ccore's picture
Upload folder using huggingface_hub
e126367
raw
history blame contribute delete
No virus
63.1 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.9633507853403143,
"eval_steps": 500,
"global_step": 3000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 0.001,
"loss": 1.3845,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 0.001,
"loss": 1.6284,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 0.001,
"loss": 1.3152,
"step": 15
},
{
"epoch": 0.01,
"learning_rate": 0.001,
"loss": 1.6192,
"step": 20
},
{
"epoch": 0.02,
"learning_rate": 0.001,
"loss": 1.7394,
"step": 25
},
{
"epoch": 0.02,
"learning_rate": 0.001,
"loss": 1.9925,
"step": 30
},
{
"epoch": 0.02,
"learning_rate": 0.001,
"loss": 1.902,
"step": 35
},
{
"epoch": 0.03,
"learning_rate": 0.001,
"loss": 1.822,
"step": 40
},
{
"epoch": 0.03,
"learning_rate": 0.001,
"loss": 2.137,
"step": 45
},
{
"epoch": 0.03,
"learning_rate": 0.001,
"loss": 2.2556,
"step": 50
},
{
"epoch": 0.04,
"learning_rate": 0.001,
"loss": 2.2468,
"step": 55
},
{
"epoch": 0.04,
"learning_rate": 0.001,
"loss": 2.0746,
"step": 60
},
{
"epoch": 0.04,
"learning_rate": 0.001,
"loss": 2.1092,
"step": 65
},
{
"epoch": 0.05,
"learning_rate": 0.001,
"loss": 1.9553,
"step": 70
},
{
"epoch": 0.05,
"learning_rate": 0.001,
"loss": 2.121,
"step": 75
},
{
"epoch": 0.05,
"learning_rate": 0.001,
"loss": 2.1722,
"step": 80
},
{
"epoch": 0.06,
"learning_rate": 0.001,
"loss": 2.1552,
"step": 85
},
{
"epoch": 0.06,
"learning_rate": 0.001,
"loss": 2.5971,
"step": 90
},
{
"epoch": 0.06,
"learning_rate": 0.001,
"loss": 2.4684,
"step": 95
},
{
"epoch": 0.07,
"learning_rate": 0.001,
"loss": 2.0248,
"step": 100
},
{
"epoch": 0.07,
"learning_rate": 0.001,
"loss": 2.5156,
"step": 105
},
{
"epoch": 0.07,
"learning_rate": 0.001,
"loss": 2.3322,
"step": 110
},
{
"epoch": 0.08,
"learning_rate": 0.001,
"loss": 1.7698,
"step": 115
},
{
"epoch": 0.08,
"learning_rate": 0.001,
"loss": 2.3363,
"step": 120
},
{
"epoch": 0.08,
"learning_rate": 0.001,
"loss": 2.1079,
"step": 125
},
{
"epoch": 0.09,
"learning_rate": 0.001,
"loss": 2.0998,
"step": 130
},
{
"epoch": 0.09,
"learning_rate": 0.001,
"loss": 2.8265,
"step": 135
},
{
"epoch": 0.09,
"learning_rate": 0.001,
"loss": 2.6611,
"step": 140
},
{
"epoch": 0.09,
"learning_rate": 0.001,
"loss": 2.2187,
"step": 145
},
{
"epoch": 0.1,
"learning_rate": 0.001,
"loss": 2.33,
"step": 150
},
{
"epoch": 0.1,
"learning_rate": 0.001,
"loss": 2.6119,
"step": 155
},
{
"epoch": 0.1,
"learning_rate": 0.001,
"loss": 2.2203,
"step": 160
},
{
"epoch": 0.11,
"learning_rate": 0.001,
"loss": 2.5474,
"step": 165
},
{
"epoch": 0.11,
"learning_rate": 0.001,
"loss": 2.4763,
"step": 170
},
{
"epoch": 0.11,
"learning_rate": 0.001,
"loss": 2.6068,
"step": 175
},
{
"epoch": 0.12,
"learning_rate": 0.001,
"loss": 2.1221,
"step": 180
},
{
"epoch": 0.12,
"learning_rate": 0.001,
"loss": 2.1185,
"step": 185
},
{
"epoch": 0.12,
"learning_rate": 0.001,
"loss": 1.9481,
"step": 190
},
{
"epoch": 0.13,
"learning_rate": 0.001,
"loss": 2.823,
"step": 195
},
{
"epoch": 0.13,
"learning_rate": 0.001,
"loss": 2.4702,
"step": 200
},
{
"epoch": 0.13,
"learning_rate": 0.001,
"loss": 2.3721,
"step": 205
},
{
"epoch": 0.14,
"learning_rate": 0.001,
"loss": 2.4932,
"step": 210
},
{
"epoch": 0.14,
"learning_rate": 0.001,
"loss": 2.3977,
"step": 215
},
{
"epoch": 0.14,
"learning_rate": 0.001,
"loss": 2.5299,
"step": 220
},
{
"epoch": 0.15,
"learning_rate": 0.001,
"loss": 2.4818,
"step": 225
},
{
"epoch": 0.15,
"learning_rate": 0.001,
"loss": 2.3404,
"step": 230
},
{
"epoch": 0.15,
"learning_rate": 0.001,
"loss": 2.3614,
"step": 235
},
{
"epoch": 0.16,
"learning_rate": 0.001,
"loss": 2.1509,
"step": 240
},
{
"epoch": 0.16,
"learning_rate": 0.001,
"loss": 2.5301,
"step": 245
},
{
"epoch": 0.16,
"learning_rate": 0.001,
"loss": 2.3328,
"step": 250
},
{
"epoch": 0.17,
"learning_rate": 0.001,
"loss": 2.412,
"step": 255
},
{
"epoch": 0.17,
"learning_rate": 0.001,
"loss": 2.5923,
"step": 260
},
{
"epoch": 0.17,
"learning_rate": 0.001,
"loss": 2.2655,
"step": 265
},
{
"epoch": 0.18,
"learning_rate": 0.001,
"loss": 2.7524,
"step": 270
},
{
"epoch": 0.18,
"learning_rate": 0.001,
"loss": 2.6456,
"step": 275
},
{
"epoch": 0.18,
"learning_rate": 0.001,
"loss": 2.3814,
"step": 280
},
{
"epoch": 0.19,
"learning_rate": 0.001,
"loss": 2.3805,
"step": 285
},
{
"epoch": 0.19,
"learning_rate": 0.001,
"loss": 2.6727,
"step": 290
},
{
"epoch": 0.19,
"learning_rate": 0.001,
"loss": 2.3931,
"step": 295
},
{
"epoch": 0.2,
"learning_rate": 0.001,
"loss": 1.9526,
"step": 300
},
{
"epoch": 0.2,
"learning_rate": 0.001,
"loss": 2.7219,
"step": 305
},
{
"epoch": 0.2,
"learning_rate": 0.001,
"loss": 2.4869,
"step": 310
},
{
"epoch": 0.21,
"learning_rate": 0.001,
"loss": 2.4303,
"step": 315
},
{
"epoch": 0.21,
"learning_rate": 0.001,
"loss": 2.5916,
"step": 320
},
{
"epoch": 0.21,
"learning_rate": 0.001,
"loss": 2.3917,
"step": 325
},
{
"epoch": 0.22,
"learning_rate": 0.001,
