2385 lines
41 KiB
JSON
2385 lines
41 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 4.998391765841107,
|
|
"global_step": 3885,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1e-05,
|
|
"loss": 2.1179,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.875,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.8748,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.9221,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.8121,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.8224,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7931,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.8394,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.8047,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7649,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7028,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7688,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.761,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7407,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6848,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.811,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7782,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7352,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7782,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7228,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7137,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6885,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6691,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7616,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7547,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.649,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.796,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.7022,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.707,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5932,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6794,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6577,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6662,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6114,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6776,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6073,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6783,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6673,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6344,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5606,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.584,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.595,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.551,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6187,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6669,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5301,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5117,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5942,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5738,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5742,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5397,
|
|
"step": 510
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5526,
|
|
"step": 520
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.527,
|
|
"step": 530
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5189,
|
|
"step": 540
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5508,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6197,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6338,
|
|
"step": 570
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.515,
|
|
"step": 580
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5122,
|
|
"step": 590
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5044,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5313,
|
|
"step": 610
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.511,
|
|
"step": 620
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5263,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5887,
|
|
"step": 640
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5424,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5115,
|
|
"step": 660
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5259,
|
|
"step": 670
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4667,
|
|
"step": 680
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4453,
|
|
"step": 690
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5575,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5078,
|
|
"step": 710
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.6226,
|
|
"step": 720
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5603,
|
|
"step": 730
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4699,
|
|
"step": 740
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5657,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.588,
|
|
"step": 760
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.5083,
|
|
"step": 770
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"eval_loss": 1.5581023693084717,
|
|
"eval_runtime": 18.0921,
|
|
"eval_samples_per_second": 28.3,
|
|
"eval_steps_per_second": 28.3,
|
|
"step": 777
