Files
mistral-7B-v0.3-finetuned/trainer_state.json

4376 lines
107 KiB
JSON
Raw Permalink Normal View History

{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 1000,
"global_step": 6123,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0016331863465621427,
"grad_norm": 145.97381805867667,
"learning_rate": 1.465798045602606e-06,
"loss": 1.6692,
"step": 10
},
{
"epoch": 0.0032663726931242854,
"grad_norm": 63.9466770045178,
"learning_rate": 3.0944625407166126e-06,
"loss": 0.5489,
"step": 20
},
{
"epoch": 0.004899559039686428,
"grad_norm": 69.53802947610717,
"learning_rate": 4.723127035830619e-06,
"loss": 0.2667,
"step": 30
},
{
"epoch": 0.006532745386248571,
"grad_norm": 16.061588061882638,
"learning_rate": 6.351791530944626e-06,
"loss": 0.1901,
"step": 40
},
{
"epoch": 0.008165931732810714,
"grad_norm": 3.8978974378230506,
"learning_rate": 7.980456026058632e-06,
"loss": 0.1569,
"step": 50
},
{
"epoch": 0.009799118079372856,
"grad_norm": 5.050891692615998,
"learning_rate": 9.609120521172639e-06,
"loss": 0.1972,
"step": 60
},
{
"epoch": 0.011432304425935,
"grad_norm": 6.8709522813864,
"learning_rate": 1.1237785016286644e-05,
"loss": 0.1833,
"step": 70
},
{
"epoch": 0.013065490772497142,
"grad_norm": 15.328524805969542,
"learning_rate": 1.2866449511400652e-05,
"loss": 0.178,
"step": 80
},
{
"epoch": 0.014698677119059285,
"grad_norm": 9.077319496812207,
"learning_rate": 1.449511400651466e-05,
"loss": 0.1688,
"step": 90
},
{
"epoch": 0.01633186346562143,
"grad_norm": 2.5767038263669377,
"learning_rate": 1.6123778501628666e-05,
"loss": 0.1501,
"step": 100
},
{
"epoch": 0.01796504981218357,
"grad_norm": 3.3702586691440404,
"learning_rate": 1.775244299674267e-05,
"loss": 0.1225,
"step": 110
},
{
"epoch": 0.019598236158745713,
"grad_norm": 15.419660529394266,
"learning_rate": 1.938110749185668e-05,
"loss": 0.1455,
"step": 120
},
{
"epoch": 0.021231422505307854,
"grad_norm": 9.204576218578334,
"learning_rate": 2.1009771986970685e-05,
"loss": 0.1232,
"step": 130
},
{
"epoch": 0.02286460885187,
"grad_norm": 4.6950400785593125,
"learning_rate": 2.263843648208469e-05,
"loss": 0.1183,
"step": 140
},
{
"epoch": 0.02449779519843214,
"grad_norm": 10.023345229462917,
"learning_rate": 2.42671009771987e-05,
"loss": 0.139,
"step": 150
},
{
"epoch": 0.026130981544994283,
"grad_norm": 25.48352980523854,
"learning_rate": 2.5895765472312705e-05,
"loss": 0.1362,
"step": 160
},
{
"epoch": 0.027764167891556425,
"grad_norm": 179.652655213722,
"learning_rate": 2.7524429967426713e-05,
"loss": 0.1686,
"step": 170
},
{
"epoch": 0.02939735423811857,
"grad_norm": 3.2472688190126,
"learning_rate": 2.915309446254072e-05,
"loss": 0.1423,
"step": 180
},
{
"epoch": 0.031030540584680712,
"grad_norm": 130.6490570657701,
"learning_rate": 3.078175895765473e-05,
"loss": 0.1394,
"step": 190
},
{
"epoch": 0.03266372693124286,
"grad_norm": 17.023448424719476,
"learning_rate": 3.241042345276873e-05,
"loss": 0.1662,
"step": 200
},
{
"epoch": 0.034296913277805,
"grad_norm": 40.447851916392466,
"learning_rate": 3.403908794788274e-05,
"loss": 0.1537,
"step": 210
},
{
"epoch": 0.03593009962436714,
"grad_norm": 1.7806735692879652,
"learning_rate": 3.566775244299674e-05,
"loss": 0.1613,
"step": 220
},
{
"epoch": 0.03756328597092928,
"grad_norm": 5.439746759275804,
"learning_rate": 3.729641693811075e-05,
"loss": 0.1249,
"step": 230
},
{
"epoch": 0.039196472317491425,
"grad_norm": 4.633337483961239,
"learning_rate": 3.892508143322476e-05,
"loss": 0.137,
"step": 240
},
{
"epoch": 0.04082965866405357,
"grad_norm": 0.9111761023777424,
"learning_rate": 4.0553745928338767e-05,
"loss": 0.157,
"step": 250
},
{
"epoch": 0.04246284501061571,
"grad_norm": 3.9227244169283555,
"learning_rate": 4.2182410423452775e-05,
"loss": 0.1509,
"step": 260
},
{
"epoch": 0.04409603135717785,
"grad_norm": 5.7305735330406735,
"learning_rate": 4.381107491856678e-05,
"loss": 0.1432,
"step": 270
},
{
"epoch": 0.04572921770374,
"grad_norm": 35.150516987852924,
"learning_rate": 4.543973941368078e-05,
"loss": 0.4369,
"step": 280
},
{
"epoch": 0.04736240405030214,
"grad_norm": 2.4082560421040284,
"learning_rate": 4.706840390879479e-05,
"loss": 0.1753,
"step": 290
},
{
"epoch": 0.04899559039686428,
"grad_norm": 6.3919055929048545,
"learning_rate": 4.86970684039088e-05,
"loss": 0.1577,
"step": 300
},
{
"epoch": 0.050628776743426425,
"grad_norm": 1.7049096159399162,
"learning_rate": 4.999998541115495e-05,
"loss": 0.1407,
"step": 310
},
{
"epoch": 0.05226196308998857,
"grad_norm": 1.1785931753061112,
"learning_rate": 4.999947480336581e-05,
"loss": 0.1297,
"step": 320
},
{
"epoch": 0.05389514943655071,
"grad_norm": 1.6792978348080347,
"learning_rate": 4.9998234770350544e-05,
"loss": 0.1377,
"step": 330
},
{
"epoch": 0.05552833578311285,
"grad_norm": 3.3597358418220686,
"learning_rate": 4.999626534829037e-05,
"loss": 0.1441,
"step": 340
},
{
"epoch": 0.057161522129675,
"grad_norm": 0.6121380644736275,
"learning_rate": 4.9993566594648356e-05,
"loss": 0.1219,
"step": 350
},
{
"epoch": 0.05879470847623714,
"grad_norm": 0.7770612425638073,
"learning_rate": 4.9990138588167696e-05,
"loss": 0.1165,
"step": 360
},
{
"epoch": 0.06042789482279928,
"grad_norm": 0.6774948858473921,
"learning_rate": 4.998598142886947e-05,
"loss": 0.1332,
"step": 370
},
{
"epoch": 0.062061081169361425,
"grad_norm": 0.7888908355244016,
"learning_rate": 4.9981095238049714e-05,
"loss": 0.1154,
"step": 380
},
{
"epoch": 0.06369426751592357,
"grad_norm": 0.6906625554166081,
"learning_rate": 4.997548015827583e-05,
"loss": 0.1168,
"step": 390
},
{
"epoch": 0.06532745386248572,
"grad_norm": 0.46256145095399775,
"learning_rate": 4.9969136353382526e-05,
"loss": 0.112,
"step": 400
},
{
"epoch": 0.06696064020904785,
"grad_norm": 4.435885009892541,
"learning_rate": 4.996206400846692e-05,
"loss": 0.1147,
"step": 410
},
{
"epoch": 0.06859382655561,
"grad_norm": 0.8503068116512648,
"learning_rate": 4.995426332988323e-05,
"loss": 0.1191,
"step": 420
},
{
"epoch": 0.07022701290217213,
"grad_norm": 13.519548375410174,
"learning_rate": 4.99457345452367e-05,
"loss": 0.1231,
"step": 430
},
{
"epoch": 0.07186019924873428,
"grad_norm": 1.1495715676750418,
"learning_rate": 4.9936477903376985e-05,
"loss": 0.1195,
"step": 440
},
{
"epoch": 0.07349338559529642,
"grad_norm": 0.49053087808698165,
"learning_rate": 4.992649367439089e-05,
"loss": 0.1064,
"step": 450
},
{
"epoch": 0.07512657194185857,
"grad_norm": 1.2284082829295608,
"learning_rate": 4.9915782149594467e-05,
"loss": 0.1075,
"step": 460
},
{
"epoch": 0.07675975828842072,
"grad_norm": 0.41690962193899994,
"learning_rate": 4.990434364152455e-05,
"loss": 0.108,
"step": 470
},
{
"epoch": 0.07839294463498285,
"grad_norm": 0.3957345635557366,
"learning_rate": 4.9892178483929584e-05,
"loss": 0.103,
"step": 480
},
{
"epoch": 0.080026130981545,
"grad_norm": 0.531084266886797,
"learning_rate": 4.987928703175995e-05,
"loss": 0.1018,
"step": 490
},
{
"epoch": 0.08165931732810713,
"grad_norm": 3.298265212812673,
"learning_rate": 4.9865669661157565e-05,
"loss": 0.1028,
"step": 500
},
{
"epoch": 0.08329250367466928,
"grad_norm": 0.3815905332260957,
"learning_rate": 4.985132676944492e-05,
"loss": 0.1097,
"step": 510
},
{
"epoch": 0.08492569002123142,
"grad_norm": 0.5264684990218298,
"learning_rate": 4.983625877511349e-05,
"loss": 0.1018,
"step": 520
},
{
"epoch": 0.08655887636779357,
"grad_norm": 0.4544320955641461,
"learning_rate": 4.982046611781152e-05,
"loss": 0.0992,
"step": 530
},
{
"epoch": 0.0881920627143557,
"grad_norm": 0.5141683211211776,
"learning_rate": 4.9803949258331185e-05,
"loss": 0.0989,
"step": 540
},
{
"epoch": 0.08982524906091785,
"grad_norm": 1.3180089092544367,
"learning_rate": 4.9786708678595185e-05,
"loss": 0.114,
"step": 550
},
{
"epoch": 0.09145843540748,
"grad_norm": 2.179679867180841,
"learning_rate": 4.9768744881642616e-05,
"loss": 0.1232,
"step": 560
},
{
"epoch": 0.09309162175404213,
"grad_norm": 0.4945952085408727,
"learning_rate": 4.9750058391614375e-05,
"loss": 0.1036,
"step": 570
},
{
"epoch": 0.09472480810060428,
"grad_norm": 4.958533967596306,
"learning_rate": 4.97306497537378e-05,
"loss": 0.1057,
"step": 580
},
{
"epoch": 0.09635799444716642,
"grad_norm": 0.5604914542274124,
"learning_rate": 4.9710519534310785e-05,
"loss": 0.1425,
"step": 590
},
{
"epoch": 0.09799118079372857,
"grad_norm": 0.4221819769513211,
"learning_rate": 4.968966832068527e-05,
"loss": 0.1,
"step": 600
},
{
"epoch": 0.0996243671402907,
"grad_norm": 0.36525945886033623,
"learning_rate": 4.9668096721250075e-05,
"loss": 0.0969,
"step": 610
},
{
"epoch": 0.10125755348685285,
"grad_norm": 1.037999790663953,
"learning_rate": 4.9645805365413177e-05,
"loss": 0.0969,
"step": 620
},
{
"epoch": 0.102890739833415,
"grad_norm": 2.5605550328005426,
"learning_rate": 4.962279490358334e-05,
"loss": 0.1015,
"step": 630
},
{
"epoch": 0.10452392617997713,
"grad_norm": 0.40479139773747297,
"learning_rate": 4.959906600715112e-05,
"loss": 0.1054,
"step": 640
},
{
"epoch": 0.10615711252653928,
"grad_norm": 0.434556601481739,
"learning_rate": 4.9574619368469276e-05,
"loss": 0.0989,
"step": 650
},
{
"epoch": 0.10779029887310142,
"grad_norm": 0.316717605393854,
"learning_rate": 4.9549455700832595e-05,
"loss": 0.097,
"step": 660
},
{
"epoch": 0.10942348521966357,
"grad_norm": 0.3545571114759296,
"learning_rate": 4.9523575738457065e-05,
"loss": 0.0942,
"step": 670
},
{
"epoch": 0.1110566715662257,
"grad_norm": 0.4064905156454397,
"learning_rate": 4.9496980236458445e-05,
"loss": 0.0892,
"step": 680
},
{
"epoch": 0.11268985791278785,
"grad_norm": 0.3424891203682006,
"learning_rate": 4.9469669970830235e-05,
"loss": 0.0932,
"step": 690
},
{
"epoch": 0.11432304425935,
"grad_norm": 0.39610757732073953,
"learning_rate": 4.944164573842105e-05,
"loss": 0.091,
"step": 700
},
{
"epoch": 0.11595623060591213,
"grad_norm": 0.45954292386117307,
"learning_rate": 4.9412908356911337e-05,
"loss": 0.0906,
"step": 710
},
{
"epoch": 0.11758941695247428,
"grad_norm": 0.34587941820280377,
"learning_rate": 4.9383458664789564e-05,
"loss": 0.091,
"step": 720
},
{
"epoch": 0.11922260329903642,
"grad_norm": 0.3363193663907574,
"learning_rate": 4.9353297521327704e-05,
"loss": 0.0914,
"step": 730
},
{
"epoch": 0.12085578964559857,
"grad_norm": 0.34006860574325676,
"learning_rate": 4.932242580655621e-05,
"loss": 0.0908,
"step": 740
},
{
"epoch": 0.1224889759921607,
"grad_norm": 0.3211702232736895,
"learning_rate": 4.92908444212383e-05,
"loss": 0.0915,
"step": 750
},
{
"epoch": 0.12412216233872285,
"grad_norm": 0.3042137881352104,
"learning_rate": 4.925855428684371e-05,
"loss": 0.0871,
"step": 760
},
{
"epoch": 0.12575534868528498,
"grad_norm": 0.3110002714995785,
"learning_rate": 4.922555634552176e-05,
"loss": 0.087,
"step": 770
},
{
