MultimodalAgent's picture
Upload ensemble_teacher1_1/trainer_state.json with huggingface_hub
ec7b6ea verified
Raw
History Blame Contribute Delete
267 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 15165,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0019784350578692256,
"grad_norm": 18.83415773016879,
"learning_rate": 5.932762030323006e-07,
"loss": 8.2608,
"step": 10
},
{
"epoch": 0.003956870115738451,
"grad_norm": 18.953139891565925,
"learning_rate": 1.2524719841793013e-06,
"loss": 8.2021,
"step": 20
},
{
"epoch": 0.005935305173607676,
"grad_norm": 17.047190545054274,
"learning_rate": 1.9116677653263018e-06,
"loss": 7.8828,
"step": 30
},
{
"epoch": 0.007913740231476903,
"grad_norm": 7.170721178917912,
"learning_rate": 2.570863546473303e-06,
"loss": 7.1412,
"step": 40
},
{
"epoch": 0.009892175289346126,
"grad_norm": 4.134672335987729,
"learning_rate": 3.2300593276203036e-06,
"loss": 6.6072,
"step": 50
},
{
"epoch": 0.011870610347215352,
"grad_norm": 2.9944810910906696,
"learning_rate": 3.889255108767304e-06,
"loss": 6.2712,
"step": 60
},
{
"epoch": 0.013849045405084578,
"grad_norm": 2.7970513693315793,
"learning_rate": 4.548450889914305e-06,
"loss": 5.9829,
"step": 70
},
{
"epoch": 0.015827480462953805,
"grad_norm": 3.647632982797199,
"learning_rate": 5.207646671061305e-06,
"loss": 5.6505,
"step": 80
},
{
"epoch": 0.01780591552082303,
"grad_norm": 3.740214396660077,
"learning_rate": 5.866842452208306e-06,
"loss": 5.1825,
"step": 90
},
{
"epoch": 0.019784350578692253,
"grad_norm": 6.12614078808338,
"learning_rate": 6.5260382333553066e-06,
"loss": 4.6773,
"step": 100
},
{
"epoch": 0.02176278563656148,
"grad_norm": 5.15959689995949,
"learning_rate": 7.1852340145023085e-06,
"loss": 4.1938,
"step": 110
},
{
"epoch": 0.023741220694430704,
"grad_norm": 6.16531113423169,
"learning_rate": 7.844429795649308e-06,
"loss": 3.7884,
"step": 120
},
{
"epoch": 0.02571965575229993,
"grad_norm": 4.8684776151555855,
"learning_rate": 8.503625576796309e-06,
"loss": 3.4738,
"step": 130
},
{
"epoch": 0.027698090810169156,
"grad_norm": 4.484420845519575,
"learning_rate": 9.162821357943308e-06,
"loss": 3.2115,
"step": 140
},
{
"epoch": 0.029676525868038383,
"grad_norm": 6.851328613123179,
"learning_rate": 9.82201713909031e-06,
"loss": 3.0015,
"step": 150
},
{
"epoch": 0.03165496092590761,
"grad_norm": 5.812087061911436,
"learning_rate": 1.048121292023731e-05,
"loss": 2.8175,
"step": 160
},
{
"epoch": 0.033633395983776834,
"grad_norm": 6.9325494255558375,
"learning_rate": 1.1140408701384312e-05,
"loss": 2.698,
"step": 170
},
{
"epoch": 0.03561183104164606,
"grad_norm": 4.685368838386482,
"learning_rate": 1.1799604482531313e-05,
"loss": 2.59,
"step": 180
},
{
"epoch": 0.03759026609951528,
"grad_norm": 7.094727681417276,
"learning_rate": 1.2458800263678314e-05,
"loss": 2.5067,
"step": 190
},
{
"epoch": 0.039568701157384506,
"grad_norm": 10.805621903825395,
"learning_rate": 1.3117996044825315e-05,
"loss": 2.4395,
"step": 200
},
{
"epoch": 0.04154713621525374,
"grad_norm": 16.15779288881059,
"learning_rate": 1.3777191825972313e-05,
"loss": 2.3869,
"step": 210
},
{
"epoch": 0.04352557127312296,
"grad_norm": 15.926165373706107,
"learning_rate": 1.4436387607119314e-05,
"loss": 2.3372,
"step": 220
},
{
"epoch": 0.045504006330992185,
"grad_norm": 17.804252127547205,
"learning_rate": 1.5095583388266315e-05,
"loss": 2.2883,
"step": 230
},
{
"epoch": 0.04748244138886141,
"grad_norm": 10.023736576926494,
"learning_rate": 1.5754779169413318e-05,
"loss": 2.2571,
"step": 240
},
{
"epoch": 0.04946087644673064,
"grad_norm": 12.163162680085504,
"learning_rate": 1.6413974950560317e-05,
"loss": 2.2232,
"step": 250
},
{
"epoch": 0.05143931150459986,
"grad_norm": 12.292818936624279,
"learning_rate": 1.707317073170732e-05,
"loss": 2.1898,
"step": 260
},
{
"epoch": 0.05341774656246909,
"grad_norm": 10.778223723930257,
"learning_rate": 1.7732366512854316e-05,
"loss": 2.1514,
"step": 270
},
{
"epoch": 0.05539618162033831,
"grad_norm": 32.81585475834664,
"learning_rate": 1.839156229400132e-05,
"loss": 2.1189,
"step": 280
},
{
"epoch": 0.057374616678207535,
"grad_norm": 21.314336729686293,
"learning_rate": 1.905075807514832e-05,
"loss": 2.0858,
"step": 290
},
{
"epoch": 0.059353051736076766,
"grad_norm": 22.593539823222432,
"learning_rate": 1.970995385629532e-05,
"loss": 2.0285,
"step": 300
},
{
"epoch": 0.06133148679394599,
"grad_norm": 19.62204024434294,
"learning_rate": 2.0369149637442324e-05,
"loss": 1.9786,
"step": 310
},
{
"epoch": 0.06330992185181522,
"grad_norm": 15.655459266100811,
"learning_rate": 2.102834541858932e-05,
"loss": 1.8985,
"step": 320
},
{
"epoch": 0.06528835690968444,
"grad_norm": 10.573693160261442,
"learning_rate": 2.1687541199736323e-05,
"loss": 1.812,
"step": 330
},
{
"epoch": 0.06726679196755367,
"grad_norm": 12.326829961937381,
"learning_rate": 2.2346736980883322e-05,
"loss": 1.734,
"step": 340
},
{
"epoch": 0.06924522702542289,
"grad_norm": 12.445007337826206,
"learning_rate": 2.3005932762030325e-05,
"loss": 1.6281,
"step": 350
},
{
"epoch": 0.07122366208329212,
"grad_norm": 13.3317136074336,
"learning_rate": 2.3665128543177324e-05,
"loss": 1.5338,
"step": 360
},
{
"epoch": 0.07320209714116134,
"grad_norm": 13.259614152330807,
"learning_rate": 2.4324324324324327e-05,
"loss": 1.4389,
"step": 370
},
{
"epoch": 0.07518053219903056,
"grad_norm": 13.958238353110612,
"learning_rate": 2.4983520105471327e-05,
"loss": 1.3512,
"step": 380
},
{
"epoch": 0.07715896725689979,
"grad_norm": 12.700525607370855,
"learning_rate": 2.564271588661833e-05,
"loss": 1.2876,
"step": 390
},
{
"epoch": 0.07913740231476901,
"grad_norm": 12.36342135304624,
"learning_rate": 2.630191166776533e-05,
"loss": 1.2281,
"step": 400
},
{
"epoch": 0.08111583737263825,
"grad_norm": 9.833896833489808,
"learning_rate": 2.6961107448912325e-05,
"loss": 1.1754,
"step": 410
},
{
"epoch": 0.08309427243050747,
"grad_norm": 12.131912328906797,
"learning_rate": 2.762030323005933e-05,
"loss": 1.1288,
"step": 420
},
{
"epoch": 0.0850727074883767,
"grad_norm": 7.104029936016102,
"learning_rate": 2.8279499011206327e-05,
"loss": 1.0823,
"step": 430
},
{
"epoch": 0.08705114254624592,
"grad_norm": 10.941139456454286,
"learning_rate": 2.8938694792353333e-05,
"loss": 1.0503,
"step": 440
},
{
"epoch": 0.08902957760411515,
"grad_norm": 10.932553725172617,
"learning_rate": 2.959789057350033e-05,
"loss": 1.0045,
"step": 450
},
{
"epoch": 0.09100801266198437,
"grad_norm": 9.044273713781383,
"learning_rate": 3.025708635464733e-05,
"loss": 0.9812,
"step": 460
},
{
"epoch": 0.0929864477198536,
"grad_norm": 9.673733589225867,
"learning_rate": 3.0916282135794335e-05,
"loss": 0.9496,
"step": 470
},
{
"epoch": 0.09496488277772282,
"grad_norm": 7.579162633133298,
"learning_rate": 3.1575477916941334e-05,
"loss": 0.9047,
"step": 480
},
{
"epoch": 0.09694331783559204,
"grad_norm": 9.248881829798938,
"learning_rate": 3.2234673698088334e-05,
"loss": 0.8734,
"step": 490
},
{
"epoch": 0.09892175289346128,
"grad_norm": 9.71806705902721,
"learning_rate": 3.289386947923533e-05,
"loss": 0.854,
"step": 500
},
{
"epoch": 0.1009001879513305,
"grad_norm": 6.152025374759133,
"learning_rate": 3.355306526038233e-05,
"loss": 0.8301,
"step": 510
},
{
"epoch": 0.10287862300919973,
"grad_norm": 8.275016927215734,
"learning_rate": 3.421226104152934e-05,
"loss": 0.7976,
"step": 520
},
{
"epoch": 0.10485705806706895,
"grad_norm": 11.742420999853534,
"learning_rate": 3.487145682267634e-05,
"loss": 0.7727,
"step": 530
},
{
"epoch": 0.10683549312493817,
"grad_norm": 7.4488520112511125,
"learning_rate": 3.553065260382334e-05,
"loss": 0.7593,
"step": 540
},
{
"epoch": 0.1088139281828074,
"grad_norm": 7.442067995474561,
"learning_rate": 3.618984838497034e-05,
"loss": 0.7326,
"step": 550
},
{
"epoch": 0.11079236324067662,
"grad_norm": 6.896696791848505,
"learning_rate": 3.684904416611734e-05,
"loss": 0.7145,
"step": 560
},
{
"epoch": 0.11277079829854585,
"grad_norm": 6.633140470473098,
"learning_rate": 3.750823994726434e-05,
"loss": 0.6938,
"step": 570
},
{
"epoch": 0.11474923335641507,
"grad_norm": 8.980769856651557,
"learning_rate": 3.8167435728411335e-05,
"loss": 0.6633,
"step": 580
},
{
"epoch": 0.11672766841428431,
"grad_norm": 9.180690905342084,
"learning_rate": 3.882663150955834e-05,
"loss": 0.6565,
"step": 590
},
{
"epoch": 0.11870610347215353,
"grad_norm": 5.76539877614404,
"learning_rate": 3.948582729070534e-05,
"loss": 0.6327,
"step": 600
},
{
"epoch": 0.12068453853002276,
"grad_norm": 6.76924777896912,
"learning_rate": 4.014502307185235e-05,
"loss": 0.6225,
"step": 610
},
{
"epoch": 0.12266297358789198,
"grad_norm": 7.797329103364111,
"learning_rate": 4.0804218852999346e-05,
"loss": 0.5904,
"step": 620
},
{
"epoch": 0.1246414086457612,
"grad_norm": 9.162843826493228,
"learning_rate": 4.146341463414634e-05,
"loss": 0.59,
"step": 630
},
{
"epoch": 0.12661984370363044,
"grad_norm": 6.401453349202934,
"learning_rate": 4.2122610415293345e-05,
"loss": 0.5682,
"step": 640
},
{
"epoch": 0.12859827876149965,
"grad_norm": 4.640311787030178,
"learning_rate": 4.2781806196440345e-05,
"loss": 0.5408,
"step": 650
},
{
"epoch": 0.1305767138193689,
"grad_norm": 7.001907345720152,
"learning_rate": 4.344100197758735e-05,
"loss": 0.5306,
"step": 660
},
{
"epoch": 0.1325551488772381,
"grad_norm": 4.876722497966865,
"learning_rate": 4.4100197758734343e-05,
"loss": 0.5033,
"step": 670
},
{
"epoch": 0.13453358393510734,
"grad_norm": 6.970440746846238,
"learning_rate": 4.475939353988134e-05,
"loss": 0.4995,
"step": 680
},
{
"epoch": 0.13651201899297655,
"grad_norm": 5.591068958664834,
"learning_rate": 4.541858932102835e-05,
"loss": 0.4735,
"step": 690
},
{
"epoch": 0.13849045405084578,
"grad_norm": 5.698931024382154,
"learning_rate": 4.607778510217535e-05,
"loss": 0.4475,
"step": 700
},
{
"epoch": 0.140468889108715,
"grad_norm": 5.018558043426356,
"learning_rate": 4.673698088332235e-05,
"loss": 0.4399,
"step": 710
},
{
"epoch": 0.14244732416658423,
"grad_norm": 6.419667369431367,
"learning_rate": 4.739617666446935e-05,
"loss": 0.4213,
"step": 720
},
{
"epoch": 0.14442575922445347,
"grad_norm": 5.315142718439434,
"learning_rate": 4.805537244561635e-05,
"loss": 0.4066,
"step": 730
},
{
"epoch": 0.14640419428232268,
"grad_norm": 3.8652691981941536,
"learning_rate": 4.871456822676335e-05,
"loss": 0.3793,
"step": 740
},
{
"epoch": 0.14838262934019192,
"grad_norm": 3.7423552705313305,
"learning_rate": 4.937376400791035e-05,
"loss": 0.3687,
"step": 750
},
{
"epoch": 0.15036106439806113,
"grad_norm": 4.678259355369585,
"learning_rate": 5.003295978905735e-05,
"loss": 0.3539,
"step": 760
},
{
"epoch": 0.15233949945593037,
"grad_norm": 5.018421006615824,
"learning_rate": 5.069215557020436e-05,
"loss": 0.3388,
"step": 770
},
{
"epoch": 0.15431793451379958,
"grad_norm": 4.688296251930949,
"learning_rate": 5.135135135135135e-05,
"loss": 0.3233,
"step": 780
},
{
"epoch": 0.15629636957166881,
"grad_norm": 3.5639269672081393,
"learning_rate": 5.201054713249836e-05,
"loss": 0.3063,
"step": 790
},
{
"epoch": 0.15827480462953802,
"grad_norm": 3.093982237274519,
"learning_rate": 5.2669742913645356e-05,
"loss": 0.2967,
"step": 800
},
{
"epoch": 0.16025323968740726,
"grad_norm": 4.841647966694258,
"learning_rate": 5.332893869479235e-05,
"loss": 0.2875,
"step": 810
},
{
"epoch": 0.1622316747452765,
"grad_norm": 3.6707841393198986,
"learning_rate": 5.3988134475939355e-05,
"loss": 0.2752,
"step": 820
},
{
"epoch": 0.1642101098031457,
"grad_norm": 3.7027407015847653,
"learning_rate": 5.464733025708636e-05,
"loss": 0.2627,
"step": 830
},
{
"epoch": 0.16618854486101495,
"grad_norm": 3.4682352972135626,
"learning_rate": 5.5306526038233354e-05,
"loss": 0.2456,
"step": 840
},
{
"epoch": 0.16816697991888416,
"grad_norm": 2.5685858726084185,
"learning_rate": 5.596572181938036e-05,
"loss": 0.2379,
"step": 850
},
{
"epoch": 0.1701454149767534,
"grad_norm": 2.7669856479108974,
"learning_rate": 5.662491760052736e-05,
"loss": 0.2384,
"step": 860
},
{
"epoch": 0.1721238500346226,
"grad_norm": 3.632736462475164,
"learning_rate": 5.7284113381674366e-05,
"loss": 0.2244,
"step": 870
},
{
"epoch": 0.17410228509249184,
"grad_norm": 3.8919237298640827,
"learning_rate": 5.794330916282136e-05,
"loss": 0.208,
"step": 880
},
{
"epoch": 0.17608072015036105,
"grad_norm": 2.144447970969846,
"learning_rate": 5.8602504943968364e-05,
"loss": 0.2012,
"step": 890
},
{
"epoch": 0.1780591552082303,
"grad_norm": 3.1698975403714478,
"learning_rate": 5.9261700725115364e-05,
"loss": 0.2011,
"step": 900
},
{
"epoch": 0.18003759026609953,
"grad_norm": 3.0521994638295626,
"learning_rate": 5.9920896506262356e-05,
"loss": 0.1884,
"step": 910
},
{
"epoch": 0.18201602532396874,
"grad_norm": 2.3014525129540435,
"learning_rate": 6.058009228740936e-05,
"loss": 0.1785,
"step": 920
},
{
"epoch": 0.18399446038183798,
"grad_norm": 2.4378132759781614,
"learning_rate": 6.123928806855637e-05,
"loss": 0.1665,
"step": 930
},
{
"epoch": 0.1859728954397072,
"grad_norm": 2.754759063327128,
"learning_rate": 6.189848384970337e-05,
"loss": 0.1583,
"step": 940
},
{
"epoch": 0.18795133049757642,
"grad_norm": 2.3723879939555865,
"learning_rate": 6.255767963085037e-05,
"loss": 0.1502,
"step": 950
},
{
"epoch": 0.18992976555544563,
"grad_norm": 3.1764341430309204,
"learning_rate": 6.321687541199737e-05,
"loss": 0.1467,
"step": 960
},
{
"epoch": 0.19190820061331487,
"grad_norm": 1.5557945069282222,
"learning_rate": 6.387607119314437e-05,
"loss": 0.1398,
"step": 970
},
{
"epoch": 0.19388663567118408,
"grad_norm": 1.7572434405286825,
"learning_rate": 6.453526697429137e-05,
"loss": 0.1317,
"step": 980
},
{
"epoch": 0.19586507072905332,
"grad_norm": 1.9904172176521575,
"learning_rate": 6.519446275543837e-05,
"loss": 0.1275,
"step": 990
},
{
"epoch": 0.19784350578692256,
"grad_norm": 1.8097847409572874,
"learning_rate": 6.585365853658538e-05,
"loss": 0.1222,
"step": 1000
},
{
"epoch": 0.19982194084479177,
"grad_norm": 1.569415184450286,
"learning_rate": 6.651285431773236e-05,
"loss": 0.1151,
"step": 1010
},
{
"epoch": 0.201800375902661,
"grad_norm": 1.355995645787709,
"learning_rate": 6.717205009887936e-05,
"loss": 0.1097,
"step": 1020
},
{
"epoch": 0.20377881096053022,
"grad_norm": 1.2927243199858764,
"learning_rate": 6.783124588002638e-05,
"loss": 0.1052,
"step": 1030
},
{
"epoch": 0.20575724601839945,
"grad_norm": 1.5555851772502252,
"learning_rate": 6.849044166117338e-05,
"loss": 0.1014,
"step": 1040
},
{
"epoch": 0.20773568107626866,
"grad_norm": 1.1359072707268356,
"learning_rate": 6.914963744232038e-05,
"loss": 0.0972,
"step": 1050
},
{
"epoch": 0.2097141161341379,
"grad_norm": 1.5831497855262247,
"learning_rate": 6.980883322346737e-05,
"loss": 0.0895,
"step": 1060
},
{
"epoch": 0.2116925511920071,
"grad_norm": 2.374223412368552,
"learning_rate": 7.046802900461437e-05,
"loss": 0.096,
"step": 1070
},
{
"epoch": 0.21367098624987635,
"grad_norm": 1.1551635768283977,
"learning_rate": 7.112722478576137e-05,
"loss": 0.0945,
"step": 1080
},
{
"epoch": 0.2156494213077456,
"grad_norm": 1.4708945190735183,
"learning_rate": 7.178642056690837e-05,
"loss": 0.0825,
"step": 1090
},
{
"epoch": 0.2176278563656148,
"grad_norm": 0.8189092893444943,
"learning_rate": 7.244561634805537e-05,
"loss": 0.0766,
"step": 1100
},
{
"epoch": 0.21960629142348403,
"grad_norm": 0.9277143882465384,
"learning_rate": 7.310481212920237e-05,
"loss": 0.0707,
"step": 1110
},
{
"epoch": 0.22158472648135324,
"grad_norm": 0.9495107827587731,
"learning_rate": 7.376400791034937e-05,
"loss": 0.0693,
"step": 1120
},
{
"epoch": 0.22356316153922248,
"grad_norm": 0.8586303321439945,
"learning_rate": 7.442320369149638e-05,
"loss": 0.0635,
"step": 1130
},
{
"epoch": 0.2255415965970917,
"grad_norm": 13.785403624417425,
"learning_rate": 7.508239947264338e-05,
"loss": 0.0938,
"step": 1140
},
{
"epoch": 0.22752003165496093,
"grad_norm": 2.034584522310342,
"learning_rate": 7.574159525379037e-05,
"loss": 0.1042,
"step": 1150
},
{
"epoch": 0.22949846671283014,
"grad_norm": 1.6478578749867223,
"learning_rate": 7.640079103493738e-05,
"loss": 0.0677,
"step": 1160
},
{
"epoch": 0.23147690177069938,
"grad_norm": 0.6137646052647932,
"learning_rate": 7.705998681608438e-05,
"loss": 0.0589,
"step": 1170
},
{
"epoch": 0.23345533682856862,
"grad_norm": 0.4923081269892374,
"learning_rate": 7.771918259723138e-05,
"loss": 0.0555,
"step": 1180
},
{
"epoch": 0.23543377188643783,
"grad_norm": 0.517370675809496,
"learning_rate": 7.837837837837838e-05,
"loss": 0.0513,
"step": 1190
},
{
"epoch": 0.23741220694430706,
"grad_norm": 0.6967099842273796,
"learning_rate": 7.903757415952538e-05,
"loss": 0.0512,
"step": 1200
},
{
"epoch": 0.23939064200217627,
"grad_norm": 0.6249731345352635,
"learning_rate": 7.969676994067239e-05,
"loss": 0.0524,
"step": 1210
},
{
"epoch": 0.2413690770600455,
"grad_norm": 0.4768002023563975,
"learning_rate": 8.035596572181938e-05,
"loss": 0.052,
"step": 1220
},
{
"epoch": 0.24334751211791472,
"grad_norm": 0.45578657738835726,
"learning_rate": 8.101516150296639e-05,
"loss": 0.0457,
"step": 1230
},
{
"epoch": 0.24532594717578396,
"grad_norm": 1.0318142304332376,
"learning_rate": 8.167435728411339e-05,
"loss": 0.0436,
"step": 1240
},
{
"epoch": 0.24730438223365317,
"grad_norm": 0.44152769256688235,
"learning_rate": 8.233355306526038e-05,
"loss": 0.0434,
"step": 1250
},
{
"epoch": 0.2492828172915224,
"grad_norm": 0.4814350309293641,
"learning_rate": 8.299274884640739e-05,
"loss": 0.0411,
"step": 1260
},
{
"epoch": 0.2512612523493916,
"grad_norm": 0.40008126480912476,
"learning_rate": 8.365194462755439e-05,
"loss": 0.0402,
"step": 1270
},
{
"epoch": 0.2532396874072609,
"grad_norm": 0.40157146433302776,
"learning_rate": 8.431114040870139e-05,
"loss": 0.0389,
"step": 1280
},
{
"epoch": 0.2552181224651301,
"grad_norm": 0.4176117259399137,
"learning_rate": 8.497033618984839e-05,
"loss": 0.0374,
"step": 1290
},
{
"epoch": 0.2571965575229993,
"grad_norm": 0.4759232382708195,
"learning_rate": 8.562953197099539e-05,
"loss": 0.0369,
"step": 1300
},
{
"epoch": 0.2591749925808685,
"grad_norm": 0.3460435615914539,
"learning_rate": 8.62887277521424e-05,
"loss": 0.0357,
"step": 1310
},
{
"epoch": 0.2611534276387378,
"grad_norm": 0.43159827825501135,
"learning_rate": 8.694792353328939e-05,
"loss": 0.0345,
"step": 1320
},
{
"epoch": 0.263131862696607,
"grad_norm": 0.35272735708968916,
"learning_rate": 8.760711931443639e-05,
"loss": 0.0337,
"step": 1330
},
{
"epoch": 0.2651102977544762,
"grad_norm": 0.4422254181186775,
"learning_rate": 8.82663150955834e-05,
"loss": 0.0328,
"step": 1340
},
{
"epoch": 0.2670887328123454,
"grad_norm": 0.3975108888949372,
"learning_rate": 8.892551087673038e-05,
"loss": 0.0326,
"step": 1350
},
{
"epoch": 0.2690671678702147,
"grad_norm": 0.36651052199526324,
"learning_rate": 8.95847066578774e-05,
"loss": 0.0317,
"step": 1360
},
{
"epoch": 0.2710456029280839,
"grad_norm": 0.3580105732034045,
"learning_rate": 9.02439024390244e-05,
"loss": 0.0305,
"step": 1370
},
{
"epoch": 0.2730240379859531,
"grad_norm": 0.355790284341662,
"learning_rate": 9.09030982201714e-05,
"loss": 0.0301,
"step": 1380
},
{
"epoch": 0.27500247304382236,
"grad_norm": 0.38069949044724,
"learning_rate": 9.15622940013184e-05,
"loss": 0.0297,
"step": 1390
},
{
"epoch": 0.27698090810169157,
"grad_norm": 0.34078656528352297,
"learning_rate": 9.22214897824654e-05,
"loss": 0.0289,
"step": 1400
},
{
"epoch": 0.2789593431595608,
"grad_norm": 0.35019855957815205,
"learning_rate": 9.288068556361241e-05,
"loss": 0.028,
"step": 1410
},
{
"epoch": 0.28093777821743,
"grad_norm": 0.3821242222939998,
"learning_rate": 9.35398813447594e-05,
"loss": 0.0269,
"step": 1420
},
{
"epoch": 0.28291621327529926,
"grad_norm": 0.3105521006323455,
"learning_rate": 9.41990771259064e-05,
"loss": 0.0269,
"step": 1430
},
{
"epoch": 0.28489464833316847,
"grad_norm": 0.33057511952729735,
"learning_rate": 9.48582729070534e-05,
"loss": 0.0261,
"step": 1440
},
{
"epoch": 0.2868730833910377,
"grad_norm": 0.323904016262063,
"learning_rate": 9.551746868820039e-05,
"loss": 0.0252,
"step": 1450
},
{
"epoch": 0.28885151844890694,
"grad_norm": 0.35316727297089406,
"learning_rate": 9.61766644693474e-05,
"loss": 0.0252,
"step": 1460
},
{
"epoch": 0.29082995350677615,
"grad_norm": 0.3155832069438333,
"learning_rate": 9.68358602504944e-05,
"loss": 0.0249,
"step": 1470
},
{
"epoch": 0.29280838856464536,
"grad_norm": 0.322689328264037,
"learning_rate": 9.74950560316414e-05,
"loss": 0.0243,
"step": 1480
},
{
"epoch": 0.29478682362251457,
"grad_norm": 0.2867308185290371,
"learning_rate": 9.81542518127884e-05,
"loss": 0.0235,
"step": 1490
},
{
"epoch": 0.29676525868038384,
"grad_norm": 0.3113558025898631,
"learning_rate": 9.88134475939354e-05,
"loss": 0.0234,
"step": 1500
},
{
"epoch": 0.29874369373825305,
"grad_norm": 0.28333983923838824,
"learning_rate": 9.94726433750824e-05,
"loss": 0.0228,
"step": 1510
},
{
"epoch": 0.30072212879612226,
"grad_norm": 0.30872625535716053,
"learning_rate": 9.999999470139298e-05,
"loss": 0.0227,
"step": 1520
},
{
"epoch": 0.30270056385399147,
"grad_norm": 0.6582002837233047,
"learning_rate": 9.99998092502654e-05,
"loss": 0.035,
"step": 1530
},
{
"epoch": 0.30467899891186073,
"grad_norm": 2.1618122315683537,
"learning_rate": 9.999935886991012e-05,
"loss": 0.046,
"step": 1540
},
{
"epoch": 0.30665743396972994,
"grad_norm": 0.8119867128049029,
"learning_rate": 9.99986435627135e-05,
"loss": 0.0425,
"step": 1550
},
{
"epoch": 0.30863586902759915,
"grad_norm": 0.3422995315184839,
"learning_rate": 9.99976633324657e-05,
"loss": 0.028,
"step": 1560
},
{
"epoch": 0.3106143040854684,
"grad_norm": 0.26980420163880464,
"learning_rate": 9.999641818436056e-05,
"loss": 0.0236,
"step": 1570
},
{
"epoch": 0.31259273914333763,
"grad_norm": 0.2744372140945502,
"learning_rate": 9.999490812499562e-05,
"loss": 0.0225,
"step": 1580
},
{
"epoch": 0.31457117420120684,
"grad_norm": 0.23314464787269223,
"learning_rate": 9.999313316237211e-05,
"loss": 0.0213,
"step": 1590
},
{
"epoch": 0.31654960925907605,
"grad_norm": 0.2081293503015357,
"learning_rate": 9.999109330589484e-05,
"loss": 0.0209,
"step": 1600
},
{
"epoch": 0.3185280443169453,
"grad_norm": 0.241383144178934,
"learning_rate": 9.998878856637221e-05,
"loss": 0.0201,
"step": 1610
},
{
"epoch": 0.3205064793748145,
"grad_norm": 0.22263560050505837,
"learning_rate": 9.998621895601612e-05,
"loss": 0.0195,
"step": 1620
},
{
"epoch": 0.32248491443268373,
"grad_norm": 0.2136949453588079,
"learning_rate": 9.998338448844192e-05,
"loss": 0.0189,
"step": 1630
},
{
"epoch": 0.324463349490553,
"grad_norm": 0.2569413695395457,
"learning_rate": 9.998028517866833e-05,
"loss": 0.0188,
"step": 1640
},
{
"epoch": 0.3264417845484222,
"grad_norm": 0.22811074888077165,
"learning_rate": 9.997692104311738e-05,
"loss": 0.0184,
"step": 1650
},
{
"epoch": 0.3284202196062914,
"grad_norm": 0.22735849058145127,
"learning_rate": 9.997329209961429e-05,
"loss": 0.0185,
"step": 1660
},
{
"epoch": 0.33039865466416063,
"grad_norm": 0.21047594753508164,
"learning_rate": 9.996939836738737e-05,
"loss": 0.0181,
"step": 1670
},
{
"epoch": 0.3323770897220299,
"grad_norm": 0.21848018513930142,
"learning_rate": 9.9965239867068e-05,
"loss": 0.0179,
"step": 1680
},
{
"epoch": 0.3343555247798991,
"grad_norm": 0.2178222222650017,
"learning_rate": 9.996081662069043e-05,
"loss": 0.0174,
"step": 1690
},
{
"epoch": 0.3363339598377683,
"grad_norm": 0.21376939711406978,
"learning_rate": 9.995612865169168e-05,
"loss": 0.0174,
"step": 1700
},
{
"epoch": 0.3383123948956375,
"grad_norm": 0.18339505375881116,
"learning_rate": 9.995117598491145e-05,
"loss": 0.0169,
"step": 1710
},
{
"epoch": 0.3402908299535068,
"grad_norm": 0.19004418186542446,
"learning_rate": 9.994595864659197e-05,
"loss": 0.0166,
"step": 1720
},
{
"epoch": 0.342269265011376,
"grad_norm": 0.20157365808434016,
"learning_rate": 9.994047666437785e-05,
"loss": 0.0162,
"step": 1730
},
{
"epoch": 0.3442477000692452,
"grad_norm": 0.20267124963539498,
"learning_rate": 9.993473006731592e-05,
"loss": 0.0161,
"step": 1740
},
{
"epoch": 0.3462261351271145,
"grad_norm": 0.19749515081082383,
"learning_rate": 9.992871888585518e-05,
"loss": 0.0159,
"step": 1750
},
{
"epoch": 0.3482045701849837,
"grad_norm": 0.19101220373630584,
"learning_rate": 9.992244315184647e-05,
"loss": 0.016,
"step": 1760
},
{
"epoch": 0.3501830052428529,
"grad_norm": 0.19179527513815997,
"learning_rate": 9.991590289854243e-05,
"loss": 0.0157,
"step": 1770
},
{
"epoch": 0.3521614403007221,
"grad_norm": 0.2193536295864431,
"learning_rate": 9.990909816059725e-05,
"loss": 0.0155,
"step": 1780
},
{
"epoch": 0.35413987535859137,
"grad_norm": 0.16842383099513045,
"learning_rate": 9.99020289740666e-05,
"loss": 0.0155,
"step": 1790
},
{
"epoch": 0.3561183104164606,
"grad_norm": 0.193419432595525,
"learning_rate": 9.989469537640728e-05,
"loss": 0.0151,
"step": 1800
},
{
"epoch": 0.3580967454743298,
"grad_norm": 0.16655548648483312,
"learning_rate": 9.988709740647713e-05,
"loss": 0.0149,
"step": 1810
},
{
"epoch": 0.36007518053219906,
"grad_norm": 0.17973703250202813,
"learning_rate": 9.987923510453478e-05,
"loss": 0.0147,
"step": 1820
},
{
"epoch": 0.36205361559006827,
"grad_norm": 0.14887685366056683,
"learning_rate": 9.987110851223947e-05,
"loss": 0.0145,
"step": 1830
},
{
"epoch": 0.3640320506479375,
"grad_norm": 0.17505643874211468,
"learning_rate": 9.986271767265079e-05,
"loss": 0.0143,
"step": 1840
},
{
"epoch": 0.3660104857058067,
"grad_norm": 0.17847793431919987,
"learning_rate": 9.985406263022849e-05,
"loss": 0.0141,
"step": 1850
},
{
"epoch": 0.36798892076367595,
"grad_norm": 0.16335653040417208,
"learning_rate": 9.984514343083221e-05,
"loss": 0.014,
"step": 1860
},
{
"epoch": 0.36996735582154516,
"grad_norm": 0.20948386295085683,
"learning_rate": 9.983596012172127e-05,
"loss": 0.0139,
"step": 1870
},
{
"epoch": 0.3719457908794144,
"grad_norm": 0.18643107902421685,
"learning_rate": 9.982651275155441e-05,
"loss": 0.0137,
"step": 1880
},
{
"epoch": 0.3739242259372836,
"grad_norm": 0.1744714660839781,
"learning_rate": 9.981680137038949e-05,
"loss": 0.0139,
"step": 1890
},
{
"epoch": 0.37590266099515285,
"grad_norm": 0.18580387097988882,
"learning_rate": 9.980682602968329e-05,
"loss": 0.0136,
"step": 1900
},
{
"epoch": 0.37788109605302206,
"grad_norm": 0.1783515847185619,
