2888 lines
57 KiB
JSON
2888 lines
57 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 2.9998415967052114,
|
|
"global_step": 2367,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 0.00010416666666666667,
|
|
"loss": 5.6655,
|
|
"step": 5
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 0.00020833333333333335,
|
|
"loss": 4.5516,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 0.0003125,
|
|
"loss": 3.7337,
|
|
"step": 15
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 0.0004166666666666667,
|
|
"loss": 3.2268,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 0.0005208333333333334,
|
|
"loss": 2.9287,
|
|
"step": 25
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 0.000625,
|
|
"loss": 2.8609,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 0.0007291666666666666,
|
|
"loss": 2.7533,
|
|
"step": 35
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 0.0008333333333333334,
|
|
"loss": 2.7291,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 0.0009375,
|
|
"loss": 2.5916,
|
|
"step": 45
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 0.0009999981647383803,
|
|
"loss": 2.605,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 0.0009999775181998842,
|
|
"loss": 2.6466,
|
|
"step": 55
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 0.0009999339319963169,
|
|
"loss": 2.5633,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 0.000999867408127474,
|
|
"loss": 2.628,
|
|
"step": 65
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 0.0009997779496455639,
|
|
"loss": 2.5335,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 0.0009996655606550656,
|
|
"loss": 2.5589,
|
|
"step": 75
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 0.000999530246312543,
|
|
"loss": 2.5707,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 0.0009993720128264065,
|
|
"loss": 2.428,
|
|
"step": 85
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 0.000999190867456629,
|
|
"loss": 2.4371,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 0.0009989868185144122,
|
|
"loss": 2.4955,
|
|
"step": 95
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 0.0009987598753618061,
|
|
"loss": 2.4295,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 0.0009985100484112787,
|
|
"loss": 2.3623,
|
|
"step": 105
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 0.0009982373491252382,
|
|
"loss": 2.3422,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 0.0009979417900155081,
|
|
"loss": 2.3134,
|
|
"step": 115
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 0.0009976233846427521,
|
|
"loss": 2.4091,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 0.000997282147615852,
|
|
"loss": 2.4004,
|
|
"step": 125
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 0.0009969180945912381,
|
|
"loss": 2.3233,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 0.0009965312422721704,
|
|
"loss": 2.3183,
|
|
"step": 135
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 0.0009961216084079724,
|
|
"loss": 2.3771,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 0.0009956892117932161,
|
|
"loss": 2.2779,
|
|
"step": 145
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 0.000995234072266861,
|
|
"loss": 2.303,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 0.0009947562107113423,
|
|
"loss": 2.3452,
|
|
"step": 155
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 0.000994255649051614,
|
|
"loss": 2.2872,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 0.0009937324102541423,
|
|
"loss": 2.3187,
|
|
"step": 165
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 0.0009931865183258524,
|
|
"loss": 2.2837,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 0.0009926179983130263,
|
|
"loss": 2.4024,
|
|
"step": 175
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 0.0009920268763001542,
|
|
"loss": 2.347,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 0.0009914131794087377,
|
|
"loss": 2.3126,
|
|
"step": 185
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 0.0009907769357960449,
|
|
"loss": 2.2042,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 0.0009901181746538195,
|
|
"loss": 2.1674,
|
|
"step": 195
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 0.0009894369262069401,
|
|
"loss": 2.1963,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 0.0009887332217120347,
|
|
"loss": 2.2971,
|
|
"step": 205
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 0.0009880070934560458,
|
|
"loss": 2.1949,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 0.0009872585747547494,
|
|
"loss": 2.1544,
|
|
"step": 215
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 0.0009864876999512264,
|
|
"loss": 2.1866,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 0.0009856945044142864,
|
|
"loss": 2.198,
|
|
"step": 225
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 0.000984879024536846,
|
|
"loss": 2.2507,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 0.0009840412977342582,
|
|
"loss": 2.1326,
|
|
"step": 235
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 0.000983181362442595,
|
|
"loss": 2.204,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 0.0009822992581168867,
|
|
"loss": 2.2197,
|
|
"step": 245
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 0.0009813950252293075,
|
|
"loss": 2.1562,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 0.0009804687052673228,
|
|
"loss": 2.2193,
|
|
"step": 255
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 0.0009795203407317828,
|
|
"loss": 2.2668,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 0.0009785499751349732,
|
|
"loss": 2.2434,
|
|
"step": 265
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 0.0009775576529986198,
|
|
"loss": 2.1776,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 0.0009765434198518448,
|
|
"loss": 2.2592,
|
|
"step": 275
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 0.0009755073222290775,
|
|
"loss": 2.1072,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 0.0009744494076679204,
|
|
"loss": 2.0784,
|
|
"step": 285
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 0.0009733697247069677,
|
|
"loss": 2.1776,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 0.0009722683228835774,
|
|
"loss": 2.1752,
|
|
"step": 295
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 0.0009711452527315997,
|
|
"loss": 2.0958,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 0.0009700005657790578,
|
|
"loss": 2.1055,
|
|
"step": 305
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 0.0009688343145457837,
|
|
"loss": 2.1135,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 0.0009676465525410088,
|
|
"loss": 2.0275,
|
|
"step": 315
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 0.0009664373342609086,