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4882,
|
|
"step": 780
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4555,
|
|
"step": 790
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4035,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4673,
|
|
"step": 810
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3792,
|
|
"step": 820
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3867,
|
|
"step": 830
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.389,
|
|
"step": 840
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4264,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3682,
|
|
"step": 860
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4074,
|
|
"step": 870
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3506,
|
|
"step": 880
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3164,
|
|
"step": 890
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4228,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3693,
|
|
"step": 910
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4301,
|
|
"step": 920
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.47,
|
|
"step": 930
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4198,
|
|
"step": 940
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3479,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4116,
|
|
"step": 960
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4352,
|
|
"step": 970
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4085,
|
|
"step": 980
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4118,
|
|
"step": 990
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4072,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4231,
|
|
"step": 1010
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3781,
|
|
"step": 1020
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4377,
|
|
"step": 1030
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3402,
|
|
"step": 1040
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4094,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3511,
|
|
"step": 1060
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3832,
|
|
"step": 1070
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4262,
|
|
"step": 1080
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.383,
|
|
"step": 1090
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4542,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3757,
|
|
"step": 1110
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3886,
|
|
"step": 1120
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3838,
|
|
"step": 1130
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3351,
|
|
"step": 1140
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4089,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4081,
|
|
"step": 1160
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.419,
|
|
"step": 1170
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.391,
|
|
"step": 1180
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4382,
|
|
"step": 1190
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.379,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4071,
|
|
"step": 1210
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3658,
|
|
"step": 1220
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3861,
|
|
"step": 1230
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.385,
|
|
"step": 1240
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3712,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3088,
|
|
"step": 1260
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3961,
|
|
"step": 1270
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4149,
|
|
"step": 1280
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4143,
|
|
"step": 1290
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.401,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4208,
|
|
"step": 1310
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3487,
|
|
"step": 1320
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4031,
|
|
"step": 1330
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4645,
|
|
"step": 1340
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.331,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4235,
|
|
"step": 1360
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3116,
|
|
"step": 1370
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3436,
|
|
"step": 1380
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4319,
|
|
"step": 1390
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4129,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3383,
|
|
"step": 1410
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3584,
|
|
"step": 1420
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.416,
|
|
"step": 1430
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3045,
|
|
"step": 1440
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3298,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3451,
|
|
"step": 1460
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3974,
|
|
"step": 1470
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3717,
|
|
"step": 1480
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4008,
|
|