"learning_rate": 9.97965867822912e-05,
"loss": 0.0134,
"step": 1910
},
{
"epoch": 0.37985953111089127,
"grad_norm": 0.16439964155241193,
"learning_rate": 9.978608368246693e-05,
"loss": 0.0131,
"step": 1920
},
{
"epoch": 0.38183796616876053,
"grad_norm": 0.162855559762921,
"learning_rate": 9.977531678586227e-05,
"loss": 0.0131,
"step": 1930
},
{
"epoch": 0.38381640122662974,
"grad_norm": 0.2023138293577081,
"learning_rate": 9.976428614952675e-05,
"loss": 0.0131,
"step": 1940
},
{
"epoch": 0.38579483628449895,
"grad_norm": 0.4581132017426544,
"learning_rate": 9.975299183190734e-05,
"loss": 0.0168,
"step": 1950
},
{
"epoch": 0.38777327134236816,
"grad_norm": 0.1940710815604147,
"learning_rate": 9.974143389284818e-05,
"loss": 0.0147,
"step": 1960
},
{
"epoch": 0.38975170640023743,
"grad_norm": 0.14450779481997,
"learning_rate": 9.97296123935902e-05,
"loss": 0.0134,
"step": 1970
},
{
"epoch": 0.39173014145810664,
"grad_norm": 0.15486261737329882,
"learning_rate": 9.971752739677088e-05,
"loss": 0.0128,
"step": 1980
},
{
"epoch": 0.39370857651597585,
"grad_norm": 0.15321661449255397,
"learning_rate": 9.970517896642382e-05,
"loss": 0.0125,
"step": 1990
},
{
"epoch": 0.3956870115738451,
"grad_norm": 0.15770777393843685,
"learning_rate": 9.969256716797848e-05,
"loss": 0.0124,
"step": 2000
},
{
"epoch": 0.3976654466317143,
"grad_norm": 0.14487487091051574,
"learning_rate": 9.967969206825977e-05,
"loss": 0.0124,
"step": 2010
},
{
"epoch": 0.39964388168958354,
"grad_norm": 0.13953052863798998,
"learning_rate": 9.96665537354878e-05,
"loss": 0.0121,
"step": 2020
},
{
"epoch": 0.40162231674745275,
"grad_norm": 0.16936616757764028,
"learning_rate": 9.965315223927741e-05,
"loss": 0.0122,
"step": 2030
},
{
"epoch": 0.403600751805322,
"grad_norm": 0.14660923952333818,
"learning_rate": 9.963948765063778e-05,
"loss": 0.012,
"step": 2040
},
{
"epoch": 0.4055791868631912,
"grad_norm": 0.1560709624380365,
"learning_rate": 9.962556004197217e-05,
"loss": 0.012,
"step": 2050
},
{
"epoch": 0.40755762192106043,
"grad_norm": 0.16808223853144952,
"learning_rate": 9.961136948707752e-05,
"loss": 0.0118,
"step": 2060
},
{
"epoch": 0.40953605697892964,
"grad_norm": 0.14966917651227296,
"learning_rate": 9.959691606114393e-05,
"loss": 0.0119,
"step": 2070
},
{
"epoch": 0.4115144920367989,
"grad_norm": 0.16576027603766436,
"learning_rate": 9.958219984075441e-05,
"loss": 0.0116,
"step": 2080
},
{
"epoch": 0.4134929270946681,
"grad_norm": 0.13201379755549364,
"learning_rate": 9.956722090388438e-05,
"loss": 0.0113,
"step": 2090
},
{
"epoch": 0.4154713621525373,
"grad_norm": 0.1399999993709743,
"learning_rate": 9.955197932990132e-05,
"loss": 0.0113,
"step": 2100
},
{
"epoch": 0.4174497972104066,
"grad_norm": 0.15704807005212315,
"learning_rate": 9.953647519956433e-05,
"loss": 0.0114,
"step": 2110
},
{
"epoch": 0.4194282322682758,
"grad_norm": 0.14369096849347784,
"learning_rate": 9.952070859502361e-05,
"loss": 0.0113,
"step": 2120
},
{
"epoch": 0.421406667326145,
"grad_norm": 0.1431325790659893,
"learning_rate": 9.950467959982021e-05,
"loss": 0.0112,
"step": 2130
},
{
"epoch": 0.4233851023840142,
"grad_norm": 0.1291143217645934,
"learning_rate": 9.948838829888543e-05,
"loss": 0.011,
"step": 2140
},
{
"epoch": 0.4253635374418835,
"grad_norm": 0.14157885878823356,
"learning_rate": 9.947183477854042e-05,
"loss": 0.0111,
"step": 2150
},
{
"epoch": 0.4273419724997527,
"grad_norm": 0.17377404638170016,
"learning_rate": 9.945501912649579e-05,
"loss": 0.0108,
"step": 2160
},
{
"epoch": 0.4293204075576219,
"grad_norm": 0.12707883307617196,
"learning_rate": 9.943794143185096e-05,
"loss": 0.0108,
"step": 2170
},
{
"epoch": 0.4312988426154912,
"grad_norm": 0.14301379327253447,
"learning_rate": 9.942060178509396e-05,
"loss": 0.0109,
"step": 2180
},
{
"epoch": 0.4332772776733604,
"grad_norm": 0.13765310177597165,
"learning_rate": 9.940300027810067e-05,
"loss": 0.0107,
"step": 2190
},
{
"epoch": 0.4352557127312296,
"grad_norm": 0.1265382268710168,
"learning_rate": 9.938513700413454e-05,
"loss": 0.0105,
"step": 2200
},
{
"epoch": 0.4372341477890988,
"grad_norm": 0.14082965936667094,
"learning_rate": 9.9367012057846e-05,
"loss": 0.0105,
"step": 2210
},
{
"epoch": 0.43921258284696807,
"grad_norm": 0.13257078783884468,
"learning_rate": 9.934862553527198e-05,
"loss": 0.0105,
"step": 2220
},
{
"epoch": 0.4411910179048373,
"grad_norm": 0.13907806492688254,
"learning_rate": 9.932997753383537e-05,
"loss": 0.0104,
"step": 2230
},
{
"epoch": 0.4431694529627065,
"grad_norm": 0.12466793088917975,
"learning_rate": 9.931106815234459e-05,
"loss": 0.0104,
"step": 2240
},
{
"epoch": 0.4451478880205757,
"grad_norm": 0.14907419462845958,
"learning_rate": 9.929189749099296e-05,
"loss": 0.0106,
"step": 2250
},
{
"epoch": 0.44712632307844496,
"grad_norm": 0.1411640174312852,
"learning_rate": 9.927246565135826e-05,
"loss": 0.0103,
"step": 2260
},
{
"epoch": 0.4491047581363142,
"grad_norm": 0.12001937823367097,
"learning_rate": 9.92527727364021e-05,
"loss": 0.0101,
"step": 2270
},
{
"epoch": 0.4510831931941834,
"grad_norm": 0.13372810312081806,
"learning_rate": 9.923281885046949e-05,
"loss": 0.0102,
"step": 2280
},
{
"epoch": 0.45306162825205265,
"grad_norm": 0.1416610121062831,
"learning_rate": 9.921260409928814e-05,
"loss": 0.0103,
"step": 2290
},
{
"epoch": 0.45504006330992186,
"grad_norm": 0.12600073771582693,
"learning_rate": 9.919212858996804e-05,
"loss": 0.0101,
"step": 2300
},
{
"epoch": 0.45701849836779107,
"grad_norm": 0.13172926879715913,
"learning_rate": 9.917139243100087e-05,
"loss": 0.01,
"step": 2310
},
{
"epoch": 0.4589969334256603,
"grad_norm": 0.1369213495175742,
"learning_rate": 9.915039573225926e-05,
"loss": 0.0099,
"step": 2320
},
{
"epoch": 0.46097536848352955,
"grad_norm": 0.12156941746596098,
"learning_rate": 9.912913860499649e-05,
"loss": 0.0098,
"step": 2330
},
{
"epoch": 0.46295380354139876,
"grad_norm": 0.13601737926609253,
"learning_rate": 9.910762116184562e-05,
"loss": 0.0098,
"step": 2340
},
{
"epoch": 0.46493223859926797,
"grad_norm": 0.1436159433972993,
"learning_rate": 9.908584351681911e-05,
"loss": 0.0098,
"step": 2350
},
{
"epoch": 0.46691067365713723,
"grad_norm": 0.12316795660806547,
"learning_rate": 9.906380578530809e-05,
"loss": 0.0096,
"step": 2360
},
{
"epoch": 0.46888910871500644,
"grad_norm": 0.12357887141349247,
"learning_rate": 9.904150808408178e-05,
"loss": 0.0096,
"step": 2370
},
{
"epoch": 0.47086754377287565,
"grad_norm": 0.12108889135295853,
"learning_rate": 9.901895053128689e-05,
"loss": 0.0097,
"step": 2380
},
{
"epoch": 0.47284597883074486,
"grad_norm": 0.12666207059842866,
"learning_rate": 9.899613324644698e-05,
"loss": 0.0096,
"step": 2390
},
{
"epoch": 0.4748244138886141,
"grad_norm": 0.12204117589676883,
"learning_rate": 9.897305635046183e-05,
"loss": 0.0095,
"step": 2400
},
{
"epoch": 0.47680284894648334,
"grad_norm": 0.11286113484881569,
"learning_rate": 9.894971996560678e-05,
"loss": 0.0095,
"step": 2410
},
{
"epoch": 0.47878128400435255,
"grad_norm": 0.11557713188185322,
"learning_rate": 9.89261242155321e-05,
"loss": 0.0094,
"step": 2420
},
{
"epoch": 0.48075971906222176,
"grad_norm": 0.13322061066535654,
"learning_rate": 9.890226922526237e-05,
"loss": 0.0095,
"step": 2430
},
{
"epoch": 0.482738154120091,
"grad_norm": 0.12773567932628888,
"learning_rate": 9.887815512119574e-05,
"loss": 0.0095,
"step": 2440
},
{
"epoch": 0.48471658917796023,
"grad_norm": 0.13887970782774162,
"learning_rate": 9.885378203110331e-05,
"loss": 0.0095,
"step": 2450
},
{
"epoch": 0.48669502423582944,
"grad_norm": 0.10987030717218844,
"learning_rate": 9.882915008412847e-05,
"loss": 0.0093,
"step": 2460
},
{
"epoch": 0.4886734592936987,
"grad_norm": 0.10499917426827143,
"learning_rate": 9.880425941078617e-05,
"loss": 0.0093,
"step": 2470
},
{
"epoch": 0.4906518943515679,
"grad_norm": 0.14223242252473678,
"learning_rate": 9.877911014296223e-05,
"loss": 0.0092,
"step": 2480
},
{
"epoch": 0.49263032940943713,
"grad_norm": 0.1089284507102967,
"learning_rate": 9.875370241391269e-05,
"loss": 0.0092,
"step": 2490
},
{
"epoch": 0.49460876446730634,
"grad_norm": 0.12386900212544813,
"learning_rate": 9.872803635826307e-05,
"loss": 0.009,
"step": 2500
},
{
"epoch": 0.4965871995251756,
"grad_norm": 0.11902748524120127,
"learning_rate": 9.870211211200766e-05,
"loss": 0.0088,
"step": 2510
},
{
"epoch": 0.4985656345830448,
"grad_norm": 0.13804362869132278,
"learning_rate": 9.867592981250878e-05,
"loss": 0.009,
"step": 2520
},
{
"epoch": 0.500544069640914,
"grad_norm": 0.17786582032115625,
"learning_rate": 9.864948959849609e-05,
"loss": 0.0092,
"step": 2530
},
{
"epoch": 0.5025225046987832,
"grad_norm": 0.1479464905058761,
"learning_rate": 9.862279161006585e-05,
"loss": 0.0091,
"step": 2540
},
{
"epoch": 0.5045009397566524,
"grad_norm": 0.1127130413803928,
"learning_rate": 9.859583598868012e-05,
"loss": 0.009,
"step": 2550
},
{
"epoch": 0.5064793748145218,
"grad_norm": 0.11795012464385328,
"learning_rate": 9.856862287716613e-05,
"loss": 0.0089,
"step": 2560
},
{
"epoch": 0.508457809872391,
"grad_norm": 0.1187241172057776,
"learning_rate": 9.854115241971535e-05,
"loss": 0.0088,
"step": 2570
},
{
"epoch": 0.5104362449302602,
"grad_norm": 0.11702810745118249,
"learning_rate": 9.85134247618829e-05,
"loss": 0.0089,
"step": 2580
},
{
"epoch": 0.5124146799881294,
"grad_norm": 0.13143294794621907,
"learning_rate": 9.848544005058668e-05,
"loss": 0.0086,
"step": 2590
},
{
"epoch": 0.5143931150459986,
"grad_norm": 0.11228694387667242,
"learning_rate": 9.845719843410663e-05,
"loss": 0.0087,
"step": 2600
},
{
"epoch": 0.5163715501038678,
"grad_norm": 0.13848703274101368,
"learning_rate": 9.842870006208389e-05,
"loss": 0.0086,
"step": 2610
},
{
"epoch": 0.518349985161737,
"grad_norm": 0.13592484040648178,
"learning_rate": 9.839994508552006e-05,
"loss": 0.0086,
"step": 2620
},
{
"epoch": 0.5203284202196063,
"grad_norm": 0.11618883084830398,
"learning_rate": 9.837093365677644e-05,
"loss": 0.0085,
"step": 2630
},
{
"epoch": 0.5223068552774756,
"grad_norm": 0.10259762776414111,
"learning_rate": 9.83416659295731e-05,
"loss": 0.0085,
"step": 2640
},
{
"epoch": 0.5242852903353448,
"grad_norm": 0.1316745995816759,
"learning_rate": 9.831214205898814e-05,
"loss": 0.0086,
"step": 2650
},
{
"epoch": 0.526263725393214,
"grad_norm": 0.11806848092795394,
"learning_rate": 9.82823622014569e-05,
"loss": 0.0085,
"step": 2660
},
{
"epoch": 0.5282421604510832,
"grad_norm": 0.1203492280099884,
"learning_rate": 9.825232651477109e-05,
"loss": 0.0085,
"step": 2670
},
{
"epoch": 0.5302205955089524,
"grad_norm": 0.10874728446824328,
"learning_rate": 9.82220351580779e-05,
"loss": 0.0083,
"step": 2680
},
{
"epoch": 0.5321990305668216,
"grad_norm": 0.10676879787290053,
"learning_rate": 9.819148829187932e-05,
"loss": 0.0085,
"step": 2690
},
{
"epoch": 0.5341774656246908,
"grad_norm": 0.11702157099605656,
"learning_rate": 9.816068607803107e-05,
"loss": 0.0082,
"step": 2700
},
{
"epoch": 0.5361559006825601,
"grad_norm": 0.11105417188799853,
"learning_rate": 9.812962867974191e-05,
"loss": 0.0084,
"step": 2710
},
{
"epoch": 0.5381343357404293,
"grad_norm": 0.1116345983572604,
"learning_rate": 9.809831626157274e-05,
"loss": 0.0082,
"step": 2720
},
{
"epoch": 0.5401127707982986,
"grad_norm": 0.10809782779937041,
"learning_rate": 9.806674898943569e-05,
"loss": 0.0081,
"step": 2730
},
{
"epoch": 0.5420912058561678,
"grad_norm": 0.11301172939493037,
"learning_rate": 9.803492703059323e-05,
"loss": 0.0082,
"step": 2740
},
{
"epoch": 0.544069640914037,
"grad_norm": 0.10945048282187327,
"learning_rate": 9.800285055365736e-05,
"loss": 0.0084,
"step": 2750
},
{
"epoch": 0.5460480759719062,
"grad_norm": 0.1130153961206554,
"learning_rate": 9.797051972858867e-05,
"loss": 0.0082,
"step": 2760
},
{
"epoch": 0.5480265110297754,
"grad_norm": 0.10116713488780331,
"learning_rate": 9.79379347266954e-05,
"loss": 0.0081,
"step": 2770
},
{
"epoch": 0.5500049460876447,
"grad_norm": 0.09573931812955483,
"learning_rate": 9.79050957206326e-05,
"loss": 0.0082,
"step": 2780
},
{
"epoch": 0.5519833811455139,
"grad_norm": 0.12457809740640126,
"learning_rate": 9.787200288440119e-05,
"loss": 0.0082,
"step": 2790
},
{
"epoch": 0.5539618162033831,
"grad_norm": 0.09611529181251109,
"learning_rate": 9.783865639334705e-05,
"loss": 0.0082,
"step": 2800
},
{
"epoch": 0.5559402512612523,
"grad_norm": 0.10372741899823032,
"learning_rate": 9.780505642416001e-05,
"loss": 0.0081,
"step": 2810
},
{
"epoch": 0.5579186863191216,
"grad_norm": 0.10860018971843517,
"learning_rate": 9.777120315487306e-05,
"loss": 0.008,
"step": 2820
},
{
"epoch": 0.5598971213769908,
"grad_norm": 0.10147808865925607,
"learning_rate": 9.773709676486129e-05,
"loss": 0.008,
"step": 2830
},
{
"epoch": 0.56187555643486,
"grad_norm": 0.10320091768483318,
"learning_rate": 9.770273743484097e-05,
"loss": 0.0079,
"step": 2840
},
{
"epoch": 0.5638539914927293,
"grad_norm": 0.1123966678473519,
"learning_rate": 9.766812534686863e-05,
"loss": 0.0078,
"step": 2850
},
{
"epoch": 0.5658324265505985,
"grad_norm": 0.10082372136797269,
"learning_rate": 9.763326068434e-05,
"loss": 0.0078,
"step": 2860
},
{
"epoch": 0.5678108616084677,
"grad_norm": 0.11124059319263015,
"learning_rate": 9.75981436319892e-05,
"loss": 0.0078,
"step": 2870
},
{
"epoch": 0.5697892966663369,
"grad_norm": 0.09801225165766604,
"learning_rate": 9.756277437588759e-05,
"loss": 0.0077,
"step": 2880
},
{
"epoch": 0.5717677317242061,
"grad_norm": 0.09992038113390543,
"learning_rate": 9.752715310344288e-05,
"loss": 0.0077,
"step": 2890
},
{
"epoch": 0.5737461667820754,
"grad_norm": 0.09870244684241487,
"learning_rate": 9.74912800033981e-05,
"loss": 0.0078,
"step": 2900
},
{
"epoch": 0.5757246018399446,
"grad_norm": 0.09535814395990586,
"learning_rate": 9.745515526583065e-05,
"loss": 0.0077,
"step": 2910
},
{
"epoch": 0.5777030368978139,
"grad_norm": 0.14553012896395245,
"learning_rate": 9.741877908215123e-05,
"loss": 0.0079,
"step": 2920
},
{
"epoch": 0.5796814719556831,
"grad_norm": 0.10043959284050716,
"learning_rate": 9.738215164510286e-05,
"loss": 0.0078,
"step": 2930
},
{
"epoch": 0.5816599070135523,
"grad_norm": 0.09954688262786084,
"learning_rate": 9.734527314875985e-05,
"loss": 0.0079,
"step": 2940
},
{
"epoch": 0.5836383420714215,
"grad_norm": 0.12156383753192526,
"learning_rate": 9.730814378852678e-05,
"loss": 0.0079,
"step": 2950
},
{
"epoch": 0.5856167771292907,
"grad_norm": 0.10532029280879827,
"learning_rate": 9.727076376113745e-05,
"loss": 0.0079,
"step": 2960
},
{
"epoch": 0.5875952121871599,
"grad_norm": 0.11122784045807846,
"learning_rate": 9.723313326465384e-05,
"loss": 0.0077,
"step": 2970
},
{
"epoch": 0.5895736472450291,
"grad_norm": 0.10363430784725526,
"learning_rate": 9.71952524984651e-05,
"loss": 0.0076,
"step": 2980
},
{
"epoch": 0.5915520823028984,
"grad_norm": 1.9791612899166353,
"learning_rate": 9.715712166328644e-05,
"loss": 0.0738,
"step": 2990
},
{
"epoch": 0.5935305173607677,
"grad_norm": 0.33249499212056244,
"learning_rate": 9.711874096115805e-05,
"loss": 0.0416,
"step": 3000
},
{
"epoch": 0.5955089524186369,
"grad_norm": 0.6363427991171072,
"learning_rate": 9.708011059544415e-05,
"loss": 0.0156,
"step": 3010
},
{
"epoch": 0.5974873874765061,
"grad_norm": 0.11169367784507092,
"learning_rate": 9.704123077083176e-05,
"loss": 0.0104,
"step": 3020
},
{
"epoch": 0.5994658225343753,
"grad_norm": 0.10944845384274775,
"learning_rate": 9.700210169332968e-05,
"loss": 0.0091,
"step": 3030
},
{
"epoch": 0.6014442575922445,
"grad_norm": 0.08558371324106481,
"learning_rate": 9.696272357026746e-05,
"loss": 0.0085,
"step": 3040
},
{
"epoch": 0.6034226926501137,
"grad_norm": 0.10013853375803404,
"learning_rate": 9.692309661029421e-05,
"loss": 0.0082,
"step": 3050
},
{
"epoch": 0.6054011277079829,
"grad_norm": 0.09845792141125358,
"learning_rate": 9.68832210233775e-05,
"loss": 0.0081,
"step": 3060
},
{
"epoch": 0.6073795627658523,
"grad_norm": 0.09501224521899478,
"learning_rate": 9.684309702080233e-05,
"loss": 0.0079,
"step": 3070
},
{
"epoch": 0.6093579978237215,
"grad_norm": 0.08761270510034475,
"learning_rate": 9.680272481516992e-05,
"loss": 0.0078,
"step": 3080
},
{
"epoch": 0.6113364328815907,
"grad_norm": 0.08274578764622441,
"learning_rate": 9.676210462039664e-05,
"loss": 0.0077,
"step": 3090
},
{
"epoch": 0.6133148679394599,
"grad_norm": 0.09099628298861405,
"learning_rate": 9.672123665171285e-05,
"loss": 0.0077,
"step": 3100
},
{
"epoch": 0.6152933029973291,
"grad_norm": 0.08896116288013423,
"learning_rate": 9.668012112566174e-05,
"loss": 0.0077,
"step": 3110
},
{
"epoch": 0.6172717380551983,
"grad_norm": 0.08875923423810328,
"learning_rate": 9.663875826009824e-05,
"loss": 0.0076,
"step": 3120
},
{
"epoch": 0.6192501731130675,
"grad_norm": 0.08482396985171677,
"learning_rate": 9.659714827418786e-05,
"loss": 0.0073,
"step": 3130
},
{
"epoch": 0.6212286081709368,
"grad_norm": 0.08025221535296631,
"learning_rate": 9.655529138840543e-05,
"loss": 0.0074,
"step": 3140
},
{
"epoch": 0.623207043228806,
"grad_norm": 0.08574096671384178,
"learning_rate": 9.651318782453407e-05,
"loss": 0.0074,
"step": 3150
},
{
"epoch": 0.6251854782866753,
"grad_norm": 0.08722130485524873,
"learning_rate": 9.64708378056639e-05,
"loss": 0.0072,
"step": 3160
},
{
"epoch": 0.6271639133445445,
"grad_norm": 0.0734150187550043,
"learning_rate": 9.642824155619095e-05,
"loss": 0.0072,
"step": 3170
},
{
"epoch": 0.6291423484024137,
"grad_norm": 0.08275773960040779,
"learning_rate": 9.638539930181592e-05,
"loss": 0.0072,
"step": 3180
},
{
"epoch": 0.6311207834602829,
"grad_norm": 0.07852787709316977,
"learning_rate": 9.634231126954296e-05,
"loss": 0.0073,
"step": 3190
},
{
"epoch": 0.6330992185181521,
"grad_norm": 0.08846367509899719,
"learning_rate": 9.629897768767853e-05,
"loss": 0.0072,
"step": 3200
},
{
"epoch": 0.6350776535760214,
"grad_norm": 0.08489879464653804,
"learning_rate": 9.625539878583016e-05,
"loss": 0.0072,
"step": 3210
},
{
"epoch": 0.6370560886338906,
"grad_norm": 0.07261154147793239,
"learning_rate": 9.621157479490521e-05,
"loss": 0.0072,
"step": 3220
},
{
"epoch": 0.6390345236917598,
"grad_norm": 0.07974019430917753,
"learning_rate": 9.616750594710973e-05,
"loss": 0.0071,
"step": 3230
},
{
"epoch": 0.641012958749629,
"grad_norm": 0.07263826747062917,
"learning_rate": 9.612319247594707e-05,
"loss": 0.007,
"step": 3240
},
{
"epoch": 0.6429913938074983,
"grad_norm": 0.09671443059620426,
"learning_rate": 9.607863461621685e-05,
"loss": 0.007,
"step": 3250
},
{
"epoch": 0.6449698288653675,
"grad_norm": 0.08478277389943766,
"learning_rate": 9.603383260401352e-05,
"loss": 0.0072,
"step": 3260
},
{
"epoch": 0.6469482639232367,
"grad_norm": 0.07758836162131409,
"learning_rate": 9.598878667672526e-05,
"loss": 0.007,
"step": 3270
},
{
"epoch": 0.648926698981106,
"grad_norm": 0.08272995261354078,
"learning_rate": 9.594349707303262e-05,
"loss": 0.0069,
"step": 3280
},
{
"epoch": 0.6509051340389752,
"grad_norm": 0.07973060907851731,
"learning_rate": 9.589796403290732e-05,
"loss": 0.0069,
"step": 3290
},
{
"epoch": 0.6528835690968444,
"grad_norm": 0.07093674420545966,
"learning_rate": 9.585218779761093e-05,
"loss": 0.0068,
"step": 3300
},
{
"epoch": 0.6548620041547136,
"grad_norm": 0.06913411096953943,
"learning_rate": 9.580616860969364e-05,
"loss": 0.0067,
"step": 3310
},
{
"epoch": 0.6568404392125828,
"grad_norm": 0.06931061663072934,
"learning_rate": 9.575990671299295e-05,
"loss": 0.0068,
"step": 3320
},
{
"epoch": 0.658818874270452,
"grad_norm": 0.08243282586915512,
"learning_rate": 9.571340235263236e-05,
"loss": 0.007,
"step": 3330
},
{
"epoch": 0.6607973093283213,
"grad_norm": 0.07631827658066849,
"learning_rate": 9.566665577502008e-05,
"loss": 0.0069,
"step": 3340
},
{
"epoch": 0.6627757443861905,
"grad_norm": 0.08020496809113337,
"learning_rate": 9.561966722784774e-05,
"loss": 0.0068,
"step": 3350
},
{
"epoch": 0.6647541794440598,
"grad_norm": 0.0656960620994247,
"learning_rate": 9.557243696008912e-05,
"loss": 0.0067,
"step": 3360
},
{
"epoch": 0.666732614501929,
"grad_norm": 0.08361766659034513,
"learning_rate": 9.552496522199872e-05,
"loss": 0.0067,
"step": 3370
},
{
"epoch": 0.6687110495597982,
"grad_norm": 0.07446598640403798,
"learning_rate": 9.54772522651105e-05,
"loss": 0.0067,
"step": 3380
},
{
"epoch": 0.6706894846176674,
"grad_norm": 0.07947178181210834,
"learning_rate": 9.54292983422366e-05,
"loss": 0.0066,
"step": 3390
},
{
"epoch": 0.6726679196755366,
"grad_norm": 0.08383495020962163,
"learning_rate": 9.538110370746587e-05,
"loss": 0.0067,
"step": 3400
},
{
"epoch": 0.6746463547334058,
"grad_norm": 0.07200727112961013,
"learning_rate": 9.533266861616264e-05,
"loss": 0.0067,
"step": 3410
},
{
"epoch": 0.676624789791275,
"grad_norm": 0.07580548430145595,
"learning_rate": 9.528399332496533e-05,
"loss": 0.0066,
"step": 3420
},
{
"epoch": 0.6786032248491444,
"grad_norm": 0.08917135842105278,
"learning_rate": 9.523507809178506e-05,
"loss": 0.0066,
"step": 3430
},
{
"epoch": 0.6805816599070136,
"grad_norm": 0.09348688177706456,
"learning_rate": 9.518592317580432e-05,
"loss": 0.0066,
"step": 3440
},
{
"epoch": 0.6825600949648828,
"grad_norm": 0.07982621923070499,
"learning_rate": 9.513652883747558e-05,
"loss": 0.0066,
"step": 3450
},
{
"epoch": 0.684538530022752,
"grad_norm": 0.07410164219157499,
"learning_rate": 9.508689533851991e-05,
"loss": 0.0066,
"step": 3460
},
{
"epoch": 0.6865169650806212,
"grad_norm": 0.06852760263521068,
"learning_rate": 9.503702294192563e-05,
"loss": 0.0065,
"step": 3470
},
{
"epoch": 0.6884954001384904,
"grad_norm": 0.07022509922084559,
"learning_rate": 9.498691191194682e-05,
"loss": 0.0066,
"step": 3480
},
{
"epoch": 0.6904738351963596,
"grad_norm": 0.08425781544919872,
"learning_rate": 9.493656251410204e-05,
"loss": 0.0066,
"step": 3490
},
{
"epoch": 0.692452270254229,
"grad_norm": 0.07319082992630636,
"learning_rate": 9.488597501517287e-05,
"loss": 0.0066,
"step": 3500
},
{
"epoch": 0.6944307053120982,
"grad_norm": 0.06724749413870058,
"learning_rate": 9.483514968320243e-05,
"loss": 0.0065,
"step": 3510
},
{
"epoch": 0.6964091403699674,
"grad_norm": 0.06473946157875621,
"learning_rate": 9.478408678749412e-05,
"loss": 0.0064,
"step": 3520
},
{
"epoch": 0.6983875754278366,
"grad_norm": 0.065302161234294,
"learning_rate": 9.473278659861e-05,
"loss": 0.0065,
"step": 3530
},
{
"epoch": 0.7003660104857058,
"grad_norm": 0.0645433505038465,
"learning_rate": 9.46812493883695e-05,
"loss": 0.0064,
"step": 3540
},
{
"epoch": 0.702344445543575,
"grad_norm": 0.06621536065179888,
"learning_rate": 9.462947542984795e-05,
"loss": 0.0064,
"step": 3550
},
{
"epoch": 0.7043228806014442,
"grad_norm": 0.06461586763701248,
"learning_rate": 9.457746499737508e-05,
"loss": 0.0065,
"step": 3560
},
{
"epoch": 0.7063013156593135,
"grad_norm": 0.06664972077141973,
"learning_rate": 9.45252183665336e-05,
"loss": 0.0064,
"step": 3570
},
{
"epoch": 0.7082797507171827,
"grad_norm": 0.07494743504615466,
"learning_rate": 9.44727358141578e-05,
"loss": 0.0065,
"step": 3580
},
{
"epoch": 0.710258185775052,
"grad_norm": 0.07721049683526093,
"learning_rate": 9.442001761833194e-05,
"loss": 0.0065,
"step": 3590
},
{
"epoch": 0.7122366208329212,
"grad_norm": 0.0771302625844104,
"learning_rate": 9.436706405838892e-05,
"loss": 0.0064,
"step": 3600
},
{
"epoch": 0.7142150558907904,
"grad_norm": 0.062417889604423196,
"learning_rate": 9.431387541490872e-05,
"loss": 0.0063,
"step": 3610
},
{
"epoch": 0.7161934909486596,
"grad_norm": 0.054946677778002946,
"learning_rate": 9.426045196971695e-05,
"loss": 0.0063,
"step": 3620
},
{
"epoch": 0.7181719260065288,
"grad_norm": 0.06722951793272056,
"learning_rate": 9.420679400588334e-05,
"loss": 0.0063,
"step": 3630
},
{
"epoch": 0.7201503610643981,
"grad_norm": 0.06370642015482036,
"learning_rate": 9.41529018077202e-05,
"loss": 0.0063,
"step": 3640
},
{
"epoch": 0.7221287961222673,
"grad_norm": 0.07957638627416937,
"learning_rate": 9.409877566078101e-05,
"loss": 0.0063,
"step": 3650
},
{
"epoch": 0.7241072311801365,
"grad_norm": 0.07321813903269636,
"learning_rate": 9.404441585185883e-05,
"loss": 0.0063,
"step": 3660
},
{
"epoch": 0.7260856662380057,
"grad_norm": 0.07111333434355711,
"learning_rate": 9.398982266898481e-05,
"loss": 0.0063,
"step": 3670
},
{
"epoch": 0.728064101295875,
"grad_norm": 0.07481536876192033,
"learning_rate": 9.393499640142662e-05,
"loss": 0.0064,
"step": 3680
},
{
"epoch": 0.7300425363537442,
"grad_norm": 0.060891954106065224,
"learning_rate": 9.387993733968702e-05,
"loss": 0.0062,
"step": 3690
},
{
"epoch": 0.7320209714116134,
"grad_norm": 0.07230775898494173,
"learning_rate": 9.382464577550218e-05,
"loss": 0.0063,
"step": 3700
},
{
"epoch": 0.7339994064694826,
"grad_norm": 0.06562749379265283,
"learning_rate": 9.376912200184029e-05,
"loss": 0.0063,
"step": 3710
},
{
"epoch": 0.7359778415273519,
"grad_norm": 0.06584883346618361,
"learning_rate": 9.371336631289984e-05,
"loss": 0.0063,
"step": 3720
},
{
"epoch": 0.7379562765852211,
"grad_norm": 0.06693297403000101,
"learning_rate": 9.365737900410821e-05,
"loss": 0.0063,
"step": 3730
},
{
"epoch": 0.7399347116430903,
"grad_norm": 0.06927594449263381,
"learning_rate": 9.360116037212002e-05,
"loss": 0.0063,
"step": 3740
},
{
"epoch": 0.7419131467009595,
"grad_norm": 0.07024182461941782,
"learning_rate": 9.354471071481557e-05,
"loss": 0.0063,
"step": 3750
},
{
"epoch": 0.7438915817588287,
"grad_norm": 0.059150396082578924,
"learning_rate": 9.348803033129929e-05,
"loss": 0.0062,
"step": 3760
},
{
"epoch": 0.745870016816698,
"grad_norm": 0.07940128446257239,
"learning_rate": 9.343111952189814e-05,
"loss": 0.0067,
"step": 3770
},
{
"epoch": 0.7478484518745672,
"grad_norm": 0.1905347171436949,
"learning_rate": 9.337397858816e-05,
"loss": 0.0067,
"step": 3780
},
{
"epoch": 0.7498268869324365,
"grad_norm": 0.6061998486250697,
"learning_rate": 9.331660783285208e-05,
"loss": 0.0124,
"step": 3790
},
{
"epoch": 0.7518053219903057,
"grad_norm": 0.1258837853994195,
"learning_rate": 9.325900755995938e-05,
"loss": 0.0095,
"step": 3800
},
{
"epoch": 0.7537837570481749,
"grad_norm": 0.0727378209583613,
"learning_rate": 9.320117807468292e-05,
"loss": 0.0075,
"step": 3810
},
{
"epoch": 0.7557621921060441,
"grad_norm": 0.08534436558727057,
"learning_rate": 9.314311968343833e-05,
"loss": 0.0067,
"step": 3820
