sft_00 / trainer_state.json
iHateNLP's picture
End of training
d4fa5be verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.9985964912280703,
"eval_steps": 623,
"global_step": 2492,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0008020050125313284,
"grad_norm": 0.09629630297422409,
"learning_rate": 0.000499999801338539,
"loss": 0.8462,
"step": 1
},
{
"epoch": 0.0040100250626566416,
"grad_norm": 0.09335128962993622,
"learning_rate": 0.0004999950334792631,
"loss": 0.9167,
"step": 5
},
{
"epoch": 0.008020050125313283,
"grad_norm": 0.0814199447631836,
"learning_rate": 0.0004999801341143832,
"loss": 0.8982,
"step": 10
},
{
"epoch": 0.012030075187969926,
"grad_norm": 0.0830962136387825,
"learning_rate": 0.0004999553024973442,
"loss": 0.8568,
"step": 15
},
{
"epoch": 0.016040100250626566,
"grad_norm": 0.07535357028245926,
"learning_rate": 0.00049992053961476,
"loss": 0.8423,
"step": 20
},
{
"epoch": 0.020050125313283207,
"grad_norm": 0.08100854605436325,
"learning_rate": 0.0004998758468478354,
"loss": 0.822,
"step": 25
},
{
"epoch": 0.02406015037593985,
"grad_norm": 0.0720616802573204,
"learning_rate": 0.0004998212259723108,
"loss": 0.801,
"step": 30
},
{
"epoch": 0.028070175438596492,
"grad_norm": 0.07352820038795471,
"learning_rate": 0.0004997566791583916,
"loss": 0.819,
"step": 35
},
{
"epoch": 0.03208020050125313,
"grad_norm": 0.07747522741556168,
"learning_rate": 0.0004996822089706628,
"loss": 0.7917,
"step": 40
},
{
"epoch": 0.03609022556390978,
"grad_norm": 0.0775732770562172,
"learning_rate": 0.0004995978183679864,
"loss": 0.8007,
"step": 45
},
{
"epoch": 0.040100250626566414,
"grad_norm": 0.08353012800216675,
"learning_rate": 0.0004995035107033833,
"loss": 0.7939,
"step": 50
},
{
"epoch": 0.04411027568922306,
"grad_norm": 0.09015482664108276,
"learning_rate": 0.0004993992897239018,
"loss": 0.7844,
"step": 55
},
{
"epoch": 0.0481203007518797,
"grad_norm": 0.08346273750066757,
"learning_rate": 0.0004992851595704668,
"loss": 0.7805,
"step": 60
},
{
"epoch": 0.05213032581453634,
"grad_norm": 0.08374553918838501,
"learning_rate": 0.0004991611247777166,
"loss": 0.7758,
"step": 65
},
{
"epoch": 0.056140350877192984,
"grad_norm": 0.0855209082365036,
"learning_rate": 0.0004990271902738223,
"loss": 0.7799,
"step": 70
},
{
"epoch": 0.06015037593984962,
"grad_norm": 0.08624681085348129,
"learning_rate": 0.0004988833613802917,
"loss": 0.7461,
"step": 75
},
{
"epoch": 0.06416040100250626,
"grad_norm": 0.08796326816082001,
"learning_rate": 0.0004987296438117581,
"loss": 0.7181,
"step": 80
},
{
"epoch": 0.06817042606516291,
"grad_norm": 0.08574078977108002,
"learning_rate": 0.0004985660436757538,
"loss": 0.7209,
"step": 85
},
{
"epoch": 0.07218045112781955,
"grad_norm": 0.09403970092535019,
"learning_rate": 0.0004983925674724662,
"loss": 0.7297,
"step": 90
},
{
"epoch": 0.0761904761904762,
"grad_norm": 0.1051216572523117,
"learning_rate": 0.0004982092220944808,
"loss": 0.7077,
"step": 95
},
{
"epoch": 0.08020050125313283,
"grad_norm": 0.10198145359754562,
"learning_rate": 0.0004980160148265066,
"loss": 0.7369,
"step": 100
},
{
"epoch": 0.08421052631578947,
"grad_norm": 0.10444864630699158,
"learning_rate": 0.0004978129533450866,
"loss": 0.7044,
"step": 105
},
{
"epoch": 0.08822055137844612,
"grad_norm": 0.09810823947191238,
"learning_rate": 0.0004976000457182935,
"loss": 0.7055,
"step": 110
},
{
"epoch": 0.09223057644110276,
"grad_norm": 0.10619106888771057,
"learning_rate": 0.0004973773004054081,
"loss": 0.6951,
"step": 115
},
{
"epoch": 0.0962406015037594,
"grad_norm": 0.11827559024095535,
"learning_rate": 0.0004971447262565846,
"loss": 0.7285,
"step": 120
},
{
"epoch": 0.10025062656641603,
"grad_norm": 0.09872109442949295,
"learning_rate": 0.0004969023325124974,
"loss": 0.7065,
"step": 125
},
{
"epoch": 0.10426065162907268,
"grad_norm": 0.11186440289020538,
"learning_rate": 0.0004966501288039749,
"loss": 0.6663,
"step": 130
},
{
"epoch": 0.10827067669172932,
"grad_norm": 0.10218355059623718,
"learning_rate": 0.0004963881251516169,
"loss": 0.6763,
"step": 135
},
{
"epoch": 0.11228070175438597,
"grad_norm": 0.11396008729934692,
"learning_rate": 0.0004961163319653958,
"loss": 0.6704,
"step": 140
},
{
"epoch": 0.11629072681704261,
"grad_norm": 0.10969991981983185,
"learning_rate": 0.0004958347600442437,
"loss": 0.6779,
"step": 145
},
{
"epoch": 0.12030075187969924,
"grad_norm": 0.11111714690923691,
"learning_rate": 0.0004955434205756226,
"loss": 0.6835,
"step": 150
},
{
"epoch": 0.12431077694235589,
"grad_norm": 0.11582209914922714,
"learning_rate": 0.000495242325135081,
"loss": 0.6966,
"step": 155
},
{
"epoch": 0.12832080200501253,
"grad_norm": 0.11664141714572906,
"learning_rate": 0.0004949314856857926,
"loss": 0.641,
"step": 160
},
{
"epoch": 0.13233082706766916,
"grad_norm": 0.11299869418144226,
"learning_rate": 0.000494610914578082,
"loss": 0.6349,
"step": 165
},
{
"epoch": 0.13634085213032582,
"grad_norm": 0.10604275017976761,
"learning_rate": 0.0004942806245489338,
"loss": 0.6498,
"step": 170
},
{
"epoch": 0.14035087719298245,
"grad_norm": 0.11253874003887177,
"learning_rate": 0.000493940628721486,
"loss": 0.6536,
"step": 175
},
{
"epoch": 0.1443609022556391,
"grad_norm": 0.12596091628074646,
"learning_rate": 0.0004935909406045095,
"loss": 0.6348,
"step": 180
},
{
"epoch": 0.14837092731829574,
"grad_norm": 0.12171223014593124,
"learning_rate": 0.0004932315740918703,
"loss": 0.64,
"step": 185
},
{
"epoch": 0.1523809523809524,
"grad_norm": 0.12628182768821716,
"learning_rate": 0.0004928625434619785,
"loss": 0.6225,
"step": 190
},
{
"epoch": 0.15639097744360902,
"grad_norm": 0.12631121277809143,
"learning_rate": 0.0004924838633772201,
"loss": 0.6197,
"step": 195
},
{
"epoch": 0.16040100250626566,
"grad_norm": 0.12037768959999084,
"learning_rate": 0.0004920955488833753,
"loss": 0.6201,
"step": 200
},
{
"epoch": 0.1644110275689223,
"grad_norm": 0.11302505433559418,
"learning_rate": 0.0004916976154090198,
"loss": 0.5972,
"step": 205
},
{
"epoch": 0.16842105263157894,
"grad_norm": 0.15172865986824036,
"learning_rate": 0.0004912900787649124,
"loss": 0.6322,
"step": 210
},
{
"epoch": 0.1724310776942356,
"grad_norm": 0.12461714446544647,
"learning_rate": 0.0004908729551433668,
"loss": 0.5681,
"step": 215
},
{
"epoch": 0.17644110275689223,
"grad_norm": 0.11686723679304123,
"learning_rate": 0.0004904462611176079,
"loss": 0.6112,
"step": 220
},
{
"epoch": 0.18045112781954886,
"grad_norm": 0.12082964181900024,
"learning_rate": 0.0004900100136411134,
"loss": 0.59,
"step": 225
},
{
"epoch": 0.18446115288220552,
"grad_norm": 0.14786957204341888,
"learning_rate": 0.0004895642300469405,
"loss": 0.6008,
"step": 230
},
{
"epoch": 0.18847117794486215,
"grad_norm": 0.11940759420394897,
"learning_rate": 0.0004891089280470369,
"loss": 0.5781,
"step": 235
},
{
"epoch": 0.1924812030075188,
"grad_norm": 0.12067493796348572,
"learning_rate": 0.0004886441257315373,
"loss": 0.5963,
"step": 240
},
{
"epoch": 0.19649122807017544,
"grad_norm": 0.12630167603492737,
"learning_rate": 0.0004881698415680442,
"loss": 0.5767,
"step": 245
},
{
"epoch": 0.20050125313283207,
"grad_norm": 0.12695525586605072,
"learning_rate": 0.00048768609440089474,
"loss": 0.6039,
"step": 250
},
{
"epoch": 0.20451127819548873,
"grad_norm": 0.13445709645748138,
"learning_rate": 0.00048719290345041166,
"loss": 0.6087,
"step": 255
},
{
"epoch": 0.20852130325814536,
"grad_norm": 0.13432475924491882,
"learning_rate": 0.0004866902883121397,
"loss": 0.5491,
"step": 260
},
{
"epoch": 0.21253132832080202,
"grad_norm": 0.15156829357147217,
"learning_rate": 0.0004861782689560667,
"loss": 0.5492,
"step": 265
},
{
"epoch": 0.21654135338345865,
"grad_norm": 0.13093040883541107,
"learning_rate": 0.0004856568657258308,
"loss": 0.6045,
"step": 270
},
{
"epoch": 0.22055137844611528,
"grad_norm": 0.1290189027786255,
"learning_rate": 0.00048512609933791165,
"loss": 0.5433,
"step": 275
},
{
"epoch": 0.22456140350877193,
"grad_norm": 0.13374298810958862,
"learning_rate": 0.00048458599088080736,
"loss": 0.5575,
"step": 280
},
{
"epoch": 0.22857142857142856,
"grad_norm": 0.1266585886478424,
"learning_rate": 0.00048403656181419686,
"loss": 0.5474,
"step": 285
},
{
"epoch": 0.23258145363408522,
"grad_norm": 0.13824708759784698,
"learning_rate": 0.00048347783396808687,
"loss": 0.5465,
"step": 290
},
{
"epoch": 0.23659147869674185,
"grad_norm": 0.12988659739494324,
"learning_rate": 0.00048290982954194485,
"loss": 0.5605,
"step": 295
},
{
"epoch": 0.24060150375939848,
"grad_norm": 0.13609863817691803,
"learning_rate": 0.0004823325711038169,
"loss": 0.5812,
"step": 300
},
{
"epoch": 0.24461152882205514,
"grad_norm": 0.14271323382854462,
"learning_rate": 0.00048174608158943114,
"loss": 0.5485,
"step": 305
},
{
"epoch": 0.24862155388471177,
