frz-nt-glm / trainer_state.json
zjzju's picture
Upload 27 files
2d445d4
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 5.997246443322625,
"global_step": 4356,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 0.00019999739928881725,
"loss": 2.1906,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 0.00019998959729054295,
"loss": 1.8919,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 0.00019997659441099206,
"loss": 1.7855,
"step": 30
},
{
"epoch": 0.06,
"learning_rate": 0.00019995839132649917,
"loss": 1.7235,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 0.0001999349889838836,
"loss": 1.7027,
"step": 50
},
{
"epoch": 0.08,
"learning_rate": 0.00019990638860040006,
"loss": 1.7118,
"step": 60
},
{
"epoch": 0.1,
"learning_rate": 0.00019987259166367533,
"loss": 1.689,
"step": 70
},
{
"epoch": 0.11,
"learning_rate": 0.00019983359993163078,
"loss": 1.6845,
"step": 80
},
{
"epoch": 0.12,
"learning_rate": 0.0001997894154323911,
"loss": 1.6655,
"step": 90
},
{
"epoch": 0.14,
"learning_rate": 0.0001997400404641787,
"loss": 1.6518,
"step": 100
},
{
"epoch": 0.15,
"learning_rate": 0.00019968547759519425,
"loss": 1.6571,
"step": 110
},
{
"epoch": 0.17,
"learning_rate": 0.000199625729663483,
"loss": 1.6559,
"step": 120
},
{
"epoch": 0.18,
"learning_rate": 0.00019956079977678722,
"loss": 1.6417,
"step": 130
},
{
"epoch": 0.19,
"learning_rate": 0.0001994906913123846,
"loss": 1.6313,
"step": 140
},
{
"epoch": 0.21,
"learning_rate": 0.00019941540791691245,
"loss": 1.6243,
"step": 150
},
{
"epoch": 0.22,
"learning_rate": 0.00019933495350617813,
"loss": 1.6189,
"step": 160
},
{
"epoch": 0.23,
"learning_rate": 0.0001992493322649554,
"loss": 1.628,
"step": 170
},
{
"epoch": 0.25,
"learning_rate": 0.00019915854864676664,
"loss": 1.6149,
"step": 180
},
{
"epoch": 0.26,
"learning_rate": 0.00019906260737365122,
"loss": 1.6032,
"step": 190
},
{
"epoch": 0.28,
"learning_rate": 0.00019896151343592008,
"loss": 1.626,
"step": 200
},
{
"epoch": 0.29,
"learning_rate": 0.0001988552720918958,
"loss": 1.6369,
"step": 210
},
{
"epoch": 0.3,
"learning_rate": 0.00019874388886763944,
"loss": 1.5757,
"step": 220
},
{
"epoch": 0.32,
"learning_rate": 0.00019862736955666296,
"loss": 1.5873,
"step": 230
},
{
"epoch": 0.33,
"learning_rate": 0.00019850572021962788,
"loss": 1.6041,
"step": 240
},
{
"epoch": 0.34,
"learning_rate": 0.00019837894718402997,
"loss": 1.5905,
"step": 250
},
{
"epoch": 0.36,
"learning_rate": 0.00019824705704387028,
"loss": 1.6131,
"step": 260
},
{
"epoch": 0.37,
"learning_rate": 0.00019811005665931205,
"loss": 1.6045,
"step": 270
},
{
"epoch": 0.39,
"learning_rate": 0.00019796795315632395,
"loss": 1.5942,
"step": 280
},
{
"epoch": 0.4,
"learning_rate": 0.00019782075392630935,
"loss": 1.5768,
"step": 290
},
{
"epoch": 0.41,
"learning_rate": 0.00019766846662572191,
"loss": 1.5967,
"step": 300
},
{
"epoch": 0.43,
"learning_rate": 0.00019751109917566737,
"loss": 1.5606,
"step": 310
},
{
"epoch": 0.44,
"learning_rate": 0.00019734865976149145,
"loss": 1.58,
"step": 320
},
{
"epoch": 0.45,
"learning_rate": 0.00019718115683235417,
"loss": 1.5809,
"step": 330
},
{
"epoch": 0.47,
"learning_rate": 0.00019700859910079036,
"loss": 1.5758,
"step": 340
},
{
"epoch": 0.48,
"learning_rate": 0.00019683099554225649,
"loss": 1.57,
"step": 350
},
{
"epoch": 0.5,
"learning_rate": 0.0001966483553946637,
"loss": 1.5783,
"step": 360
},
{
"epoch": 0.51,
"learning_rate": 0.00019646068815789755,
"loss": 1.5699,
"step": 370
},
{
"epoch": 0.52,
"learning_rate": 0.00019626800359332362,
"loss": 1.5767,
"step": 380
},
{
"epoch": 0.54,
"learning_rate": 0.00019607031172327996,
"loss": 1.5635,
"step": 390
},
{
"epoch": 0.55,
"learning_rate": 0.00019586762283055573,
"loss": 1.5763,
"step": 400
},
{
"epoch": 0.56,
"learning_rate": 0.0001956599474578563,
"loss": 1.5719,
"step": 410
},
{
"epoch": 0.58,
"learning_rate": 0.00019544729640725498,
"loss": 1.5712,
"step": 420
},
{
"epoch": 0.59,
"learning_rate": 0.00019522968073963106,
"loss": 1.5699,
"step": 430
},
{
"epoch": 0.61,
"learning_rate": 0.00019500711177409454,
"loss": 1.539,
"step": 440
},
{
"epoch": 0.62,
"learning_rate": 0.0001947796010873974,
"loss": 1.5515,
"step": 450
},
{
"epoch": 0.63,
"learning_rate": 0.00019454716051333135,
"loss": 1.5691,
"step": 460
},
{
"epoch": 0.65,
"learning_rate": 0.0001943098021421124,
"loss": 1.573,
"step": 470
},
{
"epoch": 0.66,
"learning_rate": 0.00019406753831975203,
"loss": 1.5532,
"step": 480
},
{
"epoch": 0.67,
"learning_rate": 0.00019382038164741477,
"loss": 1.5778,
"step": 490
},
{
"epoch": 0.69,
"learning_rate": 0.0001935683449807631,
"loss": 1.5514,
"step": 500
},
{
"epoch": 0.7,
"learning_rate": 0.00019331144142928854,
"loss": 1.5493,
"step": 510
},
{
"epoch": 0.72,
"learning_rate": 0.00019304968435562993,
"loss": 1.5698,
"step": 520
},
{
"epoch": 0.73,
"learning_rate": 0.00019278308737487822,
"loss": 1.5699,
"step": 530
},
{
"epoch": 0.74,