},
{
"epoch": 0.7577406271639133,
"grad_norm": 0.9380709749756719,
"learning_rate": 9.308483269385406e-05,
"loss": 0.0087,
"step": 3830
},
{
"epoch": 0.7597190622217825,
"grad_norm": 0.1775752299868515,
"learning_rate": 9.302631741476984e-05,
"loss": 0.0109,
"step": 3840
},
{
"epoch": 0.7616974972796517,
"grad_norm": 0.08354447847852846,
"learning_rate": 9.2967574156235e-05,
"loss": 0.0083,
"step": 3850
},
{
"epoch": 0.7636759323375211,
"grad_norm": 0.08170636335755797,
"learning_rate": 9.290860322950687e-05,
"loss": 0.0072,
"step": 3860
},
{
"epoch": 0.7656543673953903,
"grad_norm": 0.07503365800401021,
"learning_rate": 9.284940494704906e-05,
"loss": 0.0068,
"step": 3870
},
{
"epoch": 0.7676328024532595,
"grad_norm": 0.08751262611483802,
"learning_rate": 9.278997962252986e-05,
"loss": 0.0067,
"step": 3880
},
{
"epoch": 0.7696112375111287,
"grad_norm": 0.0643986915522455,
"learning_rate": 9.273032757082062e-05,
"loss": 0.0065,
"step": 3890
},
{
"epoch": 0.7715896725689979,
"grad_norm": 0.0694722501084022,
"learning_rate": 9.267044910799395e-05,
"loss": 0.0065,
"step": 3900
},
{
"epoch": 0.7735681076268671,
"grad_norm": 0.06320928291707667,
"learning_rate": 9.261034455132218e-05,
"loss": 0.0065,
"step": 3910
},
{
"epoch": 0.7755465426847363,
"grad_norm": 0.059612881648971584,
"learning_rate": 9.255001421927558e-05,
"loss": 0.0065,
"step": 3920
},
{
"epoch": 0.7775249777426056,
"grad_norm": 0.06129585573339859,
"learning_rate": 9.248945843152075e-05,
"loss": 0.0062,
"step": 3930
},
{
"epoch": 0.7795034128004749,
"grad_norm": 0.05949962038028326,
"learning_rate": 9.242867750891887e-05,
"loss": 0.0062,
"step": 3940
},
{
"epoch": 0.7814818478583441,
"grad_norm": 0.06148268871166804,
"learning_rate": 9.236767177352403e-05,
"loss": 0.0061,
"step": 3950
},
{
"epoch": 0.7834602829162133,
"grad_norm": 0.06850010112899138,
"learning_rate": 9.230644154858152e-05,
"loss": 0.0061,
"step": 3960
},
{
"epoch": 0.7854387179740825,
"grad_norm": 0.06198631327573319,
"learning_rate": 9.224498715852607e-05,
"loss": 0.0061,
"step": 3970
},
{
"epoch": 0.7874171530319517,
"grad_norm": 0.06418108872128661,
"learning_rate": 9.218330892898024e-05,
"loss": 0.0061,
"step": 3980
},
{
"epoch": 0.7893955880898209,
"grad_norm": 0.06014633324181408,
"learning_rate": 9.212140718675257e-05,
"loss": 0.0061,
"step": 3990
},
{
"epoch": 0.7913740231476902,
"grad_norm": 0.050895513285315275,
"learning_rate": 9.205928225983593e-05,
"loss": 0.0061,
"step": 4000
},
{
"epoch": 0.7933524582055594,
"grad_norm": 0.056344806495317715,
"learning_rate": 9.199693447740577e-05,
"loss": 0.0061,
"step": 4010
},
{
"epoch": 0.7953308932634287,
"grad_norm": 0.0567846189542034,
"learning_rate": 9.193436416981833e-05,
"loss": 0.0062,
"step": 4020
},
{
"epoch": 0.7973093283212979,
"grad_norm": 0.05291176992845361,
"learning_rate": 9.187157166860894e-05,
"loss": 0.006,
"step": 4030
},
{
"epoch": 0.7992877633791671,
"grad_norm": 0.06336161577717206,
"learning_rate": 9.180855730649025e-05,
"loss": 0.0059,
"step": 4040
},
{
"epoch": 0.8012661984370363,
"grad_norm": 0.05390528708739402,
"learning_rate": 9.174532141735045e-05,
"loss": 0.0059,
"step": 4050
},
{
"epoch": 0.8032446334949055,
"grad_norm": 0.057088492185258743,
"learning_rate": 9.168186433625156e-05,
"loss": 0.006,
"step": 4060
},
{
"epoch": 0.8052230685527747,
"grad_norm": 0.05710566257077743,
"learning_rate": 9.161818639942753e-05,
"loss": 0.0059,
"step": 4070
},
{
"epoch": 0.807201503610644,
"grad_norm": 0.0504202356540121,
"learning_rate": 9.155428794428258e-05,
"loss": 0.0059,
"step": 4080
},
{
"epoch": 0.8091799386685132,
"grad_norm": 0.05776764292317763,
"learning_rate": 9.14901693093894e-05,
"loss": 0.006,
"step": 4090
},
{
"epoch": 0.8111583737263824,
"grad_norm": 0.057237936957186215,
"learning_rate": 9.14258308344873e-05,
"loss": 0.0059,
"step": 4100
},
{
"epoch": 0.8131368087842517,
"grad_norm": 0.053832097684184975,
"learning_rate": 9.136127286048038e-05,
"loss": 0.006,
"step": 4110
},
{
"epoch": 0.8151152438421209,
"grad_norm": 0.05720418064991198,
"learning_rate": 9.129649572943587e-05,
"loss": 0.006,
"step": 4120
},
{
"epoch": 0.8170936788999901,
"grad_norm": 0.057875409325671534,
"learning_rate": 9.123149978458216e-05,
"loss": 0.0059,
"step": 4130
},
{
"epoch": 0.8190721139578593,
"grad_norm": 0.06088402107697045,
"learning_rate": 9.116628537030711e-05,
"loss": 0.0058,
"step": 4140
},
{
"epoch": 0.8210505490157286,
"grad_norm": 0.060371580920135476,
"learning_rate": 9.11008528321561e-05,
"loss": 0.0059,
"step": 4150
},
{
"epoch": 0.8230289840735978,
"grad_norm": 0.07281581558960042,
"learning_rate": 9.103520251683029e-05,
"loss": 0.0059,
"step": 4160
},
{
"epoch": 0.825007419131467,
"grad_norm": 0.08503219195645913,
"learning_rate": 9.096933477218475e-05,
"loss": 0.006,
"step": 4170
},
{
"epoch": 0.8269858541893362,
"grad_norm": 0.052294168407101764,
"learning_rate": 9.090324994722663e-05,
"loss": 0.0058,
"step": 4180
},
{
"epoch": 0.8289642892472054,
"grad_norm": 0.053160237974548545,
"learning_rate": 9.083694839211329e-05,
"loss": 0.0058,
"step": 4190
},
{
"epoch": 0.8309427243050747,
"grad_norm": 0.06315834443397667,
"learning_rate": 9.07704304581505e-05,
"loss": 0.0059,
"step": 4200
},
{
"epoch": 0.8329211593629439,
"grad_norm": 0.06331923652121475,
"learning_rate": 9.070369649779047e-05,
"loss": 0.0059,
"step": 4210
},
{
"epoch": 0.8348995944208132,
"grad_norm": 0.04923613089320947,
"learning_rate": 9.063674686463008e-05,
"loss": 0.0059,
"step": 4220
},
{
"epoch": 0.8368780294786824,
"grad_norm": 0.052338047560489,
"learning_rate": 9.056958191340901e-05,
"loss": 0.0058,
"step": 4230
},
{
"epoch": 0.8388564645365516,
"grad_norm": 0.05258170327574389,
"learning_rate": 9.050220200000774e-05,
"loss": 0.0058,
"step": 4240
},
{
"epoch": 0.8408348995944208,
"grad_norm": 0.06392282804688217,
"learning_rate": 9.043460748144584e-05,
"loss": 0.0057,
"step": 4250
},
{
"epoch": 0.84281333465229,
"grad_norm": 0.05866588737784968,
"learning_rate": 9.036679871587992e-05,
"loss": 0.0058,
"step": 4260
},
{
"epoch": 0.8447917697101592,
"grad_norm": 0.047982007299136466,
"learning_rate": 9.029877606260186e-05,
"loss": 0.0057,
"step": 4270
},
{
"epoch": 0.8467702047680284,
"grad_norm": 0.0481439483078887,
"learning_rate": 9.023053988203679e-05,
"loss": 0.0056,
"step": 4280
},
{
"epoch": 0.8487486398258978,
"grad_norm": 0.05933704536249277,
"learning_rate": 9.016209053574127e-05,
"loss": 0.0058,
"step": 4290
},
{
"epoch": 0.850727074883767,
"grad_norm": 0.04849073032706896,
"learning_rate": 9.009342838640134e-05,
"loss": 0.0057,
"step": 4300
},
{
"epoch": 0.8527055099416362,
"grad_norm": 0.0505850899657574,
"learning_rate": 9.002455379783057e-05,
"loss": 0.0057,
"step": 4310
},
{
"epoch": 0.8546839449995054,
"grad_norm": 0.04913108593749979,
"learning_rate": 8.99554671349682e-05,
"loss": 0.0056,
"step": 4320
},
{
"epoch": 0.8566623800573746,
"grad_norm": 0.04862354828441595,
"learning_rate": 8.988616876387717e-05,
"loss": 0.0057,
"step": 4330
},
{
"epoch": 0.8586408151152438,
"grad_norm": 0.06972418460787164,
"learning_rate": 8.98166590517421e-05,
"loss": 0.0058,
"step": 4340
},
{
"epoch": 0.860619250173113,
"grad_norm": 0.056146887280498346,
"learning_rate": 8.974693836686754e-05,
"loss": 0.0057,
"step": 4350
},
{
"epoch": 0.8625976852309823,
"grad_norm": 0.04812819285856473,
"learning_rate": 8.967700707867584e-05,
"loss": 0.0057,
"step": 4360
},
{
"epoch": 0.8645761202888516,
"grad_norm": 0.044773594452000846,
"learning_rate": 8.960686555770523e-05,
"loss": 0.0056,
"step": 4370
},
{
"epoch": 0.8665545553467208,
"grad_norm": 0.04840383299594874,
"learning_rate": 8.953651417560791e-05,
"loss": 0.0058,
"step": 4380
},
{
"epoch": 0.86853299040459,
"grad_norm": 0.04970761171149387,
"learning_rate": 8.946595330514806e-05,
"loss": 0.0057,
"step": 4390
},
{
"epoch": 0.8705114254624592,
"grad_norm": 0.04894324636554783,
"learning_rate": 8.939518332019985e-05,
"loss": 0.0057,
"step": 4400
},
{
"epoch": 0.8724898605203284,
"grad_norm": 0.04853103103678262,
"learning_rate": 8.932420459574544e-05,
"loss": 0.0057,
"step": 4410
},
{
"epoch": 0.8744682955781976,
"grad_norm": 0.07762326562453636,
"learning_rate": 8.925301750787306e-05,
"loss": 0.0057,
"step": 4420
},
{
"epoch": 0.8764467306360668,
"grad_norm": 0.05054879529912945,
"learning_rate": 8.918162243377493e-05,
"loss": 0.0056,
"step": 4430
},
{
"epoch": 0.8784251656939361,
"grad_norm": 0.05478832570194223,
"learning_rate": 8.911001975174536e-05,
"loss": 0.0057,
"step": 4440
},
{
"epoch": 0.8804036007518053,
"grad_norm": 0.04657812508941677,
"learning_rate": 8.903820984117864e-05,
"loss": 0.0057,
"step": 4450
},
{
"epoch": 0.8823820358096746,
"grad_norm": 0.052093793523986735,
"learning_rate": 8.896619308256711e-05,
"loss": 0.0057,
"step": 4460
},
{
"epoch": 0.8843604708675438,
"grad_norm": 0.04394197204665351,
"learning_rate": 8.88939698574991e-05,
"loss": 0.0057,
"step": 4470
},
{
"epoch": 0.886338905925413,
"grad_norm": 0.05656331821710276,
"learning_rate": 8.882154054865695e-05,
"loss": 0.0056,
"step": 4480
},
{
"epoch": 0.8883173409832822,
"grad_norm": 0.049449670379517814,
"learning_rate": 8.874890553981492e-05,
"loss": 0.0056,
"step": 4490
},
{
"epoch": 0.8902957760411514,
"grad_norm": 0.05150676354280675,
"learning_rate": 8.867606521583725e-05,
"loss": 0.0057,
"step": 4500
},
{
"epoch": 0.8922742110990207,
"grad_norm": 0.0498183975208156,
"learning_rate": 8.860301996267602e-05,
"loss": 0.0055,
"step": 4510
},
{
"epoch": 0.8942526461568899,
"grad_norm": 0.04934528183420388,
"learning_rate": 8.852977016736912e-05,
"loss": 0.0056,
"step": 4520
},
{
"epoch": 0.8962310812147591,
"grad_norm": 0.06325224717534036,
"learning_rate": 8.84563162180383e-05,
"loss": 0.0056,
"step": 4530
},
{
"epoch": 0.8982095162726283,
"grad_norm": 0.062115892926322844,
"learning_rate": 8.838265850388698e-05,
"loss": 0.0057,
"step": 4540
},
{
"epoch": 0.9001879513304976,
"grad_norm": 0.050098760517275694,
"learning_rate": 8.830879741519832e-05,
"loss": 0.0056,
"step": 4550
},
{
"epoch": 0.9021663863883668,
"grad_norm": 0.05053393187481689,
"learning_rate": 8.823473334333298e-05,
"loss": 0.0056,
"step": 4560
},
{
"epoch": 0.904144821446236,
"grad_norm": 0.049606634644607885,
"learning_rate": 8.816046668072726e-05,
"loss": 0.0055,
"step": 4570
},
{
"epoch": 0.9061232565041053,
"grad_norm": 0.04940433857433825,
"learning_rate": 8.808599782089081e-05,
"loss": 0.0056,
"step": 4580
},
{
"epoch": 0.9081016915619745,
"grad_norm": 0.05210440765536462,
"learning_rate": 8.80113271584047e-05,
"loss": 0.0056,
"step": 4590
},
{
"epoch": 0.9100801266198437,
"grad_norm": 0.048872359152922505,
"learning_rate": 8.793645508891926e-05,
"loss": 0.0056,
"step": 4600
},
{
"epoch": 0.9120585616777129,
"grad_norm": 0.0474524651210216,
"learning_rate": 8.786138200915199e-05,
"loss": 0.0055,
"step": 4610
},
{
"epoch": 0.9140369967355821,
"grad_norm": 0.05611126919252264,
"learning_rate": 8.778610831688548e-05,
"loss": 0.0055,
"step": 4620
},
{
"epoch": 0.9160154317934514,
"grad_norm": 0.053673933071949814,
"learning_rate": 8.771063441096526e-05,
"loss": 0.0055,
"step": 4630
},
{
"epoch": 0.9179938668513206,
"grad_norm": 0.049145678875860736,
"learning_rate": 8.763496069129773e-05,
"loss": 0.0055,
"step": 4640
},
{
"epoch": 0.9199723019091899,
"grad_norm": 0.046677926343471225,
"learning_rate": 8.755908755884804e-05,
"loss": 0.0055,
"step": 4650
},
{
"epoch": 0.9219507369670591,
"grad_norm": 0.05231084152679913,
"learning_rate": 8.74830154156379e-05,
"loss": 0.0056,
"step": 4660
},
{
"epoch": 0.9239291720249283,
"grad_norm": 0.051854254850395756,
"learning_rate": 8.740674466474356e-05,
"loss": 0.0056,
"step": 4670
},
{
"epoch": 0.9259076070827975,
"grad_norm": 0.05727564581566202,
"learning_rate": 8.733027571029358e-05,
"loss": 0.0057,
"step": 4680
},
{
"epoch": 0.9278860421406667,
"grad_norm": 0.04350759098972187,
"learning_rate": 8.725360895746671e-05,
"loss": 0.0056,
"step": 4690
},
{
"epoch": 0.9298644771985359,
"grad_norm": 0.05429845095061911,
"learning_rate": 8.717674481248978e-05,
"loss": 0.0056,
"step": 4700
},
{
"epoch": 0.9318429122564051,
"grad_norm": 0.04977968896245699,
"learning_rate": 8.709968368263553e-05,
"loss": 0.0055,
"step": 4710
},
{
"epoch": 0.9338213473142745,
"grad_norm": 0.03990415499375446,
"learning_rate": 8.70224259762204e-05,
"loss": 0.0055,
"step": 4720
},
{
"epoch": 0.9357997823721437,
"grad_norm": 0.04900443011417874,
"learning_rate": 8.694497210260246e-05,
"loss": 0.0056,
"step": 4730
},
{
"epoch": 0.9377782174300129,
"grad_norm": 0.04941519243560236,
"learning_rate": 8.686732247217917e-05,
"loss": 0.0056,
"step": 4740
},
{
"epoch": 0.9397566524878821,
"grad_norm": 0.045555626835663394,
"learning_rate": 8.678947749638526e-05,
"loss": 0.0056,
"step": 4750
},
{
"epoch": 0.9417350875457513,
"grad_norm": 0.05047988632250821,
"learning_rate": 8.671143758769044e-05,
"loss": 0.0055,
"step": 4760
},
{
"epoch": 0.9437135226036205,
"grad_norm": 0.04171320978771468,
"learning_rate": 8.663320315959736e-05,
"loss": 0.0055,
"step": 4770
},
{
"epoch": 0.9456919576614897,
"grad_norm": 0.05320952004003585,
"learning_rate": 8.655477462663937e-05,
"loss": 0.0055,
"step": 4780
},
{
"epoch": 0.9476703927193589,
"grad_norm": 0.0568546245001232,
"learning_rate": 8.647615240437821e-05,
"loss": 0.0059,
"step": 4790
},
{
"epoch": 0.9496488277772283,
"grad_norm": 0.055441453083329505,
"learning_rate": 8.6397336909402e-05,
"loss": 0.0055,
"step": 4800
},
{
"epoch": 0.9516272628350975,
"grad_norm": 0.038512301024341014,
"learning_rate": 8.631832855932289e-05,
"loss": 0.0055,
"step": 4810
},
{
"epoch": 0.9536056978929667,
"grad_norm": 0.0424237121563684,
"learning_rate": 8.62391277727749e-05,
"loss": 0.0055,
"step": 4820
},
{
"epoch": 0.9555841329508359,
"grad_norm": 0.09686022503746353,
"learning_rate": 8.615973496941169e-05,
"loss": 0.0056,
"step": 4830
},
{
"epoch": 0.9575625680087051,
"grad_norm": 0.04613490071274156,
"learning_rate": 8.608015056990437e-05,
"loss": 0.0056,
"step": 4840
},
{
"epoch": 0.9595410030665743,
"grad_norm": 0.050863793466427845,
"learning_rate": 8.600037499593918e-05,
"loss": 0.0055,
"step": 4850
},
{
"epoch": 0.9615194381244435,
"grad_norm": 0.059377991249387384,
"learning_rate": 8.592040867021538e-05,
"loss": 0.0056,
"step": 4860
},
{
"epoch": 0.9634978731823128,
"grad_norm": 0.046539793320152095,
"learning_rate": 8.584025201644292e-05,
"loss": 0.0055,
"step": 4870
},
{
"epoch": 0.965476308240182,
"grad_norm": 0.05141020110495577,
"learning_rate": 8.575990545934023e-05,
"loss": 0.0055,
"step": 4880
},
{
"epoch": 0.9674547432980513,
"grad_norm": 0.04361738539207239,
"learning_rate": 8.567936942463195e-05,
"loss": 0.0054,
"step": 4890
},
{
"epoch": 0.9694331783559205,
"grad_norm": 0.033705374589094644,
"learning_rate": 8.559864433904671e-05,
"loss": 0.0054,
"step": 4900
},
{
"epoch": 0.9714116134137897,
"grad_norm": 0.05423815204708426,
"learning_rate": 8.551773063031484e-05,
"loss": 0.0055,
"step": 4910
},
{
"epoch": 0.9733900484716589,
"grad_norm": 0.04106940909075412,
"learning_rate": 8.543662872716607e-05,
"loss": 0.0055,
"step": 4920
},
{
"epoch": 0.9753684835295281,
"grad_norm": 0.05696790752316582,
"learning_rate": 8.535533905932738e-05,
"loss": 0.0055,
"step": 4930
},
{
"epoch": 0.9773469185873974,
"grad_norm": 0.04726106326585974,
"learning_rate": 8.527386205752056e-05,
"loss": 0.0055,
"step": 4940
},
{
"epoch": 0.9793253536452666,
"grad_norm": 0.03894347337686866,
"learning_rate": 8.519219815346004e-05,
"loss": 0.0054,
"step": 4950
},
{
"epoch": 0.9813037887031358,
"grad_norm": 0.03699857632746051,
"learning_rate": 8.511034777985058e-05,
"loss": 0.0054,
"step": 4960
},
{
"epoch": 0.983282223761005,
"grad_norm": 0.04905840661301598,
"learning_rate": 8.502831137038496e-05,
"loss": 0.0055,
"step": 4970
},
{
"epoch": 0.9852606588188743,
"grad_norm": 0.05148826569417279,
"learning_rate": 8.494608935974168e-05,
"loss": 0.0054,
"step": 4980
},
{
"epoch": 0.9872390938767435,
"grad_norm": 0.047222232898975446,
"learning_rate": 8.486368218358268e-05,
"loss": 0.0054,
"step": 4990
},
{
"epoch": 0.9892175289346127,
"grad_norm": 0.05142550662959016,
"learning_rate": 8.478109027855103e-05,
"loss": 0.0055,
"step": 5000
},
{
"epoch": 0.991195963992482,
"grad_norm": 0.043002068756616836,
"learning_rate": 8.469831408226858e-05,
"loss": 0.0056,
"step": 5010
},
{
"epoch": 0.9931743990503512,
"grad_norm": 0.040134524434816944,
"learning_rate": 8.46153540333337e-05,
"loss": 0.0055,
"step": 5020
},
{
"epoch": 0.9951528341082204,
"grad_norm": 0.048260363711960075,
"learning_rate": 8.453221057131886e-05,
"loss": 0.0054,
"step": 5030
},
{
"epoch": 0.9971312691660896,
"grad_norm": 0.04421006993683667,
"learning_rate": 8.444888413676844e-05,
"loss": 0.0055,
"step": 5040
},
{
"epoch": 0.9991097042239588,
"grad_norm": 0.04369413439191883,
"learning_rate": 8.436537517119624e-05,
"loss": 0.0055,
"step": 5050
},
{
"epoch": 1.0009892175289345,
"grad_norm": 0.0719856709001207,
"learning_rate": 8.428168411708332e-05,
"loss": 0.0055,
"step": 5060
},
{
"epoch": 1.0029676525868039,
"grad_norm": 0.047900843298213036,
"learning_rate": 8.419781141787548e-05,
"loss": 0.0054,
"step": 5070
},
{
"epoch": 1.004946087644673,
"grad_norm": 0.05550072548945524,
"learning_rate": 8.411375751798099e-05,
"loss": 0.0054,
"step": 5080
},
{
"epoch": 1.0069245227025423,
"grad_norm": 0.0585174912559527,
"learning_rate": 8.402952286276824e-05,
"loss": 0.0053,
"step": 5090
},
{
"epoch": 1.0089029577604116,
"grad_norm": 0.05977794537241512,
"learning_rate": 8.394510789856339e-05,
"loss": 0.0054,
"step": 5100
},
{
"epoch": 1.0108813928182807,
"grad_norm": 0.05483601101494429,
"learning_rate": 8.386051307264797e-05,
"loss": 0.0054,
"step": 5110
},
{
"epoch": 1.01285982787615,
"grad_norm": 0.044091733123439364,
"learning_rate": 8.377573883325653e-05,
"loss": 0.0054,
"step": 5120
},
{
"epoch": 1.0148382629340191,
"grad_norm": 0.04918536166175687,
"learning_rate": 8.369078562957425e-05,
"loss": 0.0055,
"step": 5130
},
{
"epoch": 1.0168166979918885,
"grad_norm": 0.05652860370839811,
"learning_rate": 8.360565391173459e-05,
"loss": 0.0055,
"step": 5140
},
{
"epoch": 1.0187951330497576,
"grad_norm": 0.04760498157974804,
"learning_rate": 8.352034413081686e-05,
"loss": 0.0054,
"step": 5150
},
{
"epoch": 1.0207735681076269,
"grad_norm": 0.04385626763045335,
"learning_rate": 8.343485673884388e-05,
"loss": 0.0053,
"step": 5160
},
{
"epoch": 1.0227520031654962,
"grad_norm": 0.04244162914461471,
"learning_rate": 8.334919218877957e-05,
"loss": 0.0053,
"step": 5170
},
{
"epoch": 1.0247304382233653,
"grad_norm": 0.03999930730392116,
"learning_rate": 8.326335093452651e-05,
"loss": 0.0052,
"step": 5180
},
{
"epoch": 1.0267088732812346,
"grad_norm": 0.052515463653386975,
"learning_rate": 8.317733343092357e-05,
"loss": 0.0053,
"step": 5190
},
{
"epoch": 1.0286873083391037,
"grad_norm": 0.04577942213576051,
"learning_rate": 8.309114013374353e-05,
"loss": 0.0052,
"step": 5200
},
{
"epoch": 1.030665743396973,
"grad_norm": 0.04993962938218267,
"learning_rate": 8.300477149969057e-05,
"loss": 0.0054,
"step": 5210
},
{
"epoch": 1.0326441784548421,
"grad_norm": 0.045792798191854917,
"learning_rate": 8.291822798639797e-05,
"loss": 0.0054,
"step": 5220
},
{
"epoch": 1.0346226135127115,
"grad_norm": 0.04030041547592181,
"learning_rate": 8.28315100524256e-05,
"loss": 0.0053,
"step": 5230
},
{
"epoch": 1.0366010485705808,
"grad_norm": 0.044053287325595,
"learning_rate": 8.274461815725751e-05,
"loss": 0.0053,
"step": 5240
},
{
"epoch": 1.0385794836284499,
"grad_norm": 0.03583641958219779,
"learning_rate": 8.265755276129953e-05,
"loss": 0.0053,
"step": 5250
},
{
"epoch": 1.0405579186863192,
"grad_norm": 0.03943441378422594,
"learning_rate": 8.257031432587676e-05,
"loss": 0.0053,
"step": 5260
},
{
"epoch": 1.0425363537441883,
"grad_norm": 0.046702443146808284,
"learning_rate": 8.24829033132312e-05,
"loss": 0.0053,
"step": 5270
},
{
"epoch": 1.0445147888020576,
"grad_norm": 0.04002421382617951,
"learning_rate": 8.239532018651924e-05,
"loss": 0.0054,
"step": 5280
},
{
"epoch": 1.0464932238599267,
"grad_norm": 0.049898395275073566,
"learning_rate": 8.230756540980929e-05,
"loss": 0.0053,
"step": 5290
},
{
"epoch": 1.048471658917796,
"grad_norm": 0.044099315155096776,
"learning_rate": 8.22196394480792e-05,
"loss": 0.0053,
"step": 5300
},
{
"epoch": 1.0504500939756654,
"grad_norm": 0.038843947411654166,
"learning_rate": 8.213154276721388e-05,
"loss": 0.0052,
"step": 5310
},
{
"epoch": 1.0524285290335345,
"grad_norm": 0.03811263250797667,
"learning_rate": 8.204327583400287e-05,
"loss": 0.0053,
"step": 5320
},
{
"epoch": 1.0544069640914038,
"grad_norm": 0.0467458274561098,
"learning_rate": 8.195483911613772e-05,
"loss": 0.0053,
"step": 5330
},
{
"epoch": 1.0563853991492729,
"grad_norm": 0.0373503533099714,
"learning_rate": 8.186623308220967e-05,
"loss": 0.0053,
"step": 5340
},
{
"epoch": 1.0583638342071422,
"grad_norm": 0.04039130744819984,
"learning_rate": 8.177745820170705e-05,
"loss": 0.0052,
"step": 5350
},
{
"epoch": 1.0603422692650113,
"grad_norm": 0.03964891194283436,
"learning_rate": 8.168851494501287e-05,
"loss": 0.0053,
"step": 5360
},
{
"epoch": 1.0623207043228806,
"grad_norm": 0.04084344156391512,
"learning_rate": 8.159940378340232e-05,
"loss": 0.0054,
"step": 5370
},
{
"epoch": 1.0642991393807497,
"grad_norm": 0.043231825218949894,
"learning_rate": 8.15101251890402e-05,
"loss": 0.0054,
"step": 5380
},
{
"epoch": 1.066277574438619,
"grad_norm": 0.051759476483110474,
"learning_rate": 8.142067963497851e-05,
"loss": 0.0055,
"step": 5390
},
{
"epoch": 1.0682560094964884,
"grad_norm": 0.04396815532879462,
"learning_rate": 8.133106759515388e-05,
"loss": 0.0053,
"step": 5400
},
{
"epoch": 1.0702344445543575,
"grad_norm": 0.044125384198911356,
"learning_rate": 8.124128954438508e-05,
"loss": 0.0054,
"step": 5410
},
{
"epoch": 1.0722128796122268,
"grad_norm": 0.05351443836665567,
"learning_rate": 8.115134595837055e-05,
"loss": 0.0053,
"step": 5420
},
{
"epoch": 1.0741913146700959,
"grad_norm": 0.04323835485788573,
"learning_rate": 8.106123731368578e-05,
"loss": 0.0053,
"step": 5430
},
{
"epoch": 1.0761697497279652,
"grad_norm": 0.048674364979576765,
"learning_rate": 8.097096408778086e-05,
"loss": 0.0053,
"step": 5440
},
{
"epoch": 1.0781481847858343,
"grad_norm": 0.03619513568830504,
"learning_rate": 8.088052675897795e-05,
"loss": 0.0053,
"step": 5450
},
{
"epoch": 1.0801266198437036,
"grad_norm": 0.04103062352002762,
"learning_rate": 8.078992580646871e-05,
"loss": 0.0052,
"step": 5460
},
{
"epoch": 1.082105054901573,
"grad_norm": 0.031033634520249744,
"learning_rate": 8.06991617103118e-05,
"loss": 0.0052,
"step": 5470
},
{
"epoch": 1.084083489959442,
"grad_norm": 0.030030582487575556,
"learning_rate": 8.060823495143024e-05,
"loss": 0.0052,
"step": 5480
},
{
"epoch": 1.0860619250173114,
"grad_norm": 0.03582363331333947,
"learning_rate": 8.051714601160903e-05,
"loss": 0.0051,
"step": 5490
},
{
"epoch": 1.0880403600751805,
"grad_norm": 0.027794893239300178,
"learning_rate": 8.042589537349243e-05,
"loss": 0.0053,
"step": 5500
},
{
"epoch": 1.0900187951330498,
"grad_norm": 0.03990402884030837,
"learning_rate": 8.033448352058154e-05,
"loss": 0.0053,
"step": 5510
},
{
"epoch": 1.0919972301909189,
"grad_norm": 0.06247549764208659,
"learning_rate": 8.024291093723162e-05,
"loss": 0.0053,
"step": 5520
},
{
"epoch": 1.0939756652487882,
"grad_norm": 0.04277503107762733,
"learning_rate": 8.01511781086496e-05,
"loss": 0.0053,
"step": 5530
},
{
"epoch": 1.0959541003066575,
"grad_norm": 0.0403199829465898,
"learning_rate": 8.005928552089148e-05,
"loss": 0.0052,
"step": 5540
},
{
"epoch": 1.0979325353645266,
"grad_norm": 0.03888083981959423,
"learning_rate": 7.996723366085977e-05,
"loss": 0.0053,
"step": 5550
},
{
"epoch": 1.099910970422396,
"grad_norm": 0.043107681407633894,
"learning_rate": 7.98750230163009e-05,
"loss": 0.0053,
"step": 5560
},
{
"epoch": 1.101889405480265,
"grad_norm": 0.03641841144993682,
"learning_rate": 7.978265407580261e-05,
"loss": 0.0052,
"step": 5570
},
{
"epoch": 1.1038678405381344,
"grad_norm": 0.041363570145620976,
"learning_rate": 7.969012732879144e-05,
"loss": 0.0052,
"step": 5580
},
{
"epoch": 1.1058462755960035,
"grad_norm": 0.041161510981752886,
"learning_rate": 7.959744326553001e-05,
"loss": 0.0053,
"step": 5590
},
{
"epoch": 1.1078247106538728,
"grad_norm": 0.04133755286702683,
"learning_rate": 7.950460237711458e-05,
"loss": 0.0052,
"step": 5600
},
{
"epoch": 1.109803145711742,
"grad_norm": 0.056566462073514165,
"learning_rate": 7.94116051554723e-05,
"loss": 0.0056,
"step": 5610
},
{
"epoch": 1.1117815807696112,
"grad_norm": 0.043585300028570556,
"learning_rate": 7.931845209335871e-05,
"loss": 0.0055,
"step": 5620
},
{
"epoch": 1.1137600158274805,
"grad_norm": 0.03449315055234552,
"learning_rate": 7.922514368435505e-05,
"loss": 0.0054,
"step": 5630
},
{
"epoch": 1.1157384508853496,
"grad_norm": 0.0402489204427218,
"learning_rate": 7.91316804228657e-05,
"loss": 0.0053,
"step": 5640
},
{
"epoch": 1.117716885943219,
"grad_norm": 0.04255509433498578,
"learning_rate": 7.903806280411557e-05,
"loss": 0.0053,
"step": 5650
},
{
"epoch": 1.119695321001088,
"grad_norm": 0.037709339966754185,
"learning_rate": 7.89442913241474e-05,
"loss": 0.0053,
"step": 5660
},
{
"epoch": 1.1216737560589574,
"grad_norm": 0.04756551929330449,
"learning_rate": 7.885036647981921e-05,
"loss": 0.0054,
"step": 5670
},
{
"epoch": 1.1236521911168267,
"grad_norm": 0.04710819234980247,
"learning_rate": 7.875628876880161e-05,
"loss": 0.0053,
"step": 5680
},
{
"epoch": 1.1256306261746958,
"grad_norm": 0.11839642722567298,
"learning_rate": 7.866205868957524e-05,
"loss": 0.0054,
"step": 5690
},
{
"epoch": 1.127609061232565,
"grad_norm": 0.046477740341715125,
"learning_rate": 7.8567676741428e-05,
"loss": 0.0054,
"step": 5700
},
{
"epoch": 1.1295874962904342,
"grad_norm": 0.0383229716806028,
"learning_rate": 7.847314342445258e-05,
"loss": 0.0053,
"step": 5710
},
{
"epoch": 1.1315659313483035,
"grad_norm": 0.03645158350961823,
"learning_rate": 7.837845923954364e-05,
"loss": 0.0052,
"step": 5720
},
{
"epoch": 1.1335443664061726,
"grad_norm": 0.0450113726262194,
"learning_rate": 7.828362468839525e-05,
"loss": 0.0052,
"step": 5730
},
{