"grad_norm": 0.13438940048217773,
"learning_rate": 0.0004811503843012861,
"loss": 0.5325,
"step": 310
},
{
"epoch": 0.25263157894736843,
"grad_norm": 0.138273686170578,
"learning_rate": 0.0004805455029077255,
"loss": 0.5283,
"step": 315
},
{
"epoch": 0.25664160401002506,
"grad_norm": 0.15901249647140503,
"learning_rate": 0.0004799314614419969,
"loss": 0.5717,
"step": 320
},
{
"epoch": 0.2606516290726817,
"grad_norm": 0.14647658169269562,
"learning_rate": 0.0004793082843012979,
"loss": 0.5378,
"step": 325
},
{
"epoch": 0.2646616541353383,
"grad_norm": 0.1300746202468872,
"learning_rate": 0.0004786759962458059,
"loss": 0.5183,
"step": 330
},
{
"epoch": 0.268671679197995,
"grad_norm": 0.13711751997470856,
"learning_rate": 0.000478034622397695,
"loss": 0.5442,
"step": 335
},
{
"epoch": 0.27268170426065164,
"grad_norm": 0.150499626994133,
"learning_rate": 0.0004773841882401372,
"loss": 0.524,
"step": 340
},
{
"epoch": 0.27669172932330827,
"grad_norm": 0.13912473618984222,
"learning_rate": 0.00047672471961629037,
"loss": 0.5412,
"step": 345
},
{
"epoch": 0.2807017543859649,
"grad_norm": 0.1455031782388687,
"learning_rate": 0.0004760562427282712,
"loss": 0.5321,
"step": 350
},
{
"epoch": 0.2847117794486215,
"grad_norm": 0.12485536932945251,
"learning_rate": 0.0004753787841361145,
"loss": 0.5113,
"step": 355
},
{
"epoch": 0.2887218045112782,
"grad_norm": 0.13250643014907837,
"learning_rate": 0.00047469237075671724,
"loss": 0.5402,
"step": 360
},
{
"epoch": 0.29273182957393484,
"grad_norm": 0.13680434226989746,
"learning_rate": 0.00047399702986276984,
"loss": 0.5751,
"step": 365
},
{
"epoch": 0.2967418546365915,
"grad_norm": 0.14655643701553345,
"learning_rate": 0.0004732927890816719,
"loss": 0.5371,
"step": 370
},
{
"epoch": 0.3007518796992481,
"grad_norm": 0.14064128696918488,
"learning_rate": 0.00047257967639443513,
"loss": 0.531,
"step": 375
},
{
"epoch": 0.3047619047619048,
"grad_norm": 0.14199021458625793,
"learning_rate": 0.00047185772013457096,
"loss": 0.4934,
"step": 380
},
{
"epoch": 0.3087719298245614,
"grad_norm": 0.14177988469600677,
"learning_rate": 0.00047112694898696533,
"loss": 0.5249,
"step": 385
},
{
"epoch": 0.31278195488721805,
"grad_norm": 0.14135052263736725,
"learning_rate": 0.00047038739198673876,
"loss": 0.5272,
"step": 390
},
{
"epoch": 0.3167919799498747,
"grad_norm": 0.12976044416427612,
"learning_rate": 0.00046963907851809263,
"loss": 0.5263,
"step": 395
},
{
"epoch": 0.3208020050125313,
"grad_norm": 0.15720346570014954,
"learning_rate": 0.0004688820383131418,
"loss": 0.4858,
"step": 400
},
{
"epoch": 0.324812030075188,
"grad_norm": 0.14797399938106537,
"learning_rate": 0.0004681163014507334,
"loss": 0.5201,
"step": 405
},
{
"epoch": 0.3288220551378446,
"grad_norm": 0.15095673501491547,
"learning_rate": 0.0004673418983552513,
"loss": 0.5366,
"step": 410
},
{
"epoch": 0.33283208020050126,
"grad_norm": 0.16217216849327087,
"learning_rate": 0.00046655885979540783,
"loss": 0.4836,
"step": 415
},
{
"epoch": 0.3368421052631579,
"grad_norm": 0.14743471145629883,
"learning_rate": 0.00046576721688302105,
"loss": 0.5067,
"step": 420
},
{
"epoch": 0.3408521303258145,
"grad_norm": 0.15485595166683197,
"learning_rate": 0.00046496700107177835,
"loss": 0.4614,
"step": 425
},
{
"epoch": 0.3448621553884712,
"grad_norm": 0.15458199381828308,
"learning_rate": 0.0004641582441559872,
"loss": 0.4897,
"step": 430
},
{
"epoch": 0.34887218045112783,
"grad_norm": 0.13934871554374695,
"learning_rate": 0.00046334097826931144,
"loss": 0.522,
"step": 435
},
{
"epoch": 0.35288220551378446,
"grad_norm": 0.1473606377840042,
"learning_rate": 0.00046251523588349485,
"loss": 0.4851,
"step": 440
},
{
"epoch": 0.3568922305764411,
"grad_norm": 0.14684675633907318,
"learning_rate": 0.00046168104980707104,
"loss": 0.4888,
"step": 445
},
{
"epoch": 0.3609022556390977,
"grad_norm": 0.15129096806049347,
"learning_rate": 0.0004608384531840595,
"loss": 0.5236,
"step": 450
},
{
"epoch": 0.3649122807017544,
"grad_norm": 0.15553347766399384,
"learning_rate": 0.00045998747949264895,
"loss": 0.4791,
"step": 455
},
{
"epoch": 0.36892230576441104,
"grad_norm": 0.16991403698921204,
"learning_rate": 0.00045912816254386747,
"loss": 0.4672,
"step": 460
},
{
"epoch": 0.37293233082706767,
"grad_norm": 0.15001340210437775,
"learning_rate": 0.0004582605364802385,
"loss": 0.4989,
"step": 465
},
{
"epoch": 0.3769423558897243,
"grad_norm": 0.15301191806793213,
"learning_rate": 0.00045738463577442467,
"loss": 0.4885,
"step": 470
},
{
"epoch": 0.38095238095238093,
"grad_norm": 0.1510782241821289,
"learning_rate": 0.00045650049522785827,
"loss": 0.4827,
"step": 475
},
{
"epoch": 0.3849624060150376,
"grad_norm": 0.14636896550655365,
"learning_rate": 0.0004556081499693581,
"loss": 0.4791,
"step": 480
},
{
"epoch": 0.38897243107769425,
"grad_norm": 0.1476137489080429,
"learning_rate": 0.00045470763545373395,
"loss": 0.4709,
"step": 485
},
{
"epoch": 0.3929824561403509,
"grad_norm": 0.15410684049129486,
"learning_rate": 0.000453798987460378,
"loss": 0.4794,
"step": 490
},
{
"epoch": 0.3969924812030075,
"grad_norm": 0.14843297004699707,
"learning_rate": 0.00045288224209184315,
"loss": 0.5077,
"step": 495
},
{
"epoch": 0.40100250626566414,
"grad_norm": 0.1490759700536728,
"learning_rate": 0.00045195743577240835,
"loss": 0.464,
"step": 500
},
{
"epoch": 0.4050125313283208,
"grad_norm": 0.17478294670581818,
"learning_rate": 0.0004510246052466317,
"loss": 0.4754,
"step": 505
},
{
"epoch": 0.40902255639097745,
"grad_norm": 0.15610414743423462,
"learning_rate": 0.0004500837875778905,
"loss": 0.488,
"step": 510
},
{
"epoch": 0.4130325814536341,
"grad_norm": 0.15166425704956055,
"learning_rate": 0.00044913502014690837,
"loss": 0.4772,
"step": 515
},
{
"epoch": 0.4170426065162907,
"grad_norm": 0.14028780162334442,
"learning_rate": 0.0004481783406502703,
"loss": 0.5013,
"step": 520
},
{
"epoch": 0.42105263157894735,
"grad_norm": 0.17099718749523163,
"learning_rate": 0.0004472137870989247,
"loss": 0.4811,
"step": 525
},
{
"epoch": 0.42506265664160403,
"grad_norm": 0.1719314604997635,
"learning_rate": 0.0004462413978166734,
"loss": 0.4735,
"step": 530
},
{
"epoch": 0.42907268170426066,
"grad_norm": 0.17192725837230682,
"learning_rate": 0.0004452612114386485,
"loss": 0.4945,
"step": 535
},
{
"epoch": 0.4330827067669173,
"grad_norm": 0.1921142190694809,
"learning_rate": 0.000444273266909778,
"loss": 0.4975,
"step": 540
},
{
"epoch": 0.4370927318295739,
"grad_norm": 0.16015449166297913,
"learning_rate": 0.00044327760348323767,
"loss": 0.4566,
"step": 545
},
{
"epoch": 0.44110275689223055,
"grad_norm": 0.15686671435832977,
"learning_rate": 0.000442274260718892,
"loss": 0.449,
"step": 550
},
{
"epoch": 0.44511278195488724,
"grad_norm": 0.1583927422761917,
"learning_rate": 0.0004412632784817221,
"loss": 0.4738,
"step": 555
},
{
"epoch": 0.44912280701754387,
"grad_norm": 0.1498887687921524,
"learning_rate": 0.00044024469694024196,
"loss": 0.4416,
"step": 560
},
{
"epoch": 0.4531328320802005,
"grad_norm": 0.16318589448928833,
"learning_rate": 0.0004392185565649024,
"loss": 0.4527,
"step": 565
},
{
"epoch": 0.45714285714285713,
"grad_norm": 0.14862380921840668,
"learning_rate": 0.000438184898126483,
"loss": 0.4613,
"step": 570
},
{
"epoch": 0.46115288220551376,
"grad_norm": 0.17321060597896576,
"learning_rate": 0.00043714376269447233,
"loss": 0.4587,
"step": 575
},
{
"epoch": 0.46516290726817044,
"grad_norm": 0.15772369503974915,
"learning_rate": 0.0004360951916354361,
"loss": 0.4546,
"step": 580
},
{
"epoch": 0.4691729323308271,
"grad_norm": 0.1607331931591034,
"learning_rate": 0.0004350392266113736,
"loss": 0.4836,
"step": 585
},
{
"epoch": 0.4731829573934837,
"grad_norm": 0.16422398388385773,
"learning_rate": 0.0004339759095780623,
"loss": 0.446,
"step": 590
},
{
"epoch": 0.47719298245614034,
"grad_norm": 0.15711261332035065,
"learning_rate": 0.00043290528278339097,
"loss": 0.4308,
"step": 595
},
{
"epoch": 0.48120300751879697,
"grad_norm": 0.15732958912849426,
"learning_rate": 0.0004318273887656811,
"loss": 0.4687,
"step": 600
},
{
"epoch": 0.48521303258145365,
"grad_norm": 0.15295931696891785,
"learning_rate": 0.00043074227035199656,
"loss": 0.4474,
"step": 605
},
{
"epoch": 0.4892230576441103,
"grad_norm": 0.16534899175167084,
"learning_rate": 0.00042964997065644204,
"loss": 0.4552,
"step": 610
},
{
"epoch": 0.4932330827067669,
"grad_norm": 0.15630511939525604,
"learning_rate": 0.00042855053307845056,
"loss": 0.4842,
"step": 615
},
{
"epoch": 0.49724310776942354,
"grad_norm": 0.1624716818332672,
"learning_rate": 0.0004274440013010581,
"loss": 0.4228,
"step": 620
},
{
"epoch": 0.49964912280701756,
"eval_loss": 0.33371835947036743,
"eval_runtime": 8.4047,
"eval_samples_per_second": 6.544,