"loss": 2.4881,
"step": 330
},
{
"epoch": 0.22,
"learning_rate": 0.001,
"loss": 2.3369,
"step": 335
},
{
"epoch": 0.22,
"learning_rate": 0.001,
"loss": 2.45,
"step": 340
},
{
"epoch": 0.23,
"learning_rate": 0.001,
"loss": 2.1916,
"step": 345
},
{
"epoch": 0.23,
"learning_rate": 0.001,
"loss": 2.2606,
"step": 350
},
{
"epoch": 0.23,
"learning_rate": 0.001,
"loss": 2.5533,
"step": 355
},
{
"epoch": 0.24,
"learning_rate": 0.001,
"loss": 2.1873,
"step": 360
},
{
"epoch": 0.24,
"learning_rate": 0.001,
"loss": 2.6057,
"step": 365
},
{
"epoch": 0.24,
"learning_rate": 0.001,
"loss": 2.4611,
"step": 370
},
{
"epoch": 0.25,
"learning_rate": 0.001,
"loss": 2.606,
"step": 375
},
{
"epoch": 0.25,
"learning_rate": 0.001,
"loss": 2.5645,
"step": 380
},
{
"epoch": 0.25,
"learning_rate": 0.001,
"loss": 2.7591,
"step": 385
},
{
"epoch": 0.26,
"learning_rate": 0.001,
"loss": 2.3399,
"step": 390
},
{
"epoch": 0.26,
"learning_rate": 0.001,
"loss": 2.3675,
"step": 395
},
{
"epoch": 0.26,
"learning_rate": 0.001,
"loss": 2.5141,
"step": 400
},
{
"epoch": 0.27,
"learning_rate": 0.001,
"loss": 2.2407,
"step": 405
},
{
"epoch": 0.27,
"learning_rate": 0.001,
"loss": 2.6321,
"step": 410
},
{
"epoch": 0.27,
"learning_rate": 0.001,
"loss": 2.6431,
"step": 415
},
{
"epoch": 0.27,
"learning_rate": 0.001,
"loss": 2.4707,
"step": 420
},
{
"epoch": 0.28,
"learning_rate": 0.001,
"loss": 2.3744,
"step": 425
},
{
"epoch": 0.28,
"learning_rate": 0.001,
"loss": 2.5658,
"step": 430
},
{
"epoch": 0.28,
"learning_rate": 0.001,
"loss": 1.9922,
"step": 435
},
{
"epoch": 0.29,
"learning_rate": 0.001,
"loss": 2.5948,
"step": 440
},
{
"epoch": 0.29,
"learning_rate": 0.001,
"loss": 2.2519,
"step": 445
},
{
"epoch": 0.29,
"learning_rate": 0.001,
"loss": 2.6248,
"step": 450
},
{
"epoch": 0.3,
"learning_rate": 0.001,
"loss": 2.5877,
"step": 455
},
{
"epoch": 0.3,
"learning_rate": 0.001,
"loss": 2.1311,
"step": 460
},
{
"epoch": 0.3,
"learning_rate": 0.001,
"loss": 2.6091,
"step": 465
},
{
"epoch": 0.31,
"learning_rate": 0.001,
"loss": 2.5488,
"step": 470
},
{
"epoch": 0.31,
"learning_rate": 0.001,
"loss": 2.499,
"step": 475
},
{
"epoch": 0.31,
"learning_rate": 0.001,
"loss": 2.5031,
"step": 480
},
{
"epoch": 0.32,
"learning_rate": 0.001,
"loss": 2.5783,
"step": 485
},
{
"epoch": 0.32,
"learning_rate": 0.001,
"loss": 2.6303,
"step": 490
},
{
"epoch": 0.32,
"learning_rate": 0.001,
"loss": 2.6344,
"step": 495
},
{
"epoch": 0.33,
"learning_rate": 0.001,
"loss": 2.7031,
"step": 500
},
{
"epoch": 0.33,
"learning_rate": 0.001,
"loss": 2.6677,
"step": 505
},
{
"epoch": 0.33,
"learning_rate": 0.001,
"loss": 2.7617,
"step": 510
},
{
"epoch": 0.34,
"learning_rate": 0.001,
"loss": 2.3122,
"step": 515
},
{
"epoch": 0.34,
"learning_rate": 0.001,
"loss": 2.3887,
"step": 520
},
{
"epoch": 0.34,
"learning_rate": 0.001,
"loss": 2.5569,
"step": 525
},
{
"epoch": 0.35,
"learning_rate": 0.001,
"loss": 2.4038,
"step": 530
},
{
"epoch": 0.35,
"learning_rate": 0.001,
"loss": 2.7413,
"step": 535
},
{
"epoch": 0.35,
"learning_rate": 0.001,
"loss": 2.5264,
"step": 540
},
{
"epoch": 0.36,
"learning_rate": 0.001,
"loss": 2.335,
"step": 545
},
{
"epoch": 0.36,
"learning_rate": 0.001,
"loss": 2.7574,
"step": 550
},
{
"epoch": 0.36,
"learning_rate": 0.001,
"loss": 2.4542,
"step": 555
},
{
"epoch": 0.37,
"learning_rate": 0.001,
"loss": 2.0885,
"step": 560
},
{
"epoch": 0.37,
"learning_rate": 0.001,
"loss": 2.6057,
"step": 565
},
{
"epoch": 0.37,
"learning_rate": 0.001,
"loss": 2.4971,
"step": 570
},
{
"epoch": 0.38,
"learning_rate": 0.001,
"loss": 2.6251,
"step": 575
},
{
"epoch": 0.38,
"learning_rate": 0.001,
"loss": 2.5784,
"step": 580
},
{
"epoch": 0.38,
"learning_rate": 0.001,
"loss": 2.3878,
"step": 585
},
{
"epoch": 0.39,
"learning_rate": 0.001,
"loss": 2.4271,
"step": 590
},
{
"epoch": 0.39,
"learning_rate": 0.001,
"loss": 2.6751,
"step": 595
},
{
"epoch": 0.39,
"learning_rate": 0.001,
"loss": 2.17,
"step": 600
},
{
"epoch": 0.4,
"learning_rate": 0.001,
"loss": 2.0794,
"step": 605
},
{
"epoch": 0.4,
"learning_rate": 0.001,
"loss": 2.6073,
"step": 610
},
{
"epoch": 0.4,
"learning_rate": 0.001,
"loss": 2.6717,
"step": 615
},
{
"epoch": 0.41,
"learning_rate": 0.001,
"loss": 2.8236,
"step": 620
},
{
"epoch": 0.41,
"learning_rate": 0.001,
"loss": 2.4193,
"step": 625
},
{
"epoch": 0.41,
"learning_rate": 0.001,
"loss": 2.7181,
"step": 630
},
{
"epoch": 0.42,
"learning_rate": 0.001,
"loss": 2.6567,
"step": 635
},
{
"epoch": 0.42,
"learning_rate": 0.001,
"loss": 2.1919,
"step": 640
},
{
"epoch": 0.42,
"learning_rate": 0.001,
"loss": 2.5645,
"step": 645
},
{
"epoch": 0.43,
"learning_rate": 0.001,
"loss": 2.3313,
"step": 650
},
{
"epoch": 0.43,
"learning_rate": 0.001,
"loss": 2.533,
"step": 655
},
{
"epoch": 0.43,
"learning_rate": 0.001,
"loss": 2.621,
"step": 660
},
{
"epoch": 0.44,
"learning_rate": 0.001,
"loss": 2.5693,
"step": 665
},
{
"epoch": 0.44,
"learning_rate": 0.001,
"loss": 2.6943,
"step": 670
},
{