"epoch": 1.135522801464042,
"grad_norm": 0.03550411578940989,
"learning_rate": 7.818864027349822e-05,
"loss": 0.0052,
"step": 5740
},
{
"epoch": 1.1375012365219113,
"grad_norm": 0.027251965595028085,
"learning_rate": 7.809350649813743e-05,
"loss": 0.0051,
"step": 5750
},
{
"epoch": 1.1394796715797804,
"grad_norm": 0.03621476292087555,
"learning_rate": 7.799822386638915e-05,
"loss": 0.0052,
"step": 5760
},
{
"epoch": 1.1414581066376497,
"grad_norm": 0.03272113423901916,
"learning_rate": 7.790279288311837e-05,
"loss": 0.0053,
"step": 5770
},
{
"epoch": 1.1434365416955188,
"grad_norm": 0.040204161620133606,
"learning_rate": 7.78072140539762e-05,
"loss": 0.0052,
"step": 5780
},
{
"epoch": 1.145414976753388,
"grad_norm": 0.040160939364404485,
"learning_rate": 7.771148788539704e-05,
"loss": 0.0051,
"step": 5790
},
{
"epoch": 1.1473934118112572,
"grad_norm": 0.03518171511287162,
"learning_rate": 7.761561488459603e-05,
"loss": 0.0052,
"step": 5800
},
{
"epoch": 1.1493718468691265,
"grad_norm": 0.04520262485290099,
"learning_rate": 7.751959555956632e-05,
"loss": 0.0053,
"step": 5810
},
{
"epoch": 1.1513502819269958,
"grad_norm": 0.042900190491880326,
"learning_rate": 7.742343041907634e-05,
"loss": 0.0053,
"step": 5820
},
{
"epoch": 1.153328716984865,
"grad_norm": 0.04683151967273415,
"learning_rate": 7.732711997266719e-05,
"loss": 0.0052,
"step": 5830
},
{
"epoch": 1.1553071520427343,
"grad_norm": 0.03505999101080227,
"learning_rate": 7.723066473064986e-05,
"loss": 0.0053,
"step": 5840
},
{
"epoch": 1.1572855871006034,
"grad_norm": 0.04656635214059782,
"learning_rate": 7.713406520410252e-05,
"loss": 0.0053,
"step": 5850
},
{
"epoch": 1.1592640221584727,
"grad_norm": 0.04179336547034277,
"learning_rate": 7.703732190486792e-05,
"loss": 0.0052,
"step": 5860
},
{
"epoch": 1.1612424572163418,
"grad_norm": 0.03116060651278895,
"learning_rate": 7.694043534555055e-05,
"loss": 0.0052,
"step": 5870
},
{
"epoch": 1.163220892274211,
"grad_norm": 0.04690720920569915,
"learning_rate": 7.684340603951399e-05,
"loss": 0.0052,
"step": 5880
},
{
"epoch": 1.1651993273320804,
"grad_norm": 0.04098129335597052,
"learning_rate": 7.674623450087822e-05,
"loss": 0.0053,
"step": 5890
},
{
"epoch": 1.1671777623899495,
"grad_norm": 0.03646895298311979,
"learning_rate": 7.664892124451676e-05,
"loss": 0.0052,
"step": 5900
},
{
"epoch": 1.1691561974478188,
"grad_norm": 0.04723789514096997,
"learning_rate": 7.655146678605414e-05,
"loss": 0.0053,
"step": 5910
},
{
"epoch": 1.171134632505688,
"grad_norm": 5.375195684350047,
"learning_rate": 7.645387164186298e-05,
"loss": 0.0204,
"step": 5920
},
{
"epoch": 1.1731130675635573,
"grad_norm": 2.634267593555541,
"learning_rate": 7.635613632906141e-05,
"loss": 0.0855,
"step": 5930
},
{
"epoch": 1.1750915026214264,
"grad_norm": 0.7590969824054792,
"learning_rate": 7.625826136551022e-05,
"loss": 0.0589,
"step": 5940
},
{
"epoch": 1.1770699376792957,
"grad_norm": 0.25748384500101734,
"learning_rate": 7.616024726981015e-05,
"loss": 0.0222,
"step": 5950
},
{
"epoch": 1.179048372737165,
"grad_norm": 0.17703171961522354,
"learning_rate": 7.606209456129914e-05,
"loss": 0.0159,
"step": 5960
},
{
"epoch": 1.181026807795034,
"grad_norm": 0.08672887220585662,
"learning_rate": 7.596380376004965e-05,
"loss": 0.0095,
"step": 5970
},
{
"epoch": 1.1830052428529034,
"grad_norm": 0.05402458437791095,
"learning_rate": 7.586537538686574e-05,
"loss": 0.0072,
"step": 5980
},
{
"epoch": 1.1849836779107725,
"grad_norm": 0.04452390263920157,
"learning_rate": 7.576680996328045e-05,
"loss": 0.0063,
"step": 5990
},
{
"epoch": 1.1869621129686418,
"grad_norm": 0.038050350345016797,
"learning_rate": 7.566810801155304e-05,
"loss": 0.0058,
"step": 6000
},
{
"epoch": 1.188940548026511,
"grad_norm": 0.03852856640194162,
"learning_rate": 7.556927005466612e-05,
"loss": 0.0056,
"step": 6010
},
{
"epoch": 1.1909189830843803,
"grad_norm": 0.041189680009927,
"learning_rate": 7.547029661632298e-05,
"loss": 0.0056,
"step": 6020
},
{
"epoch": 1.1928974181422496,
"grad_norm": 0.03825003704039207,
"learning_rate": 7.537118822094474e-05,
"loss": 0.0054,
"step": 6030
},
{
"epoch": 1.1948758532001187,
"grad_norm": 0.0322276898731417,
"learning_rate": 7.52719453936676e-05,
"loss": 0.0054,
"step": 6040
},
{
"epoch": 1.196854288257988,
"grad_norm": 0.029959977277491553,
"learning_rate": 7.517256866034011e-05,
"loss": 0.0054,
"step": 6050
},
{
"epoch": 1.198832723315857,
"grad_norm": 0.03582231070302725,
"learning_rate": 7.507305854752027e-05,
"loss": 0.0053,
"step": 6060
},
{
"epoch": 1.2008111583737264,
"grad_norm": 0.032178891262669276,
"learning_rate": 7.497341558247286e-05,
"loss": 0.0052,
"step": 6070
},
{
"epoch": 1.2027895934315955,
"grad_norm": 0.027720539585402305,
"learning_rate": 7.487364029316655e-05,
"loss": 0.0052,
"step": 6080
},
{
"epoch": 1.2047680284894648,
"grad_norm": 0.02224977013330438,
"learning_rate": 7.47737332082712e-05,
"loss": 0.0052,
"step": 6090
},
{
"epoch": 1.2067464635473342,
"grad_norm": 0.033059221452976024,
"learning_rate": 7.467369485715492e-05,
"loss": 0.0052,
"step": 6100
},
{
"epoch": 1.2087248986052033,
"grad_norm": 0.028734354844642866,
"learning_rate": 7.457352576988144e-05,
"loss": 0.0051,
"step": 6110
},
{
"epoch": 1.2107033336630726,
"grad_norm": 0.036005894532009496,
"learning_rate": 7.447322647720714e-05,
"loss": 0.0051,
"step": 6120
},
{
"epoch": 1.2126817687209417,
"grad_norm": 0.029898724956941146,
"learning_rate": 7.437279751057833e-05,
"loss": 0.0051,
"step": 6130
},
{
"epoch": 1.214660203778811,
"grad_norm": 0.03332085027695129,
"learning_rate": 7.42722394021284e-05,
"loss": 0.0052,
"step": 6140
},
{
"epoch": 1.21663863883668,
"grad_norm": 0.03513978106939451,
"learning_rate": 7.417155268467505e-05,
"loss": 0.0051,
"step": 6150
},
{
"epoch": 1.2186170738945494,
"grad_norm": 0.026482979419996774,
"learning_rate": 7.407073789171737e-05,
"loss": 0.0051,
"step": 6160
},
{
"epoch": 1.2205955089524188,
"grad_norm": 0.032409915864019645,
"learning_rate": 7.396979555743315e-05,
"loss": 0.0052,
"step": 6170
},
{
"epoch": 1.2225739440102878,
"grad_norm": 0.032793057837405055,
"learning_rate": 7.386872621667588e-05,
"loss": 0.0051,
"step": 6180
},
{
"epoch": 1.2245523790681572,
"grad_norm": 0.0245869749900172,
"learning_rate": 7.376753040497206e-05,
"loss": 0.005,
"step": 6190
},
{
"epoch": 1.2265308141260263,
"grad_norm": 0.023890421006088333,
"learning_rate": 7.366620865851832e-05,
"loss": 0.005,
"step": 6200
},
{
"epoch": 1.2285092491838956,
"grad_norm": 0.026261323252168123,
"learning_rate": 7.356476151417851e-05,
"loss": 0.0051,
"step": 6210
},
{
"epoch": 1.2304876842417647,
"grad_norm": 0.03538561422814988,
"learning_rate": 7.3463189509481e-05,
"loss": 0.0051,
"step": 6220
},
{
"epoch": 1.232466119299634,
"grad_norm": 0.027412102190070085,
"learning_rate": 7.336149318261565e-05,
"loss": 0.0051,
"step": 6230
},
{
"epoch": 1.2344445543575033,
"grad_norm": 0.030928126594571684,
"learning_rate": 7.325967307243116e-05,
"loss": 0.005,
"step": 6240
},
{
"epoch": 1.2364229894153724,
"grad_norm": 0.027918922259256834,
"learning_rate": 7.3157729718432e-05,
"loss": 0.0051,
"step": 6250
},
{
"epoch": 1.2384014244732418,
"grad_norm": 0.03164611925092058,
"learning_rate": 7.305566366077573e-05,
"loss": 0.0051,
"step": 6260
},
{
"epoch": 1.2403798595311109,
"grad_norm": 0.02929783367850166,
"learning_rate": 7.295347544027006e-05,
"loss": 0.005,
"step": 6270
},
{
"epoch": 1.2423582945889802,
"grad_norm": 0.02681089065088164,
"learning_rate": 7.285116559836996e-05,
"loss": 0.005,
"step": 6280
},
{
"epoch": 1.2443367296468493,
"grad_norm": 0.028841188530230485,
"learning_rate": 7.274873467717487e-05,
"loss": 0.005,
"step": 6290
},
{
"epoch": 1.2463151647047186,
"grad_norm": 0.026804379927279805,
"learning_rate": 7.264618321942575e-05,
"loss": 0.005,
"step": 6300
},
{
"epoch": 1.248293599762588,
"grad_norm": 0.02265970969382068,
"learning_rate": 7.254351176850222e-05,
"loss": 0.0049,
"step": 6310
},
{
"epoch": 1.250272034820457,
"grad_norm": 0.03051592326677612,
"learning_rate": 7.244072086841977e-05,
"loss": 0.005,
"step": 6320
},
{
"epoch": 1.2522504698783261,
"grad_norm": 0.020292357045255553,
"learning_rate": 7.23378110638267e-05,
"loss": 0.005,
"step": 6330
},
{
"epoch": 1.2542289049361954,
"grad_norm": 0.027122627547033654,
"learning_rate": 7.223478290000142e-05,
"loss": 0.005,
"step": 6340
},
{
"epoch": 1.2562073399940648,
"grad_norm": 0.02896953450896817,
"learning_rate": 7.213163692284943e-05,
"loss": 0.0049,
"step": 6350
},
{
"epoch": 1.2581857750519339,
"grad_norm": 0.02642217638220959,
"learning_rate": 7.202837367890054e-05,
"loss": 0.005,
"step": 6360
},
{
"epoch": 1.2601642101098032,
"grad_norm": 0.022906069808378036,
"learning_rate": 7.192499371530581e-05,
"loss": 0.005,
"step": 6370
},
{
"epoch": 1.2621426451676725,
"grad_norm": 0.03247356086204418,
"learning_rate": 7.182149757983484e-05,
"loss": 0.005,
"step": 6380
},
{
"epoch": 1.2641210802255416,
"grad_norm": 0.026005157648908206,
"learning_rate": 7.171788582087277e-05,
"loss": 0.005,
"step": 6390
},
{
"epoch": 1.2660995152834107,
"grad_norm": 0.026694932641269483,
"learning_rate": 7.16141589874173e-05,
"loss": 0.005,
"step": 6400
},
{
"epoch": 1.26807795034128,
"grad_norm": 0.026616390376041472,
"learning_rate": 7.151031762907597e-05,
"loss": 0.0049,
"step": 6410
},
{
"epoch": 1.2700563853991493,
"grad_norm": 0.02420463518625067,
"learning_rate": 7.140636229606308e-05,
"loss": 0.005,
"step": 6420
},
{
"epoch": 1.2720348204570184,
"grad_norm": 0.029239487195380758,
"learning_rate": 7.130229353919684e-05,
"loss": 0.0049,
"step": 6430
},
{
"epoch": 1.2740132555148878,
"grad_norm": 0.029675011691943694,
"learning_rate": 7.119811190989649e-05,
"loss": 0.0049,
"step": 6440
},
{
"epoch": 1.275991690572757,
"grad_norm": 0.02775416898861546,
"learning_rate": 7.10938179601793e-05,
"loss": 0.0049,
"step": 6450
},
{
"epoch": 1.2779701256306262,
"grad_norm": 0.02011469561071511,
"learning_rate": 7.098941224265769e-05,
"loss": 0.0049,
"step": 6460
},
{
"epoch": 1.2799485606884953,
"grad_norm": 0.01811387979720197,
"learning_rate": 7.088489531053627e-05,
"loss": 0.0049,
"step": 6470
},
{
"epoch": 1.2819269957463646,
"grad_norm": 0.028062698888850948,
"learning_rate": 7.0780267717609e-05,
"loss": 0.005,
"step": 6480
},
{
"epoch": 1.283905430804234,
"grad_norm": 0.026082725678708722,
"learning_rate": 7.067553001825609e-05,
"loss": 0.005,
"step": 6490
},
{
"epoch": 1.285883865862103,
"grad_norm": 0.033257603107593174,
"learning_rate": 7.057068276744123e-05,
"loss": 0.005,
"step": 6500
},
{
"epoch": 1.2878623009199723,
"grad_norm": 0.020643115120901275,
"learning_rate": 7.04657265207086e-05,
"loss": 0.0049,
"step": 6510
},
{
"epoch": 1.2898407359778417,
"grad_norm": 0.020926731286904143,
"learning_rate": 7.036066183417981e-05,
"loss": 0.0049,
"step": 6520
},
{
"epoch": 1.2918191710357108,
"grad_norm": 0.02209364178561641,
"learning_rate": 7.025548926455117e-05,
"loss": 0.0049,
"step": 6530
},
{
"epoch": 1.2937976060935799,
"grad_norm": 0.020661649256744814,
"learning_rate": 7.015020936909052e-05,
"loss": 0.0049,
"step": 6540
},
{
"epoch": 1.2957760411514492,
"grad_norm": 0.036527743248369486,
"learning_rate": 7.004482270563441e-05,
"loss": 0.0049,
"step": 6550
},
{
"epoch": 1.2977544762093185,
"grad_norm": 0.03451686256423913,
"learning_rate": 6.993932983258514e-05,
"loss": 0.0049,
"step": 6560
},
{
"epoch": 1.2997329112671876,
"grad_norm": 0.02602752753469148,
"learning_rate": 6.983373130890774e-05,
"loss": 0.0049,
"step": 6570
},
{
"epoch": 1.301711346325057,
"grad_norm": 0.022789421472611756,
"learning_rate": 6.972802769412706e-05,
"loss": 0.0049,
"step": 6580
},
{
"epoch": 1.3036897813829262,
"grad_norm": 0.023072822386535248,
"learning_rate": 6.962221954832476e-05,
"loss": 0.0049,
"step": 6590
},
{
"epoch": 1.3056682164407953,
"grad_norm": 0.0250162002156148,
"learning_rate": 6.95163074321364e-05,
"loss": 0.0049,
"step": 6600
},
{
"epoch": 1.3076466514986644,
"grad_norm": 0.018797653270342333,
"learning_rate": 6.941029190674844e-05,
"loss": 0.0048,
"step": 6610
},
{
"epoch": 1.3096250865565338,
"grad_norm": 0.017446599276522104,
"learning_rate": 6.93041735338952e-05,
"loss": 0.0049,
"step": 6620
},
{
"epoch": 1.311603521614403,
"grad_norm": 0.016439118459896422,
"learning_rate": 6.919795287585604e-05,
"loss": 0.0049,
"step": 6630
},
{
"epoch": 1.3135819566722722,
"grad_norm": 0.021968697449193792,
"learning_rate": 6.909163049545223e-05,
"loss": 0.0049,
"step": 6640
},
{
"epoch": 1.3155603917301415,
"grad_norm": 0.02610794667572913,
"learning_rate": 6.898520695604404e-05,
"loss": 0.0049,
"step": 6650
},
{
"epoch": 1.3175388267880108,
"grad_norm": 0.03022332265027175,
"learning_rate": 6.887868282152775e-05,
"loss": 0.005,
"step": 6660
},
{
"epoch": 1.31951726184588,
"grad_norm": 0.02348724398891657,
"learning_rate": 6.877205865633262e-05,
"loss": 0.005,
"step": 6670
},
{
"epoch": 1.321495696903749,
"grad_norm": 0.023404871124257823,
"learning_rate": 6.866533502541802e-05,
"loss": 0.005,
"step": 6680
},
{
"epoch": 1.3234741319616183,
"grad_norm": 0.021073399068651837,
"learning_rate": 6.855851249427024e-05,
"loss": 0.0049,
"step": 6690
},
{
"epoch": 1.3254525670194877,
"grad_norm": 0.0290891333074333,
"learning_rate": 6.845159162889965e-05,
"loss": 0.005,
"step": 6700
},
{
"epoch": 1.3274310020773568,
"grad_norm": 0.02295109531842834,
"learning_rate": 6.834457299583768e-05,
"loss": 0.0049,
"step": 6710
},
{
"epoch": 1.329409437135226,
"grad_norm": 0.02485541562666531,
"learning_rate": 6.823745716213377e-05,
"loss": 0.0049,
"step": 6720
},
{
"epoch": 1.3313878721930952,
"grad_norm": 0.02498484412071572,
"learning_rate": 6.813024469535236e-05,
"loss": 0.0049,
"step": 6730
},
{
"epoch": 1.3333663072509645,
"grad_norm": 0.027189658475958337,
"learning_rate": 6.802293616356997e-05,
"loss": 0.0049,
"step": 6740
},
{
"epoch": 1.3353447423088336,
"grad_norm": 0.029563245055805085,
"learning_rate": 6.791553213537209e-05,
"loss": 0.005,
"step": 6750
},
{
"epoch": 1.337323177366703,
"grad_norm": 0.028342723767653663,
"learning_rate": 6.780803317985021e-05,
"loss": 0.0049,
"step": 6760
},
{
"epoch": 1.3393016124245722,
"grad_norm": 0.027594338923669113,
"learning_rate": 6.77004398665988e-05,
"loss": 0.0049,
"step": 6770
},
{
"epoch": 1.3412800474824413,
"grad_norm": 0.023770431354564612,
"learning_rate": 6.759275276571233e-05,
"loss": 0.005,
"step": 6780
},
{
"epoch": 1.3432584825403107,
"grad_norm": 0.02535148607418614,
"learning_rate": 6.748497244778214e-05,
"loss": 0.0049,
"step": 6790
},
{
"epoch": 1.3452369175981798,
"grad_norm": 0.02412061229590485,
"learning_rate": 6.737709948389356e-05,
"loss": 0.0048,
"step": 6800
},
{
"epoch": 1.347215352656049,
"grad_norm": 0.028408998303324638,
"learning_rate": 6.726913444562283e-05,
"loss": 0.0049,
"step": 6810
},
{
"epoch": 1.3491937877139182,
"grad_norm": 0.022292081115544763,
"learning_rate": 6.716107790503398e-05,
"loss": 0.0049,
"step": 6820
},
{
"epoch": 1.3511722227717875,
"grad_norm": 0.020630433745954694,
"learning_rate": 6.705293043467589e-05,
"loss": 0.0048,
"step": 6830
},
{
"epoch": 1.3531506578296568,
"grad_norm": 0.026083963085026748,
"learning_rate": 6.69446926075793e-05,
"loss": 0.0049,
"step": 6840
},
{
"epoch": 1.355129092887526,
"grad_norm": 0.027846745422480855,
"learning_rate": 6.683636499725365e-05,
"loss": 0.0049,
"step": 6850
},
{
"epoch": 1.3571075279453952,
"grad_norm": 0.02211691484926181,
"learning_rate": 6.672794817768414e-05,
"loss": 0.0048,
"step": 6860
},
{
"epoch": 1.3590859630032643,
"grad_norm": 0.022552927649894396,
"learning_rate": 6.661944272332867e-05,
"loss": 0.0049,
"step": 6870
},
{
"epoch": 1.3610643980611337,
"grad_norm": 0.0223102769227007,
"learning_rate": 6.651084920911473e-05,
"loss": 0.0049,
"step": 6880
},
{
"epoch": 1.3630428331190028,
"grad_norm": 0.026766798445256215,
"learning_rate": 6.640216821043646e-05,
"loss": 0.0048,
"step": 6890
},
{
"epoch": 1.365021268176872,
"grad_norm": 0.02627206385174102,
"learning_rate": 6.629340030315148e-05,
"loss": 0.0049,
"step": 6900
},
{
"epoch": 1.3669997032347414,
"grad_norm": 0.024950582097184044,
"learning_rate": 6.618454606357796e-05,
"loss": 0.0049,
"step": 6910
},
{
"epoch": 1.3689781382926105,
"grad_norm": 0.022283301061069117,
"learning_rate": 6.607560606849149e-05,
"loss": 0.0049,
"step": 6920
},
{
"epoch": 1.3709565733504798,
"grad_norm": 0.02703236330449557,
"learning_rate": 6.596658089512208e-05,
"loss": 0.0049,
"step": 6930
},
{
"epoch": 1.372935008408349,
"grad_norm": 0.023734236378066768,
"learning_rate": 6.585747112115096e-05,
"loss": 0.0049,
"step": 6940
},
{
"epoch": 1.3749134434662182,
"grad_norm": 0.026125558866355304,
"learning_rate": 6.574827732470779e-05,
"loss": 0.0049,
"step": 6950
},
{
"epoch": 1.3768918785240873,
"grad_norm": 0.023051062226622353,
"learning_rate": 6.563900008436727e-05,
"loss": 0.0049,
"step": 6960
},
{
"epoch": 1.3788703135819567,
"grad_norm": 0.020542191153885114,
"learning_rate": 6.552963997914635e-05,
"loss": 0.0049,
"step": 6970
},
{
"epoch": 1.380848748639826,
"grad_norm": 0.026052113733475147,
"learning_rate": 6.542019758850097e-05,
"loss": 0.0049,
"step": 6980
},
{
"epoch": 1.382827183697695,
"grad_norm": 0.02496784580776075,
"learning_rate": 6.531067349232313e-05,
"loss": 0.0048,
"step": 6990
},
{
"epoch": 1.3848056187555644,
"grad_norm": 0.01590636747145238,
"learning_rate": 6.52010682709377e-05,
"loss": 0.0049,
"step": 7000
},
{
"epoch": 1.3867840538134335,
"grad_norm": 0.027469350244893512,
"learning_rate": 6.509138250509946e-05,
"loss": 0.0049,
"step": 7010
},
{
"epoch": 1.3887624888713028,
"grad_norm": 0.0264551588766423,
"learning_rate": 6.498161677598993e-05,
"loss": 0.0048,
"step": 7020
},
{
"epoch": 1.390740923929172,
"grad_norm": 0.021346071187265794,
"learning_rate": 6.48717716652143e-05,
"loss": 0.0048,
"step": 7030
},
{
"epoch": 1.3927193589870412,
"grad_norm": 0.02522732494632548,
"learning_rate": 6.476184775479841e-05,
"loss": 0.0049,
"step": 7040
},
{
"epoch": 1.3946977940449106,
"grad_norm": 0.02205428976797274,
"learning_rate": 6.465184562718563e-05,
"loss": 0.0048,
"step": 7050
},
{
"epoch": 1.3966762291027797,
"grad_norm": 0.028252858099771048,
"learning_rate": 6.454176586523373e-05,
"loss": 0.0049,
"step": 7060
},
{
"epoch": 1.398654664160649,
"grad_norm": 0.02833434603097972,
"learning_rate": 6.443160905221187e-05,
"loss": 0.005,
"step": 7070
},
{
"epoch": 1.400633099218518,
"grad_norm": 0.024838281820746037,
"learning_rate": 6.432137577179749e-05,
"loss": 0.0049,
"step": 7080
},
{
"epoch": 1.4026115342763874,
"grad_norm": 0.026727042321422482,
"learning_rate": 6.421106660807314e-05,
"loss": 0.0049,
"step": 7090
},
{
"epoch": 1.4045899693342565,
"grad_norm": 0.025912633826529467,
"learning_rate": 6.410068214552351e-05,
"loss": 0.0049,
"step": 7100
},
{
"epoch": 1.4065684043921258,
"grad_norm": 0.02591054082753992,
"learning_rate": 6.399022296903224e-05,
"loss": 0.0048,
"step": 7110
},
{
"epoch": 1.4085468394499951,
"grad_norm": 0.024004154745299346,
"learning_rate": 6.387968966387881e-05,
"loss": 0.0049,
"step": 7120
},
{
"epoch": 1.4105252745078642,
"grad_norm": 0.01942513431760325,
"learning_rate": 6.376908281573556e-05,
"loss": 0.0049,
"step": 7130
},
{
"epoch": 1.4125037095657336,
"grad_norm": 0.022023011373024166,
"learning_rate": 6.365840301066445e-05,
"loss": 0.005,
"step": 7140
},
{
"epoch": 1.4144821446236027,
"grad_norm": 0.02433637151571739,
"learning_rate": 6.354765083511401e-05,
"loss": 0.0048,
"step": 7150
},
{
"epoch": 1.416460579681472,
"grad_norm": 0.018612166637251227,
"learning_rate": 6.343682687591626e-05,
"loss": 0.0049,
"step": 7160
},
{
"epoch": 1.418439014739341,
"grad_norm": 0.01818341757388756,
"learning_rate": 6.332593172028356e-05,
"loss": 0.0049,
"step": 7170
},
{
"epoch": 1.4204174497972104,
"grad_norm": 0.02239642667811102,
"learning_rate": 6.32149659558055e-05,
"loss": 0.0049,
"step": 7180
},
{
"epoch": 1.4223958848550797,
"grad_norm": 0.024426867036918657,
"learning_rate": 6.310393017044582e-05,
"loss": 0.0049,
"step": 7190
},
{
"epoch": 1.4243743199129488,
"grad_norm": 0.03248723228388698,
"learning_rate": 6.299282495253925e-05,
"loss": 0.0048,
"step": 7200
},
{
"epoch": 1.4263527549708181,
"grad_norm": 0.022518368780678488,
"learning_rate": 6.288165089078842e-05,
"loss": 0.0049,
"step": 7210
},
{
"epoch": 1.4283311900286872,
"grad_norm": 0.030563054125574175,
"learning_rate": 6.277040857426077e-05,
"loss": 0.0049,
"step": 7220
},
{
"epoch": 1.4303096250865566,
"grad_norm": 0.01591548925733016,
"learning_rate": 6.265909859238535e-05,
"loss": 0.0049,
"step": 7230
},
{
"epoch": 1.4322880601444257,
"grad_norm": 0.023737749693154418,
"learning_rate": 6.254772153494977e-05,
"loss": 0.0048,
"step": 7240
},
{
"epoch": 1.434266495202295,
"grad_norm": 0.024566415006400374,
"learning_rate": 6.243627799209703e-05,
"loss": 0.0048,
"step": 7250
},
{
"epoch": 1.4362449302601643,
"grad_norm": 0.02103585773093873,
"learning_rate": 6.232476855432243e-05,
"loss": 0.0049,
"step": 7260
},
{
"epoch": 1.4382233653180334,
"grad_norm": 0.02425126065125599,
"learning_rate": 6.22131938124704e-05,
"loss": 0.0048,
"step": 7270
},
{
"epoch": 1.4402018003759027,
"grad_norm": 0.023383511669526156,
"learning_rate": 6.210155435773138e-05,
"loss": 0.0048,
"step": 7280
},
{
"epoch": 1.4421802354337718,
"grad_norm": 0.023465475534607146,
"learning_rate": 6.198985078163876e-05,
"loss": 0.0048,
"step": 7290
},
{
"epoch": 1.4441586704916411,
"grad_norm": 0.025555048305625532,
"learning_rate": 6.187808367606561e-05,
"loss": 0.0049,
"step": 7300
},
{
"epoch": 1.4461371055495102,
"grad_norm": 0.023020287781069366,
"learning_rate": 6.176625363322163e-05,
"loss": 0.0049,
"step": 7310
},
{
"epoch": 1.4481155406073796,
"grad_norm": 0.02736573969516962,
"learning_rate": 6.165436124565007e-05,
"loss": 0.0049,
"step": 7320
},
{
"epoch": 1.450093975665249,
"grad_norm": 0.02028226804230375,
"learning_rate": 6.154240710622445e-05,
"loss": 0.0049,
"step": 7330
},
{
"epoch": 1.452072410723118,
"grad_norm": 0.024787393447659962,
"learning_rate": 6.143039180814548e-05,
"loss": 0.0048,
"step": 7340
},
{
"epoch": 1.4540508457809873,
"grad_norm": 0.01651030937799272,
"learning_rate": 6.131831594493799e-05,
"loss": 0.0048,
"step": 7350
},
{
"epoch": 1.4560292808388564,
"grad_norm": 0.027560274693314745,
"learning_rate": 6.120618011044766e-05,
"loss": 0.0048,
"step": 7360
},
{
"epoch": 1.4580077158967257,
"grad_norm": 0.03016108676990367,
"learning_rate": 6.109398489883795e-05,
"loss": 0.0048,
"step": 7370
},
{
"epoch": 1.4599861509545948,
"grad_norm": 0.02818302216156242,
"learning_rate": 6.0981730904586954e-05,
"loss": 0.0048,
"step": 7380
},
{
"epoch": 1.4619645860124642,
"grad_norm": 0.019359601225490804,
"learning_rate": 6.086941872248424e-05,
"loss": 0.0048,
"step": 7390
},
{
"epoch": 1.4639430210703335,
"grad_norm": 0.0336985821241703,
"learning_rate": 6.0757048947627636e-05,
"loss": 0.0049,
"step": 7400
},
{
"epoch": 1.4659214561282026,
"grad_norm": 0.026841868540368317,
"learning_rate": 6.0644622175420204e-05,
"loss": 0.0049,
"step": 7410
},
{
"epoch": 1.467899891186072,
"grad_norm": 0.016752856016812572,
"learning_rate": 6.0532139001566946e-05,
"loss": 0.0049,
"step": 7420
},
{
"epoch": 1.469878326243941,
"grad_norm": 0.02663374770137727,
"learning_rate": 6.041960002207174e-05,
"loss": 0.0049,
"step": 7430
},
{
"epoch": 1.4718567613018103,
"grad_norm": 0.020649300754154273,
"learning_rate": 6.0307005833234195e-05,
"loss": 0.0049,
"step": 7440
},
{
"epoch": 1.4738351963596794,
"grad_norm": 0.0264696416108313,
"learning_rate": 6.019435703164639e-05,
"loss": 0.0049,
"step": 7450
},
{
"epoch": 1.4758136314175487,
"grad_norm": 0.03021177336898658,
"learning_rate": 6.008165421418981e-05,
"loss": 0.0049,
"step": 7460
},
{
"epoch": 1.477792066475418,
"grad_norm": 0.02442800169356723,
"learning_rate": 5.9968897978032135e-05,
"loss": 0.0048,
"step": 7470
},
{
"epoch": 1.4797705015332872,
"grad_norm": 0.02780062082819394,
"learning_rate": 5.9856088920624096e-05,
"loss": 0.0049,
"step": 7480
},
{
"epoch": 1.4817489365911565,
"grad_norm": 0.02719074869474753,
"learning_rate": 5.974322763969631e-05,
"loss": 0.0049,
"step": 7490
},
{
"epoch": 1.4837273716490256,
"grad_norm": 0.023632284032513724,
"learning_rate": 5.963031473325609e-05,
"loss": 0.0048,
"step": 7500
},
{
"epoch": 1.485705806706895,
"grad_norm": 0.02663730935588133,
"learning_rate": 5.9517350799584307e-05,
"loss": 0.0048,
"step": 7510
},
{
"epoch": 1.487684241764764,
"grad_norm": 0.020707526797521154,
"learning_rate": 5.9404336437232175e-05,
"loss": 0.0049,
"step": 7520
},
{
"epoch": 1.4896626768226333,
"grad_norm": 0.020957378628548653,
"learning_rate": 5.929127224501818e-05,
"loss": 0.0047,
"step": 7530
},
{
"epoch": 1.4916411118805026,
"grad_norm": 0.01683572071594991,
"learning_rate": 5.917815882202477e-05,
"loss": 0.0048,
"step": 7540
},
{
"epoch": 1.4936195469383717,
"grad_norm": 0.023100409964261073,
"learning_rate": 5.9064996767595236e-05,
"loss": 0.0048,
"step": 7550
},
{
"epoch": 1.495597981996241,
"grad_norm": 0.021538180426099042,
"learning_rate": 5.8951786681330625e-05,
"loss": 0.0048,
"step": 7560
},
{
"epoch": 1.4975764170541102,
"grad_norm": 0.025934500250164153,
"learning_rate": 5.883852916308641e-05,
"loss": 0.0048,
"step": 7570
},
{
"epoch": 1.4995548521119795,
"grad_norm": 0.023801998834826102,
"learning_rate": 5.872522481296941e-05,
"loss": 0.0048,
"step": 7580
},
{
"epoch": 1.5015332871698486,
"grad_norm": 0.028302345715488564,
"learning_rate": 5.861187423133464e-05,
"loss": 0.0049,
"step": 7590
},
{
"epoch": 1.503511722227718,
"grad_norm": 0.021453927969748942,
"learning_rate": 5.8498478018782e-05,
"loss": 0.0048,
"step": 7600
},
{
"epoch": 1.5054901572855872,
"grad_norm": 0.031183956890326207,
"learning_rate": 5.83850367761532e-05,
"loss": 0.0048,
"step": 7610
},
{
"epoch": 1.5074685923434563,
"grad_norm": 0.02520879546696497,
"learning_rate": 5.8271551104528554e-05,
"loss": 0.0048,
"step": 7620
},
{
"epoch": 1.5094470274013254,
"grad_norm": 0.017077779870025487,
"learning_rate": 5.815802160522379e-05,
"loss": 0.0048,
"step": 7630
},
{
"epoch": 1.5114254624591947,
"grad_norm": 0.024311487059294992,
"learning_rate": 5.804444887978684e-05,
"loss": 0.0048,
"step": 7640
},
{
