mlfoundations-dev/oh-dcft-v3.1-llama-3.3-70b-qwen
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity
Files
b594b8ca9145d76d89fd478f3889764756807e82
oh-dcft-v3.1-llama-3.3-70b-…/trainer_state.json
ModelHub XC b594b8ca91 初始化项目，由ModelHub XC社区提供模型 
Model: mlfoundations-dev/oh-dcft-v3.1-llama-3.3-70b-qwen
Source: Original Platform
2026-08-08 18:55:13 +08:00

3973 lines
88 KiB
JSON
Raw Blame History

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.999194198227236,
"eval_steps": 500,
"global_step": 5583,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.005372011818426001,
"grad_norm": 25.23110830120061,
"learning_rate": 5e-06,
"loss": 0.5524,
"step": 10
},
{
"epoch": 0.010744023636852002,
"grad_norm": 2.7826435546794013,
"learning_rate": 5e-06,
"loss": 0.5237,
"step": 20
},
{
"epoch": 0.016116035455278,
"grad_norm": 0.8773125581852239,
"learning_rate": 5e-06,
"loss": 0.5011,
"step": 30
},
{
"epoch": 0.021488047273704004,
"grad_norm": 0.905480676990675,
"learning_rate": 5e-06,
"loss": 0.4903,
"step": 40
},
{
"epoch": 0.026860059092130004,
"grad_norm": 0.9003527250388368,
"learning_rate": 5e-06,
"loss": 0.4725,
"step": 50
},
{
"epoch": 0.032232070910556,
"grad_norm": 0.7690412778021365,
"learning_rate": 5e-06,
"loss": 0.4683,
"step": 60
},
{
"epoch": 0.037604082728982004,
"grad_norm": 0.674228450328133,
"learning_rate": 5e-06,
"loss": 0.465,
"step": 70
},
{
"epoch": 0.04297609454740801,
"grad_norm": 0.5682520559652751,
"learning_rate": 5e-06,
"loss": 0.4744,
"step": 80
},
{
"epoch": 0.048348106365834004,
"grad_norm": 0.5487808393256164,
"learning_rate": 5e-06,
"loss": 0.4564,
"step": 90
},
{
"epoch": 0.05372011818426001,
"grad_norm": 0.5444906907383498,
"learning_rate": 5e-06,
"loss": 0.4466,
"step": 100
},
{
"epoch": 0.059092130002686004,
"grad_norm": 0.5101815571273731,
"learning_rate": 5e-06,
"loss": 0.4594,
"step": 110
},
{
"epoch": 0.064464141821112,
"grad_norm": 0.5299045266679747,
"learning_rate": 5e-06,
"loss": 0.4608,
"step": 120
},
{
"epoch": 0.069836153639538,
"grad_norm": 0.4867455311822639,
"learning_rate": 5e-06,
"loss": 0.4594,
"step": 130
},
{
"epoch": 0.07520816545796401,
"grad_norm": 0.5226511293151176,
"learning_rate": 5e-06,
"loss": 0.4537,
"step": 140
},
{
"epoch": 0.08058017727639001,
"grad_norm": 0.5413759464697727,
"learning_rate": 5e-06,
"loss": 0.4614,
"step": 150
},
{
"epoch": 0.08595218909481601,
"grad_norm": 0.5253819183696734,
"learning_rate": 5e-06,
"loss": 0.4677,
"step": 160
},
{
"epoch": 0.091324200913242,
"grad_norm": 0.512774010217095,
"learning_rate": 5e-06,
"loss": 0.4663,
"step": 170
},
{
"epoch": 0.09669621273166801,
"grad_norm": 0.5161443496807266,
"learning_rate": 5e-06,
"loss": 0.4445,
"step": 180
},
{
"epoch": 0.10206822455009401,
"grad_norm": 0.53161510586257,
"learning_rate": 5e-06,
"loss": 0.4386,
"step": 190
},
{
"epoch": 0.10744023636852001,
"grad_norm": 0.5033874795161914,
"learning_rate": 5e-06,
"loss": 0.4358,
"step": 200
},
{
"epoch": 0.11281224818694602,
"grad_norm": 0.5030823104996639,
"learning_rate": 5e-06,
"loss": 0.4509,
"step": 210
},
{
"epoch": 0.11818426000537201,
"grad_norm": 0.4921742024096051,
"learning_rate": 5e-06,
"loss": 0.454,
"step": 220
},
{
"epoch": 0.12355627182379801,
"grad_norm": 0.5151849417876855,
"learning_rate": 5e-06,
"loss": 0.4402,
"step": 230
},
{
"epoch": 0.128928283642224,
"grad_norm": 0.5509880033121557,
"learning_rate": 5e-06,
"loss": 0.4492,
"step": 240
},
{
"epoch": 0.13430029546065,
"grad_norm": 0.4657702949495027,
"learning_rate": 5e-06,
"loss": 0.4525,
"step": 250
},
{
"epoch": 0.139672307279076,
"grad_norm": 0.4907871132824897,
"learning_rate": 5e-06,
"loss": 0.4423,
"step": 260
},
{
"epoch": 0.145044319097502,
"grad_norm": 0.5002770016321901,
"learning_rate": 5e-06,
"loss": 0.4454,
"step": 270
},
{
"epoch": 0.15041633091592801,
"grad_norm": 0.5189467430778508,
"learning_rate": 5e-06,
"loss": 0.4421,
"step": 280
},
{
"epoch": 0.15578834273435402,
"grad_norm": 0.5292508408955507,
"learning_rate": 5e-06,
"loss": 0.4446,
"step": 290
},
{
"epoch": 0.16116035455278002,
"grad_norm": 0.49312368458399464,
"learning_rate": 5e-06,
"loss": 0.436,
"step": 300
},
{
"epoch": 0.16653236637120603,
"grad_norm": 0.5129893583398213,
"learning_rate": 5e-06,
"loss": 0.4299,
"step": 310
},
{
"epoch": 0.17190437818963203,
"grad_norm": 0.5066217704400877,
"learning_rate": 5e-06,
"loss": 0.472,
"step": 320
},
{
"epoch": 0.177276390008058,
"grad_norm": 0.523311153155544,
"learning_rate": 5e-06,
"loss": 0.451,
"step": 330
},
{
"epoch": 0.182648401826484,
"grad_norm": 0.48044358190446446,
"learning_rate": 5e-06,
"loss": 0.4402,
"step": 340
},
{
"epoch": 0.18802041364491,
"grad_norm": 0.5084268017861858,
"learning_rate": 5e-06,
"loss": 0.4427,
"step": 350
},
{
"epoch": 0.19339242546333602,
"grad_norm": 0.5590494266460699,
"learning_rate": 5e-06,
"loss": 0.435,
"step": 360
},
{
"epoch": 0.19876443728176202,
"grad_norm": 0.5150238173121026,
"learning_rate": 5e-06,
"loss": 0.457,
"step": 370
},
{
"epoch": 0.20413644910018802,
"grad_norm": 0.5179681470188343,
"learning_rate": 5e-06,
"loss": 0.4345,
"step": 380
},
{
"epoch": 0.20950846091861403,
"grad_norm": 0.4959644692499422,
"learning_rate": 5e-06,
"loss": 0.4653,
"step": 390
},
{
"epoch": 0.21488047273704003,
"grad_norm": 0.4887248921897578,
"learning_rate": 5e-06,
"loss": 0.4329,
"step": 400
},
{
"epoch": 0.22025248455546603,
"grad_norm": 0.5246025423036973,
"learning_rate": 5e-06,
"loss": 0.4372,
"step": 410
},
{
"epoch": 0.22562449637389204,
"grad_norm": 0.5231208256198969,
"learning_rate": 5e-06,
"loss": 0.4483,
"step": 420
},
{
"epoch": 0.230996508192318,
"grad_norm": 0.4905322052458498,
"learning_rate": 5e-06,
"loss": 0.4468,
"step": 430
},
{
"epoch": 0.23636852001074402,
"grad_norm": 0.4936919209174855,
"learning_rate": 5e-06,
"loss": 0.4233,
"step": 440
},
{
"epoch": 0.24174053182917002,
"grad_norm": 0.48253252781988465,
"learning_rate": 5e-06,
"loss": 0.4282,
"step": 450
},
{
"epoch": 0.24711254364759602,
"grad_norm": 0.5320793975315983,
"learning_rate": 5e-06,
"loss": 0.438,
"step": 460
},
{
"epoch": 0.252484555466022,
"grad_norm": 0.5669630042331335,
"learning_rate": 5e-06,
"loss": 0.4405,
"step": 470
},
{
"epoch": 0.257856567284448,
"grad_norm": 0.5087900877086022,
"learning_rate": 5e-06,
"loss": 0.4249,
"step": 480
},
{
"epoch": 0.263228579102874,
"grad_norm": 0.5450204053078138,
"learning_rate": 5e-06,
"loss": 0.4444,
"step": 490
},
{
"epoch": 0.2686005909213,
"grad_norm": 0.5058928148426334,
"learning_rate": 5e-06,
"loss": 0.4445,
"step": 500
},
{
"epoch": 0.273972602739726,
"grad_norm": 0.5051451139496426,
"learning_rate": 5e-06,
"loss": 0.4425,
"step": 510
},
{
"epoch": 0.279344614558152,
"grad_norm": 0.5457866711066754,
"learning_rate": 5e-06,
"loss": 0.4389,
"step": 520
},
{
"epoch": 0.284716626376578,
"grad_norm": 0.4849183989554118,
"learning_rate": 5e-06,
"loss": 0.4288,
"step": 530
},
{
"epoch": 0.290088638195004,
"grad_norm": 0.5051790388747184,
"learning_rate": 5e-06,
"loss": 0.4375,
"step": 540
},
{
"epoch": 0.29546065001343,
"grad_norm": 0.5147432880032614,
"learning_rate": 5e-06,
"loss": 0.4318,
"step": 550
},
{
"epoch": 0.30083266183185603,
"grad_norm": 0.5151984027160786,
"learning_rate": 5e-06,
"loss": 0.4296,
"step": 560
},
{
"epoch": 0.30620467365028203,
"grad_norm": 0.487287573842156,
"learning_rate": 5e-06,
"loss": 0.4335,
"step": 570
},
{
"epoch": 0.31157668546870804,
"grad_norm": 0.48311459111349353,
"learning_rate": 5e-06,
"loss": 0.4249,
"step": 580
},
{
"epoch": 0.31694869728713404,
"grad_norm": 0.48793456986947,
"learning_rate": 5e-06,
"loss": 0.4284,
"step": 590
},
{
"epoch": 0.32232070910556004,
"grad_norm": 0.52641572654466,
"learning_rate": 5e-06,
"loss": 0.4307,
"step": 600
},
{
"epoch": 0.32769272092398605,
"grad_norm": 0.4882559371631369,
"learning_rate": 5e-06,
"loss": 0.4263,
"step": 610
},
{
"epoch": 0.33306473274241205,
"grad_norm": 0.4840381990603004,
"learning_rate": 5e-06,
"loss": 0.4231,
"step": 620
},
{
"epoch": 0.33843674456083805,
"grad_norm": 0.47841166192032697,
"learning_rate": 5e-06,
"loss": 0.4243,
"step": 630
},
{
"epoch": 0.34380875637926406,
"grad_norm": 0.4974356233716661,
"learning_rate": 5e-06,
"loss": 0.4219,
"step": 640
},
{
"epoch": 0.34918076819769006,
"grad_norm": 0.503552480449389,
"learning_rate": 5e-06,
"loss": 0.4213,
"step": 650
},
{
"epoch": 0.354552780016116,
"grad_norm": 0.48878619440735516,
"learning_rate": 5e-06,
"loss": 0.4242,
"step": 660
},
{
"epoch": 0.359924791834542,
"grad_norm": 0.4976388993737148,
"learning_rate": 5e-06,
"loss": 0.4134,
"step": 670
},
{