"eval_steps_per_second": 0.833,
"step": 623
},
{
"epoch": 0.5012531328320802,
"grad_norm": 0.16876687109470367,
"learning_rate": 0.0004263304192891689,
"loss": 0.4628,
"step": 625
},
{
"epoch": 0.5052631578947369,
"grad_norm": 0.16050459444522858,
"learning_rate": 0.00042520983128780824,
"loss": 0.422,
"step": 630
},
{
"epoch": 0.5092731829573934,
"grad_norm": 0.1509382724761963,
"learning_rate": 0.00042408228182036446,
"loss": 0.4473,
"step": 635
},
{
"epoch": 0.5132832080200501,
"grad_norm": 0.16160839796066284,
"learning_rate": 0.00042294781568682,
"loss": 0.4264,
"step": 640
},
{
"epoch": 0.5172932330827068,
"grad_norm": 0.1524878293275833,
"learning_rate": 0.0004218064779619715,
"loss": 0.4332,
"step": 645
},
{
"epoch": 0.5213032581453634,
"grad_norm": 0.18008528649806976,
"learning_rate": 0.00042065831399363884,
"loss": 0.4365,
"step": 650
},
{
"epoch": 0.5253132832080201,
"grad_norm": 0.16481877863407135,
"learning_rate": 0.0004195033694008632,
"loss": 0.4417,
"step": 655
},
{
"epoch": 0.5293233082706766,
"grad_norm": 0.16098377108573914,
"learning_rate": 0.00041834169007209487,
"loss": 0.4459,
"step": 660
},
{
"epoch": 0.5333333333333333,
"grad_norm": 0.15441514551639557,
"learning_rate": 0.0004171733221633695,
"loss": 0.4178,
"step": 665
},
{
"epoch": 0.53734335839599,
"grad_norm": 0.15172994136810303,
"learning_rate": 0.00041599831209647475,
"loss": 0.4261,
"step": 670
},
{
"epoch": 0.5413533834586466,
"grad_norm": 0.16424188017845154,
"learning_rate": 0.0004148167065571055,
"loss": 0.4375,
"step": 675
},
{
"epoch": 0.5453634085213033,
"grad_norm": 0.1634334772825241,
"learning_rate": 0.0004136285524930091,
"loss": 0.4453,
"step": 680
},
{
"epoch": 0.5493734335839598,
"grad_norm": 0.1613541692495346,
"learning_rate": 0.00041243389711211997,
"loss": 0.4835,
"step": 685
},
{
"epoch": 0.5533834586466165,
"grad_norm": 0.16017889976501465,
"learning_rate": 0.00041123278788068375,
"loss": 0.4245,
"step": 690
},
{
"epoch": 0.5573934837092732,
"grad_norm": 0.1616198718547821,
"learning_rate": 0.0004100252725213718,
"loss": 0.4757,
"step": 695
},
{
"epoch": 0.5614035087719298,
"grad_norm": 0.15503592789173126,
"learning_rate": 0.00040881139901138467,
"loss": 0.438,
"step": 700
},
{
"epoch": 0.5654135338345865,
"grad_norm": 0.15209273993968964,
"learning_rate": 0.000407591215580546,
"loss": 0.4258,
"step": 705
},
{
"epoch": 0.569423558897243,
"grad_norm": 0.16087529063224792,
"learning_rate": 0.0004063647707093864,
"loss": 0.4319,
"step": 710
},
{
"epoch": 0.5734335839598997,
"grad_norm": 0.16061115264892578,
"learning_rate": 0.0004051321131272167,
"loss": 0.4562,
"step": 715
},
{
"epoch": 0.5774436090225564,
"grad_norm": 0.15838468074798584,
"learning_rate": 0.0004038932918101927,
"loss": 0.4171,
"step": 720
},
{
"epoch": 0.581453634085213,
"grad_norm": 0.16260045766830444,
"learning_rate": 0.00040264835597936846,
"loss": 0.4197,
"step": 725
},
{
"epoch": 0.5854636591478697,
"grad_norm": 0.20412659645080566,
"learning_rate": 0.0004013973550987408,
"loss": 0.4399,
"step": 730
},
{
"epoch": 0.5894736842105263,
"grad_norm": 0.16249172389507294,
"learning_rate": 0.0004001403388732842,
"loss": 0.4141,
"step": 735
},
{
"epoch": 0.593483709273183,
"grad_norm": 0.16174912452697754,
"learning_rate": 0.0003988773572469761,
"loss": 0.4235,
"step": 740
},
{
"epoch": 0.5974937343358396,
"grad_norm": 0.16679826378822327,
"learning_rate": 0.0003976084604008119,
"loss": 0.4078,
"step": 745
},
{
"epoch": 0.6015037593984962,
"grad_norm": 0.1594962477684021,
"learning_rate": 0.0003963336987508115,
"loss": 0.4229,
"step": 750
},
{
"epoch": 0.6055137844611529,
"grad_norm": 0.1605675369501114,
"learning_rate": 0.00039505312294601635,
"loss": 0.4084,
"step": 755
},
{
"epoch": 0.6095238095238096,
"grad_norm": 0.16422423720359802,
"learning_rate": 0.0003937667838664768,
"loss": 0.3926,
"step": 760
},
{
"epoch": 0.6135338345864662,
"grad_norm": 0.15119189023971558,
"learning_rate": 0.00039247473262123036,
"loss": 0.3998,
"step": 765
},
{
"epoch": 0.6175438596491228,
"grad_norm": 0.1493472158908844,
"learning_rate": 0.0003911770205462717,
"loss": 0.3965,
"step": 770
},
{
"epoch": 0.6215538847117794,
"grad_norm": 0.16699016094207764,
"learning_rate": 0.00038987369920251213,
"loss": 0.4089,
"step": 775
},
{
"epoch": 0.6255639097744361,
"grad_norm": 0.15896373987197876,
"learning_rate": 0.00038856482037373155,
"loss": 0.3785,
"step": 780
},
{
"epoch": 0.6295739348370928,
"grad_norm": 0.16796669363975525,
"learning_rate": 0.0003872504360645206,
"loss": 0.3973,
"step": 785
},
{
"epoch": 0.6335839598997494,
"grad_norm": 0.14833569526672363,
"learning_rate": 0.0003859305984982149,
"loss": 0.4149,
"step": 790
},
{
"epoch": 0.637593984962406,
"grad_norm": 0.1650949865579605,
"learning_rate": 0.0003846053601148194,
"loss": 0.4253,
"step": 795
},
{
"epoch": 0.6416040100250626,
"grad_norm": 0.15514247119426727,
"learning_rate": 0.00038327477356892546,
"loss": 0.3654,
"step": 800
},
{
"epoch": 0.6456140350877193,
"grad_norm": 0.16105593740940094,
"learning_rate": 0.0003819388917276185,
"loss": 0.4399,
"step": 805
},
{
"epoch": 0.649624060150376,
"grad_norm": 0.15280301868915558,
"learning_rate": 0.00038059776766837753,
"loss": 0.4166,
"step": 810
},
{
"epoch": 0.6536340852130326,
"grad_norm": 0.16363288462162018,
"learning_rate": 0.000379251454676966,
"loss": 0.3977,
"step": 815
},
{
"epoch": 0.6576441102756893,
"grad_norm": 0.16015923023223877,
"learning_rate": 0.00037790000624531505,
"loss": 0.3918,
"step": 820
},
{
"epoch": 0.6616541353383458,
"grad_norm": 0.14514611661434174,
"learning_rate": 0.0003765434760693981,
"loss": 0.3801,
"step": 825
},
{
"epoch": 0.6656641604010025,
"grad_norm": 0.16084541380405426,
"learning_rate": 0.0003751819180470969,
"loss": 0.4123,
"step": 830
},
{
"epoch": 0.6696741854636592,
"grad_norm": 0.14666476845741272,
"learning_rate": 0.000373815386276061,
"loss": 0.3712,
"step": 835
},
{
"epoch": 0.6736842105263158,
"grad_norm": 0.16997689008712769,
"learning_rate": 0.0003724439350515571,
"loss": 0.3999,
"step": 840
},
{
"epoch": 0.6776942355889725,
"grad_norm": 0.15905480086803436,
"learning_rate": 0.000371067618864313,
"loss": 0.4068,
"step": 845
},
{
"epoch": 0.681704260651629,
"grad_norm": 0.1562589704990387,
"learning_rate": 0.0003696864923983516,
"loss": 0.3843,
"step": 850
},
{
"epoch": 0.6857142857142857,
"grad_norm": 0.16755416989326477,
"learning_rate": 0.00036830061052881897,
"loss": 0.4426,
"step": 855
},
{
"epoch": 0.6897243107769424,
"grad_norm": 0.14882604777812958,
"learning_rate": 0.00036691002831980323,
"loss": 0.3939,
"step": 860
},
{
"epoch": 0.693734335839599,
"grad_norm": 0.16580046713352203,
"learning_rate": 0.0003655148010221476,
"loss": 0.3823,
"step": 865
},
{
"epoch": 0.6977443609022557,
"grad_norm": 0.15855063498020172,
"learning_rate": 0.00036411498407125435,
"loss": 0.3907,
"step": 870
},
{
"epoch": 0.7017543859649122,
"grad_norm": 0.1665424108505249,
"learning_rate": 0.00036271063308488294,
"loss": 0.4133,
"step": 875
},
{
"epoch": 0.7057644110275689,
"grad_norm": 0.1685834378004074,
"learning_rate": 0.0003613018038609398,
"loss": 0.4359,
"step": 880
},
{
"epoch": 0.7097744360902256,
"grad_norm": 0.15352065861225128,
"learning_rate": 0.0003598885523752611,
"loss": 0.4037,
"step": 885
},
{
"epoch": 0.7137844611528822,
"grad_norm": 0.17349396646022797,
"learning_rate": 0.00035847093477938953,
"loss": 0.411,
"step": 890
},
{
"epoch": 0.7177944862155389,
"grad_norm": 0.15559545159339905,
"learning_rate": 0.0003570490073983425,
"loss": 0.3817,
"step": 895
},
{
"epoch": 0.7218045112781954,
"grad_norm": 0.16843955218791962,
"learning_rate": 0.00035562282672837477,
"loss": 0.4162,
"step": 900
},
{
"epoch": 0.7258145363408521,
"grad_norm": 0.17087852954864502,
"learning_rate": 0.00035419244943473305,
"loss": 0.412,
"step": 905
},
{
"epoch": 0.7298245614035088,
"grad_norm": 0.16542008519172668,
"learning_rate": 0.00035275793234940545,
"loss": 0.421,
"step": 910
},
{
"epoch": 0.7338345864661654,
"grad_norm": 0.17914004623889923,
"learning_rate": 0.0003513193324688627,
"loss": 0.3759,
"step": 915
},
{
"epoch": 0.7378446115288221,
"grad_norm": 0.16926446557044983,
"learning_rate": 0.00034987670695179403,
"loss": 0.3988,
"step": 920
},
{
"epoch": 0.7418546365914787,
"grad_norm": 0.15922464430332184,
"learning_rate": 0.00034843011311683566,
"loss": 0.3641,
"step": 925
},
{
"epoch": 0.7458646616541353,
"grad_norm": 0.16068434715270996,
"learning_rate": 0.00034697960844029383,
"loss": 0.3746,
"step": 930
},
{
"epoch": 0.749874686716792,
"grad_norm": 0.18251362442970276,
"learning_rate": 0.0003455252505538611,
"loss": 0.4308,
"step": 935
},
{
"epoch": 0.7538847117794486,
"grad_norm": 0.1559171974658966,
"learning_rate": 0.0003440670972423262,