"epoch": 1.513403897517064,
"grad_norm": 0.022511707533268758,
"learning_rate": 5.7930833529994696e-05,
"loss": 0.0048,
"step": 7650
},
{
"epoch": 1.5153823325749332,
"grad_norm": 0.019302721849379843,
"learning_rate": 5.7817176157850194e-05,
"loss": 0.0048,
"step": 7660
},
{
"epoch": 1.5173607676328025,
"grad_norm": 0.01818277551220941,
"learning_rate": 5.770347736557884e-05,
"loss": 0.0048,
"step": 7670
},
{
"epoch": 1.5193392026906718,
"grad_norm": 0.022567983662892498,
"learning_rate": 5.758973775562557e-05,
"loss": 0.0048,
"step": 7680
},
{
"epoch": 1.521317637748541,
"grad_norm": 0.020646658163709475,
"learning_rate": 5.7475957930651646e-05,
"loss": 0.0048,
"step": 7690
},
{
"epoch": 1.52329607280641,
"grad_norm": 0.025829097938863173,
"learning_rate": 5.736213849353138e-05,
"loss": 0.0048,
"step": 7700
},
{
"epoch": 1.5252745078642793,
"grad_norm": 0.022858797715257336,
"learning_rate": 5.724828004734899e-05,
"loss": 0.0048,
"step": 7710
},
{
"epoch": 1.5272529429221486,
"grad_norm": 0.018949956202470856,
"learning_rate": 5.713438319539538e-05,
"loss": 0.0048,
"step": 7720
},
{
"epoch": 1.5292313779800177,
"grad_norm": 0.02073682615085955,
"learning_rate": 5.702044854116494e-05,
"loss": 0.0048,
"step": 7730
},
{
"epoch": 1.531209813037887,
"grad_norm": 0.01919166606947114,
"learning_rate": 5.6906476688352406e-05,
"loss": 0.0048,
"step": 7740
},
{
"epoch": 1.5331882480957564,
"grad_norm": 0.02493198525796265,
"learning_rate": 5.679246824084955e-05,
"loss": 0.0048,
"step": 7750
},
{
"epoch": 1.5351666831536255,
"grad_norm": 0.019839457914728072,
"learning_rate": 5.6678423802742074e-05,
"loss": 0.0048,
"step": 7760
},
{
"epoch": 1.5371451182114946,
"grad_norm": 0.021870198324745398,
"learning_rate": 5.65643439783064e-05,
"loss": 0.0048,
"step": 7770
},
{
"epoch": 1.539123553269364,
"grad_norm": 0.0237307286358397,
"learning_rate": 5.645022937200641e-05,
"loss": 0.0048,
"step": 7780
},
{
"epoch": 1.5411019883272332,
"grad_norm": 0.027724716672988283,
"learning_rate": 5.6336080588490326e-05,
"loss": 0.0048,
"step": 7790
},
{
"epoch": 1.5430804233851023,
"grad_norm": 0.027843426534079726,
"learning_rate": 5.622189823258743e-05,
"loss": 0.0049,
"step": 7800
},
{
"epoch": 1.5450588584429716,
"grad_norm": 0.02607362632198182,
"learning_rate": 5.610768290930488e-05,
"loss": 0.0049,
"step": 7810
},
{
"epoch": 1.547037293500841,
"grad_norm": 0.026682778654530084,
"learning_rate": 5.599343522382454e-05,
"loss": 0.0048,
"step": 7820
},
{
"epoch": 1.54901572855871,
"grad_norm": 0.02550446535152425,
"learning_rate": 5.587915578149976e-05,
"loss": 0.0048,
"step": 7830
},
{
"epoch": 1.5509941636165792,
"grad_norm": 0.036846428550849096,
"learning_rate": 5.576484518785211e-05,
"loss": 0.0048,
"step": 7840
},
{
"epoch": 1.5529725986744485,
"grad_norm": 0.022764071514795772,
"learning_rate": 5.5650504048568255e-05,
"loss": 0.0049,
"step": 7850
},
{
"epoch": 1.5549510337323178,
"grad_norm": 0.01902070429154165,
"learning_rate": 5.553613296949671e-05,
"loss": 0.0048,
"step": 7860
},
{
"epoch": 1.556929468790187,
"grad_norm": 0.021653370564554325,
"learning_rate": 5.542173255664462e-05,
"loss": 0.0048,
"step": 7870
},
{
"epoch": 1.5589079038480562,
"grad_norm": 0.018709360327366815,
"learning_rate": 5.530730341617455e-05,
"loss": 0.0049,
"step": 7880
},
{
"epoch": 1.5608863389059255,
"grad_norm": 0.024907016623923015,
"learning_rate": 5.519284615440129e-05,
"loss": 0.0049,
"step": 7890
},
{
"epoch": 1.5628647739637946,
"grad_norm": 0.020810291626932455,
"learning_rate": 5.5078361377788645e-05,
"loss": 0.0048,
"step": 7900
},
{
"epoch": 1.5648432090216637,
"grad_norm": 0.023808078170613648,
"learning_rate": 5.4963849692946166e-05,
"loss": 0.0048,
"step": 7910
},
{
"epoch": 1.566821644079533,
"grad_norm": 0.0367525581332971,
"learning_rate": 5.484931170662606e-05,
"loss": 0.0049,
"step": 7920
},
{
"epoch": 1.5688000791374024,
"grad_norm": 0.022551225199196997,
"learning_rate": 5.4734748025719804e-05,
"loss": 0.0048,
"step": 7930
},
{
"epoch": 1.5707785141952715,
"grad_norm": 0.022848839458889682,
"learning_rate": 5.462015925725509e-05,
"loss": 0.0048,
"step": 7940
},
{
"epoch": 1.5727569492531408,
"grad_norm": 0.019968881221698088,
"learning_rate": 5.4505546008392505e-05,
"loss": 0.0049,
"step": 7950
},
{
"epoch": 1.5747353843110101,
"grad_norm": 0.017322318898785605,
"learning_rate": 5.439090888642237e-05,
"loss": 0.0049,
"step": 7960
},
{
"epoch": 1.5767138193688792,
"grad_norm": 0.023095079922395647,
"learning_rate": 5.427624849876145e-05,
"loss": 0.0048,
"step": 7970
},
{
"epoch": 1.5786922544267483,
"grad_norm": 0.0271054738031546,
"learning_rate": 5.4161565452949856e-05,
"loss": 0.0049,
"step": 7980
},
{
"epoch": 1.5806706894846176,
"grad_norm": 0.016154171589055803,
"learning_rate": 5.40468603566477e-05,
"loss": 0.0048,
"step": 7990
},
{
"epoch": 1.582649124542487,
"grad_norm": 0.018963396388731128,
"learning_rate": 5.393213381763197e-05,
"loss": 0.0048,
"step": 8000
},
{
"epoch": 1.584627559600356,
"grad_norm": 0.03024017532374125,
"learning_rate": 5.3817386443793235e-05,
"loss": 0.0048,
"step": 8010
},
{
"epoch": 1.5866059946582254,
"grad_norm": 0.02005894544007958,
"learning_rate": 5.370261884313249e-05,
"loss": 0.0048,
"step": 8020
},
{
"epoch": 1.5885844297160947,
"grad_norm": 0.02270151289393821,
"learning_rate": 5.35878316237579e-05,
"loss": 0.0048,
"step": 8030
},
{
"epoch": 1.5905628647739638,
"grad_norm": 0.02694085639949798,
"learning_rate": 5.347302539388155e-05,
"loss": 0.0048,
"step": 8040
},
{
"epoch": 1.592541299831833,
"grad_norm": 0.020411613053044713,
"learning_rate": 5.335820076181629e-05,
"loss": 0.0049,
"step": 8050
},
{
"epoch": 1.5945197348897022,
"grad_norm": 0.026396643413527647,
"learning_rate": 5.324335833597245e-05,
"loss": 0.0049,
"step": 8060
},
{
"epoch": 1.5964981699475715,
"grad_norm": 0.02946601659927018,
"learning_rate": 5.312849872485468e-05,
"loss": 0.0049,
"step": 8070
},
{
"epoch": 1.5984766050054406,
"grad_norm": 0.019399403457979802,
"learning_rate": 5.3013622537058635e-05,
"loss": 0.0048,
"step": 8080
},
{
"epoch": 1.60045504006331,
"grad_norm": 0.028151245225997528,
"learning_rate": 5.289873038126786e-05,
"loss": 0.0049,
"step": 8090
},
{
"epoch": 1.6024334751211793,
"grad_norm": 0.019145229592899177,
"learning_rate": 5.2783822866250466e-05,
"loss": 0.0048,
"step": 8100
},
{
"epoch": 1.6044119101790484,
"grad_norm": 0.022618139233125348,
"learning_rate": 5.266890060085595e-05,
"loss": 0.0049,
"step": 8110
},
{
"epoch": 1.6063903452369175,
"grad_norm": 0.024380042431098786,
"learning_rate": 5.255396419401198e-05,
"loss": 0.0049,
"step": 8120
},
{
"epoch": 1.6083687802947868,
"grad_norm": 0.02186619179750482,
"learning_rate": 5.2439014254721166e-05,
"loss": 0.0048,
"step": 8130
},
{
"epoch": 1.6103472153526561,
"grad_norm": 0.02485538057290849,
"learning_rate": 5.232405139205778e-05,
"loss": 0.0049,
"step": 8140
},
{
"epoch": 1.6123256504105252,
"grad_norm": 0.017373786329301925,
"learning_rate": 5.220907621516461e-05,
"loss": 0.0048,
"step": 8150
},
{
"epoch": 1.6143040854683945,
"grad_norm": 0.023432943136548662,
"learning_rate": 5.209408933324967e-05,
"loss": 0.0048,
"step": 8160
},
{
"epoch": 1.6162825205262639,
"grad_norm": 0.02212606061353163,
"learning_rate": 5.197909135558302e-05,
"loss": 0.0048,
"step": 8170
},
{
"epoch": 1.618260955584133,
"grad_norm": 0.031994521911694954,
"learning_rate": 5.186408289149346e-05,
"loss": 0.0047,
"step": 8180
},
{
"epoch": 1.620239390642002,
"grad_norm": 0.025055712719532375,
"learning_rate": 5.1749064550365414e-05,
"loss": 0.0048,
"step": 8190
},
{
"epoch": 1.6222178256998714,
"grad_norm": 0.025094047563868113,
"learning_rate": 5.163403694163559e-05,
"loss": 0.0048,
"step": 8200
},
{
"epoch": 1.6241962607577407,
"grad_norm": 0.02689651829188429,
"learning_rate": 5.1519000674789833e-05,
"loss": 0.0049,
"step": 8210
},
{
"epoch": 1.6261746958156098,
"grad_norm": 0.019725878566961837,
"learning_rate": 5.140395635935987e-05,
"loss": 0.0048,
"step": 8220
},
{
"epoch": 1.6281531308734791,
"grad_norm": 0.019670989449714096,
"learning_rate": 5.128890460492004e-05,
"loss": 0.0048,
"step": 8230
},
{
"epoch": 1.6301315659313484,
"grad_norm": 0.02787317206467844,
"learning_rate": 5.117384602108413e-05,
"loss": 0.0048,
"step": 8240
},
{
"epoch": 1.6321100009892175,
"grad_norm": 0.020059819526115834,
"learning_rate": 5.1058781217502084e-05,
"loss": 0.0048,
"step": 8250
},
{
"epoch": 1.6340884360470866,
"grad_norm": 0.02141805420758362,
"learning_rate": 5.094371080385685e-05,
"loss": 0.0048,
"step": 8260
},
{
"epoch": 1.636066871104956,
"grad_norm": 0.035841507490396364,
"learning_rate": 5.082863538986102e-05,
"loss": 0.0049,
"step": 8270
},
{
"epoch": 1.6380453061628253,
"grad_norm": 0.021607222709096215,
"learning_rate": 5.071355558525379e-05,
"loss": 0.0048,
"step": 8280
},
{
"epoch": 1.6400237412206944,
"grad_norm": 0.020114542710176233,
"learning_rate": 5.059847199979752e-05,
"loss": 0.0048,
"step": 8290
},
{
"epoch": 1.6420021762785637,
"grad_norm": 0.01998383316755519,
"learning_rate": 5.048338524327466e-05,
"loss": 0.0048,
"step": 8300
},
{
"epoch": 1.643980611336433,
"grad_norm": 0.029085218060515686,
"learning_rate": 5.036829592548446e-05,
"loss": 0.0049,
"step": 8310
},
{
"epoch": 1.6459590463943021,
"grad_norm": 0.02468920378643903,
"learning_rate": 5.02532046562397e-05,
"loss": 0.0048,
"step": 8320
},
{
"epoch": 1.6479374814521712,
"grad_norm": 0.027048071998410975,
"learning_rate": 5.013811204536353e-05,
"loss": 0.0048,
"step": 8330
},
{
"epoch": 1.6499159165100405,
"grad_norm": 0.027812263235141313,
"learning_rate": 5.002301870268623e-05,
"loss": 0.0048,
"step": 8340
},
{
"epoch": 1.6518943515679099,
"grad_norm": 0.021718546550266834,
"learning_rate": 4.9907925238041925e-05,
"loss": 0.0049,
"step": 8350
},
{
"epoch": 1.653872786625779,
"grad_norm": 0.024253605202685102,
"learning_rate": 4.9792832261265384e-05,
"loss": 0.0048,
"step": 8360
},
{
"epoch": 1.6558512216836483,
"grad_norm": 0.03234869896233535,
"learning_rate": 4.967774038218882e-05,
"loss": 0.0048,
"step": 8370
},
{
"epoch": 1.6578296567415176,
"grad_norm": 0.025034835338017876,
"learning_rate": 4.95626502106386e-05,
"loss": 0.0048,
"step": 8380
},
{
"epoch": 1.6598080917993867,
"grad_norm": 0.0233781548695659,
"learning_rate": 4.944756235643205e-05,
"loss": 0.0048,
"step": 8390
},
{
"epoch": 1.6617865268572558,
"grad_norm": 0.01717296313291924,
"learning_rate": 4.933247742937425e-05,
"loss": 0.0047,
"step": 8400
},
{
"epoch": 1.6637649619151251,
"grad_norm": 0.015810032172554556,
"learning_rate": 4.92173960392547e-05,
"loss": 0.0048,
"step": 8410
},
{
"epoch": 1.6657433969729945,
"grad_norm": 0.028446667679901966,
"learning_rate": 4.910231879584427e-05,
"loss": 0.0049,
"step": 8420
},
{
"epoch": 1.6677218320308635,
"grad_norm": 0.026142871160596725,
"learning_rate": 4.8987246308891716e-05,
"loss": 0.0049,
"step": 8430
},
{
"epoch": 1.6697002670887326,
"grad_norm": 0.02074554910475918,
"learning_rate": 4.887217918812072e-05,
"loss": 0.0048,
"step": 8440
},
{
"epoch": 1.6716787021466022,
"grad_norm": 0.02516918432363385,
"learning_rate": 4.875711804322643e-05,
"loss": 0.0048,
"step": 8450
},
{
"epoch": 1.6736571372044713,
"grad_norm": 0.02678320276608983,
"learning_rate": 4.8642063483872415e-05,
"loss": 0.0048,
"step": 8460
},
{
"epoch": 1.6756355722623404,
"grad_norm": 0.023378992858101997,
"learning_rate": 4.85270161196873e-05,
"loss": 0.0048,
"step": 8470
},
{
"epoch": 1.6776140073202097,
"grad_norm": 0.039857865995852323,
"learning_rate": 4.841197656026159e-05,
"loss": 0.0048,
"step": 8480
},
{
"epoch": 1.679592442378079,
"grad_norm": 0.026217422646741425,
"learning_rate": 4.8296945415144466e-05,
"loss": 0.0048,
"step": 8490
},
{
"epoch": 1.6815708774359481,
"grad_norm": 0.016006677731834777,
"learning_rate": 4.818192329384047e-05,
"loss": 0.0048,
"step": 8500
},
{
"epoch": 1.6835493124938172,
"grad_norm": 0.018928220200067198,
"learning_rate": 4.806691080580638e-05,
"loss": 0.0048,
"step": 8510
},
{
"epoch": 1.6855277475516868,
"grad_norm": 0.02467817107425464,
"learning_rate": 4.795190856044792e-05,
"loss": 0.0048,
"step": 8520
},
{
"epoch": 1.6875061826095559,
"grad_norm": 0.021081572020845266,
"learning_rate": 4.783691716711652e-05,
"loss": 0.0048,
"step": 8530
},
{
"epoch": 1.689484617667425,
"grad_norm": 0.02969109387849964,
"learning_rate": 4.772193723510613e-05,
"loss": 0.0048,
"step": 8540
},
{
"epoch": 1.6914630527252943,
"grad_norm": 0.034031798126584585,
"learning_rate": 4.7606969373649986e-05,
"loss": 0.0049,
"step": 8550
},
{
"epoch": 1.6934414877831636,
"grad_norm": 0.03356129105152546,
"learning_rate": 4.7492014191917334e-05,
"loss": 0.0048,
"step": 8560
},
{
"epoch": 1.6954199228410327,
"grad_norm": 0.024449184212900554,
"learning_rate": 4.737707229901022e-05,
"loss": 0.0049,
"step": 8570
},
{
"epoch": 1.6973983578989018,
"grad_norm": 0.02761998265365707,
"learning_rate": 4.726214430396033e-05,
"loss": 0.0049,
"step": 8580
},
{
"epoch": 1.6993767929567714,
"grad_norm": 0.02310066845634119,
"learning_rate": 4.71472308157257e-05,
"loss": 0.0049,
"step": 8590
},
{
"epoch": 1.7013552280146405,
"grad_norm": 0.02509092163404365,
"learning_rate": 4.7032332443187467e-05,
"loss": 0.0049,
"step": 8600
},
{
"epoch": 1.7033336630725096,
"grad_norm": 0.027218553835803856,
"learning_rate": 4.69174497951467e-05,
"loss": 0.0049,
"step": 8610
},
{
"epoch": 1.7053120981303789,
"grad_norm": 0.02682227701689821,
"learning_rate": 4.680258348032115e-05,
"loss": 0.0048,
"step": 8620
},
{
"epoch": 1.7072905331882482,
"grad_norm": 0.0207056616105931,
"learning_rate": 4.668773410734201e-05,
"loss": 0.0048,
"step": 8630
},
{
"epoch": 1.7092689682461173,
"grad_norm": 0.042783268481699284,
"learning_rate": 4.657290228475073e-05,
"loss": 0.0049,
"step": 8640
},
{
"epoch": 1.7112474033039864,
"grad_norm": 0.023781049418847428,
"learning_rate": 4.6458088620995736e-05,
"loss": 0.0048,
"step": 8650
},
{
"epoch": 1.713225838361856,
"grad_norm": 0.0501302968719949,
"learning_rate": 4.6343293724429266e-05,
"loss": 0.0049,
"step": 8660
},
{
"epoch": 1.715204273419725,
"grad_norm": 0.0226791961271776,
"learning_rate": 4.622851820330412e-05,
"loss": 0.0049,
"step": 8670
},
{
"epoch": 1.7171827084775941,
"grad_norm": 0.027425462908513963,
"learning_rate": 4.6113762665770385e-05,
"loss": 0.0049,
"step": 8680
},
{
"epoch": 1.7191611435354635,
"grad_norm": 0.02799600495762066,
"learning_rate": 4.599902771987232e-05,
"loss": 0.0048,
"step": 8690
},
{
"epoch": 1.7211395785933328,
"grad_norm": 0.022890748171564914,
"learning_rate": 4.588431397354508e-05,
"loss": 0.0048,
"step": 8700
},
{
"epoch": 1.7231180136512019,
"grad_norm": 0.02101488357847659,
"learning_rate": 4.576962203461144e-05,
"loss": 0.0048,
"step": 8710
},
{
"epoch": 1.725096448709071,
"grad_norm": 0.02241346162126945,
"learning_rate": 4.5654952510778646e-05,
"loss": 0.0048,
"step": 8720
},
{
"epoch": 1.7270748837669405,
"grad_norm": 0.02292254077996637,
"learning_rate": 4.5540306009635194e-05,
"loss": 0.0049,
"step": 8730
},
{
"epoch": 1.7290533188248096,
"grad_norm": 0.023561897893025395,
"learning_rate": 4.54256831386476e-05,
"loss": 0.0049,
"step": 8740
},
{
"epoch": 1.7310317538826787,
"grad_norm": 0.016217048922419174,
"learning_rate": 4.531108450515712e-05,
"loss": 0.0048,
"step": 8750
},
{
"epoch": 1.733010188940548,
"grad_norm": 0.03613925827900919,
"learning_rate": 4.5196510716376645e-05,
"loss": 0.0048,
"step": 8760
},
{
"epoch": 1.7349886239984174,
"grad_norm": 0.02802485860452221,
"learning_rate": 4.508196237938741e-05,
"loss": 0.0049,
"step": 8770
},
{
"epoch": 1.7369670590562865,
"grad_norm": 0.0230326625840555,
"learning_rate": 4.4967440101135736e-05,
"loss": 0.0048,
"step": 8780
},
{
"epoch": 1.7389454941141556,
"grad_norm": 0.023795477406810844,
"learning_rate": 4.485294448842996e-05,
"loss": 0.0048,
"step": 8790
},
{
"epoch": 1.7409239291720249,
"grad_norm": 0.0254473258060912,
"learning_rate": 4.473847614793705e-05,
"loss": 0.0049,
"step": 8800
},
{
"epoch": 1.7429023642298942,
"grad_norm": 0.029128656680167786,
"learning_rate": 4.462403568617951e-05,
"loss": 0.0048,
"step": 8810
},
{
"epoch": 1.7448807992877633,
"grad_norm": 0.032541390941594,
"learning_rate": 4.4509623709532125e-05,
"loss": 0.0048,
"step": 8820
},
{
"epoch": 1.7468592343456326,
"grad_norm": 0.019850307109647934,
"learning_rate": 4.439524082421872e-05,
"loss": 0.0048,
"step": 8830
},
{
"epoch": 1.748837669403502,
"grad_norm": 0.014695076308849938,
"learning_rate": 4.4280887636309035e-05,
"loss": 0.0048,
"step": 8840
},
{
"epoch": 1.750816104461371,
"grad_norm": 0.025113867597687464,
"learning_rate": 4.416656475171537e-05,
"loss": 0.0048,
"step": 8850
},
{
"epoch": 1.7527945395192401,
"grad_norm": 0.03692869657291486,
"learning_rate": 4.405227277618953e-05,
"loss": 0.0049,
"step": 8860
},
{
"epoch": 1.7547729745771095,
"grad_norm": 0.016978215931937657,
"learning_rate": 4.393801231531952e-05,
"loss": 0.0048,
"step": 8870
},
{
"epoch": 1.7567514096349788,
"grad_norm": 0.025894459125324388,
"learning_rate": 4.382378397452635e-05,
"loss": 0.0048,
"step": 8880
},
{
"epoch": 1.7587298446928479,
"grad_norm": 0.028350587123111008,
"learning_rate": 4.3709588359060884e-05,
"loss": 0.0048,
"step": 8890
},
{
"epoch": 1.7607082797507172,
"grad_norm": 0.027603828923600936,
"learning_rate": 4.359542607400051e-05,
"loss": 0.0048,
"step": 8900
},
{
"epoch": 1.7626867148085865,
"grad_norm": 0.028736061599296748,
"learning_rate": 4.34812977242461e-05,
"loss": 0.0048,
"step": 8910
},
{
"epoch": 1.7646651498664556,
"grad_norm": 0.03338240756339308,
"learning_rate": 4.336720391451864e-05,
"loss": 0.0048,
"step": 8920
},
{
"epoch": 1.7666435849243247,
"grad_norm": 0.014374769622157058,
"learning_rate": 4.325314524935614e-05,
"loss": 0.0048,
"step": 8930
},
{
"epoch": 1.768622019982194,
"grad_norm": 0.01924029773834809,
"learning_rate": 4.3139122333110406e-05,
"loss": 0.0048,
"step": 8940
},
{
"epoch": 1.7706004550400634,
"grad_norm": 0.018963752612118974,
"learning_rate": 4.302513576994378e-05,
"loss": 0.0048,
"step": 8950
},
{
"epoch": 1.7725788900979325,
"grad_norm": 0.02075666933174052,
"learning_rate": 4.291118616382603e-05,
"loss": 0.0049,
"step": 8960
},
{
"epoch": 1.7745573251558018,
"grad_norm": 0.0210583493039699,
"learning_rate": 4.279727411853108e-05,
"loss": 0.0048,
"step": 8970
},
{
"epoch": 1.776535760213671,
"grad_norm": 0.016364707343113714,
"learning_rate": 4.268340023763381e-05,
"loss": 0.0048,
"step": 8980
},
{
"epoch": 1.7785141952715402,
"grad_norm": 0.01631848271396228,
"learning_rate": 4.256956512450696e-05,
"loss": 0.0048,
"step": 8990
},
{
"epoch": 1.7804926303294093,
"grad_norm": 0.018093566128389776,
"learning_rate": 4.2455769382317756e-05,
"loss": 0.0048,
"step": 9000
},
{
"epoch": 1.7824710653872786,
"grad_norm": 0.029270434181043904,
"learning_rate": 4.234201361402488e-05,
"loss": 0.0047,
"step": 9010
},
{
"epoch": 1.784449500445148,
"grad_norm": 0.013028248891483947,
"learning_rate": 4.2228298422375166e-05,
"loss": 0.0048,
"step": 9020
},
{
"epoch": 1.786427935503017,
"grad_norm": 0.014648738494098973,
"learning_rate": 4.21146244099005e-05,
"loss": 0.0048,
"step": 9030
},
{
"epoch": 1.7884063705608864,
"grad_norm": 0.024459817938385992,
"learning_rate": 4.2000992178914546e-05,
"loss": 0.0048,
"step": 9040
},
{
"epoch": 1.7903848056187557,
"grad_norm": 0.018185417769204123,
"learning_rate": 4.188740233150956e-05,
"loss": 0.0047,
"step": 9050
},
{
"epoch": 1.7923632406766248,
"grad_norm": 0.01832487781621057,
"learning_rate": 4.177385546955328e-05,
"loss": 0.0047,
"step": 9060
},
{
"epoch": 1.7943416757344939,
"grad_norm": 0.019095956196474563,
"learning_rate": 4.166035219468562e-05,
"loss": 0.0048,
"step": 9070
},
{
"epoch": 1.7963201107923632,
"grad_norm": 0.025944025849730306,
"learning_rate": 4.154689310831558e-05,
"loss": 0.0048,
"step": 9080
},
{
"epoch": 1.7982985458502325,
"grad_norm": 0.022190747499093458,
"learning_rate": 4.1433478811618034e-05,
"loss": 0.0048,
"step": 9090
},
{
"epoch": 1.8002769809081016,
"grad_norm": 0.02061165674153281,
"learning_rate": 4.13201099055305e-05,
"loss": 0.0048,
"step": 9100
},
{
"epoch": 1.802255415965971,
"grad_norm": 0.024088444538493854,
"learning_rate": 4.1206786990750005e-05,
"loss": 0.0048,
"step": 9110
},
{
"epoch": 1.8042338510238403,
"grad_norm": 0.0212833661696764,
"learning_rate": 4.1093510667729906e-05,
"loss": 0.0048,
"step": 9120
},
{
"epoch": 1.8062122860817094,
"grad_norm": 0.021866370318242238,
"learning_rate": 4.098028153667666e-05,
"loss": 0.0048,
"step": 9130
},
{
"epoch": 1.8081907211395785,
"grad_norm": 0.02583271102790413,
"learning_rate": 4.086710019754665e-05,
"loss": 0.0048,
"step": 9140
},
{
"epoch": 1.8101691561974478,
"grad_norm": 0.04056079661106068,
"learning_rate": 4.075396725004308e-05,
"loss": 0.0048,
"step": 9150
},
{
"epoch": 1.812147591255317,
"grad_norm": 0.022141784838086195,
"learning_rate": 4.064088329361274e-05,
"loss": 0.0048,
"step": 9160
},
{
"epoch": 1.8141260263131862,
"grad_norm": 0.02814464077929103,
"learning_rate": 4.05278489274428e-05,
"loss": 0.0048,
"step": 9170
},
{
"epoch": 1.8161044613710555,
"grad_norm": 0.02554721486384196,
"learning_rate": 4.041486475045769e-05,
"loss": 0.0048,
"step": 9180
},
{
"epoch": 1.8180828964289248,
"grad_norm": 0.014938308481380553,
"learning_rate": 4.0301931361315937e-05,
"loss": 0.0048,
"step": 9190
},
{
"epoch": 1.820061331486794,
"grad_norm": 0.022021020958978264,
"learning_rate": 4.0189049358406906e-05,
"loss": 0.0048,
"step": 9200
},
{
"epoch": 1.822039766544663,
"grad_norm": 0.02010588304298297,
"learning_rate": 4.007621933984773e-05,
"loss": 0.0048,
"step": 9210
},
{
"epoch": 1.8240182016025324,
"grad_norm": 0.014149114019488825,
"learning_rate": 3.996344190348007e-05,
"loss": 0.0047,
"step": 9220
},
{
"epoch": 1.8259966366604017,
"grad_norm": 0.014623275085233117,
"learning_rate": 3.985071764686701e-05,
"loss": 0.0047,
"step": 9230
},
{
"epoch": 1.8279750717182708,
"grad_norm": 0.019646074776575553,
"learning_rate": 3.973804716728981e-05,
"loss": 0.0048,
"step": 9240
},
{
"epoch": 1.82995350677614,
"grad_norm": 0.023531995820037635,
"learning_rate": 3.962543106174483e-05,
"loss": 0.0048,
"step": 9250
},
{
"epoch": 1.8319319418340094,
"grad_norm": 0.017616307735970345,
"learning_rate": 3.951286992694029e-05,
"loss": 0.0048,
"step": 9260
},
{
"epoch": 1.8339103768918785,
"grad_norm": 0.017570701201122932,
"learning_rate": 3.940036435929318e-05,
"loss": 0.0048,
"step": 9270
},
{
"epoch": 1.8358888119497476,
"grad_norm": 0.015403021522255823,
"learning_rate": 3.928791495492601e-05,
"loss": 0.0047,
"step": 9280
},
{
"epoch": 1.837867247007617,
"grad_norm": 0.021702505642674747,
"learning_rate": 3.9175522309663724e-05,
"loss": 0.0048,
"step": 9290
},
{
"epoch": 1.8398456820654863,
"grad_norm": 0.026565773010699154,
"learning_rate": 3.906318701903055e-05,
"loss": 0.0048,
"step": 9300
},
{
"epoch": 1.8418241171233554,
"grad_norm": 0.01346635856389289,
"learning_rate": 3.8950909678246784e-05,
"loss": 0.0047,
"step": 9310
},
{
"epoch": 1.8438025521812247,
"grad_norm": 0.013419503013464137,
"learning_rate": 3.883869088222568e-05,
"loss": 0.0048,
"step": 9320
},
{
"epoch": 1.845780987239094,
"grad_norm": 0.0221501466056421,
"learning_rate": 3.872653122557028e-05,
"loss": 0.0048,
"step": 9330
},
{
"epoch": 1.847759422296963,
"grad_norm": 0.01913869505549269,
"learning_rate": 3.861443130257031e-05,
"loss": 0.0048,
"step": 9340
},
{
"epoch": 1.8497378573548322,
"grad_norm": 0.023527625196498717,
"learning_rate": 3.850239170719891e-05,
"loss": 0.0047,
"step": 9350
},
{
"epoch": 1.8517162924127015,
"grad_norm": 0.016426117719563395,
"learning_rate": 3.8390413033109644e-05,
"loss": 0.0048,
"step": 9360
},
{
"epoch": 1.8536947274705708,
"grad_norm": 0.01361763171703126,
"learning_rate": 3.827849587363323e-05,
"loss": 0.0048,
"step": 9370
},
{
"epoch": 1.85567316252844,
"grad_norm": 0.02779640538456879,
"learning_rate": 3.816664082177447e-05,
"loss": 0.0048,
"step": 9380
},
{
"epoch": 1.8576515975863093,
"grad_norm": 0.02347271338717225,
"learning_rate": 3.8054848470209095e-05,
"loss": 0.0048,
"step": 9390
},
{
"epoch": 1.8596300326441786,
"grad_norm": 0.020592385199020687,
"learning_rate": 3.794311941128055e-05,
"loss": 0.0048,
"step": 9400
},
{
"epoch": 1.8616084677020477,
"grad_norm": 0.01573827231527125,
"learning_rate": 3.783145423699701e-05,
"loss": 0.0047,
"step": 9410
},
{
"epoch": 1.8635869027599168,
"grad_norm": 0.022476401745065717,
"learning_rate": 3.771985353902805e-05,
"loss": 0.0047,
"step": 9420
},
{
"epoch": 1.865565337817786,
"grad_norm": 0.016899493895925724,
"learning_rate": 3.7608317908701706e-05,
"loss": 0.0048,
"step": 9430
},
{
"epoch": 1.8675437728756554,
"grad_norm": 0.022699223491583473,
"learning_rate": 3.749684793700115e-05,
"loss": 0.0047,
"step": 9440
},
{
"epoch": 1.8695222079335245,
"grad_norm": 0.01311107648286897,
"learning_rate": 3.738544421456175e-05,
"loss": 0.0047,
"step": 9450
},
{
"epoch": 1.8715006429913938,
"grad_norm": 0.01352967835390026,
"learning_rate": 3.727410733166779e-05,
"loss": 0.0048,
"step": 9460
},
{
"epoch": 1.8734790780492632,
"grad_norm": 0.024742122664916426,
"learning_rate": 3.716283787824939e-05,
"loss": 0.0048,
"step": 9470
},
{
"epoch": 1.8754575131071323,
"grad_norm": 0.027873723205071692,
"learning_rate": 3.7051636443879425e-05,
"loss": 0.0048,
"step": 9480
},
{
"epoch": 1.8774359481650014,
"grad_norm": 0.024712314825693776,
"learning_rate": 3.694050361777034e-05,
"loss": 0.0048,
"step": 9490
},
{
"epoch": 1.8794143832228707,
"grad_norm": 0.025360624827392344,
"learning_rate": 3.682943998877105e-05,
"loss": 0.0048,
"step": 9500
},
{
"epoch": 1.88139281828074,
"grad_norm": 0.025580363021559388,
"learning_rate": 3.6718446145363854e-05,
"loss": 0.0048,
"step": 9510
},
{
"epoch": 1.883371253338609,
"grad_norm": 0.026910078334777474,
"learning_rate": 3.660752267566123e-05,
"loss": 0.0048,
"step": 9520
},
{
"epoch": 1.8853496883964784,
"grad_norm": 0.0124830925774804,
"learning_rate": 3.649667016740281e-05,
"loss": 0.0048,
"step": 9530
},
{
"epoch": 1.8873281234543478,
"grad_norm": 0.014082892048407143,
"learning_rate": 3.638588920795225e-05,
"loss": 0.0048,
"step": 9540
},
{
"epoch": 1.8893065585122168,