"learning_rate": 0.0001925116643538684,
"loss": 1.554,
"step": 540
},
{
"epoch": 0.76,
"learning_rate": 0.00019223542941045817,
"loss": 1.5527,
"step": 550
},
{
"epoch": 0.77,
"learning_rate": 0.00019195439691279363,
"loss": 1.5517,
"step": 560
},
{
"epoch": 0.78,
"learning_rate": 0.00019166858147856203,
"loss": 1.5673,
"step": 570
},
{
"epoch": 0.8,
"learning_rate": 0.00019137799797423126,
"loss": 1.5469,
"step": 580
},
{
"epoch": 0.81,
"learning_rate": 0.00019108266151427673,
"loss": 1.5499,
"step": 590
},
{
"epoch": 0.83,
"learning_rate": 0.00019078258746039507,
"loss": 1.5502,
"step": 600
},
{
"epoch": 0.84,
"learning_rate": 0.00019047779142070527,
"loss": 1.5402,
"step": 610
},
{
"epoch": 0.85,
"learning_rate": 0.0001901682892489367,
"loss": 1.5444,
"step": 620
},
{
"epoch": 0.87,
"learning_rate": 0.00018985409704360456,
"loss": 1.5524,
"step": 630
},
{
"epoch": 0.88,
"learning_rate": 0.00018953523114717245,
"loss": 1.5338,
"step": 640
},
{
"epoch": 0.89,
"learning_rate": 0.00018921170814520247,
"loss": 1.5416,
"step": 650
},
{
"epoch": 0.91,
"learning_rate": 0.00018888354486549237,
"loss": 1.5524,
"step": 660
},
{
"epoch": 0.92,
"learning_rate": 0.00018855075837720034,
"loss": 1.544,
"step": 670
},
{
"epoch": 0.94,
"learning_rate": 0.0001882133659899573,
"loss": 1.5317,
"step": 680
},
{
"epoch": 0.95,
"learning_rate": 0.0001878713852529663,
"loss": 1.5516,
"step": 690
},
{
"epoch": 0.96,
"learning_rate": 0.00018752483395408987,
"loss": 1.5479,
"step": 700
},
{
"epoch": 0.98,
"learning_rate": 0.00018717373011892474,
"loss": 1.5328,
"step": 710
},
{
"epoch": 0.99,
"learning_rate": 0.0001868180920098644,
"loss": 1.524,
"step": 720
},
{
"epoch": 1.01,
"learning_rate": 0.00018645793812514894,
"loss": 1.4712,
"step": 730
},
{
"epoch": 1.02,
"learning_rate": 0.0001860932871979031,
"loss": 1.3509,
"step": 740
},
{
"epoch": 1.03,
"learning_rate": 0.00018572415819516174,
"loss": 1.3227,
"step": 750
},
{
"epoch": 1.05,
"learning_rate": 0.00018535057031688335,
"loss": 1.3216,
"step": 760
},
{
"epoch": 1.06,
"learning_rate": 0.00018497254299495146,
"loss": 1.3022,
"step": 770
},
{
"epoch": 1.07,
"learning_rate": 0.00018459009589216364,
"loss": 1.308,
"step": 780
},
{
"epoch": 1.09,
"learning_rate": 0.00018420324890120916,
"loss": 1.3295,
"step": 790
},
{
"epoch": 1.1,
"learning_rate": 0.0001838120221436338,
"loss": 1.3249,
"step": 800
},
{
"epoch": 1.12,
"learning_rate": 0.00018341643596879367,
"loss": 1.3109,
"step": 810
},
{
"epoch": 1.13,
"learning_rate": 0.00018301651095279655,
"loss": 1.328,
"step": 820
},
{
"epoch": 1.14,
"learning_rate": 0.00018261226789743172,
"loss": 1.3215,
"step": 830
},
{
"epoch": 1.16,
"learning_rate": 0.00018220372782908777,
"loss": 1.3378,
"step": 840
},
{
"epoch": 1.17,
"learning_rate": 0.00018179091199765926,
"loss": 1.3289,
"step": 850
},
{
"epoch": 1.18,
"learning_rate": 0.00018137384187544116,
"loss": 1.3362,
"step": 860
},
{
"epoch": 1.2,
"learning_rate": 0.00018095253915601206,
"loss": 1.3263,
"step": 870
},
{
"epoch": 1.21,
"learning_rate": 0.00018052702575310588,
"loss": 1.3238,
"step": 880
},
{
"epoch": 1.23,
"learning_rate": 0.00018009732379947188,
"loss": 1.3278,
"step": 890
},
{
"epoch": 1.24,
"learning_rate": 0.0001796634556457236,
"loss": 1.3277,
"step": 900
},
{
"epoch": 1.25,
"learning_rate": 0.00017922544385917628,
"loss": 1.3332,
"step": 910
},
{
"epoch": 1.27,
"learning_rate": 0.00017878331122267284,
"loss": 1.3268,
"step": 920
},
{
"epoch": 1.28,
"learning_rate": 0.00017833708073339922,
"loss": 1.3316,
"step": 930
},
{
"epoch": 1.29,
"learning_rate": 0.00017788677560168784,
"loss": 1.337,
"step": 940
},
{
"epoch": 1.31,
"learning_rate": 0.0001774324192498105,
"loss": 1.3307,
"step": 950
},
{
"epoch": 1.32,
"learning_rate": 0.0001769740353107602,
"loss": 1.3295,
"step": 960
},
{
"epoch": 1.34,
"learning_rate": 0.0001765116476270216,
"loss": 1.3148,
"step": 970
},
{
"epoch": 1.35,
"learning_rate": 0.00017604528024933115,
"loss": 1.3269,
"step": 980
},
{
"epoch": 1.36,
"learning_rate": 0.00017557495743542585,
"loss": 1.3488,
"step": 990
},
{
"epoch": 1.38,
"learning_rate": 0.00017510070364878177,
"loss": 1.3377,
"step": 1000
},
{
"epoch": 1.39,
"learning_rate": 0.0001746225435573415,
"loss": 1.3396,
"step": 1010
},
{
"epoch": 1.4,
"learning_rate": 0.0001741405020322309,
"loss": 1.3375,
"step": 1020
},
{
"epoch": 1.42,
"learning_rate": 0.00017365460414646574,
"loss": 1.326,
"step": 1030
},
{
"epoch": 1.43,
"learning_rate": 0.00017316487517364721,
"loss": 1.3297,
"step": 1040
},
{
"epoch": 1.45,
"learning_rate": 0.00017267134058664775,
"loss": 1.3429,
"step": 1050
},
{
"epoch": 1.46,
"learning_rate": 0.00017217402605628572,
"loss": 1.3339,
"step": 1060
},
{
"epoch": 1.47,
"learning_rate": 0.00017167295744999027,
"loss": 1.3375,
"step": 1070
},
{
"epoch": 1.49,
"learning_rate": 0.00017116816083045602,
"loss": 1.3161,
"step": 1080
},
{
"epoch": 1.5,
"learning_rate": 0.00017065966245428723,