"loss": 0.4045,
"step": 940
},
{
"epoch": 0.7578947368421053,
"grad_norm": 0.17260219156742096,
"learning_rate": 0.00034260520644127846,
"loss": 0.3642,
"step": 945
},
{
"epoch": 0.7619047619047619,
"grad_norm": 0.1719062477350235,
"learning_rate": 0.0003411396362348056,
"loss": 0.3973,
"step": 950
},
{
"epoch": 0.7659147869674185,
"grad_norm": 0.16649049520492554,
"learning_rate": 0.0003396704448531863,
"loss": 0.4031,
"step": 955
},
{
"epoch": 0.7699248120300752,
"grad_norm": 0.15968754887580872,
"learning_rate": 0.00033819769067057617,
"loss": 0.3648,
"step": 960
},
{
"epoch": 0.7739348370927318,
"grad_norm": 0.18287472426891327,
"learning_rate": 0.0003367214322026888,
"loss": 0.3921,
"step": 965
},
{
"epoch": 0.7779448621553885,
"grad_norm": 0.1621864289045334,
"learning_rate": 0.00033524172810447055,
"loss": 0.3956,
"step": 970
},
{
"epoch": 0.7819548872180451,
"grad_norm": 0.1772567480802536,
"learning_rate": 0.00033375863716777007,
"loss": 0.3745,
"step": 975
},
{
"epoch": 0.7859649122807018,
"grad_norm": 0.17318162322044373,
"learning_rate": 0.0003322722183190025,
"loss": 0.3924,
"step": 980
},
{
"epoch": 0.7899749373433584,
"grad_norm": 0.17136339843273163,
"learning_rate": 0.0003307825306168082,
"loss": 0.3897,
"step": 985
},
{
"epoch": 0.793984962406015,
"grad_norm": 0.15933868288993835,
"learning_rate": 0.00032928963324970595,
"loss": 0.3723,
"step": 990
},
{
"epoch": 0.7979949874686717,
"grad_norm": 0.1778361201286316,
"learning_rate": 0.0003277935855337417,
"loss": 0.3822,
"step": 995
},
{
"epoch": 0.8020050125313283,
"grad_norm": 0.15671303868293762,
"learning_rate": 0.00032629444691013144,
"loss": 0.3805,
"step": 1000
},
{
"epoch": 0.806015037593985,
"grad_norm": 0.16916193068027496,
"learning_rate": 0.0003247922769428998,
"loss": 0.3931,
"step": 1005
},
{
"epoch": 0.8100250626566416,
"grad_norm": 0.15111452341079712,
"learning_rate": 0.0003232871353165129,
"loss": 0.3398,
"step": 1010
},
{
"epoch": 0.8140350877192982,
"grad_norm": 0.14872141182422638,
"learning_rate": 0.0003217790818335077,
"loss": 0.3541,
"step": 1015
},
{
"epoch": 0.8180451127819549,
"grad_norm": 0.1576554924249649,
"learning_rate": 0.00032026817641211524,
"loss": 0.3612,
"step": 1020
},
{
"epoch": 0.8220551378446115,
"grad_norm": 0.16731838881969452,
"learning_rate": 0.0003187544790838805,
"loss": 0.3742,
"step": 1025
},
{
"epoch": 0.8260651629072682,
"grad_norm": 0.17792581021785736,
"learning_rate": 0.0003172380499912768,
"loss": 0.3835,
"step": 1030
},
{
"epoch": 0.8300751879699249,
"grad_norm": 0.15929646790027618,
"learning_rate": 0.0003157189493853164,
"loss": 0.3547,
"step": 1035
},
{
"epoch": 0.8340852130325814,
"grad_norm": 0.17030423879623413,
"learning_rate": 0.00031419723762315656,
"loss": 0.3833,
"step": 1040
},
{
"epoch": 0.8380952380952381,
"grad_norm": 0.16632360219955444,
"learning_rate": 0.0003126729751657015,
"loss": 0.3851,
"step": 1045
},
{
"epoch": 0.8421052631578947,
"grad_norm": 0.16034843027591705,
"learning_rate": 0.0003111462225752,
"loss": 0.3457,
"step": 1050
},
{
"epoch": 0.8461152882205514,
"grad_norm": 0.1584872305393219,
"learning_rate": 0.0003096170405128393,
"loss": 0.3627,
"step": 1055
},
{
"epoch": 0.8501253132832081,
"grad_norm": 0.15513049066066742,
"learning_rate": 0.0003080854897363348,
"loss": 0.3571,
"step": 1060
},
{
"epoch": 0.8541353383458646,
"grad_norm": 0.18553143739700317,
"learning_rate": 0.000306551631097516,
"loss": 0.3844,
"step": 1065
},
{
"epoch": 0.8581453634085213,
"grad_norm": 0.17072825133800507,
"learning_rate": 0.00030501552553990885,
"loss": 0.3968,
"step": 1070
},
{
"epoch": 0.8621553884711779,
"grad_norm": 0.17334377765655518,
"learning_rate": 0.00030347723409631413,
"loss": 0.3759,
"step": 1075
},
{
"epoch": 0.8661654135338346,
"grad_norm": 0.16847221553325653,
"learning_rate": 0.00030193681788638274,
"loss": 0.3437,
"step": 1080
},
{
"epoch": 0.8701754385964913,
"grad_norm": 0.17665381729602814,
"learning_rate": 0.000300394338114187,
"loss": 0.3691,
"step": 1085
},
{
"epoch": 0.8741854636591478,
"grad_norm": 0.17721857130527496,
"learning_rate": 0.00029884985606578907,
"loss": 0.3776,
"step": 1090
},
{
"epoch": 0.8781954887218045,
"grad_norm": 0.14506922662258148,
"learning_rate": 0.0002973034331068061,
"loss": 0.3329,
"step": 1095
},
{
"epoch": 0.8822055137844611,
"grad_norm": 0.14164331555366516,
"learning_rate": 0.0002957551306799715,
"loss": 0.3296,
"step": 1100
},
{
"epoch": 0.8862155388471178,
"grad_norm": 0.17468836903572083,
"learning_rate": 0.0002942050103026942,
"loss": 0.3814,
"step": 1105
},
{
"epoch": 0.8902255639097745,
"grad_norm": 0.1707754135131836,
"learning_rate": 0.0002926531335646141,
"loss": 0.3288,
"step": 1110
},
{
"epoch": 0.894235588972431,
"grad_norm": 0.1548461616039276,
"learning_rate": 0.0002910995621251554,
"loss": 0.3837,
"step": 1115
},
{
"epoch": 0.8982456140350877,
"grad_norm": 0.18606321513652802,
"learning_rate": 0.000289544357711076,
"loss": 0.411,
"step": 1120
},
{
"epoch": 0.9022556390977443,
"grad_norm": 0.1885378062725067,
"learning_rate": 0.00028798758211401586,
"loss": 0.3965,
"step": 1125
},
{
"epoch": 0.906265664160401,
"grad_norm": 0.15879781544208527,
"learning_rate": 0.00028642929718804126,
"loss": 0.3217,
"step": 1130
},
{
"epoch": 0.9102756892230577,
"grad_norm": 0.16221250593662262,
"learning_rate": 0.00028486956484718735,
"loss": 0.3431,
"step": 1135
},
{
"epoch": 0.9142857142857143,
"grad_norm": 0.15938705205917358,
"learning_rate": 0.0002833084470629983,
"loss": 0.3645,
"step": 1140
},
{
"epoch": 0.9182957393483709,
"grad_norm": 0.18234948813915253,
"learning_rate": 0.0002817460058620647,
"loss": 0.3696,
"step": 1145
},
{
"epoch": 0.9223057644110275,
"grad_norm": 0.17734579741954803,
"learning_rate": 0.0002801823033235598,
"loss": 0.3939,
"step": 1150
},
{
"epoch": 0.9263157894736842,
"grad_norm": 0.17581012845039368,
"learning_rate": 0.00027861740157677204,
"loss": 0.3857,
"step": 1155
},
{
"epoch": 0.9303258145363409,
"grad_norm": 0.16594088077545166,
"learning_rate": 0.00027705136279863754,
"loss": 0.3784,
"step": 1160
},
{
"epoch": 0.9343358395989975,
"grad_norm": 0.17413030564785004,
"learning_rate": 0.0002754842492112685,
"loss": 0.3791,
"step": 1165
},
{
"epoch": 0.9383458646616541,
"grad_norm": 0.16821810603141785,
"learning_rate": 0.0002739161230794822,
"loss": 0.3896,
"step": 1170
},
{
"epoch": 0.9423558897243107,
"grad_norm": 0.18337774276733398,
"learning_rate": 0.000272347046708326,
"loss": 0.3745,
"step": 1175
},
{
"epoch": 0.9463659147869674,
"grad_norm": 0.15810732543468475,
"learning_rate": 0.00027077708244060283,
"loss": 0.3576,
"step": 1180
},
{
"epoch": 0.9503759398496241,
"grad_norm": 0.16330723464488983,
"learning_rate": 0.00026920629265439326,
"loss": 0.3641,
"step": 1185
},
{
"epoch": 0.9543859649122807,
"grad_norm": 0.168942391872406,
"learning_rate": 0.0002676347397605777,
"loss": 0.36,
"step": 1190
},
{
"epoch": 0.9583959899749374,
"grad_norm": 0.1779577136039734,
"learning_rate": 0.0002660624862003566,
"loss": 0.3644,
"step": 1195
},
{
"epoch": 0.9624060150375939,
"grad_norm": 0.175537571310997,
"learning_rate": 0.00026448959444276896,
"loss": 0.3766,
"step": 1200
},
{
"epoch": 0.9664160401002506,
"grad_norm": 0.17511993646621704,
"learning_rate": 0.0002629161269822113,
"loss": 0.3636,
"step": 1205
},
{
"epoch": 0.9704260651629073,
"grad_norm": 0.17956364154815674,
"learning_rate": 0.00026134214633595347,
"loss": 0.3916,
"step": 1210
},
{
"epoch": 0.9744360902255639,
"grad_norm": 0.15933747589588165,
"learning_rate": 0.000259767715041656,
"loss": 0.3316,
"step": 1215
},
{
"epoch": 0.9784461152882206,
"grad_norm": 0.18015708029270172,
"learning_rate": 0.000258192895654884,
"loss": 0.3791,
"step": 1220
},
{
"epoch": 0.9824561403508771,
"grad_norm": 0.15591025352478027,
"learning_rate": 0.00025661775074662275,
"loss": 0.3222,
"step": 1225
},
{
"epoch": 0.9864661654135338,
"grad_norm": 0.1800549179315567,
"learning_rate": 0.0002550423429007909,
"loss": 0.3282,
"step": 1230
},
{
"epoch": 0.9904761904761905,
"grad_norm": 0.1630546748638153,
"learning_rate": 0.0002534667347117544,
"loss": 0.3659,
"step": 1235
},
{
"epoch": 0.9944862155388471,
"grad_norm": 0.169642373919487,
"learning_rate": 0.0002518909887818393,
"loss": 0.3409,
"step": 1240
},
{
"epoch": 0.9984962406015038,
"grad_norm": 0.1650785356760025,
"learning_rate": 0.00025031516771884416,
"loss": 0.3632,
"step": 1245
},
{
"epoch": 0.9992982456140351,
"eval_loss": 0.2687419354915619,
"eval_runtime": 8.3871,
"eval_samples_per_second": 6.558,
"eval_steps_per_second": 0.835,
"step": 1246
},
{
"epoch": 1.0025062656641603,
"grad_norm": 0.1833638846874237,
"learning_rate": 0.000248739334133553,
"loss": 0.3377,
"step": 1250
},
{
"epoch": 1.006516290726817,