"epoch": 0.365296803652968,
"grad_norm": 0.4906593430866946,
"learning_rate": 5e-06,
"loss": 0.4319,
"step": 680
},
{
"epoch": 0.370668815471394,
"grad_norm": 0.48689730508962903,
"learning_rate": 5e-06,
"loss": 0.42,
"step": 690
},
{
"epoch": 0.37604082728982,
"grad_norm": 0.5087526864688174,
"learning_rate": 5e-06,
"loss": 0.4292,
"step": 700
},
{
"epoch": 0.381412839108246,
"grad_norm": 0.4826359097531591,
"learning_rate": 5e-06,
"loss": 0.4309,
"step": 710
},
{
"epoch": 0.38678485092667203,
"grad_norm": 0.5168965208054898,
"learning_rate": 5e-06,
"loss": 0.4292,
"step": 720
},
{
"epoch": 0.39215686274509803,
"grad_norm": 0.5241584710463651,
"learning_rate": 5e-06,
"loss": 0.426,
"step": 730
},
{
"epoch": 0.39752887456352404,
"grad_norm": 0.5037085747835317,
"learning_rate": 5e-06,
"loss": 0.4205,
"step": 740
},
{
"epoch": 0.40290088638195004,
"grad_norm": 0.5257737086254655,
"learning_rate": 5e-06,
"loss": 0.4319,
"step": 750
},
{
"epoch": 0.40827289820037604,
"grad_norm": 0.5053782439759247,
"learning_rate": 5e-06,
"loss": 0.4341,
"step": 760
},
{
"epoch": 0.41364491001880205,
"grad_norm": 0.48695927349136436,
"learning_rate": 5e-06,
"loss": 0.4317,
"step": 770
},
{
"epoch": 0.41901692183722805,
"grad_norm": 0.5223324375752193,
"learning_rate": 5e-06,
"loss": 0.4315,
"step": 780
},
{
"epoch": 0.42438893365565405,
"grad_norm": 0.48347269165618467,
"learning_rate": 5e-06,
"loss": 0.4213,
"step": 790
},
{
"epoch": 0.42976094547408006,
"grad_norm": 0.49513085763985204,
"learning_rate": 5e-06,
"loss": 0.4128,
"step": 800
},
{
"epoch": 0.43513295729250606,
"grad_norm": 0.5064390590783957,
"learning_rate": 5e-06,
"loss": 0.4209,
"step": 810
},
{
"epoch": 0.44050496911093207,
"grad_norm": 0.47087189867674206,
"learning_rate": 5e-06,
"loss": 0.4178,
"step": 820
},
{
"epoch": 0.44587698092935807,
"grad_norm": 0.5037177769830938,
"learning_rate": 5e-06,
"loss": 0.4154,
"step": 830
},
{
"epoch": 0.4512489927477841,
"grad_norm": 0.4836652989743429,
"learning_rate": 5e-06,
"loss": 0.4231,
"step": 840
},
{
"epoch": 0.45662100456621,
"grad_norm": 0.49268225104535635,
"learning_rate": 5e-06,
"loss": 0.4322,
"step": 850
},
{
"epoch": 0.461993016384636,
"grad_norm": 0.5177367203120576,
"learning_rate": 5e-06,
"loss": 0.4148,
"step": 860
},
{
"epoch": 0.467365028203062,
"grad_norm": 0.5287712648467915,
"learning_rate": 5e-06,
"loss": 0.4203,
"step": 870
},
{
"epoch": 0.47273704002148803,
"grad_norm": 0.5261107946601264,
"learning_rate": 5e-06,
"loss": 0.4167,
"step": 880
},
{
"epoch": 0.47810905183991403,
"grad_norm": 0.4771489742949829,
"learning_rate": 5e-06,
"loss": 0.4167,
"step": 890
},
{
"epoch": 0.48348106365834004,
"grad_norm": 0.5102018630117343,
"learning_rate": 5e-06,
"loss": 0.4229,
"step": 900
},
{
"epoch": 0.48885307547676604,
"grad_norm": 0.4878429347643926,
"learning_rate": 5e-06,
"loss": 0.4183,
"step": 910
},
{
"epoch": 0.49422508729519204,
"grad_norm": 0.5144994848348318,
"learning_rate": 5e-06,
"loss": 0.4143,
"step": 920
},
{
"epoch": 0.49959709911361805,
"grad_norm": 0.503185719473255,
"learning_rate": 5e-06,
"loss": 0.4233,
"step": 930
},
{
"epoch": 0.504969110932044,
"grad_norm": 0.4877537965487305,
"learning_rate": 5e-06,
"loss": 0.4111,
"step": 940
},
{
"epoch": 0.51034112275047,
"grad_norm": 0.49337702492530877,
"learning_rate": 5e-06,
"loss": 0.4178,
"step": 950
},
{
"epoch": 0.515713134568896,
"grad_norm": 0.5058601662368037,
"learning_rate": 5e-06,
"loss": 0.4202,
"step": 960
},
{
"epoch": 0.5210851463873221,
"grad_norm": 0.510997345591431,
"learning_rate": 5e-06,
"loss": 0.4105,
"step": 970
},
{
"epoch": 0.526457158205748,
"grad_norm": 0.4859028721289983,
"learning_rate": 5e-06,
"loss": 0.4117,
"step": 980
},
{
"epoch": 0.5318291700241741,
"grad_norm": 0.4779893349116978,
"learning_rate": 5e-06,
"loss": 0.4173,
"step": 990
},
{
"epoch": 0.5372011818426,
"grad_norm": 0.4764031353354688,
"learning_rate": 5e-06,
"loss": 0.425,
"step": 1000
},
{
"epoch": 0.5425731936610261,
"grad_norm": 0.5006562105139756,
"learning_rate": 5e-06,
"loss": 0.4159,
"step": 1010
},
{
"epoch": 0.547945205479452,
"grad_norm": 0.5152775958116914,
"learning_rate": 5e-06,
"loss": 0.4188,
"step": 1020
},
{
"epoch": 0.5533172172978781,
"grad_norm": 0.5289018387715745,
"learning_rate": 5e-06,
"loss": 0.4139,
"step": 1030
},
{
"epoch": 0.558689229116304,
"grad_norm": 0.5011636819261501,
"learning_rate": 5e-06,
"loss": 0.4366,
"step": 1040
},
{
"epoch": 0.5640612409347301,
"grad_norm": 0.5231733582695839,
"learning_rate": 5e-06,
"loss": 0.4151,
"step": 1050
},
{
"epoch": 0.569433252753156,
"grad_norm": 0.5188191978407911,
"learning_rate": 5e-06,
"loss": 0.4228,
"step": 1060
},
{
"epoch": 0.5748052645715821,
"grad_norm": 0.5151903183336549,
"learning_rate": 5e-06,
"loss": 0.415,
"step": 1070
},
{
"epoch": 0.580177276390008,
"grad_norm": 0.4813910302235839,
"learning_rate": 5e-06,
"loss": 0.4096,
"step": 1080
},
{
"epoch": 0.5855492882084341,
"grad_norm": 0.49401253214817437,
"learning_rate": 5e-06,
"loss": 0.423,
"step": 1090
},
{
"epoch": 0.59092130002686,
"grad_norm": 0.47241411535211064,
"learning_rate": 5e-06,
"loss": 0.4269,
"step": 1100
},
{
"epoch": 0.5962933118452861,
"grad_norm": 0.5122976717434686,
"learning_rate": 5e-06,
"loss": 0.4118,
"step": 1110
},
{
"epoch": 0.6016653236637121,
"grad_norm": 0.4830895458893607,
"learning_rate": 5e-06,
"loss": 0.4045,
"step": 1120
},
{
"epoch": 0.607037335482138,
"grad_norm": 0.45858227235019566,
"learning_rate": 5e-06,
"loss": 0.4066,
"step": 1130
},
{
"epoch": 0.6124093473005641,
"grad_norm": 0.4930799485366545,
"learning_rate": 5e-06,
"loss": 0.411,
"step": 1140
},
{
"epoch": 0.61778135911899,
"grad_norm": 0.552477707583534,
"learning_rate": 5e-06,
"loss": 0.419,
"step": 1150
},
{
"epoch": 0.6231533709374161,
"grad_norm": 0.5513588473268155,
"learning_rate": 5e-06,
"loss": 0.4245,
"step": 1160
},
{
"epoch": 0.628525382755842,
"grad_norm": 0.4959903508276244,
"learning_rate": 5e-06,
"loss": 0.4283,
"step": 1170
},
{
"epoch": 0.6338973945742681,
"grad_norm": 0.49870079552977137,
"learning_rate": 5e-06,
"loss": 0.4125,
"step": 1180
},
{
"epoch": 0.639269406392694,
"grad_norm": 0.5203331550271634,
"learning_rate": 5e-06,
"loss": 0.4146,
"step": 1190
},
{
"epoch": 0.6446414182111201,
"grad_norm": 0.5159373689413201,
"learning_rate": 5e-06,
"loss": 0.4164,
"step": 1200
},
{
"epoch": 0.650013430029546,
"grad_norm": 0.48339146055972076,
"learning_rate": 5e-06,
"loss": 0.4174,
"step": 1210
},
{
"epoch": 0.6553854418479721,
"grad_norm": 0.47997830375508066,
"learning_rate": 5e-06,
"loss": 0.4214,
"step": 1220
},
{
"epoch": 0.660757453666398,
"grad_norm": 0.49306877946793926,
"learning_rate": 5e-06,
"loss": 0.4063,
"step": 1230
},
{
"epoch": 0.6661294654848241,
"grad_norm": 0.4688085012843865,
"learning_rate": 5e-06,
"loss": 0.4194,
"step": 1240
},
{
"epoch": 0.67150147730325,
"grad_norm": 0.5065404141632286,
"learning_rate": 5e-06,
"loss": 0.4174,
"step": 1250
},
{
"epoch": 0.6768734891216761,
"grad_norm": 0.4737642817754586,
"learning_rate": 5e-06,
"loss": 0.4088,
"step": 1260
},
{
"epoch": 0.6822455009401021,
"grad_norm": 0.4810016610698945,
"learning_rate": 5e-06,
"loss": 0.4239,
"step": 1270
},
{
"epoch": 0.6876175127585281,
"grad_norm": 0.4808641926976229,
"learning_rate": 5e-06,
"loss": 0.4128,
"step": 1280
},
{
"epoch": 0.6929895245769541,
"grad_norm": 0.5075803008791022,
"learning_rate": 5e-06,
"loss": 0.4184,
"step": 1290
},
{
"epoch": 0.6983615363953801,
"grad_norm": 0.48462584119976126,
"learning_rate": 5e-06,
"loss": 0.4124,
"step": 1300
},
{
"epoch": 0.7037335482138061,
"grad_norm": 0.505895005934085,
"learning_rate": 5e-06,
"loss": 0.4257,
"step": 1310
},
{
"epoch": 0.709105560032232,
"grad_norm": 0.48583666914062235,
"learning_rate": 5e-06,
"loss": 0.4098,
"step": 1320
},
{
"epoch": 0.7144775718506581,
"grad_norm": 0.4970285423060719,
"learning_rate": 5e-06,
"loss": 0.4089,
"step": 1330
},
{
"epoch": 0.719849583669084,
"grad_norm": 0.49766486891278666,
"learning_rate": 5e-06,
"loss": 0.4088,
"step": 1340
},
{
"epoch": 0.7252215954875101,
"grad_norm": 0.4855825885601451,
"learning_rate": 5e-06,
"loss": 0.4025,
"step": 1350
},
{
"epoch": 0.730593607305936,
"grad_norm": 0.5029858180262275,
"learning_rate": 5e-06,
"loss": 0.4141,
"step": 1360
},
{
"epoch": 0.7359656191243621,
"grad_norm": 0.4974844805071485,
"learning_rate": 5e-06,
"loss": 0.4222,
"step": 1370
},
{
"epoch": 0.741337630942788,
"grad_norm": 0.5195759396465923,
"learning_rate": 5e-06,
"loss": 0.4088,
"step": 1380
},
{
"epoch": 0.7467096427612141,
"grad_norm": 0.5192006492882316,