"grad_norm": 0.04497905294618717,
"learning_rate": 3.627518038429403e-05,
"loss": 0.0048,
"step": 9550
},
{
"epoch": 1.891284993570086,
"grad_norm": 0.016823527573695946,
"learning_rate": 3.61645442830305e-05,
"loss": 0.0048,
"step": 9560
},
{
"epoch": 1.8932634286279553,
"grad_norm": 0.022836559482041434,
"learning_rate": 3.6053981490378577e-05,
"loss": 0.0048,
"step": 9570
},
{
"epoch": 1.8952418636858246,
"grad_norm": 0.023863382955751446,
"learning_rate": 3.5943492592166836e-05,
"loss": 0.0047,
"step": 9580
},
{
"epoch": 1.8972202987436937,
"grad_norm": 0.01899874777479604,
"learning_rate": 3.5833078173832256e-05,
"loss": 0.0048,
"step": 9590
},
{
"epoch": 1.899198733801563,
"grad_norm": 0.02039452349002635,
"learning_rate": 3.572273882041723e-05,
"loss": 0.0048,
"step": 9600
},
{
"epoch": 1.9011771688594323,
"grad_norm": 0.01732008600770316,
"learning_rate": 3.561247511656637e-05,
"loss": 0.0048,
"step": 9610
},
{
"epoch": 1.9031556039173014,
"grad_norm": 0.02201774356438948,
"learning_rate": 3.5502287646523455e-05,
"loss": 0.0047,
"step": 9620
},
{
"epoch": 1.9051340389751705,
"grad_norm": 0.015155689072533451,
"learning_rate": 3.5392176994128354e-05,
"loss": 0.0047,
"step": 9630
},
{
"epoch": 1.9071124740330399,
"grad_norm": 0.013115977680650344,
"learning_rate": 3.528214374281388e-05,
"loss": 0.0047,
"step": 9640
},
{
"epoch": 1.9090909090909092,
"grad_norm": 0.010999176250657388,
"learning_rate": 3.517218847560274e-05,
"loss": 0.0048,
"step": 9650
},
{
"epoch": 1.9110693441487783,
"grad_norm": 0.011947217253396389,
"learning_rate": 3.506231177510446e-05,
"loss": 0.0048,
"step": 9660
},
{
"epoch": 1.9130477792066476,
"grad_norm": 0.029758818636993766,
"learning_rate": 3.4952514223512235e-05,
"loss": 0.0048,
"step": 9670
},
{
"epoch": 1.915026214264517,
"grad_norm": 0.01424277225916946,
"learning_rate": 3.484279640259989e-05,
"loss": 0.0047,
"step": 9680
},
{
"epoch": 1.917004649322386,
"grad_norm": 0.012542964009551502,
"learning_rate": 3.473315889371881e-05,
"loss": 0.0047,
"step": 9690
},
{
"epoch": 1.918983084380255,
"grad_norm": 0.015739425506641095,
"learning_rate": 3.462360227779482e-05,
"loss": 0.0047,
"step": 9700
},
{
"epoch": 1.9209615194381244,
"grad_norm": 0.018040740676101067,
"learning_rate": 3.4514127135325103e-05,
"loss": 0.0047,
"step": 9710
},
{
"epoch": 1.9229399544959938,
"grad_norm": 0.013773542810568412,
"learning_rate": 3.440473404637519e-05,
"loss": 0.0047,
"step": 9720
},
{
"epoch": 1.9249183895538629,
"grad_norm": 0.010301154019948177,
"learning_rate": 3.4295423590575826e-05,
"loss": 0.0048,
"step": 9730
},
{
"epoch": 1.9268968246117322,
"grad_norm": 0.014492891464198633,
"learning_rate": 3.4186196347119906e-05,
"loss": 0.0047,
"step": 9740
},
{
"epoch": 1.9288752596696015,
"grad_norm": 0.014825966526851987,
"learning_rate": 3.407705289475942e-05,
"loss": 0.0048,
"step": 9750
},
{
"epoch": 1.9308536947274706,
"grad_norm": 0.013704571410445562,
"learning_rate": 3.3967993811802406e-05,
"loss": 0.0048,
"step": 9760
},
{
"epoch": 1.9328321297853397,
"grad_norm": 0.010149497713848347,
"learning_rate": 3.385901967610982e-05,
"loss": 0.0047,
"step": 9770
},
{
"epoch": 1.934810564843209,
"grad_norm": 0.009617942359634816,
"learning_rate": 3.375013106509254e-05,
"loss": 0.0047,
"step": 9780
},
{
"epoch": 1.9367889999010783,
"grad_norm": 0.020580453273161353,
"learning_rate": 3.364132855570829e-05,
"loss": 0.0047,
"step": 9790
},
{
"epoch": 1.9387674349589474,
"grad_norm": 0.016168354389408276,
"learning_rate": 3.353261272445854e-05,
"loss": 0.0048,
"step": 9800
},
{
"epoch": 1.9407458700168165,
"grad_norm": 0.00931203790164073,
"learning_rate": 3.342398414738555e-05,
"loss": 0.0047,
"step": 9810
},
{
"epoch": 1.942724305074686,
"grad_norm": 0.016274801257452898,
"learning_rate": 3.3315443400069186e-05,
"loss": 0.0048,
"step": 9820
},
{
"epoch": 1.9447027401325552,
"grad_norm": 0.01792017909148388,
"learning_rate": 3.3206991057623974e-05,
"loss": 0.0048,
"step": 9830
},
{
"epoch": 1.9466811751904243,
"grad_norm": 0.016452817520733878,
"learning_rate": 3.3098627694696046e-05,
"loss": 0.0047,
"step": 9840
},
{
"epoch": 1.9486596102482936,
"grad_norm": 0.018872462062766507,
"learning_rate": 3.299035388545999e-05,
"loss": 0.0047,
"step": 9850
},
{
"epoch": 1.950638045306163,
"grad_norm": 0.021168624040013857,
"learning_rate": 3.288217020361594e-05,
"loss": 0.0047,
"step": 9860
},
{
"epoch": 1.952616480364032,
"grad_norm": 0.02228806039538108,
"learning_rate": 3.277407722238646e-05,
"loss": 0.0047,
"step": 9870
},
{
"epoch": 1.9545949154219011,
"grad_norm": 0.012821479139683076,
"learning_rate": 3.266607551451357e-05,
"loss": 0.0047,
"step": 9880
},
{
"epoch": 1.9565733504797707,
"grad_norm": 0.015156990266566596,
"learning_rate": 3.255816565225559e-05,
"loss": 0.0047,
"step": 9890
},
{
"epoch": 1.9585517855376398,
"grad_norm": 0.017650079249370197,
"learning_rate": 3.245034820738426e-05,
"loss": 0.0048,
"step": 9900
},
{
"epoch": 1.9605302205955089,
"grad_norm": 0.026689281445409448,
"learning_rate": 3.234262375118161e-05,
"loss": 0.0047,
"step": 9910
},
{
"epoch": 1.9625086556533782,
"grad_norm": 0.01813300899379979,
"learning_rate": 3.223499285443694e-05,
"loss": 0.0048,
"step": 9920
},
{
"epoch": 1.9644870907112475,
"grad_norm": 0.015076337810720509,
"learning_rate": 3.212745608744385e-05,
"loss": 0.0047,
"step": 9930
},
{
"epoch": 1.9664655257691166,
"grad_norm": 0.013752218397611057,
"learning_rate": 3.202001401999715e-05,
"loss": 0.0047,
"step": 9940
},
{
"epoch": 1.9684439608269857,
"grad_norm": 0.01923809362888309,
"learning_rate": 3.191266722138989e-05,
"loss": 0.0048,
"step": 9950
},
{
"epoch": 1.9704223958848552,
"grad_norm": 0.014573927120732374,
"learning_rate": 3.180541626041035e-05,
"loss": 0.0048,
"step": 9960
},
{
"epoch": 1.9724008309427243,
"grad_norm": 0.01880093437329018,
"learning_rate": 3.169826170533896e-05,
"loss": 0.0047,
"step": 9970
},
{
"epoch": 1.9743792660005934,
"grad_norm": 0.01499167441998306,
"learning_rate": 3.1591204123945383e-05,
"loss": 0.0048,
"step": 9980
},
{
"epoch": 1.9763577010584628,
"grad_norm": 0.018225863641493614,
"learning_rate": 3.1484244083485396e-05,
"loss": 0.0048,
"step": 9990
},
{
"epoch": 1.978336136116332,
"grad_norm": 0.01043715453185582,
"learning_rate": 3.137738215069801e-05,
"loss": 0.0047,
"step": 10000
},
{
"epoch": 1.9803145711742012,
"grad_norm": 0.016168999868731465,
"learning_rate": 3.127061889180235e-05,
"loss": 0.0047,
"step": 10010
},
{
"epoch": 1.9822930062320703,
"grad_norm": 0.009931453200800768,
"learning_rate": 3.1163954872494726e-05,
"loss": 0.0047,
"step": 10020
},
{
"epoch": 1.9842714412899398,
"grad_norm": 0.018526102530768276,
"learning_rate": 3.1057390657945653e-05,
"loss": 0.0047,
"step": 10030
},
{
"epoch": 1.986249876347809,
"grad_norm": 0.018038498840359813,
"learning_rate": 3.095092681279674e-05,
"loss": 0.0047,
"step": 10040
},
{
"epoch": 1.988228311405678,
"grad_norm": 0.03859590549103417,
"learning_rate": 3.084456390115786e-05,
"loss": 0.0047,
"step": 10050
},
{
"epoch": 1.9902067464635473,
"grad_norm": 0.025768249428268056,
"learning_rate": 3.0738302486604045e-05,
"loss": 0.0047,
"step": 10060
},
{
"epoch": 1.9921851815214167,
"grad_norm": 0.020338897132106657,
"learning_rate": 3.0632143132172504e-05,
"loss": 0.0047,
"step": 10070
},
{
"epoch": 1.9941636165792858,
"grad_norm": 0.026378658857932825,
"learning_rate": 3.052608640035972e-05,
"loss": 0.0047,
"step": 10080
},
{
"epoch": 1.9961420516371549,
"grad_norm": 0.013912534793320736,
"learning_rate": 3.042013285311839e-05,
"loss": 0.0048,
"step": 10090
},
{
"epoch": 1.9981204866950244,
"grad_norm": 0.014286281874039046,
"learning_rate": 3.0314283051854485e-05,
"loss": 0.0047,
"step": 10100
},
{
"epoch": 2.0,
"grad_norm": 0.028305034408510336,
"learning_rate": 3.0208537557424277e-05,
"loss": 0.0047,
"step": 10110
},
{
"epoch": 2.001978435057869,
"grad_norm": 0.04217146027291757,
"learning_rate": 3.0102896930131354e-05,
"loss": 0.0047,
"step": 10120
},
{
"epoch": 2.0039568701157386,
"grad_norm": 0.017573659708441665,
"learning_rate": 2.9997361729723606e-05,
"loss": 0.0047,
"step": 10130
},
{
"epoch": 2.0059353051736077,
"grad_norm": 0.022835477288420534,
"learning_rate": 2.9891932515390382e-05,
"loss": 0.0047,
"step": 10140
},
{
"epoch": 2.007913740231477,
"grad_norm": 0.011355341282447933,
"learning_rate": 2.9786609845759418e-05,
"loss": 0.0047,
"step": 10150
},
{
"epoch": 2.009892175289346,
"grad_norm": 0.00904882167976142,
"learning_rate": 2.9681394278893903e-05,
"loss": 0.0047,
"step": 10160
},
{
"epoch": 2.0118706103472155,
"grad_norm": 0.021351774103845745,
"learning_rate": 2.9576286372289542e-05,
"loss": 0.0047,
"step": 10170
},
{
"epoch": 2.0138490454050846,
"grad_norm": 0.015173884113918224,
"learning_rate": 2.947128668287161e-05,
"loss": 0.0047,
"step": 10180
},
{
"epoch": 2.0158274804629537,
"grad_norm": 0.017705574481519378,
"learning_rate": 2.936639576699194e-05,
"loss": 0.0047,
"step": 10190
},
{
"epoch": 2.0178059155208232,
"grad_norm": 0.015106130424750308,
"learning_rate": 2.9261614180426034e-05,
"loss": 0.0047,
"step": 10200
},
{
"epoch": 2.0197843505786923,
"grad_norm": 0.010738979130028022,
"learning_rate": 2.915694247837014e-05,
"loss": 0.0047,
"step": 10210
},
{
"epoch": 2.0217627856365614,
"grad_norm": 0.01109188646303964,
"learning_rate": 2.90523812154382e-05,
"loss": 0.0047,
"step": 10220
},
{
"epoch": 2.0237412206944305,
"grad_norm": 0.017806430975803623,
"learning_rate": 2.894793094565904e-05,
"loss": 0.0047,
"step": 10230
},
{
"epoch": 2.0257196557523,
"grad_norm": 0.005973699848521098,
"learning_rate": 2.8843592222473354e-05,
"loss": 0.0047,
"step": 10240
},
{
"epoch": 2.027698090810169,
"grad_norm": 0.015121347790510714,
"learning_rate": 2.8739365598730818e-05,
"loss": 0.0047,
"step": 10250
},
{
"epoch": 2.0296765258680383,
"grad_norm": 0.00834478178610599,
"learning_rate": 2.8635251626687076e-05,
"loss": 0.0047,
"step": 10260
},
{
"epoch": 2.031654960925908,
"grad_norm": 0.01738693186882847,
"learning_rate": 2.853125085800096e-05,
"loss": 0.0047,
"step": 10270
},
{
"epoch": 2.033633395983777,
"grad_norm": 0.01438471804585251,
"learning_rate": 2.8427363843731393e-05,
"loss": 0.0047,
"step": 10280
},
{
"epoch": 2.035611831041646,
"grad_norm": 0.009647924539496942,
"learning_rate": 2.832359113433463e-05,
"loss": 0.0046,
"step": 10290
},
{
"epoch": 2.037590266099515,
"grad_norm": 0.02554478092589615,
"learning_rate": 2.8219933279661247e-05,
"loss": 0.0047,
"step": 10300
},
{
"epoch": 2.0395687011573846,
"grad_norm": 0.015837355489976022,
"learning_rate": 2.8116390828953253e-05,
"loss": 0.0047,
"step": 10310
},
{
"epoch": 2.0415471362152537,
"grad_norm": 0.01001247257165805,
"learning_rate": 2.801296433084115e-05,
"loss": 0.0047,
"step": 10320
},
{
"epoch": 2.043525571273123,
"grad_norm": 0.01010035603588263,
"learning_rate": 2.7909654333341084e-05,
"loss": 0.0047,
"step": 10330
},
{
"epoch": 2.0455040063309924,
"grad_norm": 0.008408202103864596,
"learning_rate": 2.7806461383851895e-05,
"loss": 0.0046,
"step": 10340
},
{
"epoch": 2.0474824413888615,
"grad_norm": 0.014348954341192956,
"learning_rate": 2.7703386029152246e-05,
"loss": 0.0046,
"step": 10350
},
{
"epoch": 2.0494608764467306,
"grad_norm": 0.010672904187580984,
"learning_rate": 2.76004288153977e-05,
"loss": 0.0046,
"step": 10360
},
{
"epoch": 2.0514393115045997,
"grad_norm": 0.0148026562622368,
"learning_rate": 2.749759028811786e-05,
"loss": 0.0047,
"step": 10370
},
{
"epoch": 2.0534177465624692,
"grad_norm": 0.010898271195079048,
"learning_rate": 2.7394870992213394e-05,
"loss": 0.0047,
"step": 10380
},
{
"epoch": 2.0553961816203383,
"grad_norm": 0.012277255260417229,
"learning_rate": 2.7292271471953284e-05,
"loss": 0.0047,
"step": 10390
},
{
"epoch": 2.0573746166782074,
"grad_norm": 0.007469823067613679,
"learning_rate": 2.7189792270971827e-05,
"loss": 0.0047,
"step": 10400
},
{
"epoch": 2.059353051736077,
"grad_norm": 0.01184380141139781,
"learning_rate": 2.7087433932265825e-05,
"loss": 0.0047,
"step": 10410
},
{
"epoch": 2.061331486793946,
"grad_norm": 0.0077864232583511285,
"learning_rate": 2.698519699819163e-05,
"loss": 0.0047,
"step": 10420
},
{
"epoch": 2.063309921851815,
"grad_norm": 0.004479031550808949,
"learning_rate": 2.688308201046236e-05,
"loss": 0.0046,
"step": 10430
},
{
"epoch": 2.0652883569096843,
"grad_norm": 0.006111590212230939,
"learning_rate": 2.6781089510144996e-05,
"loss": 0.0047,
"step": 10440
},
{
"epoch": 2.067266791967554,
"grad_norm": 0.01661410494137613,
"learning_rate": 2.667922003765745e-05,
"loss": 0.0046,
"step": 10450
},
{
"epoch": 2.069245227025423,
"grad_norm": 0.013910762751716382,
"learning_rate": 2.657747413276581e-05,
"loss": 0.0047,
"step": 10460
},
{
"epoch": 2.071223662083292,
"grad_norm": 0.011114231723358839,
"learning_rate": 2.647585233458142e-05,
"loss": 0.0047,
"step": 10470
},
{
"epoch": 2.0732020971411615,
"grad_norm": 0.01802270818902327,
"learning_rate": 2.637435518155802e-05,
"loss": 0.0047,
"step": 10480
},
{
"epoch": 2.0751805321990306,
"grad_norm": 0.005906448584090342,
"learning_rate": 2.627298321148893e-05,
"loss": 0.0047,
"step": 10490
},
{
"epoch": 2.0771589672568997,
"grad_norm": 0.008470163532462832,
"learning_rate": 2.617173696150411e-05,
"loss": 0.0046,
"step": 10500
},
{
"epoch": 2.079137402314769,
"grad_norm": 0.012929295253028657,
"learning_rate": 2.6070616968067445e-05,
"loss": 0.0046,
"step": 10510
},
{
"epoch": 2.0811158373726384,
"grad_norm": 0.009714522416840426,
"learning_rate": 2.5969623766973815e-05,
"loss": 0.0046,
"step": 10520
},
{
"epoch": 2.0830942724305075,
"grad_norm": 0.009301438133805403,
"learning_rate": 2.5868757893346284e-05,
"loss": 0.0046,
"step": 10530
},
{
"epoch": 2.0850727074883766,
"grad_norm": 0.01128390001414039,
"learning_rate": 2.5768019881633237e-05,
"loss": 0.0046,
"step": 10540
},
{
"epoch": 2.087051142546246,
"grad_norm": 0.010287632493257289,
"learning_rate": 2.5667410265605617e-05,
"loss": 0.0046,
"step": 10550
},
{
"epoch": 2.0890295776041152,
"grad_norm": 0.009402280847741517,
"learning_rate": 2.556692957835398e-05,
"loss": 0.0046,
"step": 10560
},
{
"epoch": 2.0910080126619843,
"grad_norm": 0.015441746174070056,
"learning_rate": 2.5466578352285774e-05,
"loss": 0.0046,
"step": 10570
},
{
"epoch": 2.0929864477198534,
"grad_norm": 0.028184524395839527,
"learning_rate": 2.5366357119122486e-05,
"loss": 0.0047,
"step": 10580
},
{
"epoch": 2.094964882777723,
"grad_norm": 0.02073783762391385,
"learning_rate": 2.5266266409896834e-05,
"loss": 0.0047,
"step": 10590
},
{
"epoch": 2.096943317835592,
"grad_norm": 0.02916093634683859,
"learning_rate": 2.5166306754949908e-05,
"loss": 0.0047,
"step": 10600
},
{
"epoch": 2.098921752893461,
"grad_norm": 0.0054648201367024085,
"learning_rate": 2.5066478683928457e-05,
"loss": 0.0047,
"step": 10610
},
{
"epoch": 2.1009001879513307,
"grad_norm": 0.016652479531144366,
"learning_rate": 2.4966782725781916e-05,
"loss": 0.0047,
"step": 10620
},
{
"epoch": 2.1028786230092,
"grad_norm": 0.01587776707835437,
"learning_rate": 2.4867219408759794e-05,
"loss": 0.0047,
"step": 10630
},
{
"epoch": 2.104857058067069,
"grad_norm": 0.008087790977336787,
"learning_rate": 2.4767789260408753e-05,
"loss": 0.0047,
"step": 10640
},
{
"epoch": 2.106835493124938,
"grad_norm": 0.01775556044778209,
"learning_rate": 2.466849280756985e-05,
"loss": 0.0047,
"step": 10650
},
{
"epoch": 2.1088139281828076,
"grad_norm": 0.011531282202047057,
"learning_rate": 2.4569330576375743e-05,
"loss": 0.0047,
"step": 10660
},
{
"epoch": 2.1107923632406767,
"grad_norm": 0.024343108263367108,
"learning_rate": 2.4470303092247926e-05,
"loss": 0.0047,
"step": 10670
},
{
"epoch": 2.1127707982985457,
"grad_norm": 0.01134330463013958,
"learning_rate": 2.4371410879893857e-05,
"loss": 0.0047,
"step": 10680
},
{
"epoch": 2.114749233356415,
"grad_norm": 0.010622519012440863,
"learning_rate": 2.4272654463304328e-05,
"loss": 0.0047,
"step": 10690
},
{
"epoch": 2.1167276684142844,
"grad_norm": 0.009123330378693074,
"learning_rate": 2.4174034365750513e-05,
"loss": 0.0047,
"step": 10700
},
{
"epoch": 2.1187061034721535,
"grad_norm": 0.010037715720386623,
"learning_rate": 2.4075551109781358e-05,
"loss": 0.0047,
"step": 10710
},
{
"epoch": 2.1206845385300226,
"grad_norm": 0.014617360861586,
"learning_rate": 2.3977205217220707e-05,
"loss": 0.0046,
"step": 10720
},
{
"epoch": 2.122662973587892,
"grad_norm": 0.009556467982072088,
"learning_rate": 2.387899720916459e-05,
"loss": 0.0047,
"step": 10730
},
{
"epoch": 2.1246414086457612,
"grad_norm": 0.012145946598867387,
"learning_rate": 2.3780927605978397e-05,
"loss": 0.0047,
"step": 10740
},
{
"epoch": 2.1266198437036303,
"grad_norm": 0.00781970123249589,
"learning_rate": 2.3682996927294214e-05,
"loss": 0.0046,
"step": 10750
},
{
"epoch": 2.1285982787614994,
"grad_norm": 0.011447440088680892,
"learning_rate": 2.3585205692008e-05,
"loss": 0.0047,
"step": 10760
},
{
"epoch": 2.130576713819369,
"grad_norm": 0.018074848156235425,
"learning_rate": 2.348755441827686e-05,
"loss": 0.0047,
"step": 10770
},
{
"epoch": 2.132555148877238,
"grad_norm": 0.011619022989755234,
"learning_rate": 2.3390043623516293e-05,
"loss": 0.0047,
"step": 10780
},
{
"epoch": 2.134533583935107,
"grad_norm": 0.015546160131111557,
"learning_rate": 2.3292673824397487e-05,
"loss": 0.0047,
"step": 10790
},
{
"epoch": 2.1365120189929767,
"grad_norm": 0.010791279098775282,
"learning_rate": 2.319544553684449e-05,
"loss": 0.0047,
"step": 10800
},
{
"epoch": 2.138490454050846,
"grad_norm": 0.018762180209708577,
"learning_rate": 2.3098359276031578e-05,
"loss": 0.0047,
"step": 10810
},
{
"epoch": 2.140468889108715,
"grad_norm": 0.009995295158968416,
"learning_rate": 2.3001415556380484e-05,
"loss": 0.0047,
"step": 10820
},
{
"epoch": 2.142447324166584,
"grad_norm": 0.003447407000126886,
"learning_rate": 2.290461489155768e-05,
"loss": 0.0047,
"step": 10830
},
{
"epoch": 2.1444257592244536,
"grad_norm": 0.01483660543511331,
"learning_rate": 2.280795779447159e-05,
"loss": 0.0046,
"step": 10840
},
{
"epoch": 2.1464041942823227,
"grad_norm": 0.00939370979608112,
"learning_rate": 2.271144477726999e-05,
"loss": 0.0047,
"step": 10850
},
{
"epoch": 2.1483826293401918,
"grad_norm": 0.009093690594601004,
"learning_rate": 2.261507635133724e-05,
"loss": 0.0047,
"step": 10860
},
{
"epoch": 2.1503610643980613,
"grad_norm": 0.008690068267563337,
"learning_rate": 2.2518853027291487e-05,
"loss": 0.0047,
"step": 10870
},
{
"epoch": 2.1523394994559304,
"grad_norm": 0.009831730088965715,
"learning_rate": 2.242277531498212e-05,
"loss": 0.0046,
"step": 10880
},
{
"epoch": 2.1543179345137995,
"grad_norm": 0.006917591633862752,
"learning_rate": 2.2326843723486973e-05,
"loss": 0.0047,
"step": 10890
},
{
"epoch": 2.1562963695716686,
"grad_norm": 0.01064728957715656,
"learning_rate": 2.223105876110962e-05,
"loss": 0.0046,
"step": 10900
},
{
"epoch": 2.158274804629538,
"grad_norm": 0.007098212917479849,
"learning_rate": 2.213542093537675e-05,
"loss": 0.0047,
"step": 10910
},
{
"epoch": 2.1602532396874072,
"grad_norm": 0.0055247141488649985,
"learning_rate": 2.2039930753035354e-05,
"loss": 0.0046,
"step": 10920
},
{
"epoch": 2.1622316747452763,
"grad_norm": 0.011592474145949556,
"learning_rate": 2.1944588720050197e-05,
"loss": 0.0047,
"step": 10930
},
{
"epoch": 2.164210109803146,
"grad_norm": 0.010012347903896369,
"learning_rate": 2.1849395341601026e-05,
"loss": 0.0046,
"step": 10940
},
{
"epoch": 2.166188544861015,
"grad_norm": 0.004596310017660314,
"learning_rate": 2.1754351122079923e-05,
"loss": 0.0046,
"step": 10950
},
{
"epoch": 2.168166979918884,
"grad_norm": 0.003557529455894823,
"learning_rate": 2.1659456565088654e-05,
"loss": 0.0047,
"step": 10960
},
{
"epoch": 2.170145414976753,
"grad_norm": 0.01269444300931225,
"learning_rate": 2.156471217343598e-05,
"loss": 0.0047,
"step": 10970
},
{
"epoch": 2.1721238500346227,
"grad_norm": 0.0050745579977574075,
"learning_rate": 2.1470118449134974e-05,
"loss": 0.0046,
"step": 10980
},
{
"epoch": 2.174102285092492,
"grad_norm": 0.016641791964316,
"learning_rate": 2.1375675893400372e-05,
"loss": 0.0046,
"step": 10990
},
{
"epoch": 2.176080720150361,
"grad_norm": 0.0154486493080561,
"learning_rate": 2.128138500664597e-05,
"loss": 0.0047,
"step": 11000
},
{
"epoch": 2.1780591552082305,
"grad_norm": 0.004789583207744623,
"learning_rate": 2.1187246288481917e-05,
"loss": 0.0046,
"step": 11010
},
{
"epoch": 2.1800375902660996,
"grad_norm": 0.020081582753526478,
"learning_rate": 2.109326023771206e-05,
"loss": 0.0047,
"step": 11020
},
{
"epoch": 2.1820160253239687,
"grad_norm": 0.02172580760270673,
"learning_rate": 2.0999427352331362e-05,
"loss": 0.0046,
"step": 11030
},
{
"epoch": 2.1839944603818378,
"grad_norm": 0.0048152339835073315,
"learning_rate": 2.0905748129523166e-05,
"loss": 0.0047,
"step": 11040
},
{
"epoch": 2.1859728954397073,
"grad_norm": 0.006783945927678857,
"learning_rate": 2.081222306565666e-05,
"loss": 0.0047,
"step": 11050
},
{
"epoch": 2.1879513304975764,
"grad_norm": 0.004275369860240765,
"learning_rate": 2.0718852656284195e-05,
"loss": 0.0046,
"step": 11060
},
{
"epoch": 2.1899297655554455,
"grad_norm": 0.041407881499763835,
"learning_rate": 2.0625637396138664e-05,
"loss": 0.0046,
"step": 11070
},
{
"epoch": 2.191908200613315,
"grad_norm": 0.006777557889043144,
"learning_rate": 2.053257777913089e-05,
"loss": 0.0046,
"step": 11080
},
{
"epoch": 2.193886635671184,
"grad_norm": 0.011671020435615794,
"learning_rate": 2.043967429834703e-05,
"loss": 0.0047,
"step": 11090
},
{
"epoch": 2.1958650707290532,
"grad_norm": 0.007182724806823058,
"learning_rate": 2.034692744604586e-05,
"loss": 0.0047,
"step": 11100
},
{
"epoch": 2.1978435057869223,
"grad_norm": 0.00796205340491484,
"learning_rate": 2.025433771365632e-05,
"loss": 0.0047,
"step": 11110
},
{
"epoch": 2.199821940844792,
"grad_norm": 0.009320322615979934,
"learning_rate": 2.0161905591774826e-05,
"loss": 0.0047,
"step": 11120
},
{
"epoch": 2.201800375902661,
"grad_norm": 0.003151786585119805,
"learning_rate": 2.0069631570162624e-05,
"loss": 0.0047,
"step": 11130
},
{
"epoch": 2.20377881096053,
"grad_norm": 0.004409080235535024,
"learning_rate": 1.9977516137743302e-05,
"loss": 0.0047,
"step": 11140
},
{
"epoch": 2.2057572460183996,
"grad_norm": 0.004165929193044647,
"learning_rate": 1.9885559782600126e-05,
"loss": 0.0046,
"step": 11150
},
{
"epoch": 2.2077356810762687,
"grad_norm": 0.011966781650276712,
"learning_rate": 1.97937629919735e-05,
"loss": 0.0047,
"step": 11160
},
{
"epoch": 2.209714116134138,
"grad_norm": 0.0037043329676071336,
"learning_rate": 1.97021262522583e-05,
"loss": 0.0046,
"step": 11170
},
{
"epoch": 2.211692551192007,
"grad_norm": 0.010174871194566485,
"learning_rate": 1.9610650049001407e-05,
"loss": 0.0047,
"step": 11180
},
{
"epoch": 2.2136709862498765,
"grad_norm": 0.008710086999336323,
"learning_rate": 1.951933486689907e-05,
"loss": 0.0046,
"step": 11190
},
{
"epoch": 2.2156494213077456,
"grad_norm": 0.014194782978234026,
"learning_rate": 1.9428181189794353e-05,
"loss": 0.0046,
"step": 11200
},
{
"epoch": 2.2176278563656147,
"grad_norm": 0.011387381936850627,
"learning_rate": 1.9337189500674573e-05,
"loss": 0.0047,
"step": 11210
},
{
"epoch": 2.219606291423484,
"grad_norm": 0.006379389322977933,
"learning_rate": 1.9246360281668697e-05,
"loss": 0.0047,
"step": 11220
},
{
"epoch": 2.2215847264813533,
"grad_norm": 0.003721761619797733,
"learning_rate": 1.9155694014044877e-05,
"loss": 0.0047,
"step": 11230
},
{
"epoch": 2.2235631615392224,
"grad_norm": 0.008853495991480328,
"learning_rate": 1.9065191178207825e-05,
"loss": 0.0046,
"step": 11240
},
{
"epoch": 2.2255415965970915,
"grad_norm": 0.009349768237205016,
"learning_rate": 1.8974852253696307e-05,
"loss": 0.0046,
"step": 11250
},
{
"epoch": 2.227520031654961,
"grad_norm": 0.010397581588081985,
"learning_rate": 1.8884677719180594e-05,
"loss": 0.0047,
"step": 11260
},
{
"epoch": 2.22949846671283,
"grad_norm": 0.014052249551951048,
"learning_rate": 1.879466805245986e-05,
"loss": 0.0046,
"step": 11270
},
{
"epoch": 2.2314769017706992,
"grad_norm": 0.0048661913342422725,
"learning_rate": 1.87048237304598e-05,
"loss": 0.0047,
"step": 11280
},
{
"epoch": 2.233455336828569,
"grad_norm": 0.01499066906501541,
"learning_rate": 1.8615145229229935e-05,
"loss": 0.0046,
"step": 11290
},
{
"epoch": 2.235433771886438,
"grad_norm": 0.01200186678792473,
"learning_rate": 1.8525633023941208e-05,
"loss": 0.0047,
"step": 11300
},
{
"epoch": 2.237412206944307,
"grad_norm": 0.012987348420137719,
"learning_rate": 1.8436287588883417e-05,
"loss": 0.0046,
"step": 11310
},
{
"epoch": 2.239390642002176,
"grad_norm": 0.011466142782628959,
"learning_rate": 1.8347109397462713e-05,
"loss": 0.0047,
"step": 11320
},
{
"epoch": 2.2413690770600456,
"grad_norm": 0.016293413696584172,
"learning_rate": 1.8258098922199107e-05,
"loss": 0.0047,
"step": 11330
},
{
"epoch": 2.2433475121179147,
"grad_norm": 0.007467726192115216,
"learning_rate": 1.8169256634723887e-05,
"loss": 0.0046,
"step": 11340
},
{
"epoch": 2.245325947175784,
"grad_norm": 0.010963496045312278,
"learning_rate": 1.808058300577724e-05,
"loss": 0.0046,
"step": 11350
},
{
"epoch": 2.2473043822336534,
"grad_norm": 0.00917753987060109,
"learning_rate": 1.7992078505205683e-05,
"loss": 0.0046,
"step": 11360
},
{
"epoch": 2.2492828172915225,
"grad_norm": 0.007862677030743285,
"learning_rate": 1.7903743601959577e-05,
"loss": 0.0047,
"step": 11370
},
{
"epoch": 2.2512612523493916,
"grad_norm": 0.007747796654084662,
"learning_rate": 1.7815578764090674e-05,
"loss": 0.0047,
"step": 11380
},
{
"epoch": 2.2532396874072607,
"grad_norm": 0.013272367008158384,
"learning_rate": 1.7727584458749608e-05,
"loss": 0.0046,
"step": 11390
},
{
"epoch": 2.25521812246513,
"grad_norm": 0.01083285952756501,
"learning_rate": 1.7639761152183415e-05,
"loss": 0.0046,
"step": 11400
},
{
"epoch": 2.2571965575229993,
"grad_norm": 0.01136955458229527,
"learning_rate": 1.7552109309733066e-05,
"loss": 0.0046,
"step": 11410
},
{
"epoch": 2.2591749925808684,
"grad_norm": 0.006665322172386251,
"learning_rate": 1.746462939583105e-05,
"loss": 0.0046,
"step": 11420
},
{
"epoch": 2.261153427638738,
"grad_norm": 0.012013862447185601,
"learning_rate": 1.737732187399886e-05,
"loss": 0.0046,
"step": 11430
},
{