"epoch": 0.44,
"learning_rate": 0.001,
"loss": 2.2715,
"step": 675
},
{
"epoch": 0.45,
"learning_rate": 0.001,
"loss": 2.5614,
"step": 680
},
{
"epoch": 0.45,
"learning_rate": 0.001,
"loss": 2.5784,
"step": 685
},
{
"epoch": 0.45,
"learning_rate": 0.001,
"loss": 2.0613,
"step": 690
},
{
"epoch": 0.45,
"learning_rate": 0.001,
"loss": 2.582,
"step": 695
},
{
"epoch": 0.46,
"learning_rate": 0.001,
"loss": 2.5178,
"step": 700
},
{
"epoch": 0.46,
"learning_rate": 0.001,
"loss": 2.432,
"step": 705
},
{
"epoch": 0.46,
"learning_rate": 0.001,
"loss": 2.1979,
"step": 710
},
{
"epoch": 0.47,
"learning_rate": 0.001,
"loss": 2.4317,
"step": 715
},
{
"epoch": 0.47,
"learning_rate": 0.001,
"loss": 2.285,
"step": 720
},
{
"epoch": 0.47,
"learning_rate": 0.001,
"loss": 2.6127,
"step": 725
},
{
"epoch": 0.48,
"learning_rate": 0.001,
"loss": 2.6221,
"step": 730
},
{
"epoch": 0.48,
"learning_rate": 0.001,
"loss": 2.7519,
"step": 735
},
{
"epoch": 0.48,
"learning_rate": 0.001,
"loss": 2.7351,
"step": 740
},
{
"epoch": 0.49,
"learning_rate": 0.001,
"loss": 2.5224,
"step": 745
},
{
"epoch": 0.49,
"learning_rate": 0.001,
"loss": 2.5145,
"step": 750
},
{
"epoch": 0.49,
"learning_rate": 0.001,
"loss": 2.5122,
"step": 755
},
{
"epoch": 0.5,
"learning_rate": 0.001,
"loss": 2.2971,
"step": 760
},
{
"epoch": 0.5,
"learning_rate": 0.001,
"loss": 2.5628,
"step": 765
},
{
"epoch": 0.5,
"learning_rate": 0.001,
"loss": 2.5803,
"step": 770
},
{
"epoch": 0.51,
"learning_rate": 0.001,
"loss": 2.5209,
"step": 775
},
{
"epoch": 0.51,
"learning_rate": 0.001,
"loss": 2.5051,
"step": 780
},
{
"epoch": 0.51,
"learning_rate": 0.001,
"loss": 2.5275,
"step": 785
},
{
"epoch": 0.52,
"learning_rate": 0.001,
"loss": 2.5019,
"step": 790
},
{
"epoch": 0.52,
"learning_rate": 0.001,
"loss": 2.1714,
"step": 795
},
{
"epoch": 0.52,
"learning_rate": 0.001,
"loss": 2.87,
"step": 800
},
{
"epoch": 0.53,
"learning_rate": 0.001,
"loss": 2.6468,
"step": 805
},
{
"epoch": 0.53,
"learning_rate": 0.001,
"loss": 2.4115,
"step": 810
},
{
"epoch": 0.53,
"learning_rate": 0.001,
"loss": 2.5377,
"step": 815
},
{
"epoch": 0.54,
"learning_rate": 0.001,
"loss": 2.4986,
"step": 820
},
{
"epoch": 0.54,
"learning_rate": 0.001,
"loss": 2.6485,
"step": 825
},
{
"epoch": 0.54,
"learning_rate": 0.001,
"loss": 2.6314,
"step": 830
},
{
"epoch": 0.55,
"learning_rate": 0.001,
"loss": 2.3516,
"step": 835
},
{
"epoch": 0.55,
"learning_rate": 0.001,
"loss": 2.39,
"step": 840
},
{
"epoch": 0.55,
"learning_rate": 0.001,
"loss": 2.3759,
"step": 845
},
{
"epoch": 0.56,
"learning_rate": 0.001,
"loss": 2.5348,
"step": 850
},
{
"epoch": 0.56,
"learning_rate": 0.001,
"loss": 2.6843,
"step": 855
},
{
"epoch": 0.56,
"learning_rate": 0.001,
"loss": 2.6289,
"step": 860
},
{
"epoch": 0.57,
"learning_rate": 0.001,
"loss": 2.4023,
"step": 865
},
{
"epoch": 0.57,
"learning_rate": 0.001,
"loss": 2.5979,
"step": 870
},
{
"epoch": 0.57,
"learning_rate": 0.001,
"loss": 2.4683,
"step": 875
},
{
"epoch": 0.58,
"learning_rate": 0.001,
"loss": 2.7099,
"step": 880
},
{
"epoch": 0.58,
"learning_rate": 0.001,
"loss": 2.3129,
"step": 885
},
{
"epoch": 0.58,
"learning_rate": 0.001,
"loss": 2.4508,
"step": 890
},
{
"epoch": 0.59,
"learning_rate": 0.001,
"loss": 2.6484,
"step": 895
},
{
"epoch": 0.59,
"learning_rate": 0.001,
"loss": 2.4541,
"step": 900
},
{
"epoch": 0.59,
"learning_rate": 0.001,
"loss": 2.5101,
"step": 905
},
{
"epoch": 0.6,
"learning_rate": 0.001,
"loss": 2.6002,
"step": 910
},
{
"epoch": 0.6,
"learning_rate": 0.001,
"loss": 2.742,
"step": 915
},
{
"epoch": 0.6,
"learning_rate": 0.001,
"loss": 2.4772,
"step": 920
},
{
"epoch": 0.61,
"learning_rate": 0.001,
"loss": 2.4328,
"step": 925
},
{
"epoch": 0.61,
"learning_rate": 0.001,
"loss": 2.7404,
"step": 930
},
{
"epoch": 0.61,
"learning_rate": 0.001,
"loss": 2.5969,
"step": 935
},
{
"epoch": 0.62,
"learning_rate": 0.001,
"loss": 2.1367,
"step": 940
},
{
"epoch": 0.62,
"learning_rate": 0.001,
"loss": 2.703,
"step": 945
},
{
"epoch": 0.62,
"learning_rate": 0.001,
"loss": 2.5638,
"step": 950
},
{
"epoch": 0.62,
"learning_rate": 0.001,
"loss": 1.8322,
"step": 955
},
{
"epoch": 0.63,
"learning_rate": 0.001,
"loss": 2.5764,
"step": 960
},
{
"epoch": 0.63,
"learning_rate": 0.001,
"loss": 2.4679,
"step": 965
},
{
"epoch": 0.63,
"learning_rate": 0.001,
"loss": 2.1691,
"step": 970
},
{
"epoch": 0.64,
"learning_rate": 0.001,
"loss": 2.3252,
"step": 975
},
{
"epoch": 0.64,
"learning_rate": 0.001,
"loss": 2.3988,
"step": 980
},
{
"epoch": 0.64,
"learning_rate": 0.001,
"loss": 2.221,
"step": 985
},
{
"epoch": 0.65,
"learning_rate": 0.001,
"loss": 2.4715,
"step": 990
},
{
"epoch": 0.65,
"learning_rate": 0.001,
"loss": 2.2479,
"step": 995
},
{
"epoch": 0.65,
"learning_rate": 0.001,
"loss": 2.3859,
"step": 1000
},
{
"epoch": 0.66,
"learning_rate": 0.001,
"loss": 2.7889,
"step": 1005
},
{
"epoch": 0.66,
"learning_rate": 0.001,
"loss": 2.625,
"step": 1010
},
{
"epoch": 0.66,
"learning_rate": 0.001,
"loss": 2.324,