"epoch": 0.12738853503184713,
"grad_norm": 0.3004893613603362,
"learning_rate": 4.919185156007392e-05,
"loss": 0.0884,
"step": 780
},
{
"epoch": 0.12902172137840928,
"grad_norm": 0.3536816004856033,
"learning_rate": 4.915744091392567e-05,
"loss": 0.0905,
"step": 790
},
{
"epoch": 0.13065490772497143,
"grad_norm": 0.28502698340252075,
"learning_rate": 4.9122325411097836e-05,
"loss": 0.0914,
"step": 800
},
{
"epoch": 0.13228809407153355,
"grad_norm": 0.3367365084978498,
"learning_rate": 4.908650607617729e-05,
"loss": 0.0885,
"step": 810
},
{
"epoch": 0.1339212804180957,
"grad_norm": 0.3028341194791023,
"learning_rate": 4.904998395428706e-05,
"loss": 0.086,
"step": 820
},
{
"epoch": 0.13555446676465785,
"grad_norm": 1.7470342688373222,
"learning_rate": 4.901276011105579e-05,
"loss": 0.0852,
"step": 830
},
{
"epoch": 0.13718765311122,
"grad_norm": 0.36170689370113385,
"learning_rate": 4.8974835632586716e-05,
"loss": 0.0857,
"step": 840
},
{
"epoch": 0.13882083945778212,
"grad_norm": 0.2990351248657847,
"learning_rate": 4.8936211625425944e-05,
"loss": 0.0858,
"step": 850
},
{
"epoch": 0.14045402580434427,
"grad_norm": 0.2707938653873345,
"learning_rate": 4.889688921653014e-05,
"loss": 0.0875,
"step": 860
},
{
"epoch": 0.14208721215090642,
"grad_norm": 0.3644861025739434,
"learning_rate": 4.8856869553233706e-05,
"loss": 0.0868,
"step": 870
},
{
"epoch": 0.14372039849746857,
"grad_norm": 0.27651233473925974,
"learning_rate": 4.8816153803215235e-05,
"loss": 0.0861,
"step": 880
},
{
"epoch": 0.14535358484403071,
"grad_norm": 0.29441174421804195,
"learning_rate": 4.8774743154463495e-05,
"loss": 0.0856,
"step": 890
},
{
"epoch": 0.14698677119059284,
"grad_norm": 0.24428157703477268,
"learning_rate": 4.873263881524275e-05,
"loss": 0.0838,
"step": 900
},
{
"epoch": 0.14861995753715498,
"grad_norm": 0.2528107283013204,
"learning_rate": 4.8689842014057485e-05,
"loss": 0.0811,
"step": 910
},
{
"epoch": 0.15025314388371713,
"grad_norm": 0.2807606608583151,
"learning_rate": 4.864635399961659e-05,
"loss": 0.0854,
"step": 920
},
{
"epoch": 0.15188633023027928,
"grad_norm": 0.5980205654545195,
"learning_rate": 4.860217604079691e-05,
"loss": 0.0839,
"step": 930
},
{
"epoch": 0.15351951657684143,
"grad_norm": 0.3525389540420788,
"learning_rate": 4.8557309426606234e-05,
"loss": 0.0862,
"step": 940
},
{
"epoch": 0.15515270292340355,
"grad_norm": 0.27967418293424284,
"learning_rate": 4.851175546614567e-05,
"loss": 0.0851,
"step": 950
},
{
"epoch": 0.1567858892699657,
"grad_norm": 2.851285121692407,
"learning_rate": 4.846551548857145e-05,
"loss": 0.0832,
"step": 960
},
{
"epoch": 0.15841907561652785,
"grad_norm": 0.490434282346677,
"learning_rate": 4.8418590843056165e-05,
"loss": 0.0872,
"step": 970
},
{
"epoch": 0.16005226196309,
"grad_norm": 0.3244649357958235,
"learning_rate": 4.8370982898749397e-05,
"loss": 0.0953,
"step": 980
},
{
"epoch": 0.16168544830965212,
"grad_norm": 0.38464410153848855,
"learning_rate": 4.832269304473773e-05,
"loss": 0.0871,
"step": 990
},
{
"epoch": 0.16331863465621427,
"grad_norm": 0.2993500124169986,
"learning_rate": 4.8273722690004286e-05,
"loss": 0.0817,
"step": 1000
},
{
"epoch": 0.16331863465621427,
"eval_loss": 0.084870345890522,
"eval_runtime": 547.1785,
"eval_samples_per_second": 11.053,
"eval_steps_per_second": 0.691,
"step": 1000
},
{
"epoch": 0.16495182100277642,
"grad_norm": 0.35498874701174743,
"learning_rate": 4.822407326338755e-05,
"loss": 0.0859,
"step": 1010
},
{
"epoch": 0.16658500734933857,
"grad_norm": 0.35696399125886435,
"learning_rate": 4.817374621353974e-05,
"loss": 0.0852,
"step": 1020
},
{
"epoch": 0.16821819369590071,
"grad_norm": 0.31404362940408626,
"learning_rate": 4.8122743008884485e-05,
"loss": 0.0822,
"step": 1030
},
{
"epoch": 0.16985138004246284,
"grad_norm": 0.22468450893926695,
"learning_rate": 4.807106513757401e-05,
"loss": 0.0821,
"step": 1040
},
{
"epoch": 0.17148456638902498,
"grad_norm": 0.28059441036386645,
"learning_rate": 4.8018714107445716e-05,
"loss": 0.0834,
"step": 1050
},
{
"epoch": 0.17311775273558713,
"grad_norm": 0.4710450267893097,
"learning_rate": 4.7965691445978164e-05,
"loss": 0.0792,
"step": 1060
},
{
"epoch": 0.17475093908214928,
"grad_norm": 0.24495352736497042,
"learning_rate": 4.791199870024654e-05,
"loss": 0.0797,
"step": 1070
},
{
"epoch": 0.1763841254287114,
"grad_norm": 0.20885215808474264,
"learning_rate": 4.7857637436877465e-05,
"loss": 0.0818,
"step": 1080
},
{
"epoch": 0.17801731177527355,
"grad_norm": 0.215691238720452,
"learning_rate": 4.780260924200336e-05,
"loss": 0.0807,
"step": 1090
},
{
"epoch": 0.1796504981218357,
"grad_norm": 0.2490550206899775,
"learning_rate": 4.7746915721216065e-05,
"loss": 0.0827,
"step": 1100
},
{
"epoch": 0.18128368446839785,
"grad_norm": 0.39652823253024566,
"learning_rate": 4.769055849952009e-05,
"loss": 0.0803,
"step": 1110
},
{
"epoch": 0.18291687081496,
"grad_norm": 0.33783505428961685,
"learning_rate": 4.763353922128515e-05,
"loss": 0.079,
"step": 1120
},
{
"epoch": 0.18455005716152212,
"grad_norm": 1.4193607684304164,
"learning_rate": 4.75758595501982e-05,
"loss": 0.0831,
"step": 1130
},
{
"epoch": 0.18618324350808427,
"grad_norm": 0.22157513322444444,
"learning_rate": 4.751752116921486e-05,
"loss": 0.0823,
"step": 1140
},
{
"epoch": 0.18781642985464642,
"grad_norm": 0.24437544021044588,
"learning_rate": 4.745852578051037e-05,
"loss": 0.083,
"step": 1150
},
{
"epoch": 0.18944961620120856,
"grad_norm": 0.40759179364886006,
"learning_rate": 4.739887510542988e-05,
"loss": 0.08,
"step": 1160
},
{
"epoch": 0.1910828025477707,
"grad_norm": 0.29537676355806713,
"learning_rate": 4.7338570884438246e-05,
"loss": 0.0799,
"step": 1170
},
{
"epoch": 0.19271598889433283,
"grad_norm": 0.22588452973624223,
"learning_rate": 4.727761487706923e-05,
"loss": 0.0798,
"step": 1180
},
{
"epoch": 0.19434917524089498,
"grad_norm": 0.21344757100162806,
"learning_rate": 4.721600886187417e-05,
"loss": 0.0767,
"step": 1190
},
{
"epoch": 0.19598236158745713,
"grad_norm": 0.25991504119395453,
"learning_rate": 4.715375463637008e-05,
"loss": 0.0808,
"step": 1200
},
{
"epoch": 0.19761554793401928,
"grad_norm": 0.2475588246591517,
"learning_rate": 4.7090854016987253e-05,
"loss": 0.0809,
"step": 1210
},
{
"epoch": 0.1992487342805814,
"grad_norm": 0.23614583141411066,
"learning_rate": 4.702730883901615e-05,
"loss": 0.0791,
"step": 1220
},
{
"epoch": 0.20088192062714355,
"grad_norm": 0.2631528314417221,
"learning_rate": 4.6963120956553974e-05,
"loss": 0.078,
"step": 1230
},
{
"epoch": 0.2025151069737057,
"grad_norm": 0.2025288663201872,
"learning_rate": 4.689829224245049e-05,
"loss": 0.0803,
"step": 1240
},
{
"epoch": 0.20414829332026785,
"grad_norm": 0.26782894744915264,
"learning_rate": 4.683282458825342e-05,
"loss": 0.0763,
"step": 1250
},
{
"epoch": 0.20578147966683,
"grad_norm": 0.22528867982449569,
"learning_rate": 4.676671990415322e-05,
"loss": 0.0767,
"step": 1260
},
{
"epoch": 0.20741466601339212,
"grad_norm": 0.3121589421154957,
"learning_rate": 4.6699980118927385e-05,
"loss": 0.0815,
"step": 1270
},
{
"epoch": 0.20904785235995427,
"grad_norm": 0.24114689802808423,
"learning_rate": 4.6632607179884133e-05,
"loss": 0.0786,
"step": 1280
},
{
"epoch": 0.21068103870651642,
"grad_norm": 0.3511063691183293,
"learning_rate": 4.656460305280562e-05,
"loss": 0.0805,
"step": 1290
},
{
"epoch": 0.21231422505307856,
"grad_norm": 0.255683441381279,
"learning_rate": 4.649596972189056e-05,
"loss": 0.0767,
"step": 1300
},
{
"epoch": 0.21394741139964069,
"grad_norm": 0.22565237584222458,
"learning_rate": 4.642670918969634e-05,
"loss": 0.0818,
"step": 1310
},
{
"epoch": 0.21558059774620283,
"grad_norm": 0.23155726918596503,
"learning_rate": 4.635682347708057e-05,
"loss": 0.0815,
"step": 1320
},
{
"epoch": 0.21721378409276498,
"grad_norm": 0.3095169921385205,
"learning_rate": 4.628631462314219e-05,
"loss": 0.0791,
"step": 1330
},
{
"epoch": 0.21884697043932713,
"grad_norm": 0.2139118613106089,
"learning_rate": 4.621518468516186e-05,
"loss": 0.0767,
"step": 1340
},
{
"epoch": 0.22048015678588928,
"grad_norm": 0.26563327412444715,
"learning_rate": 4.614343573854204e-05,
"loss": 0.0805,
"step": 1350
},
{
"epoch": 0.2221133431324514,
"grad_norm": 0.21487063050629832,
"learning_rate": 4.607106987674636e-05,
"loss": 0.0808,
"step": 1360
},
{
"epoch": 0.22374652947901355,
"grad_norm": 2.790160881644759,
"learning_rate": 4.599808921123859e-05,
"loss": 0.0828,
"step": 1370
},
{
"epoch": 0.2253797158255757,
"grad_norm": 1.208495521153815,
"learning_rate": 4.5924495871421e-05,
"loss": 0.1373,
"step": 1380
},
{
"epoch": 0.22701290217213785,
"grad_norm": 0.3264121207623596,
"learning_rate": 4.5850292004572245e-05,
"loss": 0.0891,
"step": 1390
},
{
"epoch": 0.2286460885187,
"grad_norm": 0.3384338046246461,
"learning_rate": 4.57754797757847e-05,
"loss": 0.0835,
"step": 1400
},
{
"epoch": 0.23027927486526212,
"grad_norm": 0.35089637077362557,
"learning_rate": 4.5700061367901306e-05,
"loss": 0.0817,
"step": 1410
},
{
"epoch": 0.23191246121182427,
"grad_norm": 0.29673242863655525,
"learning_rate": 4.562403898145183e-05,
"loss": 0.0858,
"step": 1420
},
{
"epoch": 0.23354564755838642,
"grad_norm": 0.25458614622387543,
"learning_rate": 4.554741483458876e-05,
"loss": 0.0812,
"step": 1430
},
{
"epoch": 0.23517883390494856,
"grad_norm": 0.22141136828335095,
"learning_rate": 4.547019116302249e-05,
"loss": 0.0792,
"step": 1440
},
{
"epoch": 0.23681202025151069,
"grad_norm": 0.32725138896683775,
"learning_rate": 4.539237021995611e-05,
"loss": 0.0797,
"step": 1450
},
{
"epoch": 0.23844520659807283,
"grad_norm": 0.4376257084969803,
"learning_rate": 4.531395427601966e-05,
"loss": 0.0782,
"step": 1460
},
{
"epoch": 0.24007839294463498,
"grad_norm": 0.26301635435465603,
"learning_rate": 4.523494561920394e-05,
"loss": 0.0786,
"step": 1470
},
{
"epoch": 0.24171157929119713,
"grad_norm": 0.20962291850199896,
"learning_rate": 4.515534655479366e-05,
"loss": 0.0785,
"step": 1480
},
{
"epoch": 0.24334476563775928,
"grad_norm": 0.2057571065641452,
"learning_rate": 4.5075159405300244e-05,
"loss": 0.0787,
"step": 1490
},
{
"epoch": 0.2449779519843214,
"grad_norm": 0.22026811864950177,
"learning_rate": 4.499438651039401e-05,
"loss": 0.078,
"step": 1500
},
{
"epoch": 0.24661113833088355,
"grad_norm": 0.25783917702875475,
"learning_rate": 4.491303022683598e-05,
"loss": 0.0772,
"step": 1510
},
{
"epoch": 0.2482443246774457,
"grad_norm": 0.24307453420631892,
"learning_rate": 4.483109292840903e-05,
"loss": 0.0772,
"step": 1520
},
{
"epoch": 0.24987751102400785,
"grad_norm": 0.2742753889756254,
"learning_rate": 4.474857700584868e-05,
"loss": 0.0772,
"step": 1530
},
{
"epoch": 0.25151069737056997,
"grad_norm": 0.19445102825688804,
"learning_rate": 4.466548486677332e-05,