"step": 1490
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3392,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4095,
|
|
"step": 1510
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4033,
|
|
"step": 1520
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3972,
|
|
"step": 1530
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.4225,
|
|
"step": 1540
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.33,
|
|
"step": 1550
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"eval_loss": 1.5331286191940308,
|
|
"eval_runtime": 17.7428,
|
|
"eval_samples_per_second": 28.857,
|
|
"eval_steps_per_second": 28.857,
|
|
"step": 1554
|
|
},
|
|
{
|
|
"epoch": 2.01,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2771,
|
|
"step": 1560
|
|
},
|
|
{
|
|
"epoch": 2.02,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1553,
|
|
"step": 1570
|
|
},
|
|
{
|
|
"epoch": 2.03,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2805,
|
|
"step": 1580
|
|
},
|
|
{
|
|
"epoch": 2.05,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2333,
|
|
"step": 1590
|
|
},
|
|
{
|
|
"epoch": 2.06,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3188,
|
|
"step": 1600
|
|
},
|
|
{
|
|
"epoch": 2.07,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2829,
|
|
"step": 1610
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2259,
|
|
"step": 1620
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1629,
|
|
"step": 1630
|
|
},
|
|
{
|
|
"epoch": 2.11,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2151,
|
|
"step": 1640
|
|
},
|
|
{
|
|
"epoch": 2.12,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2491,
|
|
"step": 1650
|
|
},
|
|
{
|
|
"epoch": 2.14,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2415,
|
|
"step": 1660
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2224,
|
|
"step": 1670
|
|
},
|
|
{
|
|
"epoch": 2.16,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1897,
|
|
"step": 1680
|
|
},
|
|
{
|
|
"epoch": 2.17,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2426,
|
|
"step": 1690
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2183,
|
|
"step": 1700
|
|
},
|
|
{
|
|
"epoch": 2.2,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1904,
|
|
"step": 1710
|
|
},
|
|
{
|
|
"epoch": 2.21,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2171,
|
|
"step": 1720
|
|
},
|
|
{
|
|
"epoch": 2.23,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2448,
|
|
"step": 1730
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2509,
|
|
"step": 1740
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2487,
|
|
"step": 1750
|
|
},
|
|
{
|
|
"epoch": 2.26,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2461,
|
|
"step": 1760
|
|
},
|
|
{
|
|
"epoch": 2.28,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2159,
|
|
"step": 1770
|
|
},
|
|
{
|
|
"epoch": 2.29,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2734,
|
|
"step": 1780
|
|
},
|
|
{
|
|
"epoch": 2.3,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2041,
|
|
"step": 1790
|
|
},
|
|
{
|
|
"epoch": 2.32,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2908,
|
|
"step": 1800
|
|
},
|
|
{
|
|
"epoch": 2.33,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2412,
|
|
"step": 1810
|
|
},
|
|
{
|
|
"epoch": 2.34,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.209,
|
|
"step": 1820
|
|
},
|
|
{
|
|
"epoch": 2.35,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2671,
|
|
"step": 1830
|
|
},
|
|
{
|
|
"epoch": 2.37,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2154,
|
|
"step": 1840
|
|
},
|
|
{
|
|
"epoch": 2.38,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2206,
|
|
"step": 1850
|
|
},
|
|
{
|
|
"epoch": 2.39,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1998,
|
|
"step": 1860
|
|
},
|
|
{
|
|
"epoch": 2.41,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2266,
|
|
"step": 1870
|
|
},
|
|
{
|
|
"epoch": 2.42,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2652,
|
|
"step": 1880
|
|
},
|
|
{
|
|
"epoch": 2.43,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2762,
|
|
"step": 1890
|
|
},
|
|
{
|
|
"epoch": 2.44,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1745,
|
|
"step": 1900
|
|
},
|
|
{
|
|
"epoch": 2.46,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2094,
|
|
"step": 1910
|
|
},
|
|
{
|
|
"epoch": 2.47,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2166,
|
|
"step": 1920
|
|
},
|
|
{
|
|
"epoch": 2.48,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2525,
|
|
"step": 1930
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.201,
|
|
"step": 1940
|
|
},
|
|
{
|
|
"epoch": 2.51,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1967,
|
|
"step": 1950
|
|
},
|
|
{
|
|
"epoch": 2.52,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.258,
|
|
"step": 1960
|
|
},
|
|
{
|
|
"epoch": 2.53,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2531,
|
|
"step": 1970
|
|
},
|
|
{
|
|
"epoch": 2.55,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2636,
|
|
"step": 1980
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2029,
|
|
"step": 1990
|
|
},
|
|
{
|
|