"learning_rate": 5e-06,
"loss": 0.4122,
"step": 1390
},
{
"epoch": 0.75208165457964,
"grad_norm": 0.48436457791566156,
"learning_rate": 5e-06,
"loss": 0.4318,
"step": 1400
},
{
"epoch": 0.7574536663980661,
"grad_norm": 0.5000911567395471,
"learning_rate": 5e-06,
"loss": 0.4097,
"step": 1410
},
{
"epoch": 0.762825678216492,
"grad_norm": 0.49274648962111117,
"learning_rate": 5e-06,
"loss": 0.4135,
"step": 1420
},
{
"epoch": 0.7681976900349181,
"grad_norm": 0.4979579934605875,
"learning_rate": 5e-06,
"loss": 0.4009,
"step": 1430
},
{
"epoch": 0.7735697018533441,
"grad_norm": 0.46471442953412,
"learning_rate": 5e-06,
"loss": 0.4061,
"step": 1440
},
{
"epoch": 0.7789417136717701,
"grad_norm": 0.5116814519358561,
"learning_rate": 5e-06,
"loss": 0.4081,
"step": 1450
},
{
"epoch": 0.7843137254901961,
"grad_norm": 0.47123618031767167,
"learning_rate": 5e-06,
"loss": 0.411,
"step": 1460
},
{
"epoch": 0.7896857373086221,
"grad_norm": 0.5492150909290102,
"learning_rate": 5e-06,
"loss": 0.4107,
"step": 1470
},
{
"epoch": 0.7950577491270481,
"grad_norm": 0.49793785975302846,
"learning_rate": 5e-06,
"loss": 0.4068,
"step": 1480
},
{
"epoch": 0.8004297609454741,
"grad_norm": 0.4975848694312907,
"learning_rate": 5e-06,
"loss": 0.4118,
"step": 1490
},
{
"epoch": 0.8058017727639001,
"grad_norm": 0.5613329259556228,
"learning_rate": 5e-06,
"loss": 0.4053,
"step": 1500
},
{
"epoch": 0.811173784582326,
"grad_norm": 0.49862206188730673,
"learning_rate": 5e-06,
"loss": 0.4085,
"step": 1510
},
{
"epoch": 0.8165457964007521,
"grad_norm": 0.5114041258349515,
"learning_rate": 5e-06,
"loss": 0.4127,
"step": 1520
},
{
"epoch": 0.821917808219178,
"grad_norm": 0.4789962956827193,
"learning_rate": 5e-06,
"loss": 0.4194,
"step": 1530
},
{
"epoch": 0.8272898200376041,
"grad_norm": 0.540275990173432,
"learning_rate": 5e-06,
"loss": 0.4226,
"step": 1540
},
{
"epoch": 0.83266183185603,
"grad_norm": 0.5180207397875711,
"learning_rate": 5e-06,
"loss": 0.4099,
"step": 1550
},
{
"epoch": 0.8380338436744561,
"grad_norm": 0.4847352059935786,
"learning_rate": 5e-06,
"loss": 0.4083,
"step": 1560
},
{
"epoch": 0.843405855492882,
"grad_norm": 0.5611357081200975,
"learning_rate": 5e-06,
"loss": 0.4054,
"step": 1570
},
{
"epoch": 0.8487778673113081,
"grad_norm": 0.527298877110818,
"learning_rate": 5e-06,
"loss": 0.4187,
"step": 1580
},
{
"epoch": 0.8541498791297341,
"grad_norm": 0.5160056675790338,
"learning_rate": 5e-06,
"loss": 0.4085,
"step": 1590
},
{
"epoch": 0.8595218909481601,
"grad_norm": 0.4979226916104278,
"learning_rate": 5e-06,
"loss": 0.4086,
"step": 1600
},
{
"epoch": 0.8648939027665861,
"grad_norm": 0.48561293825487145,
"learning_rate": 5e-06,
"loss": 0.4048,
"step": 1610
},
{
"epoch": 0.8702659145850121,
"grad_norm": 0.5264507223222344,
"learning_rate": 5e-06,
"loss": 0.4025,
"step": 1620
},
{
"epoch": 0.8756379264034381,
"grad_norm": 0.45827475378643506,
"learning_rate": 5e-06,
"loss": 0.4107,
"step": 1630
},
{
"epoch": 0.8810099382218641,
"grad_norm": 0.4598952847745578,
"learning_rate": 5e-06,
"loss": 0.4114,
"step": 1640
},
{
"epoch": 0.8863819500402901,
"grad_norm": 0.4897282955229834,
"learning_rate": 5e-06,
"loss": 0.4088,
"step": 1650
},
{
"epoch": 0.8917539618587161,
"grad_norm": 0.5208215132708872,
"learning_rate": 5e-06,
"loss": 0.3976,
"step": 1660
},
{
"epoch": 0.8971259736771421,
"grad_norm": 0.4798522944815239,
"learning_rate": 5e-06,
"loss": 0.4153,
"step": 1670
},
{
"epoch": 0.9024979854955681,
"grad_norm": 0.4860389540678452,
"learning_rate": 5e-06,
"loss": 0.4192,
"step": 1680
},
{
"epoch": 0.9078699973139941,
"grad_norm": 0.46106648686259355,
"learning_rate": 5e-06,
"loss": 0.4226,
"step": 1690
},
{
"epoch": 0.91324200913242,
"grad_norm": 0.49535127626693193,
"learning_rate": 5e-06,
"loss": 0.404,
"step": 1700
},
{
"epoch": 0.9186140209508461,
"grad_norm": 0.4543561154266515,
"learning_rate": 5e-06,
"loss": 0.4126,
"step": 1710
},
{
"epoch": 0.923986032769272,
"grad_norm": 0.5200028022866499,
"learning_rate": 5e-06,
"loss": 0.4072,
"step": 1720
},
{
"epoch": 0.9293580445876981,
"grad_norm": 0.5114668126414205,
"learning_rate": 5e-06,
"loss": 0.4037,
"step": 1730
},
{
"epoch": 0.934730056406124,
"grad_norm": 0.5165691248841157,
"learning_rate": 5e-06,
"loss": 0.4129,
"step": 1740
},
{
"epoch": 0.9401020682245501,
"grad_norm": 0.5191922236594055,
"learning_rate": 5e-06,
"loss": 0.4089,
"step": 1750
},
{
"epoch": 0.9454740800429761,
"grad_norm": 0.48259570729073553,
"learning_rate": 5e-06,
"loss": 0.4031,
"step": 1760
},
{
"epoch": 0.9508460918614021,
"grad_norm": 0.518843327631672,
"learning_rate": 5e-06,
"loss": 0.398,
"step": 1770
},
{
"epoch": 0.9562181036798281,
"grad_norm": 0.5364105770372746,
"learning_rate": 5e-06,
"loss": 0.3988,
"step": 1780
},
{
"epoch": 0.9615901154982541,
"grad_norm": 0.5020255273268895,
"learning_rate": 5e-06,
"loss": 0.4076,
"step": 1790
},
{
"epoch": 0.9669621273166801,
"grad_norm": 0.5073523384433892,
"learning_rate": 5e-06,
"loss": 0.4061,
"step": 1800
},
{
"epoch": 0.9723341391351061,
"grad_norm": 0.4727913677316271,
"learning_rate": 5e-06,
"loss": 0.4014,
"step": 1810
},
{
"epoch": 0.9777061509535321,
"grad_norm": 0.5050437658596597,
"learning_rate": 5e-06,
"loss": 0.4166,
"step": 1820
},
{
"epoch": 0.9830781627719581,
"grad_norm": 0.4907596747431724,
"learning_rate": 5e-06,
"loss": 0.4177,
"step": 1830
},
{
"epoch": 0.9884501745903841,
"grad_norm": 0.4688237071483825,
"learning_rate": 5e-06,
"loss": 0.402,
"step": 1840
},
{
"epoch": 0.9938221864088101,
"grad_norm": 0.4643104668565543,
"learning_rate": 5e-06,
"loss": 0.4026,
"step": 1850
},
{
"epoch": 0.9991941982272361,
"grad_norm": 0.49500486010624006,
"learning_rate": 5e-06,
"loss": 0.4024,
"step": 1860
},
{
"epoch": 0.9997313994090787,
"eval_loss": 0.4042987525463104,
"eval_runtime": 471.547,
"eval_samples_per_second": 26.593,
"eval_steps_per_second": 0.416,
"step": 1861
},
{
"epoch": 1.004566210045662,
"grad_norm": 0.4895276245416368,
"learning_rate": 5e-06,
"loss": 0.3782,
"step": 1870
},
{
"epoch": 1.009938221864088,
"grad_norm": 0.4763690228214777,
"learning_rate": 5e-06,
"loss": 0.3515,
"step": 1880
},
{
"epoch": 1.0153102336825142,
"grad_norm": 0.478887231667028,
"learning_rate": 5e-06,
"loss": 0.3602,
"step": 1890
},
{
"epoch": 1.02068224550094,
"grad_norm": 0.4406116481047118,
"learning_rate": 5e-06,
"loss": 0.3583,
"step": 1900
},
{
"epoch": 1.026054257319366,
"grad_norm": 0.45230055522726487,
"learning_rate": 5e-06,
"loss": 0.3428,
"step": 1910
},
{
"epoch": 1.031426269137792,
"grad_norm": 0.47360183462868854,
"learning_rate": 5e-06,
"loss": 0.3615,
"step": 1920
},
{
"epoch": 1.0367982809562182,
"grad_norm": 0.4565501019154625,
"learning_rate": 5e-06,
"loss": 0.3629,
"step": 1930
},
{
"epoch": 1.0421702927746441,
"grad_norm": 0.4980427239755443,
"learning_rate": 5e-06,
"loss": 0.3695,
"step": 1940
},
{
"epoch": 1.04754230459307,
"grad_norm": 0.48230806698810685,
"learning_rate": 5e-06,
"loss": 0.3434,
"step": 1950
},
{
"epoch": 1.052914316411496,
"grad_norm": 0.470470012779878,
"learning_rate": 5e-06,
"loss": 0.3509,
"step": 1960
},
{
"epoch": 1.0582863282299222,
"grad_norm": 0.4479895760634986,
"learning_rate": 5e-06,
"loss": 0.3424,
"step": 1970
},
{
"epoch": 1.0636583400483481,
"grad_norm": 0.4719779566218419,
"learning_rate": 5e-06,
"loss": 0.3543,
"step": 1980
},
{
"epoch": 1.069030351866774,
"grad_norm": 0.48397615308885894,
"learning_rate": 5e-06,
"loss": 0.3514,
"step": 1990
},
{
"epoch": 1.0744023636852,
"grad_norm": 0.43678232465571554,
"learning_rate": 5e-06,
"loss": 0.3441,
"step": 2000
},
{
"epoch": 1.0797743755036262,
"grad_norm": 0.44851060997131426,
"learning_rate": 5e-06,
"loss": 0.3519,
"step": 2010
},
{
"epoch": 1.0851463873220522,
"grad_norm": 0.4661433912641436,
"learning_rate": 5e-06,
"loss": 0.3501,
"step": 2020
},
{
"epoch": 1.090518399140478,
"grad_norm": 0.5116441181752102,
"learning_rate": 5e-06,
"loss": 0.3583,
"step": 2030
},
{
"epoch": 1.095890410958904,
"grad_norm": 0.48794994234515765,
"learning_rate": 5e-06,
"loss": 0.3503,
"step": 2040
},
{
"epoch": 1.1012624227773302,
"grad_norm": 0.47447458677616894,
"learning_rate": 5e-06,
"loss": 0.3563,
"step": 2050
},
{
"epoch": 1.1066344345957562,
"grad_norm": 0.4781262245860992,
"learning_rate": 5e-06,
"loss": 0.3605,
"step": 2060
},
{
"epoch": 1.1120064464141821,
"grad_norm": 0.4697691917656968,
"learning_rate": 5e-06,
"loss": 0.3576,
"step": 2070
},
{
"epoch": 1.117378458232608,
"grad_norm": 0.4418186659255878,
"learning_rate": 5e-06,
"loss": 0.3581,
"step": 2080
},
{
"epoch": 1.122750470051034,
