generated from xuyuqing/ailab
1020 lines
20 KiB
JSON
1020 lines
20 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 4.988272087568413,
|
|
"global_step": 1595,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 8.333333333333334e-06,
|
|
"loss": 4.054,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 1.8750000000000002e-05,
|
|
"loss": 3.5891,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 2.916666666666667e-05,
|
|
"loss": 3.1017,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.958333333333333e-05,
|
|
"loss": 2.8749,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 2.7578,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.999484517168163e-05,
|
|
"loss": 2.6894,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.997938281250692e-05,
|
|
"loss": 2.5445,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.9953619298940405e-05,
|
|
"loss": 2.587,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.991756525550125e-05,
|
|
"loss": 2.5327,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.9871235550381775e-05,
|
|
"loss": 2.5389,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.981464928931607e-05,
|
|
"loss": 2.5013,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.9747829807701e-05,
|
|
"loss": 2.4917,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.967080466097304e-05,
|
|
"loss": 2.5222,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.958360561324481e-05,
|
|
"loss": 2.4736,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.9486268624205955e-05,
|
|
"loss": 2.4863,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.937883383429387e-05,
|
|
"loss": 2.4699,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.9261345548140355e-05,
|
|
"loss": 2.4269,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.9133852216300965e-05,
|
|
"loss": 2.426,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.89964064152747e-05,
|
|
"loss": 2.4609,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.8849064825822146e-05,
|
|
"loss": 2.4219,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.869188820959112e-05,
|
|
"loss": 2.4061,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.8524941384059415e-05,
|
|
"loss": 2.4538,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.834829319580494e-05,
|
|
"loss": 2.4333,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.816201649211435e-05,
|
|
"loss": 2.461,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.796618809094183e-05,
|
|
"loss": 2.4167,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 4.7760888749230416e-05,
|
|
"loss": 2.4521,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 4.7546203129608926e-05,
|
|
"loss": 2.4466,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 4.7322219765478295e-05,
|
|
"loss": 2.4138,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 4.708903102450159e-05,
|
|
"loss": 2.4727,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 4.6846733070512827e-05,
|
|
"loss": 2.3853,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 4.659542582386041e-05,
|
|
"loss": 2.4041,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"eval_loss": 2.40828013420105,
|
|
"eval_runtime": 17.7449,
|
|
"eval_samples_per_second": 576.561,
|
|
"eval_steps_per_second": 18.033,
|
|
"step": 319
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.633521292020126e-05,
|
|
"loss": 2.4217,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 4.606620166776294e-05,
|
|
"loss": 2.4063,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 4.578850300309123e-05,
|
|
"loss": 2.4221,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 4.550223144530136e-05,
|
|
"loss": 2.4377,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 4.520750504885196e-05,
|
|