"epoch": 2.57,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3107,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"epoch": 2.59,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2099,
|
|
"step": 2010
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2283,
|
|
"step": 2020
|
|
},
|
|
{
|
|
"epoch": 2.61,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2111,
|
|
"step": 2030
|
|
},
|
|
{
|
|
"epoch": 2.62,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2211,
|
|
"step": 2040
|
|
},
|
|
{
|
|
"epoch": 2.64,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2691,
|
|
"step": 2050
|
|
},
|
|
{
|
|
"epoch": 2.65,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2478,
|
|
"step": 2060
|
|
},
|
|
{
|
|
"epoch": 2.66,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2589,
|
|
"step": 2070
|
|
},
|
|
{
|
|
"epoch": 2.68,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2469,
|
|
"step": 2080
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.253,
|
|
"step": 2090
|
|
},
|
|
{
|
|
"epoch": 2.7,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2449,
|
|
"step": 2100
|
|
},
|
|
{
|
|
"epoch": 2.71,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1944,
|
|
"step": 2110
|
|
},
|
|
{
|
|
"epoch": 2.73,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2346,
|
|
"step": 2120
|
|
},
|
|
{
|
|
"epoch": 2.74,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2193,
|
|
"step": 2130
|
|
},
|
|
{
|
|
"epoch": 2.75,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2029,
|
|
"step": 2140
|
|
},
|
|
{
|
|
"epoch": 2.77,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2675,
|
|
"step": 2150
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1958,
|
|
"step": 2160
|
|
},
|
|
{
|
|
"epoch": 2.79,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2286,
|
|
"step": 2170
|
|
},
|
|
{
|
|
"epoch": 2.8,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1888,
|
|
"step": 2180
|
|
},
|
|
{
|
|
"epoch": 2.82,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1694,
|
|
"step": 2190
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.225,
|
|
"step": 2200
|
|
},
|
|
{
|
|
"epoch": 2.84,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2131,
|
|
"step": 2210
|
|
},
|
|
{
|
|
"epoch": 2.86,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.3042,
|
|
"step": 2220
|
|
},
|
|
{
|
|
"epoch": 2.87,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2324,
|
|
"step": 2230
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1676,
|
|
"step": 2240
|
|
},
|
|
{
|
|
"epoch": 2.89,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2519,
|
|
"step": 2250
|
|
},
|
|
{
|
|
"epoch": 2.91,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2414,
|
|
"step": 2260
|
|
},
|
|
{
|
|
"epoch": 2.92,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2363,
|
|
"step": 2270
|
|
},
|
|
{
|
|
"epoch": 2.93,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2504,
|
|
"step": 2280
|
|
},
|
|
{
|
|
"epoch": 2.95,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.196,
|
|
"step": 2290
|
|
},
|
|
{
|
|
"epoch": 2.96,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2527,
|
|
"step": 2300
|
|
},
|
|
{
|
|
"epoch": 2.97,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2256,
|
|
"step": 2310
|
|
},
|
|
{
|
|
"epoch": 2.98,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1909,
|
|
"step": 2320
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.2779,
|
|
"step": 2330
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"eval_loss": 1.548167109489441,
|
|
"eval_runtime": 17.7658,
|
|
"eval_samples_per_second": 28.819,
|
|
"eval_steps_per_second": 28.819,
|
|
"step": 2331
|
|
},
|
|
{
|
|
"epoch": 3.01,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1213,
|
|
"step": 2340
|
|
},
|
|
{
|
|
"epoch": 3.02,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1144,
|
|
"step": 2350
|
|
},
|
|
{
|
|
"epoch": 3.04,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0383,
|
|
"step": 2360
|
|
},
|
|
{
|
|
"epoch": 3.05,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1255,
|
|
"step": 2370
|
|
},
|
|
{
|
|
"epoch": 3.06,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0925,
|
|
"step": 2380
|
|
},
|
|
{
|
|
"epoch": 3.07,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0786,
|
|
"step": 2390
|
|
},
|
|
{
|
|
"epoch": 3.09,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0565,
|
|
"step": 2400
|
|
},
|
|
{
|
|
"epoch": 3.1,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0401,
|
|
"step": 2410
|
|
},
|
|
{
|
|
"epoch": 3.11,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0217,
|
|
"step": 2420
|
|
},
|
|
{
|
|
"epoch": 3.13,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0722,
|
|
"step": 2430
|
|
},
|
|
{
|
|
"epoch": 3.14,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0483,
|
|
"step": 2440
|
|
},
|
|
{
|
|
"epoch": 3.15,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1119,
|
|
"step": 2450
|
|
},
|
|
{
|
|
"epoch": 3.17,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1065,
|
|
"step": 2460
|
|
},
|
|
{
|
|
"epoch": 3.18,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0412,
|
|
"step": 2470
|
|
},
|
|
{
|
|
"epoch": 3.19,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0226,