"step": 1015
},
{
"epoch": 0.67,
"learning_rate": 0.001,
"loss": 2.5879,
"step": 1020
},
{
"epoch": 0.67,
"learning_rate": 0.001,
"loss": 2.4721,
"step": 1025
},
{
"epoch": 0.67,
"learning_rate": 0.001,
"loss": 2.6365,
"step": 1030
},
{
"epoch": 0.68,
"learning_rate": 0.001,
"loss": 2.6356,
"step": 1035
},
{
"epoch": 0.68,
"learning_rate": 0.001,
"loss": 2.4022,
"step": 1040
},
{
"epoch": 0.68,
"learning_rate": 0.001,
"loss": 2.7109,
"step": 1045
},
{
"epoch": 0.69,
"learning_rate": 0.001,
"loss": 2.373,
"step": 1050
},
{
"epoch": 0.69,
"learning_rate": 0.001,
"loss": 2.523,
"step": 1055
},
{
"epoch": 0.69,
"learning_rate": 0.001,
"loss": 2.536,
"step": 1060
},
{
"epoch": 0.7,
"learning_rate": 0.001,
"loss": 2.6047,
"step": 1065
},
{
"epoch": 0.7,
"learning_rate": 0.001,
"loss": 2.5583,
"step": 1070
},
{
"epoch": 0.7,
"learning_rate": 0.001,
"loss": 2.7925,
"step": 1075
},
{
"epoch": 0.71,
"learning_rate": 0.001,
"loss": 2.0572,
"step": 1080
},
{
"epoch": 0.71,
"learning_rate": 0.001,
"loss": 2.3024,
"step": 1085
},
{
"epoch": 0.71,
"learning_rate": 0.001,
"loss": 2.0677,
"step": 1090
},
{
"epoch": 0.72,
"learning_rate": 0.001,
"loss": 2.5241,
"step": 1095
},
{
"epoch": 0.72,
"learning_rate": 0.001,
"loss": 2.7447,
"step": 1100
},
{
"epoch": 0.72,
"learning_rate": 0.001,
"loss": 2.6435,
"step": 1105
},
{
"epoch": 0.73,
"learning_rate": 0.001,
"loss": 2.4226,
"step": 1110
},
{
"epoch": 0.73,
"learning_rate": 0.001,
"loss": 2.6891,
"step": 1115
},
{
"epoch": 0.73,
"learning_rate": 0.001,
"loss": 2.458,
"step": 1120
},
{
"epoch": 0.74,
"learning_rate": 0.001,
"loss": 2.6884,
"step": 1125
},
{
"epoch": 0.74,
"learning_rate": 0.001,
"loss": 2.7058,
"step": 1130
},
{
"epoch": 0.74,
"learning_rate": 0.001,
"loss": 2.5828,
"step": 1135
},
{
"epoch": 0.75,
"learning_rate": 0.001,
"loss": 2.7107,
"step": 1140
},
{
"epoch": 0.75,
"learning_rate": 0.001,
"loss": 2.3635,
"step": 1145
},
{
"epoch": 0.75,
"learning_rate": 0.001,
"loss": 2.5258,
"step": 1150
},
{
"epoch": 0.76,
"learning_rate": 0.001,
"loss": 2.6046,
"step": 1155
},
{
"epoch": 0.76,
"learning_rate": 0.001,
"loss": 2.5215,
"step": 1160
},
{
"epoch": 0.76,
"learning_rate": 0.001,
"loss": 2.1417,
"step": 1165
},
{
"epoch": 0.77,
"learning_rate": 0.001,
"loss": 2.5903,
"step": 1170
},
{
"epoch": 0.77,
"learning_rate": 0.001,
"loss": 2.5376,
"step": 1175
},
{
"epoch": 0.77,
"learning_rate": 0.001,
"loss": 2.5716,
"step": 1180
},
{
"epoch": 0.78,
"learning_rate": 0.001,
"loss": 2.5904,
"step": 1185
},
{
"epoch": 0.78,
"learning_rate": 0.001,
"loss": 2.6511,
"step": 1190
},
{
"epoch": 0.78,
"learning_rate": 0.001,
"loss": 2.1679,
"step": 1195
},
{
"epoch": 0.79,
"learning_rate": 0.001,
"loss": 2.4169,
"step": 1200
},
{
"epoch": 0.79,
"learning_rate": 0.001,
"loss": 2.7519,
"step": 1205
},
{
"epoch": 0.79,
"learning_rate": 0.001,
"loss": 2.4385,
"step": 1210
},
{
"epoch": 0.8,
"learning_rate": 0.001,
"loss": 1.9145,
"step": 1215
},
{
"epoch": 0.8,
"learning_rate": 0.001,
"loss": 2.5898,
"step": 1220
},
{
"epoch": 0.8,
"learning_rate": 0.001,
"loss": 2.4768,
"step": 1225
},
{
"epoch": 0.8,
"learning_rate": 0.001,
"loss": 2.5546,
"step": 1230
},
{
"epoch": 0.81,
"learning_rate": 0.001,
"loss": 2.5279,
"step": 1235
},
{
"epoch": 0.81,
"learning_rate": 0.001,
"loss": 2.498,
"step": 1240
},
{
"epoch": 0.81,
"learning_rate": 0.001,
"loss": 2.4277,
"step": 1245
},
{
"epoch": 0.82,
"learning_rate": 0.001,
"loss": 2.9934,
"step": 1250
},
{
"epoch": 0.82,
"learning_rate": 0.001,
"loss": 2.3367,
"step": 1255
},
{
"epoch": 0.82,
"learning_rate": 0.001,
"loss": 2.012,
"step": 1260
},
{
"epoch": 0.83,
"learning_rate": 0.001,
"loss": 2.4148,
"step": 1265
},
{
"epoch": 0.83,
"learning_rate": 0.001,
"loss": 2.3207,
"step": 1270
},
{
"epoch": 0.83,
"learning_rate": 0.001,
"loss": 2.5629,
"step": 1275
},
{
"epoch": 0.84,
"learning_rate": 0.001,
"loss": 2.4338,
"step": 1280
},
{
"epoch": 0.84,
"learning_rate": 0.001,
"loss": 2.4612,
"step": 1285
},
{
"epoch": 0.84,
"learning_rate": 0.001,
"loss": 2.5406,
"step": 1290
},
{
"epoch": 0.85,
"learning_rate": 0.001,
"loss": 2.5438,
"step": 1295
},
{
"epoch": 0.85,
"learning_rate": 0.001,
"loss": 2.6348,
"step": 1300
},
{
"epoch": 0.85,
"learning_rate": 0.001,
"loss": 2.1436,
"step": 1305
},
{
"epoch": 0.86,
"learning_rate": 0.001,
"loss": 2.4825,
"step": 1310
},
{
"epoch": 0.86,
"learning_rate": 0.001,
"loss": 2.2579,
"step": 1315
},
{
"epoch": 0.86,
"learning_rate": 0.001,
"loss": 2.5234,
"step": 1320
},
{
"epoch": 0.87,
"learning_rate": 0.001,
"loss": 2.5127,
"step": 1325
},
{
"epoch": 0.87,
"learning_rate": 0.001,
"loss": 2.3415,
"step": 1330
},
{
"epoch": 0.87,
"learning_rate": 0.001,
"loss": 2.4005,
"step": 1335
},
{
"epoch": 0.88,
"learning_rate": 0.001,
"loss": 2.4731,
"step": 1340
},
{
"epoch": 0.88,
"learning_rate": 0.001,
"loss": 2.8025,
"step": 1345
},
{
"epoch": 0.88,
"learning_rate": 0.001,
"loss": 2.1733,
"step": 1350
},
{
"epoch": 0.89,