|
|
"loss": 2.0031,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 0.0009652067151861023,
|
|
"loss": 2.1489,
|
|
"step": 325
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 0.0009639547517791077,
|
|
"loss": 2.1305,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 0.0009626815014817496,
|
|
"loss": 2.0894,
|
|
"step": 335
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 0.0009613870227125259,
|
|
"loss": 2.0822,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 0.0009600713748639258,
|
|
"loss": 2.1499,
|
|
"step": 345
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 0.0009587346182997054,
|
|
"loss": 2.0848,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 0.0009573768143521182,
|
|
"loss": 2.0754,
|
|
"step": 355
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 0.000955998025319101,
|
|
"loss": 2.1364,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 0.0009545983144614152,
|
|
"loss": 2.066,
|
|
"step": 365
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 0.0009531777459997453,
|
|
"loss": 2.1016,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 0.0009517363851117511,
|
|
"loss": 2.0479,
|
|
"step": 375
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 0.0009502742979290782,
|
|
"loss": 2.0646,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 0.0009487915515343235,
|
|
"loss": 2.1756,
|
|
"step": 385
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 0.0009472882139579572,
|
|
"loss": 2.0339,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 0.0009457643541752017,
|
|
"loss": 2.1112,
|
|
"step": 395
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 0.0009442200421028666,
|
|
"loss": 2.1327,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 0.0009426553485961416,
|
|
"loss": 2.0274,
|
|
"step": 405
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 0.0009410703454453442,
|
|
"loss": 1.9823,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 0.0009394651053726273,
|
|
"loss": 1.9665,
|
|
"step": 415
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 0.0009378397020286417,
|
|
"loss": 2.0468,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 0.0009361942099891571,
|
|
"loss": 1.9778,
|
|
"step": 425
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 0.0009345287047516411,
|
|
"loss": 2.0264,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 0.0009328432627317937,
|
|
"loss": 2.1288,
|
|
"step": 435
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 0.0009311379612600432,
|
|
"loss": 1.9488,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 0.0009294128785779969,
|
|
"loss": 1.9984,
|
|
"step": 445
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 0.0009276680938348512,
|
|
"loss": 2.0063,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 0.000925903687083761,
|
|
"loss": 2.052,
|
|
"step": 455
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 0.0009241197392781662,
|
|
"loss": 2.0739,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 0.000922316332268077,
|
|
"loss": 1.9291,
|
|
"step": 465
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 0.00092049354879632,
|
|
"loss": 2.0575,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 0.0009186514724947399,
|
|
"loss": 1.9736,
|
|
"step": 475
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 0.0009167901878803637,
|
|
"loss": 2.019,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 0.0009149097803515225,
|
|
"loss": 1.8767,
|
|
"step": 485
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 0.0009130103361839337,
|
|
"loss": 2.144,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 0.0009110919425267414,
|
|
"loss": 1.9369,
|
|
"step": 495
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 0.0009091546873985196,
|
|
"loss": 2.0787,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 0.0009071986596832321,
|
|
"loss": 1.8462,
|
|
"step": 505
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 0.0009052239491261558,
|
|
"loss": 1.9782,
|
|
"step": 510
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 0.0009032306463297619,
|
|
"loss": 1.912,
|
|
"step": 515
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 0.0009012188427495597,
|
|
"loss": 2.0624,
|
|
"step": 520
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 0.0008991886306899002,
|
|
"loss": 1.9532,
|
|
"step": 525
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 0.0008971401032997409,
|
|
"loss": 2.0872,
|
|
"step": 530
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 0.0008950733545683725,
|
|
"loss": 2.0329,
|
|
"step": 535
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 0.0008929884793211061,
|
|
"loss": 1.9024,
|
|
"step": 540
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 0.0008908855732149224,
|
|
"loss": 1.9682,
|
|
"step": 545
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 0.0008887647327340832,
|
|
"loss": 1.9982,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 0.0008866260551857045,
|
|
"loss": 1.9342,
|
|
"step": 555
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 0.0008844696386952916,
|
|
"loss": 1.8963,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 0.0008822955822022372,
|
|
"loss": 1.9586,
|
|
"step": 565
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 0.0008801039854552822,
|
|
"loss": 1.8984,
|
|
"step": 570
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 0.0008778949490079383,
|
|
"loss": 1.8707,
|
|
"step": 575
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 0.0008756685742138751,
|
|
"loss": 1.9667,
|
|
"step": 580
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 0.0008734249632222702,
|
|
"loss": 1.931,
|
|
"step": 585
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 0.0008711642189731211,
|
|
"loss": 1.9109,
|
|
"step": 590
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 0.0008688864451925238,
|
|
"loss": 1.8908,
|
|
"step": 595
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 0.0008665917463879124,
|
|
"loss": 2.0963,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 0.0008642802278432649,
|
|
"loss": 1.9331,
|
|
"step": 605
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 0.0008619519956142725,
|
|
"loss": 2.0053,
|
|
"step": 610
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 0.0008596071565234733,
|
|
"loss": 1.9626,
|
|
"step": 615
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 0.0008572458181553517,
|
|
"loss": 1.9638,
|
|
"step": 620
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 0.0008548680888514013,
|
|
"loss": 1.8846,
|
|
"step": 625
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 0.0008524740777051554,
|
|
"loss": 2.0264,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 0.0008500638945571804,