"loss": 0.082,
"step": 1540
},
{
"epoch": 0.25314388371713215,
"grad_norm": 0.26544323510101464,
"learning_rate": 4.4581818935613996e-05,
"loss": 0.0784,
"step": 1550
},
{
"epoch": 0.25477707006369427,
"grad_norm": 0.20869952354674196,
"learning_rate": 4.44975816535436e-05,
"loss": 0.078,
"step": 1560
},
{
"epoch": 0.2564102564102564,
"grad_norm": 0.2721904679146718,
"learning_rate": 4.441277547840573e-05,
"loss": 0.0766,
"step": 1570
},
{
"epoch": 0.25804344275681856,
"grad_norm": 0.23389498183862642,
"learning_rate": 4.432740288464289e-05,
"loss": 0.0756,
"step": 1580
},
{
"epoch": 0.2596766291033807,
"grad_norm": 0.21544093606177983,
"learning_rate": 4.424146636322435e-05,
"loss": 0.0758,
"step": 1590
},
{
"epoch": 0.26130981544994286,
"grad_norm": 0.2033437549488562,
"learning_rate": 4.4154968421573454e-05,
"loss": 0.0786,
"step": 1600
},
{
"epoch": 0.262943001796505,
"grad_norm": 0.1846338526255751,
"learning_rate": 4.406791158349446e-05,
"loss": 0.0767,
"step": 1610
},
{
"epoch": 0.2645761881430671,
"grad_norm": 0.22171443387025816,
"learning_rate": 4.398029838909886e-05,
"loss": 0.074,
"step": 1620
},
{
"epoch": 0.2662093744896293,
"grad_norm": 0.19164895267513063,
"learning_rate": 4.389213139473133e-05,
"loss": 0.0745,
"step": 1630
},
{
"epoch": 0.2678425608361914,
"grad_norm": 0.19989423444665402,
"learning_rate": 4.3803413172895115e-05,
"loss": 0.075,
"step": 1640
},
{
"epoch": 0.2694757471827536,
"grad_norm": 0.1835447353098617,
"learning_rate": 4.371414631217695e-05,
"loss": 0.0752,
"step": 1650
},
{
"epoch": 0.2711089335293157,
"grad_norm": 0.2321196677317487,
"learning_rate": 4.3624333417171545e-05,
"loss": 0.0746,
"step": 1660
},
{
"epoch": 0.2727421198758778,
"grad_norm": 0.21794759890790977,
"learning_rate": 4.353397710840562e-05,
"loss": 0.0739,
"step": 1670
},
{
"epoch": 0.27437530622244,
"grad_norm": 0.19905577560374166,
"learning_rate": 4.3443080022261384e-05,
"loss": 0.0734,
"step": 1680
},
{
"epoch": 0.2760084925690021,
"grad_norm": 0.23240645012002362,
"learning_rate": 4.335164481089966e-05,
"loss": 0.0728,
"step": 1690
},
{
"epoch": 0.27764167891556424,
"grad_norm": 0.1979143614885727,
"learning_rate": 4.325967414218249e-05,
"loss": 0.0732,
"step": 1700
},
{
"epoch": 0.2792748652621264,
"grad_norm": 0.19875940224295757,
"learning_rate": 4.316717069959527e-05,
"loss": 0.0733,
"step": 1710
},
{
"epoch": 0.28090805160868854,
"grad_norm": 0.24050388467988038,
"learning_rate": 4.307413718216849e-05,
"loss": 0.0726,
"step": 1720
},
{
"epoch": 0.2825412379552507,
"grad_norm": 0.21607286419311014,
"learning_rate": 4.2980576304398956e-05,
"loss": 0.0746,
"step": 1730
},
{
"epoch": 0.28417442430181283,
"grad_norm": 0.19231119779606923,
"learning_rate": 4.28864907961706e-05,
"loss": 0.0742,
"step": 1740
},
{
"epoch": 0.28580761064837495,
"grad_norm": 0.18911943531852765,
"learning_rate": 4.2791883402674794e-05,
"loss": 0.0735,
"step": 1750
},
{
"epoch": 0.28744079699493713,
"grad_norm": 0.20472268070153826,
"learning_rate": 4.269675688433032e-05,
"loss": 0.0751,
"step": 1760
},
{
"epoch": 0.28907398334149925,
"grad_norm": 0.19152926789870695,
"learning_rate": 4.2601114016702784e-05,
"loss": 0.0745,
"step": 1770
},
{
"epoch": 0.29070716968806143,
"grad_norm": 0.18791435253520175,
"learning_rate": 4.25049575904236e-05,
"loss": 0.0757,
"step": 1780
},
{
"epoch": 0.29234035603462355,
"grad_norm": 0.18920170317167004,
"learning_rate": 4.240829041110865e-05,
"loss": 0.0758,
"step": 1790
},
{
"epoch": 0.29397354238118567,
"grad_norm": 0.2294151779466235,
"learning_rate": 4.231111529927634e-05,
"loss": 0.0732,
"step": 1800
},
{
"epoch": 0.29560672872774785,
"grad_norm": 0.22098437127780643,
"learning_rate": 4.2213435090265354e-05,
"loss": 0.0725,
"step": 1810
},
{
"epoch": 0.29723991507430997,
"grad_norm": 0.18235963750759424,
"learning_rate": 4.2115252634151906e-05,
"loss": 0.0732,
"step": 1820
},
{
"epoch": 0.29887310142087214,
"grad_norm": 0.19186025440050816,
"learning_rate": 4.201657079566659e-05,
"loss": 0.0732,
"step": 1830
},
{
"epoch": 0.30050628776743427,
"grad_norm": 0.19346487311465732,
"learning_rate": 4.191739245411078e-05,
"loss": 0.0706,
"step": 1840
},
{
"epoch": 0.3021394741139964,
"grad_norm": 0.16781546733849126,
"learning_rate": 4.181772050327265e-05,
"loss": 0.0725,
"step": 1850
},
{
"epoch": 0.30377266046055856,
"grad_norm": 0.18317681417047219,
"learning_rate": 4.17175578513427e-05,
"loss": 0.0719,
"step": 1860
},
{
"epoch": 0.3054058468071207,
"grad_norm": 0.17729724134286734,
"learning_rate": 4.161690742082893e-05,
"loss": 0.0719,
"step": 1870
},
{
"epoch": 0.30703903315368286,
"grad_norm": 0.19907664130638136,
"learning_rate": 4.151577214847156e-05,
"loss": 0.0705,
"step": 1880
},
{
"epoch": 0.308672219500245,
"grad_norm": 0.18425449682288675,
"learning_rate": 4.141415498515732e-05,
"loss": 0.0728,
"step": 1890
},
{
"epoch": 0.3103054058468071,
"grad_norm": 0.22372272636270119,
"learning_rate": 4.131205889583341e-05,
"loss": 0.0735,
"step": 1900
},
{
"epoch": 0.3119385921933693,
"grad_norm": 3.2795529491854967,
"learning_rate": 4.1209486859420907e-05,
"loss": 0.1313,
"step": 1910
},
{
"epoch": 0.3135717785399314,
"grad_norm": 2.1759170956944285,
"learning_rate": 4.110644186872795e-05,
"loss": 0.167,
"step": 1920
},
{
"epoch": 0.3152049648864935,
"grad_norm": 20.495153904336536,
"learning_rate": 4.100292693036232e-05,
"loss": 0.1183,
"step": 1930
},
{
"epoch": 0.3168381512330557,
"grad_norm": 0.27114573703810224,
"learning_rate": 4.089894506464374e-05,
"loss": 0.0946,
"step": 1940
},
{
"epoch": 0.3184713375796178,
"grad_norm": 0.2293491438034897,
"learning_rate": 4.079449930551579e-05,
"loss": 0.0764,
"step": 1950
},
{
"epoch": 0.32010452392618,
"grad_norm": 0.4416270695480235,
"learning_rate": 4.068959270045737e-05,
"loss": 0.0815,
"step": 1960
},
{
"epoch": 0.3217377102727421,
"grad_norm": 0.8003415051699548,
"learning_rate": 4.0584228310393726e-05,
"loss": 0.0792,
"step": 1970
},
{
"epoch": 0.32337089661930424,
"grad_norm": 0.23605878133294136,
"learning_rate": 4.047840920960721e-05,
"loss": 0.0785,
"step": 1980
},
{
"epoch": 0.3250040829658664,
"grad_norm": 0.25368460073934873,
"learning_rate": 4.037213848564755e-05,
"loss": 0.0794,
"step": 1990
},
{
"epoch": 0.32663726931242854,
"grad_norm": 0.20878030346362816,
"learning_rate": 4.026541923924175e-05,
"loss": 0.0766,
"step": 2000
},
{
"epoch": 0.32663726931242854,
"eval_loss": 0.07618946582078934,
"eval_runtime": 546.48,
"eval_samples_per_second": 11.067,
"eval_steps_per_second": 0.692,
"step": 2000
},
{
"epoch": 0.3282704556589907,
"grad_norm": 0.2118040356804158,
"learning_rate": 4.015825458420366e-05,
"loss": 0.075,
"step": 2010
},
{
"epoch": 0.32990364200555283,
"grad_norm": 0.23167408945719953,
"learning_rate": 4.005064764734307e-05,
"loss": 0.0734,
"step": 2020
},
{
"epoch": 0.33153682835211495,
"grad_norm": 0.18108982532172427,
"learning_rate": 3.9942601568374485e-05,
"loss": 0.0743,
"step": 2030
},
{
"epoch": 0.33317001469867713,
"grad_norm": 0.18835200914864414,
"learning_rate": 3.9834119499825595e-05,
"loss": 0.0724,
"step": 2040
},
{
"epoch": 0.33480320104523925,
"grad_norm": 0.1745221233943252,
"learning_rate": 3.972520460694516e-05,
"loss": 0.072,
"step": 2050
},
{
"epoch": 0.33643638739180143,
"grad_norm": 0.19466679755892982,
"learning_rate": 3.9615860067610765e-05,
"loss": 0.073,
"step": 2060
},
{
"epoch": 0.33806957373836355,
"grad_norm": 0.20465085954535606,
"learning_rate": 3.9506089072236055e-05,
"loss": 0.07,
"step": 2070
},
{
"epoch": 0.33970276008492567,
"grad_norm": 0.1989786071594349,
"learning_rate": 3.939589482367764e-05,
"loss": 0.0727,
"step": 2080
},
{
"epoch": 0.34133594643148785,
"grad_norm": 0.1866798202275284,
"learning_rate": 3.9285280537141634e-05,
"loss": 0.073,
"step": 2090
},
{
"epoch": 0.34296913277804997,
"grad_norm": 0.17298028095975354,
"learning_rate": 3.9174249440089895e-05,
"loss": 0.0734,
"step": 2100
},
{
"epoch": 0.34460231912461214,
"grad_norm": 0.18501438171644455,
"learning_rate": 3.906280477214579e-05,
"loss": 0.0728,
"step": 2110
},
{
"epoch": 0.34623550547117427,
"grad_norm": 0.21393648416112718,
"learning_rate": 3.8950949784999724e-05,
"loss": 0.0705,
"step": 2120
},
{
"epoch": 0.3478686918177364,
"grad_norm": 0.198375922247382,
"learning_rate": 3.883868774231424e-05,
"loss": 0.0742,
"step": 2130
},
{
"epoch": 0.34950187816429856,
"grad_norm": 0.19619688147392647,
"learning_rate": 3.8726021919628766e-05,
"loss": 0.0714,
"step": 2140
},
{
"epoch": 0.3511350645108607,
"grad_norm": 0.17739705847901396,
"learning_rate": 3.8612955604264104e-05,
"loss": 0.0702,
"step": 2150
},
{
"epoch": 0.3527682508574228,
"grad_norm": 0.1772284299737398,
"learning_rate": 3.849949209522645e-05,
"loss": 0.072,
"step": 2160
},
{
"epoch": 0.354401437203985,
"grad_norm": 0.1869906916346273,
"learning_rate": 3.838563470311119e-05,
"loss": 0.0716,
"step": 2170
},
{
"epoch": 0.3560346235505471,
"grad_norm": 0.17508616518212625,
"learning_rate": 3.8271386750006274e-05,
"loss": 0.0702,
"step": 2180
},
{
"epoch": 0.3576678098971093,
"grad_norm": 0.32221825956274597,
"learning_rate": 3.81567515693953e-05,
"loss": 0.0677,
"step": 2190
},
{
"epoch": 0.3593009962436714,
"grad_norm": 0.20363659025292177,
"learning_rate": 3.804173250606023e-05,
"loss": 0.0697,
"step": 2200
},
{
"epoch": 0.3609341825902335,
"grad_norm": 0.17911680885734255,
"learning_rate": 3.7926332915983805e-05,
"loss": 0.069,
"step": 2210
},
{
"epoch": 0.3625673689367957,
"grad_norm": 0.17113681475424364,
"learning_rate": 3.781055616625166e-05,
"loss": 0.0714,
"step": 2220
},
{
"epoch": 0.3642005552833578,
"grad_norm": 0.16661148091329733,
"learning_rate": 3.7694405634954034e-05,
"loss": 0.0684,
"step": 2230
},
{
"epoch": 0.36583374162992,
"grad_norm": 0.16481369051049344,
"learning_rate": 3.7577884711087216e-05,
"loss": 0.0697,
"step": 2240
},
{
"epoch": 0.3674669279764821,
"grad_norm": 0.17366227474771695,
"learning_rate": 3.7460996794454687e-05,
"loss": 0.0697,
"step": 2250
},
{
"epoch": 0.36910011432304424,
"grad_norm": 0.16074168453264664,
"learning_rate": 3.734374529556789e-05,
"loss": 0.0689,
"step": 2260
},
{
"epoch": 0.3707333006696064,
"grad_norm": 0.17246487237625335,
"learning_rate": 3.722613363554673e-05,
"loss": 0.069,
"step": 2270
},
{
"epoch": 0.37236648701616853,
"grad_norm": 0.16428198944119007,
"learning_rate": 3.71081652460198e-05,
"loss": 0.0712,
"step": 2280
},
{
"epoch": 0.3739996733627307,
"grad_norm": 0.1864287972632789,
"learning_rate": 3.698984356902414e-05,
"loss": 0.0685,
"step": 2290
},
{
"epoch": 0.37563285970929283,
"grad_norm": 0.17069131735021167,
"learning_rate": 3.687117205690494e-05,
"loss": 0.0697,
"step": 2300
},
{
"epoch": 0.37726604605585495,