"grad_norm": 0.4525720645706372,
"learning_rate": 5e-06,
"loss": 0.3482,
"step": 2090
},
{
"epoch": 1.1281224818694602,
"grad_norm": 0.47164729198556593,
"learning_rate": 5e-06,
"loss": 0.3518,
"step": 2100
},
{
"epoch": 1.1334944936878861,
"grad_norm": 0.4780362991120794,
"learning_rate": 5e-06,
"loss": 0.3605,
"step": 2110
},
{
"epoch": 1.138866505506312,
"grad_norm": 0.45101508118677897,
"learning_rate": 5e-06,
"loss": 0.3539,
"step": 2120
},
{
"epoch": 1.144238517324738,
"grad_norm": 0.46243427496308614,
"learning_rate": 5e-06,
"loss": 0.3471,
"step": 2130
},
{
"epoch": 1.1496105291431642,
"grad_norm": 0.45895312831755847,
"learning_rate": 5e-06,
"loss": 0.3494,
"step": 2140
},
{
"epoch": 1.1549825409615901,
"grad_norm": 0.46425226076163784,
"learning_rate": 5e-06,
"loss": 0.3426,
"step": 2150
},
{
"epoch": 1.160354552780016,
"grad_norm": 0.5051239504730862,
"learning_rate": 5e-06,
"loss": 0.356,
"step": 2160
},
{
"epoch": 1.165726564598442,
"grad_norm": 0.4560291857740493,
"learning_rate": 5e-06,
"loss": 0.3436,
"step": 2170
},
{
"epoch": 1.1710985764168682,
"grad_norm": 0.46586472412952123,
"learning_rate": 5e-06,
"loss": 0.3554,
"step": 2180
},
{
"epoch": 1.1764705882352942,
"grad_norm": 0.47463013503202,
"learning_rate": 5e-06,
"loss": 0.3549,
"step": 2190
},
{
"epoch": 1.18184260005372,
"grad_norm": 0.48738878390795715,
"learning_rate": 5e-06,
"loss": 0.3616,
"step": 2200
},
{
"epoch": 1.187214611872146,
"grad_norm": 0.48890429518366846,
"learning_rate": 5e-06,
"loss": 0.3456,
"step": 2210
},
{
"epoch": 1.192586623690572,
"grad_norm": 0.47045438323568006,
"learning_rate": 5e-06,
"loss": 0.3575,
"step": 2220
},
{
"epoch": 1.1979586355089982,
"grad_norm": 0.46643562303520814,
"learning_rate": 5e-06,
"loss": 0.3572,
"step": 2230
},
{
"epoch": 1.2033306473274241,
"grad_norm": 0.4711541973427813,
"learning_rate": 5e-06,
"loss": 0.3463,
"step": 2240
},
{
"epoch": 1.20870265914585,
"grad_norm": 0.48043545998856163,
"learning_rate": 5e-06,
"loss": 0.3509,
"step": 2250
},
{
"epoch": 1.2140746709642762,
"grad_norm": 0.48163337441004483,
"learning_rate": 5e-06,
"loss": 0.3607,
"step": 2260
},
{
"epoch": 1.2194466827827022,
"grad_norm": 0.4736587577149829,
"learning_rate": 5e-06,
"loss": 0.3663,
"step": 2270
},
{
"epoch": 1.2248186946011281,
"grad_norm": 0.5118805431884204,
"learning_rate": 5e-06,
"loss": 0.3608,
"step": 2280
},
{
"epoch": 1.230190706419554,
"grad_norm": 0.48267765503443366,
"learning_rate": 5e-06,
"loss": 0.3574,
"step": 2290
},
{
"epoch": 1.23556271823798,
"grad_norm": 0.4487058222257102,
"learning_rate": 5e-06,
"loss": 0.3493,
"step": 2300
},
{
"epoch": 1.2409347300564062,
"grad_norm": 0.47007542745542785,
"learning_rate": 5e-06,
"loss": 0.3514,
"step": 2310
},
{
"epoch": 1.2463067418748321,
"grad_norm": 0.4757736251978376,
"learning_rate": 5e-06,
"loss": 0.3804,
"step": 2320
},
{
"epoch": 1.251678753693258,
"grad_norm": 0.48922728114483144,
"learning_rate": 5e-06,
"loss": 0.3558,
"step": 2330
},
{
"epoch": 1.257050765511684,
"grad_norm": 0.4982569914182343,
"learning_rate": 5e-06,
"loss": 0.355,
"step": 2340
},
{
"epoch": 1.26242277733011,
"grad_norm": 0.4793876371419449,
"learning_rate": 5e-06,
"loss": 0.3561,
"step": 2350
},
{
"epoch": 1.2677947891485362,
"grad_norm": 0.46545152398636225,
"learning_rate": 5e-06,
"loss": 0.3472,
"step": 2360
},
{
"epoch": 1.273166800966962,
"grad_norm": 0.4967452966425127,
"learning_rate": 5e-06,
"loss": 0.3523,
"step": 2370
},
{
"epoch": 1.278538812785388,
"grad_norm": 0.4771247760513103,
"learning_rate": 5e-06,
"loss": 0.3665,
"step": 2380
},
{
"epoch": 1.2839108246038142,
"grad_norm": 0.4958016938945244,
"learning_rate": 5e-06,
"loss": 0.3533,
"step": 2390
},
{
"epoch": 1.2892828364222402,
"grad_norm": 0.5105520260750973,
"learning_rate": 5e-06,
"loss": 0.3644,
"step": 2400
},
{
"epoch": 1.2946548482406661,
"grad_norm": 0.4536892978464728,
"learning_rate": 5e-06,
"loss": 0.358,
"step": 2410
},
{
"epoch": 1.300026860059092,
"grad_norm": 0.46265183496455864,
"learning_rate": 5e-06,
"loss": 0.3718,
"step": 2420
},
{
"epoch": 1.305398871877518,
"grad_norm": 0.4625786404713504,
"learning_rate": 5e-06,
"loss": 0.3498,
"step": 2430
},
{
"epoch": 1.3107708836959442,
"grad_norm": 0.4487573165277923,
"learning_rate": 5e-06,
"loss": 0.3453,
"step": 2440
},
{
"epoch": 1.3161428955143701,
"grad_norm": 0.48228276168484907,
"learning_rate": 5e-06,
"loss": 0.3574,
"step": 2450
},
{
"epoch": 1.321514907332796,
"grad_norm": 0.45803164032303556,
"learning_rate": 5e-06,
"loss": 0.3585,
"step": 2460
},
{
"epoch": 1.3268869191512223,
"grad_norm": 0.4961528659344489,
"learning_rate": 5e-06,
"loss": 0.3447,
"step": 2470
},
{
"epoch": 1.3322589309696482,
"grad_norm": 0.46995905220607553,
"learning_rate": 5e-06,
"loss": 0.3572,
"step": 2480
},
{
"epoch": 1.3376309427880742,
"grad_norm": 0.4735815907276366,
"learning_rate": 5e-06,
"loss": 0.3568,
"step": 2490
},
{
"epoch": 1.3430029546065,
"grad_norm": 0.45401301977667363,
"learning_rate": 5e-06,
"loss": 0.3557,
"step": 2500
},
{
"epoch": 1.348374966424926,
"grad_norm": 0.442403966397837,
"learning_rate": 5e-06,
"loss": 0.3539,
"step": 2510
},
{
"epoch": 1.3537469782433522,
"grad_norm": 0.5002564542847203,
"learning_rate": 5e-06,
"loss": 0.3527,
"step": 2520
},
{
"epoch": 1.3591189900617782,
"grad_norm": 0.4645813178315932,
"learning_rate": 5e-06,
"loss": 0.3601,
"step": 2530
},
{
"epoch": 1.3644910018802041,
"grad_norm": 0.47453517344411733,
"learning_rate": 5e-06,
"loss": 0.3615,
"step": 2540
},
{
"epoch": 1.36986301369863,
"grad_norm": 0.4604202509493409,
"learning_rate": 5e-06,
"loss": 0.3529,
"step": 2550
},
{
"epoch": 1.375235025517056,
"grad_norm": 0.453130113647468,
"learning_rate": 5e-06,
"loss": 0.3444,
"step": 2560
},
{
"epoch": 1.3806070373354822,
"grad_norm": 0.44091309480166135,
"learning_rate": 5e-06,
"loss": 0.3625,
"step": 2570
},
{
"epoch": 1.3859790491539081,
"grad_norm": 0.4517474265184944,
"learning_rate": 5e-06,
"loss": 0.3634,
"step": 2580
},
{
"epoch": 1.391351060972334,
"grad_norm": 0.4709274208742667,
"learning_rate": 5e-06,
"loss": 0.3576,
"step": 2590
},
{
"epoch": 1.3967230727907602,
"grad_norm": 0.45649673901571647,
"learning_rate": 5e-06,
"loss": 0.3552,
"step": 2600
},
{
"epoch": 1.4020950846091862,
"grad_norm": 0.4816316434749545,
"learning_rate": 5e-06,
"loss": 0.3674,
"step": 2610
},
{
"epoch": 1.4074670964276121,
"grad_norm": 0.4509867756276843,
"learning_rate": 5e-06,
"loss": 0.3627,
"step": 2620
},
{
"epoch": 1.412839108246038,
"grad_norm": 0.47322108910920707,
"learning_rate": 5e-06,
"loss": 0.3561,
"step": 2630
},
{
"epoch": 1.418211120064464,
"grad_norm": 0.4568009857383218,
"learning_rate": 5e-06,
"loss": 0.3548,
"step": 2640
},
{
"epoch": 1.4235831318828902,
"grad_norm": 0.4870747914342431,
"learning_rate": 5e-06,
"loss": 0.365,
"step": 2650
},
{
"epoch": 1.4289551437013162,
"grad_norm": 0.4766156244040559,
"learning_rate": 5e-06,
"loss": 0.3612,
"step": 2660
},
{
"epoch": 1.434327155519742,
"grad_norm": 0.4741487895978261,
"learning_rate": 5e-06,
"loss": 0.3552,
"step": 2670
},
{
"epoch": 1.4396991673381683,
"grad_norm": 0.45805481522590397,
"learning_rate": 5e-06,
"loss": 0.3712,
"step": 2680
},
{
"epoch": 1.4450711791565942,
"grad_norm": 0.48703760968615994,
"learning_rate": 5e-06,
"loss": 0.3555,
"step": 2690
},
{
"epoch": 1.4504431909750202,
"grad_norm": 0.476047193063988,
"learning_rate": 5e-06,
"loss": 0.3526,
"step": 2700
},
{
"epoch": 1.4558152027934461,
"grad_norm": 0.4501144882135316,
"learning_rate": 5e-06,
"loss": 0.3509,
"step": 2710
},
{
"epoch": 1.461187214611872,
"grad_norm": 0.4807022793546781,
"learning_rate": 5e-06,
"loss": 0.3587,
"step": 2720
},
{
"epoch": 1.4665592264302982,
"grad_norm": 0.4866933520193109,
"learning_rate": 5e-06,
"loss": 0.3541,
"step": 2730
},
{
"epoch": 1.4719312382487242,
"grad_norm": 0.4692699336726864,
"learning_rate": 5e-06,
"loss": 0.3562,
"step": 2740
},
{
"epoch": 1.4773032500671501,
"grad_norm": 0.47926277815969454,
"learning_rate": 5e-06,
"loss": 0.346,
"step": 2750
},
{
"epoch": 1.482675261885576,
"grad_norm": 0.4799455318722434,
"learning_rate": 5e-06,
"loss": 0.3502,
"step": 2760
},
{
"epoch": 1.488047273704002,
"grad_norm": 0.46636973311006896,
"learning_rate": 5e-06,
"loss": 0.3508,
"step": 2770
},
{
"epoch": 1.4934192855224282,
"grad_norm": 0.45313120266777845,
"learning_rate": 5e-06,
"loss": 0.35,
"step": 2780
},
{
"epoch": 1.4987912973408541,
"grad_norm": 0.49984202987539106,
"learning_rate": 5e-06,
"loss": 0.3531,
"step": 2790
},
{
"epoch": 1.50416330915928,
"grad_norm": 0.5107078124846389,