|
|
"loss": 1.9709,
|
|
"step": 635
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 0.0008476376499900369,
|
|
"loss": 1.8799,
|
|
"step": 640
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 0.0008451954553232055,
|
|
"loss": 1.9257,
|
|
"step": 645
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 0.0008427374226079803,
|
|
"loss": 1.9897,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 0.0008402636646223262,
|
|
"loss": 1.9573,
|
|
"step": 655
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 0.000837774294865706,
|
|
"loss": 1.9529,
|
|
"step": 660
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 0.0008352694275538723,
|
|
"loss": 2.0161,
|
|
"step": 665
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 0.0008327491776136267,
|
|
"loss": 1.942,
|
|
"step": 670
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 0.0008302136606775479,
|
|
"loss": 1.8917,
|
|
"step": 675
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 0.000827662993078685,
|
|
"loss": 1.9157,
|
|
"step": 680
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 0.0008250972918452207,
|
|
"loss": 1.9969,
|
|
"step": 685
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 0.0008225166746951022,
|
|
"loss": 1.8957,
|
|
"step": 690
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 0.0008199212600306396,
|
|
"loss": 1.8494,
|
|
"step": 695
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 0.0008173111669330733,
|
|
"loss": 1.8495,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 0.0008146865151571108,
|
|
"loss": 1.8407,
|
|
"step": 705
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 0.0008120474251254319,
|
|
"loss": 1.7461,
|
|
"step": 710
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 0.0008093940179231643,
|
|
"loss": 1.8608,
|
|
"step": 715
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 0.0008067264152923268,
|
|
"loss": 1.9029,
|
|
"step": 720
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 0.0008040447396262442,
|
|
"loss": 1.8156,
|
|
"step": 725
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 0.0008013491139639324,
|
|
"loss": 1.8412,
|
|
"step": 730
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 0.0007986396619844518,
|
|
"loss": 1.8522,
|
|
"step": 735
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 0.0007959165080012341,
|
|
"loss": 1.9613,
|
|
"step": 740
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 0.0007931797769563777,
|
|
"loss": 1.917,
|
|
"step": 745
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 0.0007904295944149156,
|
|
"loss": 1.8676,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 0.0007876660865590539,
|
|
"loss": 1.839,
|
|
"step": 755
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 0.0007848893801823833,
|
|
"loss": 1.8689,
|
|
"step": 760
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 0.0007820996026840606,
|
|
"loss": 1.9667,
|
|
"step": 765
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 0.000779296882062964,
|
|
"loss": 1.8947,
|
|
"step": 770
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 0.0007764813469118201,
|
|
"loss": 1.8329,
|
|
"step": 775
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 0.0007736531264113041,
|
|
"loss": 1.7263,
|
|
"step": 780
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 0.0007708123503241124,
|
|
"loss": 1.8701,
|
|
"step": 785
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"eval_loss": 1.8378170728683472,
|
|
"eval_runtime": 367.2627,
|
|
"eval_samples_per_second": 34.874,
|
|
"eval_steps_per_second": 2.181,
|
|
"step": 789
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 0.0007679591489890098,
|
|
"loss": 1.8124,
|
|
"step": 790
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 0.0007650936533148485,
|
|
"loss": 1.7062,
|
|
"step": 795
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 0.0007622159947745617,
|
|
"loss": 1.6682,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 0.0007593263053991325,
|
|
"loss": 1.6572,
|
|
"step": 805
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 0.0007564247177715349,
|
|
"loss": 1.6605,
|
|
"step": 810
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 0.0007535113650206517,
|
|
"loss": 1.7364,
|
|
"step": 815
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 0.0007505863808151661,
|
|
"loss": 1.628,
|
|
"step": 820
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 0.0007476498993574277,
|
|
"loss": 1.5975,
|
|
"step": 825
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 0.0007447020553772973,
|
|
"loss": 1.6411,
|
|
"step": 830
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 0.0007417429841259631,
|
|
"loss": 1.6513,
|
|
"step": 835
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 0.0007387728213697366,
|
|
"loss": 1.6971,
|
|
"step": 840
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 0.0007357917033838228,
|
|
"loss": 1.6577,
|
|
"step": 845
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 0.0007327997669460682,
|
|
"loss": 1.6068,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 0.0007297971493306848,
|
|
"loss": 1.5742,
|
|
"step": 855
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 0.0007267839883019515,
|
|
"loss": 1.6666,
|
|
"step": 860
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 0.0007237604221078942,
|
|
"loss": 1.7017,
|
|
"step": 865
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 0.000720726589473942,
|
|
"loss": 1.6573,
|
|
"step": 870
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 0.000717682629596563,
|
|
"loss": 1.4819,
|
|
"step": 875
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 0.0007146286821368767,
|
|
"loss": 1.6235,
|
|
"step": 880
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 0.0007115648872142474,
|
|
"loss": 1.6736,
|
|
"step": 885
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 0.0007084913853998542,
|
|
"loss": 1.6477,
|
|
"step": 890
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 0.0007054083177102423,
|
|
"loss": 1.6538,
|
|
"step": 895
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 0.0007023158256008521,
|
|
"loss": 1.5688,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 0.0006992140509595303,
|
|
"loss": 1.5525,
|
|
"step": 905
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 0.0006961031361000184,
|
|
"loss": 1.7425,
|
|
"step": 910
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 0.0006929832237554241,
|
|
"loss": 1.559,
|
|
"step": 915
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 0.0006898544570716722,
|
|
"loss": 1.5379,
|
|
"step": 920
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 0.0006867169796009369,
|
|
"loss": 1.586,
|
|
"step": 925
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 0.0006835709352950558,