"loss": 1.3281,
"step": 1090
},
{
"epoch": 1.51,
"learning_rate": 0.00017014748877063214,
"loss": 1.322,
"step": 1100
},
{
"epoch": 1.53,
"learning_rate": 0.0001696316664198073,
"loss": 1.3344,
"step": 1110
},
{
"epoch": 1.54,
"learning_rate": 0.00016911222223191182,
"loss": 1.3607,
"step": 1120
},
{
"epoch": 1.56,
"learning_rate": 0.00016858918322543186,
"loss": 1.34,
"step": 1130
},
{
"epoch": 1.57,
"learning_rate": 0.00016806257660583534,
"loss": 1.3324,
"step": 1140
},
{
"epoch": 1.58,
"learning_rate": 0.00016753242976415666,
"loss": 1.3249,
"step": 1150
},
{
"epoch": 1.6,
"learning_rate": 0.00016699877027557226,
"loss": 1.3363,
"step": 1160
},
{
"epoch": 1.61,
"learning_rate": 0.00016646162589796615,
"loss": 1.3403,
"step": 1170
},
{
"epoch": 1.62,
"learning_rate": 0.0001659210245704861,
"loss": 1.3283,
"step": 1180
},
{
"epoch": 1.64,
"learning_rate": 0.00016537699441209047,
"loss": 1.3258,
"step": 1190
},
{
"epoch": 1.65,
"learning_rate": 0.0001648295637200856,
"loss": 1.3258,
"step": 1200
},
{
"epoch": 1.67,
"learning_rate": 0.00016427876096865394,
"loss": 1.3581,
"step": 1210
},
{
"epoch": 1.68,
"learning_rate": 0.00016372461480737297,
"loss": 1.3337,
"step": 1220
},
{
"epoch": 1.69,
"learning_rate": 0.00016316715405972508,
"loss": 1.3432,
"step": 1230
},
{
"epoch": 1.71,
"learning_rate": 0.0001626064077215983,
"loss": 1.3385,
"step": 1240
},
{
"epoch": 1.72,
"learning_rate": 0.00016204240495977805,
"loss": 1.3474,
"step": 1250
},
{
"epoch": 1.73,
"learning_rate": 0.0001614751751104301,
"loss": 1.3198,
"step": 1260
},
{
"epoch": 1.75,
"learning_rate": 0.00016090474767757474,
"loss": 1.3407,
"step": 1270
},
{
"epoch": 1.76,
"learning_rate": 0.00016033115233155202,
"loss": 1.3203,
"step": 1280
},
{
"epoch": 1.78,
"learning_rate": 0.00015975441890747855,
"loss": 1.3311,
"step": 1290
},
{
"epoch": 1.79,
"learning_rate": 0.00015917457740369565,
"loss": 1.3271,
"step": 1300
},
{
"epoch": 1.8,
"learning_rate": 0.000158591657980209,
"loss": 1.3432,
"step": 1310
},
{
"epoch": 1.82,
"learning_rate": 0.00015800569095711982,
"loss": 1.3302,
"step": 1320
},
{
"epoch": 1.83,
"learning_rate": 0.00015741670681304796,
"loss": 1.3326,
"step": 1330
},
{
"epoch": 1.84,
"learning_rate": 0.00015682473618354635,
"loss": 1.3358,
"step": 1340
},
{
"epoch": 1.86,
"learning_rate": 0.0001562298098595078,
"loss": 1.3343,
"step": 1350
},
{
"epoch": 1.87,
"learning_rate": 0.0001556319587855631,
"loss": 1.3214,
"step": 1360
},
{
"epoch": 1.89,
"learning_rate": 0.0001550312140584718,
"loss": 1.3342,
"step": 1370
},
{
"epoch": 1.9,
"learning_rate": 0.00015442760692550443,
"loss": 1.3253,
"step": 1380
},
{
"epoch": 1.91,
"learning_rate": 0.0001538211687828174,
"loss": 1.3494,
"step": 1390
},
{
"epoch": 1.93,
"learning_rate": 0.00015321193117381996,
"loss": 1.324,
"step": 1400
},
{
"epoch": 1.94,
"learning_rate": 0.00015259992578753334,
"loss": 1.3267,
"step": 1410
},
{
"epoch": 1.96,
"learning_rate": 0.00015198518445694255,
"loss": 1.343,
"step": 1420
},
{
"epoch": 1.97,
"learning_rate": 0.00015136773915734066,
"loss": 1.3209,
"step": 1430
},
{
"epoch": 1.98,
"learning_rate": 0.00015074762200466556,
"loss": 1.3224,
"step": 1440
},
{
"epoch": 2.0,
"learning_rate": 0.0001501248652538296,
"loss": 1.3481,
"step": 1450
},
{
"epoch": 2.01,
"learning_rate": 0.00014949950129704162,
"loss": 1.1061,
"step": 1460
},
{
"epoch": 2.02,
"learning_rate": 0.00014887156266212237,
"loss": 0.9927,
"step": 1470
},
{
"epoch": 2.04,
"learning_rate": 0.00014824108201081247,
"loss": 0.9946,
"step": 1480
},
{
"epoch": 2.05,
"learning_rate": 0.00014760809213707344,
"loss": 0.9764,
"step": 1490
},
{
"epoch": 2.07,
"learning_rate": 0.00014697262596538227,
"loss": 0.9851,
"step": 1500
},
{
"epoch": 2.08,
"learning_rate": 0.00014633471654901842,
"loss": 0.9941,
"step": 1510
},
{
"epoch": 2.09,
"learning_rate": 0.000145694397068345,
"loss": 0.9875,
"step": 1520
},
{
"epoch": 2.11,
"learning_rate": 0.0001450517008290827,
"loss": 0.9828,
"step": 1530
},
{
"epoch": 2.12,
"learning_rate": 0.00014440666126057744,
"loss": 0.9913,
"step": 1540
},
{
"epoch": 2.13,
"learning_rate": 0.00014375931191406159,
"loss": 0.9881,
"step": 1550
},
{
"epoch": 2.15,
"learning_rate": 0.00014310968646090883,
"loss": 0.9866,
"step": 1560
},
{
"epoch": 2.16,
"learning_rate": 0.0001424578186908828,
"loss": 0.9934,
"step": 1570
},
{
"epoch": 2.18,
"learning_rate": 0.0001418037425103795,
"loss": 0.9963,
"step": 1580
},
{
"epoch": 2.19,
"learning_rate": 0.00014114749194066363,
"loss": 0.9791,
"step": 1590
},
{
"epoch": 2.2,
"learning_rate": 0.00014048910111609915,
"loss": 0.9851,
"step": 1600
},
{
"epoch": 2.22,
"learning_rate": 0.0001398286042823736,
"loss": 0.9829,
"step": 1610
},
{
"epoch": 2.23,
"learning_rate": 0.00013916603579471705,
"loss": 0.9995,
"step": 1620
},
{
"epoch": 2.24,
"learning_rate": 0.00013850143011611497,
"loss": 0.9843,
"step": 1630
},
{
"epoch": 2.26,