"learning_rate": 0.001,
"loss": 2.5599,
"step": 1355
},
{
"epoch": 0.89,
"learning_rate": 0.001,
"loss": 2.6287,
"step": 1360
},
{
"epoch": 0.89,
"learning_rate": 0.001,
"loss": 2.5367,
"step": 1365
},
{
"epoch": 0.9,
"learning_rate": 0.001,
"loss": 2.3331,
"step": 1370
},
{
"epoch": 0.9,
"learning_rate": 0.001,
"loss": 2.2061,
"step": 1375
},
{
"epoch": 0.9,
"learning_rate": 0.001,
"loss": 2.3905,
"step": 1380
},
{
"epoch": 0.91,
"learning_rate": 0.001,
"loss": 2.6162,
"step": 1385
},
{
"epoch": 0.91,
"learning_rate": 0.001,
"loss": 2.2712,
"step": 1390
},
{
"epoch": 0.91,
"learning_rate": 0.001,
"loss": 2.3674,
"step": 1395
},
{
"epoch": 0.92,
"learning_rate": 0.001,
"loss": 2.4791,
"step": 1400
},
{
"epoch": 0.92,
"learning_rate": 0.001,
"loss": 2.6425,
"step": 1405
},
{
"epoch": 0.92,
"learning_rate": 0.001,
"loss": 2.4496,
"step": 1410
},
{
"epoch": 0.93,
"learning_rate": 0.001,
"loss": 2.5499,
"step": 1415
},
{
"epoch": 0.93,
"learning_rate": 0.001,
"loss": 2.5993,
"step": 1420
},
{
"epoch": 0.93,
"learning_rate": 0.001,
"loss": 2.5543,
"step": 1425
},
{
"epoch": 0.94,
"learning_rate": 0.001,
"loss": 2.2981,
"step": 1430
},
{
"epoch": 0.94,
"learning_rate": 0.001,
"loss": 2.7974,
"step": 1435
},
{
"epoch": 0.94,
"learning_rate": 0.001,
"loss": 2.0898,
"step": 1440
},
{
"epoch": 0.95,
"learning_rate": 0.001,
"loss": 2.561,
"step": 1445
},
{
"epoch": 0.95,
"learning_rate": 0.001,
"loss": 2.2922,
"step": 1450
},
{
"epoch": 0.95,
"learning_rate": 0.001,
"loss": 2.3195,
"step": 1455
},
{
"epoch": 0.96,
"learning_rate": 0.001,
"loss": 2.4302,
"step": 1460
},
{
"epoch": 0.96,
"learning_rate": 0.001,
"loss": 2.6357,
"step": 1465
},
{
"epoch": 0.96,
"learning_rate": 0.001,
"loss": 2.1727,
"step": 1470
},
{
"epoch": 0.97,
"learning_rate": 0.001,
"loss": 2.5076,
"step": 1475
},
{
"epoch": 0.97,
"learning_rate": 0.001,
"loss": 2.7291,
"step": 1480
},
{
"epoch": 0.97,
"learning_rate": 0.001,
"loss": 2.443,
"step": 1485
},
{
"epoch": 0.98,
"learning_rate": 0.001,
"loss": 2.4889,
"step": 1490
},
{
"epoch": 0.98,
"learning_rate": 0.001,
"loss": 2.4709,
"step": 1495
},
{
"epoch": 0.98,
"learning_rate": 0.001,
"loss": 2.3124,
"step": 1500
},
{
"epoch": 0.98,
"learning_rate": 0.001,
"loss": 2.5515,
"step": 1505
},
{
"epoch": 0.99,
"learning_rate": 0.001,
"loss": 2.3861,
"step": 1510
},
{
"epoch": 0.99,
"learning_rate": 0.001,
"loss": 2.4269,
"step": 1515
},
{
"epoch": 0.99,
"learning_rate": 0.001,
"loss": 2.4074,
"step": 1520
},
{
"epoch": 1.0,
"learning_rate": 0.001,
"loss": 2.2783,
"step": 1525
},
{
"epoch": 1.0,
"learning_rate": 0.001,
"loss": 2.4139,
"step": 1530
},
{
"epoch": 1.0,
"learning_rate": 0.001,
"loss": 1.9251,
"step": 1535
},
{
"epoch": 1.01,
"learning_rate": 0.001,
"loss": 2.1038,
"step": 1540
},
{
"epoch": 1.01,
"learning_rate": 0.001,
"loss": 1.9436,
"step": 1545
},
{
"epoch": 1.01,
"learning_rate": 0.001,
"loss": 1.9581,
"step": 1550
},
{
"epoch": 1.02,
"learning_rate": 0.001,
"loss": 2.1705,
"step": 1555
},
{
"epoch": 1.02,
"learning_rate": 0.001,
"loss": 2.1393,
"step": 1560
},
{
"epoch": 1.02,
"learning_rate": 0.001,
"loss": 1.741,
"step": 1565
},
{
"epoch": 1.03,
"learning_rate": 0.001,
"loss": 1.9232,
"step": 1570
},
{
"epoch": 1.03,
"learning_rate": 0.001,
"loss": 1.9046,
"step": 1575
},
{
"epoch": 1.03,
"learning_rate": 0.001,
"loss": 1.8102,
"step": 1580
},
{
"epoch": 1.04,
"learning_rate": 0.001,
"loss": 1.7362,
"step": 1585
},
{
"epoch": 1.04,
"learning_rate": 0.001,
"loss": 1.8057,
"step": 1590
},
{
"epoch": 1.04,
"learning_rate": 0.001,
"loss": 2.0777,
"step": 1595
},
{
"epoch": 1.05,
"learning_rate": 0.001,
"loss": 2.1163,
"step": 1600
},
{
"epoch": 1.05,
"learning_rate": 0.001,
"loss": 1.8468,
"step": 1605
},
{
"epoch": 1.05,
"learning_rate": 0.001,
"loss": 2.3334,
"step": 1610
},
{
"epoch": 1.06,
"learning_rate": 0.001,
"loss": 2.1232,
"step": 1615
},
{
"epoch": 1.06,
"learning_rate": 0.001,
"loss": 1.7049,
"step": 1620
},
{
"epoch": 1.06,
"learning_rate": 0.001,
"loss": 2.1545,
"step": 1625
},
{
"epoch": 1.07,
"learning_rate": 0.001,
"loss": 2.3166,
"step": 1630
},
{
"epoch": 1.07,
"learning_rate": 0.001,
"loss": 2.103,
"step": 1635
},
{
"epoch": 1.07,
"learning_rate": 0.001,
"loss": 1.9295,
"step": 1640
},
{
"epoch": 1.08,
"learning_rate": 0.001,
"loss": 2.0598,
"step": 1645
},
{
"epoch": 1.08,
"learning_rate": 0.001,
"loss": 2.3646,
"step": 1650
},
{
"epoch": 1.08,
"learning_rate": 0.001,
"loss": 2.1853,
"step": 1655
},
{
"epoch": 1.09,
"learning_rate": 0.001,
"loss": 2.0635,
"step": 1660
},
{
"epoch": 1.09,
"learning_rate": 0.001,
"loss": 1.9439,
"step": 1665
},
{
"epoch": 1.09,
"learning_rate": 0.001,
"loss": 2.469,
"step": 1670
},
{
"epoch": 1.1,
"learning_rate": 0.001,
"loss": 2.1583,
"step": 1675
},
{
"epoch": 1.1,
"learning_rate": 0.001,
"loss": 2.079,
"step": 1680
},
{
"epoch": 1.1,
"learning_rate": 0.001,
"loss": 2.0118,
"step": 1685
},
{
"epoch": 1.11,
"learning_rate": 0.001,
"loss": 2.011,
"step": 1690
},
{