|
|
"loss": 1.6424,
|
|
"step": 930
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 0.000680416468498924,
|
|
"loss": 1.6002,
|
|
"step": 935
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 0.0006772537239438732,
|
|
"loss": 1.5361,
|
|
"step": 940
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 0.0006740828467410294,
|
|
"loss": 1.5436,
|
|
"step": 945
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 0.0006709039823746564,
|
|
"loss": 1.5845,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 0.0006677172766954794,
|
|
"loss": 1.5275,
|
|
"step": 955
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 0.0006645228759139949,
|
|
"loss": 1.5879,
|
|
"step": 960
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 0.0006613209265937611,
|
|
"loss": 1.5809,
|
|
"step": 965
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 0.0006581115756446733,
|
|
"loss": 1.6111,
|
|
"step": 970
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 0.000654894970316224,
|
|
"loss": 1.5794,
|
|
"step": 975
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 0.0006516712581907471,
|
|
"loss": 1.5756,
|
|
"step": 980
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 0.000648440587176646,
|
|
"loss": 1.5913,
|
|
"step": 985
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 0.0006452031055016072,
|
|
"loss": 1.615,
|
|
"step": 990
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 0.0006419589617058008,
|
|
"loss": 1.6578,
|
|
"step": 995
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 0.000638708304635063,
|
|
"loss": 1.6285,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 0.0006354512834340694,
|
|
"loss": 1.5911,
|
|
"step": 1005
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 0.0006321880475394899,
|
|
"loss": 1.6383,
|
|
"step": 1010
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 0.0006289187466731334,
|
|
"loss": 1.5337,
|
|
"step": 1015
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 0.0006256435308350785,
|
|
"loss": 1.6334,
|
|
"step": 1020
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 0.0006223625502967903,
|
|
"loss": 1.7418,
|
|
"step": 1025
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 0.000619075955594227,
|
|
"loss": 1.5987,
|
|
"step": 1030
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 0.0006157838975209322,
|
|
"loss": 1.6512,
|
|
"step": 1035
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 0.0006124865271211165,
|
|
"loss": 1.68,
|
|
"step": 1040
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 0.0006091839956827278,
|
|
"loss": 1.5891,
|
|
"step": 1045
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 0.0006058764547305088,
|
|
"loss": 1.6131,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 0.0006025640560190467,
|
|
"loss": 1.6289,
|
|
"step": 1055
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 0.0005992469515258089,
|
|
"loss": 1.5907,
|
|
"step": 1060
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 0.0005959252934441706,
|
|
"loss": 1.6156,
|
|
"step": 1065
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 0.0005925992341764323,
|
|
"loss": 1.5184,
|
|
"step": 1070
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 0.000589268926326827,
|
|
"loss": 1.5729,
|
|
"step": 1075
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 0.0005859345226945177,
|
|
"loss": 1.5053,
|
|
"step": 1080
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 0.000582596176266589,
|
|
"loss": 1.6205,
|
|
"step": 1085
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 0.0005792540402110257,
|
|
"loss": 1.5363,
|
|
"step": 1090
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 0.0005759082678696857,
|
|
"loss": 1.6992,
|
|
"step": 1095
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 0.0005725590127512657,
|
|
"loss": 1.5543,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 0.0005692064285242558,
|
|
"loss": 1.6408,
|
|
"step": 1105
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 0.0005658506690098916,
|
|
"loss": 1.6324,
|
|
"step": 1110
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 0.0005624918881750943,
|
|
"loss": 1.5604,
|
|
"step": 1115
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 0.0005591302401254076,
|
|
"loss": 1.5888,
|
|
"step": 1120
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 0.000555765879097928,
|
|
"loss": 1.661,
|
|
"step": 1125
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 0.0005523989594542258,
|
|
"loss": 1.6155,
|
|
"step": 1130
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 0.0005490296356732649,
|
|
"loss": 1.5069,
|
|
"step": 1135
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 0.0005456580623443145,
|
|
"loss": 1.4833,
|
|
"step": 1140
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 0.000542284394159856,
|
|
"loss": 1.4966,
|
|
"step": 1145
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 0.0005389087859084854,
|
|
"loss": 1.6364,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 0.0005355313924678123,
|
|
"loss": 1.6311,
|
|
"step": 1155
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 0.0005321523687973526,
|
|
"loss": 1.5597,
|
|
"step": 1160
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 0.00052877186993142,
|
|
"loss": 1.6759,
|
|
"step": 1165
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 0.0005253900509720118,
|
|
"loss": 1.5734,
|
|
"step": 1170
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 0.0005220070670816932,
|
|
"loss": 1.6767,
|
|
"step": 1175
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 0.0005186230734764782,
|
|
"loss": 1.603,
|
|
"step": 1180
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 0.0005152382254187076,
|
|
"loss": 1.5121,
|
|
"step": 1185
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 0.0005118526782099258,
|
|
"loss": 1.5978,
|
|
"step": 1190
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 0.0005084665871837553,
|
|
"loss": 1.4273,
|
|
"step": 1195
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 0.0005050801076987699,
|
|
"loss": 1.5823,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 0.0005016933951313659,
|
|
"loss": 1.6582,
|
|
"step": 1205
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 0.0004983066048686342,
|
|
"loss": 1.5263,
|
|
"step": 1210
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 0.0004949198923012302,
|
|
"loss": 1.6287,
|
|
"step": 1215
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 0.0004915334128162448,
|
|
"loss": 1.5244,
|
|
"step": 1220
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 0.00048814732179007443,
|
|
"loss": 1.5054,
|
|
"step": 1225
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 0.0004847617745812925,