"learning_rate": 5e-06,
"loss": 0.3554,
"step": 2800
},
{
"epoch": 1.5095353209777063,
"grad_norm": 0.49609121410014656,
"learning_rate": 5e-06,
"loss": 0.3597,
"step": 2810
},
{
"epoch": 1.514907332796132,
"grad_norm": 0.44000526898502273,
"learning_rate": 5e-06,
"loss": 0.3488,
"step": 2820
},
{
"epoch": 1.5202793446145582,
"grad_norm": 0.48110675970521016,
"learning_rate": 5e-06,
"loss": 0.3571,
"step": 2830
},
{
"epoch": 1.525651356432984,
"grad_norm": 0.44724848382400667,
"learning_rate": 5e-06,
"loss": 0.347,
"step": 2840
},
{
"epoch": 1.53102336825141,
"grad_norm": 0.4634129922845378,
"learning_rate": 5e-06,
"loss": 0.3636,
"step": 2850
},
{
"epoch": 1.5363953800698362,
"grad_norm": 0.4665778548932393,
"learning_rate": 5e-06,
"loss": 0.3589,
"step": 2860
},
{
"epoch": 1.5417673918882622,
"grad_norm": 0.48219157934372814,
"learning_rate": 5e-06,
"loss": 0.3448,
"step": 2870
},
{
"epoch": 1.5471394037066881,
"grad_norm": 0.47689653413221034,
"learning_rate": 5e-06,
"loss": 0.3568,
"step": 2880
},
{
"epoch": 1.5525114155251143,
"grad_norm": 0.4817413690543133,
"learning_rate": 5e-06,
"loss": 0.3554,
"step": 2890
},
{
"epoch": 1.55788342734354,
"grad_norm": 0.48101456064388537,
"learning_rate": 5e-06,
"loss": 0.3679,
"step": 2900
},
{
"epoch": 1.5632554391619662,
"grad_norm": 0.4610166853876066,
"learning_rate": 5e-06,
"loss": 0.369,
"step": 2910
},
{
"epoch": 1.5686274509803921,
"grad_norm": 0.46324728897514716,
"learning_rate": 5e-06,
"loss": 0.3557,
"step": 2920
},
{
"epoch": 1.573999462798818,
"grad_norm": 0.4792758625068545,
"learning_rate": 5e-06,
"loss": 0.3564,
"step": 2930
},
{
"epoch": 1.5793714746172443,
"grad_norm": 0.4760438491041952,
"learning_rate": 5e-06,
"loss": 0.3516,
"step": 2940
},
{
"epoch": 1.5847434864356702,
"grad_norm": 0.4768587833700611,
"learning_rate": 5e-06,
"loss": 0.3736,
"step": 2950
},
{
"epoch": 1.5901154982540961,
"grad_norm": 0.4702173398495168,
"learning_rate": 5e-06,
"loss": 0.3588,
"step": 2960
},
{
"epoch": 1.5954875100725223,
"grad_norm": 0.5073117917455773,
"learning_rate": 5e-06,
"loss": 0.3576,
"step": 2970
},
{
"epoch": 1.600859521890948,
"grad_norm": 0.4735397279022496,
"learning_rate": 5e-06,
"loss": 0.3518,
"step": 2980
},
{
"epoch": 1.6062315337093742,
"grad_norm": 0.49501392208596307,
"learning_rate": 5e-06,
"loss": 0.3574,
"step": 2990
},
{
"epoch": 1.6116035455278002,
"grad_norm": 0.45154332778333167,
"learning_rate": 5e-06,
"loss": 0.3469,
"step": 3000
},
{
"epoch": 1.6169755573462261,
"grad_norm": 0.4821594714198928,
"learning_rate": 5e-06,
"loss": 0.3665,
"step": 3010
},
{
"epoch": 1.6223475691646523,
"grad_norm": 0.4834902526253989,
"learning_rate": 5e-06,
"loss": 0.3627,
"step": 3020
},
{
"epoch": 1.627719580983078,
"grad_norm": 0.5010431988707854,
"learning_rate": 5e-06,
"loss": 0.3481,
"step": 3030
},
{
"epoch": 1.6330915928015042,
"grad_norm": 0.4842957701772517,
"learning_rate": 5e-06,
"loss": 0.3494,
"step": 3040
},
{
"epoch": 1.6384636046199301,
"grad_norm": 0.42862130200724274,
"learning_rate": 5e-06,
"loss": 0.3477,
"step": 3050
},
{
"epoch": 1.643835616438356,
"grad_norm": 0.45715763363239004,
"learning_rate": 5e-06,
"loss": 0.3605,
"step": 3060
},
{
"epoch": 1.6492076282567822,
"grad_norm": 0.48270644830974097,
"learning_rate": 5e-06,
"loss": 0.3552,
"step": 3070
},
{
"epoch": 1.6545796400752082,
"grad_norm": 0.48420910087305813,
"learning_rate": 5e-06,
"loss": 0.357,
"step": 3080
},
{
"epoch": 1.6599516518936341,
"grad_norm": 0.4633746055111087,
"learning_rate": 5e-06,
"loss": 0.3481,
"step": 3090
},
{
"epoch": 1.6653236637120603,
"grad_norm": 0.46936521480607657,
"learning_rate": 5e-06,
"loss": 0.3455,
"step": 3100
},
{
"epoch": 1.670695675530486,
"grad_norm": 0.46328974064931927,
"learning_rate": 5e-06,
"loss": 0.3539,
"step": 3110
},
{
"epoch": 1.6760676873489122,
"grad_norm": 0.4649283617312367,
"learning_rate": 5e-06,
"loss": 0.3589,
"step": 3120
},
{
"epoch": 1.6814396991673382,
"grad_norm": 0.4703978382827451,
"learning_rate": 5e-06,
"loss": 0.3436,
"step": 3130
},
{
"epoch": 1.686811710985764,
"grad_norm": 0.47847168406654755,
"learning_rate": 5e-06,
"loss": 0.37,
"step": 3140
},
{
"epoch": 1.6921837228041903,
"grad_norm": 0.4778436891961714,
"learning_rate": 5e-06,
"loss": 0.3488,
"step": 3150
},
{
"epoch": 1.6975557346226162,
"grad_norm": 0.44977599188705814,
"learning_rate": 5e-06,
"loss": 0.3571,
"step": 3160
},
{
"epoch": 1.7029277464410422,
"grad_norm": 0.5011217127797689,
"learning_rate": 5e-06,
"loss": 0.3524,
"step": 3170
},
{
"epoch": 1.7082997582594683,
"grad_norm": 0.4686231916011371,
"learning_rate": 5e-06,
"loss": 0.352,
"step": 3180
},
{
"epoch": 1.713671770077894,
"grad_norm": 0.4567419474384379,
"learning_rate": 5e-06,
"loss": 0.3446,
"step": 3190
},
{
"epoch": 1.7190437818963202,
"grad_norm": 0.4920146270483708,
"learning_rate": 5e-06,
"loss": 0.3453,
"step": 3200
},
{
"epoch": 1.7244157937147462,
"grad_norm": 0.4714208322948418,
"learning_rate": 5e-06,
"loss": 0.3592,
"step": 3210
},
{
"epoch": 1.7297878055331721,
"grad_norm": 0.5151502729310895,
"learning_rate": 5e-06,
"loss": 0.3621,
"step": 3220
},
{
"epoch": 1.7351598173515983,
"grad_norm": 0.47377609015255395,
"learning_rate": 5e-06,
"loss": 0.3521,
"step": 3230
},
{
"epoch": 1.740531829170024,
"grad_norm": 0.4896598896220437,
"learning_rate": 5e-06,
"loss": 0.3556,
"step": 3240
},
{
"epoch": 1.7459038409884502,
"grad_norm": 0.4626924876759191,
"learning_rate": 5e-06,
"loss": 0.3624,
"step": 3250
},
{
"epoch": 1.7512758528068761,
"grad_norm": 0.4893799058510227,
"learning_rate": 5e-06,
"loss": 0.355,
"step": 3260
},
{
"epoch": 1.756647864625302,
"grad_norm": 0.4732147945130505,
"learning_rate": 5e-06,
"loss": 0.3584,
"step": 3270
},
{
"epoch": 1.7620198764437283,
"grad_norm": 0.46485063144719596,
"learning_rate": 5e-06,
"loss": 0.3562,
"step": 3280
},
{
"epoch": 1.7673918882621542,
"grad_norm": 0.49823480469474907,
"learning_rate": 5e-06,
"loss": 0.3596,
"step": 3290
},
{
"epoch": 1.7727639000805802,
"grad_norm": 0.46620308535633026,
"learning_rate": 5e-06,
"loss": 0.3492,
"step": 3300
},
{
"epoch": 1.7781359118990063,
"grad_norm": 0.47634133927314953,
"learning_rate": 5e-06,
"loss": 0.3445,
"step": 3310
},
{
"epoch": 1.783507923717432,
"grad_norm": 0.4668848160515941,
"learning_rate": 5e-06,
"loss": 0.3571,
"step": 3320
},
{
"epoch": 1.7888799355358582,
"grad_norm": 0.4585027323046419,
"learning_rate": 5e-06,
"loss": 0.3552,
"step": 3330
},
{
"epoch": 1.7942519473542842,
"grad_norm": 0.5084050187589441,
"learning_rate": 5e-06,
"loss": 0.3595,
"step": 3340
},
{
"epoch": 1.7996239591727101,
"grad_norm": 0.4653354008421696,
"learning_rate": 5e-06,
"loss": 0.3513,
"step": 3350
},
{
"epoch": 1.8049959709911363,
"grad_norm": 0.4856936785111354,
"learning_rate": 5e-06,
"loss": 0.3584,
"step": 3360
},
{
"epoch": 1.810367982809562,
"grad_norm": 0.4595385848192909,
"learning_rate": 5e-06,
"loss": 0.3691,
"step": 3370
},
{
"epoch": 1.8157399946279882,
"grad_norm": 0.5004383593079554,
"learning_rate": 5e-06,
"loss": 0.3582,
"step": 3380
},
{
"epoch": 1.8211120064464141,
"grad_norm": 0.4854729302325183,
"learning_rate": 5e-06,
"loss": 0.3469,
"step": 3390
},
{
"epoch": 1.82648401826484,
"grad_norm": 0.4643207432920559,
"learning_rate": 5e-06,
"loss": 0.3593,
"step": 3400
},
{
"epoch": 1.8318560300832663,
"grad_norm": 0.4665815903863702,
"learning_rate": 5e-06,
"loss": 0.3531,
"step": 3410
},
{
"epoch": 1.8372280419016922,
"grad_norm": 0.48774803768728303,
"learning_rate": 5e-06,
"loss": 0.3512,
"step": 3420
},
{
"epoch": 1.8426000537201181,
"grad_norm": 0.4709529294030036,
"learning_rate": 5e-06,
"loss": 0.3535,
"step": 3430
},
{
"epoch": 1.8479720655385443,
"grad_norm": 0.46736498303364754,
"learning_rate": 5e-06,
"loss": 0.3646,
"step": 3440
},
{
"epoch": 1.85334407735697,
"grad_norm": 0.488258930586327,
"learning_rate": 5e-06,
"loss": 0.3521,
"step": 3450
},
{
"epoch": 1.8587160891753962,
"grad_norm": 0.4700766855772819,
"learning_rate": 5e-06,
"loss": 0.3453,
"step": 3460
},
{
"epoch": 1.8640881009938222,
"grad_norm": 0.46996238530410667,
"learning_rate": 5e-06,
"loss": 0.3642,
"step": 3470
},
{
"epoch": 1.869460112812248,
"grad_norm": 0.4254367637586878,
"learning_rate": 5e-06,
"loss": 0.3573,
"step": 3480
},
{
"epoch": 1.8748321246306743,
"grad_norm": 0.4908636458552527,
"learning_rate": 5e-06,
"loss": 0.3638,
"step": 3490
},
{
"epoch": 1.8802041364491002,
"grad_norm": 0.4801639417628621,
"learning_rate": 5e-06,
"loss": 0.3497,
"step": 3500
},
{
"epoch": 1.8855761482675262,
"grad_norm": 0.45396361000408497,
"learning_rate": 5e-06,