|
|
"step": 2480
|
|
},
|
|
{
|
|
"epoch": 3.2,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0748,
|
|
"step": 2490
|
|
},
|
|
{
|
|
"epoch": 3.22,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.055,
|
|
"step": 2500
|
|
},
|
|
{
|
|
"epoch": 3.23,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0492,
|
|
"step": 2510
|
|
},
|
|
{
|
|
"epoch": 3.24,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0837,
|
|
"step": 2520
|
|
},
|
|
{
|
|
"epoch": 3.26,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0462,
|
|
"step": 2530
|
|
},
|
|
{
|
|
"epoch": 3.27,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0538,
|
|
"step": 2540
|
|
},
|
|
{
|
|
"epoch": 3.28,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0349,
|
|
"step": 2550
|
|
},
|
|
{
|
|
"epoch": 3.29,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0526,
|
|
"step": 2560
|
|
},
|
|
{
|
|
"epoch": 3.31,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.069,
|
|
"step": 2570
|
|
},
|
|
{
|
|
"epoch": 3.32,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.112,
|
|
"step": 2580
|
|
},
|
|
{
|
|
"epoch": 3.33,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0861,
|
|
"step": 2590
|
|
},
|
|
{
|
|
"epoch": 3.35,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0818,
|
|
"step": 2600
|
|
},
|
|
{
|
|
"epoch": 3.36,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.054,
|
|
"step": 2610
|
|
},
|
|
{
|
|
"epoch": 3.37,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1018,
|
|
"step": 2620
|
|
},
|
|
{
|
|
"epoch": 3.38,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0581,
|
|
"step": 2630
|
|
},
|
|
{
|
|
"epoch": 3.4,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0861,
|
|
"step": 2640
|
|
},
|
|
{
|
|
"epoch": 3.41,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0755,
|
|
"step": 2650
|
|
},
|
|
{
|
|
"epoch": 3.42,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0656,
|
|
"step": 2660
|
|
},
|
|
{
|
|
"epoch": 3.44,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0681,
|
|
"step": 2670
|
|
},
|
|
{
|
|
"epoch": 3.45,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0875,
|
|
"step": 2680
|
|
},
|
|
{
|
|
"epoch": 3.46,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0687,
|
|
"step": 2690
|
|
},
|
|
{
|
|
"epoch": 3.47,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1273,
|
|
"step": 2700
|
|
},
|
|
{
|
|
"epoch": 3.49,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1803,
|
|
"step": 2710
|
|
},
|
|
{
|
|
"epoch": 3.5,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0955,
|
|
"step": 2720
|
|
},
|
|
{
|
|
"epoch": 3.51,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0973,
|
|
"step": 2730
|
|
},
|
|
{
|
|
"epoch": 3.53,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1065,
|
|
"step": 2740
|
|
},
|
|
{
|
|
"epoch": 3.54,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0539,
|
|
"step": 2750
|
|
},
|
|
{
|
|
"epoch": 3.55,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0589,
|
|
"step": 2760
|
|
},
|
|
{
|
|
"epoch": 3.56,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.094,
|
|
"step": 2770
|
|
},
|
|
{
|
|
"epoch": 3.58,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0977,
|
|
"step": 2780
|
|
},
|
|
{
|
|
"epoch": 3.59,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0545,
|
|
"step": 2790
|
|
},
|
|
{
|
|
"epoch": 3.6,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1171,
|
|
"step": 2800
|
|
},
|
|
{
|
|
"epoch": 3.62,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0773,
|
|
"step": 2810
|
|
},
|
|
{
|
|
"epoch": 3.63,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0672,
|
|
"step": 2820
|
|
},
|
|
{
|
|
"epoch": 3.64,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0719,
|
|
"step": 2830
|
|
},
|
|
{
|
|
"epoch": 3.65,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0666,
|
|
"step": 2840
|
|
},
|
|
{
|
|
"epoch": 3.67,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1267,
|
|
"step": 2850
|
|
},
|
|
{
|
|
"epoch": 3.68,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1158,
|
|
"step": 2860
|
|
},
|
|
{
|
|
"epoch": 3.69,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0238,
|
|
"step": 2870
|
|
},
|
|
{
|
|
"epoch": 3.71,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1239,
|
|
"step": 2880
|
|
},
|
|
{
|
|
"epoch": 3.72,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0598,
|
|
"step": 2890
|
|
},
|
|
{
|
|
"epoch": 3.73,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0201,
|
|
"step": 2900
|
|
},
|
|
{
|
|
"epoch": 3.74,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0351,
|
|
"step": 2910
|
|
},
|
|
{
|
|
"epoch": 3.76,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0387,
|
|
"step": 2920
|
|
},
|
|
{
|
|
"epoch": 3.77,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0522,
|
|
"step": 2930
|
|
},
|
|
{
|
|
"epoch": 3.78,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0885,
|
|
"step": 2940
|
|
},
|
|
{
|
|
"epoch": 3.8,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0656,
|
|
"step": 2950
|
|
},
|
|
{
|
|
"epoch": 3.81,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0732,
|
|
"step": 2960
|
|
},
|
|
{
|
|