"grad_norm": 0.16308297216892242,
"learning_rate": 0.00024716355063724707,
"loss": 0.2999,
"step": 1255
},
{
"epoch": 1.0105263157894737,
"grad_norm": 0.16679681837558746,
"learning_rate": 0.00024558787983921783,
"loss": 0.2986,
"step": 1260
},
{
"epoch": 1.0145363408521304,
"grad_norm": 0.17338265478610992,
"learning_rate": 0.0002440123843442788,
"loss": 0.2845,
"step": 1265
},
{
"epoch": 1.0185463659147869,
"grad_norm": 0.16227693855762482,
"learning_rate": 0.00024243712675027827,
"loss": 0.3051,
"step": 1270
},
{
"epoch": 1.0225563909774436,
"grad_norm": 0.15637050569057465,
"learning_rate": 0.00024086216964561236,
"loss": 0.2896,
"step": 1275
},
{
"epoch": 1.0265664160401002,
"grad_norm": 0.1684030145406723,
"learning_rate": 0.00023928757560673808,
"loss": 0.272,
"step": 1280
},
{
"epoch": 1.030576441102757,
"grad_norm": 0.1560041755437851,
"learning_rate": 0.00023771340719568688,
"loss": 0.2795,
"step": 1285
},
{
"epoch": 1.0345864661654136,
"grad_norm": 0.15775898098945618,
"learning_rate": 0.00023613972695757919,
"loss": 0.2721,
"step": 1290
},
{
"epoch": 1.03859649122807,
"grad_norm": 0.17277134954929352,
"learning_rate": 0.00023456659741813944,
"loss": 0.2726,
"step": 1295
},
{
"epoch": 1.0426065162907268,
"grad_norm": 0.15520408749580383,
"learning_rate": 0.00023299408108121134,
"loss": 0.2681,
"step": 1300
},
{
"epoch": 1.0466165413533834,
"grad_norm": 0.16784396767616272,
"learning_rate": 0.00023142224042627495,
"loss": 0.2817,
"step": 1305
},
{
"epoch": 1.0506265664160401,
"grad_norm": 0.16616085171699524,
"learning_rate": 0.00022985113790596393,
"loss": 0.2863,
"step": 1310
},
{
"epoch": 1.0546365914786968,
"grad_norm": 0.15373535454273224,
"learning_rate": 0.00022828083594358416,
"loss": 0.2837,
"step": 1315
},
{
"epoch": 1.0586466165413533,
"grad_norm": 0.15861855447292328,
"learning_rate": 0.00022671139693063386,
"loss": 0.2878,
"step": 1320
},
{
"epoch": 1.06265664160401,
"grad_norm": 0.15624883770942688,
"learning_rate": 0.00022514288322432414,
"loss": 0.2952,
"step": 1325
},
{
"epoch": 1.0666666666666667,
"grad_norm": 0.17166998982429504,
"learning_rate": 0.0002235753571451018,
"loss": 0.2957,
"step": 1330
},
{
"epoch": 1.0706766917293233,
"grad_norm": 0.15910623967647552,
"learning_rate": 0.00022200888097417305,
"loss": 0.2954,
"step": 1335
},
{
"epoch": 1.07468671679198,
"grad_norm": 0.154972106218338,
"learning_rate": 0.00022044351695102907,
"loss": 0.2624,
"step": 1340
},
{
"epoch": 1.0786967418546367,
"grad_norm": 0.16736608743667603,
"learning_rate": 0.00021887932727097278,
"loss": 0.2677,
"step": 1345
},
{
"epoch": 1.0827067669172932,
"grad_norm": 0.17113475501537323,
"learning_rate": 0.00021731637408264822,
"loss": 0.2613,
"step": 1350
},
{
"epoch": 1.0867167919799499,
"grad_norm": 0.17451037466526031,
"learning_rate": 0.00021575471948557062,
"loss": 0.291,
"step": 1355
},
{
"epoch": 1.0907268170426065,
"grad_norm": 0.15798042714595795,
"learning_rate": 0.00021419442552765948,
"loss": 0.2628,
"step": 1360
},
{
"epoch": 1.0947368421052632,
"grad_norm": 0.17537885904312134,
"learning_rate": 0.00021263555420277335,
"loss": 0.2764,
"step": 1365
},
{
"epoch": 1.09874686716792,
"grad_norm": 0.14623847603797913,
"learning_rate": 0.00021107816744824617,
"loss": 0.29,
"step": 1370
},
{
"epoch": 1.1027568922305764,
"grad_norm": 0.17818230390548706,
"learning_rate": 0.00020952232714242685,
"loss": 0.2866,
"step": 1375
},
{
"epoch": 1.106766917293233,
"grad_norm": 0.16421560943126678,
"learning_rate": 0.00020796809510222048,
"loss": 0.2718,
"step": 1380
},
{
"epoch": 1.1107769423558898,
"grad_norm": 0.16824981570243835,
"learning_rate": 0.00020641553308063247,
"loss": 0.2976,
"step": 1385
},
{
"epoch": 1.1147869674185464,
"grad_norm": 0.15799076855182648,
"learning_rate": 0.00020486470276431446,
"loss": 0.2934,
"step": 1390
},
{
"epoch": 1.1187969924812031,
"grad_norm": 0.18710538744926453,
"learning_rate": 0.000203315665771114,
"loss": 0.2792,
"step": 1395
},
{
"epoch": 1.1228070175438596,
"grad_norm": 0.18664434552192688,
"learning_rate": 0.00020176848364762578,
"loss": 0.3125,
"step": 1400
},
{
"epoch": 1.1268170426065163,
"grad_norm": 0.14843712747097015,
"learning_rate": 0.00020022321786674664,
"loss": 0.2636,
"step": 1405
},
{
"epoch": 1.130827067669173,
"grad_norm": 0.18187515437602997,
"learning_rate": 0.000198679929825233,
"loss": 0.3186,
"step": 1410
},
{
"epoch": 1.1348370927318296,
"grad_norm": 0.15874765813350677,
"learning_rate": 0.0001971386808412612,
"loss": 0.2578,
"step": 1415
},
{
"epoch": 1.1388471177944863,
"grad_norm": 0.17650403082370758,
"learning_rate": 0.00019559953215199168,
"loss": 0.3012,
"step": 1420
},
{
"epoch": 1.1428571428571428,
"grad_norm": 0.17228610813617706,
"learning_rate": 0.0001940625449111354,
"loss": 0.2666,
"step": 1425
},
{
"epoch": 1.1468671679197995,
"grad_norm": 0.15302354097366333,
"learning_rate": 0.00019252778018652447,
"loss": 0.272,
"step": 1430
},
{
"epoch": 1.1508771929824562,
"grad_norm": 0.17536306381225586,
"learning_rate": 0.0001909952989576855,
"loss": 0.3183,
"step": 1435
},
{
"epoch": 1.1548872180451129,
"grad_norm": 0.15911608934402466,
"learning_rate": 0.00018946516211341695,
"loss": 0.3022,
"step": 1440
},
{
"epoch": 1.1588972431077695,
"grad_norm": 0.15123997628688812,
"learning_rate": 0.00018793743044936972,
"loss": 0.289,
"step": 1445
},
{
"epoch": 1.162907268170426,
"grad_norm": 0.16768620908260345,
"learning_rate": 0.00018641216466563183,
"loss": 0.2694,
"step": 1450
},
{
"epoch": 1.1669172932330827,
"grad_norm": 0.16061028838157654,
"learning_rate": 0.00018488942536431618,
"loss": 0.2824,
"step": 1455
},
{
"epoch": 1.1709273182957394,
"grad_norm": 0.15412503480911255,
"learning_rate": 0.00018336927304715356,
"loss": 0.3025,
"step": 1460
},
{
"epoch": 1.174937343358396,
"grad_norm": 0.17830166220664978,
"learning_rate": 0.0001818517681130879,
"loss": 0.3046,
"step": 1465
},
{
"epoch": 1.1789473684210527,
"grad_norm": 0.17978624999523163,
"learning_rate": 0.00018033697085587696,
"loss": 0.2966,
"step": 1470
},
{
"epoch": 1.1829573934837092,
"grad_norm": 0.1743180751800537,
"learning_rate": 0.00017882494146169678,
"loss": 0.3072,
"step": 1475
},
{
"epoch": 1.186967418546366,
"grad_norm": 0.16470114886760712,
"learning_rate": 0.00017731574000674996,
"loss": 0.2797,
"step": 1480
},
{
"epoch": 1.1909774436090226,
"grad_norm": 0.18040384352207184,
"learning_rate": 0.0001758094264548792,
"loss": 0.3093,
"step": 1485
},
{
"epoch": 1.1949874686716793,
"grad_norm": 0.17426797747612,
"learning_rate": 0.00017430606065518435,
"loss": 0.318,
"step": 1490
},
{
"epoch": 1.198997493734336,
"grad_norm": 0.1632205694913864,
"learning_rate": 0.00017280570233964485,
"loss": 0.2691,
"step": 1495
},
{
"epoch": 1.2030075187969924,
"grad_norm": 0.1813742071390152,
"learning_rate": 0.00017130841112074602,
"loss": 0.2769,
"step": 1500
},
{
"epoch": 1.207017543859649,
"grad_norm": 0.17400313913822174,
"learning_rate": 0.00016981424648911112,
"loss": 0.2723,
"step": 1505
},
{
"epoch": 1.2110275689223058,
"grad_norm": 0.15530717372894287,
"learning_rate": 0.0001683232678111371,
"loss": 0.2971,
"step": 1510
},
{
"epoch": 1.2150375939849625,
"grad_norm": 0.17954862117767334,
"learning_rate": 0.00016683553432663616,
"loss": 0.2866,
"step": 1515
},
{
"epoch": 1.2190476190476192,
"grad_norm": 0.16163484752178192,
"learning_rate": 0.0001653511051464819,
"loss": 0.2557,
"step": 1520
},
{
"epoch": 1.2230576441102756,
"grad_norm": 0.16374574601650238,
"learning_rate": 0.00016387003925026077,
"loss": 0.2956,
"step": 1525
},
{
"epoch": 1.2270676691729323,
"grad_norm": 0.17792537808418274,
"learning_rate": 0.0001623923954839287,
"loss": 0.2777,
"step": 1530
},
{
"epoch": 1.231077694235589,
"grad_norm": 0.18563567101955414,
"learning_rate": 0.00016091823255747292,
"loss": 0.301,
"step": 1535
},
{
"epoch": 1.2350877192982457,
"grad_norm": 0.15728415548801422,
"learning_rate": 0.00015944760904257942,
"loss": 0.2784,
"step": 1540
},
{
"epoch": 1.2390977443609024,
"grad_norm": 0.15651676058769226,
"learning_rate": 0.00015798058337030551,
"loss": 0.2876,
"step": 1545
},
{
"epoch": 1.2431077694235588,
"grad_norm": 0.1799081265926361,
"learning_rate": 0.00015651721382875874,
"loss": 0.2919,
"step": 1550
},
{
"epoch": 1.2471177944862155,
"grad_norm": 0.18436437845230103,
"learning_rate": 0.00015505755856078048,
"loss": 0.2934,
"step": 1555
},
{
"epoch": 1.2511278195488722,
"grad_norm": 0.16269451379776,
"learning_rate": 0.00015360167556163583,
"loss": 0.2719,
"step": 1560
},
{
"epoch": 1.2551378446115289,
"grad_norm": 0.15247991681098938,
"learning_rate": 0.0001521496226767098,
"loss": 0.2678,
"step": 1565
},
{
"epoch": 1.2591478696741856,
"grad_norm": 0.1753372848033905,
"learning_rate": 0.0001507014575992085,