"loss": 0.357,
"step": 3510
},
{
"epoch": 1.8909481600859523,
"grad_norm": 0.4632915688904155,
"learning_rate": 5e-06,
"loss": 0.3583,
"step": 3520
},
{
"epoch": 1.896320171904378,
"grad_norm": 0.491440659929986,
"learning_rate": 5e-06,
"loss": 0.3492,
"step": 3530
},
{
"epoch": 1.9016921837228042,
"grad_norm": 0.4891492614008297,
"learning_rate": 5e-06,
"loss": 0.3526,
"step": 3540
},
{
"epoch": 1.9070641955412302,
"grad_norm": 0.5171252758216774,
"learning_rate": 5e-06,
"loss": 0.3631,
"step": 3550
},
{
"epoch": 1.9124362073596561,
"grad_norm": 0.4461868765641311,
"learning_rate": 5e-06,
"loss": 0.3523,
"step": 3560
},
{
"epoch": 1.9178082191780823,
"grad_norm": 0.48770468521805,
"learning_rate": 5e-06,
"loss": 0.3594,
"step": 3570
},
{
"epoch": 1.923180230996508,
"grad_norm": 0.5105974923297489,
"learning_rate": 5e-06,
"loss": 0.3554,
"step": 3580
},
{
"epoch": 1.9285522428149342,
"grad_norm": 0.4457466990354959,
"learning_rate": 5e-06,
"loss": 0.3507,
"step": 3590
},
{
"epoch": 1.9339242546333602,
"grad_norm": 0.47606633626915407,
"learning_rate": 5e-06,
"loss": 0.3694,
"step": 3600
},
{
"epoch": 1.939296266451786,
"grad_norm": 0.4484067546170701,
"learning_rate": 5e-06,
"loss": 0.3629,
"step": 3610
},
{
"epoch": 1.9446682782702123,
"grad_norm": 0.461791387706203,
"learning_rate": 5e-06,
"loss": 0.3526,
"step": 3620
},
{
"epoch": 1.9500402900886382,
"grad_norm": 0.44496474942160236,
"learning_rate": 5e-06,
"loss": 0.3401,
"step": 3630
},
{
"epoch": 1.9554123019070642,
"grad_norm": 0.5163785217403414,
"learning_rate": 5e-06,
"loss": 0.342,
"step": 3640
},
{
"epoch": 1.9607843137254903,
"grad_norm": 0.47380717094158326,
"learning_rate": 5e-06,
"loss": 0.3543,
"step": 3650
},
{
"epoch": 1.966156325543916,
"grad_norm": 0.46954718282044644,
"learning_rate": 5e-06,
"loss": 0.3705,
"step": 3660
},
{
"epoch": 1.9715283373623422,
"grad_norm": 0.4906972194683541,
"learning_rate": 5e-06,
"loss": 0.355,
"step": 3670
},
{
"epoch": 1.9769003491807682,
"grad_norm": 0.4665996203057171,
"learning_rate": 5e-06,
"loss": 0.3546,
"step": 3680
},
{
"epoch": 1.9822723609991941,
"grad_norm": 0.4511962208632972,
"learning_rate": 5e-06,
"loss": 0.3518,
"step": 3690
},
{
"epoch": 1.9876443728176203,
"grad_norm": 0.5129655066415513,
"learning_rate": 5e-06,
"loss": 0.3664,
"step": 3700
},
{
"epoch": 1.9930163846360462,
"grad_norm": 0.4858845690258609,
"learning_rate": 5e-06,
"loss": 0.3609,
"step": 3710
},
{
"epoch": 1.9983883964544722,
"grad_norm": 0.4870693370290478,
"learning_rate": 5e-06,
"loss": 0.3514,
"step": 3720
},
{
"epoch": 2.0,
"eval_loss": 0.39740535616874695,
"eval_runtime": 468.5266,
"eval_samples_per_second": 26.765,
"eval_steps_per_second": 0.418,
"step": 3723
},
{
"epoch": 2.0037604082728984,
"grad_norm": 0.5148318273487177,
"learning_rate": 5e-06,
"loss": 0.3332,
"step": 3730
},
{
"epoch": 2.009132420091324,
"grad_norm": 0.44757575228737206,
"learning_rate": 5e-06,
"loss": 0.3166,
"step": 3740
},
{
"epoch": 2.0145044319097503,
"grad_norm": 0.49417525446020677,
"learning_rate": 5e-06,
"loss": 0.2932,
"step": 3750
},
{
"epoch": 2.019876443728176,
"grad_norm": 0.45003732267652785,
"learning_rate": 5e-06,
"loss": 0.2966,
"step": 3760
},
{
"epoch": 2.025248455546602,
"grad_norm": 0.4446601372854061,
"learning_rate": 5e-06,
"loss": 0.3013,
"step": 3770
},
{
"epoch": 2.0306204673650283,
"grad_norm": 0.4520286680577658,
"learning_rate": 5e-06,
"loss": 0.287,
"step": 3780
},
{
"epoch": 2.035992479183454,
"grad_norm": 0.45274387988815035,
"learning_rate": 5e-06,
"loss": 0.301,
"step": 3790
},
{
"epoch": 2.04136449100188,
"grad_norm": 0.47200858969384085,
"learning_rate": 5e-06,
"loss": 0.2955,
"step": 3800
},
{
"epoch": 2.0467365028203064,
"grad_norm": 0.44150919063824945,
"learning_rate": 5e-06,
"loss": 0.2935,
"step": 3810
},
{
"epoch": 2.052108514638732,
"grad_norm": 0.4724986754383593,
"learning_rate": 5e-06,
"loss": 0.3013,
"step": 3820
},
{
"epoch": 2.0574805264571583,
"grad_norm": 0.4658048986243489,
"learning_rate": 5e-06,
"loss": 0.3082,
"step": 3830
},
{
"epoch": 2.062852538275584,
"grad_norm": 0.4311596560424912,
"learning_rate": 5e-06,
"loss": 0.2919,
"step": 3840
},
{
"epoch": 2.06822455009401,
"grad_norm": 0.4755135160095483,
"learning_rate": 5e-06,
"loss": 0.3087,
"step": 3850
},
{
"epoch": 2.0735965619124364,
"grad_norm": 0.4651593055702794,
"learning_rate": 5e-06,
"loss": 0.3031,
"step": 3860
},
{
"epoch": 2.078968573730862,
"grad_norm": 0.44604165006256646,
"learning_rate": 5e-06,
"loss": 0.2934,
"step": 3870
},
{
"epoch": 2.0843405855492882,
"grad_norm": 0.47944778982671077,
"learning_rate": 5e-06,
"loss": 0.302,
"step": 3880
},
{
"epoch": 2.0897125973677144,
"grad_norm": 0.45900387672803933,
"learning_rate": 5e-06,
"loss": 0.295,
"step": 3890
},
{
"epoch": 2.09508460918614,
"grad_norm": 0.4390114277911646,
"learning_rate": 5e-06,
"loss": 0.2952,
"step": 3900
},
{
"epoch": 2.1004566210045663,
"grad_norm": 0.46198748965614295,
"learning_rate": 5e-06,
"loss": 0.2961,
"step": 3910
},
{
"epoch": 2.105828632822992,
"grad_norm": 0.46567184763785124,
"learning_rate": 5e-06,
"loss": 0.3072,
"step": 3920
},
{
"epoch": 2.111200644641418,
"grad_norm": 0.4605817205911234,
"learning_rate": 5e-06,
"loss": 0.2989,
"step": 3930
},
{
"epoch": 2.1165726564598444,
"grad_norm": 0.4659945936319118,
"learning_rate": 5e-06,
"loss": 0.3057,
"step": 3940
},
{
"epoch": 2.12194466827827,
"grad_norm": 0.4698714737543665,
"learning_rate": 5e-06,
"loss": 0.2945,
"step": 3950
},
{
"epoch": 2.1273166800966963,
"grad_norm": 0.4603923117486592,
"learning_rate": 5e-06,
"loss": 0.2972,
"step": 3960
},
{
"epoch": 2.1326886919151224,
"grad_norm": 0.4444434238101049,
"learning_rate": 5e-06,
"loss": 0.2931,
"step": 3970
},
{
"epoch": 2.138060703733548,
"grad_norm": 0.46375448199682273,
"learning_rate": 5e-06,
"loss": 0.289,
"step": 3980
},
{
"epoch": 2.1434327155519743,
"grad_norm": 0.4515353697839556,
"learning_rate": 5e-06,
"loss": 0.2981,
"step": 3990
},
{
"epoch": 2.1488047273704,
"grad_norm": 0.45865800213441305,
"learning_rate": 5e-06,
"loss": 0.2946,
"step": 4000
},
{
"epoch": 2.1541767391888262,
"grad_norm": 0.47415589821389625,
"learning_rate": 5e-06,
"loss": 0.2963,
"step": 4010
},
{
"epoch": 2.1595487510072524,
"grad_norm": 0.45650337082160813,
"learning_rate": 5e-06,
"loss": 0.2965,
"step": 4020
},
{
"epoch": 2.164920762825678,
"grad_norm": 0.5007055892618506,
"learning_rate": 5e-06,
"loss": 0.3075,
"step": 4030
},
{
"epoch": 2.1702927746441043,
"grad_norm": 0.4633408694498252,
"learning_rate": 5e-06,
"loss": 0.3034,
"step": 4040
},
{
"epoch": 2.17566478646253,
"grad_norm": 0.47949376553994594,
"learning_rate": 5e-06,
"loss": 0.2959,
"step": 4050
},
{
"epoch": 2.181036798280956,
"grad_norm": 0.4534552216810433,
"learning_rate": 5e-06,
"loss": 0.2927,
"step": 4060
},
{
"epoch": 2.1864088100993824,
"grad_norm": 0.45779290627720376,
"learning_rate": 5e-06,
"loss": 0.3002,
"step": 4070
},
{
"epoch": 2.191780821917808,
"grad_norm": 0.45467471253139063,
"learning_rate": 5e-06,
"loss": 0.2941,
"step": 4080
},
{
"epoch": 2.1971528337362343,
"grad_norm": 0.4909879550475429,
"learning_rate": 5e-06,
"loss": 0.2959,
"step": 4090
},
{
"epoch": 2.2025248455546604,
"grad_norm": 0.4564057151231639,
"learning_rate": 5e-06,
"loss": 0.3004,
"step": 4100
},
{
"epoch": 2.207896857373086,
"grad_norm": 0.47859377699517,
"learning_rate": 5e-06,
"loss": 0.3006,
"step": 4110
},
{
"epoch": 2.2132688691915123,
"grad_norm": 0.46099184457974013,
"learning_rate": 5e-06,
"loss": 0.3043,
"step": 4120
},
{
"epoch": 2.218640881009938,
"grad_norm": 0.47055531304375103,
"learning_rate": 5e-06,
"loss": 0.3026,
"step": 4130
},
{
"epoch": 2.2240128928283642,
"grad_norm": 0.44859460535001255,
"learning_rate": 5e-06,
"loss": 0.2997,
"step": 4140
},
{
"epoch": 2.2293849046467904,
"grad_norm": 0.4523869768759771,
"learning_rate": 5e-06,
"loss": 0.3053,
"step": 4150
},
{
"epoch": 2.234756916465216,
"grad_norm": 0.4578556718341696,
"learning_rate": 5e-06,
"loss": 0.2935,
"step": 4160
},
{
"epoch": 2.2401289282836423,
"grad_norm": 0.47101805169234406,
"learning_rate": 5e-06,
"loss": 0.2984,
"step": 4170
},
{
"epoch": 2.245500940102068,
"grad_norm": 0.4721691371474998,
"learning_rate": 5e-06,
"loss": 0.3002,
"step": 4180
},
{
"epoch": 2.250872951920494,
"grad_norm": 0.4637692829174677,
"learning_rate": 5e-06,
"loss": 0.2942,
"step": 4190
},
{
"epoch": 2.2562449637389204,
"grad_norm": 0.4388979055074455,
"learning_rate": 5e-06,
"loss": 0.3084,
"step": 4200
},
{
"epoch": 2.261616975557346,
"grad_norm": 0.4855595548331312,