"epoch": 1.11,
"learning_rate": 0.001,
"loss": 2.3133,
"step": 1695
},
{
"epoch": 1.11,
"learning_rate": 0.001,
"loss": 2.1873,
"step": 1700
},
{
"epoch": 1.12,
"learning_rate": 0.001,
"loss": 1.7311,
"step": 1705
},
{
"epoch": 1.12,
"learning_rate": 0.001,
"loss": 1.7269,
"step": 1710
},
{
"epoch": 1.12,
"learning_rate": 0.001,
"loss": 2.3935,
"step": 1715
},
{
"epoch": 1.13,
"learning_rate": 0.001,
"loss": 2.1104,
"step": 1720
},
{
"epoch": 1.13,
"learning_rate": 0.001,
"loss": 2.2349,
"step": 1725
},
{
"epoch": 1.13,
"learning_rate": 0.001,
"loss": 1.6555,
"step": 1730
},
{
"epoch": 1.14,
"learning_rate": 0.001,
"loss": 1.9011,
"step": 1735
},
{
"epoch": 1.14,
"learning_rate": 0.001,
"loss": 2.2142,
"step": 1740
},
{
"epoch": 1.14,
"learning_rate": 0.001,
"loss": 2.1273,
"step": 1745
},
{
"epoch": 1.15,
"learning_rate": 0.001,
"loss": 2.4855,
"step": 1750
},
{
"epoch": 1.15,
"learning_rate": 0.001,
"loss": 2.3025,
"step": 1755
},
{
"epoch": 1.15,
"learning_rate": 0.001,
"loss": 1.9496,
"step": 1760
},
{
"epoch": 1.16,
"learning_rate": 0.001,
"loss": 1.9856,
"step": 1765
},
{
"epoch": 1.16,
"learning_rate": 0.001,
"loss": 2.1494,
"step": 1770
},
{
"epoch": 1.16,
"learning_rate": 0.001,
"loss": 1.9905,
"step": 1775
},
{
"epoch": 1.16,
"learning_rate": 0.001,
"loss": 2.1177,
"step": 1780
},
{
"epoch": 1.17,
"learning_rate": 0.001,
"loss": 1.9396,
"step": 1785
},
{
"epoch": 1.17,
"learning_rate": 0.001,
"loss": 1.8821,
"step": 1790
},
{
"epoch": 1.17,
"learning_rate": 0.001,
"loss": 2.1909,
"step": 1795
},
{
"epoch": 1.18,
"learning_rate": 0.001,
"loss": 1.9459,
"step": 1800
},
{
"epoch": 1.18,
"learning_rate": 0.001,
"loss": 2.1047,
"step": 1805
},
{
"epoch": 1.18,
"learning_rate": 0.001,
"loss": 1.833,
"step": 1810
},
{
"epoch": 1.19,
"learning_rate": 0.001,
"loss": 2.0839,
"step": 1815
},
{
"epoch": 1.19,
"learning_rate": 0.001,
"loss": 2.0736,
"step": 1820
},
{
"epoch": 1.19,
"learning_rate": 0.001,
"loss": 2.0034,
"step": 1825
},
{
"epoch": 1.2,
"learning_rate": 0.001,
"loss": 2.0788,
"step": 1830
},
{
"epoch": 1.2,
"learning_rate": 0.001,
"loss": 2.1903,
"step": 1835
},
{
"epoch": 1.2,
"learning_rate": 0.001,
"loss": 2.0829,
"step": 1840
},
{
"epoch": 1.21,
"learning_rate": 0.001,
"loss": 2.2449,
"step": 1845
},
{
"epoch": 1.21,
"learning_rate": 0.001,
"loss": 2.1561,
"step": 1850
},
{
"epoch": 1.21,
"learning_rate": 0.001,
"loss": 2.1827,
"step": 1855
},
{
"epoch": 1.22,
"learning_rate": 0.001,
"loss": 2.1885,
"step": 1860
},
{
"epoch": 1.22,
"learning_rate": 0.001,
"loss": 2.049,
"step": 1865
},
{
"epoch": 1.22,
"learning_rate": 0.001,
"loss": 2.2107,
"step": 1870
},
{
"epoch": 1.23,
"learning_rate": 0.001,
"loss": 2.0679,
"step": 1875
},
{
"epoch": 1.23,
"learning_rate": 0.001,
"loss": 2.2887,
"step": 1880
},
{
"epoch": 1.23,
"learning_rate": 0.001,
"loss": 2.2472,
"step": 1885
},
{
"epoch": 1.24,
"learning_rate": 0.001,
"loss": 2.1993,
"step": 1890
},
{
"epoch": 1.24,
"learning_rate": 0.001,
"loss": 2.1636,
"step": 1895
},
{
"epoch": 1.24,
"learning_rate": 0.001,
"loss": 1.9547,
"step": 1900
},
{
"epoch": 1.25,
"learning_rate": 0.001,
"loss": 2.3164,
"step": 1905
},
{
"epoch": 1.25,
"learning_rate": 0.001,
"loss": 2.2847,
"step": 1910
},
{
"epoch": 1.25,
"learning_rate": 0.001,
"loss": 2.2196,
"step": 1915
},
{
"epoch": 1.26,
"learning_rate": 0.001,
"loss": 2.1823,
"step": 1920
},
{
"epoch": 1.26,
"learning_rate": 0.001,
"loss": 2.2532,
"step": 1925
},
{
"epoch": 1.26,
"learning_rate": 0.001,
"loss": 2.2451,
"step": 1930
},
{
"epoch": 1.27,
"learning_rate": 0.001,
"loss": 1.984,
"step": 1935
},
{
"epoch": 1.27,
"learning_rate": 0.001,
"loss": 2.0976,
"step": 1940
},
{
"epoch": 1.27,
"learning_rate": 0.001,
"loss": 2.0658,
"step": 1945
},
{
"epoch": 1.28,
"learning_rate": 0.001,
"loss": 2.316,
"step": 1950
},
{
"epoch": 1.28,
"learning_rate": 0.001,
"loss": 2.2036,
"step": 1955
},
{
"epoch": 1.28,
"learning_rate": 0.001,
"loss": 2.0602,
"step": 1960
},
{
"epoch": 1.29,
"learning_rate": 0.001,
"loss": 1.971,
"step": 1965
},
{
"epoch": 1.29,
"learning_rate": 0.001,
"loss": 2.1463,
"step": 1970
},
{
"epoch": 1.29,
"learning_rate": 0.001,
"loss": 1.7689,
"step": 1975
},
{
"epoch": 1.3,
"learning_rate": 0.001,
"loss": 2.014,
"step": 1980
},
{
"epoch": 1.3,
"learning_rate": 0.001,
"loss": 2.3631,
"step": 1985
},
{
"epoch": 1.3,
"learning_rate": 0.001,
"loss": 2.2294,
"step": 1990
},
{
"epoch": 1.31,
"learning_rate": 0.001,
"loss": 1.8407,
"step": 1995
},
{
"epoch": 1.31,
"learning_rate": 0.001,
"loss": 1.9365,
"step": 2000
},
{
"epoch": 1.31,
"learning_rate": 0.001,
"loss": 2.0895,
"step": 2005
},
{
"epoch": 1.32,
"learning_rate": 0.001,
"loss": 2.1625,
"step": 2010
},
{
"epoch": 1.32,
"learning_rate": 0.001,
"loss": 2.053,
"step": 2015
},
{
"epoch": 1.32,
"learning_rate": 0.001,
"loss": 2.0925,
"step": 2020
},
{
"epoch": 1.33,
"learning_rate": 0.001,
"loss": 1.9764,
"step": 2025
},
{
"epoch": 1.33,
"learning_rate": 0.001,
"loss": 2.1212,
