,
"grad_norm": 10.342720782810092,
"learning_rate": 7.507838513331049e-08,
"loss": 0.233,
"step": 730
},
{
"epoch": 3.383199079401611,
"grad_norm": 12.679510689599434,
"learning_rate": 6.986464836107547e-08,
"loss": 0.2433,
"step": 735
},
{
"epoch": 3.4062140391254316,
"grad_norm": 10.38040904275821,
"learning_rate": 6.48249031495462e-08,
"loss": 0.2071,
"step": 740
},
{
"epoch": 3.429228998849252,
"grad_norm": 12.061157277304762,
"learning_rate": 5.996118809687894e-08,
"loss": 0.2537,
"step": 745
},
{
"epoch": 3.4522439585730726,
"grad_norm": 14.129792097128606,
"learning_rate": 5.527547059629012e-08,
"loss": 0.2384,
"step": 750
},
{
"epoch": 3.475258918296893,
"grad_norm": 12.922540776147939,
"learning_rate": 5.0769646040236424e-08,
"loss": 0.2387,
"step": 755
},
{
"epoch": 3.4982738780207137,
"grad_norm": 11.979253675702983,
"learning_rate": 4.6445537053721396e-08,
"loss": 0.2294,
"step": 760
},
{
"epoch": 3.521288837744534,
"grad_norm": 11.457518174839912,
"learning_rate": 4.2304892757035636e-08,
"loss": 0.2264,
"step": 765
},
{
"epoch": 3.5443037974683547,
"grad_norm": 18.542321221268033,
"learning_rate": 3.834938805823029e-08,
"loss": 0.2424,
"step": 770
},
{
"epoch": 3.567318757192175,
"grad_norm": 13.26259377110348,
"learning_rate": 3.4580622975609375e-08,
"loss": 0.2435,
"step": 775
},
{
"epoch": 3.5903337169159952,
"grad_norm": 11.283666064750703,
"learning_rate": 3.100012199051627e-08,
"loss": 0.2321,
"step": 780
},
{
"epoch": 3.613348676639816,
"grad_norm": 33.16980829715486,
"learning_rate": 2.7609333430674232e-08,
"loss": 0.2396,
"step": 785
},
{
"epoch": 3.6363636363636362,
"grad_norm": 15.962685581367262,
"learning_rate": 2.4409628884331625e-08,
"loss": 0.2303,
"step": 790
},
{
"epoch": 3.659378596087457,
"grad_norm": 12.558852196507443,
"learning_rate": 2.14023026454489e-08,
"loss": 0.2455,
"step": 795
},
{
"epoch": 3.6823935558112773,
"grad_norm": 13.379534877412999,
"learning_rate": 1.8588571190151337e-08,
"loss": 0.2358,
"step": 800
},
{
"epoch": 3.705408515535098,
"grad_norm": 10.55910332644343,
"learning_rate": 1.5969572684658984e-08,
"loss": 0.2253,
"step": 805
},
{
"epoch": 3.7284234752589183,
"grad_norm": 14.737759161492995,
"learning_rate": 1.3546366524893827e-08,
"loss": 0.25,
"step": 810
},
{
"epoch": 3.751438434982739,
"grad_norm": 11.622900035948389,
"learning_rate": 1.1319932907949859e-08,
"loss": 0.2408,
"step": 815
},
{
"epoch": 3.7744533947065593,
"grad_norm": 12.11809870039906,
"learning_rate": 9.291172435598904e-09,
"loss": 0.2441,
"step": 820
},
{
"epoch": 3.7974683544303796,
"grad_norm": 18.918951484361397,
"learning_rate": 7.460905749993474e-09,
"loss": 0.2338,
"step": 825
},
{
"epoch": 3.8204833141542003,
"grad_norm": 14.793042348598215,
"learning_rate": 5.8298732017137975e-09,
"loss": 0.2405,
"step": 830
},
{
"epoch": 3.8434982738780206,
"grad_norm": 12.864105686917561,
"learning_rate": 4.398734550292715e-09,
"loss": 0.2329,
"step": 835
},
{
"epoch": 3.8665132336018413,
"grad_norm": 14.835178736224917,
"learning_rate": 3.168068697340043e-09,
"loss": 0.2528,
"step": 840
},
{
"epoch": 3.8895281933256616,
"grad_norm": 9.895215072689167,
"learning_rate": 2.1383734523748308e-09,
"loss": 0.2318,
"step": 845
},
{
"epoch": 3.9125431530494823,
"grad_norm": 11.324908576826193,
"learning_rate": 1.3100653314587761e-09,
"loss": 0.2492,
"step": 850
},
{
"epoch": 3.9355581127733026,
"grad_norm": 10.840567294920715,
"learning_rate": 6.834793887142143e-10,
"loss": 0.2295,
"step": 855
},
{
"epoch": 3.9585730724971233,
"grad_norm": 9.775718572541846,
"learning_rate": 2.5886908079308934e-10,
"loss": 0.2134,
"step": 860
},
{
"epoch": 3.9815880322209436,
"grad_norm": 12.104927788108858,
"learning_rate": 3.640616435257593e-11,
"loss": 0.2589,
"step": 865
},
{
"epoch": 3.995397008055236,
"eval_loss": 0.7718632221221924,
"eval_runtime": 19.3388,
"eval_samples_per_second": 7.343,
"eval_steps_per_second": 0.931,
"step": 868
},
{
"epoch": 3.995397008055236,
"step": 868,
"total_flos": 29536675078144.0,
"train_loss": 0.387863024725892,
"train_runtime": 11116.6413,
"train_samples_per_second": 2.501,
"train_steps_per_second": 0.078
}
],
"logging_steps": 5,
"max_steps": 868,
"num_input_tokens_seen": 0,
"num_train_epochs": 4,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": false,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 29536675078144.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}