"epoch": 2.263131862696607,
"grad_norm": 0.003825761152740482,
"learning_rate": 1.7290187206844538e-05,
"loss": 0.0046,
"step": 11440
},
{
"epoch": 2.265110297754476,
"grad_norm": 0.0033022974152408942,
"learning_rate": 1.720322585606027e-05,
"loss": 0.0047,
"step": 11450
},
{
"epoch": 2.2670887328123452,
"grad_norm": 0.004757366327253904,
"learning_rate": 1.711643828241985e-05,
"loss": 0.0046,
"step": 11460
},
{
"epoch": 2.269067167870215,
"grad_norm": 0.009707249028171383,
"learning_rate": 1.7029824945776345e-05,
"loss": 0.0046,
"step": 11470
},
{
"epoch": 2.271045602928084,
"grad_norm": 0.010769447473014023,
"learning_rate": 1.6943386305059593e-05,
"loss": 0.0047,
"step": 11480
},
{
"epoch": 2.273024037985953,
"grad_norm": 0.018443000639790572,
"learning_rate": 1.6857122818273792e-05,
"loss": 0.0047,
"step": 11490
},
{
"epoch": 2.2750024730438225,
"grad_norm": 0.009485489411969795,
"learning_rate": 1.6771034942495064e-05,
"loss": 0.0047,
"step": 11500
},
{
"epoch": 2.2769809081016916,
"grad_norm": 0.008794964077977611,
"learning_rate": 1.6685123133869044e-05,
"loss": 0.0047,
"step": 11510
},
{
"epoch": 2.2789593431595607,
"grad_norm": 0.005800324329560414,
"learning_rate": 1.659938784760843e-05,
"loss": 0.0046,
"step": 11520
},
{
"epoch": 2.28093777821743,
"grad_norm": 0.0018831175677115214,
"learning_rate": 1.651382953799063e-05,
"loss": 0.0047,
"step": 11530
},
{
"epoch": 2.2829162132752994,
"grad_norm": 0.010287259082322803,
"learning_rate": 1.642844865835531e-05,
"loss": 0.0046,
"step": 11540
},
{
"epoch": 2.2848946483331685,
"grad_norm": 0.005658588686005756,
"learning_rate": 1.634324566110203e-05,
"loss": 0.0046,
"step": 11550
},
{
"epoch": 2.2868730833910376,
"grad_norm": 0.005822856525905536,
"learning_rate": 1.6258220997687756e-05,
"loss": 0.0047,
"step": 11560
},
{
"epoch": 2.288851518448907,
"grad_norm": 0.008590712477345202,
"learning_rate": 1.6173375118624594e-05,
"loss": 0.0047,
"step": 11570
},
{
"epoch": 2.290829953506776,
"grad_norm": 0.004876996416456864,
"learning_rate": 1.6088708473477342e-05,
"loss": 0.0047,
"step": 11580
},
{
"epoch": 2.2928083885646453,
"grad_norm": 0.00842810259789916,
"learning_rate": 1.6004221510861057e-05,
"loss": 0.0046,
"step": 11590
},
{
"epoch": 2.2947868236225144,
"grad_norm": 0.0071233585204164476,
"learning_rate": 1.5919914678438776e-05,
"loss": 0.0046,
"step": 11600
},
{
"epoch": 2.296765258680384,
"grad_norm": 0.0046822498861528224,
"learning_rate": 1.583578842291908e-05,
"loss": 0.0046,
"step": 11610
},
{
"epoch": 2.298743693738253,
"grad_norm": 0.003287457544874044,
"learning_rate": 1.5751843190053754e-05,
"loss": 0.0046,
"step": 11620
},
{
"epoch": 2.300722128796122,
"grad_norm": 0.008738807750511,
"learning_rate": 1.5668079424635424e-05,
"loss": 0.0046,
"step": 11630
},
{
"epoch": 2.3027005638539917,
"grad_norm": 0.00991795770675324,
"learning_rate": 1.5584497570495143e-05,
"loss": 0.0046,
"step": 11640
},
{
"epoch": 2.304678998911861,
"grad_norm": 0.006274378398565224,
"learning_rate": 1.5501098070500135e-05,
"loss": 0.0046,
"step": 11650
},
{
"epoch": 2.30665743396973,
"grad_norm": 0.009355205636403045,
"learning_rate": 1.5417881366551395e-05,
"loss": 0.0047,
"step": 11660
},
{
"epoch": 2.308635869027599,
"grad_norm": 0.012756388051471,
"learning_rate": 1.5334847899581346e-05,
"loss": 0.0047,
"step": 11670
},
{
"epoch": 2.3106143040854685,
"grad_norm": 0.010696814881384912,
"learning_rate": 1.5251998109551508e-05,
"loss": 0.0046,
"step": 11680
},
{
"epoch": 2.3125927391433376,
"grad_norm": 0.00887318432765735,
"learning_rate": 1.516933243545019e-05,
"loss": 0.0046,
"step": 11690
},
{
"epoch": 2.3145711742012067,
"grad_norm": 0.008904011953672748,
"learning_rate": 1.5086851315290117e-05,
"loss": 0.0047,
"step": 11700
},
{
"epoch": 2.3165496092590763,
"grad_norm": 0.010194824412785934,
"learning_rate": 1.5004555186106122e-05,
"loss": 0.0046,
"step": 11710
},
{
"epoch": 2.3185280443169454,
"grad_norm": 0.009306345299286168,
"learning_rate": 1.4922444483952886e-05,
"loss": 0.0046,
"step": 11720
},
{
"epoch": 2.3205064793748145,
"grad_norm": 0.008858233044818227,
"learning_rate": 1.4840519643902573e-05,
"loss": 0.0046,
"step": 11730
},
{
"epoch": 2.3224849144326836,
"grad_norm": 0.007499926336122923,
"learning_rate": 1.4758781100042517e-05,
"loss": 0.0047,
"step": 11740
},
{
"epoch": 2.324463349490553,
"grad_norm": 0.003439327321785818,
"learning_rate": 1.4677229285472988e-05,
"loss": 0.0047,
"step": 11750
},
{
"epoch": 2.326441784548422,
"grad_norm": 0.008077477973948968,
"learning_rate": 1.459586463230478e-05,
"loss": 0.0047,
"step": 11760
},
{
"epoch": 2.3284202196062913,
"grad_norm": 0.008390366452460865,
"learning_rate": 1.4514687571657049e-05,
"loss": 0.0046,
"step": 11770
},
{
"epoch": 2.330398654664161,
"grad_norm": 0.012358088456361188,
"learning_rate": 1.4433698533654955e-05,
"loss": 0.0047,
"step": 11780
},
{
"epoch": 2.33237708972203,
"grad_norm": 0.011113571851676381,
"learning_rate": 1.4352897947427396e-05,
"loss": 0.0046,
"step": 11790
},
{
"epoch": 2.334355524779899,
"grad_norm": 0.006301177619714851,
"learning_rate": 1.4272286241104754e-05,
"loss": 0.0046,
"step": 11800
},
{
"epoch": 2.336333959837768,
"grad_norm": 0.007973220688055543,
"learning_rate": 1.4191863841816611e-05,
"loss": 0.0046,
"step": 11810
},
{
"epoch": 2.3383123948956377,
"grad_norm": 0.013374203713257387,
"learning_rate": 1.4111631175689443e-05,
"loss": 0.0047,
"step": 11820
},
{
"epoch": 2.340290829953507,
"grad_norm": 0.008823285665961677,
"learning_rate": 1.4031588667844476e-05,
"loss": 0.0047,
"step": 11830
},
{
"epoch": 2.342269265011376,
"grad_norm": 0.007793166136000498,
"learning_rate": 1.3951736742395277e-05,
"loss": 0.0046,
"step": 11840
},
{
"epoch": 2.3442477000692454,
"grad_norm": 0.006305742768568126,
"learning_rate": 1.3872075822445668e-05,
"loss": 0.0046,
"step": 11850
},
{
"epoch": 2.3462261351271145,
"grad_norm": 0.014624922013159392,
"learning_rate": 1.3792606330087376e-05,
"loss": 0.0047,
"step": 11860
},
{
"epoch": 2.3482045701849836,
"grad_norm": 0.008948432366295627,
"learning_rate": 1.371332868639783e-05,
"loss": 0.0046,
"step": 11870
},
{
"epoch": 2.3501830052428527,
"grad_norm": 0.010464569469698813,
"learning_rate": 1.3634243311437945e-05,
"loss": 0.0047,
"step": 11880
},
{
"epoch": 2.3521614403007223,
"grad_norm": 0.001176931665766798,
"learning_rate": 1.355535062424984e-05,
"loss": 0.0047,
"step": 11890
},
{
"epoch": 2.3541398753585914,
"grad_norm": 0.0018597977610570586,
"learning_rate": 1.3476651042854699e-05,
"loss": 0.0046,
"step": 11900
},
{
"epoch": 2.3561183104164605,
"grad_norm": 0.0037584440939611,
"learning_rate": 1.3398144984250494e-05,
"loss": 0.0046,
"step": 11910
},
{
"epoch": 2.35809674547433,
"grad_norm": 0.007028202994452381,
"learning_rate": 1.3319832864409804e-05,
"loss": 0.0047,
"step": 11920
},
{
"epoch": 2.360075180532199,
"grad_norm": 0.004301144570970643,
"learning_rate": 1.3241715098277613e-05,
"loss": 0.0046,
"step": 11930
},
{
"epoch": 2.362053615590068,
"grad_norm": 0.005720534246820603,
"learning_rate": 1.3163792099769067e-05,
"loss": 0.0047,
"step": 11940
},
{
"epoch": 2.3640320506479373,
"grad_norm": 0.0036595375655008476,
"learning_rate": 1.3086064281767346e-05,
"loss": 0.0046,
"step": 11950
},
{
"epoch": 2.366010485705807,
"grad_norm": 0.006605882668477732,
"learning_rate": 1.3008532056121437e-05,
"loss": 0.0047,
"step": 11960
},
{
"epoch": 2.367988920763676,
"grad_norm": 0.006439834054296722,
"learning_rate": 1.2931195833643978e-05,
"loss": 0.0047,
"step": 11970
},
{
"epoch": 2.369967355821545,
"grad_norm": 0.01164829339688394,
"learning_rate": 1.2854056024109013e-05,
"loss": 0.0046,
"step": 11980
},
{
"epoch": 2.3719457908794146,
"grad_norm": 0.007061335189498727,
"learning_rate": 1.2777113036249928e-05,
"loss": 0.0046,
"step": 11990
},
{
"epoch": 2.3739242259372837,
"grad_norm": 0.010406290978172413,
"learning_rate": 1.2700367277757209e-05,
"loss": 0.0046,
"step": 12000
},
{
"epoch": 2.375902660995153,
"grad_norm": 0.011454031986549013,
"learning_rate": 1.2623819155276267e-05,
"loss": 0.0046,
"step": 12010
},
{
"epoch": 2.377881096053022,
"grad_norm": 0.004989127330545308,
"learning_rate": 1.2547469074405376e-05,
"loss": 0.0046,
"step": 12020
},
{
"epoch": 2.3798595311108914,
"grad_norm": 0.011825126089196485,
"learning_rate": 1.2471317439693436e-05,
"loss": 0.0046,
"step": 12030
},
{
"epoch": 2.3818379661687605,
"grad_norm": 0.012403611866636173,
"learning_rate": 1.2395364654637853e-05,
"loss": 0.0046,
"step": 12040
},
{
"epoch": 2.3838164012266296,
"grad_norm": 0.006538135528063423,
"learning_rate": 1.2319611121682439e-05,
"loss": 0.0046,
"step": 12050
},
{
"epoch": 2.385794836284499,
"grad_norm": 0.007189966090924128,
"learning_rate": 1.224405724221519e-05,
"loss": 0.0046,
"step": 12060
},
{
"epoch": 2.3877732713423683,
"grad_norm": 0.006479844069213243,
"learning_rate": 1.2168703416566274e-05,
"loss": 0.0046,
"step": 12070
},
{
"epoch": 2.3897517064002374,
"grad_norm": 0.005722721260927207,
"learning_rate": 1.2093550044005818e-05,
"loss": 0.0046,
"step": 12080
},
{
"epoch": 2.3917301414581065,
"grad_norm": 0.009360093295392754,
"learning_rate": 1.2018597522741854e-05,
"loss": 0.0046,
"step": 12090
},
{
"epoch": 2.393708576515976,
"grad_norm": 0.00532075773237555,
"learning_rate": 1.1943846249918162e-05,
"loss": 0.0046,
"step": 12100
},
{
"epoch": 2.395687011573845,
"grad_norm": 0.008994493597559465,
"learning_rate": 1.1869296621612209e-05,
"loss": 0.0046,
"step": 12110
},
{
"epoch": 2.397665446631714,
"grad_norm": 0.004332218300834066,
"learning_rate": 1.1794949032832991e-05,
"loss": 0.0046,
"step": 12120
},
{
"epoch": 2.3996438816895838,
"grad_norm": 0.010964609332513163,
"learning_rate": 1.1720803877518992e-05,
"loss": 0.0047,
"step": 12130
},
{
"epoch": 2.401622316747453,
"grad_norm": 0.0020947477783036334,
"learning_rate": 1.1646861548536092e-05,
"loss": 0.0046,
"step": 12140
},
{
"epoch": 2.403600751805322,
"grad_norm": 0.006348605673775687,
"learning_rate": 1.1573122437675466e-05,
"loss": 0.0046,
"step": 12150
},
{
"epoch": 2.405579186863191,
"grad_norm": 0.011627207685052689,
"learning_rate": 1.1499586935651529e-05,
"loss": 0.0046,
"step": 12160
},
{
"epoch": 2.4075576219210606,
"grad_norm": 0.00850296909354686,
"learning_rate": 1.1426255432099848e-05,
"loss": 0.0047,
"step": 12170
},
{
"epoch": 2.4095360569789297,
"grad_norm": 0.017998982131959208,
"learning_rate": 1.1353128315575052e-05,
"loss": 0.0046,
"step": 12180
},
{
"epoch": 2.411514492036799,
"grad_norm": 0.007560345548260199,
"learning_rate": 1.128020597354884e-05,
"loss": 0.0046,
"step": 12190
},
{
"epoch": 2.4134929270946683,
"grad_norm": 0.010247610791769808,
"learning_rate": 1.1207488792407889e-05,
"loss": 0.0046,
"step": 12200
},
{
"epoch": 2.4154713621525374,
"grad_norm": 0.006900847800161894,
"learning_rate": 1.1134977157451793e-05,
"loss": 0.0046,
"step": 12210
},
{
"epoch": 2.4174497972104065,
"grad_norm": 0.0043800703649640805,
"learning_rate": 1.1062671452891055e-05,
"loss": 0.0046,
"step": 12220
},
{
"epoch": 2.4194282322682756,
"grad_norm": 0.007739840397167808,
"learning_rate": 1.0990572061845034e-05,
"loss": 0.0047,
"step": 12230
},
{
"epoch": 2.421406667326145,
"grad_norm": 0.011511601545750502,
"learning_rate": 1.091867936633988e-05,
"loss": 0.0046,
"step": 12240
},
{
"epoch": 2.4233851023840143,
"grad_norm": 0.006358682922317654,
"learning_rate": 1.0846993747306589e-05,
"loss": 0.0046,
"step": 12250
},
{
"epoch": 2.4253635374418834,
"grad_norm": 0.0038046601233852695,
"learning_rate": 1.0775515584578933e-05,
"loss": 0.0046,
"step": 12260
},
{
"epoch": 2.427341972499753,
"grad_norm": 0.001964349557245093,
"learning_rate": 1.0704245256891421e-05,
"loss": 0.0046,
"step": 12270
},
{
"epoch": 2.429320407557622,
"grad_norm": 0.005797155707508102,
"learning_rate": 1.0633183141877362e-05,
"loss": 0.0046,
"step": 12280
},
{
"epoch": 2.431298842615491,
"grad_norm": 0.028477816384047804,
"learning_rate": 1.0562329616066824e-05,
"loss": 0.0046,
"step": 12290
},
{
"epoch": 2.43327727767336,
"grad_norm": 0.0028578249925270837,
"learning_rate": 1.0491685054884647e-05,
"loss": 0.0046,
"step": 12300
},
{
"epoch": 2.4352557127312298,
"grad_norm": 0.0071730247161914795,
"learning_rate": 1.0421249832648417e-05,
"loss": 0.0047,
"step": 12310
},
{
"epoch": 2.437234147789099,
"grad_norm": 0.00223552588818684,
"learning_rate": 1.0351024322566538e-05,
"loss": 0.0046,
"step": 12320
},
{
"epoch": 2.439212582846968,
"grad_norm": 0.003016329882750013,
"learning_rate": 1.028100889673625e-05,
"loss": 0.0046,
"step": 12330
},
{
"epoch": 2.4411910179048375,
"grad_norm": 0.007336778710179138,
"learning_rate": 1.0211203926141604e-05,
"loss": 0.0046,
"step": 12340
},
{
"epoch": 2.4431694529627066,
"grad_norm": 0.0020722976431069113,
"learning_rate": 1.0141609780651584e-05,
"loss": 0.0047,
"step": 12350
},
{
"epoch": 2.4451478880205757,
"grad_norm": 0.0033094010028402064,
"learning_rate": 1.0072226829018016e-05,
"loss": 0.0046,
"step": 12360
},
{
"epoch": 2.447126323078445,
"grad_norm": 0.00693168764093832,
"learning_rate": 1.0003055438873765e-05,
"loss": 0.0046,
"step": 12370
},
{
"epoch": 2.4491047581363143,
"grad_norm": 0.024056625868265586,
"learning_rate": 9.934095976730678e-06,
"loss": 0.0046,
"step": 12380
},
{
"epoch": 2.4510831931941834,
"grad_norm": 0.007632457949908046,
"learning_rate": 9.865348807977698e-06,
"loss": 0.0046,
"step": 12390
},
{
"epoch": 2.4530616282520525,
"grad_norm": 0.008973982418681662,
"learning_rate": 9.79681429687891e-06,
"loss": 0.0046,
"step": 12400
},
{
"epoch": 2.455040063309922,
"grad_norm": 0.005678267277215895,
"learning_rate": 9.728492806571576e-06,
"loss": 0.0047,
"step": 12410
},
{
"epoch": 2.457018498367791,
"grad_norm": 0.006094494122925294,
"learning_rate": 9.6603846990643e-06,
"loss": 0.0046,
"step": 12420
},
{
"epoch": 2.4589969334256603,
"grad_norm": 0.004594800387657895,
"learning_rate": 9.592490335234994e-06,
"loss": 0.0046,
"step": 12430
},
{
"epoch": 2.4609753684835294,
"grad_norm": 0.009623913566046914,
"learning_rate": 9.524810074829072e-06,
"loss": 0.0046,
"step": 12440
},
{
"epoch": 2.462953803541399,
"grad_norm": 0.005726250909890168,
"learning_rate": 9.457344276457485e-06,
"loss": 0.0046,
"step": 12450
},
{
"epoch": 2.464932238599268,
"grad_norm": 0.0018835439971334395,
"learning_rate": 9.390093297594837e-06,
"loss": 0.0046,
"step": 12460
},
{
"epoch": 2.466910673657137,
"grad_norm": 0.004978908477820304,
"learning_rate": 9.323057494577498e-06,
"loss": 0.0046,
"step": 12470
},
{
"epoch": 2.4688891087150067,
"grad_norm": 0.0019000572362372168,
"learning_rate": 9.25623722260166e-06,
"loss": 0.0046,
"step": 12480
},
{
"epoch": 2.4708675437728758,
"grad_norm": 0.007019374895689543,
"learning_rate": 9.18963283572155e-06,
"loss": 0.0046,
"step": 12490
},
{
"epoch": 2.472845978830745,
"grad_norm": 0.004441425720479358,
"learning_rate": 9.123244686847499e-06,
"loss": 0.0046,
"step": 12500
},
{
"epoch": 2.474824413888614,
"grad_norm": 0.0016514227531753782,
"learning_rate": 9.057073127744064e-06,
"loss": 0.0046,
"step": 12510
},
{
"epoch": 2.4768028489464835,
"grad_norm": 0.012274366959032817,
"learning_rate": 8.991118509028179e-06,
"loss": 0.0046,
"step": 12520
},
{
"epoch": 2.4787812840043526,
"grad_norm": 0.0064800657037888225,
"learning_rate": 8.925381180167319e-06,
"loss": 0.0046,
"step": 12530
},
{
"epoch": 2.4807597190622217,
"grad_norm": 0.0018258658761540242,
"learning_rate": 8.859861489477594e-06,
"loss": 0.0046,
"step": 12540
},
{
"epoch": 2.4827381541200912,
"grad_norm": 0.004008320625324715,
"learning_rate": 8.794559784121937e-06,
"loss": 0.0046,
"step": 12550
},
{
"epoch": 2.4847165891779603,
"grad_norm": 0.0044649927140816455,
"learning_rate": 8.729476410108284e-06,
"loss": 0.0046,
"step": 12560
},
{
"epoch": 2.4866950242358294,
"grad_norm": 0.006569710852024697,
"learning_rate": 8.664611712287719e-06,
"loss": 0.0046,
"step": 12570
},
{
"epoch": 2.4886734592936985,
"grad_norm": 0.005099193229009867,
"learning_rate": 8.599966034352635e-06,
"loss": 0.0046,
"step": 12580
},
{
"epoch": 2.490651894351568,
"grad_norm": 0.009905677228630658,
"learning_rate": 8.535539718834929e-06,
"loss": 0.0046,
"step": 12590
},
{
"epoch": 2.492630329409437,
"grad_norm": 0.009623970168333062,
"learning_rate": 8.471333107104196e-06,
"loss": 0.0047,
"step": 12600
},
{
"epoch": 2.4946087644673063,
"grad_norm": 0.00867519635238705,
"learning_rate": 8.407346539365869e-06,
"loss": 0.0046,
"step": 12610
},
{
"epoch": 2.496587199525176,
"grad_norm": 0.009834560303540758,
"learning_rate": 8.343580354659491e-06,
"loss": 0.0047,
"step": 12620
},
{
"epoch": 2.498565634583045,
"grad_norm": 0.006520270589696586,
"learning_rate": 8.280034890856885e-06,
"loss": 0.0047,
"step": 12630
},
{
"epoch": 2.500544069640914,
"grad_norm": 0.004413024361466315,
"learning_rate": 8.216710484660334e-06,
"loss": 0.0047,
"step": 12640
},
{
"epoch": 2.502522504698783,
"grad_norm": 0.004699601254489097,
"learning_rate": 8.15360747160086e-06,
"loss": 0.0046,
"step": 12650
},
{
"epoch": 2.5045009397566522,
"grad_norm": 0.0021297358088347915,
"learning_rate": 8.090726186036357e-06,
"loss": 0.0046,
"step": 12660
},
{
"epoch": 2.5064793748145218,
"grad_norm": 0.001725797505883062,
"learning_rate": 8.02806696114992e-06,
"loss": 0.0046,
"step": 12670
},
{
"epoch": 2.508457809872391,
"grad_norm": 0.0025007111009153417,
"learning_rate": 7.965630128948038e-06,
"loss": 0.0046,
"step": 12680
},
{
"epoch": 2.5104362449302604,
"grad_norm": 0.004211695061312644,
"learning_rate": 7.90341602025877e-06,
"loss": 0.0046,
"step": 12690
},
{
"epoch": 2.5124146799881295,
"grad_norm": 0.002620254708977969,
"learning_rate": 7.841424964730105e-06,
"loss": 0.0046,
"step": 12700
},
{
"epoch": 2.5143931150459986,
"grad_norm": 0.0074673850614855935,
"learning_rate": 7.779657290828146e-06,
"loss": 0.0046,
"step": 12710
},
{
"epoch": 2.5163715501038677,
"grad_norm": 0.0036764764678667273,
"learning_rate": 7.718113325835396e-06,
"loss": 0.0047,
"step": 12720
},
{
"epoch": 2.518349985161737,
"grad_norm": 0.004989372993712167,
"learning_rate": 7.656793395848983e-06,
"loss": 0.0047,
"step": 12730
},
{
"epoch": 2.5203284202196063,
"grad_norm": 0.0014799549281851857,
"learning_rate": 7.595697825778986e-06,
"loss": 0.0046,
"step": 12740
},
{
"epoch": 2.5223068552774754,
"grad_norm": 0.009395371871205475,
"learning_rate": 7.5348269393466896e-06,
"loss": 0.0046,
"step": 12750
},
{
"epoch": 2.524285290335345,
"grad_norm": 0.0038557645080086107,
"learning_rate": 7.474181059082852e-06,
"loss": 0.0046,
"step": 12760
},
{
"epoch": 2.526263725393214,
"grad_norm": 0.016693831375106376,
"learning_rate": 7.413760506326051e-06,
"loss": 0.0046,
"step": 12770
},
{
"epoch": 2.528242160451083,
"grad_norm": 0.005085258880392611,
"learning_rate": 7.353565601220874e-06,
"loss": 0.0046,
"step": 12780
},
{
"epoch": 2.5302205955089523,
"grad_norm": 0.0036812262170036426,
"learning_rate": 7.293596662716362e-06,
"loss": 0.0046,
"step": 12790
},
{
"epoch": 2.5321990305668214,
"grad_norm": 0.003589363697366034,
"learning_rate": 7.233854008564206e-06,
"loss": 0.0046,
"step": 12800
},
{
"epoch": 2.534177465624691,
"grad_norm": 0.005658782813637028,
"learning_rate": 7.174337955317123e-06,
"loss": 0.0046,
"step": 12810
},
{
"epoch": 2.53615590068256,
"grad_norm": 0.002906395001247396,
"learning_rate": 7.115048818327152e-06,
"loss": 0.0046,
"step": 12820
},
{
"epoch": 2.5381343357404296,
"grad_norm": 0.012177601457515666,
"learning_rate": 7.055986911744017e-06,
"loss": 0.0047,
"step": 12830
},
{
"epoch": 2.5401127707982987,
"grad_norm": 0.004325725086228941,
"learning_rate": 6.997152548513397e-06,
"loss": 0.0046,
"step": 12840
},
{
"epoch": 2.5420912058561678,
"grad_norm": 0.004739538342428766,
"learning_rate": 6.938546040375327e-06,
"loss": 0.0046,
"step": 12850
},
{
"epoch": 2.544069640914037,
"grad_norm": 0.003353231724063483,
"learning_rate": 6.8801676978625386e-06,
"loss": 0.0046,
"step": 12860
},
{
"epoch": 2.546048075971906,
"grad_norm": 0.0048593626802096,
"learning_rate": 6.822017830298788e-06,
"loss": 0.0046,
"step": 12870
},
{
"epoch": 2.5480265110297755,
"grad_norm": 0.004047908092400987,
"learning_rate": 6.764096745797255e-06,
"loss": 0.0046,
"step": 12880
},
{
"epoch": 2.5500049460876446,
"grad_norm": 0.004217057603307179,
"learning_rate": 6.706404751258871e-06,
"loss": 0.0046,
"step": 12890
},
{
"epoch": 2.551983381145514,
"grad_norm": 0.011257414327678017,
"learning_rate": 6.648942152370702e-06,
"loss": 0.0046,
"step": 12900
},
{
"epoch": 2.5539618162033833,
"grad_norm": 0.007478103176638387,
"learning_rate": 6.591709253604356e-06,
"loss": 0.0047,
"step": 12910
},
{
"epoch": 2.5559402512612523,
"grad_norm": 0.00364625396648389,
"learning_rate": 6.534706358214337e-06,
"loss": 0.0046,
"step": 12920
},
{
"epoch": 2.5579186863191214,
"grad_norm": 0.006126419293936818,
"learning_rate": 6.477933768236471e-06,
"loss": 0.0046,
"step": 12930
},
{
"epoch": 2.5598971213769905,
"grad_norm": 0.006356960378968002,
"learning_rate": 6.421391784486264e-06,
"loss": 0.0046,
"step": 12940
},
{
"epoch": 2.56187555643486,
"grad_norm": 0.0011702888526363324,
"learning_rate": 6.365080706557352e-06,
"loss": 0.0046,
"step": 12950
},
{
"epoch": 2.563853991492729,
"grad_norm": 0.0024361448309283126,
"learning_rate": 6.309000832819867e-06,
"loss": 0.0046,
"step": 12960
},
{
"epoch": 2.5658324265505987,
"grad_norm": 0.0040352959295605955,
"learning_rate": 6.25315246041891e-06,
"loss": 0.0046,
"step": 12970
},
{
"epoch": 2.567810861608468,
"grad_norm": 0.009601593938324712,
"learning_rate": 6.197535885272909e-06,
"loss": 0.0046,
"step": 12980
},
{
"epoch": 2.569789296666337,
"grad_norm": 0.005550560624107649,
"learning_rate": 6.142151402072132e-06,
"loss": 0.0047,
"step": 12990
},
{
"epoch": 2.571767731724206,
"grad_norm": 0.0023240202866294615,
"learning_rate": 6.086999304277069e-06,
"loss": 0.0046,
"step": 13000
},
{
"epoch": 2.573746166782075,
"grad_norm": 0.013348757565488888,
"learning_rate": 6.032079884116876e-06,
"loss": 0.0046,
"step": 13010
},
{
"epoch": 2.5757246018399447,
"grad_norm": 0.0033202367734176948,
"learning_rate": 5.977393432587874e-06,
"loss": 0.0046,
"step": 13020
},
{
"epoch": 2.5777030368978138,
"grad_norm": 0.0037734430486638187,
"learning_rate": 5.922940239451935e-06,
"loss": 0.0046,
"step": 13030
},
{
"epoch": 2.5796814719556833,
"grad_norm": 0.0028729598545328416,
"learning_rate": 5.868720593235017e-06,
"loss": 0.0046,
"step": 13040
},
{
"epoch": 2.5816599070135524,
"grad_norm": 0.013239324259412395,
"learning_rate": 5.814734781225589e-06,
"loss": 0.0046,
"step": 13050
},
{
"epoch": 2.5836383420714215,
"grad_norm": 0.006597090652450134,
"learning_rate": 5.760983089473143e-06,
"loss": 0.0047,
"step": 13060
},
{
"epoch": 2.5856167771292906,
"grad_norm": 0.003193389822996803,
"learning_rate": 5.707465802786654e-06,
"loss": 0.0046,
"step": 13070
},
{
"epoch": 2.5875952121871597,
"grad_norm": 0.003367478022575721,
"learning_rate": 5.654183204733049e-06,
"loss": 0.0046,
"step": 13080
},
{
"epoch": 2.5895736472450293,
"grad_norm": 0.006941548480813207,
"learning_rate": 5.601135577635775e-06,
"loss": 0.0046,
"step": 13090
},
{
"epoch": 2.5915520823028984,
"grad_norm": 0.0023608550306655944,
"learning_rate": 5.54832320257323e-06,
"loss": 0.0046,
"step": 13100
},
{
"epoch": 2.593530517360768,
"grad_norm": 0.003730927128171682,
"learning_rate": 5.495746359377335e-06,
"loss": 0.0046,
"step": 13110
},
{
"epoch": 2.595508952418637,
"grad_norm": 0.0032545840626603127,
"learning_rate": 5.443405326631984e-06,
"loss": 0.0046,
"step": 13120
},
{
"epoch": 2.597487387476506,
"grad_norm": 0.004475302653521189,
"learning_rate": 5.391300381671627e-06,
"loss": 0.0046,
"step": 13130
},
{
"epoch": 2.599465822534375,
"grad_norm": 0.008785230892289806,
"learning_rate": 5.339431800579781e-06,
"loss": 0.0046,
"step": 13140
},
{
"epoch": 2.6014442575922443,
"grad_norm": 0.0027249975041392024,
"learning_rate": 5.287799858187548e-06,
"loss": 0.0046,
"step": 13150
},
{
"epoch": 2.603422692650114,
"grad_norm": 0.010394431276799456,
"learning_rate": 5.236404828072183e-06,
"loss": 0.0046,
"step": 13160
},
{
"epoch": 2.605401127707983,
"grad_norm": 0.004885147960158267,
"learning_rate": 5.185246982555641e-06,
"loss": 0.0046,
"step": 13170
},
{
"epoch": 2.6073795627658525,
"grad_norm": 0.0027018424703615864,
"learning_rate": 5.134326592703131e-06,
"loss": 0.0046,
"step": 13180
},
{
"epoch": 2.6093579978237216,
"grad_norm": 0.005594548049832172,
"learning_rate": 5.083643928321679e-06,
"loss": 0.0046,
"step": 13190
},
{
"epoch": 2.6113364328815907,
"grad_norm": 0.0029596645625657694,
"learning_rate": 5.033199257958671e-06,
"loss": 0.0047,
"step": 13200
},
{
"epoch": 2.6133148679394598,
"grad_norm": 0.003918016696516807,
"learning_rate": 4.9829928489004975e-06,
"loss": 0.0046,
"step": 13210
},
{
"epoch": 2.615293302997329,
"grad_norm": 0.008896906582424303,
"learning_rate": 4.933024967171063e-06,
"loss": 0.0046,
"step": 13220
},
{
"epoch": 2.6172717380551984,
"grad_norm": 0.00472676698518972,
"learning_rate": 4.88329587753043e-06,
"loss": 0.0046,
"step": 13230
},
{
"epoch": 2.6192501731130675,
"grad_norm": 0.0017922171971847063,
"learning_rate": 4.8338058434733935e-06,
"loss": 0.0046,
"step": 13240
},
{
"epoch": 2.621228608170937,
"grad_norm": 0.008520221902307513,
"learning_rate": 4.784555127228091e-06,
"loss": 0.0047,
"step": 13250
},
{
"epoch": 2.623207043228806,
"grad_norm": 0.00364322772777207,
"learning_rate": 4.735543989754587e-06,
"loss": 0.0046,
"step": 13260
},
{
"epoch": 2.6251854782866753,
"grad_norm": 0.004317864316878868,
"learning_rate": 4.686772690743529e-06,
"loss": 0.0046,
"step": 13270
},
{
"epoch": 2.6271639133445444,
"grad_norm": 0.00866624590340705,
"learning_rate": 4.638241488614758e-06,
"loss": 0.0047,
"step": 13280
},
{
"epoch": 2.6291423484024135,
"grad_norm": 0.009585852042433162,
"learning_rate": 4.5899506405159335e-06,
"loss": 0.0046,
"step": 13290
},
{
"epoch": 2.631120783460283,
"grad_norm": 0.0045117722923979225,
"learning_rate": 4.5419004023211695e-06,
"loss": 0.0046,
"step": 13300
},
{
"epoch": 2.633099218518152,
"grad_norm": 0.001237195117880036,
"learning_rate": 4.494091028629699e-06,
"loss": 0.0046,
"step": 13310
},
{
"epoch": 2.6350776535760216,