|
|
"loss": 1.5065,
|
|
"step": 1230
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 0.00048137692652352196,
|
|
"loss": 1.649,
|
|
"step": 1235
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 0.0004779929329183068,
|
|
"loss": 1.5417,
|
|
"step": 1240
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 0.0004746099490279883,
|
|
"loss": 1.4152,
|
|
"step": 1245
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 0.00047122813006858017,
|
|
"loss": 1.5171,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 0.0004678476312026475,
|
|
"loss": 1.5013,
|
|
"step": 1255
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 0.00046446860753218797,
|
|
"loss": 1.5518,
|
|
"step": 1260
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 0.0004610912140915146,
|
|
"loss": 1.5121,
|
|
"step": 1265
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 0.00045771560584014414,
|
|
"loss": 1.4661,
|
|
"step": 1270
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 0.0004543419376556855,
|
|
"loss": 1.5079,
|
|
"step": 1275
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 0.00045097036432673515,
|
|
"loss": 1.5807,
|
|
"step": 1280
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 0.0004476010405457744,
|
|
"loss": 1.5803,
|
|
"step": 1285
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 0.00044423412090207216,
|
|
"loss": 1.5011,
|
|
"step": 1290
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 0.0004408697598745924,
|
|
"loss": 1.5691,
|
|
"step": 1295
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 0.0004375081118249058,
|
|
"loss": 1.5332,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 0.0004341493309901085,
|
|
"loss": 1.5385,
|
|
"step": 1305
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 0.00043079357147574434,
|
|
"loss": 1.6153,
|
|
"step": 1310
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 0.0004274409872487345,
|
|
"loss": 1.6695,
|
|
"step": 1315
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 0.0004240917321303143,
|
|
"loss": 1.5654,
|
|
"step": 1320
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 0.0004207459597889744,
|
|
"loss": 1.5585,
|
|
"step": 1325
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 0.000417403823733411,
|
|
"loss": 1.5072,
|
|
"step": 1330
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 0.0004140654773054824,
|
|
"loss": 1.5723,
|
|
"step": 1335
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 0.00041073107367317315,
|
|
"loss": 1.6242,
|
|
"step": 1340
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 0.0004074007658235678,
|
|
"loss": 1.5305,
|
|
"step": 1345
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 0.00040407470655582934,
|
|
"loss": 1.5084,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 0.00040075304847419114,
|
|
"loss": 1.5424,
|
|
"step": 1355
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 0.0003974359439809534,
|
|
"loss": 1.5136,
|
|
"step": 1360
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 0.0003941235452694913,
|
|
"loss": 1.5349,
|
|
"step": 1365
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 0.0003908160043172725,
|
|
"loss": 1.4194,
|
|
"step": 1370
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 0.0003875134728788834,
|
|
"loss": 1.6139,
|
|
"step": 1375
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 0.0003842161024790679,
|
|
"loss": 1.4876,
|
|
"step": 1380
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 0.00038092404440577315,
|
|
"loss": 1.5042,
|
|
"step": 1385
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 0.0003776374497032097,
|
|
"loss": 1.5229,
|
|
"step": 1390
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 0.0003743564691649216,
|
|
"loss": 1.5647,
|
|
"step": 1395
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 0.0003710812533268666,
|
|
"loss": 1.5142,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 0.0003678119524605102,
|
|
"loss": 1.5012,
|
|
"step": 1405
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 0.0003645487165659305,
|
|
"loss": 1.4997,
|
|
"step": 1410
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 0.000361291695364937,
|
|
"loss": 1.3816,
|
|
"step": 1415
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 0.00035804103829419943,
|
|
"loss": 1.4572,
|
|
"step": 1420
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 0.0003547968944983927,
|
|
"loss": 1.5515,
|
|
"step": 1425
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 0.00035155941282335415,
|
|
"loss": 1.6266,
|
|
"step": 1430
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 0.0003483287418092528,
|
|
"loss": 1.4554,
|
|
"step": 1435
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 0.000345105029683776,
|
|
"loss": 1.4924,
|
|
"step": 1440
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 0.00034188842435532686,
|
|
"loss": 1.4861,
|
|
"step": 1445
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 0.000338679073406239,
|
|
"loss": 1.4455,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 0.0003354771240860052,
|
|
"loss": 1.5319,
|
|
"step": 1455
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 0.0003322827233045206,
|
|
"loss": 1.5121,
|
|
"step": 1460
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 0.0003290960176253438,
|
|
"loss": 1.4594,
|
|
"step": 1465
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 0.0003259171532589707,
|
|
"loss": 1.5755,
|
|
"step": 1470
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 0.00032274627605612685,
|
|
"loss": 1.6141,
|
|
"step": 1475
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 0.00031958353150107597,
|
|
"loss": 1.4925,
|
|
"step": 1480
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 0.0003164290647049443,
|
|
"loss": 1.5114,
|
|
"step": 1485
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 0.0003132830203990631,
|
|
"loss": 1.4687,
|
|
"step": 1490
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 0.00031014554292832795,
|
|
"loss": 1.392,
|
|
"step": 1495
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 0.0003070167762445759,
|
|
"loss": 1.4322,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 0.0003038968638999816,
|
|
"loss": 1.5359,
|
|
"step": 1505
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 0.00030078594904046967,
|
|
"loss": 1.5745,
|
|
"step": 1510
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 0.0002976841743991479,
|
|
"loss": 1.4865,
|
|
"step": 1515
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 0.0002945916822897579,
|
|
"loss": 1.4958,
|
|
"step": 1520
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 0.0002915086146001458,
|
|
"loss": 1.4678,
|
|
"step": 1525
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 0.0002884351127857526,