"epoch": 3.82,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0434,
|
|
"step": 2970
|
|
},
|
|
{
|
|
"epoch": 3.83,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.049,
|
|
"step": 2980
|
|
},
|
|
{
|
|
"epoch": 3.85,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0529,
|
|
"step": 2990
|
|
},
|
|
{
|
|
"epoch": 3.86,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0482,
|
|
"step": 3000
|
|
},
|
|
{
|
|
"epoch": 3.87,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1163,
|
|
"step": 3010
|
|
},
|
|
{
|
|
"epoch": 3.89,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1045,
|
|
"step": 3020
|
|
},
|
|
{
|
|
"epoch": 3.9,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0549,
|
|
"step": 3030
|
|
},
|
|
{
|
|
"epoch": 3.91,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0867,
|
|
"step": 3040
|
|
},
|
|
{
|
|
"epoch": 3.92,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0974,
|
|
"step": 3050
|
|
},
|
|
{
|
|
"epoch": 3.94,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1031,
|
|
"step": 3060
|
|
},
|
|
{
|
|
"epoch": 3.95,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0481,
|
|
"step": 3070
|
|
},
|
|
{
|
|
"epoch": 3.96,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0923,
|
|
"step": 3080
|
|
},
|
|
{
|
|
"epoch": 3.98,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.1012,
|
|
"step": 3090
|
|
},
|
|
{
|
|
"epoch": 3.99,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0483,
|
|
"step": 3100
|
|
},
|
|
{
|
|
"epoch": 4.0,
|
|
"eval_loss": 1.5938072204589844,
|
|
"eval_runtime": 17.7442,
|
|
"eval_samples_per_second": 28.854,
|
|
"eval_steps_per_second": 28.854,
|
|
"step": 3109
|
|
},
|
|
{
|
|
"epoch": 4.0,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.0378,
|
|
"step": 3110
|
|
},
|
|
{
|
|
"epoch": 4.01,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9629,
|
|
"step": 3120
|
|
},
|
|
{
|
|
"epoch": 4.03,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9207,
|
|
"step": 3130
|
|
},
|
|
{
|
|
"epoch": 4.04,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9342,
|
|
"step": 3140
|
|
},
|
|
{
|
|
"epoch": 4.05,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9223,
|
|
"step": 3150
|
|
},
|
|
{
|
|
"epoch": 4.07,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8871,
|
|
"step": 3160
|
|
},
|
|
{
|
|
"epoch": 4.08,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9001,
|
|
"step": 3170
|
|
},
|
|
{
|
|
"epoch": 4.09,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9807,
|
|
"step": 3180
|
|
},
|
|
{
|
|
"epoch": 4.1,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8635,
|
|
"step": 3190
|
|
},
|
|
{
|
|
"epoch": 4.12,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9653,
|
|
"step": 3200
|
|
},
|
|
{
|
|
"epoch": 4.13,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9224,
|
|
"step": 3210
|
|
},
|
|
{
|
|
"epoch": 4.14,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8851,
|
|
"step": 3220
|
|
},
|
|
{
|
|
"epoch": 4.16,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.951,
|
|
"step": 3230
|
|
},
|
|
{
|
|
"epoch": 4.17,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9046,
|
|
"step": 3240
|
|
},
|
|
{
|
|
"epoch": 4.18,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9218,
|
|
"step": 3250
|
|
},
|
|
{
|
|
"epoch": 4.19,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9024,
|
|
"step": 3260
|
|
},
|
|
{
|
|
"epoch": 4.21,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9042,
|
|
"step": 3270
|
|
},
|
|
{
|
|
"epoch": 4.22,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8998,
|
|
"step": 3280
|
|
},
|
|
{
|
|
"epoch": 4.23,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.89,
|
|
"step": 3290
|
|
},
|
|
{
|
|
"epoch": 4.25,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8829,
|
|
"step": 3300
|
|
},
|
|
{
|
|
"epoch": 4.26,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9323,
|
|
"step": 3310
|
|
},
|
|
{
|
|
"epoch": 4.27,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9605,
|
|
"step": 3320
|
|
},
|
|
{
|
|
"epoch": 4.28,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.918,
|
|
"step": 3330
|
|
},
|
|
{
|
|
"epoch": 4.3,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9138,
|
|
"step": 3340
|
|
},
|
|
{
|
|
"epoch": 4.31,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9218,
|
|
"step": 3350
|
|
},
|
|
{
|
|
"epoch": 4.32,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8901,
|
|
"step": 3360
|
|
},
|
|
{
|
|
"epoch": 4.34,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8635,
|
|
"step": 3370
|
|
},
|
|
{
|
|
"epoch": 4.35,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9265,
|
|
"step": 3380
|
|
},
|
|
{
|
|
"epoch": 4.36,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9359,
|
|
"step": 3390
|
|
},
|
|
{
|
|
"epoch": 4.37,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9046,
|
|
"step": 3400
|
|
},
|
|
{
|
|
"epoch": 4.39,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9513,
|
|
"step": 3410
|
|
},
|
|
{
|
|
"epoch": 4.4,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9364,
|
|
"step": 3420
|
|
},
|
|
{
|
|
"epoch": 4.41,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9361,
|
|
"step": 3430