"learning_rate": 0.0001378348218155158,
"loss": 1.0025,
"step": 1640
},
{
"epoch": 2.27,
"learning_rate": 0.00013716624556603274,
"loss": 1.0135,
"step": 1650
},
{
"epoch": 2.29,
"learning_rate": 0.00013649573614314044,
"loss": 1.0007,
"step": 1660
},
{
"epoch": 2.3,
"learning_rate": 0.00013582332842286592,
"loss": 1.0117,
"step": 1670
},
{
"epoch": 2.31,
"learning_rate": 0.00013514905737997473,
"loss": 1.0068,
"step": 1680
},
{
"epoch": 2.33,
"learning_rate": 0.0001344729580861517,
"loss": 0.998,
"step": 1690
},
{
"epoch": 2.34,
"learning_rate": 0.0001337950657081768,
"loss": 0.9995,
"step": 1700
},
{
"epoch": 2.35,
"learning_rate": 0.00013311541550609565,
"loss": 0.9975,
"step": 1710
},
{
"epoch": 2.37,
"learning_rate": 0.00013243404283138597,
"loss": 1.014,
"step": 1720
},
{
"epoch": 2.38,
"learning_rate": 0.0001317509831251184,
"loss": 0.9954,
"step": 1730
},
{
"epoch": 2.4,
"learning_rate": 0.00013106627191611332,
"loss": 1.0051,
"step": 1740
},
{
"epoch": 2.41,
"learning_rate": 0.00013037994481909264,
"loss": 1.0156,
"step": 1750
},
{
"epoch": 2.42,
"learning_rate": 0.0001296920375328275,
"loss": 1.0155,
"step": 1760
},
{
"epoch": 2.44,
"learning_rate": 0.00012900258583828137,
"loss": 1.0223,
"step": 1770
},
{
"epoch": 2.45,
"learning_rate": 0.00012831162559674887,
"loss": 1.0181,
"step": 1780
},
{
"epoch": 2.46,
"learning_rate": 0.00012761919274799054,
"loss": 1.0023,
"step": 1790
},
{
"epoch": 2.48,
"learning_rate": 0.00012692532330836346,
"loss": 0.9917,
"step": 1800
},
{
"epoch": 2.49,
"learning_rate": 0.0001262300533689478,
"loss": 1.0132,
"step": 1810
},
{
"epoch": 2.51,
"learning_rate": 0.00012553341909366978,
"loss": 1.0136,
"step": 1820
},
{
"epoch": 2.52,
"learning_rate": 0.0001248354567174203,
"loss": 0.9961,
"step": 1830
},
{
"epoch": 2.53,
"learning_rate": 0.00012413620254417057,
"loss": 1.0244,
"step": 1840
},
{
"epoch": 2.55,
"learning_rate": 0.0001234356929450835,
"loss": 1.0126,
"step": 1850
},
{
"epoch": 2.56,
"learning_rate": 0.00012273396435662212,
"loss": 1.0121,
"step": 1860
},
{
"epoch": 2.57,
"learning_rate": 0.00012203105327865407,
"loss": 1.0293,
"step": 1870
},
{
"epoch": 2.59,
"learning_rate": 0.00012132699627255347,
"loss": 1.0104,
"step": 1880
},
{
"epoch": 2.6,
"learning_rate": 0.00012062182995929882,
"loss": 1.0119,
"step": 1890
},
{
"epoch": 2.62,
"learning_rate": 0.00011991559101756852,
"loss": 1.0047,
"step": 1900
},
{
"epoch": 2.63,
"learning_rate": 0.00011920831618183282,
"loss": 1.0095,
"step": 1910
},
{
"epoch": 2.64,
"learning_rate": 0.00011850004224044315,
"loss": 1.0112,
"step": 1920
},
{
"epoch": 2.66,
"learning_rate": 0.0001177908060337188,
"loss": 1.0131,
"step": 1930
},
{
"epoch": 2.67,
"learning_rate": 0.00011708064445203042,
"loss": 1.0082,
"step": 1940
},
{
"epoch": 2.68,
"learning_rate": 0.00011636959443388132,
"loss": 1.0246,
"step": 1950
},
{
"epoch": 2.7,
"learning_rate": 0.00011565769296398618,
"loss": 1.0102,
"step": 1960
},
{
"epoch": 2.71,
"learning_rate": 0.00011494497707134731,
"loss": 1.0103,
"step": 1970
},
{
"epoch": 2.73,
"learning_rate": 0.00011423148382732853,
"loss": 1.0081,
"step": 1980
},
{
"epoch": 2.74,
"learning_rate": 0.000113517250343727,
"loss": 1.0092,
"step": 1990
},
{
"epoch": 2.75,
"learning_rate": 0.0001128023137708429,
"loss": 0.9983,
"step": 2000
},
{
"epoch": 2.77,
"learning_rate": 0.00011208671129554702,
"loss": 1.0136,
"step": 2010
},
{
"epoch": 2.78,
"learning_rate": 0.00011137048013934656,
"loss": 1.0348,
"step": 2020
},
{
"epoch": 2.79,
"learning_rate": 0.00011065365755644906,
"loss": 0.9996,
"step": 2030
},
{
"epoch": 2.81,
"learning_rate": 0.00010993628083182467,
"loss": 1.0142,
"step": 2040
},
{
"epoch": 2.82,
"learning_rate": 0.00010921838727926681,
"loss": 1.0101,
"step": 2050
},
{
"epoch": 2.84,
"learning_rate": 0.00010850001423945126,
"loss": 1.0126,
"step": 2060
},
{
"epoch": 2.85,
"learning_rate": 0.00010778119907799398,
"loss": 1.0188,
"step": 2070
},
{
"epoch": 2.86,
"learning_rate": 0.00010706197918350758,
"loss": 1.0232,
"step": 2080
},
{
"epoch": 2.88,
"learning_rate": 0.00010634239196565646,
"loss": 1.0161,
"step": 2090
},
{
"epoch": 2.89,
"learning_rate": 0.00010562247485321115,
"loss": 1.0213,
"step": 2100
},
{
"epoch": 2.91,
"learning_rate": 0.0001049022652921013,
"loss": 0.9976,
"step": 2110
},
{
"epoch": 2.92,
"learning_rate": 0.00010418180074346815,
"loss": 1.0324,
"step": 2120
},
{
"epoch": 2.93,
"learning_rate": 0.00010346111868171584,
"loss": 1.0132,
"step": 2130
},
{
"epoch": 2.95,
"learning_rate": 0.00010274025659256232,
"loss": 1.0003,
"step": 2140
},
{
"epoch": 2.96,
"learning_rate": 0.00010201925197108953,
"loss": 1.0046,
"step": 2150
},
{
"epoch": 2.97,
"learning_rate": 0.0001012981423197931,
"loss": 1.0093,
"step": 2160
},
{
"epoch": 2.99,
"learning_rate": 0.00010057696514663169,
"loss": 1.0065,
"step": 2170
},
{
"epoch": 3.0,
"learning_rate": 9.985575796307615e-05,
"loss": 0.9698,
"step": 2180
},
{