"grad_norm": 0.005616253652264756,
"learning_rate": 4.446522772764494e-06,
"loss": 0.0046,
"step": 13320
},
{
"epoch": 2.6370560886338907,
"grad_norm": 0.006385849142279609,
"learning_rate": 4.3991958867709345e-06,
"loss": 0.0046,
"step": 13330
},
{
"epoch": 2.63903452369176,
"grad_norm": 0.0007570184678947039,
"learning_rate": 4.352110621415484e-06,
"loss": 0.0046,
"step": 13340
},
{
"epoch": 2.641012958749629,
"grad_norm": 0.0011536038966337653,
"learning_rate": 4.305267226184356e-06,
"loss": 0.0046,
"step": 13350
},
{
"epoch": 2.642991393807498,
"grad_norm": 0.003190238337374958,
"learning_rate": 4.258665949282192e-06,
"loss": 0.0046,
"step": 13360
},
{
"epoch": 2.6449698288653676,
"grad_norm": 0.009766269099607174,
"learning_rate": 4.212307037630747e-06,
"loss": 0.0046,
"step": 13370
},
{
"epoch": 2.6469482639232367,
"grad_norm": 0.004237523484823364,
"learning_rate": 4.1661907368675535e-06,
"loss": 0.0046,
"step": 13380
},
{
"epoch": 2.648926698981106,
"grad_norm": 0.002847146495047342,
"learning_rate": 4.1203172913446775e-06,
"loss": 0.0046,
"step": 13390
},
{
"epoch": 2.6509051340389753,
"grad_norm": 0.005977544860245569,
"learning_rate": 4.074686944127365e-06,
"loss": 0.0046,
"step": 13400
},
{
"epoch": 2.6528835690968444,
"grad_norm": 0.0024874558915793724,
"learning_rate": 4.029299936992792e-06,
"loss": 0.0046,
"step": 13410
},
{
"epoch": 2.6548620041547135,
"grad_norm": 0.0010007971892299517,
"learning_rate": 3.984156510428777e-06,
"loss": 0.0046,
"step": 13420
},
{
"epoch": 2.6568404392125826,
"grad_norm": 0.0058827649768598625,
"learning_rate": 3.939256903632494e-06,
"loss": 0.0046,
"step": 13430
},
{
"epoch": 2.658818874270452,
"grad_norm": 0.0049870793585777475,
"learning_rate": 3.89460135450922e-06,
"loss": 0.0046,
"step": 13440
},
{
"epoch": 2.6607973093283213,
"grad_norm": 0.00414610423933033,
"learning_rate": 3.850190099671047e-06,
"loss": 0.0046,
"step": 13450
},
{
"epoch": 2.6627757443861904,
"grad_norm": 0.002512780578889755,
"learning_rate": 3.8060233744356633e-06,
"loss": 0.0046,
"step": 13460
},
{
"epoch": 2.66475417944406,
"grad_norm": 0.005225356787057185,
"learning_rate": 3.7621014128250977e-06,
"loss": 0.0046,
"step": 13470
},
{
"epoch": 2.666732614501929,
"grad_norm": 0.002370669417036777,
"learning_rate": 3.7184244475644616e-06,
"loss": 0.0047,
"step": 13480
},
{
"epoch": 2.668711049559798,
"grad_norm": 0.003913797432239136,
"learning_rate": 3.6749927100807434e-06,
"loss": 0.0046,
"step": 13490
},
{
"epoch": 2.670689484617667,
"grad_norm": 0.006789128461192986,
"learning_rate": 3.6318064305015332e-06,
"loss": 0.0046,
"step": 13500
},
{
"epoch": 2.6726679196755367,
"grad_norm": 0.0009129940892869895,
"learning_rate": 3.5888658376538655e-06,
"loss": 0.0046,
"step": 13510
},
{
"epoch": 2.674646354733406,
"grad_norm": 0.00809943653017259,
"learning_rate": 3.5461711590629774e-06,
"loss": 0.0046,
"step": 13520
},
{
"epoch": 2.676624789791275,
"grad_norm": 0.0035324219826922926,
"learning_rate": 3.5037226209510887e-06,
"loss": 0.0046,
"step": 13530
},
{
"epoch": 2.6786032248491445,
"grad_norm": 0.0015469609256242396,
"learning_rate": 3.4615204482362386e-06,
"loss": 0.0046,
"step": 13540
},
{
"epoch": 2.6805816599070136,
"grad_norm": 0.00239016138245589,
"learning_rate": 3.419564864531044e-06,
"loss": 0.0046,
"step": 13550
},
{
"epoch": 2.6825600949648827,
"grad_norm": 0.005110117118259523,
"learning_rate": 3.3778560921415738e-06,
"loss": 0.0046,
"step": 13560
},
{
"epoch": 2.684538530022752,
"grad_norm": 0.004585127100803052,
"learning_rate": 3.3363943520661132e-06,
"loss": 0.0046,
"step": 13570
},
{
"epoch": 2.6865169650806213,
"grad_norm": 0.004567544664598693,
"learning_rate": 3.2951798639940467e-06,
"loss": 0.0046,
"step": 13580
},
{
"epoch": 2.6884954001384904,
"grad_norm": 0.002907993514367838,
"learning_rate": 3.2542128463046497e-06,
"loss": 0.0046,
"step": 13590
},
{
"epoch": 2.6904738351963595,
"grad_norm": 0.001599103437056259,
"learning_rate": 3.2134935160659686e-06,
"loss": 0.0046,
"step": 13600
},
{
"epoch": 2.692452270254229,
"grad_norm": 0.0035874295221592803,
"learning_rate": 3.173022089033645e-06,
"loss": 0.0046,
"step": 13610
},
{
"epoch": 2.694430705312098,
"grad_norm": 0.0038263997418943405,
"learning_rate": 3.1327987796497537e-06,
"loss": 0.0046,
"step": 13620
},
{
"epoch": 2.6964091403699673,
"grad_norm": 0.0021494062192137053,
"learning_rate": 3.0928238010417275e-06,
"loss": 0.0047,
"step": 13630
},
{
"epoch": 2.6983875754278364,
"grad_norm": 0.0034280419224544207,
"learning_rate": 3.0530973650211793e-06,
"loss": 0.0046,
"step": 13640
},
{
"epoch": 2.700366010485706,
"grad_norm": 0.006481271365825322,
"learning_rate": 3.0136196820827855e-06,
"loss": 0.0047,
"step": 13650
},
{
"epoch": 2.702344445543575,
"grad_norm": 0.009859141799020046,
"learning_rate": 2.974390961403195e-06,
"loss": 0.0047,
"step": 13660
},
{
"epoch": 2.704322880601444,
"grad_norm": 0.007348156009129425,
"learning_rate": 2.935411410839889e-06,
"loss": 0.0046,
"step": 13670
},
{
"epoch": 2.7063013156593136,
"grad_norm": 0.005944030781047453,
"learning_rate": 2.8966812369301043e-06,
"loss": 0.0046,
"step": 13680
},
{
"epoch": 2.7082797507171827,
"grad_norm": 0.003474008647576606,
"learning_rate": 2.8582006448897135e-06,
"loss": 0.0047,
"step": 13690
},
{
"epoch": 2.710258185775052,
"grad_norm": 0.005453809638263781,
"learning_rate": 2.8199698386121687e-06,
"loss": 0.0046,
"step": 13700
},
{
"epoch": 2.712236620832921,
"grad_norm": 0.008950266832599228,
"learning_rate": 2.7819890206674084e-06,
"loss": 0.0046,
"step": 13710
},
{
"epoch": 2.7142150558907905,
"grad_norm": 0.003649536582463252,
"learning_rate": 2.7442583923007746e-06,
"loss": 0.0046,
"step": 13720
},
{
"epoch": 2.7161934909486596,
"grad_norm": 0.007433459031096395,
"learning_rate": 2.706778153431949e-06,
"loss": 0.0046,
"step": 13730
},
{
"epoch": 2.7181719260065287,
"grad_norm": 0.0044163527790853216,
"learning_rate": 2.669548502653918e-06,
"loss": 0.0046,
"step": 13740
},
{
"epoch": 2.7201503610643982,
"grad_norm": 0.0073784528177694665,
"learning_rate": 2.6325696372318688e-06,
"loss": 0.0047,
"step": 13750
},
{
"epoch": 2.7221287961222673,
"grad_norm": 0.0036631848222943514,
"learning_rate": 2.5958417531022017e-06,
"loss": 0.0047,
"step": 13760
},
{
"epoch": 2.7241072311801364,
"grad_norm": 0.005939601159110849,
"learning_rate": 2.559365044871459e-06,
"loss": 0.0046,
"step": 13770
},
{
"epoch": 2.7260856662380055,
"grad_norm": 0.007108086775249662,
"learning_rate": 2.523139705815297e-06,
"loss": 0.0046,
"step": 13780
},
{
"epoch": 2.728064101295875,
"grad_norm": 0.0017446179007564247,
"learning_rate": 2.4871659278774883e-06,
"loss": 0.0046,
"step": 13790
},
{
"epoch": 2.730042536353744,
"grad_norm": 0.0038899068621689407,
"learning_rate": 2.4514439016688374e-06,
"loss": 0.0046,
"step": 13800
},
{
"epoch": 2.7320209714116133,
"grad_norm": 0.0014983331526322956,
"learning_rate": 2.415973816466255e-06,
"loss": 0.0046,
"step": 13810
},
{
"epoch": 2.733999406469483,
"grad_norm": 0.0028712086220151634,
"learning_rate": 2.3807558602117143e-06,
"loss": 0.0046,
"step": 13820
},
{
"epoch": 2.735977841527352,
"grad_norm": 0.0025180894925366844,
"learning_rate": 2.3457902195112236e-06,
"loss": 0.0046,
"step": 13830
},
{
"epoch": 2.737956276585221,
"grad_norm": 0.006069978522393488,
"learning_rate": 2.311077079633911e-06,
"loss": 0.0046,
"step": 13840
},
{
"epoch": 2.73993471164309,
"grad_norm": 0.003449993664559199,
"learning_rate": 2.2766166245109744e-06,
"loss": 0.0046,
"step": 13850
},
{
"epoch": 2.7419131467009596,
"grad_norm": 0.005155431711304329,
"learning_rate": 2.2424090367347605e-06,
"loss": 0.0046,
"step": 13860
},
{
"epoch": 2.7438915817588287,
"grad_norm": 0.002802048529433583,
"learning_rate": 2.2084544975577383e-06,
"loss": 0.0046,
"step": 13870
},
{
"epoch": 2.745870016816698,
"grad_norm": 0.004629123753612135,
"learning_rate": 2.174753186891604e-06,
"loss": 0.0047,
"step": 13880
},
{
"epoch": 2.7478484518745674,
"grad_norm": 0.005571463177970553,
"learning_rate": 2.1413052833062773e-06,
"loss": 0.0046,
"step": 13890
},
{
"epoch": 2.7498268869324365,
"grad_norm": 0.0037513764127805735,
"learning_rate": 2.1081109640289755e-06,
"loss": 0.0046,
"step": 13900
},
{
"epoch": 2.7518053219903056,
"grad_norm": 0.010786774430351889,
"learning_rate": 2.0751704049432943e-06,
"loss": 0.0046,
"step": 13910
},
{
"epoch": 2.7537837570481747,
"grad_norm": 0.0021090460165309583,
"learning_rate": 2.0424837805882224e-06,
"loss": 0.0046,
"step": 13920
},
{
"epoch": 2.7557621921060442,
"grad_norm": 0.0013570228629476366,
"learning_rate": 2.0100512641572587e-06,
"loss": 0.0046,
"step": 13930
},
{
"epoch": 2.7577406271639133,
"grad_norm": 0.005942438914260372,
"learning_rate": 1.9778730274975012e-06,
"loss": 0.0046,
"step": 13940
},
{
"epoch": 2.7597190622217824,
"grad_norm": 0.002213171678604495,
"learning_rate": 1.945949241108708e-06,
"loss": 0.0046,
"step": 13950
},
{
"epoch": 2.761697497279652,
"grad_norm": 0.002961786041624848,
"learning_rate": 1.914280074142405e-06,
"loss": 0.0046,
"step": 13960
},
{
"epoch": 2.763675932337521,
"grad_norm": 0.0054852252790529715,
"learning_rate": 1.8828656944009814e-06,
"loss": 0.0046,
"step": 13970
},
{
"epoch": 2.76565436739539,
"grad_norm": 0.002527203885260355,
"learning_rate": 1.8517062683368335e-06,
"loss": 0.0046,
"step": 13980
},
{
"epoch": 2.7676328024532593,
"grad_norm": 0.005844519735218453,
"learning_rate": 1.820801961051427e-06,
"loss": 0.0047,
"step": 13990
},
{
"epoch": 2.769611237511129,
"grad_norm": 0.003273203480884671,
"learning_rate": 1.7901529362944869e-06,
"loss": 0.0046,
"step": 14000
},
{
"epoch": 2.771589672568998,
"grad_norm": 0.0029042101372606616,
"learning_rate": 1.759759356463081e-06,
"loss": 0.0046,
"step": 14010
},
{
"epoch": 2.773568107626867,
"grad_norm": 0.0038404767019665663,
"learning_rate": 1.7296213826007712e-06,
"loss": 0.0046,
"step": 14020
},
{
"epoch": 2.7755465426847366,
"grad_norm": 0.0012990129318038214,
"learning_rate": 1.6997391743967695e-06,
"loss": 0.0046,
"step": 14030
},
{
"epoch": 2.7775249777426056,
"grad_norm": 0.004486312870847904,
"learning_rate": 1.6701128901850937e-06,
"loss": 0.0046,
"step": 14040
},
{
"epoch": 2.7795034128004747,
"grad_norm": 0.0034379690990209606,
"learning_rate": 1.6407426869437026e-06,
"loss": 0.0046,
"step": 14050
},
{
"epoch": 2.781481847858344,
"grad_norm": 0.012135636046883379,
"learning_rate": 1.6116287202937063e-06,
"loss": 0.0046,
"step": 14060
},
{
"epoch": 2.7834602829162134,
"grad_norm": 0.005929944311555382,
"learning_rate": 1.5827711444985016e-06,
"loss": 0.0046,
"step": 14070
},
{
"epoch": 2.7854387179740825,
"grad_norm": 0.0036208389926315936,
"learning_rate": 1.554170112462977e-06,
"loss": 0.0046,
"step": 14080
},
{
"epoch": 2.7874171530319516,
"grad_norm": 0.00465174528997413,
"learning_rate": 1.5258257757327033e-06,
"loss": 0.0046,
"step": 14090
},
{
"epoch": 2.789395588089821,
"grad_norm": 0.0022969630516721977,
"learning_rate": 1.4977382844931166e-06,
"loss": 0.0046,
"step": 14100
},
{
"epoch": 2.7913740231476902,
"grad_norm": 0.005575498935687786,
"learning_rate": 1.4699077875687251e-06,
"loss": 0.0046,
"step": 14110
},
{
"epoch": 2.7933524582055593,
"grad_norm": 0.001755016235640258,
"learning_rate": 1.442334432422332e-06,
"loss": 0.0046,
"step": 14120
},
{
"epoch": 2.7953308932634284,
"grad_norm": 0.00458177304562813,
"learning_rate": 1.415018365154258e-06,
"loss": 0.0046,
"step": 14130
},
{
"epoch": 2.797309328321298,
"grad_norm": 0.005779460730728936,
"learning_rate": 1.3879597305015368e-06,
"loss": 0.0046,
"step": 14140
},
{
"epoch": 2.799287763379167,
"grad_norm": 0.004323932484579805,
"learning_rate": 1.3611586718371871e-06,
"loss": 0.0047,
"step": 14150
},
{
"epoch": 2.801266198437036,
"grad_norm": 0.002088461725705357,
"learning_rate": 1.3346153311694309e-06,
"loss": 0.0046,
"step": 14160
},
{
"epoch": 2.8032446334949057,
"grad_norm": 0.0023946939333232076,
"learning_rate": 1.3083298491409213e-06,
"loss": 0.0046,
"step": 14170
},
{
"epoch": 2.805223068552775,
"grad_norm": 0.0015444352083862514,
"learning_rate": 1.2823023650280541e-06,
"loss": 0.0046,
"step": 14180
},
{
"epoch": 2.807201503610644,
"grad_norm": 0.004584827016385754,
"learning_rate": 1.2565330167401745e-06,
"loss": 0.0046,
"step": 14190
},
{
"epoch": 2.809179938668513,
"grad_norm": 0.006582194237980956,
"learning_rate": 1.2310219408188773e-06,
"loss": 0.0046,
"step": 14200
},
{
"epoch": 2.8111583737263826,
"grad_norm": 0.005090877994607883,
"learning_rate": 1.2057692724372794e-06,
"loss": 0.0046,
"step": 14210
},
{
"epoch": 2.8131368087842517,
"grad_norm": 0.004558257657543348,
"learning_rate": 1.1807751453992654e-06,
"loss": 0.0046,
"step": 14220
},
{
"epoch": 2.8151152438421208,
"grad_norm": 0.004312430223801912,
"learning_rate": 1.156039692138855e-06,
"loss": 0.0046,
"step": 14230
},
{
"epoch": 2.8170936788999903,
"grad_norm": 0.004023197361355825,
"learning_rate": 1.1315630437194359e-06,
"loss": 0.0046,
"step": 14240
},
{
"epoch": 2.8190721139578594,
"grad_norm": 0.010775752029811077,
"learning_rate": 1.1073453298330982e-06,
"loss": 0.0046,
"step": 14250
},
{
"epoch": 2.8210505490157285,
"grad_norm": 0.005477207284372358,
"learning_rate": 1.0833866787999248e-06,
"loss": 0.0047,
"step": 14260
},
{
"epoch": 2.8230289840735976,
"grad_norm": 0.001716182302177534,
"learning_rate": 1.0596872175673455e-06,
"loss": 0.0046,
"step": 14270
},
{
"epoch": 2.825007419131467,
"grad_norm": 0.00453893662140068,
"learning_rate": 1.0362470717094397e-06,
"loss": 0.0046,
"step": 14280
},
{
"epoch": 2.8269858541893362,
"grad_norm": 0.005175445572724438,
"learning_rate": 1.0130663654262684e-06,
"loss": 0.0046,
"step": 14290
},
{
"epoch": 2.8289642892472053,
"grad_norm": 0.007451324176885912,
"learning_rate": 9.901452215432428e-07,
"loss": 0.0046,
"step": 14300
},
{
"epoch": 2.830942724305075,
"grad_norm": 0.002051765979202371,
"learning_rate": 9.674837615104349e-07,
"loss": 0.0046,
"step": 14310
},
{
"epoch": 2.832921159362944,
"grad_norm": 0.004166722993673084,
"learning_rate": 9.450821054019732e-07,
"loss": 0.0046,
"step": 14320
},
{
"epoch": 2.834899594420813,
"grad_norm": 0.0020158192332341714,
"learning_rate": 9.229403719153817e-07,
"loss": 0.0046,
"step": 14330
},
{
"epoch": 2.836878029478682,
"grad_norm": 0.002314492220451588,
"learning_rate": 9.010586783709418e-07,
"loss": 0.0046,
"step": 14340
},
{
"epoch": 2.8388564645365517,
"grad_norm": 0.006785113150448129,
"learning_rate": 8.794371407111091e-07,
"loss": 0.0046,
"step": 14350
},
{
"epoch": 2.840834899594421,
"grad_norm": 0.0007663110571174921,
"learning_rate": 8.580758734998639e-07,
"loss": 0.0046,
"step": 14360
},
{
"epoch": 2.84281333465229,
"grad_norm": 0.004119548797094098,
"learning_rate": 8.36974989922118e-07,
"loss": 0.0046,
"step": 14370
},
{
"epoch": 2.8447917697101595,
"grad_norm": 0.007005562455027145,
"learning_rate": 8.161346017831195e-07,
"loss": 0.0046,
"step": 14380
},
{
"epoch": 2.8467702047680286,
"grad_norm": 0.002528753473523976,
"learning_rate": 7.955548195078433e-07,
"loss": 0.0046,
"step": 14390
},
{
"epoch": 2.8487486398258977,
"grad_norm": 0.0018960262272628438,
"learning_rate": 7.752357521404186e-07,
"loss": 0.0046,
"step": 14400
},
{
"epoch": 2.8507270748837668,
"grad_norm": 0.005897476313959371,
"learning_rate": 7.551775073435574e-07,
"loss": 0.0046,
"step": 14410
},
{
"epoch": 2.8527055099416363,
"grad_norm": 0.0054337151694665116,
"learning_rate": 7.353801913979719e-07,
"loss": 0.0046,
"step": 14420
},
{
"epoch": 2.8546839449995054,
"grad_norm": 0.006764251262869732,
"learning_rate": 7.158439092018076e-07,
"loss": 0.0046,
"step": 14430
},
{
"epoch": 2.8566623800573745,
"grad_norm": 0.0010954146697080148,
"learning_rate": 6.965687642701113e-07,
"loss": 0.0046,
"step": 14440
},
{
"epoch": 2.858640815115244,
"grad_norm": 0.0025798782274373206,
"learning_rate": 6.775548587342529e-07,
"loss": 0.0046,
"step": 14450
},
{
"epoch": 2.860619250173113,
"grad_norm": 0.001403265665669007,
"learning_rate": 6.588022933414095e-07,
"loss": 0.0046,
"step": 14460
},
{
"epoch": 2.8625976852309822,
"grad_norm": 0.004191335772000475,
"learning_rate": 6.403111674539997e-07,
"loss": 0.0046,
"step": 14470
},
{
"epoch": 2.8645761202888513,
"grad_norm": 0.005768077099999293,
"learning_rate": 6.220815790492051e-07,
"loss": 0.0046,
"step": 14480
},
{
"epoch": 2.866554555346721,
"grad_norm": 0.0029096217336345423,
"learning_rate": 6.041136247184054e-07,
"loss": 0.0047,
"step": 14490
},
{
"epoch": 2.86853299040459,
"grad_norm": 0.002569374881028189,
"learning_rate": 5.86407399666683e-07,
"loss": 0.0046,
"step": 14500
},
{
"epoch": 2.870511425462459,
"grad_norm": 0.00585231025762138,
"learning_rate": 5.689629977123412e-07,
"loss": 0.0046,
"step": 14510
},
{
"epoch": 2.8724898605203286,
"grad_norm": 0.004988123311533302,
"learning_rate": 5.517805112863539e-07,
"loss": 0.0046,
"step": 14520
},
{
"epoch": 2.8744682955781977,
"grad_norm": 0.005182944566904667,
"learning_rate": 5.348600314319385e-07,
"loss": 0.0047,
"step": 14530
},
{
"epoch": 2.876446730636067,
"grad_norm": 0.008808716121352934,
"learning_rate": 5.182016478040231e-07,
"loss": 0.0046,
"step": 14540
},
{
"epoch": 2.878425165693936,
"grad_norm": 0.008366614382327059,
"learning_rate": 5.018054486687962e-07,
"loss": 0.0046,
"step": 14550
},
{
"epoch": 2.8804036007518055,
"grad_norm": 0.0006534242313928812,
"learning_rate": 4.85671520903247e-07,
"loss": 0.0046,
"step": 14560
},
{
"epoch": 2.8823820358096746,
"grad_norm": 0.007238183348467795,
"learning_rate": 4.697999499946704e-07,
"loss": 0.0046,
"step": 14570
},
{
"epoch": 2.8843604708675437,
"grad_norm": 0.007835929584908773,
"learning_rate": 4.5419082004026226e-07,
"loss": 0.0046,
"step": 14580
},
{
"epoch": 2.886338905925413,
"grad_norm": 0.008277987543645947,
"learning_rate": 4.3884421374661975e-07,
"loss": 0.0046,
"step": 14590
},
{
"epoch": 2.8883173409832823,
"grad_norm": 0.0020163978499294338,
"learning_rate": 4.2376021242934714e-07,
"loss": 0.0046,
"step": 14600
},
{
"epoch": 2.8902957760411514,
"grad_norm": 0.0042552763507915595,
"learning_rate": 4.089388960126117e-07,
"loss": 0.0046,
"step": 14610
},
{
"epoch": 2.8922742110990205,
"grad_norm": 0.0060658041438372,
"learning_rate": 3.943803430287052e-07,
"loss": 0.0046,
"step": 14620
},
{
"epoch": 2.89425264615689,
"grad_norm": 0.005596470744621037,
"learning_rate": 3.8008463061764977e-07,
"loss": 0.0046,
"step": 14630
},
{
"epoch": 2.896231081214759,
"grad_norm": 0.008598652307480305,
"learning_rate": 3.660518345267705e-07,
"loss": 0.0046,
"step": 14640
},
{
"epoch": 2.8982095162726282,
"grad_norm": 0.003172914604657869,
"learning_rate": 3.5228202911030685e-07,
"loss": 0.0046,
"step": 14650
},
{
"epoch": 2.900187951330498,
"grad_norm": 0.004793924110057877,
"learning_rate": 3.38775287329024e-07,
"loss": 0.0046,
"step": 14660
},
{
"epoch": 2.902166386388367,
"grad_norm": 0.004140086966086256,
"learning_rate": 3.255316807498077e-07,
"loss": 0.0047,
"step": 14670
},
{
"epoch": 2.904144821446236,
"grad_norm": 0.0028868173544492758,
"learning_rate": 3.125512795452812e-07,
"loss": 0.0046,
"step": 14680
},
{
"epoch": 2.906123256504105,
"grad_norm": 0.005336724594207578,
"learning_rate": 2.998341524934778e-07,
"loss": 0.0047,
"step": 14690
},
{
"epoch": 2.9081016915619746,
"grad_norm": 0.004909440284560435,
"learning_rate": 2.873803669774133e-07,
"loss": 0.0047,
"step": 14700
},
{
"epoch": 2.9100801266198437,
"grad_norm": 0.004056939910888324,
"learning_rate": 2.751899889847864e-07,
"loss": 0.0046,
"step": 14710
},
{
"epoch": 2.912058561677713,
"grad_norm": 0.0031709793919959886,
"learning_rate": 2.632630831075844e-07,
"loss": 0.0046,
"step": 14720
},
{
"epoch": 2.9140369967355824,
"grad_norm": 0.006708328109396972,
"learning_rate": 2.5159971254176684e-07,
"loss": 0.0047,
"step": 14730
},
{
"epoch": 2.9160154317934515,
"grad_norm": 0.007981747819856977,
"learning_rate": 2.401999390869325e-07,
"loss": 0.0046,
"step": 14740
},
{
"epoch": 2.9179938668513206,
"grad_norm": 0.0007989972259983827,
"learning_rate": 2.290638231459641e-07,
"loss": 0.0046,
"step": 14750
},
{
"epoch": 2.9199723019091897,
"grad_norm": 0.005320786995718429,
"learning_rate": 2.1819142372475066e-07,
"loss": 0.0046,
"step": 14760
},
{
"epoch": 2.921950736967059,
"grad_norm": 0.004589876050172932,
"learning_rate": 2.0758279843183793e-07,
"loss": 0.0046,
"step": 14770
},
{
"epoch": 2.9239291720249283,
"grad_norm": 0.005953966071911301,
"learning_rate": 1.972380034781285e-07,
"loss": 0.0046,
"step": 14780
},
{
"epoch": 2.9259076070827974,
"grad_norm": 0.005471478263219146,
"learning_rate": 1.8715709367660982e-07,
"loss": 0.0046,
"step": 14790
},
{
"epoch": 2.927886042140667,
"grad_norm": 0.006238885812858415,
"learning_rate": 1.773401224420379e-07,
"loss": 0.0046,
"step": 14800
},
{
"epoch": 2.929864477198536,
"grad_norm": 0.003119978559663112,
"learning_rate": 1.6778714179065957e-07,
"loss": 0.0046,
"step": 14810
},
{
"epoch": 2.931842912256405,
"grad_norm": 0.002903269015219757,
"learning_rate": 1.584982023399517e-07,
"loss": 0.0046,
"step": 14820
},
{
"epoch": 2.9338213473142742,
"grad_norm": 0.003790039070911083,
"learning_rate": 1.4947335330832702e-07,
"loss": 0.0046,
"step": 14830
},
{
"epoch": 2.935799782372144,
"grad_norm": 0.0016732999939370978,
"learning_rate": 1.4071264251489523e-07,
"loss": 0.0046,
"step": 14840
},
{
"epoch": 2.937778217430013,
"grad_norm": 0.004290456627577342,
"learning_rate": 1.3221611637919685e-07,
"loss": 0.0046,
"step": 14850
},
{
"epoch": 2.939756652487882,
"grad_norm": 0.003443634280001439,
"learning_rate": 1.2398381992096975e-07,
"loss": 0.0046,
"step": 14860
},
{
"epoch": 2.9417350875457515,
"grad_norm": 0.003652422871644758,
"learning_rate": 1.160157967598996e-07,
"loss": 0.0046,
"step": 14870
},
{
"epoch": 2.9437135226036206,
"grad_norm": 0.004129312067433668,
"learning_rate": 1.0831208911539215e-07,
"loss": 0.0046,
"step": 14880
},
{
"epoch": 2.9456919576614897,
"grad_norm": 0.005116448093157166,
"learning_rate": 1.0087273780634566e-07,
"loss": 0.0046,
"step": 14890
},
{
"epoch": 2.947670392719359,
"grad_norm": 0.007496650483840799,
"learning_rate": 9.36977822509455e-08,
"loss": 0.0046,
"step": 14900
},
{
"epoch": 2.9496488277772284,
"grad_norm": 0.004263858709885515,
"learning_rate": 8.678726046644214e-08,
"loss": 0.0046,
"step": 14910
},
{
"epoch": 2.9516272628350975,
"grad_norm": 0.001472401412939499,
"learning_rate": 8.014120906896239e-08,
"loss": 0.0046,
"step": 14920
},
{
"epoch": 2.9536056978929666,
"grad_norm": 0.002624416110714285,
"learning_rate": 7.375966327330952e-08,
"loss": 0.0046,
"step": 14930
},
{
"epoch": 2.955584132950836,
"grad_norm": 0.003920433741185325,
"learning_rate": 6.76426568927635e-08,
"loss": 0.0046,
"step": 14940
},
{
"epoch": 2.957562568008705,
"grad_norm": 0.006638768765171172,
"learning_rate": 6.179022233893106e-08,
"loss": 0.0047,
"step": 14950
},
{
"epoch": 2.9595410030665743,
"grad_norm": 0.007957061711529703,
"learning_rate": 5.6202390621551415e-08,
"loss": 0.0047,
"step": 14960
},
{
"epoch": 2.9615194381244434,
"grad_norm": 0.008210271781902008,
"learning_rate": 5.0879191348324194e-08,
"loss": 0.0046,
"step": 14970
},
{
"epoch": 2.963497873182313,
"grad_norm": 0.005418351565007964,
"learning_rate": 4.5820652724787305e-08,
"loss": 0.0046,
"step": 14980
},
{
"epoch": 2.965476308240182,
"grad_norm": 0.0013190207343898369,
"learning_rate": 4.102680155413929e-08,
"loss": 0.0046,
"step": 14990
},
{
"epoch": 2.967454743298051,
"grad_norm": 0.0035468614380098234,
"learning_rate": 3.649766323710058e-08,
"loss": 0.0046,
"step": 15000
},
{
"epoch": 2.9694331783559207,
"grad_norm": 0.00400458791463646,
"learning_rate": 3.223326177178021e-08,
"loss": 0.0046,
"step": 15010
},
{
"epoch": 2.97141161341379,
"grad_norm": 0.007328477008627146,
"learning_rate": 2.8233619753564867e-08,
"loss": 0.0046,
"step": 15020
},
{
"epoch": 2.973390048471659,
"grad_norm": 0.0013854486422206437,
"learning_rate": 2.4498758374968954e-08,
"loss": 0.0046,
"step": 15030
},
{
"epoch": 2.975368483529528,
"grad_norm": 0.001931643826234835,
"learning_rate": 2.1028697425551356e-08,
"loss": 0.0046,
"step": 15040
},
{
"epoch": 2.9773469185873975,
"grad_norm": 0.005365205623647741,
"learning_rate": 1.78234552917933e-08,
"loss": 0.0047,
"step": 15050
},
{
"epoch": 2.9793253536452666,
"grad_norm": 0.001869691699893443,
"learning_rate": 1.4883048957003986e-08,
"loss": 0.0046,
"step": 15060
},
{
"epoch": 2.9813037887031357,
"grad_norm": 0.004698255884120171,
"learning_rate": 1.220749400123733e-08,
"loss": 0.0046,
"step": 15070
},
{
"epoch": 2.9832822237610053,
"grad_norm": 0.003309400652961775,
"learning_rate": 9.796804601197584e-09,
"loss": 0.0046,
"step": 15080
},
{
"epoch": 2.9852606588188744,
"grad_norm": 0.004195019249944128,
"learning_rate": 7.650993530178285e-09,
"loss": 0.0046,
"step": 15090
},
{
"epoch": 2.9872390938767435,
"grad_norm": 0.002538325018404864,
"learning_rate": 5.77007215798453e-09,
"loss": 0.0046,
"step": 15100
},
{
"epoch": 2.9892175289346126,
"grad_norm": 0.01122177791468114,
"learning_rate": 4.154050450871916e-09,
"loss": 0.0046,
"step": 15110
},
{
"epoch": 2.991195963992482,
"grad_norm": 0.0019075859604775073,
"learning_rate": 2.8029369715021345e-09,
"loss": 0.0047,
"step": 15120
},
{
"epoch": 2.993174399050351,
"grad_norm": 0.00439061740265406,
"learning_rate": 1.7167388788985606e-09,
"loss": 0.0046,
"step": 15130
},
{
"epoch": 2.9951528341082203,
"grad_norm": 0.009913037704454265,
"learning_rate": 8.954619283851884e-10,
"loss": 0.0046,
"step": 15140
},
{
"epoch": 2.99713126916609,
"grad_norm": 0.006363931291033015,
"learning_rate": 3.3911047158663447e-10,
"loss": 0.0046,
"step": 15150
},
{
"epoch": 2.999109704223959,
"grad_norm": 0.005973523259733178,
"learning_rate": 4.7687456389278186e-11,
"loss": 0.0046,
"step": 15160
},
{
"epoch": 3.0,
"step": 15165,
"total_flos": 1485582470479872.0,
"train_loss": 0.11580274374157613,
"train_runtime": 45663.8712,
"train_samples_per_second": 170.004,
"train_steps_per_second": 0.332
}
],
"logging_steps": 10,
"max_steps": 15165,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 1800,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1485582470479872.0,
"train_batch_size": 128,
"trial_name": null,
"trial_params": null
}