|
|
"loss": 1.4629,
|
|
"step": 1530
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 0.00028537131786312333,
|
|
"loss": 1.5381,
|
|
"step": 1535
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 0.00028231737040343717,
|
|
"loss": 1.5322,
|
|
"step": 1540
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 0.000279273410526058,
|
|
"loss": 1.4523,
|
|
"step": 1545
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 0.000276239577892106,
|
|
"loss": 1.552,
|
|
"step": 1550
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 0.00027321601169804873,
|
|
"loss": 1.4431,
|
|
"step": 1555
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 0.00027020285066931525,
|
|
"loss": 1.5036,
|
|
"step": 1560
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 0.0002672002330539318,
|
|
"loss": 1.5313,
|
|
"step": 1565
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 0.0002642082966161772,
|
|
"loss": 1.5092,
|
|
"step": 1570
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"learning_rate": 0.0002612271786302636,
|
|
"loss": 1.5065,
|
|
"step": 1575
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"eval_loss": 1.6175580024719238,
|
|
"eval_runtime": 366.4829,
|
|
"eval_samples_per_second": 34.948,
|
|
"eval_steps_per_second": 2.186,
|
|
"step": 1578
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"learning_rate": 0.0002582570158740372,
|
|
"loss": 1.5297,
|
|
"step": 1580
|
|
},
|
|
{
|
|
"epoch": 2.01,
|
|
"learning_rate": 0.00025529794462270285,
|
|
"loss": 1.3664,
|
|
"step": 1585
|
|
},
|
|
{
|
|
"epoch": 2.02,
|
|
"learning_rate": 0.0002523501006425724,
|
|
"loss": 1.2522,
|
|
"step": 1590
|
|
},
|
|
{
|
|
"epoch": 2.02,
|
|
"learning_rate": 0.000249413619184834,
|
|
"loss": 1.2119,
|
|
"step": 1595
|
|
},
|
|
{
|
|
"epoch": 2.03,
|
|
"learning_rate": 0.00024648863497934824,
|
|
"loss": 1.3337,
|
|
"step": 1600
|
|
},
|
|
{
|
|
"epoch": 2.03,
|
|
"learning_rate": 0.00024357528222846515,
|
|
"loss": 1.2604,
|
|
"step": 1605
|
|
},
|
|
{
|
|
"epoch": 2.04,
|
|
"learning_rate": 0.00024067369460086769,
|
|
"loss": 1.2815,
|
|
"step": 1610
|
|
},
|
|
{
|
|
"epoch": 2.05,
|
|
"learning_rate": 0.00023778400522543846,
|
|
"loss": 1.2383,
|
|
"step": 1615
|
|
},
|
|
{
|
|
"epoch": 2.05,
|
|
"learning_rate": 0.00023490634668515154,
|
|
"loss": 1.2959,
|
|
"step": 1620
|
|
},
|
|
{
|
|
"epoch": 2.06,
|
|
"learning_rate": 0.00023204085101099015,
|
|
"loss": 1.3136,
|
|
"step": 1625
|
|
},
|
|
{
|
|
"epoch": 2.07,
|
|
"learning_rate": 0.00022918764967588763,
|
|
"loss": 1.2596,
|
|
"step": 1630
|
|
},
|
|
{
|
|
"epoch": 2.07,
|
|
"learning_rate": 0.00022634687358869595,
|
|
"loss": 1.272,
|
|
"step": 1635
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"learning_rate": 0.00022351865308817991,
|
|
"loss": 1.218,
|
|
"step": 1640
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"learning_rate": 0.00022070311793703608,
|
|
"loss": 1.3847,
|
|
"step": 1645
|
|
},
|
|
{
|
|
"epoch": 2.09,
|
|
"learning_rate": 0.00021790039731593946,
|
|
"loss": 1.2601,
|
|
"step": 1650
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"learning_rate": 0.00021511061981761666,
|
|
"loss": 1.3791,
|
|
"step": 1655
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"learning_rate": 0.0002123339134409461,
|
|
"loss": 1.2843,
|
|
"step": 1660
|
|
},
|
|
{
|
|
"epoch": 2.11,
|
|
"learning_rate": 0.0002095704055850846,
|
|
"loss": 1.2707,
|
|
"step": 1665
|
|
},
|
|
{
|
|
"epoch": 2.12,
|
|
"learning_rate": 0.00020682022304362226,
|
|
"loss": 1.2781,
|
|
"step": 1670
|
|
},
|
|
{
|
|
"epoch": 2.12,
|
|
"learning_rate": 0.00020408349199876587,
|
|
"loss": 1.3632,
|
|
"step": 1675
|
|
},
|
|
{
|
|
"epoch": 2.13,
|
|
"learning_rate": 0.0002013603380155482,
|
|
"loss": 1.2773,
|
|
"step": 1680
|
|
},
|
|
{
|
|
"epoch": 2.14,
|
|
"learning_rate": 0.00019865088603606773,
|
|
"loss": 1.2757,
|
|
"step": 1685
|
|
},
|
|
{
|
|
"epoch": 2.14,
|
|
"learning_rate": 0.00019595526037375588,
|
|
"loss": 1.299,
|
|
"step": 1690
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 0.00019327358470767325,
|
|
"loss": 1.2752,
|
|
"step": 1695
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 0.0001906059820768357,
|
|
"loss": 1.3311,
|
|
"step": 1700
|
|
},
|
|
{
|
|
"epoch": 2.16,
|
|
"learning_rate": 0.00018795257487456797,
|
|
"loss": 1.2696,
|
|
"step": 1705
|
|
},
|
|
{
|
|
"epoch": 2.17,
|
|
"learning_rate": 0.00018531348484288925,
|
|
"loss": 1.2492,
|
|
"step": 1710
|
|
},
|
|
{
|
|
"epoch": 2.17,
|
|
"learning_rate": 0.00018268883306692674,
|
|
"loss": 1.3153,
|
|
"step": 1715
|
|
},
|
|
{
|
|
"epoch": 2.18,
|
|
"learning_rate": 0.0001800787399693604,
|
|
"loss": 1.274,
|
|
"step": 1720
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"learning_rate": 0.0001774833253048978,
|
|
"loss": 1.3302,
|
|
"step": 1725
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"learning_rate": 0.0001749027081547792,
|
|
"loss": 1.1726,
|
|
"step": 1730
|
|
},
|
|
{
|
|
"epoch": 2.2,
|
|
"learning_rate": 0.0001723370069213151,
|
|
"loss": 1.2367,
|
|
"step": 1735
|
|
},
|
|
{
|
|
"epoch": 2.21,
|
|
"learning_rate": 0.0001697863393224522,
|
|
"loss": 1.2647,
|
|
"step": 1740
|
|
},
|
|
{
|
|
"epoch": 2.21,
|
|
"learning_rate": 0.0001672508223863732,
|
|
"loss": 1.2689,
|
|
"step": 1745
|
|
},
|
|
{
|
|
"epoch": 2.22,
|
|
"learning_rate": 0.00016473057244612778,
|
|
"loss": 1.2523,
|
|
"step": 1750
|
|
},
|
|
{
|
|
"epoch": 2.22,
|
|
"learning_rate": 0.00016222570513429403,
|
|
"loss": 1.2597,
|
|
"step": 1755
|
|
},
|
|
{
|
|
"epoch": 2.23,
|
|
"learning_rate": 0.00015973633537767396,
|
|
"loss": 1.2528,
|
|
"step": 1760
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"learning_rate": 0.00015726257739201994,
|
|
"loss": 1.2704,
|
|
"step": 1765
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"learning_rate": 0.0001548045446767945,
|
|
"loss": 1.1525,
|
|
"step": 1770
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 0.0001523623500099633,
|
|
"loss": 1.2283,
|
|
"step": 1775
|
|
},
|
|
{
|
|
"epoch": 2.26,
|
|
"learning_rate": 0.0001499361054428196,
|
|
"loss": 1.3046,
|
|
"step": 1780
|
|
},
|
|
{
|
|
"epoch": 2.26,
|
|
"learning_rate": 0.00014752592229484463,
|
|
"loss": 1.3047,
|
|
"step": 1785
|
|
},
|
|
{
|
|
"epoch": 2.27,
|
|
"learning_rate": 0.00014513191114859869,
|
|
"loss": 1.2188,
|
|
"step": 1790
|
|
},
|
|
{
|
|
"epoch": 2.27,
|
|
"learning_rate": 0.0001427541818446484,
|
|
"loss": 1.2769,
|
|
"step": 1795
|
|
},
|
|
{
|
|
"epoch": 2.28,
|
|
"learning_rate": 0.0001403928434765267,
|
|
"loss": 1.2688,
|
|
"step": 1800
|
|
},
|
|
{
|
|
"epoch": 2.29,
|
|
"learning_rate": 0.00013804800438572745,
|
|
"loss": 1.2727,
|
|
"step": 1805
|
|
},
|
|
{
|
|
"epoch": 2.29,
|
|