"epoch": 3.02,
"learning_rate": 9.913455828215814e-05,
"loss": 0.6643,
"step": 2190
},
{
"epoch": 3.03,
"learning_rate": 9.84134036165192e-05,
"loss": 0.6469,
"step": 2200
},
{
"epoch": 3.04,
"learning_rate": 9.769233147645943e-05,
"loss": 0.6258,
"step": 2210
},
{
"epoch": 3.06,
"learning_rate": 9.697137936798634e-05,
"loss": 0.6403,
"step": 2220
},
{
"epoch": 3.07,
"learning_rate": 9.625058479086418e-05,
"loss": 0.6438,
"step": 2230
},
{
"epoch": 3.08,
"learning_rate": 9.552998523666326e-05,
"loss": 0.6249,
"step": 2240
},
{
"epoch": 3.1,
"learning_rate": 9.480961818681004e-05,
"loss": 0.6426,
"step": 2250
},
{
"epoch": 3.11,
"learning_rate": 9.408952111063727e-05,
"loss": 0.652,
"step": 2260
},
{
"epoch": 3.13,
"learning_rate": 9.336973146343537e-05,
"loss": 0.644,
"step": 2270
},
{
"epoch": 3.14,
"learning_rate": 9.265028668450402e-05,
"loss": 0.6364,
"step": 2280
},
{
"epoch": 3.15,
"learning_rate": 9.193122419520485e-05,
"loss": 0.6513,
"step": 2290
},
{
"epoch": 3.17,
"learning_rate": 9.121258139701502e-05,
"loss": 0.6501,
"step": 2300
},
{
"epoch": 3.18,
"learning_rate": 9.049439566958175e-05,
"loss": 0.6431,
"step": 2310
},
{
"epoch": 3.19,
"learning_rate": 8.977670436877811e-05,
"loss": 0.652,
"step": 2320
},
{
"epoch": 3.21,
"learning_rate": 8.905954482475991e-05,
"loss": 0.6494,
"step": 2330
},
{
"epoch": 3.22,
"learning_rate": 8.83429543400241e-05,
"loss": 0.6486,
"step": 2340
},
{
"epoch": 3.24,
"learning_rate": 8.76269701874684e-05,
"loss": 0.6399,
"step": 2350
},
{
"epoch": 3.25,
"learning_rate": 8.691162960845264e-05,
"loss": 0.6452,
"step": 2360
},
{
"epoch": 3.26,
"learning_rate": 8.619696981086172e-05,
"loss": 0.6448,
"step": 2370
},
{
"epoch": 3.28,
"learning_rate": 8.548302796717019e-05,
"loss": 0.6541,
"step": 2380
},
{
"epoch": 3.29,
"learning_rate": 8.476984121250875e-05,
"loss": 0.6405,
"step": 2390
},
{
"epoch": 3.3,
"learning_rate": 8.405744664273278e-05,
"loss": 0.6419,
"step": 2400
},
{
"epoch": 3.32,
"learning_rate": 8.334588131249277e-05,
"loss": 0.6369,
"step": 2410
},
{
"epoch": 3.33,
"learning_rate": 8.263518223330697e-05,
"loss": 0.6585,
"step": 2420
},
{
"epoch": 3.35,
"learning_rate": 8.192538637163621e-05,
"loss": 0.6461,
"step": 2430
},
{
"epoch": 3.36,
"learning_rate": 8.121653064696118e-05,
"loss": 0.6614,
"step": 2440
},
{
"epoch": 3.37,
"learning_rate": 8.050865192986211e-05,
"loss": 0.6536,
"step": 2450
},
{
"epoch": 3.39,
"learning_rate": 7.980178704010089e-05,
"loss": 0.6559,
"step": 2460
},
{
"epoch": 3.4,
"learning_rate": 7.9095972744706e-05,
"loss": 0.6605,
"step": 2470
},
{
"epoch": 3.41,
"learning_rate": 7.839124575606004e-05,
"loss": 0.6631,
"step": 2480
},
{
"epoch": 3.43,
"learning_rate": 7.76876427299903e-05,
"loss": 0.6666,
"step": 2490
},
{
"epoch": 3.44,
"learning_rate": 7.69852002638618e-05,
"loss": 0.6646,
"step": 2500
},
{
"epoch": 3.46,
"learning_rate": 7.62839548946742e-05,
"loss": 0.6526,
"step": 2510
},
{
"epoch": 3.47,
"learning_rate": 7.558394309716088e-05,
"loss": 0.657,
"step": 2520
},
{
"epoch": 3.48,
"learning_rate": 7.488520128189209e-05,
"loss": 0.6546,
"step": 2530
},
{
"epoch": 3.5,
"learning_rate": 7.41877657933809e-05,
"loss": 0.657,
"step": 2540
},
{
"epoch": 3.51,
"learning_rate": 7.349167290819274e-05,
"loss": 0.6594,
"step": 2550
},
{
"epoch": 3.52,
"learning_rate": 7.279695883305866e-05,
"loss": 0.6676,
"step": 2560
},
{
"epoch": 3.54,
"learning_rate": 7.210365970299194e-05,
"loss": 0.6527,
"step": 2570
},
{
"epoch": 3.55,
"learning_rate": 7.141181157940859e-05,
"loss": 0.6494,
"step": 2580
},
{
"epoch": 3.57,
"learning_rate": 7.072145044825162e-05,
"loss": 0.6525,
"step": 2590
},
{
"epoch": 3.58,
"learning_rate": 7.003261221811934e-05,
"loss": 0.6521,
"step": 2600
},
{
"epoch": 3.59,
"learning_rate": 6.934533271839752e-05,
"loss": 0.6669,
"step": 2610
},
{
"epoch": 3.61,
"learning_rate": 6.865964769739575e-05,
"loss": 0.6548,
"step": 2620
},
{
"epoch": 3.62,
"learning_rate": 6.797559282048806e-05,
"loss": 0.6668,
"step": 2630
},
{
"epoch": 3.63,
"learning_rate": 6.729320366825784e-05,
"loss": 0.6516,
"step": 2640
},
{
"epoch": 3.65,
"learning_rate": 6.661251573464706e-05,
"loss": 0.655,
"step": 2650
},
{
"epoch": 3.66,
"learning_rate": 6.593356442511015e-05,
"loss": 0.6641,
"step": 2660
},
{
"epoch": 3.68,
"learning_rate": 6.525638505477231e-05,
"loss": 0.6592,
"step": 2670
},
{
"epoch": 3.69,
"learning_rate": 6.458101284659286e-05,
"loss": 0.6582,
"step": 2680
},
{
"epoch": 3.7,
"learning_rate": 6.390748292953284e-05,
"loss": 0.6814,
"step": 2690
},
{
"epoch": 3.72,
"learning_rate": 6.323583033672799e-05,
"loss": 0.658,
"step": 2700
},
{
"epoch": 3.73,
"learning_rate": 6.256609000366649e-05,
"loss": 0.6653,
"step": 2710
},
{
"epoch": 3.74,
"learning_rate": 6.189829676637182e-05,
"loss": 0.654,
"step": 2720
},
{
"epoch": 3.76,
"learning_rate": 6.123248535959083e-05,
"loss": 0.6626,
"step": 2730
},
{
"epoch": 3.77,
"learning_rate": 6.056869041498687e-05,