"grad_norm": 0.18350071694510064,
"learning_rate": 3.6752154172214705e-05,
"loss": 0.0716,
"step": 2310
},
{
"epoch": 0.37889923240241713,
"grad_norm": 0.16635451409707538,
"learning_rate": 3.66327933876123e-05,
"loss": 0.0719,
"step": 2320
},
{
"epoch": 0.38053241874897925,
"grad_norm": 0.16203581726590074,
"learning_rate": 3.651309318576158e-05,
"loss": 0.0699,
"step": 2330
},
{
"epoch": 0.3821656050955414,
"grad_norm": 0.17386948871129598,
"learning_rate": 3.639305705922978e-05,
"loss": 0.0689,
"step": 2340
},
{
"epoch": 0.38379879144210355,
"grad_norm": 0.17823241253435881,
"learning_rate": 3.6272688510385635e-05,
"loss": 0.0696,
"step": 2350
},
{
"epoch": 0.38543197778866567,
"grad_norm": 0.1612558901404048,
"learning_rate": 3.615199105129716e-05,
"loss": 0.0693,
"step": 2360
},
{
"epoch": 0.38706516413522785,
"grad_norm": 0.1998665108410326,
"learning_rate": 3.603096820362919e-05,
"loss": 0.0695,
"step": 2370
},
{
"epoch": 0.38869835048178997,
"grad_norm": 0.16693730449924057,
"learning_rate": 3.590962349854064e-05,
"loss": 0.0695,
"step": 2380
},
{
"epoch": 0.3903315368283521,
"grad_norm": 0.18223635832428517,
"learning_rate": 3.578796047658143e-05,
"loss": 0.0685,
"step": 2390
},
{
"epoch": 0.39196472317491426,
"grad_norm": 0.14730518464437817,
"learning_rate": 3.566598268758924e-05,
"loss": 0.0669,
"step": 2400
},
{
"epoch": 0.3935979095214764,
"grad_norm": 0.17125342490081105,
"learning_rate": 3.554369369058589e-05,
"loss": 0.0686,
"step": 2410
},
{
"epoch": 0.39523109586803856,
"grad_norm": 0.1942368814991241,
"learning_rate": 3.54210970536735e-05,
"loss": 0.0677,
"step": 2420
},
{
"epoch": 0.3968642822146007,
"grad_norm": 0.17868491153926427,
"learning_rate": 3.5298196353930416e-05,
"loss": 0.0676,
"step": 2430
},
{
"epoch": 0.3984974685611628,
"grad_norm": 0.1548907079737275,
"learning_rate": 3.517499517730678e-05,
"loss": 0.0672,
"step": 2440
},
{
"epoch": 0.400130654907725,
"grad_norm": 0.15526071918815793,
"learning_rate": 3.5051497118519974e-05,
"loss": 0.0669,
"step": 2450
},
{
"epoch": 0.4017638412542871,
"grad_norm": 0.15914304652905847,
"learning_rate": 3.492770578094966e-05,
"loss": 0.0672,
"step": 2460
},
{
"epoch": 0.4033970276008493,
"grad_norm": 0.1475358802947237,
"learning_rate": 3.48036247765327e-05,
"loss": 0.0677,
"step": 2470
},
{
"epoch": 0.4050302139474114,
"grad_norm": 0.15770440682045228,
"learning_rate": 3.467925772565774e-05,
"loss": 0.0707,
"step": 2480
},
{
"epoch": 0.4066634002939735,
"grad_norm": 0.1670524982695031,
"learning_rate": 3.4554608257059576e-05,
"loss": 0.0689,
"step": 2490
},
{
"epoch": 0.4082965866405357,
"grad_norm": 0.17230407977910778,
"learning_rate": 3.442968000771329e-05,
"loss": 0.0676,
"step": 2500
},
{
"epoch": 0.4099297729870978,
"grad_norm": 0.1662305093737705,
"learning_rate": 3.430447662272811e-05,
"loss": 0.0686,
"step": 2510
},
{
"epoch": 0.41156295933366,
"grad_norm": 0.15536263881649395,
"learning_rate": 3.417900175524109e-05,
"loss": 0.0687,
"step": 2520
},
{
"epoch": 0.4131961456802221,
"grad_norm": 0.1470849185615423,
"learning_rate": 3.405325906631048e-05,
"loss": 0.0659,
"step": 2530
},
{
"epoch": 0.41482933202678424,
"grad_norm": 0.17103058386957173,
"learning_rate": 3.392725222480891e-05,
"loss": 0.0664,
"step": 2540
},
{
"epoch": 0.4164625183733464,
"grad_norm": 0.16641775909066878,
"learning_rate": 3.3800984907316404e-05,
"loss": 0.0688,
"step": 2550
},
{
"epoch": 0.41809570471990853,
"grad_norm": 0.17294586249257893,
"learning_rate": 3.3674460798013004e-05,
"loss": 0.0682,
"step": 2560
},
{
"epoch": 0.4197288910664707,
"grad_norm": 0.17968489651382255,
"learning_rate": 3.3547683588571344e-05,
"loss": 0.0659,
"step": 2570
},
{
"epoch": 0.42136207741303283,
"grad_norm": 0.16041005749667095,
"learning_rate": 3.342065697804892e-05,
"loss": 0.0655,
"step": 2580
},
{
"epoch": 0.42299526375959495,
"grad_norm": 0.20578138721659112,
"learning_rate": 3.329338467278017e-05,
"loss": 0.0683,
"step": 2590
},
{
"epoch": 0.42462845010615713,
"grad_norm": 0.15112983660887272,
"learning_rate": 3.31658703862683e-05,
"loss": 0.0655,
"step": 2600
},
{
"epoch": 0.42626163645271925,
"grad_norm": 0.2065576108641922,
"learning_rate": 3.3038117839076954e-05,
"loss": 0.0674,
"step": 2610
},
{
"epoch": 0.42789482279928137,
"grad_norm": 0.15949477874268886,
"learning_rate": 3.291013075872168e-05,
"loss": 0.0658,
"step": 2620
},
{
"epoch": 0.42952800914584355,
"grad_norm": 0.14668997404113893,
"learning_rate": 3.2781912879561125e-05,
"loss": 0.0652,
"step": 2630
},
{
"epoch": 0.43116119549240567,
"grad_norm": 0.16937137291290755,
"learning_rate": 3.265346794268809e-05,
"loss": 0.0668,
"step": 2640
},
{
"epoch": 0.43279438183896785,
"grad_norm": 0.1620844258263254,
"learning_rate": 3.252479969582041e-05,
"loss": 0.0669,
"step": 2650
},
{
"epoch": 0.43442756818552997,
"grad_norm": 0.15335706393425183,
"learning_rate": 3.2395911893191536e-05,
"loss": 0.0672,
"step": 2660
},
{
"epoch": 0.4360607545320921,
"grad_norm": 0.18456990420937391,
"learning_rate": 3.226680829544108e-05,
"loss": 0.0666,
"step": 2670
},
{
"epoch": 0.43769394087865426,
"grad_norm": 0.1632988305948867,
"learning_rate": 3.213749266950499e-05,
"loss": 0.0656,
"step": 2680
},
{
"epoch": 0.4393271272252164,
"grad_norm": 0.15752563421624427,
"learning_rate": 3.200796878850574e-05,
"loss": 0.0661,
"step": 2690
},
{
"epoch": 0.44096031357177856,
"grad_norm": 0.19844882135201591,
"learning_rate": 3.1878240431642154e-05,
"loss": 0.0666,
"step": 2700
},
{
"epoch": 0.4425934999183407,
"grad_norm": 0.1730140755168181,
"learning_rate": 3.1748311384079195e-05,
"loss": 0.0675,
"step": 2710
},
{
"epoch": 0.4442266862649028,
"grad_norm": 0.15956060007262066,
"learning_rate": 3.161818543683751e-05,
"loss": 0.0658,
"step": 2720
},
{
"epoch": 0.445859872611465,
"grad_norm": 0.18151986570145462,
"learning_rate": 3.1487866386682786e-05,
"loss": 0.0673,
"step": 2730
},
{
"epoch": 0.4474930589580271,
"grad_norm": 0.1674221615790595,
"learning_rate": 3.1357358036015025e-05,
"loss": 0.0664,
"step": 2740
},
{
"epoch": 0.4491262453045893,
"grad_norm": 0.16858339287890442,
"learning_rate": 3.1226664192757545e-05,
"loss": 0.066,
"step": 2750
},
{
"epoch": 0.4507594316511514,
"grad_norm": 0.15937279780570335,
"learning_rate": 3.109578867024589e-05,
"loss": 0.0654,
"step": 2760
},
{
"epoch": 0.4523926179977135,
"grad_norm": 0.1615797662602226,
"learning_rate": 3.0964735287116604e-05,
"loss": 0.0662,
"step": 2770
},
{
"epoch": 0.4540258043442757,
"grad_norm": 0.16129197340752968,
"learning_rate": 3.083350786719575e-05,
"loss": 0.0689,
"step": 2780
},
{
"epoch": 0.4556589906908378,
"grad_norm": 0.15789137332520725,
"learning_rate": 3.0702110239387394e-05,
"loss": 0.0646,
"step": 2790
},
{
"epoch": 0.4572921770374,
"grad_norm": 0.15491751303807894,
"learning_rate": 3.0570546237561836e-05,
"loss": 0.0662,
"step": 2800
},
{
"epoch": 0.4589253633839621,
"grad_norm": 0.16085385949586323,
"learning_rate": 3.0438819700443805e-05,
"loss": 0.0647,
"step": 2810
},
{
"epoch": 0.46055854973052424,
"grad_norm": 0.1513816140716184,
"learning_rate": 3.0306934471500408e-05,
"loss": 0.0661,
"step": 2820
},
{
"epoch": 0.4621917360770864,
"grad_norm": 0.17156042835851865,
"learning_rate": 3.017489439882899e-05,
"loss": 0.0676,
"step": 2830
},
{
"epoch": 0.46382492242364853,
"grad_norm": 0.156470849204471,
"learning_rate": 3.0042703335044903e-05,
"loss": 0.0645,
"step": 2840
},
{
"epoch": 0.46545810877021065,
"grad_norm": 0.15662861670187953,
"learning_rate": 2.9910365137169015e-05,
"loss": 0.0652,
"step": 2850
},
{
"epoch": 0.46709129511677283,
"grad_norm": 0.1458907521549837,
"learning_rate": 2.977788366651526e-05,
"loss": 0.0645,
"step": 2860
},
{
"epoch": 0.46872448146333495,
"grad_norm": 0.14968819358834828,
"learning_rate": 2.9645262788577897e-05,
"loss": 0.0649,
"step": 2870
},
{
"epoch": 0.47035766780989713,
"grad_norm": 0.16239239238357633,
"learning_rate": 2.9512506372918787e-05,
"loss": 0.0658,
"step": 2880
},
{
"epoch": 0.47199085415645925,
"grad_norm": 0.16153443829874786,
"learning_rate": 2.9379618293054428e-05,
"loss": 0.0674,
"step": 2890
},
{
"epoch": 0.47362404050302137,
"grad_norm": 0.14506335232836579,
"learning_rate": 2.9246602426342983e-05,
"loss": 0.0654,
"step": 2900
},
{
"epoch": 0.47525722684958355,
"grad_norm": 0.21112748856920222,
"learning_rate": 2.9113462653871154e-05,
"loss": 0.0652,
"step": 2910
},
{
"epoch": 0.47689041319614567,
"grad_norm": 0.16608513871804156,
"learning_rate": 2.898020286034087e-05,
"loss": 0.0657,
"step": 2920
},
{
"epoch": 0.47852359954270784,
"grad_norm": 0.19450860779048604,
"learning_rate": 2.8846826933956035e-05,
"loss": 0.0649,
"step": 2930
},
{
"epoch": 0.48015678588926997,
"grad_norm": 0.14683068668245297,
"learning_rate": 2.8713338766309e-05,
"loss": 0.0644,
"step": 2940
},
{
"epoch": 0.4817899722358321,
"grad_norm": 0.25474565288997614,
"learning_rate": 2.8579742252267067e-05,
"loss": 0.0665,
"step": 2950
},
{
"epoch": 0.48342315858239426,
"grad_norm": 0.14739524186632513,
"learning_rate": 2.8446041289858826e-05,
"loss": 0.0626,
"step": 2960
},
{
"epoch": 0.4850563449289564,
"grad_norm": 0.15780995045090349,
"learning_rate": 2.8312239780160414e-05,
"loss": 0.0649,
"step": 2970
},
{
"epoch": 0.48668953127551856,
"grad_norm": 0.1610087197823336,
"learning_rate": 2.8178341627181714e-05,
"loss": 0.0653,
"step": 2980
},
{
"epoch": 0.4883227176220807,
"grad_norm": 0.15569888331898027,
"learning_rate": 2.8044350737752418e-05,
"loss": 0.064,
"step": 2990
},
{
"epoch": 0.4899559039686428,
"grad_norm": 0.157473800823855,
"learning_rate": 2.791027102140804e-05,
"loss": 0.0628,
"step": 3000
},
{
"epoch": 0.4899559039686428,
"eval_loss": 0.06539792567491531,
"eval_runtime": 544.9481,
"eval_samples_per_second": 11.098,
"eval_steps_per_second": 0.694,
"step": 3000
},
{
"epoch": 0.491589090315205,
"grad_norm": 0.14483631317903772,
"learning_rate": 2.7776106390275875e-05,
"loss": 0.0647,
"step": 3010
},
{
"epoch": 0.4932222766617671,
"grad_norm": 0.15002701914108177,
"learning_rate": 2.7641860758960804e-05,
"loss": 0.0653,
"step": 3020
},
{
"epoch": 0.4948554630083293,
"grad_norm": 0.15372927314782364,
"learning_rate": 2.7507538044431125e-05,
"loss": 0.0656,
"step": 3030
},
{
"epoch": 0.4964886493548914,
"grad_norm": 0.15276783280802592,
"learning_rate": 2.737314216590422e-05,
"loss": 0.0643,
"step": 3040
},
{
"epoch": 0.4981218357014535,
"grad_norm": 0.15067228131913446,
"learning_rate": 2.723867704473224e-05,
"loss": 0.0679,
"step": 3050
},
{
"epoch": 0.4997550220480157,
"grad_norm": 0.16823849990730033,
"learning_rate": 2.7104146604287663e-05,
"loss": 0.0647,
"step": 3060
},
{
"epoch": 0.5013882083945779,
"grad_norm": 0.15649373838210812,
"learning_rate": 2.696955476984881e-05,