"loss": 0.3065,
"step": 1570
},
{
"epoch": 1.263157894736842,
"grad_norm": 0.16919946670532227,
"learning_rate": 0.0001492572378678669,
"loss": 0.2659,
"step": 1575
},
{
"epoch": 1.2671679197994987,
"grad_norm": 0.17600032687187195,
"learning_rate": 0.00014781702086466314,
"loss": 0.2987,
"step": 1580
},
{
"epoch": 1.2711779448621554,
"grad_norm": 0.15909601747989655,
"learning_rate": 0.000146380863812538,
"loss": 0.2811,
"step": 1585
},
{
"epoch": 1.275187969924812,
"grad_norm": 0.165226012468338,
"learning_rate": 0.00014494882377312168,
"loss": 0.2682,
"step": 1590
},
{
"epoch": 1.2791979949874688,
"grad_norm": 0.17594917118549347,
"learning_rate": 0.00014352095764446675,
"loss": 0.2602,
"step": 1595
},
{
"epoch": 1.2832080200501252,
"grad_norm": 0.154309943318367,
"learning_rate": 0.0001420973221587869,
"loss": 0.2579,
"step": 1600
},
{
"epoch": 1.287218045112782,
"grad_norm": 0.16517868638038635,
"learning_rate": 0.00014067797388020353,
"loss": 0.2929,
"step": 1605
},
{
"epoch": 1.2912280701754386,
"grad_norm": 0.1717553734779358,
"learning_rate": 0.00013926296920249796,
"loss": 0.2849,
"step": 1610
},
{
"epoch": 1.2952380952380953,
"grad_norm": 0.16944153606891632,
"learning_rate": 0.0001378523643468708,
"loss": 0.2527,
"step": 1615
},
{
"epoch": 1.299248120300752,
"grad_norm": 0.16444195806980133,
"learning_rate": 0.00013644621535970804,
"loss": 0.2665,
"step": 1620
},
{
"epoch": 1.3032581453634084,
"grad_norm": 0.17226529121398926,
"learning_rate": 0.0001350445781103547,
"loss": 0.2758,
"step": 1625
},
{
"epoch": 1.3072681704260651,
"grad_norm": 0.18174055218696594,
"learning_rate": 0.0001336475082888943,
"loss": 0.2831,
"step": 1630
},
{
"epoch": 1.3112781954887218,
"grad_norm": 0.16900789737701416,
"learning_rate": 0.00013225506140393696,
"loss": 0.2882,
"step": 1635
},
{
"epoch": 1.3152882205513785,
"grad_norm": 0.16686789691448212,
"learning_rate": 0.00013086729278041304,
"loss": 0.2556,
"step": 1640
},
{
"epoch": 1.3192982456140352,
"grad_norm": 0.16731151938438416,
"learning_rate": 0.00012948425755737592,
"loss": 0.2665,
"step": 1645
},
{
"epoch": 1.3233082706766917,
"grad_norm": 0.17180107533931732,
"learning_rate": 0.00012810601068581056,
"loss": 0.3001,
"step": 1650
},
{
"epoch": 1.3273182957393483,
"grad_norm": 0.1701911836862564,
"learning_rate": 0.0001267326069264501,
"loss": 0.2977,
"step": 1655
},
{
"epoch": 1.331328320802005,
"grad_norm": 0.19208161532878876,
"learning_rate": 0.0001253641008476007,
"loss": 0.2752,
"step": 1660
},
{
"epoch": 1.3353383458646617,
"grad_norm": 0.15200825035572052,
"learning_rate": 0.00012400054682297298,
"loss": 0.2875,
"step": 1665
},
{
"epoch": 1.3393483709273184,
"grad_norm": 0.17044006288051605,
"learning_rate": 0.00012264199902952154,
"loss": 0.2501,
"step": 1670
},
{
"epoch": 1.3433583959899749,
"grad_norm": 0.16576853394508362,
"learning_rate": 0.00012128851144529257,
"loss": 0.2758,
"step": 1675
},
{
"epoch": 1.3473684210526315,
"grad_norm": 0.17262569069862366,
"learning_rate": 0.00011994013784727947,
"loss": 0.2661,
"step": 1680
},
{
"epoch": 1.3513784461152882,
"grad_norm": 0.161615252494812,
"learning_rate": 0.00011859693180928574,
"loss": 0.2689,
"step": 1685
},
{
"epoch": 1.355388471177945,
"grad_norm": 0.16010525822639465,
"learning_rate": 0.00011725894669979639,
"loss": 0.2623,
"step": 1690
},
{
"epoch": 1.3593984962406016,
"grad_norm": 0.1819092184305191,
"learning_rate": 0.00011592623567985783,
"loss": 0.2667,
"step": 1695
},
{
"epoch": 1.363408521303258,
"grad_norm": 0.16817572712898254,
"learning_rate": 0.00011459885170096537,
"loss": 0.2756,
"step": 1700
},
{
"epoch": 1.3674185463659148,
"grad_norm": 0.17827191948890686,
"learning_rate": 0.00011327684750295958,
"loss": 0.3166,
"step": 1705
},
{
"epoch": 1.3714285714285714,
"grad_norm": 0.15747743844985962,
"learning_rate": 0.00011196027561193045,
"loss": 0.2952,
"step": 1710
},
{
"epoch": 1.3754385964912281,
"grad_norm": 0.17462944984436035,
"learning_rate": 0.00011064918833813073,
"loss": 0.2754,
"step": 1715
},
{
"epoch": 1.3794486215538848,
"grad_norm": 0.16550135612487793,
"learning_rate": 0.00010934363777389752,
"loss": 0.2658,
"step": 1720
},
{
"epoch": 1.3834586466165413,
"grad_norm": 0.19006794691085815,
"learning_rate": 0.00010804367579158256,
"loss": 0.2679,
"step": 1725
},
{
"epoch": 1.387468671679198,
"grad_norm": 0.2023632973432541,
"learning_rate": 0.00010674935404149075,
"loss": 0.2739,
"step": 1730
},
{
"epoch": 1.3914786967418546,
"grad_norm": 0.17994622886180878,
"learning_rate": 0.00010546072394982872,
"loss": 0.2585,
"step": 1735
},
{
"epoch": 1.3954887218045113,
"grad_norm": 0.1834658533334732,
"learning_rate": 0.00010417783671666114,
"loss": 0.2567,
"step": 1740
},
{
"epoch": 1.399498746867168,
"grad_norm": 0.16949954628944397,
"learning_rate": 0.00010290074331387617,
"loss": 0.2462,
"step": 1745
},
{
"epoch": 1.4035087719298245,
"grad_norm": 0.17096513509750366,
"learning_rate": 0.00010162949448316089,
"loss": 0.2818,
"step": 1750
},
{
"epoch": 1.4075187969924812,
"grad_norm": 0.16146983206272125,
"learning_rate": 0.00010036414073398479,
"loss": 0.2691,
"step": 1755
},
{
"epoch": 1.4115288220551379,
"grad_norm": 0.15868982672691345,
"learning_rate": 9.910473234159286e-05,
"loss": 0.2857,
"step": 1760
},
{
"epoch": 1.4155388471177945,
"grad_norm": 0.1619461625814438,
"learning_rate": 9.785131934500818e-05,
"loss": 0.2807,
"step": 1765
},
{
"epoch": 1.4195488721804512,
"grad_norm": 0.1557190865278244,
"learning_rate": 9.660395154504401e-05,
"loss": 0.2548,
"step": 1770
},
{
"epoch": 1.4235588972431077,
"grad_norm": 0.17061957716941833,
"learning_rate": 9.536267850232472e-05,
"loss": 0.2801,
"step": 1775
},
{
"epoch": 1.4275689223057644,
"grad_norm": 0.17641028761863708,
"learning_rate": 9.412754953531663e-05,
"loss": 0.2924,
"step": 1780
},
{
"epoch": 1.431578947368421,
"grad_norm": 0.15920333564281464,
"learning_rate": 9.289861371836885e-05,
"loss": 0.2563,
"step": 1785
},
{
"epoch": 1.4355889724310777,
"grad_norm": 0.15198518335819244,
"learning_rate": 9.167591987976312e-05,
"loss": 0.2357,
"step": 1790
},
{
"epoch": 1.4395989974937344,
"grad_norm": 0.1758808046579361,
"learning_rate": 9.045951659977397e-05,
"loss": 0.2535,
"step": 1795
},
{
"epoch": 1.443609022556391,
"grad_norm": 0.18127872049808502,
"learning_rate": 8.924945220873823e-05,
"loss": 0.2888,
"step": 1800
},
{
"epoch": 1.4476190476190476,
"grad_norm": 0.1601109355688095,
"learning_rate": 8.804577478513492e-05,
"loss": 0.2528,
"step": 1805
},
{
"epoch": 1.4516290726817043,
"grad_norm": 0.16406309604644775,
"learning_rate": 8.684853215367522e-05,
"loss": 0.2766,
"step": 1810
},
{
"epoch": 1.455639097744361,
"grad_norm": 0.158805251121521,
"learning_rate": 8.565777188340207e-05,
"loss": 0.2747,
"step": 1815
},
{
"epoch": 1.4596491228070176,
"grad_norm": 0.17480885982513428,
"learning_rate": 8.44735412857999e-05,
"loss": 0.3079,
"step": 1820
},
{
"epoch": 1.463659147869674,
"grad_norm": 0.1617252379655838,
"learning_rate": 8.329588741291535e-05,
"loss": 0.2701,
"step": 1825
},
{
"epoch": 1.4676691729323308,
"grad_norm": 0.1652277261018753,
"learning_rate": 8.212485705548756e-05,
"loss": 0.25,
"step": 1830
},
{
"epoch": 1.4716791979949875,
"grad_norm": 0.16213206946849823,
"learning_rate": 8.096049674108877e-05,
"loss": 0.2683,
"step": 1835
},
{
"epoch": 1.4756892230576442,
"grad_norm": 0.167341947555542,
"learning_rate": 7.980285273227633e-05,
"loss": 0.2724,
"step": 1840
},
{
"epoch": 1.4796992481203008,
"grad_norm": 0.1590610146522522,
"learning_rate": 7.865197102475388e-05,
"loss": 0.2518,
"step": 1845
},
{
"epoch": 1.4837092731829573,
"grad_norm": 0.16797657310962677,
"learning_rate": 7.75078973455445e-05,
"loss": 0.2651,
"step": 1850
},
{
"epoch": 1.487719298245614,
"grad_norm": 0.17504721879959106,
"learning_rate": 7.637067715117327e-05,
"loss": 0.3017,
"step": 1855
},
{
"epoch": 1.4917293233082707,
"grad_norm": 0.16186365485191345,
"learning_rate": 7.52403556258617e-05,
"loss": 0.2344,
"step": 1860
},
{
"epoch": 1.4957393483709274,
"grad_norm": 0.17528776824474335,
"learning_rate": 7.41169776797322e-05,
"loss": 0.264,
"step": 1865
},
{
"epoch": 1.4989473684210526,
"eval_loss": 0.2417936623096466,
"eval_runtime": 8.3922,
"eval_samples_per_second": 6.554,
"eval_steps_per_second": 0.834,
"step": 1869
},
{
"epoch": 1.499749373433584,
"grad_norm": 0.1840500384569168,
"learning_rate": 7.300058794702352e-05,
"loss": 0.27,
"step": 1870
},
{
"epoch": 1.5037593984962405,
"grad_norm": 0.1910168081521988,
"learning_rate": 7.189123078431784e-05,
"loss": 0.2823,
"step": 1875
},
{
"epoch": 1.5077694235588972,
"grad_norm": 0.16170190274715424,
"learning_rate": 7.078895026877804e-05,
"loss": 0.253,
"step": 1880
},
{
"epoch": 1.511779448621554,