"loss": 0.6696,
"step": 2740
},
{
"epoch": 3.79,
"learning_rate": 5.9906946459338656e-05,
"loss": 0.6635,
"step": 2750
},
{
"epoch": 3.8,
"learning_rate": 5.924728791274432e-05,
"loss": 0.6593,
"step": 2760
},
{
"epoch": 3.81,
"learning_rate": 5.858974908683105e-05,
"loss": 0.6622,
"step": 2770
},
{
"epoch": 3.83,
"learning_rate": 5.79343641829704e-05,
"loss": 0.6605,
"step": 2780
},
{
"epoch": 3.84,
"learning_rate": 5.728116729049928e-05,
"loss": 0.6664,
"step": 2790
},
{
"epoch": 3.85,
"learning_rate": 5.663019238494704e-05,
"loss": 0.6575,
"step": 2800
},
{
"epoch": 3.87,
"learning_rate": 5.5981473326267976e-05,
"loss": 0.6792,
"step": 2810
},
{
"epoch": 3.88,
"learning_rate": 5.533504385708024e-05,
"loss": 0.6841,
"step": 2820
},
{
"epoch": 3.9,
"learning_rate": 5.4690937600910905e-05,
"loss": 0.6625,
"step": 2830
},
{
"epoch": 3.91,
"learning_rate": 5.404918806044679e-05,
"loss": 0.6637,
"step": 2840
},
{
"epoch": 3.92,
"learning_rate": 5.340982861579199e-05,
"loss": 0.6587,
"step": 2850
},
{
"epoch": 3.94,
"learning_rate": 5.277289252273174e-05,
"loss": 0.6641,
"step": 2860
},
{
"epoch": 3.95,
"learning_rate": 5.213841291100239e-05,
"loss": 0.6515,
"step": 2870
},
{
"epoch": 3.97,
"learning_rate": 5.1506422782568345e-05,
"loss": 0.6596,
"step": 2880
},
{
"epoch": 3.98,
"learning_rate": 5.087695500990555e-05,
"loss": 0.6508,
"step": 2890
},
{
"epoch": 3.99,
"learning_rate": 5.025004233429145e-05,
"loss": 0.6552,
"step": 2900
},
{
"epoch": 4.01,
"learning_rate": 4.962571736410223e-05,
"loss": 0.5383,
"step": 2910
},
{
"epoch": 4.02,
"learning_rate": 4.90040125731165e-05,
"loss": 0.4324,
"step": 2920
},
{
"epoch": 4.03,
"learning_rate": 4.8384960298826274e-05,
"loss": 0.4194,
"step": 2930
},
{
"epoch": 4.05,
"learning_rate": 4.776859274075506e-05,
"loss": 0.4238,
"step": 2940
},
{
"epoch": 4.06,
"learning_rate": 4.715494195878285e-05,
"loss": 0.419,
"step": 2950
},
{
"epoch": 4.08,
"learning_rate": 4.654403987147865e-05,
"loss": 0.4239,
"step": 2960
},
{
"epoch": 4.09,
"learning_rate": 4.593591825444028e-05,
"loss": 0.4261,
"step": 2970
},
{
"epoch": 4.1,
"learning_rate": 4.5330608738641486e-05,
"loss": 0.4415,
"step": 2980
},
{
"epoch": 4.12,
"learning_rate": 4.472814280878689e-05,
"loss": 0.4236,
"step": 2990
},
{
"epoch": 4.13,
"learning_rate": 4.412855180167406e-05,
"loss": 0.4334,
"step": 3000
},
{
"epoch": 4.14,
"learning_rate": 4.353186690456371e-05,
"loss": 0.422,
"step": 3010
},
{
"epoch": 4.16,
"learning_rate": 4.293811915355761e-05,
"loss": 0.4361,
"step": 3020
},
{
"epoch": 4.17,
"learning_rate": 4.234733943198399e-05,
"loss": 0.4339,
"step": 3030
},
{
"epoch": 4.19,
"learning_rate": 4.175955846879151e-05,
"loss": 0.44,
"step": 3040
},
{
"epoch": 4.2,
"learning_rate": 4.11748068369506e-05,
"loss": 0.4328,
"step": 3050
},
{
"epoch": 4.21,
"learning_rate": 4.059311495186338e-05,
"loss": 0.4297,
"step": 3060
},
{
"epoch": 4.23,
"learning_rate": 4.001451306978174e-05,
"loss": 0.4225,
"step": 3070
},
{
"epoch": 4.24,
"learning_rate": 3.943903128623335e-05,
"loss": 0.4288,
"step": 3080
},
{
"epoch": 4.25,
"learning_rate": 3.886669953445637e-05,
"loss": 0.4198,
"step": 3090
},
{
"epoch": 4.27,
"learning_rate": 3.829754758384262e-05,
"loss": 0.4199,
"step": 3100
},
{
"epoch": 4.28,
"learning_rate": 3.77316050383889e-05,
"loss": 0.4308,
"step": 3110
},
{
"epoch": 4.3,
"learning_rate": 3.7168901335157315e-05,
"loss": 0.4335,
"step": 3120
},
{
"epoch": 4.31,
"learning_rate": 3.660946574274421e-05,
"loss": 0.4267,
"step": 3130
},
{
"epoch": 4.32,
"learning_rate": 3.6053327359757535e-05,
"loss": 0.4348,
"step": 3140
},
{
"epoch": 4.34,
"learning_rate": 3.550051511330361e-05,
"loss": 0.4403,
"step": 3150
},
{
"epoch": 4.35,
"learning_rate": 3.4951057757482205e-05,
"loss": 0.4252,
"step": 3160
},
{
"epoch": 4.36,
"learning_rate": 3.440498387189111e-05,
"loss": 0.4212,
"step": 3170
},
{
"epoch": 4.38,
"learning_rate": 3.3862321860139576e-05,
"loss": 0.4336,
"step": 3180
},
{
"epoch": 4.39,
"learning_rate": 3.332309994837085e-05,
"loss": 0.4304,
"step": 3190
},
{
"epoch": 4.41,
"learning_rate": 3.278734618379402e-05,
"loss": 0.4245,
"step": 3200
},
{
"epoch": 4.42,
"learning_rate": 3.225508843322524e-05,
"loss": 0.4113,
"step": 3210
},
{
"epoch": 4.43,
"learning_rate": 3.172635438163816e-05,
"loss": 0.4389,
"step": 3220
},
{
"epoch": 4.45,
"learning_rate": 3.120117153072404e-05,
"loss": 0.427,
"step": 3230
},
{
"epoch": 4.46,
"learning_rate": 3.0679567197461134e-05,
"loss": 0.4162,
"step": 3240
},
{
"epoch": 4.47,
"learning_rate": 3.016156851269384e-05,
"loss": 0.4347,
"step": 3250
},
{
"epoch": 4.49,
"learning_rate": 2.9647202419721687e-05,
"loss": 0.4259,
"step": 3260
},
{
"epoch": 4.5,
"learning_rate": 2.913649567289759e-05,
"loss": 0.4399,
"step": 3270
},
{
"epoch": 4.52,
"learning_rate": 2.862947483623659e-05,
"loss": 0.4278,
"step": 3280
},
{
"epoch": 4.53,
"learning_rate": 2.812616628203383e-05,