"loss": 0.0626,
"step": 3070
},
{
"epoch": 0.5030213947411399,
"grad_norm": 0.1411695010499073,
"learning_rate": 2.6834905468485378e-05,
"loss": 0.0636,
"step": 3080
},
{
"epoch": 0.5046545810877021,
"grad_norm": 0.13366924762432433,
"learning_rate": 2.670020262894377e-05,
"loss": 0.0632,
"step": 3090
},
{
"epoch": 0.5062877674342643,
"grad_norm": 0.15068906105243862,
"learning_rate": 2.6565450181532525e-05,
"loss": 0.0645,
"step": 3100
},
{
"epoch": 0.5079209537808264,
"grad_norm": 0.45782370382951143,
"learning_rate": 2.6430652058007614e-05,
"loss": 0.0651,
"step": 3110
},
{
"epoch": 0.5095541401273885,
"grad_norm": 0.14741596252767994,
"learning_rate": 2.629581219145773e-05,
"loss": 0.0637,
"step": 3120
},
{
"epoch": 0.5111873264739507,
"grad_norm": 0.15694876158336823,
"learning_rate": 2.6160934516189544e-05,
"loss": 0.0657,
"step": 3130
},
{
"epoch": 0.5128205128205128,
"grad_norm": 0.17181188088763552,
"learning_rate": 2.6026022967612863e-05,
"loss": 0.066,
"step": 3140
},
{
"epoch": 0.514453699167075,
"grad_norm": 0.15342890421496308,
"learning_rate": 2.5891081482125873e-05,
"loss": 0.0639,
"step": 3150
},
{
"epoch": 0.5160868855136371,
"grad_norm": 0.15754829015156574,
"learning_rate": 2.575611399700022e-05,
"loss": 0.0659,
"step": 3160
},
{
"epoch": 0.5177200718601992,
"grad_norm": 0.15592010937213382,
"learning_rate": 2.562112445026617e-05,
"loss": 0.0617,
"step": 3170
},
{
"epoch": 0.5193532582067614,
"grad_norm": 0.15046260699905864,
"learning_rate": 2.548611678059769e-05,
"loss": 0.0628,
"step": 3180
},
{
"epoch": 0.5209864445533235,
"grad_norm": 0.14388701046840682,
"learning_rate": 2.535109492719754e-05,
"loss": 0.0638,
"step": 3190
},
{
"epoch": 0.5226196308998857,
"grad_norm": 0.18329460127035965,
"learning_rate": 2.5216062829682314e-05,
"loss": 0.0638,
"step": 3200
},
{
"epoch": 0.5242528172464478,
"grad_norm": 0.16670771434324538,
"learning_rate": 2.508102442796752e-05,
"loss": 0.0622,
"step": 3210
},
{
"epoch": 0.52588600359301,
"grad_norm": 0.1479705984123943,
"learning_rate": 2.494598366215259e-05,
"loss": 0.063,
"step": 3220
},
{
"epoch": 0.5275191899395721,
"grad_norm": 0.13883410867284673,
"learning_rate": 2.481094447240594e-05,
"loss": 0.0643,
"step": 3230
},
{
"epoch": 0.5291523762861342,
"grad_norm": 0.16886968848731043,
"learning_rate": 2.4675910798850023e-05,
"loss": 0.0637,
"step": 3240
},
{
"epoch": 0.5307855626326964,
"grad_norm": 0.15301035875326577,
"learning_rate": 2.4540886581446304e-05,
"loss": 0.0637,
"step": 3250
},
{
"epoch": 0.5324187489792586,
"grad_norm": 0.15996519192251393,
"learning_rate": 2.4405875759880377e-05,
"loss": 0.0657,
"step": 3260
},
{
"epoch": 0.5340519353258206,
"grad_norm": 0.15120638113557416,
"learning_rate": 2.4270882273446954e-05,
"loss": 0.0663,
"step": 3270
},
{
"epoch": 0.5356851216723828,
"grad_norm": 0.14754255280121978,
"learning_rate": 2.4135910060934945e-05,
"loss": 0.0627,
"step": 3280
},
{
"epoch": 0.537318308018945,
"grad_norm": 0.1469060743302255,
"learning_rate": 2.400096306051256e-05,
"loss": 0.0651,
"step": 3290
},
{
"epoch": 0.5389514943655072,
"grad_norm": 0.15648512672617107,
"learning_rate": 2.3866045209612358e-05,
"loss": 0.064,
"step": 3300
},
{
"epoch": 0.5405846807120692,
"grad_norm": 0.18874979991649526,
"learning_rate": 2.3731160444816406e-05,
"loss": 0.0631,
"step": 3310
},
{
"epoch": 0.5422178670586314,
"grad_norm": 0.14208973129907518,
"learning_rate": 2.3596312701741375e-05,
"loss": 0.0631,
"step": 3320
},
{
"epoch": 0.5438510534051936,
"grad_norm": 0.13438426748692756,
"learning_rate": 2.346150591492375e-05,
"loss": 0.0616,
"step": 3330
},
{
"epoch": 0.5454842397517556,
"grad_norm": 0.14444760891909095,
"learning_rate": 2.3326744017704986e-05,
"loss": 0.0621,
"step": 3340
},
{
"epoch": 0.5471174260983178,
"grad_norm": 0.14971466204126063,
"learning_rate": 2.3192030942116803e-05,
"loss": 0.0642,
"step": 3350
},
{
"epoch": 0.54875061244488,
"grad_norm": 0.14682006171484224,
"learning_rate": 2.3057370618766372e-05,
"loss": 0.062,
"step": 3360
},
{
"epoch": 0.5503837987914421,
"grad_norm": 0.16215277932423075,
"learning_rate": 2.2922766976721735e-05,
"loss": 0.062,
"step": 3370
},
{
"epoch": 0.5520169851380042,
"grad_norm": 0.14464775764379165,
"learning_rate": 2.2788223943397063e-05,
"loss": 0.0607,
"step": 3380
},
{
"epoch": 0.5536501714845664,
"grad_norm": 0.13563139600954116,
"learning_rate": 2.2653745444438122e-05,
"loss": 0.0622,
"step": 3390
},
{
"epoch": 0.5552833578311285,
"grad_norm": 0.16175863973370094,
"learning_rate": 2.2519335403607735e-05,
"loss": 0.0613,
"step": 3400
},
{
"epoch": 0.5569165441776907,
"grad_norm": 0.1779116341480311,
"learning_rate": 2.2384997742671256e-05,
"loss": 0.0634,
"step": 3410
},
{
"epoch": 0.5585497305242528,
"grad_norm": 0.16619950051124655,
"learning_rate": 2.2250736381282193e-05,
"loss": 0.0637,
"step": 3420
},
{
"epoch": 0.560182916870815,
"grad_norm": 0.15877432410306344,
"learning_rate": 2.2116555236867774e-05,
"loss": 0.0636,
"step": 3430
},
{
"epoch": 0.5618161032173771,
"grad_norm": 0.14167827719580398,
"learning_rate": 2.1982458224514736e-05,
"loss": 0.0626,
"step": 3440
},
{
"epoch": 0.5634492895639392,
"grad_norm": 0.15060286900862885,
"learning_rate": 2.1848449256854996e-05,
"loss": 0.0635,
"step": 3450
},
{
"epoch": 0.5650824759105014,
"grad_norm": 0.13461936290544277,
"learning_rate": 2.171453224395157e-05,
"loss": 0.0635,
"step": 3460
},
{
"epoch": 0.5667156622570635,
"grad_norm": 0.15810605736412126,
"learning_rate": 2.1580711093184437e-05,
"loss": 0.0631,
"step": 3470
},
{
"epoch": 0.5683488486036257,
"grad_norm": 0.13861141529655402,
"learning_rate": 2.1446989709136552e-05,
"loss": 0.0626,
"step": 3480
},
{
"epoch": 0.5699820349501878,
"grad_norm": 0.14740542187926117,
"learning_rate": 2.1313371993479918e-05,
"loss": 0.0647,
"step": 3490
},
{
"epoch": 0.5716152212967499,
"grad_norm": 0.1417322296880048,
"learning_rate": 2.1179861844861714e-05,
"loss": 0.0612,
"step": 3500
},
{
"epoch": 0.5732484076433121,
"grad_norm": 0.14705465120695888,
"learning_rate": 2.104646315879062e-05,
"loss": 0.0623,
"step": 3510
},
{
"epoch": 0.5748815939898743,
"grad_norm": 0.20247252344579117,
"learning_rate": 2.091317982752305e-05,
"loss": 0.0622,
"step": 3520
},
{
"epoch": 0.5765147803364364,
"grad_norm": 0.15082363484552333,
"learning_rate": 2.0780015739949675e-05,
"loss": 0.0623,
"step": 3530
},
{
"epoch": 0.5781479666829985,
"grad_norm": 0.1293200383908178,
"learning_rate": 2.06469747814819e-05,
"loss": 0.0628,
"step": 3540
},
{
"epoch": 0.5797811530295607,
"grad_norm": 0.15197046783693255,
"learning_rate": 2.0514060833938534e-05,
"loss": 0.0617,
"step": 3550
},
{
"epoch": 0.5814143393761229,
"grad_norm": 0.1282930439776445,
"learning_rate": 2.0381277775432484e-05,
"loss": 0.0623,
"step": 3560
},
{
"epoch": 0.5830475257226849,
"grad_norm": 0.14611054125101627,
"learning_rate": 2.0248629480257653e-05,
"loss": 0.0624,
"step": 3570
},
{
"epoch": 0.5846807120692471,
"grad_norm": 0.13878095055849668,
"learning_rate": 2.0116119818775837e-05,
"loss": 0.0622,
"step": 3580
},
{
"epoch": 0.5863138984158093,
"grad_norm": 0.14675779526022883,
"learning_rate": 1.9983752657303878e-05,
"loss": 0.0612,
"step": 3590
},
{
"epoch": 0.5879470847623713,
"grad_norm": 0.15825452250584593,
"learning_rate": 1.985153185800076e-05,
"loss": 0.0625,
"step": 3600
},
{
"epoch": 0.5895802711089335,
"grad_norm": 0.14523056878441204,
"learning_rate": 1.9719461278754982e-05,
"loss": 0.0609,
"step": 3610
},
{
"epoch": 0.5912134574554957,
"grad_norm": 0.14877798464966374,
"learning_rate": 1.9587544773072004e-05,
"loss": 0.0623,
"step": 3620
},
{
"epoch": 0.5928466438020578,
"grad_norm": 0.14523688812387658,
"learning_rate": 1.945578618996174e-05,
"loss": 0.0618,
"step": 3630
},
{
"epoch": 0.5944798301486199,
"grad_norm": 0.15585399103619635,
"learning_rate": 1.9324189373826353e-05,
"loss": 0.0604,
"step": 3640
},
{
"epoch": 0.5961130164951821,
"grad_norm": 0.1759354825245184,
"learning_rate": 1.919275816434797e-05,
"loss": 0.0625,
"step": 3650
},
{
"epoch": 0.5977462028417443,
"grad_norm": 0.14662108513040858,
"learning_rate": 1.9061496396376765e-05,
"loss": 0.0628,
"step": 3660
},
{
"epoch": 0.5993793891883064,
"grad_norm": 0.13485929645248515,
"learning_rate": 1.8930407899818973e-05,
"loss": 0.0603,
"step": 3670
},
{
"epoch": 0.6010125755348685,
"grad_norm": 0.13879869149008003,
"learning_rate": 1.8799496499525203e-05,
"loss": 0.0618,
"step": 3680
},
{
"epoch": 0.6026457618814307,
"grad_norm": 0.1490006143330652,
"learning_rate": 1.86687660151788e-05,
"loss": 0.061,
"step": 3690
},
{
"epoch": 0.6042789482279928,
"grad_norm": 0.1986985234105705,
"learning_rate": 1.8538220261184435e-05,
"loss": 0.0603,
"step": 3700
},
{
"epoch": 0.605912134574555,
"grad_norm": 0.1335003410988764,
"learning_rate": 1.8407863046556772e-05,
"loss": 0.0617,
"step": 3710
},
{
"epoch": 0.6075453209211171,
"grad_norm": 0.14607058326958405,
"learning_rate": 1.8277698174809322e-05,
"loss": 0.0621,
"step": 3720
},
{
"epoch": 0.6091785072676792,
"grad_norm": 0.1437268280141882,
"learning_rate": 1.8147729443843538e-05,
"loss": 0.0611,
"step": 3730
},
{
"epoch": 0.6108116936142414,
"grad_norm": 0.13900887847494084,
"learning_rate": 1.8017960645837906e-05,
"loss": 0.0605,
"step": 3740
},
{
"epoch": 0.6124448799608035,
"grad_norm": 0.12914279154010508,
"learning_rate": 1.7888395567137374e-05,
"loss": 0.0604,
"step": 3750
},
{
"epoch": 0.6140780663073657,
"grad_norm": 0.17891531756072188,
"learning_rate": 1.7759037988142824e-05,
"loss": 0.0626,
"step": 3760
},
{
"epoch": 0.6157112526539278,
"grad_norm": 0.17165013766562734,
"learning_rate": 1.7629891683200804e-05,
"loss": 0.0636,
"step": 3770
},
{
"epoch": 0.61734443900049,
"grad_norm": 0.13966562040029765,
"learning_rate": 1.7500960420493384e-05,
"loss": 0.0602,
"step": 3780
},
{
"epoch": 0.6189776253470521,
"grad_norm": 0.14105273362926696,
"learning_rate": 1.737224796192821e-05,
"loss": 0.0623,
"step": 3790
},
{
"epoch": 0.6206108116936142,
"grad_norm": 0.1393512729626591,
"learning_rate": 1.724375806302875e-05,
"loss": 0.0591,
"step": 3800
},
{
"epoch": 0.6222439980401764,
"grad_norm": 0.13985026080803073,
"learning_rate": 1.7115494472824707e-05,
"loss": 0.0613,
"step": 3810
},
{
"epoch": 0.6238771843867386,
"grad_norm": 0.13791849756072955,
"learning_rate": 1.698746093374262e-05,
"loss": 0.0602,
"step": 3820
},
{
"epoch": 0.6255103707333006,
"grad_norm": 0.13385763382500676,
"learning_rate": 1.68596611814967e-05,
"loss": 0.0588,
"step": 3830
},
{
"epoch": 0.6271435570798628,
"grad_norm": 0.14715865761931357,
"learning_rate": 1.673209894497982e-05,
"loss": 0.0616,
"step": 3840
},
{
"epoch": 0.628776743426425,
"grad_norm": 0.12797307459385493,