"step": 2030
},
{
"epoch": 1.33,
"learning_rate": 0.001,
"loss": 1.923,
"step": 2035
},
{
"epoch": 1.34,
"learning_rate": 0.001,
"loss": 2.2899,
"step": 2040
},
{
"epoch": 1.34,
"learning_rate": 0.001,
"loss": 2.3056,
"step": 2045
},
{
"epoch": 1.34,
"learning_rate": 0.001,
"loss": 2.0217,
"step": 2050
},
{
"epoch": 1.34,
"learning_rate": 0.001,
"loss": 2.1757,
"step": 2055
},
{
"epoch": 1.35,
"learning_rate": 0.001,
"loss": 1.7524,
"step": 2060
},
{
"epoch": 1.35,
"learning_rate": 0.001,
"loss": 2.3042,
"step": 2065
},
{
"epoch": 1.35,
"learning_rate": 0.001,
"loss": 2.164,
"step": 2070
},
{
"epoch": 1.36,
"learning_rate": 0.001,
"loss": 2.3265,
"step": 2075
},
{
"epoch": 1.36,
"learning_rate": 0.001,
"loss": 2.4762,
"step": 2080
},
{
"epoch": 1.36,
"learning_rate": 0.001,
"loss": 2.2577,
"step": 2085
},
{
"epoch": 1.37,
"learning_rate": 0.001,
"loss": 2.4857,
"step": 2090
},
{
"epoch": 1.37,
"learning_rate": 0.001,
"loss": 2.1887,
"step": 2095
},
{
"epoch": 1.37,
"learning_rate": 0.001,
"loss": 2.3144,
"step": 2100
},
{
"epoch": 1.38,
"learning_rate": 0.001,
"loss": 2.2387,
"step": 2105
},
{
"epoch": 1.38,
"learning_rate": 0.001,
"loss": 2.5008,
"step": 2110
},
{
"epoch": 1.38,
"learning_rate": 0.001,
"loss": 2.0604,
"step": 2115
},
{
"epoch": 1.39,
"learning_rate": 0.001,
"loss": 2.414,
"step": 2120
},
{
"epoch": 1.39,
"learning_rate": 0.001,
"loss": 2.2392,
"step": 2125
},
{
"epoch": 1.39,
"learning_rate": 0.001,
"loss": 2.1842,
"step": 2130
},
{
"epoch": 1.4,
"learning_rate": 0.001,
"loss": 2.0674,
"step": 2135
},
{
"epoch": 1.4,
"learning_rate": 0.001,
"loss": 2.2088,
"step": 2140
},
{
"epoch": 1.4,
"learning_rate": 0.001,
"loss": 2.363,
"step": 2145
},
{
"epoch": 1.41,
"learning_rate": 0.001,
"loss": 2.1834,
"step": 2150
},
{
"epoch": 1.41,
"learning_rate": 0.001,
"loss": 2.1047,
"step": 2155
},
{
"epoch": 1.41,
"learning_rate": 0.001,
"loss": 1.8624,
"step": 2160
},
{
"epoch": 1.42,
"learning_rate": 0.001,
"loss": 2.0746,
"step": 2165
},
{
"epoch": 1.42,
"learning_rate": 0.001,
"loss": 2.4054,
"step": 2170
},
{
"epoch": 1.42,
"learning_rate": 0.001,
"loss": 2.326,
"step": 2175
},
{
"epoch": 1.43,
"learning_rate": 0.001,
"loss": 2.1054,
"step": 2180
},
{
"epoch": 1.43,
"learning_rate": 0.001,
"loss": 1.7848,
"step": 2185
},
{
"epoch": 1.43,
"learning_rate": 0.001,
"loss": 2.2795,
"step": 2190
},
{
"epoch": 1.44,
"learning_rate": 0.001,
"loss": 1.9866,
"step": 2195
},
{
"epoch": 1.44,
"learning_rate": 0.001,
"loss": 2.5626,
"step": 2200
},
{
"epoch": 1.44,
"learning_rate": 0.001,
"loss": 2.2101,
"step": 2205
},
{
"epoch": 1.45,
"learning_rate": 0.001,
"loss": 2.286,
"step": 2210
},
{
"epoch": 1.45,
"learning_rate": 0.001,
"loss": 1.8204,
"step": 2215
},
{
"epoch": 1.45,
"learning_rate": 0.001,
"loss": 2.1886,
"step": 2220
},
{
"epoch": 1.46,
"learning_rate": 0.001,
"loss": 2.0936,
"step": 2225
},
{
"epoch": 1.46,
"learning_rate": 0.001,
"loss": 2.5121,
"step": 2230
},
{
"epoch": 1.46,
"learning_rate": 0.001,
"loss": 2.2165,
"step": 2235
},
{
"epoch": 1.47,
"learning_rate": 0.001,
"loss": 1.8987,
"step": 2240
},
{
"epoch": 1.47,
"learning_rate": 0.001,
"loss": 1.8333,
"step": 2245
},
{
"epoch": 1.47,
"learning_rate": 0.001,
"loss": 2.2611,
"step": 2250
},
{
"epoch": 1.48,
"learning_rate": 0.001,
"loss": 2.2649,
"step": 2255
},
{
"epoch": 1.48,
"learning_rate": 0.001,
"loss": 2.0907,
"step": 2260
},
{
"epoch": 1.48,
"learning_rate": 0.001,
"loss": 2.2594,
"step": 2265
},
{
"epoch": 1.49,
"learning_rate": 0.001,
"loss": 2.1826,
"step": 2270
},
{
"epoch": 1.49,
"learning_rate": 0.001,
"loss": 2.0087,
"step": 2275
},
{
"epoch": 1.49,
"learning_rate": 0.001,
"loss": 2.0936,
"step": 2280
},
{
"epoch": 1.5,
"learning_rate": 0.001,
"loss": 2.1456,
"step": 2285
},
{
"epoch": 1.5,
"learning_rate": 0.001,
"loss": 2.3667,
"step": 2290
},
{
"epoch": 1.5,
"learning_rate": 0.001,
"loss": 1.9434,
"step": 2295
},
{
"epoch": 1.51,
"learning_rate": 0.001,
"loss": 2.2934,
"step": 2300
},
{
"epoch": 1.51,
"learning_rate": 0.001,
"loss": 2.2615,
"step": 2305
},
{
"epoch": 1.51,
"learning_rate": 0.001,
"loss": 2.0636,
"step": 2310
},
{
"epoch": 1.52,
"learning_rate": 0.001,
"loss": 2.0307,
"step": 2315
},
{
"epoch": 1.52,
"learning_rate": 0.001,
"loss": 2.0033,
"step": 2320
},
{
"epoch": 1.52,
"learning_rate": 0.001,
"loss": 2.0915,
"step": 2325
},
{
"epoch": 1.52,
"learning_rate": 0.001,
"loss": 2.0934,
"step": 2330
},
{
"epoch": 1.53,
"learning_rate": 0.001,
"loss": 2.2811,
"step": 2335
},
{
"epoch": 1.53,
"learning_rate": 0.001,
"loss": 1.962,
"step": 2340
},
{
"epoch": 1.53,
"learning_rate": 0.001,
"loss": 1.8715,
"step": 2345
},
{
"epoch": 1.54,
"learning_rate": 0.001,
"loss": 2.2623,
"step": 2350
},
{
"epoch": 1.54,
"learning_rate": 0.001,
"loss": 2.1543,
"step": 2355
},
{
"epoch": 1.54,
"learning_rate": 0.001,
"loss": 2.1026,
"step": 2360
},
{
"epoch": 1.55,
"learning_rate": 0.001,
"loss": 2.4501,
"step": 2365
},
{
"epoch": 1.55,
"learning_rate": 0.001,