"learning_rate": 0.00013571977215673508,
|
|
"loss": 1.1562,
|
|
"step": 1810
|
|
},
|
|
{
|
|
"epoch": 2.3,
|
|
"learning_rate": 0.00013340825361208758,
|
|
"loss": 1.3305,
|
|
"step": 1815
|
|
},
|
|
{
|
|
"epoch": 2.31,
|
|
"learning_rate": 0.00013111355480747622,
|
|
"loss": 1.2663,
|
|
"step": 1820
|
|
},
|
|
{
|
|
"epoch": 2.31,
|
|
"learning_rate": 0.0001288357810268791,
|
|
"loss": 1.2736,
|
|
"step": 1825
|
|
},
|
|
{
|
|
"epoch": 2.32,
|
|
"learning_rate": 0.00012657503677772992,
|
|
"loss": 1.2877,
|
|
"step": 1830
|
|
},
|
|
{
|
|
"epoch": 2.33,
|
|
"learning_rate": 0.00012433142578612493,
|
|
"loss": 1.3585,
|
|
"step": 1835
|
|
},
|
|
{
|
|
"epoch": 2.33,
|
|
"learning_rate": 0.00012210505099206177,
|
|
"loss": 1.3242,
|
|
"step": 1840
|
|
},
|
|
{
|
|
"epoch": 2.34,
|
|
"learning_rate": 0.00011989601454471788,
|
|
"loss": 1.3037,
|
|
"step": 1845
|
|
},
|
|
{
|
|
"epoch": 2.34,
|
|
"learning_rate": 0.00011770441779776281,
|
|
"loss": 1.2969,
|
|
"step": 1850
|
|
},
|
|
{
|
|
"epoch": 2.35,
|
|
"learning_rate": 0.0001155303613047084,
|
|
"loss": 1.2783,
|
|
"step": 1855
|
|
},
|
|
{
|
|
"epoch": 2.36,
|
|
"learning_rate": 0.00011337394481429563,
|
|
"loss": 1.2409,
|
|
"step": 1860
|
|
},
|
|
{
|
|
"epoch": 2.36,
|
|
"learning_rate": 0.00011123526726591682,
|
|
"loss": 1.2276,
|
|
"step": 1865
|
|
},
|
|
{
|
|
"epoch": 2.37,
|
|
"learning_rate": 0.00010911442678507771,
|
|
"loss": 1.261,
|
|
"step": 1870
|
|
},
|
|
{
|
|
"epoch": 2.38,
|
|
"learning_rate": 0.00010701152067889408,
|
|
"loss": 1.2741,
|
|
"step": 1875
|
|
},
|
|
{
|
|
"epoch": 2.38,
|
|
"learning_rate": 0.0001049266454316275,
|
|
"loss": 1.2337,
|
|
"step": 1880
|
|
},
|
|
{
|
|
"epoch": 2.39,
|
|
"learning_rate": 0.00010285989670025914,
|
|
"loss": 1.2259,
|
|
"step": 1885
|
|
},
|
|
{
|
|
"epoch": 2.4,
|
|
"learning_rate": 0.00010081136931009982,
|
|
"loss": 1.3068,
|
|
"step": 1890
|
|
},
|
|
{
|
|
"epoch": 2.4,
|
|
"learning_rate": 9.878115725044045e-05,
|
|
"loss": 1.2444,
|
|
"step": 1895
|
|
},
|
|
{
|
|
"epoch": 2.41,
|
|
"learning_rate": 9.676935367023826e-05,
|
|
"loss": 1.2301,
|
|
"step": 1900
|
|
},
|
|
{
|
|
"epoch": 2.41,
|
|
"learning_rate": 9.477605087384428e-05,
|
|
"loss": 1.2882,
|
|
"step": 1905
|
|
},
|
|
{
|
|
"epoch": 2.42,
|
|
"learning_rate": 9.280134031676795e-05,
|
|
"loss": 1.2423,
|
|
"step": 1910
|
|
},
|
|
{
|
|
"epoch": 2.43,
|
|
"learning_rate": 9.084531260148043e-05,
|
|
"loss": 1.2471,
|
|
"step": 1915
|
|
},
|
|
{
|
|
"epoch": 2.43,
|
|
"learning_rate": 8.890805747325864e-05,
|
|
"loss": 1.2215,
|
|
"step": 1920
|
|
},
|
|
{
|
|
"epoch": 2.44,
|
|
"learning_rate": 8.698966381606633e-05,
|
|
"loss": 1.359,
|
|
"step": 1925
|
|
},
|
|
{
|
|
"epoch": 2.45,
|
|
"learning_rate": 8.509021964847752e-05,
|
|
"loss": 1.2103,
|
|
"step": 1930
|
|
},
|
|
{
|
|
"epoch": 2.45,
|
|
"learning_rate": 8.320981211963647e-05,
|
|
"loss": 1.291,
|
|
"step": 1935
|
|
},
|
|
{
|
|
"epoch": 2.46,
|
|
"learning_rate": 8.134852750526017e-05,
|
|
"loss": 1.2588,
|
|
"step": 1940
|
|
},
|
|
{
|
|
"epoch": 2.47,
|
|
"learning_rate": 7.950645120368011e-05,
|
|
"loss": 1.3012,
|
|
"step": 1945
|
|
},
|
|
{
|
|
"epoch": 2.47,
|
|
"learning_rate": 7.768366773192286e-05,
|
|
"loss": 1.2732,
|
|
"step": 1950
|
|
},
|
|
{
|
|
"epoch": 2.48,
|
|
"learning_rate": 7.58802607218339e-05,
|
|
"loss": 1.2364,
|
|
"step": 1955
|
|
},
|
|
{
|
|
"epoch": 2.48,
|
|
"learning_rate": 7.409631291623903e-05,
|
|
"loss": 1.3049,
|
|
"step": 1960
|
|
},
|
|
{
|
|
"epoch": 2.49,
|
|
"learning_rate": 7.23319061651489e-05,
|
|
"loss": 1.334,
|
|
"step": 1965
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 7.058712142200324e-05,
|
|
"loss": 1.1297,
|
|
"step": 1970
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 6.886203873995672e-05,
|
|
"loss": 1.3076,
|
|
"step": 1975
|
|
},
|
|
{
|
|
"epoch": 2.51,
|
|
"learning_rate": 6.715673726820626e-05,
|
|
"loss": 1.2343,
|
|
"step": 1980
|
|
},
|
|
{
|
|
"epoch": 2.52,
|
|
"learning_rate": 6.547129524835904e-05,
|
|
"loss": 1.2474,
|
|
"step": 1985
|
|
},
|
|
{
|
|
"epoch": 2.52,
|
|
"learning_rate": 6.38057900108428e-05,
|
|
"loss": 1.2312,
|
|
"step": 1990
|
|
},
|
|
{
|
|
"epoch": 2.53,
|
|
"learning_rate": 6.216029797135842e-05,
|
|
"loss": 1.1307,
|
|
"step": 1995
|
|
},
|
|
{
|
|
"epoch": 2.53,
|
|
"learning_rate": 6.053489462737283e-05,
|
|
"loss": 1.2532,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"epoch": 2.54,
|
|
"learning_rate": 5.8929654554655985e-05,
|
|
"loss": 1.2598,
|
|
"step": 2005
|
|
},
|
|
{
|
|
"epoch": 2.55,
|
|
"learning_rate": 5.734465140385864e-05,
|
|
"loss": 1.1858,
|
|
"step": 2010
|
|
},
|
|
{
|
|
"epoch": 2.55,
|
|
"learning_rate": 5.5779957897133415e-05,
|
|
"loss": 1.225,
|
|
"step": 2015
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"learning_rate": 5.423564582479845e-05,
|
|
"loss": 1.2459,
|
|
"step": 2020
|
|
},
|
|
{
|
|
"epoch": 2.57,
|
|
"learning_rate": 5.2711786042042854e-05,
|
|
"loss": 1.1665,
|
|
"step": 2025
|
|
},
|
|
{
|
|
"epoch": 2.57,
|
|
"learning_rate": 5.120844846567657e-05,
|
|
"loss": 1.2943,
|
|
"step": 2030
|
|
},
|
|
{
|
|
"epoch": 2.58,
|
|
"learning_rate": 4.972570207092186e-05,
|
|
"loss": 1.3287,
|
|
"step": 2035
|
|
},
|
|
{
|
|
"epoch": 2.59,
|
|
"learning_rate": 4.826361488824898e-05,
|
|
"loss": 1.2245,
|
|
"step": 2040
|
|
},
|
|
{
|
|
"epoch": 2.59,
|
|
"learning_rate": 4.68222540002548e-05,
|
|
"loss": 1.1462,
|
|
"step": 2045
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"learning_rate": 4.5401685538584756e-05,
|
|
"loss": 1.2929,
|
|
"step": 2050
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"learning_rate": 4.400197468089906e-05,
|
|
"loss": 1.2585,
|
|
"step": 2055
|
|
},
|
|
{
|
|
"epoch": 2.61,
|
|
"learning_rate": 4.26231856478817e-05,
|
|
"loss": 1.1936,
|
|
"step": 2060
|
|
},
|
|
{
|
|
"epoch": 2.62,
|
|
"learning_rate": 4.126538170029453e-05,
|
|
"loss": 1.3289,
|
|
"step": 2065
|
|
},
|
|
{
|
|
"epoch": 2.62,
|
|
"learning_rate": 3.99286251360742e-05,
|
|
"loss": 1.2575,
|
|
"step": 2070
|
|
},
|
|
{
|
|
"epoch": 2.63,
|
|
"learning_rate": 3.8612977287474093e-05,
|
|
"loss": 1.2051,
|
|
"step": 2075
|
|
},
|
|
{
|
|
"epoch": 2.64,
|
|
"learning_rate": 3.731849851825042e-05,
|
|
"loss": 1.265,
|
|
"step": 2080
|
|
},
|
|
{
|
|
"epoch": 2.64,
|
|
"learning_rate": 3.604524822089245e-05,
|
|
"loss": 1.273,
|
|
"step": 2085
|
|
},
|
|
{
|
|
"epoch": 2.65,
|
|
"learning_rate": 3.479328481389771e-05,
|
|
"loss": 1.2193,
|
|
"step": 2090
|
|
},
|
|
{
|
|
"epoch": 2.66,
|
|
"learning_rate": 3.35626657390915e-05,
|
|
"loss": 1.3077,
|
|
"step": 2095
|
|
},
|
|
{
|
|
"epoch": 2.66,
|
|
"learning_rate": 3.235344745899116e-05,
|
|
"loss": 1.177,
|
|
"step": 2100
|
|
},
|
|
{
|
|
"epoch": 2.67,
|
|