"grad_norm": 0.17806147038936615,
"learning_rate": 6.969379019639635e-05,
"loss": 0.2836,
"step": 1885
},
{
"epoch": 1.5157894736842106,
"grad_norm": 0.15123490989208221,
"learning_rate": 6.860579408025436e-05,
"loss": 0.2495,
"step": 1890
},
{
"epoch": 1.5197994987468673,
"grad_norm": 0.16695557534694672,
"learning_rate": 6.752500514879437e-05,
"loss": 0.2773,
"step": 1895
},
{
"epoch": 1.5238095238095237,
"grad_norm": 0.16708271205425262,
"learning_rate": 6.645146634410151e-05,
"loss": 0.252,
"step": 1900
},
{
"epoch": 1.5278195488721804,
"grad_norm": 0.1510598361492157,
"learning_rate": 6.538522032019759e-05,
"loss": 0.2373,
"step": 1905
},
{
"epoch": 1.531829573934837,
"grad_norm": 0.1686723679304123,
"learning_rate": 6.432630944134654e-05,
"loss": 0.2743,
"step": 1910
},
{
"epoch": 1.5358395989974938,
"grad_norm": 0.16884025931358337,
"learning_rate": 6.327477578037106e-05,
"loss": 0.2673,
"step": 1915
},
{
"epoch": 1.5398496240601505,
"grad_norm": 0.16477200388908386,
"learning_rate": 6.223066111698111e-05,
"loss": 0.2713,
"step": 1920
},
{
"epoch": 1.543859649122807,
"grad_norm": 0.18342123925685883,
"learning_rate": 6.119400693611357e-05,
"loss": 0.2791,
"step": 1925
},
{
"epoch": 1.5478696741854636,
"grad_norm": 0.16829492151737213,
"learning_rate": 6.0164854426284447e-05,
"loss": 0.2702,
"step": 1930
},
{
"epoch": 1.5518796992481203,
"grad_norm": 0.16608929634094238,
"learning_rate": 5.914324447795186e-05,
"loss": 0.2741,
"step": 1935
},
{
"epoch": 1.555889724310777,
"grad_norm": 0.16408754885196686,
"learning_rate": 5.8129217681891886e-05,
"loss": 0.2661,
"step": 1940
},
{
"epoch": 1.5598997493734337,
"grad_norm": 0.1709819734096527,
"learning_rate": 5.7122814327585316e-05,
"loss": 0.2557,
"step": 1945
},
{
"epoch": 1.5639097744360901,
"grad_norm": 0.187583327293396,
"learning_rate": 5.612407440161721e-05,
"loss": 0.2623,
"step": 1950
},
{
"epoch": 1.5679197994987468,
"grad_norm": 0.16979658603668213,
"learning_rate": 5.513303758608804e-05,
"loss": 0.2615,
"step": 1955
},
{
"epoch": 1.5719298245614035,
"grad_norm": 0.13367925584316254,
"learning_rate": 5.414974325703686e-05,
"loss": 0.2276,
"step": 1960
},
{
"epoch": 1.5759398496240602,
"grad_norm": 0.17686474323272705,
"learning_rate": 5.317423048287717e-05,
"loss": 0.2444,
"step": 1965
},
{
"epoch": 1.5799498746867169,
"grad_norm": 0.17005467414855957,
"learning_rate": 5.220653802284439e-05,
"loss": 0.2581,
"step": 1970
},
{
"epoch": 1.5839598997493733,
"grad_norm": 0.1664769947528839,
"learning_rate": 5.124670432545578e-05,
"loss": 0.2628,
"step": 1975
},
{
"epoch": 1.58796992481203,
"grad_norm": 0.17945070564746857,
"learning_rate": 5.0294767526983006e-05,
"loss": 0.276,
"step": 1980
},
{
"epoch": 1.5919799498746867,
"grad_norm": 0.1758703887462616,
"learning_rate": 4.9350765449936915e-05,
"loss": 0.2744,
"step": 1985
},
{
"epoch": 1.5959899749373434,
"grad_norm": 0.1651277393102646,
"learning_rate": 4.84147356015647e-05,
"loss": 0.2872,
"step": 1990
},
{
"epoch": 1.6,
"grad_norm": 0.16701501607894897,
"learning_rate": 4.748671517235948e-05,
"loss": 0.2676,
"step": 1995
},
{
"epoch": 1.6040100250626566,
"grad_norm": 0.16929472982883453,
"learning_rate": 4.656674103458291e-05,
"loss": 0.2498,
"step": 2000
},
{
"epoch": 1.6080200501253132,
"grad_norm": 0.15207117795944214,
"learning_rate": 4.5654849740800145e-05,
"loss": 0.26,
"step": 2005
},
{
"epoch": 1.61203007518797,
"grad_norm": 0.15770703554153442,
"learning_rate": 4.47510775224273e-05,
"loss": 0.2644,
"step": 2010
},
{
"epoch": 1.6160401002506266,
"grad_norm": 0.16916294395923615,
"learning_rate": 4.38554602882921e-05,
"loss": 0.3026,
"step": 2015
},
{
"epoch": 1.6200501253132833,
"grad_norm": 0.16087712347507477,
"learning_rate": 4.2968033623206974e-05,
"loss": 0.2632,
"step": 2020
},
{
"epoch": 1.6240601503759398,
"grad_norm": 0.18064051866531372,
"learning_rate": 4.2088832786555456e-05,
"loss": 0.2654,
"step": 2025
},
{
"epoch": 1.6280701754385964,
"grad_norm": 0.17459240555763245,
"learning_rate": 4.121789271089113e-05,
"loss": 0.276,
"step": 2030
},
{
"epoch": 1.6320802005012531,
"grad_norm": 0.15657363831996918,
"learning_rate": 4.0355248000549434e-05,
"loss": 0.241,
"step": 2035
},
{
"epoch": 1.6360902255639098,
"grad_norm": 0.1620418131351471,
"learning_rate": 3.950093293027318e-05,
"loss": 0.2821,
"step": 2040
},
{
"epoch": 1.6401002506265665,
"grad_norm": 0.18404340744018555,
"learning_rate": 3.865498144385049e-05,
"loss": 0.2935,
"step": 2045
},
{
"epoch": 1.644110275689223,
"grad_norm": 0.16540312767028809,
"learning_rate": 3.7817427152766094e-05,
"loss": 0.2583,
"step": 2050
},
{
"epoch": 1.6481203007518797,
"grad_norm": 0.16649490594863892,
"learning_rate": 3.6988303334866054e-05,
"loss": 0.2774,
"step": 2055
},
{
"epoch": 1.6521303258145363,
"grad_norm": 0.1626511514186859,
"learning_rate": 3.6167642933035486e-05,
"loss": 0.2467,
"step": 2060
},
{
"epoch": 1.656140350877193,
"grad_norm": 0.16354456543922424,
"learning_rate": 3.535547855388963e-05,
"loss": 0.2544,
"step": 2065
},
{
"epoch": 1.6601503759398497,
"grad_norm": 0.16961799561977386,
"learning_rate": 3.455184246647822e-05,
"loss": 0.2389,
"step": 2070
},
{
"epoch": 1.6641604010025062,
"grad_norm": 0.17093317210674286,
"learning_rate": 3.375676660100366e-05,
"loss": 0.2767,
"step": 2075
},
{
"epoch": 1.6681704260651629,
"grad_norm": 0.1904112696647644,
"learning_rate": 3.297028254755221e-05,
"loss": 0.29,
"step": 2080
},
{
"epoch": 1.6721804511278195,
"grad_norm": 0.18005013465881348,
"learning_rate": 3.219242155483868e-05,
"loss": 0.249,
"step": 2085
},
{
"epoch": 1.6761904761904762,
"grad_norm": 0.1755589246749878,
"learning_rate": 3.142321452896504e-05,
"loss": 0.2956,
"step": 2090
},
{
"epoch": 1.680200501253133,
"grad_norm": 0.15832673013210297,
"learning_rate": 3.0662692032192516e-05,
"loss": 0.2561,
"step": 2095
},
{
"epoch": 1.6842105263157894,
"grad_norm": 0.15485481917858124,
"learning_rate": 2.9910884281727225e-05,
"loss": 0.2502,
"step": 2100
},
{
"epoch": 1.688220551378446,
"grad_norm": 0.16627748310565948,
"learning_rate": 2.916782114851918e-05,
"loss": 0.2939,
"step": 2105
},
{
"epoch": 1.6922305764411028,
"grad_norm": 0.17444783449172974,
"learning_rate": 2.843353215607619e-05,
"loss": 0.2731,
"step": 2110
},
{
"epoch": 1.6962406015037594,
"grad_norm": 0.17445623874664307,
"learning_rate": 2.7708046479290316e-05,
"loss": 0.2752,
"step": 2115
},
{
"epoch": 1.7002506265664161,
"grad_norm": 0.17704740166664124,
"learning_rate": 2.6991392943278874e-05,
"loss": 0.2423,
"step": 2120
},
{
"epoch": 1.7042606516290726,
"grad_norm": 0.16555817425251007,
"learning_rate": 2.6283600022238925e-05,
"loss": 0.2811,
"step": 2125
},
{
"epoch": 1.7082706766917293,
"grad_norm": 0.1633770912885666,
"learning_rate": 2.5584695838316323e-05,
"loss": 0.2463,
"step": 2130
},
{
"epoch": 1.712280701754386,
"grad_norm": 0.17787782847881317,
"learning_rate": 2.4894708160488057e-05,
"loss": 0.266,
"step": 2135
},
{
"epoch": 1.7162907268170426,
"grad_norm": 0.15796466171741486,
"learning_rate": 2.4213664403458903e-05,
"loss": 0.2574,
"step": 2140
},
{
"epoch": 1.7203007518796993,
"grad_norm": 0.1675783395767212,
"learning_rate": 2.3541591626572518e-05,
"loss": 0.251,
"step": 2145
},
{
"epoch": 1.7243107769423558,
"grad_norm": 0.17418669164180756,
"learning_rate": 2.287851653273587e-05,
"loss": 0.2552,
"step": 2150
},
{
"epoch": 1.7283208020050125,
"grad_norm": 0.17587542533874512,
"learning_rate": 2.222446546735868e-05,
"loss": 0.272,
"step": 2155
},
{
"epoch": 1.7323308270676692,
"grad_norm": 0.17269235849380493,
"learning_rate": 2.1579464417306266e-05,
"loss": 0.2586,
"step": 2160
},
{
"epoch": 1.7363408521303259,
"grad_norm": 0.1612161546945572,
"learning_rate": 2.0943539009867423e-05,
"loss": 0.2783,
"step": 2165
},
{
"epoch": 1.7403508771929825,
"grad_norm": 0.15821614861488342,
"learning_rate": 2.0316714511736002e-05,
"loss": 0.2634,
"step": 2170
},
{
"epoch": 1.744360902255639,
"grad_norm": 0.17940925061702728,
"learning_rate": 1.9699015828006788e-05,
"loss": 0.2632,
"step": 2175
},
{
"epoch": 1.7483709273182957,
"grad_norm": 0.1658666729927063,
"learning_rate": 1.909046750118648e-05,
"loss": 0.2666,
"step": 2180
},
{
"epoch": 1.7523809523809524,
"grad_norm": 0.16111963987350464,
"learning_rate": 1.849109371021815e-05,
"loss": 0.2561,
"step": 2185
},
{
"epoch": 1.756390977443609,
"grad_norm": 0.17400680482387543,
"learning_rate": 1.7900918269520672e-05,
"loss": 0.2499,
"step": 2190
},
{
"epoch": 1.7604010025062657,
"grad_norm": 0.16988424956798553,
"learning_rate": 1.73199646280425e-05,
"loss": 0.2719,
"step": 2195
},
{
"epoch": 1.7644110275689222,
"grad_norm": 0.16524524986743927,
"learning_rate": 1.6748255868330148e-05,
"loss": 0.2296,
"step": 2200
},
{