"loss": 2.4232,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 4.490444535486101e-05,
|
|
"loss": 2.4304,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 4.4593177340983923e-05,
|
|
"loss": 2.4403,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 4.427382936987449e-05,
|
|
"loss": 2.3851,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 4.3946533136249926e-05,
|
|
"loss": 2.3566,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 4.361142361258169e-05,
|
|
"loss": 2.3895,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 4.326863899343479e-05,
|
|
"loss": 2.3585,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 4.291832063847817e-05,
|
|
"loss": 2.3931,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 4.2560613014189966e-05,
|
|
"loss": 2.3949,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 4.219566363428147e-05,
|
|
"loss": 2.4403,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 4.182362299886456e-05,
|
|
"loss": 2.3663,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 4.144464453238748e-05,
|
|
"loss": 2.377,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 4.105888452036471e-05,
|
|
"loss": 2.3663,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 4.066650204492698e-05,
|
|
"loss": 2.3539,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 4.0267658919217955e-05,
|
|
"loss": 2.4552,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 3.986251962066478e-05,
|
|
"loss": 2.3931,
|
|
"step": 510
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 3.949264905820697e-05,
|
|
"loss": 2.3716,
|
|
"step": 520
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 3.9117937442906835e-05,
|
|
"loss": 2.3722,
|
|
"step": 530
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 3.8696065716778966e-05,
|
|
"loss": 2.371,
|
|
"step": 540
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 3.8268545921258523e-05,
|
|
"loss": 2.3809,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 3.78355543596374e-05,
|
|
"loss": 2.3502,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 3.739726959168867e-05,
|
|
"loss": 2.3841,
|
|
"step": 570
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 3.6953872360031e-05,
|
|
"loss": 2.3895,
|
|
"step": 580
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 3.655059477901017e-05,
|
|
"loss": 2.3635,
|
|
"step": 590
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 3.60979893023808e-05,
|
|
"loss": 2.3904,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 3.5640807167388806e-05,
|
|
"loss": 2.3892,
|
|
"step": 610
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 3.5179236909667465e-05,
|
|
"loss": 2.3946,
|
|
"step": 620
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 3.471346887445162e-05,
|
|
"loss": 2.3826,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"eval_loss": 2.3646013736724854,
|
|
"eval_runtime": 18.0603,
|
|
"eval_samples_per_second": 566.491,
|
|
"eval_steps_per_second": 17.718,
|
|
"step": 639
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"learning_rate": 3.424369513808187e-05,
|
|
"loss": 2.3812,
|
|
"step": 640
|
|
},
|
|
{
|
|
"epoch": 2.03,
|
|
"learning_rate": 3.3770109428794995e-05,
|
|
"loss": 2.4335,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 2.06,
|
|
"learning_rate": 3.329290704683301e-05,
|
|
"loss": 2.3533,
|
|
"step": 660
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"learning_rate": 3.2812284783904104e-05,
|
|
"loss": 2.3331,
|
|
"step": 670
|
|
},
|
|
{
|
|
"epoch": 2.13,
|
|
"learning_rate": 3.232844084202837e-05,
|
|
"loss": 2.3991,
|
|
"step": 680
|
|
},
|
|
{
|
|
"epoch": 2.16,
|
|
"learning_rate": 3.1841574751802076e-05,
|
|
"loss": 2.3639,
|
|
"step": 690
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"learning_rate": 3.135188729011395e-05,
|
|