"loss": 0.4344,
"step": 3290
},
{
"epoch": 4.54,
"learning_rate": 2.7626596189492983e-05,
"loss": 0.4339,
"step": 3300
},
{
"epoch": 4.56,
"learning_rate": 2.7130790543364646e-05,
"loss": 0.4353,
"step": 3310
},
{
"epoch": 4.57,
"learning_rate": 2.6638775132594553e-05,
"loss": 0.4333,
"step": 3320
},
{
"epoch": 4.58,
"learning_rate": 2.6150575548982292e-05,
"loss": 0.4336,
"step": 3330
},
{
"epoch": 4.6,
"learning_rate": 2.5666217185850262e-05,
"loss": 0.4388,
"step": 3340
},
{
"epoch": 4.61,
"learning_rate": 2.5185725236722636e-05,
"loss": 0.4322,
"step": 3350
},
{
"epoch": 4.63,
"learning_rate": 2.4709124694015116e-05,
"loss": 0.4435,
"step": 3360
},
{
"epoch": 4.64,
"learning_rate": 2.423644034773498e-05,
"loss": 0.4358,
"step": 3370
},
{
"epoch": 4.65,
"learning_rate": 2.3767696784191463e-05,
"loss": 0.4471,
"step": 3380
},
{
"epoch": 4.67,
"learning_rate": 2.3302918384717177e-05,
"loss": 0.4227,
"step": 3390
},
{
"epoch": 4.68,
"learning_rate": 2.284212932439972e-05,
"loss": 0.4269,
"step": 3400
},
{
"epoch": 4.69,
"learning_rate": 2.2385353570824308e-05,
"loss": 0.4393,
"step": 3410
},
{
"epoch": 4.71,
"learning_rate": 2.1932614882827197e-05,
"loss": 0.4331,
"step": 3420
},
{
"epoch": 4.72,
"learning_rate": 2.148393680925973e-05,
"loss": 0.4407,
"step": 3430
},
{
"epoch": 4.74,
"learning_rate": 2.1039342687763586e-05,
"loss": 0.4335,
"step": 3440
},
{
"epoch": 4.75,
"learning_rate": 2.0598855643556826e-05,
"loss": 0.4196,
"step": 3450
},
{
"epoch": 4.76,
"learning_rate": 2.016249858823106e-05,
"loss": 0.4298,
"step": 3460
},
{
"epoch": 4.78,
"learning_rate": 1.973029421855981e-05,
"loss": 0.4392,
"step": 3470
},
{
"epoch": 4.79,
"learning_rate": 1.93022650153178e-05,
"loss": 0.4368,
"step": 3480
},
{
"epoch": 4.8,
"learning_rate": 1.8878433242111716e-05,
"loss": 0.4373,
"step": 3490
},
{
"epoch": 4.82,
"learning_rate": 1.8458820944222255e-05,
"loss": 0.4409,
"step": 3500
},
{
"epoch": 4.83,
"learning_rate": 1.804344994745727e-05,
"loss": 0.4263,
"step": 3510
},
{
"epoch": 4.85,
"learning_rate": 1.763234185701673e-05,
"loss": 0.4119,
"step": 3520
},
{
"epoch": 4.86,
"learning_rate": 1.7225518056368785e-05,
"loss": 0.4381,
"step": 3530
},
{
"epoch": 4.87,
"learning_rate": 1.6822999706137567e-05,
"loss": 0.4296,
"step": 3540
},
{
"epoch": 4.89,
"learning_rate": 1.6424807743002612e-05,
"loss": 0.4212,
"step": 3550
},
{
"epoch": 4.9,
"learning_rate": 1.6030962878609725e-05,
"loss": 0.4228,
"step": 3560
},
{
"epoch": 4.92,
"learning_rate": 1.5641485598493743e-05,
"loss": 0.4132,
"step": 3570
},
{
"epoch": 4.93,
"learning_rate": 1.5256396161013075e-05,
"loss": 0.4433,
"step": 3580
},
{
"epoch": 4.94,
"learning_rate": 1.487571459629582e-05,
"loss": 0.4401,
"step": 3590
},
{
"epoch": 4.96,
"learning_rate": 1.4499460705197998e-05,
"loss": 0.4329,
"step": 3600
},
{
"epoch": 4.97,
"learning_rate": 1.412765405827372e-05,
"loss": 0.43,
"step": 3610
},
{
"epoch": 4.98,
"learning_rate": 1.3760313994757001e-05,
"loss": 0.4292,
"step": 3620
},
{
"epoch": 5.0,
"learning_rate": 1.339745962155613e-05,
"loss": 0.4316,
"step": 3630
},
{
"epoch": 5.01,
"learning_rate": 1.3039109812259598e-05,
"loss": 0.353,
"step": 3640
},
{
"epoch": 5.03,
"learning_rate": 1.268528320615452e-05,
"loss": 0.3517,
"step": 3650
},
{
"epoch": 5.04,
"learning_rate": 1.2335998207257137e-05,
"loss": 0.3451,
"step": 3660
},
{
"epoch": 5.05,
"learning_rate": 1.1991272983355505e-05,
"loss": 0.3536,
"step": 3670
},
{
"epoch": 5.07,
"learning_rate": 1.1651125465064516e-05,
"loss": 0.3435,
"step": 3680
},
{
"epoch": 5.08,
"learning_rate": 1.131557334489326e-05,
"loss": 0.3361,
"step": 3690
},
{
"epoch": 5.09,
"learning_rate": 1.098463407632474e-05,
"loss": 0.3415,
"step": 3700
},
{
"epoch": 5.11,
"learning_rate": 1.0658324872908121e-05,
"loss": 0.3426,
"step": 3710
},
{
"epoch": 5.12,
"learning_rate": 1.0336662707363287e-05,
"loss": 0.3476,
"step": 3720
},
{
"epoch": 5.14,
"learning_rate": 1.0019664310698029e-05,
"loss": 0.3474,
"step": 3730
},
{
"epoch": 5.15,
"learning_rate": 9.707346171337894e-06,
"loss": 0.3476,
"step": 3740
},
{
"epoch": 5.16,
"learning_rate": 9.399724534268384e-06,
"loss": 0.3354,
"step": 3750
},
{
"epoch": 5.18,
"learning_rate": 9.096815400190172e-06,
"loss": 0.3514,
"step": 3760
},
{
"epoch": 5.19,
"learning_rate": 8.798634524686699e-06,
"loss": 0.3519,
"step": 3770
},
{
"epoch": 5.2,
"learning_rate": 8.505197417404687e-06,
"loss": 0.3387,
"step": 3780
},
{
"epoch": 5.22,
"learning_rate": 8.216519341247486e-06,
"loss": 0.3458,
"step": 3790
},
{
"epoch": 5.23,
"learning_rate": 7.932615311581126e-06,
"loss": 0.3403,
"step": 3800
},
{
"epoch": 5.25,
"learning_rate": 7.653500095453248e-06,
"loss": 0.3461,
"step": 3810
},
{
"epoch": 5.26,
"learning_rate": 7.3791882108251945e-06,
"loss": 0.3445,
"step": 3820
},
{
"epoch": 5.27,
"learning_rate": 7.109693925816651e-06,
"loss": 0.3439,
"step": 3830
},
{
"epoch": 5.29,
"learning_rate": 6.845031257963619e-06,