"learning_rate": 1.6604777946154682e-05,
"loss": 0.0599,
"step": 3850
},
{
"epoch": 0.630409929772987,
"grad_norm": 0.14532549382206625,
"learning_rate": 1.6477701899945282e-05,
"loss": 0.0623,
"step": 3860
},
{
"epoch": 0.6320431161195492,
"grad_norm": 0.12850565932372027,
"learning_rate": 1.6350874514128452e-05,
"loss": 0.0605,
"step": 3870
},
{
"epoch": 0.6336763024661114,
"grad_norm": 0.15151076931911378,
"learning_rate": 1.6224299489225726e-05,
"loss": 0.0607,
"step": 3880
},
{
"epoch": 0.6353094888126736,
"grad_norm": 0.15193327860012007,
"learning_rate": 1.609798051839531e-05,
"loss": 0.0618,
"step": 3890
},
{
"epoch": 0.6369426751592356,
"grad_norm": 0.13332656238343848,
"learning_rate": 1.597192128732442e-05,
"loss": 0.0602,
"step": 3900
},
{
"epoch": 0.6385758615057978,
"grad_norm": 0.1516588781561428,
"learning_rate": 1.5846125474121626e-05,
"loss": 0.059,
"step": 3910
},
{
"epoch": 0.64020904785236,
"grad_norm": 0.15877304886527455,
"learning_rate": 1.5720596749209626e-05,
"loss": 0.059,
"step": 3920
},
{
"epoch": 0.6418422341989221,
"grad_norm": 0.14881052217278215,
"learning_rate": 1.5595338775218106e-05,
"loss": 0.0607,
"step": 3930
},
{
"epoch": 0.6434754205454842,
"grad_norm": 0.1515722356835618,
"learning_rate": 1.5470355206876874e-05,
"loss": 0.0596,
"step": 3940
},
{
"epoch": 0.6451086068920464,
"grad_norm": 0.13392034267210046,
"learning_rate": 1.5345649690909262e-05,
"loss": 0.0611,
"step": 3950
},
{
"epoch": 0.6467417932386085,
"grad_norm": 0.14879412974569295,
"learning_rate": 1.522122586592565e-05,
"loss": 0.0595,
"step": 3960
},
{
"epoch": 0.6483749795851707,
"grad_norm": 0.15520856525461257,
"learning_rate": 1.5097087362317407e-05,
"loss": 0.0602,
"step": 3970
},
{
"epoch": 0.6500081659317328,
"grad_norm": 0.14160057027868986,
"learning_rate": 1.4973237802150835e-05,
"loss": 0.0597,
"step": 3980
},
{
"epoch": 0.651641352278295,
"grad_norm": 0.1685161116050833,
"learning_rate": 1.4849680799061607e-05,
"loss": 0.059,
"step": 3990
},
{
"epoch": 0.6532745386248571,
"grad_norm": 0.13506541156960838,
"learning_rate": 1.4726419958149244e-05,
"loss": 0.0594,
"step": 4000
},
{
"epoch": 0.6532745386248571,
"eval_loss": 0.060191765427589417,
"eval_runtime": 545.963,
"eval_samples_per_second": 11.078,
"eval_steps_per_second": 0.692,
"step": 4000
},
{
"epoch": 0.6549077249714192,
"grad_norm": 0.13904522776798348,
"learning_rate": 1.4603458875871975e-05,
"loss": 0.0573,
"step": 4010
},
{
"epoch": 0.6565409113179814,
"grad_norm": 0.1501622326545471,
"learning_rate": 1.4480801139941786e-05,
"loss": 0.0612,
"step": 4020
},
{
"epoch": 0.6581740976645435,
"grad_norm": 0.1469599359733234,
"learning_rate": 1.4358450329219709e-05,
"loss": 0.059,
"step": 4030
},
{
"epoch": 0.6598072840111057,
"grad_norm": 0.1324817837267378,
"learning_rate": 1.423641001361148e-05,
"loss": 0.0578,
"step": 4040
},
{
"epoch": 0.6614404703576678,
"grad_norm": 0.1426713095714293,
"learning_rate": 1.4114683753963277e-05,
"loss": 0.0582,
"step": 4050
},
{
"epoch": 0.6630736567042299,
"grad_norm": 0.13129457791392557,
"learning_rate": 1.3993275101957903e-05,
"loss": 0.0618,
"step": 4060
},
{
"epoch": 0.6647068430507921,
"grad_norm": 0.13722862401738084,
"learning_rate": 1.38721876000111e-05,
"loss": 0.0603,
"step": 4070
},
{
"epoch": 0.6663400293973543,
"grad_norm": 0.187664541313148,
"learning_rate": 1.3751424781168259e-05,
"loss": 0.0619,
"step": 4080
},
{
"epoch": 0.6679732157439163,
"grad_norm": 0.15387635046918335,
"learning_rate": 1.363099016900122e-05,
"loss": 0.0591,
"step": 4090
},
{
"epoch": 0.6696064020904785,
"grad_norm": 0.1378619425498789,
"learning_rate": 1.3510887277505608e-05,
"loss": 0.0587,
"step": 4100
},
{
"epoch": 0.6712395884370407,
"grad_norm": 0.12875718984341672,
"learning_rate": 1.339111961099817e-05,
"loss": 0.0585,
"step": 4110
},
{
"epoch": 0.6728727747836029,
"grad_norm": 0.1280444688661344,
"learning_rate": 1.3271690664014624e-05,
"loss": 0.0577,
"step": 4120
},
{
"epoch": 0.6745059611301649,
"grad_norm": 0.14167330016406546,
"learning_rate": 1.3152603921207633e-05,
"loss": 0.0589,
"step": 4130
},
{
"epoch": 0.6761391474767271,
"grad_norm": 0.13213187140702354,
"learning_rate": 1.3033862857245153e-05,
"loss": 0.0584,
"step": 4140
},
{
"epoch": 0.6777723338232893,
"grad_norm": 0.13969107814034695,
"learning_rate": 1.2915470936709096e-05,
"loss": 0.0608,
"step": 4150
},
{
"epoch": 0.6794055201698513,
"grad_norm": 0.1351241247827593,
"learning_rate": 1.279743161399413e-05,
"loss": 0.0581,
"step": 4160
},
{
"epoch": 0.6810387065164135,
"grad_norm": 0.14003735685257,
"learning_rate": 1.267974833320702e-05,
"loss": 0.0579,
"step": 4170
},
{
"epoch": 0.6826718928629757,
"grad_norm": 0.14470175277105873,
"learning_rate": 1.2562424528066048e-05,
"loss": 0.0592,
"step": 4180
},
{
"epoch": 0.6843050792095378,
"grad_norm": 0.14089085546993893,
"learning_rate": 1.2445463621800842e-05,
"loss": 0.0603,
"step": 4190
},
{
"epoch": 0.6859382655560999,
"grad_norm": 0.1408806560886765,
"learning_rate": 1.2328869027052511e-05,
"loss": 0.0585,
"step": 4200
},
{
"epoch": 0.6875714519026621,
"grad_norm": 0.14459110138052367,
"learning_rate": 1.2212644145774088e-05,
"loss": 0.0592,
"step": 4210
},
{
"epoch": 0.6892046382492243,
"grad_norm": 0.13725490020979408,
"learning_rate": 1.2096792369131213e-05,
"loss": 0.0589,
"step": 4220
},
{
"epoch": 0.6908378245957864,
"grad_norm": 0.11816453077448706,
"learning_rate": 1.1981317077403238e-05,
"loss": 0.058,
"step": 4230
},
{
"epoch": 0.6924710109423485,
"grad_norm": 0.1394062892473191,
"learning_rate": 1.186622163988457e-05,
"loss": 0.0575,
"step": 4240
},
{
"epoch": 0.6941041972889107,
"grad_norm": 0.13335638090583435,
"learning_rate": 1.1751509414786385e-05,
"loss": 0.059,
"step": 4250
},
{
"epoch": 0.6957373836354728,
"grad_norm": 0.29523046660399904,
"learning_rate": 1.1637183749138622e-05,
"loss": 0.0587,
"step": 4260
},
{
"epoch": 0.697370569982035,
"grad_norm": 0.12759536025073298,
"learning_rate": 1.152324797869233e-05,
"loss": 0.0564,
"step": 4270
},
{
"epoch": 0.6990037563285971,
"grad_norm": 0.13723674533510374,
"learning_rate": 1.1409705427822385e-05,
"loss": 0.0584,
"step": 4280
},
{
"epoch": 0.7006369426751592,
"grad_norm": 0.14507774593708098,
"learning_rate": 1.1296559409430374e-05,
"loss": 0.058,
"step": 4290
},
{
"epoch": 0.7022701290217214,
"grad_norm": 0.12738032638190497,
"learning_rate": 1.1183813224848083e-05,
"loss": 0.0579,
"step": 4300
},
{
"epoch": 0.7039033153682835,
"grad_norm": 0.12716135802996653,
"learning_rate": 1.107147016374106e-05,
"loss": 0.057,
"step": 4310
},
{
"epoch": 0.7055365017148456,
"grad_norm": 0.1431477422037608,
"learning_rate": 1.0959533504012675e-05,
"loss": 0.0595,
"step": 4320
},
{
"epoch": 0.7071696880614078,
"grad_norm": 0.14689233124089865,
"learning_rate": 1.0848006511708467e-05,
"loss": 0.0586,
"step": 4330
},
{
"epoch": 0.70880287440797,
"grad_norm": 0.14728650777752558,
"learning_rate": 1.0736892440920877e-05,
"loss": 0.0574,
"step": 4340
},
{
"epoch": 0.7104360607545321,
"grad_norm": 0.130847734619129,
"learning_rate": 1.062619453369426e-05,
"loss": 0.056,
"step": 4350
},
{
"epoch": 0.7120692471010942,
"grad_norm": 0.1661109840372311,
"learning_rate": 1.0515916019930288e-05,
"loss": 0.0579,
"step": 4360
},
{
"epoch": 0.7137024334476564,
"grad_norm": 0.13632756104038768,
"learning_rate": 1.0406060117293776e-05,
"loss": 0.0588,
"step": 4370
},
{
"epoch": 0.7153356197942186,
"grad_norm": 0.13515423558133835,
"learning_rate": 1.0296630031118714e-05,
"loss": 0.0563,
"step": 4380
},
{
"epoch": 0.7169688061407806,
"grad_norm": 0.12889792085200039,
"learning_rate": 1.0187628954314795e-05,
"loss": 0.0585,
"step": 4390
},
{
"epoch": 0.7186019924873428,
"grad_norm": 0.20413185019504984,
"learning_rate": 1.0079060067274228e-05,
"loss": 0.0579,
"step": 4400
},
{
"epoch": 0.720235178833905,
"grad_norm": 0.15195883603252835,
"learning_rate": 9.970926537778976e-06,
"loss": 0.0572,
"step": 4410
},
{
"epoch": 0.721868365180467,
"grad_norm": 0.13385010464663508,
"learning_rate": 9.863231520908282e-06,
"loss": 0.0576,
"step": 4420
},
{
"epoch": 0.7235015515270292,
"grad_norm": 0.13472780172806798,
"learning_rate": 9.755978158946635e-06,
"loss": 0.058,
"step": 4430
},
{
"epoch": 0.7251347378735914,
"grad_norm": 0.13561031723027941,
"learning_rate": 9.649169581292087e-06,
"loss": 0.0574,
"step": 4440
},
{
"epoch": 0.7267679242201536,
"grad_norm": 0.1349785072195776,
"learning_rate": 9.542808904364963e-06,
"loss": 0.0574,
"step": 4450
},
{
"epoch": 0.7284011105667156,
"grad_norm": 0.13755427558600514,
"learning_rate": 9.436899231516871e-06,
"loss": 0.0566,
"step": 4460
},
{
"epoch": 0.7300342969132778,
"grad_norm": 0.13145232361678383,
"learning_rate": 9.331443652940209e-06,
"loss": 0.0584,
"step": 4470
},
{
"epoch": 0.73166748325984,
"grad_norm": 0.1698156003198985,
"learning_rate": 9.226445245578003e-06,
"loss": 0.0586,
"step": 4480
},
{
"epoch": 0.7333006696064021,
"grad_norm": 0.13915759629937205,
"learning_rate": 9.121907073034088e-06,
"loss": 0.0561,
"step": 4490
},
{
"epoch": 0.7349338559529642,
"grad_norm": 0.1302554882859513,
"learning_rate": 9.017832185483747e-06,
"loss": 0.0566,
"step": 4500
},
{
"epoch": 0.7365670422995264,
"grad_norm": 0.1387683490735249,
"learning_rate": 8.914223619584697e-06,
"loss": 0.0568,
"step": 4510
},
{
"epoch": 0.7382002286460885,
"grad_norm": 0.16309735762981123,
"learning_rate": 8.811084398388542e-06,
"loss": 0.0562,
"step": 4520
},
{
"epoch": 0.7398334149926507,
"grad_norm": 0.15294056112930812,
"learning_rate": 8.708417531252452e-06,
"loss": 0.0568,
"step": 4530
},
{
"epoch": 0.7414666013392128,
"grad_norm": 0.12635089894766555,
"learning_rate": 8.606226013751495e-06,
"loss": 0.0546,
"step": 4540
},
{
"epoch": 0.7430997876857749,
"grad_norm": 0.1296223557027671,
"learning_rate": 8.504512827591138e-06,
"loss": 0.057,
"step": 4550
},
{
"epoch": 0.7447329740323371,
"grad_norm": 0.12997487575731115,
"learning_rate": 8.403280940520272e-06,
"loss": 0.0581,
"step": 4560
},
{
"epoch": 0.7463661603788992,
"grad_norm": 0.13684099591436172,
"learning_rate": 8.30253330624464e-06,
"loss": 0.0565,
"step": 4570
},
{
"epoch": 0.7479993467254614,
"grad_norm": 0.1304009105397714,
"learning_rate": 8.202272864340615e-06,
"loss": 0.0568,
"step": 4580
},
{
"epoch": 0.7496325330720235,
"grad_norm": 0.12097716741999806,
"learning_rate": 8.102502540169507e-06,
"loss": 0.0578,
"step": 4590
},
{
"epoch": 0.7512657194185857,
"grad_norm": 0.12319605953172555,
"learning_rate": 8.003225244792086e-06,
"loss": 0.0565,
"step": 4600
},
{
"epoch": 0.7528989057651478,
"grad_norm": 0.14554717048364868,
"learning_rate": 7.904443874883774e-06,
"loss": 0.0586,
"step": 4610
},
{
"epoch": 0.7545320921117099,
"grad_norm": 0.12172883945712641,