"loss": 2.3488,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 2.22,
|
|
"learning_rate": 3.085958039734758e-05,
|
|
"loss": 2.3528,
|
|
"step": 710
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 3.0463985288442475e-05,
|
|
"loss": 2.334,
|
|
"step": 720
|
|
},
|
|
{
|
|
"epoch": 2.28,
|
|
"learning_rate": 2.9967475704117753e-05,
|
|
"loss": 2.3574,
|
|
"step": 730
|
|
},
|
|
{
|
|
"epoch": 2.31,
|
|
"learning_rate": 2.9468917601038592e-05,
|
|
"loss": 2.3579,
|
|
"step": 740
|
|
},
|
|
{
|
|
"epoch": 2.35,
|
|
"learning_rate": 2.896851657771925e-05,
|
|
"loss": 2.3776,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 2.38,
|
|
"learning_rate": 2.8466478992668966e-05,
|
|
"loss": 2.3707,
|
|
"step": 760
|
|
},
|
|
{
|
|
"epoch": 2.41,
|
|
"learning_rate": 2.7963011879292573e-05,
|
|
"loss": 2.3516,
|
|
"step": 770
|
|
},
|
|
{
|
|
"epoch": 2.44,
|
|
"learning_rate": 2.745832286051273e-05,
|
|
"loss": 2.3547,
|
|
"step": 780
|
|
},
|
|
{
|
|
"epoch": 2.47,
|
|
"learning_rate": 2.695262006314912e-05,
|
|
"loss": 2.3271,
|
|
"step": 790
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 2.6446112032089782e-05,
|
|
"loss": 2.3358,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 2.53,
|
|
"learning_rate": 2.5939007644290088e-05,
|
|
"loss": 2.3812,
|
|
"step": 810
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"learning_rate": 2.5431516022634715e-05,
|
|
"loss": 2.3259,
|
|
"step": 820
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"learning_rate": 2.4923846449698278e-05,
|
|
"loss": 2.348,
|
|
"step": 830
|
|
},
|
|
{
|
|
"epoch": 2.63,
|
|
"learning_rate": 2.4416208281440053e-05,
|
|
"loss": 2.3544,
|
|
"step": 840
|
|
},
|
|
{
|
|
"epoch": 2.66,
|
|
"learning_rate": 2.390881086086846e-05,
|
|
"loss": 2.3985,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"learning_rate": 2.3401863431710863e-05,
|
|
"loss": 2.3544,
|
|
"step": 860
|
|
},
|
|
{
|
|
"epoch": 2.72,
|
|
"learning_rate": 2.2946168279918964e-05,
|
|
"loss": 2.3777,
|
|
"step": 870
|
|
},
|
|
{
|
|
"epoch": 2.75,
|
|
"learning_rate": 2.2491159179921064e-05,
|
|
"loss": 2.3595,
|
|
"step": 880
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"learning_rate": 2.1986582993616926e-05,
|
|
"loss": 2.3243,
|
|
"step": 890
|
|
},
|
|
{
|
|
"epoch": 2.81,
|
|
"learning_rate": 2.1483249499098357e-05,
|
|
"loss": 2.3614,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 2.85,
|
|
"learning_rate": 2.098136626418544e-05,
|
|
"loss": 2.3104,
|
|
"step": 910
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"learning_rate": 2.0481140258631125e-05,
|
|
"loss": 2.3463,
|
|
"step": 920
|
|
},
|
|
{
|
|
"epoch": 2.91,
|
|
"learning_rate": 1.998277776876974e-05,
|
|
"loss": 2.362,
|
|
"step": 930
|
|
},
|
|
{
|
|
"epoch": 2.94,
|
|
"learning_rate": 1.948648431244732e-05,
|
|
"loss": 2.3814,
|
|
"step": 940
|
|
},
|
|
{
|
|
"epoch": 2.97,
|
|
"learning_rate": 1.89924645542689e-05,
|
|
"loss": 2.4056,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"eval_loss": 2.3416996002197266,
|
|
"eval_runtime": 18.2852,
|
|
"eval_samples_per_second": 559.525,
|
|
"eval_steps_per_second": 17.501,
|
|
"step": 959
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"learning_rate": 1.850092222119762e-05,
|
|
"loss": 2.341,
|
|
"step": 960
|
|
},
|
|
{
|
|
"epoch": 3.03,
|
|
"learning_rate": 1.8012060018540542e-05,
|
|
"loss": 2.3657,
|
|
"step": 970
|
|
},
|
|
{
|
|
"epoch": 3.06,
|
|
"learning_rate": 1.7526079546355744e-05,
|
|
"loss": 2.3501,
|
|
"step": 980
|
|
},
|
|
{
|
|
"epoch": 3.1,
|
|
"learning_rate": 1.7091326667078695e-05,
|
|
"loss": 2.3599,
|
|
"step": 990
|
|
},
|
|
{
|
|
"epoch": 3.13,
|
|
"learning_rate": 1.6611372568138135e-05,
|
|
"loss": 2.3613,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 3.16,
|
|
"learning_rate": 1.613487782393661e-05,