"learning_rate": 5e-06,
"loss": 0.3061,
"step": 4210
},
{
"epoch": 2.2669889873757723,
"grad_norm": 0.4664870575145768,
"learning_rate": 5e-06,
"loss": 0.3088,
"step": 4220
},
{
"epoch": 2.2723609991941984,
"grad_norm": 0.47347375338045333,
"learning_rate": 5e-06,
"loss": 0.2995,
"step": 4230
},
{
"epoch": 2.277733011012624,
"grad_norm": 0.7006364769945307,
"learning_rate": 5e-06,
"loss": 0.3008,
"step": 4240
},
{
"epoch": 2.2831050228310503,
"grad_norm": 0.45676326104006293,
"learning_rate": 5e-06,
"loss": 0.3144,
"step": 4250
},
{
"epoch": 2.288477034649476,
"grad_norm": 0.5036552686895771,
"learning_rate": 5e-06,
"loss": 0.2998,
"step": 4260
},
{
"epoch": 2.293849046467902,
"grad_norm": 0.4774908704932932,
"learning_rate": 5e-06,
"loss": 0.2985,
"step": 4270
},
{
"epoch": 2.2992210582863284,
"grad_norm": 0.45656706765413957,
"learning_rate": 5e-06,
"loss": 0.3005,
"step": 4280
},
{
"epoch": 2.304593070104754,
"grad_norm": 0.48161005420123015,
"learning_rate": 5e-06,
"loss": 0.2885,
"step": 4290
},
{
"epoch": 2.3099650819231803,
"grad_norm": 0.4539357790250988,
"learning_rate": 5e-06,
"loss": 0.3021,
"step": 4300
},
{
"epoch": 2.315337093741606,
"grad_norm": 0.4611459009005326,
"learning_rate": 5e-06,
"loss": 0.2969,
"step": 4310
},
{
"epoch": 2.320709105560032,
"grad_norm": 0.4618940592079481,
"learning_rate": 5e-06,
"loss": 0.3074,
"step": 4320
},
{
"epoch": 2.3260811173784584,
"grad_norm": 0.4816712055656272,
"learning_rate": 5e-06,
"loss": 0.2983,
"step": 4330
},
{
"epoch": 2.331453129196884,
"grad_norm": 0.4813233228122113,
"learning_rate": 5e-06,
"loss": 0.3045,
"step": 4340
},
{
"epoch": 2.3368251410153102,
"grad_norm": 0.47887192506303755,
"learning_rate": 5e-06,
"loss": 0.307,
"step": 4350
},
{
"epoch": 2.3421971528337364,
"grad_norm": 0.47464931229510837,
"learning_rate": 5e-06,
"loss": 0.3021,
"step": 4360
},
{
"epoch": 2.347569164652162,
"grad_norm": 0.4660602871711936,
"learning_rate": 5e-06,
"loss": 0.304,
"step": 4370
},
{
"epoch": 2.3529411764705883,
"grad_norm": 0.4880517674465139,
"learning_rate": 5e-06,
"loss": 0.3013,
"step": 4380
},
{
"epoch": 2.3583131882890145,
"grad_norm": 0.49041029407882114,
"learning_rate": 5e-06,
"loss": 0.2961,
"step": 4390
},
{
"epoch": 2.36368520010744,
"grad_norm": 0.4662688978117843,
"learning_rate": 5e-06,
"loss": 0.3007,
"step": 4400
},
{
"epoch": 2.3690572119258664,
"grad_norm": 0.4745603064237364,
"learning_rate": 5e-06,
"loss": 0.299,
"step": 4410
},
{
"epoch": 2.374429223744292,
"grad_norm": 0.46758674576062736,
"learning_rate": 5e-06,
"loss": 0.3077,
"step": 4420
},
{
"epoch": 2.3798012355627183,
"grad_norm": 0.46274751111290063,
"learning_rate": 5e-06,
"loss": 0.2893,
"step": 4430
},
{
"epoch": 2.385173247381144,
"grad_norm": 0.4979805795649497,
"learning_rate": 5e-06,
"loss": 0.2981,
"step": 4440
},
{
"epoch": 2.39054525919957,
"grad_norm": 0.46716558522728535,
"learning_rate": 5e-06,
"loss": 0.2944,
"step": 4450
},
{
"epoch": 2.3959172710179963,
"grad_norm": 0.4645219886431316,
"learning_rate": 5e-06,
"loss": 0.3027,
"step": 4460
},
{
"epoch": 2.401289282836422,
"grad_norm": 0.43076708233547145,
"learning_rate": 5e-06,
"loss": 0.2982,
"step": 4470
},
{
"epoch": 2.4066612946548482,
"grad_norm": 0.4833331043961502,
"learning_rate": 5e-06,
"loss": 0.3055,
"step": 4480
},
{
"epoch": 2.4120333064732744,
"grad_norm": 0.47944161915551303,
"learning_rate": 5e-06,
"loss": 0.308,
"step": 4490
},
{
"epoch": 2.4174053182917,
"grad_norm": 0.4735210760345749,
"learning_rate": 5e-06,
"loss": 0.2988,
"step": 4500
},
{
"epoch": 2.4227773301101263,
"grad_norm": 0.45565434475127603,
"learning_rate": 5e-06,
"loss": 0.3021,
"step": 4510
},
{
"epoch": 2.4281493419285525,
"grad_norm": 0.4766564731916228,
"learning_rate": 5e-06,
"loss": 0.3004,
"step": 4520
},
{
"epoch": 2.433521353746978,
"grad_norm": 0.48549992148905935,
"learning_rate": 5e-06,
"loss": 0.3033,
"step": 4530
},
{
"epoch": 2.4388933655654044,
"grad_norm": 0.44183880523434893,
"learning_rate": 5e-06,
"loss": 0.3071,
"step": 4540
},
{
"epoch": 2.44426537738383,
"grad_norm": 0.47705998139459066,
"learning_rate": 5e-06,
"loss": 0.3003,
"step": 4550
},
{
"epoch": 2.4496373892022563,
"grad_norm": 0.47178717771265055,
"learning_rate": 5e-06,
"loss": 0.2979,
"step": 4560
},
{
"epoch": 2.4550094010206824,
"grad_norm": 0.47229880316656225,
"learning_rate": 5e-06,
"loss": 0.323,
"step": 4570
},
{
"epoch": 2.460381412839108,
"grad_norm": 0.47107025763530014,
"learning_rate": 5e-06,
"loss": 0.2973,
"step": 4580
},
{
"epoch": 2.4657534246575343,
"grad_norm": 0.43860774515749135,
"learning_rate": 5e-06,
"loss": 0.2923,
"step": 4590
},
{
"epoch": 2.47112543647596,
"grad_norm": 0.44310429491661396,
"learning_rate": 5e-06,
"loss": 0.3091,
"step": 4600
},
{
"epoch": 2.4764974482943862,
"grad_norm": 0.450170058844727,
"learning_rate": 5e-06,
"loss": 0.2982,
"step": 4610
},
{
"epoch": 2.4818694601128124,
"grad_norm": 0.47368860930985485,
"learning_rate": 5e-06,
"loss": 0.3109,
"step": 4620
},
{
"epoch": 2.487241471931238,
"grad_norm": 0.4448141501224447,
"learning_rate": 5e-06,
"loss": 0.3043,
"step": 4630
},
{
"epoch": 2.4926134837496643,
"grad_norm": 0.4511180428107082,
"learning_rate": 5e-06,
"loss": 0.307,
"step": 4640
},
{
"epoch": 2.4979854955680905,
"grad_norm": 0.46753646829536444,
"learning_rate": 5e-06,
"loss": 0.306,
"step": 4650
},
{
"epoch": 2.503357507386516,
"grad_norm": 0.45938971563289144,
"learning_rate": 5e-06,
"loss": 0.3007,
"step": 4660
},
{
"epoch": 2.5087295192049424,
"grad_norm": 0.47175671154199156,
"learning_rate": 5e-06,
"loss": 0.3284,
"step": 4670
},
{
"epoch": 2.514101531023368,
"grad_norm": 0.4401639803199821,
"learning_rate": 5e-06,
"loss": 0.3045,
"step": 4680
},
{
"epoch": 2.5194735428417943,
"grad_norm": 0.47017862481838113,
"learning_rate": 5e-06,
"loss": 0.3188,
"step": 4690
},
{
"epoch": 2.52484555466022,
"grad_norm": 0.45763810223180906,
"learning_rate": 5e-06,
"loss": 0.296,
"step": 4700
},
{
"epoch": 2.530217566478646,
"grad_norm": 0.4644047897844528,
"learning_rate": 5e-06,
"loss": 0.2974,
"step": 4710
},
{
"epoch": 2.5355895782970723,
"grad_norm": 0.48551837865739933,
"learning_rate": 5e-06,
"loss": 0.2979,
"step": 4720
},
{
"epoch": 2.540961590115498,
"grad_norm": 0.4466465105264453,
"learning_rate": 5e-06,
"loss": 0.3077,
"step": 4730
},
{
"epoch": 2.546333601933924,
"grad_norm": 0.49252290575630375,
"learning_rate": 5e-06,
"loss": 0.3073,
"step": 4740
},
{
"epoch": 2.5517056137523504,
"grad_norm": 0.4302316284869702,
"learning_rate": 5e-06,
"loss": 0.2943,
"step": 4750
},
{
"epoch": 2.557077625570776,
"grad_norm": 0.469588515548392,
"learning_rate": 5e-06,
"loss": 0.3137,
"step": 4760
},
{
"epoch": 2.5624496373892023,
"grad_norm": 0.48615106142894776,
"learning_rate": 5e-06,
"loss": 0.316,
"step": 4770
},
{
"epoch": 2.5678216492076285,
"grad_norm": 0.47144647236889003,
"learning_rate": 5e-06,
"loss": 0.3059,
"step": 4780
},
{
"epoch": 2.573193661026054,
"grad_norm": 0.4690758287942174,
"learning_rate": 5e-06,
"loss": 0.3019,
"step": 4790
},
{
"epoch": 2.5785656728444803,
"grad_norm": 0.46961670498579505,
"learning_rate": 5e-06,
"loss": 0.3049,
"step": 4800
},
{
"epoch": 2.5839376846629065,
"grad_norm": 0.46991982229485013,
"learning_rate": 5e-06,
"loss": 0.3122,
"step": 4810
},
{
"epoch": 2.5893096964813322,
"grad_norm": 0.4679864164801053,
"learning_rate": 5e-06,
"loss": 0.2887,
"step": 4820
},
{
"epoch": 2.5946817082997584,
"grad_norm": 0.4441047775614893,
"learning_rate": 5e-06,
"loss": 0.3054,
"step": 4830
},
{
"epoch": 2.600053720118184,
"grad_norm": 0.47658994935407895,
"learning_rate": 5e-06,
"loss": 0.2985,
"step": 4840
},
{
"epoch": 2.6054257319366103,
"grad_norm": 0.46756360598099345,
"learning_rate": 5e-06,
"loss": 0.3062,
"step": 4850
},
{
"epoch": 2.610797743755036,
"grad_norm": 0.4632538377863063,
"learning_rate": 5e-06,
"loss": 0.3003,
"step": 4860
},
{
"epoch": 2.616169755573462,
"grad_norm": 0.4449475225017694,
"learning_rate": 5e-06,
"loss": 0.3017,
"step": 4870
},
{
"epoch": 2.6215417673918884,
"grad_norm": 0.4507497733905932,
"learning_rate": 5e-06,
"loss": 0.303,
"step": 4880
},
{
"epoch": 2.626913779210314,
"grad_norm": 0.46757813747214755,
"learning_rate": 5e-06,
"loss": 0.2994,
"step": 4890
},
{
"epoch": 2.6322857910287403,
"grad_norm": 0.45377996440263907,
"learning_rate": 5e-06,
"loss": 0.3033,
"step": 4900
},
{
"epoch": 2.6376578028471664,
"grad_norm": 0.46697319931601083,
"learning_rate": 5e-06,
"loss": 0.3169,
"step": 4910
},
{
"epoch": 2.643029814665592,
"grad_norm": 0.4594680488189553,
"learning_rate": 5e-06,
"loss": 0.2897,