"learning_rate": 7.806161312650032e-06,
"loss": 0.0569,
"step": 4620
},
{
"epoch": 0.7561652784582721,
"grad_norm": 0.1961056326155406,
"learning_rate": 7.708380425742318e-06,
"loss": 0.0564,
"step": 4630
},
{
"epoch": 0.7577984648048343,
"grad_norm": 0.1551941571625073,
"learning_rate": 7.611104067174374e-06,
"loss": 0.0551,
"step": 4640
},
{
"epoch": 0.7594316511513963,
"grad_norm": 0.1427695808055181,
"learning_rate": 7.51433507523904e-06,
"loss": 0.0577,
"step": 4650
},
{
"epoch": 0.7610648374979585,
"grad_norm": 0.14505797700743758,
"learning_rate": 7.418076273425376e-06,
"loss": 0.0555,
"step": 4660
},
{
"epoch": 0.7626980238445207,
"grad_norm": 0.13555935823642642,
"learning_rate": 7.3223304703363135e-06,
"loss": 0.0553,
"step": 4670
},
{
"epoch": 0.7643312101910829,
"grad_norm": 0.1290844307632044,
"learning_rate": 7.227100459606708e-06,
"loss": 0.0563,
"step": 4680
},
{
"epoch": 0.7659643965376449,
"grad_norm": 0.13281813408633075,
"learning_rate": 7.132389019821814e-06,
"loss": 0.0572,
"step": 4690
},
{
"epoch": 0.7675975828842071,
"grad_norm": 0.13245878175927664,
"learning_rate": 7.03819891443622e-06,
"loss": 0.0564,
"step": 4700
},
{
"epoch": 0.7692307692307693,
"grad_norm": 0.14169970544865157,
"learning_rate": 6.944532891693209e-06,
"loss": 0.0562,
"step": 4710
},
{
"epoch": 0.7708639555773313,
"grad_norm": 0.12885112432051518,
"learning_rate": 6.851393684544605e-06,
"loss": 0.0555,
"step": 4720
},
{
"epoch": 0.7724971419238935,
"grad_norm": 0.13226314325547714,
"learning_rate": 6.758784010570981e-06,
"loss": 0.0566,
"step": 4730
},
{
"epoch": 0.7741303282704557,
"grad_norm": 0.12721088799884897,
"learning_rate": 6.666706571902395e-06,
"loss": 0.0558,
"step": 4740
},
{
"epoch": 0.7757635146170178,
"grad_norm": 0.16190252011114342,
"learning_rate": 6.5751640551395435e-06,
"loss": 0.0593,
"step": 4750
},
{
"epoch": 0.7773967009635799,
"grad_norm": 0.16751931275608875,
"learning_rate": 6.484159131275397e-06,
"loss": 0.0548,
"step": 4760
},
{
"epoch": 0.7790298873101421,
"grad_norm": 0.13155830986325304,
"learning_rate": 6.393694455617197e-06,
"loss": 0.0545,
"step": 4770
},
{
"epoch": 0.7806630736567042,
"grad_norm": 0.1254966927783307,
"learning_rate": 6.3037726677090764e-06,
"loss": 0.0552,
"step": 4780
},
{
"epoch": 0.7822962600032664,
"grad_norm": 0.14030676911062837,
"learning_rate": 6.214396391254965e-06,
"loss": 0.0581,
"step": 4790
},
{
"epoch": 0.7839294463498285,
"grad_norm": 0.13094470204880132,
"learning_rate": 6.125568234042073e-06,
"loss": 0.0568,
"step": 4800
},
{
"epoch": 0.7855626326963907,
"grad_norm": 0.1289368613264559,
"learning_rate": 6.037290787864794e-06,
"loss": 0.0555,
"step": 4810
},
{
"epoch": 0.7871958190429528,
"grad_norm": 0.13111818691346408,
"learning_rate": 5.949566628449079e-06,
"loss": 0.0557,
"step": 4820
},
{
"epoch": 0.788829005389515,
"grad_norm": 0.127924422151668,
"learning_rate": 5.862398315377316e-06,
"loss": 0.0563,
"step": 4830
},
{
"epoch": 0.7904621917360771,
"grad_norm": 0.13400811586086794,
"learning_rate": 5.775788392013559e-06,
"loss": 0.057,
"step": 4840
},
{
"epoch": 0.7920953780826392,
"grad_norm": 0.13006097604528882,
"learning_rate": 5.689739385429441e-06,
"loss": 0.0548,
"step": 4850
},
{
"epoch": 0.7937285644292014,
"grad_norm": 0.14100421892015846,
"learning_rate": 5.60425380633034e-06,
"loss": 0.0538,
"step": 4860
},
{
"epoch": 0.7953617507757635,
"grad_norm": 0.14150245194680788,
"learning_rate": 5.5193341489821724e-06,
"loss": 0.0563,
"step": 4870
},
{
"epoch": 0.7969949371223256,
"grad_norm": 0.15737446394116278,
"learning_rate": 5.43498289113859e-06,
"loss": 0.0554,
"step": 4880
},
{
"epoch": 0.7986281234688878,
"grad_norm": 0.1303664827537266,
"learning_rate": 5.351202493968726e-06,
"loss": 0.0549,
"step": 4890
},
{
"epoch": 0.80026130981545,
"grad_norm": 0.1579882154730711,
"learning_rate": 5.26799540198534e-06,
"loss": 0.0566,
"step": 4900
},
{
"epoch": 0.8018944961620121,
"grad_norm": 0.12320637558193934,
"learning_rate": 5.185364042973489e-06,
"loss": 0.0551,
"step": 4910
},
{
"epoch": 0.8035276825085742,
"grad_norm": 0.13561812285562463,
"learning_rate": 5.103310827919755e-06,
"loss": 0.0555,
"step": 4920
},
{
"epoch": 0.8051608688551364,
"grad_norm": 0.12657575263992352,
"learning_rate": 5.021838150941824e-06,
"loss": 0.0555,
"step": 4930
},
{
"epoch": 0.8067940552016986,
"grad_norm": 0.15109328807564862,
"learning_rate": 4.940948389218664e-06,
"loss": 0.0576,
"step": 4940
},
{
"epoch": 0.8084272415482606,
"grad_norm": 0.11927600037980748,
"learning_rate": 4.8606439029211635e-06,
"loss": 0.0562,
"step": 4950
},
{
"epoch": 0.8100604278948228,
"grad_norm": 0.13489516491237835,
"learning_rate": 4.780927035143282e-06,
"loss": 0.0568,
"step": 4960
},
{
"epoch": 0.811693614241385,
"grad_norm": 0.14125369866281293,
"learning_rate": 4.701800111833649e-06,
"loss": 0.0559,
"step": 4970
},
{
"epoch": 0.813326800587947,
"grad_norm": 0.14329929830579513,
"learning_rate": 4.62326544172772e-06,
"loss": 0.0573,
"step": 4980
},
{
"epoch": 0.8149599869345092,
"grad_norm": 0.13721815993745648,
"learning_rate": 4.545325316280408e-06,
"loss": 0.0565,
"step": 4990
},
{
"epoch": 0.8165931732810714,
"grad_norm": 0.12403207676608863,
"learning_rate": 4.467982009599256e-06,
"loss": 0.0547,
"step": 5000
},
{
"epoch": 0.8165931732810714,
"eval_loss": 0.055860333144664764,
"eval_runtime": 543.852,
"eval_samples_per_second": 11.121,
"eval_steps_per_second": 0.695,
"step": 5000
},
{
"epoch": 0.8182263596276335,
"grad_norm": 0.19887012331746975,
"learning_rate": 4.391237778378007e-06,
"loss": 0.0566,
"step": 5010
},
{
"epoch": 0.8198595459741956,
"grad_norm": 0.12467986837814826,
"learning_rate": 4.315094861830832e-06,
"loss": 0.0551,
"step": 5020
},
{
"epoch": 0.8214927323207578,
"grad_norm": 0.12071489403888272,
"learning_rate": 4.23955548162698e-06,
"loss": 0.0546,
"step": 5030
},
{
"epoch": 0.82312591866732,
"grad_norm": 0.13278086969547637,
"learning_rate": 4.164621841825931e-06,
"loss": 0.0552,
"step": 5040
},
{
"epoch": 0.824759105013882,
"grad_norm": 0.20820791200777022,
"learning_rate": 4.090296128813096e-06,
"loss": 0.0568,
"step": 5050
},
{
"epoch": 0.8263922913604442,
"grad_norm": 0.1530010429396959,
"learning_rate": 4.016580511236032e-06,
"loss": 0.0539,
"step": 5060
},
{
"epoch": 0.8280254777070064,
"grad_norm": 0.12719177224871234,
"learning_rate": 3.943477139941182e-06,
"loss": 0.0536,
"step": 5070
},
{
"epoch": 0.8296586640535685,
"grad_norm": 0.141814433546691,
"learning_rate": 3.870988147911059e-06,
"loss": 0.0561,
"step": 5080
},
{
"epoch": 0.8312918504001306,
"grad_norm": 0.12755450599107368,
"learning_rate": 3.799115650202087e-06,
"loss": 0.0549,
"step": 5090
},
{
"epoch": 0.8329250367466928,
"grad_norm": 0.1347133374311777,
"learning_rate": 3.727861743882832e-06,
"loss": 0.0536,
"step": 5100
},
{
"epoch": 0.8345582230932549,
"grad_norm": 0.13567801811178337,
"learning_rate": 3.6572285079728403e-06,
"loss": 0.0567,
"step": 5110
},
{
"epoch": 0.8361914094398171,
"grad_norm": 0.12450354118309034,
"learning_rate": 3.5872180033819648e-06,
"loss": 0.0544,
"step": 5120
},
{
"epoch": 0.8378245957863792,
"grad_norm": 0.12651598560432128,
"learning_rate": 3.5178322728502523e-06,
"loss": 0.0554,
"step": 5130
},
{
"epoch": 0.8394577821329414,
"grad_norm": 0.1209767803553846,
"learning_rate": 3.449073340888315e-06,
"loss": 0.0544,
"step": 5140
},
{
"epoch": 0.8410909684795035,
"grad_norm": 0.1306153917558242,
"learning_rate": 3.380943213718277e-06,
"loss": 0.0531,
"step": 5150
},
{
"epoch": 0.8427241548260657,
"grad_norm": 0.18773038823975557,
"learning_rate": 3.3134438792152446e-06,
"loss": 0.0545,
"step": 5160
},
{
"epoch": 0.8443573411726278,
"grad_norm": 0.12718995100041938,
"learning_rate": 3.24657730684928e-06,
"loss": 0.0575,
"step": 5170
},
{
"epoch": 0.8459905275191899,
"grad_norm": 0.14846459516101132,
"learning_rate": 3.180345447627958e-06,
"loss": 0.0572,
"step": 5180
},
{
"epoch": 0.8476237138657521,
"grad_norm": 0.139520833223967,
"learning_rate": 3.1147502340394263e-06,
"loss": 0.0537,
"step": 5190
},
{
"epoch": 0.8492569002123143,
"grad_norm": 0.1329655818459416,
"learning_rate": 3.049793579996049e-06,
"loss": 0.0583,
"step": 5200
},
{
"epoch": 0.8508900865588763,
"grad_norm": 0.11747195664034284,
"learning_rate": 2.985477380778498e-06,
"loss": 0.0548,
"step": 5210
},
{
"epoch": 0.8525232729054385,
"grad_norm": 0.13759956039154164,
"learning_rate": 2.921803512980545e-06,
"loss": 0.0539,
"step": 5220
},
{
"epoch": 0.8541564592520007,
"grad_norm": 0.12131727169672672,
"learning_rate": 2.858773834454229e-06,
"loss": 0.054,
"step": 5230
},
{
"epoch": 0.8557896455985627,
"grad_norm": 0.13293760222071632,
"learning_rate": 2.7963901842556845e-06,
"loss": 0.0562,
"step": 5240
},
{
"epoch": 0.8574228319451249,
"grad_norm": 0.15609515059366313,
"learning_rate": 2.7346543825914756e-06,
"loss": 0.0548,
"step": 5250
},
{
"epoch": 0.8590560182916871,
"grad_norm": 0.12163266798908833,
"learning_rate": 2.673568230765483e-06,
"loss": 0.0554,
"step": 5260
},
{
"epoch": 0.8606892046382493,
"grad_norm": 0.13274735674713975,
"learning_rate": 2.61313351112637e-06,
"loss": 0.0544,
"step": 5270
},
{
"epoch": 0.8623223909848113,
"grad_norm": 0.11965528370797017,
"learning_rate": 2.55335198701552e-06,
"loss": 0.0524,
"step": 5280
},
{
"epoch": 0.8639555773313735,
"grad_norm": 0.13690049978688093,
"learning_rate": 2.4942254027156634e-06,
"loss": 0.0551,
"step": 5290
},
{
"epoch": 0.8655887636779357,
"grad_norm": 0.11796501810985008,
"learning_rate": 2.4357554833999237e-06,
"loss": 0.0535,
"step": 5300
},
{
"epoch": 0.8672219500244978,
"grad_norm": 0.13289725439381544,
"learning_rate": 2.377943935081503e-06,
"loss": 0.0543,
"step": 5310
},
{
"epoch": 0.8688551363710599,
"grad_norm": 0.14419968862631125,
"learning_rate": 2.320792444563899e-06,
"loss": 0.0549,
"step": 5320
},
{
"epoch": 0.8704883227176221,
"grad_norm": 0.16902186263728,
"learning_rate": 2.264302679391714e-06,
"loss": 0.0542,
"step": 5330
},
{
"epoch": 0.8721215090641842,
"grad_norm": 0.15614660989189008,
"learning_rate": 2.2084762878019583e-06,
"loss": 0.0528,
"step": 5340
},
{
"epoch": 0.8737546954107464,
"grad_norm": 0.149797910284147,
"learning_rate": 2.153314898675973e-06,
"loss": 0.0531,
"step": 5350
},
{
"epoch": 0.8753878817573085,
"grad_norm": 0.13106985158005888,
"learning_rate": 2.098820121491929e-06,
"loss": 0.0534,
"step": 5360
},
{
"epoch": 0.8770210681038707,
"grad_norm": 0.134794687774044,
"learning_rate": 2.0449935462778374e-06,
"loss": 0.0528,
"step": 5370
},
{
"epoch": 0.8786542544504328,
"grad_norm": 0.14836220412648662,
"learning_rate": 1.991836743565156e-06,
"loss": 0.0553,
"step": 5380
},
{
"epoch": 0.880287440796995,