|
|
"loss": 2.3977,
|
|
"step": 1010
|
|
},
|
|
{
|
|
"epoch": 3.19,
|
|
"learning_rate": 1.566203893436221e-05,
|
|
"loss": 2.3344,
|
|
"step": 1020
|
|
},
|
|
{
|
|
"epoch": 3.22,
|
|
"learning_rate": 1.5193050891678773e-05,
|
|
"loss": 2.3126,
|
|
"step": 1030
|
|
},
|
|
{
|
|
"epoch": 3.25,
|
|
"learning_rate": 1.4774414010049692e-05,
|
|
"loss": 2.3947,
|
|
"step": 1040
|
|
},
|
|
{
|
|
"epoch": 3.28,
|
|
"learning_rate": 1.431327402831134e-05,
|
|
"loss": 2.3404,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"epoch": 3.32,
|
|
"learning_rate": 1.3856541105586545e-05,
|
|
"loss": 2.3144,
|
|
"step": 1060
|
|
},
|
|
{
|
|
"epoch": 3.35,
|
|
"learning_rate": 1.3404403592259634e-05,
|
|
"loss": 2.3657,
|
|
"step": 1070
|
|
},
|
|
{
|
|
"epoch": 3.38,
|
|
"learning_rate": 1.2957047943631217e-05,
|
|
"loss": 2.3721,
|
|
"step": 1080
|
|
},
|
|
{
|
|
"epoch": 3.41,
|
|
"learning_rate": 1.2514658643026555e-05,
|
|
"loss": 2.3322,
|
|
"step": 1090
|
|
},
|
|
{
|
|
"epoch": 3.44,
|
|
"learning_rate": 1.2077418125717194e-05,
|
|
"loss": 2.3658,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 3.47,
|
|
"learning_rate": 1.1645506703687214e-05,
|
|
"loss": 2.3441,
|
|
"step": 1110
|
|
},
|
|
{
|
|
"epoch": 3.5,
|
|
"learning_rate": 1.1219102491274935e-05,
|
|
"loss": 2.3189,
|
|
"step": 1120
|
|
},
|
|
{
|
|
"epoch": 3.53,
|
|
"learning_rate": 1.0798381331721109e-05,
|
|
"loss": 2.3567,
|
|
"step": 1130
|
|
},
|
|
{
|
|
"epoch": 3.57,
|
|
"learning_rate": 1.0383516724653519e-05,
|
|
"loss": 2.3262,
|
|
"step": 1140
|
|
},
|
|
{
|
|
"epoch": 3.6,
|
|
"learning_rate": 9.974679754538136e-06,
|
|
"loss": 2.337,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"epoch": 3.63,
|
|
"learning_rate": 9.572039020126281e-06,
|
|
"loss": 2.3394,
|
|
"step": 1160
|
|
},
|
|
{
|
|
"epoch": 3.66,
|
|
"learning_rate": 9.215097429546577e-06,
|
|
"loss": 2.3279,
|
|
"step": 1170
|
|
},
|
|
{
|
|
"epoch": 3.69,
|
|
"learning_rate": 8.824684825733865e-06,
|
|
"loss": 2.3321,
|
|
"step": 1180
|
|
},
|
|
{
|
|
"epoch": 3.72,
|
|
"learning_rate": 8.440942699738658e-06,
|
|
"loss": 2.395,
|
|
"step": 1190
|
|
},
|
|
{
|
|
"epoch": 3.75,
|
|
"learning_rate": 8.064029301543208e-06,
|
|
"loss": 2.3253,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"epoch": 3.78,
|
|
"learning_rate": 7.694100065056206e-06,
|
|
"loss": 2.2957,
|
|
"step": 1210
|
|
},
|
|
{
|
|
"epoch": 3.82,
|
|
"learning_rate": 7.367261356670735e-06,
|
|
"loss": 2.3385,
|
|
"step": 1220
|
|
},
|
|
{
|
|
"epoch": 3.85,
|
|
"learning_rate": 7.011019881795297e-06,
|
|
"loss": 2.3603,
|
|
"step": 1230
|
|
},
|
|
{
|
|
"epoch": 3.88,
|
|
"learning_rate": 6.662196815250421e-06,
|
|
"loss": 2.3361,
|
|
"step": 1240
|
|
},
|
|
{
|
|
"epoch": 3.91,
|
|
"learning_rate": 6.32093600687782e-06,
|
|
"loss": 2.2915,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"epoch": 3.94,
|
|
"learning_rate": 5.987378187947812e-06,
|
|
"loss": 2.3616,
|
|
"step": 1260
|
|
},
|
|
{
|
|
"epoch": 3.97,
|
|
"learning_rate": 5.661660913123673e-06,
|
|
"loss": 2.2983,
|
|
"step": 1270
|
|
},
|
|
{
|
|
"epoch": 4.0,
|
|
"eval_loss": 2.3321726322174072,
|
|
"eval_runtime": 17.7952,
|
|
"eval_samples_per_second": 574.93,
|
|
"eval_steps_per_second": 17.982,
|
|
"step": 1279
|
|
},
|
|
{
|
|
"epoch": 4.0,
|
|
"learning_rate": 5.343918503735962e-06,
|
|
"loss": 2.3394,
|
|
"step": 1280
|
|
},
|
|
{
|
|
"epoch": 4.03,
|
|
"learning_rate": 5.034281992390255e-06,
|
|
"loss": 2.3569,
|
|
"step": 1290
|
|
},
|
|
{
|
|
"epoch": 4.07,
|
|
"learning_rate": 4.732879068931131e-06,
|
|
"loss": 2.3701,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"epoch": 4.1,
|
|
"learning_rate": 4.439834027784595e-06,
|
|
"loss": 2.3255,
|
|
"step": 1310
|
|
},
|
|
{
|
|
"epoch": 4.13,
|
|