|
|
},
|
|
{
|
|
"epoch": 4.43,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8762,
|
|
"step": 3440
|
|
},
|
|
{
|
|
"epoch": 4.44,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9034,
|
|
"step": 3450
|
|
},
|
|
{
|
|
"epoch": 4.45,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8788,
|
|
"step": 3460
|
|
},
|
|
{
|
|
"epoch": 4.46,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9303,
|
|
"step": 3470
|
|
},
|
|
{
|
|
"epoch": 4.48,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9284,
|
|
"step": 3480
|
|
},
|
|
{
|
|
"epoch": 4.49,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9263,
|
|
"step": 3490
|
|
},
|
|
{
|
|
"epoch": 4.5,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9892,
|
|
"step": 3500
|
|
},
|
|
{
|
|
"epoch": 4.52,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9007,
|
|
"step": 3510
|
|
},
|
|
{
|
|
"epoch": 4.53,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9417,
|
|
"step": 3520
|
|
},
|
|
{
|
|
"epoch": 4.54,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9136,
|
|
"step": 3530
|
|
},
|
|
{
|
|
"epoch": 4.55,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9367,
|
|
"step": 3540
|
|
},
|
|
{
|
|
"epoch": 4.57,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9076,
|
|
"step": 3550
|
|
},
|
|
{
|
|
"epoch": 4.58,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9333,
|
|
"step": 3560
|
|
},
|
|
{
|
|
"epoch": 4.59,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9493,
|
|
"step": 3570
|
|
},
|
|
{
|
|
"epoch": 4.61,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9315,
|
|
"step": 3580
|
|
},
|
|
{
|
|
"epoch": 4.62,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9747,
|
|
"step": 3590
|
|
},
|
|
{
|
|
"epoch": 4.63,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8919,
|
|
"step": 3600
|
|
},
|
|
{
|
|
"epoch": 4.64,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9192,
|
|
"step": 3610
|
|
},
|
|
{
|
|
"epoch": 4.66,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9136,
|
|
"step": 3620
|
|
},
|
|
{
|
|
"epoch": 4.67,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9203,
|
|
"step": 3630
|
|
},
|
|
{
|
|
"epoch": 4.68,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.901,
|
|
"step": 3640
|
|
},
|
|
{
|
|
"epoch": 4.7,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8759,
|
|
"step": 3650
|
|
},
|
|
{
|
|
"epoch": 4.71,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.894,
|
|
"step": 3660
|
|
},
|
|
{
|
|
"epoch": 4.72,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8761,
|
|
"step": 3670
|
|
},
|
|
{
|
|
"epoch": 4.73,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9586,
|
|
"step": 3680
|
|
},
|
|
{
|
|
"epoch": 4.75,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9261,
|
|
"step": 3690
|
|
},
|
|
{
|
|
"epoch": 4.76,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.942,
|
|
"step": 3700
|
|
},
|
|
{
|
|
"epoch": 4.77,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9675,
|
|
"step": 3710
|
|
},
|
|
{
|
|
"epoch": 4.79,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9253,
|
|
"step": 3720
|
|
},
|
|
{
|
|
"epoch": 4.8,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.958,
|
|
"step": 3730
|
|
},
|
|
{
|
|
"epoch": 4.81,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8909,
|
|
"step": 3740
|
|
},
|
|
{
|
|
"epoch": 4.82,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9477,
|
|
"step": 3750
|
|
},
|
|
{
|
|
"epoch": 4.84,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.903,
|
|
"step": 3760
|
|
},
|
|
{
|
|
"epoch": 4.85,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9767,
|
|
"step": 3770
|
|
},
|
|
{
|
|
"epoch": 4.86,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8772,
|
|
"step": 3780
|
|
},
|
|
{
|
|
"epoch": 4.88,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9209,
|
|
"step": 3790
|
|
},
|
|
{
|
|
"epoch": 4.89,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9265,
|
|
"step": 3800
|
|
},
|
|
{
|
|
"epoch": 4.9,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9132,
|
|
"step": 3810
|
|
},
|
|
{
|
|
"epoch": 4.91,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9453,
|
|
"step": 3820
|
|
},
|
|
{
|
|
"epoch": 4.93,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9357,
|
|
"step": 3830
|
|
},
|
|
{
|
|
"epoch": 4.94,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9425,
|
|
"step": 3840
|
|
},
|
|
{
|
|
"epoch": 4.95,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.946,
|
|
"step": 3850
|
|
},
|
|
{
|
|
"epoch": 4.97,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8947,
|
|
"step": 3860
|
|
},
|
|
{
|
|
"epoch": 4.98,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8582,
|
|
"step": 3870
|
|
},
|
|
{
|
|
"epoch": 4.99,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.9522,
|
|
"step": 3880
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"eval_loss": 1.6564583778381348,
|
|
"eval_runtime": 17.5686,
|
|
"eval_samples_per_second": 29.143,
|
|
"eval_steps_per_second": 29.143,
|
|
"step": 3885
|
|
}
|
|
],
|
|
"max_steps": 3885,
|
|
"num_train_epochs": 5,
|
|
"total_flos": 2.3556922294213018e+17,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|