"grad_norm": 0.13155197028557478,
"learning_rate": 1.9393512643429868e-06,
"loss": 0.0536,
"step": 5390
},
{
"epoch": 0.8819206271435571,
"grad_norm": 0.1399174097266047,
"learning_rate": 1.8875386400128037e-06,
"loss": 0.0543,
"step": 5400
},
{
"epoch": 0.8835538134901192,
"grad_norm": 0.1203290152717323,
"learning_rate": 1.8364003823437726e-06,
"loss": 0.0543,
"step": 5410
},
{
"epoch": 0.8851869998366814,
"grad_norm": 0.15292401963339658,
"learning_rate": 1.7859379834286454e-06,
"loss": 0.0543,
"step": 5420
},
{
"epoch": 0.8868201861832435,
"grad_norm": 0.12643388691187696,
"learning_rate": 1.736152915640224e-06,
"loss": 0.0532,
"step": 5430
},
{
"epoch": 0.8884533725298056,
"grad_norm": 0.13489692904686948,
"learning_rate": 1.6870466315883988e-06,
"loss": 0.0545,
"step": 5440
},
{
"epoch": 0.8900865588763678,
"grad_norm": 0.1568962348673152,
"learning_rate": 1.6386205640777614e-06,
"loss": 0.0535,
"step": 5450
},
{
"epoch": 0.89171974522293,
"grad_norm": 0.13087203560425814,
"learning_rate": 1.5908761260658016e-06,
"loss": 0.0559,
"step": 5460
},
{
"epoch": 0.893352931569492,
"grad_norm": 0.15181485270273318,
"learning_rate": 1.5438147106216937e-06,
"loss": 0.0544,
"step": 5470
},
{
"epoch": 0.8949861179160542,
"grad_norm": 0.13488491623297647,
"learning_rate": 1.4974376908856225e-06,
"loss": 0.0541,
"step": 5480
},
{
"epoch": 0.8966193042626164,
"grad_norm": 0.12478759949352025,
"learning_rate": 1.4517464200287435e-06,
"loss": 0.052,
"step": 5490
},
{
"epoch": 0.8982524906091786,
"grad_norm": 0.13754659467359648,
"learning_rate": 1.4067422312136785e-06,
"loss": 0.0536,
"step": 5500
},
{
"epoch": 0.8998856769557406,
"grad_norm": 0.11500783500507243,
"learning_rate": 1.36242643755565e-06,
"loss": 0.0533,
"step": 5510
},
{
"epoch": 0.9015188633023028,
"grad_norm": 0.14787735397766552,
"learning_rate": 1.3188003320841192e-06,
"loss": 0.0538,
"step": 5520
},
{
"epoch": 0.903152049648865,
"grad_norm": 0.13202745664778504,
"learning_rate": 1.2758651877051136e-06,
"loss": 0.0528,
"step": 5530
},
{
"epoch": 0.904785235995427,
"grad_norm": 0.12668400897579846,
"learning_rate": 1.233622257164041e-06,
"loss": 0.0527,
"step": 5540
},
{
"epoch": 0.9064184223419892,
"grad_norm": 0.13619949979387838,
"learning_rate": 1.192072773009162e-06,
"loss": 0.0541,
"step": 5550
},
{
"epoch": 0.9080516086885514,
"grad_norm": 0.13242424102043152,
"learning_rate": 1.1512179475556217e-06,
"loss": 0.0524,
"step": 5560
},
{
"epoch": 0.9096847950351135,
"grad_norm": 0.13163804112161623,
"learning_rate": 1.1110589728500654e-06,
"loss": 0.0529,
"step": 5570
},
{
"epoch": 0.9113179813816756,
"grad_norm": 0.14066382987545922,
"learning_rate": 1.0715970206358922e-06,
"loss": 0.053,
"step": 5580
},
{
"epoch": 0.9129511677282378,
"grad_norm": 0.13495896425311427,
"learning_rate": 1.0328332423190119e-06,
"loss": 0.0528,
"step": 5590
},
{
"epoch": 0.9145843540748,
"grad_norm": 0.13386525771314695,
"learning_rate": 9.947687689343065e-07,
"loss": 0.0532,
"step": 5600
},
{
"epoch": 0.916217540421362,
"grad_norm": 0.1392891644981896,
"learning_rate": 9.574047111125889e-07,
"loss": 0.0535,
"step": 5610
},
{
"epoch": 0.9178507267679242,
"grad_norm": 0.13551827574659697,
"learning_rate": 9.207421590482157e-07,
"loss": 0.0537,
"step": 5620
},
{
"epoch": 0.9194839131144864,
"grad_norm": 0.12241881698567131,
"learning_rate": 8.847821824672669e-07,
"loss": 0.0534,
"step": 5630
},
{
"epoch": 0.9211170994610485,
"grad_norm": 0.12976237525116585,
"learning_rate": 8.495258305963538e-07,
"loss": 0.0551,
"step": 5640
},
{
"epoch": 0.9227502858076106,
"grad_norm": 0.1268881392563622,
"learning_rate": 8.149741321319875e-07,
"loss": 0.054,
"step": 5650
},
{
"epoch": 0.9243834721541728,
"grad_norm": 0.1281638367161884,
"learning_rate": 7.81128095210551e-07,
"loss": 0.0531,
"step": 5660
},
{
"epoch": 0.9260166585007349,
"grad_norm": 0.11506574768369364,
"learning_rate": 7.479887073789272e-07,
"loss": 0.0529,
"step": 5670
},
{
"epoch": 0.9276498448472971,
"grad_norm": 0.1262483366497081,
"learning_rate": 7.155569355656477e-07,
"loss": 0.0538,
"step": 5680
},
{
"epoch": 0.9292830311938592,
"grad_norm": 0.12799429946109006,
"learning_rate": 6.838337260526879e-07,
"loss": 0.0546,
"step": 5690
},
{
"epoch": 0.9309162175404213,
"grad_norm": 0.1347688754342538,
"learning_rate": 6.528200044478599e-07,
"loss": 0.0521,
"step": 5700
},
{
"epoch": 0.9325494038869835,
"grad_norm": 0.11754956233250828,
"learning_rate": 6.225166756578193e-07,
"loss": 0.0523,
"step": 5710
},
{
"epoch": 0.9341825902335457,
"grad_norm": 0.12377335288388047,
"learning_rate": 5.929246238616349e-07,
"loss": 0.0545,
"step": 5720
},
{
"epoch": 0.9358157765801078,
"grad_norm": 0.12859886291910413,
"learning_rate": 5.640447124850068e-07,
"loss": 0.0537,
"step": 5730
},
{
"epoch": 0.9374489629266699,
"grad_norm": 0.12382377129365348,
"learning_rate": 5.358777841750706e-07,
"loss": 0.0532,
"step": 5740
},
{
"epoch": 0.9390821492732321,
"grad_norm": 0.13896679661623798,
"learning_rate": 5.084246607758191e-07,
"loss": 0.0571,
"step": 5750
},
{
"epoch": 0.9407153356197943,
"grad_norm": 0.13229078183327786,
"learning_rate": 4.816861433041081e-07,
"loss": 0.0536,
"step": 5760
},
{
"epoch": 0.9423485219663563,
"grad_norm": 0.15723435343533537,
"learning_rate": 4.5566301192628835e-07,
"loss": 0.0534,
"step": 5770
},
{
"epoch": 0.9439817083129185,
"grad_norm": 0.11858722577359959,
"learning_rate": 4.3035602593545776e-07,
"loss": 0.0535,
"step": 5780
},
{
"epoch": 0.9456148946594807,
"grad_norm": 0.11207955419375348,
"learning_rate": 4.0576592372928435e-07,
"loss": 0.0531,
"step": 5790
},
{
"epoch": 0.9472480810060427,
"grad_norm": 0.12561048523762686,
"learning_rate": 3.818934227884735e-07,
"loss": 0.0529,
"step": 5800
},
{
"epoch": 0.9488812673526049,
"grad_norm": 0.1333913907086287,
"learning_rate": 3.587392196558348e-07,
"loss": 0.0537,
"step": 5810
},
{
"epoch": 0.9505144536991671,
"grad_norm": 0.12690886599595988,
"learning_rate": 3.363039899159592e-07,
"loss": 0.0546,
"step": 5820
},
{
"epoch": 0.9521476400457293,
"grad_norm": 0.1178252059077168,
"learning_rate": 3.145883881754935e-07,
"loss": 0.0533,
"step": 5830
},
{
"epoch": 0.9537808263922913,
"grad_norm": 0.12676258260050632,
"learning_rate": 2.935930480440635e-07,
"loss": 0.0532,
"step": 5840
},
{
"epoch": 0.9554140127388535,
"grad_norm": 0.12500973694648929,
"learning_rate": 2.7331858211576135e-07,
"loss": 0.0532,
"step": 5850
},
{
"epoch": 0.9570471990854157,
"grad_norm": 0.14229105983415347,
"learning_rate": 2.5376558195129295e-07,
"loss": 0.0528,
"step": 5860
},
{
"epoch": 0.9586803854319778,
"grad_norm": 0.12349792562255694,
"learning_rate": 2.3493461806070595e-07,
"loss": 0.0533,
"step": 5870
},
{
"epoch": 0.9603135717785399,
"grad_norm": 0.1287262043350421,
"learning_rate": 2.168262398867471e-07,
"loss": 0.0534,
"step": 5880
},
{
"epoch": 0.9619467581251021,
"grad_norm": 0.12095410936787677,
"learning_rate": 1.9944097578883102e-07,
"loss": 0.0535,
"step": 5890
},
{
"epoch": 0.9635799444716642,
"grad_norm": 0.11769702230163032,
"learning_rate": 1.8277933302761617e-07,
"loss": 0.0544,
"step": 5900
},
{
"epoch": 0.9652131308182264,
"grad_norm": 0.12997465485858162,
"learning_rate": 1.6684179775022514e-07,
"loss": 0.0531,
"step": 5910
},
{
"epoch": 0.9668463171647885,
"grad_norm": 0.11745263664727353,
"learning_rate": 1.5162883497603087e-07,
"loss": 0.0522,
"step": 5920
},
{
"epoch": 0.9684795035113506,
"grad_norm": 0.11886706684301844,
"learning_rate": 1.371408885831177e-07,
"loss": 0.0531,
"step": 5930
},
{
"epoch": 0.9701126898579128,
"grad_norm": 0.1377318555290602,
"learning_rate": 1.233783812952999e-07,
"loss": 0.0557,
"step": 5940
},
{
"epoch": 0.971745876204475,
"grad_norm": 0.11891612541784774,
"learning_rate": 1.1034171466982324e-07,
"loss": 0.053,
"step": 5950
},
{
"epoch": 0.9733790625510371,
"grad_norm": 0.12299264400379381,
"learning_rate": 9.803126908561333e-08,
"loss": 0.0547,
"step": 5960
},
{
"epoch": 0.9750122488975992,
"grad_norm": 0.13391062923056013,
"learning_rate": 8.644740373219551e-08,
"loss": 0.0545,
"step": 5970
},
{
"epoch": 0.9766454352441614,
"grad_norm": 0.11653929688194292,
"learning_rate": 7.559045659921992e-08,
"loss": 0.0535,
"step": 5980
},
{
"epoch": 0.9782786215907235,
"grad_norm": 0.12853315945107505,
"learning_rate": 6.546074446658335e-08,
"loss": 0.052,
"step": 5990
},
{
"epoch": 0.9799118079372856,
"grad_norm": 0.12591831219989558,
"learning_rate": 5.605856289519762e-08,
"loss": 0.0537,
"step": 6000
},
{
"epoch": 0.9799118079372856,
"eval_loss": 0.05418479070067406,
"eval_runtime": 545.883,
"eval_samples_per_second": 11.079,
"eval_steps_per_second": 0.692,
"step": 6000
},
{
"epoch": 0.9815449942838478,
"grad_norm": 0.1289709275515278,
"learning_rate": 4.738418621836327e-08,
"loss": 0.0559,
"step": 6010
},
{
"epoch": 0.98317818063041,
"grad_norm": 0.1270428527091105,
"learning_rate": 3.943786753376477e-08,
"loss": 0.0537,
"step": 6020
},
{
"epoch": 0.984811366976972,
"grad_norm": 0.11934175945227758,
"learning_rate": 3.221983869608203e-08,
"loss": 0.0526,
"step": 6030
},
{
"epoch": 0.9864445533235342,
"grad_norm": 0.11445880049367826,
"learning_rate": 2.573031031023465e-08,
"loss": 0.0553,
"step": 6040
},
{
"epoch": 0.9880777396700964,
"grad_norm": 0.13009550972648035,
"learning_rate": 1.996947172522856e-08,
"loss": 0.0538,
"step": 6050
},
{
"epoch": 0.9897109260166586,
"grad_norm": 0.12300433057856806,
"learning_rate": 1.49374910286354e-08,
"loss": 0.0531,
"step": 6060
},
{
"epoch": 0.9913441123632206,
"grad_norm": 0.12335232863367096,
"learning_rate": 1.0634515041682557e-08,
"loss": 0.0544,
"step": 6070
},
{
"epoch": 0.9929772987097828,
"grad_norm": 0.12952038925493475,
"learning_rate": 7.0606693149788404e-09,
"loss": 0.0531,
"step": 6080
},
{
"epoch": 0.994610485056345,
"grad_norm": 0.1314912893697476,
"learning_rate": 4.216058124842381e-09,
"loss": 0.0537,
"step": 6090
},
{
"epoch": 0.996243671402907,
"grad_norm": 0.11998223236168541,
"learning_rate": 2.100764470266969e-09,
"loss": 0.0529,
"step": 6100
},
{
"epoch": 0.9978768577494692,
"grad_norm": 0.1277215286002024,
"learning_rate": 7.14850070490658e-10,
"loss": 0.0548,
"step": 6110
},
{
"epoch": 0.9995100440960314,
"grad_norm": 0.12495906000136037,
"learning_rate": 5.835536319720447e-11,
"loss": 0.055,
"step": 6120
},
{
"epoch": 1.0,
"step": 6123,
"total_flos": 5127707767603200.0,
"train_loss": 0.07654633040222178,
"train_runtime": 76896.0345,
"train_samples_per_second": 2.548,
"train_steps_per_second": 0.08
}
],
"logging_steps": 10,
"max_steps": 6123,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 1000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 5127707767603200.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}