"learning_rate": 4.183339432819844e-06,
|
|
"loss": 2.3123,
|
|
"step": 1320
|
|
},
|
|
{
|
|
"epoch": 4.16,
|
|
"learning_rate": 3.906504414758375e-06,
|
|
"loss": 2.391,
|
|
"step": 1330
|
|
},
|
|
{
|
|
"epoch": 4.19,
|
|
"learning_rate": 3.6383680645670106e-06,
|
|
"loss": 2.2942,
|
|
"step": 1340
|
|
},
|
|
{
|
|
"epoch": 4.22,
|
|
"learning_rate": 3.3790409579938343e-06,
|
|
"loss": 2.31,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"epoch": 4.25,
|
|
"learning_rate": 3.128630037975866e-06,
|
|
"loss": 2.3248,
|
|
"step": 1360
|
|
},
|
|
{
|
|
"epoch": 4.28,
|
|
"learning_rate": 2.8872385705372514e-06,
|
|
"loss": 2.3994,
|
|
"step": 1370
|
|
},
|
|
{
|
|
"epoch": 4.32,
|
|
"learning_rate": 2.6549661022037543e-06,
|
|
"loss": 2.3534,
|
|
"step": 1380
|
|
},
|
|
{
|
|
"epoch": 4.35,
|
|
"learning_rate": 2.431908418951168e-06,
|
|
"loss": 2.3192,
|
|
"step": 1390
|
|
},
|
|
{
|
|
"epoch": 4.38,
|
|
"learning_rate": 2.218157506704466e-06,
|
|
"loss": 2.3471,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"epoch": 4.41,
|
|
"learning_rate": 2.03381190905817e-06,
|
|
"loss": 2.3081,
|
|
"step": 1410
|
|
},
|
|
{
|
|
"epoch": 4.44,
|
|
"learning_rate": 1.837983507885649e-06,
|
|
"loss": 2.2952,
|
|
"step": 1420
|
|
},
|
|
{
|
|
"epoch": 4.47,
|
|
"learning_rate": 1.651706804195066e-06,
|
|
"loss": 2.3029,
|
|
"step": 1430
|
|
},
|
|
{
|
|
"epoch": 4.5,
|
|
"learning_rate": 1.4750586159405915e-06,
|
|
"loss": 2.3195,
|
|
"step": 1440
|
|
},
|
|
{
|
|
"epoch": 4.53,
|
|
"learning_rate": 1.3081117904088808e-06,
|
|
"loss": 2.3722,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"epoch": 4.57,
|
|
"learning_rate": 1.1509351741778585e-06,
|
|
"loss": 2.327,
|
|
"step": 1460
|
|
},
|
|
{
|
|
"epoch": 4.6,
|
|
"learning_rate": 1.0035935847253014e-06,
|
|
"loss": 2.3479,
|
|
"step": 1470
|
|
},
|
|
{
|
|
"epoch": 4.63,
|
|
"learning_rate": 8.661477836990339e-07,
|
|
"loss": 2.3108,
|
|
"step": 1480
|
|
},
|
|
{
|
|
"epoch": 4.66,
|
|
"learning_rate": 7.386544518596466e-07,
|
|
"loss": 2.3649,
|
|
"step": 1490
|
|
},
|
|
{
|
|
"epoch": 4.69,
|
|
"learning_rate": 6.211661657061301e-07,
|
|
"loss": 2.3571,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 4.72,
|
|
"learning_rate": 5.137313757940465e-07,
|
|
"loss": 2.3326,
|
|
"step": 1510
|
|
},
|
|
{
|
|
"epoch": 4.75,
|
|
"learning_rate": 4.1639438675519225e-07,
|
|
"loss": 2.2972,
|
|
"step": 1520
|
|
},
|
|
{
|
|
"epoch": 4.78,
|
|
"learning_rate": 3.2919533902696364e-07,
|
|
"loss": 2.3022,
|
|
"step": 1530
|
|
},
|
|
{
|
|
"epoch": 4.82,
|
|
"learning_rate": 2.66759724360105e-07,
|
|
"loss": 2.3588,
|
|
"step": 1540
|
|
},
|
|
{
|
|
"epoch": 4.85,
|
|
"learning_rate": 1.9789677043525468e-07,
|
|
"loss": 2.323,
|
|
"step": 1550
|
|
},
|
|
{
|
|
"epoch": 4.88,
|
|
"learning_rate": 1.3926186323703905e-07,
|
|
"loss": 2.358,
|
|
"step": 1560
|
|
},
|
|
{
|
|
"epoch": 4.91,
|
|
"learning_rate": 9.087918299586773e-08,
|
|
"loss": 2.3592,
|
|
"step": 1570
|
|
},
|
|
{
|
|
"epoch": 4.94,
|
|
"learning_rate": 5.27686820645551e-08,
|
|
"loss": 2.3577,
|
|
"step": 1580
|
|
},
|
|
{
|
|
"epoch": 4.97,
|
|
"learning_rate": 2.4946076690257612e-08,
|
|
"loss": 2.3596,
|
|
"step": 1590
|
|
},
|
|
{
|
|
"epoch": 4.99,
|
|
"eval_loss": 2.3302903175354004,
|
|
"eval_runtime": 17.7344,
|
|
"eval_samples_per_second": 576.902,
|
|
"eval_steps_per_second": 18.044,
|
|
"step": 1595
|
|
},
|
|
{
|
|
"epoch": 4.99,
|
|
"step": 1595,
|
|
"total_flos": 4.262542942373151e+17,
|
|
"train_loss": 2.405394427380218,
|
|
"train_runtime": 1383.7033,
|
|
"train_samples_per_second": 147.864,
|
|
"train_steps_per_second": 1.153
|
|
}
|
|
],
|
|
"max_steps": 1595,
|
|
"num_train_epochs": 5,
|
|
"total_flos": 4.262542942373151e+17,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|