"loss": 0.3446,
"step": 3840
},
{
"epoch": 5.3,
"learning_rate": 6.585213973489335e-06,
"loss": 0.3577,
"step": 3850
},
{
"epoch": 5.31,
"learning_rate": 6.3302555865880965e-06,
"loss": 0.3368,
"step": 3860
},
{
"epoch": 5.33,
"learning_rate": 6.08016935872251e-06,
"loss": 0.338,
"step": 3870
},
{
"epoch": 5.34,
"learning_rate": 5.834968297933541e-06,
"loss": 0.3552,
"step": 3880
},
{
"epoch": 5.36,
"learning_rate": 5.594665158163992e-06,
"loss": 0.3489,
"step": 3890
},
{
"epoch": 5.37,
"learning_rate": 5.359272438595153e-06,
"loss": 0.354,
"step": 3900
},
{
"epoch": 5.38,
"learning_rate": 5.128802382996567e-06,
"loss": 0.3512,
"step": 3910
},
{
"epoch": 5.4,
"learning_rate": 4.903266979089249e-06,
"loss": 0.3333,
"step": 3920
},
{
"epoch": 5.41,
"learning_rate": 4.682677957922155e-06,
"loss": 0.3523,
"step": 3930
},
{
"epoch": 5.42,
"learning_rate": 4.467046793261931e-06,
"loss": 0.3462,
"step": 3940
},
{
"epoch": 5.44,
"learning_rate": 4.256384700996252e-06,
"loss": 0.34,
"step": 3950
},
{
"epoch": 5.45,
"learning_rate": 4.050702638550275e-06,
"loss": 0.3555,
"step": 3960
},
{
"epoch": 5.47,
"learning_rate": 3.850011304316781e-06,
"loss": 0.357,
"step": 3970
},
{
"epoch": 5.48,
"learning_rate": 3.6543211370997587e-06,
"loss": 0.3549,
"step": 3980
},
{
"epoch": 5.49,
"learning_rate": 3.4636423155712916e-06,
"loss": 0.3455,
"step": 3990
},
{
"epoch": 5.51,
"learning_rate": 3.2779847577422697e-06,
"loss": 0.3513,
"step": 4000
},
{
"epoch": 5.52,
"learning_rate": 3.0973581204464362e-06,
"loss": 0.3554,
"step": 4010
},
{
"epoch": 5.53,
"learning_rate": 2.921771798838069e-06,
"loss": 0.3433,
"step": 4020
},
{
"epoch": 5.55,
"learning_rate": 2.751234925903412e-06,
"loss": 0.3544,
"step": 4030
},
{
"epoch": 5.56,
"learning_rate": 2.585756371985493e-06,
"loss": 0.3466,
"step": 4040
},
{
"epoch": 5.58,
"learning_rate": 2.4253447443228106e-06,
"loss": 0.3459,
"step": 4050
},
{
"epoch": 5.59,
"learning_rate": 2.270008386601685e-06,
"loss": 0.3462,
"step": 4060
},
{
"epoch": 5.6,
"learning_rate": 2.119755378522137e-06,
"loss": 0.3398,
"step": 4070
},
{
"epoch": 5.62,
"learning_rate": 1.974593535377722e-06,
"loss": 0.3488,
"step": 4080
},
{
"epoch": 5.63,
"learning_rate": 1.83453040764906e-06,
"loss": 0.349,
"step": 4090
},
{
"epoch": 5.64,
"learning_rate": 1.6995732806109554e-06,
"loss": 0.3416,
"step": 4100
},
{
"epoch": 5.66,
"learning_rate": 1.569729173953638e-06,
"loss": 0.3556,
"step": 4110
},
{
"epoch": 5.67,
"learning_rate": 1.4450048414174854e-06,
"loss": 0.3341,
"step": 4120
},
{
"epoch": 5.69,
"learning_rate": 1.3254067704418283e-06,
"loss": 0.3511,
"step": 4130
},
{
"epoch": 5.7,
"learning_rate": 1.2109411818274852e-06,
"loss": 0.3536,
"step": 4140
},
{
"epoch": 5.71,
"learning_rate": 1.1016140294131894e-06,
"loss": 0.3318,
"step": 4150
},
{
"epoch": 5.73,
"learning_rate": 9.974309997658915e-07,
"loss": 0.3383,
"step": 4160
},
{
"epoch": 5.74,
"learning_rate": 8.983975118849852e-07,
"loss": 0.3465,
"step": 4170
},
{
"epoch": 5.75,
"learning_rate": 8.04518716920466e-07,
"loss": 0.3512,
"step": 4180
},
{
"epoch": 5.77,
"learning_rate": 7.157994979049898e-07,
"loss": 0.3377,
"step": 4190
},
{
"epoch": 5.78,
"learning_rate": 6.322444694998319e-07,
"loss": 0.3473,
"step": 4200
},
{
"epoch": 5.8,
"learning_rate": 5.538579777549347e-07,
"loss": 0.3423,
"step": 4210
},
{
"epoch": 5.81,
"learning_rate": 4.80644099882821e-07,
"loss": 0.3424,
"step": 4220
},
{
"epoch": 5.82,
"learning_rate": 4.126066440464982e-07,
"loss": 0.3448,
"step": 4230
},
{
"epoch": 5.84,
"learning_rate": 3.497491491614158e-07,
"loss": 0.3438,
"step": 4240
},
{
"epoch": 5.85,
"learning_rate": 2.920748847113686e-07,
"loss": 0.3514,
"step": 4250
},
{
"epoch": 5.87,
"learning_rate": 2.395868505784438e-07,
"loss": 0.3395,
"step": 4260
},
{
"epoch": 5.88,
"learning_rate": 1.9228777688700127e-07,
"loss": 0.353,
"step": 4270
},
{
"epoch": 5.89,
"learning_rate": 1.5018012386162072e-07,
"loss": 0.3474,
"step": 4280
},
{
"epoch": 5.91,
"learning_rate": 1.1326608169920372e-07,
"loss": 0.345,
"step": 4290
},
{
"epoch": 5.92,
"learning_rate": 8.154757045497619e-08,
"loss": 0.3494,
"step": 4300
},
{
"epoch": 5.93,
"learning_rate": 5.50262399426904e-08,
"loss": 0.3439,
"step": 4310
},
{
"epoch": 5.95,
"learning_rate": 3.370346964876036e-08,
"loss": 0.3474,
"step": 4320
},
{
"epoch": 5.96,
"learning_rate": 1.7580368660519152e-08,
"loss": 0.335,
"step": 4330
},
{
"epoch": 5.98,
"learning_rate": 6.657775608553962e-09,
"loss": 0.3516,
"step": 4340
},
{
"epoch": 5.99,
"learning_rate": 9.362586230632354e-10,
"loss": 0.3405,
"step": 4350
},
{
"epoch": 6.0,
"step": 4356,
"total_flos": 5.268602572020646e+18,
"train_loss": 0.8953271216487972,
"train_runtime": 33942.058,
"train_samples_per_second": 12.326,
"train_steps_per_second": 0.128
}
],
"max_steps": 4356,
"num_train_epochs": 6,
"total_flos": 5.268602572020646e+18,
"trial_name": null,
"trial_params": null
}