"epoch": 1.768421052631579,
"grad_norm": 0.18039652705192566,
"learning_rate": 1.6185814705610925e-05,
"loss": 0.2677,
"step": 2205
},
{
"epoch": 1.7724310776942356,
"grad_norm": 0.1674271821975708,
"learning_rate": 1.5632663486890407e-05,
"loss": 0.2644,
"step": 2210
},
{
"epoch": 1.7764411027568923,
"grad_norm": 0.1731635481119156,
"learning_rate": 1.5088824190064521e-05,
"loss": 0.2676,
"step": 2215
},
{
"epoch": 1.780451127819549,
"grad_norm": 0.1482301503419876,
"learning_rate": 1.455431842304647e-05,
"loss": 0.2497,
"step": 2220
},
{
"epoch": 1.7844611528822054,
"grad_norm": 0.16372230648994446,
"learning_rate": 1.4029167422908107e-05,
"loss": 0.266,
"step": 2225
},
{
"epoch": 1.788471177944862,
"grad_norm": 0.17107130587100983,
"learning_rate": 1.3513392055036072e-05,
"loss": 0.279,
"step": 2230
},
{
"epoch": 1.7924812030075188,
"grad_norm": 0.15426160395145416,
"learning_rate": 1.3007012812302737e-05,
"loss": 0.2512,
"step": 2235
},
{
"epoch": 1.7964912280701755,
"grad_norm": 0.16106005012989044,
"learning_rate": 1.25100498142523e-05,
"loss": 0.2639,
"step": 2240
},
{
"epoch": 1.8005012531328322,
"grad_norm": 0.16801631450653076,
"learning_rate": 1.2022522806301005e-05,
"loss": 0.2504,
"step": 2245
},
{
"epoch": 1.8045112781954886,
"grad_norm": 0.15076345205307007,
"learning_rate": 1.1544451158952807e-05,
"loss": 0.2516,
"step": 2250
},
{
"epoch": 1.8085213032581455,
"grad_norm": 0.16535955667495728,
"learning_rate": 1.1075853867029694e-05,
"loss": 0.2541,
"step": 2255
},
{
"epoch": 1.812531328320802,
"grad_norm": 0.14483940601348877,
"learning_rate": 1.0616749548917059e-05,
"loss": 0.2392,
"step": 2260
},
{
"epoch": 1.8165413533834587,
"grad_norm": 0.1692323386669159,
"learning_rate": 1.0167156445823777e-05,
"loss": 0.2535,
"step": 2265
},
{
"epoch": 1.8205513784461154,
"grad_norm": 0.15716001391410828,
"learning_rate": 9.727092421057627e-06,
"loss": 0.2369,
"step": 2270
},
{
"epoch": 1.8245614035087718,
"grad_norm": 0.14877931773662567,
"learning_rate": 9.296574959315463e-06,
"loss": 0.2565,
"step": 2275
},
{
"epoch": 1.8285714285714287,
"grad_norm": 0.1587458699941635,
"learning_rate": 8.875621165988474e-06,
"loss": 0.269,
"step": 2280
},
{
"epoch": 1.8325814536340852,
"grad_norm": 0.17435844242572784,
"learning_rate": 8.46424776648258e-06,
"loss": 0.2456,
"step": 2285
},
{
"epoch": 1.8365914786967419,
"grad_norm": 0.16262808442115784,
"learning_rate": 8.06247110555397e-06,
"loss": 0.2582,
"step": 2290
},
{
"epoch": 1.8406015037593986,
"grad_norm": 0.15774986147880554,
"learning_rate": 7.670307146659588e-06,
"loss": 0.2713,
"step": 2295
},
{
"epoch": 1.844611528822055,
"grad_norm": 0.16883234679698944,
"learning_rate": 7.287771471322951e-06,
"loss": 0.2732,
"step": 2300
},
{
"epoch": 1.848621553884712,
"grad_norm": 0.16868174076080322,
"learning_rate": 6.91487927851489e-06,
"loss": 0.2495,
"step": 2305
},
{
"epoch": 1.8526315789473684,
"grad_norm": 0.16309338808059692,
"learning_rate": 6.551645384049898e-06,
"loss": 0.2684,
"step": 2310
},
{
"epoch": 1.856641604010025,
"grad_norm": 0.1636061668395996,
"learning_rate": 6.198084219997374e-06,
"loss": 0.2668,
"step": 2315
},
{
"epoch": 1.8606516290726818,
"grad_norm": 0.17537201941013336,
"learning_rate": 5.854209834108087e-06,
"loss": 0.2743,
"step": 2320
},
{
"epoch": 1.8646616541353382,
"grad_norm": 0.18073910474777222,
"learning_rate": 5.5200358892561755e-06,
"loss": 0.2738,
"step": 2325
},
{
"epoch": 1.8686716791979952,
"grad_norm": 0.16718195378780365,
"learning_rate": 5.195575662896301e-06,
"loss": 0.2668,
"step": 2330
},
{
"epoch": 1.8726817042606516,
"grad_norm": 0.15906399488449097,
"learning_rate": 4.880842046536021e-06,
"loss": 0.2522,
"step": 2335
},
{
"epoch": 1.8766917293233083,
"grad_norm": 0.1543634831905365,
"learning_rate": 4.575847545223499e-06,
"loss": 0.2325,
"step": 2340
},
{
"epoch": 1.880701754385965,
"grad_norm": 0.15531331300735474,
"learning_rate": 4.2806042770509316e-06,
"loss": 0.2503,
"step": 2345
},
{
"epoch": 1.8847117794486214,
"grad_norm": 0.17484238743782043,
"learning_rate": 3.9951239726728485e-06,
"loss": 0.2491,
"step": 2350
},
{
"epoch": 1.8887218045112784,
"grad_norm": 0.15897777676582336,
"learning_rate": 3.719417974839989e-06,
"loss": 0.2708,
"step": 2355
},
{
"epoch": 1.8927318295739348,
"grad_norm": 0.16040709614753723,
"learning_rate": 3.453497237948855e-06,
"loss": 0.2655,
"step": 2360
},
{
"epoch": 1.8967418546365915,
"grad_norm": 0.14991337060928345,
"learning_rate": 3.1973723276062517e-06,
"loss": 0.2275,
"step": 2365
},
{
"epoch": 1.9007518796992482,
"grad_norm": 0.1545797735452652,
"learning_rate": 2.9510534202096263e-06,
"loss": 0.2576,
"step": 2370
},
{
"epoch": 1.9047619047619047,
"grad_norm": 0.16641472280025482,
"learning_rate": 2.71455030254264e-06,
"loss": 0.2607,
"step": 2375
},
{
"epoch": 1.9087719298245616,
"grad_norm": 0.1749178171157837,
"learning_rate": 2.487872371386424e-06,
"loss": 0.2679,
"step": 2380
},
{
"epoch": 1.912781954887218,
"grad_norm": 0.1886933445930481,
"learning_rate": 2.271028633146127e-06,
"loss": 0.2631,
"step": 2385
},
{
"epoch": 1.9167919799498747,
"grad_norm": 0.15673983097076416,
"learning_rate": 2.064027703493149e-06,
"loss": 0.2502,
"step": 2390
},
{
"epoch": 1.9208020050125314,
"grad_norm": 0.16062475740909576,
"learning_rate": 1.866877807022771e-06,
"loss": 0.2295,
"step": 2395
},
{
"epoch": 1.9248120300751879,
"grad_norm": 0.16452881693840027,
"learning_rate": 1.6795867769273665e-06,
"loss": 0.2585,
"step": 2400
},
{
"epoch": 1.9288220551378448,
"grad_norm": 0.17597217857837677,
"learning_rate": 1.50216205468523e-06,
"loss": 0.2586,
"step": 2405
},
{
"epoch": 1.9328320802005012,
"grad_norm": 0.1579117476940155,
"learning_rate": 1.3346106897648424e-06,
"loss": 0.2442,
"step": 2410
},
{
"epoch": 1.936842105263158,
"grad_norm": 0.16751398146152496,
"learning_rate": 1.1769393393448457e-06,
"loss": 0.2649,
"step": 2415
},
{
"epoch": 1.9408521303258146,
"grad_norm": 0.17506803572177887,
"learning_rate": 1.0291542680494758e-06,
"loss": 0.2709,
"step": 2420
},
{
"epoch": 1.944862155388471,
"grad_norm": 0.17089371383190155,
"learning_rate": 8.912613476997067e-07,
"loss": 0.2739,
"step": 2425
},
{
"epoch": 1.948872180451128,
"grad_norm": 0.16731411218643188,
"learning_rate": 7.632660570799366e-07,
"loss": 0.2617,
"step": 2430
},
{
"epoch": 1.9528822055137844,
"grad_norm": 0.15917718410491943,
"learning_rate": 6.451734817202736e-07,
"loss": 0.2783,
"step": 2435
},
{
"epoch": 1.9568922305764411,
"grad_norm": 0.17604759335517883,
"learning_rate": 5.369883136945309e-07,
"loss": 0.2538,
"step": 2440
},
{
"epoch": 1.9609022556390978,
"grad_norm": 0.1530042290687561,
"learning_rate": 4.387148514337358e-07,
"loss": 0.2652,
"step": 2445
},
{
"epoch": 1.9649122807017543,
"grad_norm": 0.15754002332687378,
"learning_rate": 3.503569995554068e-07,
"loss": 0.2615,
"step": 2450
},
{
"epoch": 1.9689223057644112,
"grad_norm": 0.1827680766582489,
"learning_rate": 2.7191826870834325e-07,
"loss": 0.2742,
"step": 2455
},
{
"epoch": 1.9729323308270676,
"grad_norm": 0.15967601537704468,
"learning_rate": 2.034017754332651e-07,
"loss": 0.2466,
"step": 2460
},
{
"epoch": 1.9769423558897243,
"grad_norm": 0.17333175241947174,
"learning_rate": 1.4481024203877313e-07,
"loss": 0.2764,
"step": 2465
},
{
"epoch": 1.980952380952381,
"grad_norm": 0.16165874898433685,
"learning_rate": 9.6145996493463e-08,
"loss": 0.2568,
"step": 2470
},
{
"epoch": 1.9849624060150375,
"grad_norm": 0.16772541403770447,
"learning_rate": 5.7410972333193924e-08,
"loss": 0.2407,
"step": 2475
},
{
"epoch": 1.9889724310776944,
"grad_norm": 0.14310236275196075,
"learning_rate": 2.8606708584344445e-08,
"loss": 0.2468,
"step": 2480
},
{
"epoch": 1.9929824561403509,
"grad_norm": 0.1455429494380951,
"learning_rate": 9.734349702722467e-09,
"loss": 0.2635,
"step": 2485
},
{
"epoch": 1.9969924812030075,
"grad_norm": 0.16766951978206635,
"learning_rate": 7.94645528018334e-10,
"loss": 0.2535,
"step": 2490
},
{
"epoch": 1.9985964912280703,
"eval_loss": 0.22910664975643158,
"eval_runtime": 8.4374,
"eval_samples_per_second": 6.519,
"eval_steps_per_second": 0.83,
"step": 2492
},
{
"epoch": 1.9985964912280703,
"step": 2492,
"total_flos": 8.025299762167153e+17,
"train_loss": 0.37913749282088366,
"train_runtime": 28381.1385,
"train_samples_per_second": 2.812,
"train_steps_per_second": 0.088
},
{
"epoch": 1.9985964912280703,
"eval_loss": 0.22910664975643158,
"eval_runtime": 8.378,
"eval_samples_per_second": 6.565,
"eval_steps_per_second": 0.836,
"step": 2492
}
],
"logging_steps": 5,
"max_steps": 2492,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"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": 8.025299762167153e+17,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}