"learning_rate": 3.1165685454216333e-05,
|
|
"loss": 1.317,
|
|
"step": 2105
|
|
},
|
|
{
|
|
"epoch": 2.67,
|
|
"learning_rate": 2.9999434220942255e-05,
|
|
"loss": 1.2496,
|
|
"step": 2110
|
|
},
|
|
{
|
|
"epoch": 2.68,
|
|
"learning_rate": 2.8854747268400318e-05,
|
|
"loss": 1.2356,
|
|
"step": 2115
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"learning_rate": 2.773167711642266e-05,
|
|
"loss": 1.2738,
|
|
"step": 2120
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"learning_rate": 2.663027529303236e-05,
|
|
"loss": 1.2312,
|
|
"step": 2125
|
|
},
|
|
{
|
|
"epoch": 2.7,
|
|
"learning_rate": 2.5550592332079515e-05,
|
|
"loss": 1.2528,
|
|
"step": 2130
|
|
},
|
|
{
|
|
"epoch": 2.71,
|
|
"learning_rate": 2.449267777092251e-05,
|
|
"loss": 1.1745,
|
|
"step": 2135
|
|
},
|
|
{
|
|
"epoch": 2.71,
|
|
"learning_rate": 2.3456580148155272e-05,
|
|
"loss": 1.2588,
|
|
"step": 2140
|
|
},
|
|
{
|
|
"epoch": 2.72,
|
|
"learning_rate": 2.2442347001380148e-05,
|
|
"loss": 1.1721,
|
|
"step": 2145
|
|
},
|
|
{
|
|
"epoch": 2.72,
|
|
"learning_rate": 2.145002486502684e-05,
|
|
"loss": 1.2337,
|
|
"step": 2150
|
|
},
|
|
{
|
|
"epoch": 2.73,
|
|
"learning_rate": 2.0479659268217378e-05,
|
|
"loss": 1.2546,
|
|
"step": 2155
|
|
},
|
|
{
|
|
"epoch": 2.74,
|
|
"learning_rate": 1.953129473267723e-05,
|
|
"loss": 1.2842,
|
|
"step": 2160
|
|
},
|
|
{
|
|
"epoch": 2.74,
|
|
"learning_rate": 1.8604974770692508e-05,
|
|
"loss": 1.1871,
|
|
"step": 2165
|
|
},
|
|
{
|
|
"epoch": 2.75,
|
|
"learning_rate": 1.770074188311349e-05,
|
|
"loss": 1.1609,
|
|
"step": 2170
|
|
},
|
|
{
|
|
"epoch": 2.76,
|
|
"learning_rate": 1.681863755740487e-05,
|
|
"loss": 1.1607,
|
|
"step": 2175
|
|
},
|
|
{
|
|
"epoch": 2.76,
|
|
"learning_rate": 1.5958702265741977e-05,
|
|
"loss": 1.2254,
|
|
"step": 2180
|
|
},
|
|
{
|
|
"epoch": 2.77,
|
|
"learning_rate": 1.512097546315394e-05,
|
|
"loss": 1.2322,
|
|
"step": 2185
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"learning_rate": 1.430549558571359e-05,
|
|
"loss": 1.1937,
|
|
"step": 2190
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"learning_rate": 1.3512300048773685e-05,
|
|
"loss": 1.2344,
|
|
"step": 2195
|
|
},
|
|
{
|
|
"epoch": 2.79,
|
|
"learning_rate": 1.2741425245250626e-05,
|
|
"loss": 1.2383,
|
|
"step": 2200
|
|
},
|
|
{
|
|
"epoch": 2.79,
|
|
"learning_rate": 1.199290654395424e-05,
|
|
"loss": 1.2553,
|
|
"step": 2205
|
|
},
|
|
{
|
|
"epoch": 2.8,
|
|
"learning_rate": 1.1266778287965297e-05,
|
|
"loss": 1.381,
|
|
"step": 2210
|
|
},
|
|
{
|
|
"epoch": 2.81,
|
|
"learning_rate": 1.056307379305993e-05,
|
|
"loss": 1.1926,
|
|
"step": 2215
|
|
},
|
|
{
|
|
"epoch": 2.81,
|
|
"learning_rate": 9.88182534618054e-06,
|
|
"loss": 1.1419,
|
|
"step": 2220
|
|
},
|
|
{
|
|
"epoch": 2.82,
|
|
"learning_rate": 9.223064203955078e-06,
|
|
"loss": 1.2016,
|
|
"step": 2225
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"learning_rate": 8.586820591262367e-06,
|
|
"loss": 1.2364,
|
|
"step": 2230
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"learning_rate": 7.973123699845819e-06,
|
|
"loss": 1.2936,
|
|
"step": 2235
|
|
},
|
|
{
|
|
"epoch": 2.84,
|
|
"learning_rate": 7.382001686973782e-06,
|
|
"loss": 1.2675,
|
|
"step": 2240
|
|
},
|
|
{
|
|
"epoch": 2.85,
|
|
"learning_rate": 6.8134816741476364e-06,
|
|
"loss": 1.2742,
|
|
"step": 2245
|
|
},
|
|
{
|
|
"epoch": 2.85,
|
|
"learning_rate": 6.267589745857727e-06,
|
|
"loss": 1.2404,
|
|
"step": 2250
|
|
},
|
|
{
|
|
"epoch": 2.86,
|
|
"learning_rate": 5.744350948386101e-06,
|
|
"loss": 1.311,
|
|
"step": 2255
|
|
},
|
|
{
|
|
"epoch": 2.86,
|
|
"learning_rate": 5.243789288657818e-06,
|
|
"loss": 1.2043,
|
|
"step": 2260
|
|
},
|
|
{
|
|
"epoch": 2.87,
|
|
"learning_rate": 4.765927733139108e-06,
|
|
"loss": 1.2024,
|
|
"step": 2265
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"learning_rate": 4.310788206783822e-06,
|
|
"loss": 1.3395,
|
|
"step": 2270
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"learning_rate": 3.87839159202763e-06,
|
|
"loss": 1.2397,
|
|
"step": 2275
|
|
},
|
|
{
|
|
"epoch": 2.89,
|
|
"learning_rate": 3.4687577278295635e-06,
|
|
"loss": 1.2176,
|
|
"step": 2280
|
|
},
|
|
{
|
|
"epoch": 2.9,
|
|
"learning_rate": 3.0819054087619093e-06,
|
|
"loss": 1.1944,
|
|
"step": 2285
|
|
},
|
|
{
|
|
"epoch": 2.9,
|
|
"learning_rate": 2.7178523841481227e-06,
|
|
"loss": 1.2135,
|
|
"step": 2290
|
|
},
|
|
{
|
|
"epoch": 2.91,
|
|
"learning_rate": 2.3766153572480353e-06,
|
|
"loss": 1.259,
|
|
"step": 2295
|
|
},
|
|
{
|
|
"epoch": 2.91,
|
|
"learning_rate": 2.05820998449191e-06,
|
|
"loss": 1.1943,
|
|
"step": 2300
|
|
},
|
|
{
|
|
"epoch": 2.92,
|
|
"learning_rate": 1.7626508747617954e-06,
|
|
"loss": 1.2961,
|
|
"step": 2305
|
|
},
|
|
{
|
|
"epoch": 2.93,
|
|
"learning_rate": 1.4899515887213943e-06,
|
|
"loss": 1.157,
|
|
"step": 2310
|
|
},
|
|
{
|
|
"epoch": 2.93,
|
|
"learning_rate": 1.2401246381938958e-06,
|
|
"loss": 1.1454,
|
|
"step": 2315
|
|
},
|
|
{
|
|
"epoch": 2.94,
|
|
"learning_rate": 1.0131814855877663e-06,
|
|
"loss": 1.1584,
|
|
"step": 2320
|
|
},
|
|
{
|
|
"epoch": 2.95,
|
|
"learning_rate": 8.091325433710606e-07,
|
|
"loss": 1.1784,
|
|
"step": 2325
|
|
},
|
|
{
|
|
"epoch": 2.95,
|
|
"learning_rate": 6.27987173593525e-07,
|
|
"loss": 1.2977,
|
|
"step": 2330
|
|
},
|
|
{
|
|
"epoch": 2.96,
|
|
"learning_rate": 4.6975368745705296e-07,
|
|
"loss": 1.1467,
|
|
"step": 2335
|
|
},
|
|
{
|
|
"epoch": 2.97,
|
|
"learning_rate": 3.3443934493443407e-07,
|
|
"loss": 1.2271,
|
|
"step": 2340
|
|
},
|
|
{
|
|
"epoch": 2.97,
|
|
"learning_rate": 2.2205035443623178e-07,
|
|
"loss": 1.2055,
|
|
"step": 2345
|
|
},
|
|
{
|
|
"epoch": 2.98,
|
|
"learning_rate": 1.325918725258446e-07,
|
|
"loss": 1.166,
|
|
"step": 2350
|
|
},
|
|
{
|
|
"epoch": 2.98,
|
|
"learning_rate": 6.60680036831396e-08,
|
|
"loss": 1.2394,
|
|
"step": 2355
|
|
},
|
|
{
|
|
"epoch": 2.99,
|
|
"learning_rate": 2.248180011588108e-08,
|
|
"loss": 1.2422,
|
|
"step": 2360
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"learning_rate": 1.8352616197314652e-09,
|
|
"loss": 1.1873,
|
|
"step": 2365
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"eval_loss": 1.5839784145355225,
|
|
"eval_runtime": 366.1985,
|
|
"eval_samples_per_second": 34.976,
|
|
"eval_steps_per_second": 2.187,
|
|
"step": 2367
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"step": 2367,
|
|
"total_flos": 5.627889061063557e+17,
|
|
"train_loss": 1.6624892393546513,
|
|
"train_runtime": 33680.2242,
|
|
"train_samples_per_second": 8.997,
|
|
"train_steps_per_second": 0.07
|
|
}
|
|
],
|
|
"max_steps": 2367,
|
|
"num_train_epochs": 3,
|
|
"total_flos": 5.627889061063557e+17,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|