"step": 4920
},
{
"epoch": 2.6484018264840183,
"grad_norm": 0.4977556770630546,
"learning_rate": 5e-06,
"loss": 0.3131,
"step": 4930
},
{
"epoch": 2.6537738383024445,
"grad_norm": 0.4599794430419653,
"learning_rate": 5e-06,
"loss": 0.303,
"step": 4940
},
{
"epoch": 2.6591458501208702,
"grad_norm": 0.4365617141166764,
"learning_rate": 5e-06,
"loss": 0.3038,
"step": 4950
},
{
"epoch": 2.6645178619392964,
"grad_norm": 0.4613086189177407,
"learning_rate": 5e-06,
"loss": 0.305,
"step": 4960
},
{
"epoch": 2.669889873757722,
"grad_norm": 0.46325663902687436,
"learning_rate": 5e-06,
"loss": 0.3128,
"step": 4970
},
{
"epoch": 2.6752618855761483,
"grad_norm": 0.456469424630923,
"learning_rate": 5e-06,
"loss": 0.3009,
"step": 4980
},
{
"epoch": 2.680633897394574,
"grad_norm": 0.4773037128036795,
"learning_rate": 5e-06,
"loss": 0.2999,
"step": 4990
},
{
"epoch": 2.686005909213,
"grad_norm": 0.48149380524695606,
"learning_rate": 5e-06,
"loss": 0.302,
"step": 5000
},
{
"epoch": 2.6913779210314264,
"grad_norm": 0.4516275475221354,
"learning_rate": 5e-06,
"loss": 0.2969,
"step": 5010
},
{
"epoch": 2.696749932849852,
"grad_norm": 0.4580414193871738,
"learning_rate": 5e-06,
"loss": 0.3124,
"step": 5020
},
{
"epoch": 2.7021219446682783,
"grad_norm": 0.454186120155934,
"learning_rate": 5e-06,
"loss": 0.3259,
"step": 5030
},
{
"epoch": 2.7074939564867044,
"grad_norm": 0.4359902304782414,
"learning_rate": 5e-06,
"loss": 0.3036,
"step": 5040
},
{
"epoch": 2.71286596830513,
"grad_norm": 0.4518860212887475,
"learning_rate": 5e-06,
"loss": 0.3046,
"step": 5050
},
{
"epoch": 2.7182379801235563,
"grad_norm": 0.44467216977831553,
"learning_rate": 5e-06,
"loss": 0.3018,
"step": 5060
},
{
"epoch": 2.7236099919419825,
"grad_norm": 0.47424637833011146,
"learning_rate": 5e-06,
"loss": 0.3026,
"step": 5070
},
{
"epoch": 2.7289820037604082,
"grad_norm": 0.5014084472686937,
"learning_rate": 5e-06,
"loss": 0.3005,
"step": 5080
},
{
"epoch": 2.7343540155788344,
"grad_norm": 0.481674717641741,
"learning_rate": 5e-06,
"loss": 0.313,
"step": 5090
},
{
"epoch": 2.73972602739726,
"grad_norm": 0.46503178430983816,
"learning_rate": 5e-06,
"loss": 0.3089,
"step": 5100
},
{
"epoch": 2.7450980392156863,
"grad_norm": 0.45047473457505316,
"learning_rate": 5e-06,
"loss": 0.3041,
"step": 5110
},
{
"epoch": 2.750470051034112,
"grad_norm": 0.5004805336195348,
"learning_rate": 5e-06,
"loss": 0.3116,
"step": 5120
},
{
"epoch": 2.755842062852538,
"grad_norm": 0.4276625646575985,
"learning_rate": 5e-06,
"loss": 0.301,
"step": 5130
},
{
"epoch": 2.7612140746709644,
"grad_norm": 0.44599433520202253,
"learning_rate": 5e-06,
"loss": 0.3168,
"step": 5140
},
{
"epoch": 2.76658608648939,
"grad_norm": 0.4767243040781986,
"learning_rate": 5e-06,
"loss": 0.3091,
"step": 5150
},
{
"epoch": 2.7719580983078163,
"grad_norm": 0.4539005921766262,
"learning_rate": 5e-06,
"loss": 0.3029,
"step": 5160
},
{
"epoch": 2.7773301101262424,
"grad_norm": 0.46406180742134173,
"learning_rate": 5e-06,
"loss": 0.3065,
"step": 5170
},
{
"epoch": 2.782702121944668,
"grad_norm": 0.4526636911178936,
"learning_rate": 5e-06,
"loss": 0.3181,
"step": 5180
},
{
"epoch": 2.7880741337630943,
"grad_norm": 0.45220356163061637,
"learning_rate": 5e-06,
"loss": 0.3075,
"step": 5190
},
{
"epoch": 2.7934461455815205,
"grad_norm": 0.45914982506711954,
"learning_rate": 5e-06,
"loss": 0.2989,
"step": 5200
},
{
"epoch": 2.798818157399946,
"grad_norm": 0.4852446695355985,
"learning_rate": 5e-06,
"loss": 0.3102,
"step": 5210
},
{
"epoch": 2.8041901692183724,
"grad_norm": 0.44174008476202076,
"learning_rate": 5e-06,
"loss": 0.3064,
"step": 5220
},
{
"epoch": 2.8095621810367986,
"grad_norm": 0.47195550534277947,
"learning_rate": 5e-06,
"loss": 0.3061,
"step": 5230
},
{
"epoch": 2.8149341928552243,
"grad_norm": 0.47482707999228474,
"learning_rate": 5e-06,
"loss": 0.302,
"step": 5240
},
{
"epoch": 2.82030620467365,
"grad_norm": 0.4655883266741662,
"learning_rate": 5e-06,
"loss": 0.3008,
"step": 5250
},
{
"epoch": 2.825678216492076,
"grad_norm": 0.47382383652851645,
"learning_rate": 5e-06,
"loss": 0.3009,
"step": 5260
},
{
"epoch": 2.8310502283105023,
"grad_norm": 0.45841149048452207,
"learning_rate": 5e-06,
"loss": 0.3125,
"step": 5270
},
{
"epoch": 2.836422240128928,
"grad_norm": 0.4404178325742541,
"learning_rate": 5e-06,
"loss": 0.3076,
"step": 5280
},
{
"epoch": 2.8417942519473542,
"grad_norm": 0.4691383498028934,
"learning_rate": 5e-06,
"loss": 0.3017,
"step": 5290
},
{
"epoch": 2.8471662637657804,
"grad_norm": 0.4495113094453148,
"learning_rate": 5e-06,
"loss": 0.3003,
"step": 5300
},
{
"epoch": 2.852538275584206,
"grad_norm": 0.46179624959522997,
"learning_rate": 5e-06,
"loss": 0.3078,
"step": 5310
},
{
"epoch": 2.8579102874026323,
"grad_norm": 0.4652311831624602,
"learning_rate": 5e-06,
"loss": 0.3164,
"step": 5320
},
{
"epoch": 2.8632822992210585,
"grad_norm": 0.47278598674863237,
"learning_rate": 5e-06,
"loss": 0.3035,
"step": 5330
},
{
"epoch": 2.868654311039484,
"grad_norm": 0.4910443155512238,
"learning_rate": 5e-06,
"loss": 0.3095,
"step": 5340
},
{
"epoch": 2.8740263228579104,
"grad_norm": 0.46167647985765303,
"learning_rate": 5e-06,
"loss": 0.3074,
"step": 5350
},
{
"epoch": 2.8793983346763365,
"grad_norm": 0.4954312490665213,
"learning_rate": 5e-06,
"loss": 0.3071,
"step": 5360
},
{
"epoch": 2.8847703464947623,
"grad_norm": 0.4367016608079839,
"learning_rate": 5e-06,
"loss": 0.3077,
"step": 5370
},
{
"epoch": 2.8901423583131884,
"grad_norm": 0.47686556153161486,
"learning_rate": 5e-06,
"loss": 0.303,
"step": 5380
},
{
"epoch": 2.895514370131614,
"grad_norm": 0.45520995582229345,
"learning_rate": 5e-06,
"loss": 0.2959,
"step": 5390
},
{
"epoch": 2.9008863819500403,
"grad_norm": 0.45167583945865714,
"learning_rate": 5e-06,
"loss": 0.3041,
"step": 5400
},
{
"epoch": 2.906258393768466,
"grad_norm": 0.489953863528834,
"learning_rate": 5e-06,
"loss": 0.3227,
"step": 5410
},
{
"epoch": 2.9116304055868922,
"grad_norm": 0.4842925647008196,
"learning_rate": 5e-06,
"loss": 0.3031,
"step": 5420
},
{
"epoch": 2.9170024174053184,
"grad_norm": 0.4679895193248062,
"learning_rate": 5e-06,
"loss": 0.3048,
"step": 5430
},
{
"epoch": 2.922374429223744,
"grad_norm": 0.46032428378414053,
"learning_rate": 5e-06,
"loss": 0.3001,
"step": 5440
},
{
"epoch": 2.9277464410421703,
"grad_norm": 0.49424698475244133,
"learning_rate": 5e-06,
"loss": 0.298,
"step": 5450
},
{
"epoch": 2.9331184528605965,
"grad_norm": 0.45206528520285993,
"learning_rate": 5e-06,
"loss": 0.2975,
"step": 5460
},
{
"epoch": 2.938490464679022,
"grad_norm": 0.43783656221438655,
"learning_rate": 5e-06,
"loss": 0.3056,
"step": 5470
},
{
"epoch": 2.9438624764974484,
"grad_norm": 0.4641248542873233,
"learning_rate": 5e-06,
"loss": 0.3131,
"step": 5480
},
{
"epoch": 2.9492344883158745,
"grad_norm": 0.45508868403537883,
"learning_rate": 5e-06,
"loss": 0.3026,
"step": 5490
},
{
"epoch": 2.9546065001343003,
"grad_norm": 0.4918746072980534,
"learning_rate": 5e-06,
"loss": 0.31,
"step": 5500
},
{
"epoch": 2.9599785119527264,
"grad_norm": 0.4775427997395849,
"learning_rate": 5e-06,
"loss": 0.3035,
"step": 5510
},
{
"epoch": 2.965350523771152,
"grad_norm": 0.43827768706167886,
"learning_rate": 5e-06,
"loss": 0.2982,
"step": 5520
},
{
"epoch": 2.9707225355895783,
"grad_norm": 0.4724169681109276,
"learning_rate": 5e-06,
"loss": 0.3013,
"step": 5530
},
{
"epoch": 2.976094547408004,
"grad_norm": 0.4534158098276835,
"learning_rate": 5e-06,
"loss": 0.2999,
"step": 5540
},
{
"epoch": 2.9814665592264302,
"grad_norm": 0.43461750252166137,
"learning_rate": 5e-06,
"loss": 0.2999,
"step": 5550
},
{
"epoch": 2.9868385710448564,
"grad_norm": 0.48563812071930773,
"learning_rate": 5e-06,
"loss": 0.3012,
"step": 5560
},
{
"epoch": 2.992210582863282,
"grad_norm": 0.5229982921947193,
"learning_rate": 5e-06,
"loss": 0.3139,
"step": 5570
},
{
"epoch": 2.9975825946817083,
"grad_norm": 0.46203252168577574,
"learning_rate": 5e-06,
"loss": 0.3035,
"step": 5580
},
{
"epoch": 2.999194198227236,
"eval_loss": 0.405712366104126,
"eval_runtime": 469.8508,
"eval_samples_per_second": 26.689,
"eval_steps_per_second": 0.417,
"step": 5583
},
{
"epoch": 2.999194198227236,
"step": 5583,
"total_flos": 2926650442383360.0,
"train_loss": 0.3617401811868598,
"train_runtime": 75547.7589,
"train_samples_per_second": 9.461,
"train_steps_per_second": 0.074
}
],
"logging_steps": 10,
"max_steps": 5583,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 2926650442383360.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}
Reference in New Issue View Git Blame Copy Permalink