"loss": 2.1931,
"step": 2370
},
{
"epoch": 1.55,
"learning_rate": 0.001,
"loss": 2.0309,
"step": 2375
},
{
"epoch": 1.56,
"learning_rate": 0.001,
"loss": 2.0723,
"step": 2380
},
{
"epoch": 1.56,
"learning_rate": 0.001,
"loss": 2.3666,
"step": 2385
},
{
"epoch": 1.56,
"learning_rate": 0.001,
"loss": 2.1858,
"step": 2390
},
{
"epoch": 1.57,
"learning_rate": 0.001,
"loss": 2.031,
"step": 2395
},
{
"epoch": 1.57,
"learning_rate": 0.001,
"loss": 1.6752,
"step": 2400
},
{
"epoch": 1.57,
"learning_rate": 0.001,
"loss": 2.1003,
"step": 2405
},
{
"epoch": 1.58,
"learning_rate": 0.001,
"loss": 2.4007,
"step": 2410
},
{
"epoch": 1.58,
"learning_rate": 0.001,
"loss": 2.2498,
"step": 2415
},
{
"epoch": 1.58,
"learning_rate": 0.001,
"loss": 2.2472,
"step": 2420
},
{
"epoch": 1.59,
"learning_rate": 0.001,
"loss": 2.202,
"step": 2425
},
{
"epoch": 1.59,
"learning_rate": 0.001,
"loss": 2.4824,
"step": 2430
},
{
"epoch": 1.59,
"learning_rate": 0.001,
"loss": 2.1637,
"step": 2435
},
{
"epoch": 1.6,
"learning_rate": 0.001,
"loss": 2.4039,
"step": 2440
},
{
"epoch": 1.6,
"learning_rate": 0.001,
"loss": 2.1053,
"step": 2445
},
{
"epoch": 1.6,
"learning_rate": 0.001,
"loss": 1.9858,
"step": 2450
},
{
"epoch": 1.61,
"learning_rate": 0.001,
"loss": 2.2371,
"step": 2455
},
{
"epoch": 1.61,
"learning_rate": 0.001,
"loss": 1.8063,
"step": 2460
},
{
"epoch": 1.61,
"learning_rate": 0.001,
"loss": 2.2199,
"step": 2465
},
{
"epoch": 1.62,
"learning_rate": 0.001,
"loss": 2.2837,
"step": 2470
},
{
"epoch": 1.62,
"learning_rate": 0.001,
"loss": 2.2889,
"step": 2475
},
{
"epoch": 1.62,
"learning_rate": 0.001,
"loss": 2.1687,
"step": 2480
},
{
"epoch": 1.63,
"learning_rate": 0.001,
"loss": 2.2346,
"step": 2485
},
{
"epoch": 1.63,
"learning_rate": 0.001,
"loss": 2.3059,
"step": 2490
},
{
"epoch": 1.63,
"learning_rate": 0.001,
"loss": 2.3096,
"step": 2495
},
{
"epoch": 1.64,
"learning_rate": 0.001,
"loss": 2.4252,
"step": 2500
},
{
"epoch": 1.64,
"learning_rate": 0.001,
"loss": 1.9604,
"step": 2505
},
{
"epoch": 1.64,
"learning_rate": 0.001,
"loss": 2.154,
"step": 2510
},
{
"epoch": 1.65,
"learning_rate": 0.001,
"loss": 2.3648,
"step": 2515
},
{
"epoch": 1.65,
"learning_rate": 0.001,
"loss": 2.3436,
"step": 2520
},
{
"epoch": 1.65,
"learning_rate": 0.001,
"loss": 2.3471,
"step": 2525
},
{
"epoch": 1.66,
"learning_rate": 0.001,
"loss": 2.1766,
"step": 2530
},
{
"epoch": 1.66,
"learning_rate": 0.001,
"loss": 2.1967,
"step": 2535
},
{
"epoch": 1.66,
"learning_rate": 0.001,
"loss": 2.1945,
"step": 2540
},
{
"epoch": 1.67,
"learning_rate": 0.001,
"loss": 2.1473,
"step": 2545
},
{
"epoch": 1.67,
"learning_rate": 0.001,
"loss": 2.4883,
"step": 2550
},
{
"epoch": 1.67,
"learning_rate": 0.001,
"loss": 1.9596,
"step": 2555
},
{
"epoch": 1.68,
"learning_rate": 0.001,
"loss": 2.3908,
"step": 2560
},
{
"epoch": 1.68,
"learning_rate": 0.001,
"loss": 2.2533,
"step": 2565
},
{
"epoch": 1.68,
"learning_rate": 0.001,
"loss": 2.0808,
"step": 2570
},
{
"epoch": 1.69,
"learning_rate": 0.001,
"loss": 2.028,
"step": 2575
},
{
"epoch": 1.69,
"learning_rate": 0.001,
"loss": 2.2471,
"step": 2580
},
{
"epoch": 1.69,
"learning_rate": 0.001,
"loss": 2.1975,
"step": 2585
},
{
"epoch": 1.7,
"learning_rate": 0.001,
"loss": 2.404,
"step": 2590
},
{
"epoch": 1.7,
"learning_rate": 0.001,
"loss": 2.4238,
"step": 2595
},
{
"epoch": 1.7,
"learning_rate": 0.001,
"loss": 2.2742,
"step": 2600
},
{
"epoch": 1.7,
"learning_rate": 0.001,
"loss": 2.3284,
"step": 2605
},
{
"epoch": 1.71,
"learning_rate": 0.001,
"loss": 2.0517,
"step": 2610
},
{
"epoch": 1.71,
"learning_rate": 0.001,
"loss": 2.488,
"step": 2615
},
{
"epoch": 1.71,
"learning_rate": 0.001,
"loss": 2.1215,
"step": 2620
},
{
"epoch": 1.72,
"learning_rate": 0.001,
"loss": 2.1813,
"step": 2625
},
{
"epoch": 1.72,
"learning_rate": 0.001,
"loss": 1.9404,
"step": 2630
},
{
"epoch": 1.72,
"learning_rate": 0.001,
"loss": 2.1978,
"step": 2635
},
{
"epoch": 1.73,
"learning_rate": 0.001,
"loss": 2.1462,
"step": 2640
},
{
"epoch": 1.73,
"learning_rate": 0.001,
"loss": 1.9796,
"step": 2645
},
{
"epoch": 1.73,
"learning_rate": 0.001,
"loss": 2.267,
"step": 2650
},
{
"epoch": 1.74,
"learning_rate": 0.001,
"loss": 2.0169,
"step": 2655
},
{
"epoch": 1.74,
"learning_rate": 0.001,
"loss": 2.1779,
"step": 2660
},
{
"epoch": 1.74,
"learning_rate": 0.001,
"loss": 2.2851,
"step": 2665
},
{
"epoch": 1.75,
"learning_rate": 0.001,
"loss": 2.5845,
"step": 2670
},
{
"epoch": 1.75,
"learning_rate": 0.001,
"loss": 2.006,
"step": 2675
},
{
"epoch": 1.75,
"learning_rate": 0.001,
"loss": 2.0386,
"step": 2680
},
{
"epoch": 1.76,
"learning_rate": 0.001,
"loss": 2.2271,
"step": 2685
},
{
"epoch": 1.76,
"learning_rate": 0.001,
"loss": 2.1557,
"step": 2690
},
{
"epoch": 1.76,
"learning_rate": 0.001,
"loss": 2.0532,
"step": 2695
},
{
"epoch": 1.77,
"learning_rate": 0.001,
"loss": 2.14,
"step": 2700
},
{
"epoch": 1.77,
"learning_rate": 0.001,
"loss": 1.995,
"step": 2705
},
{
"epoch": 1.77,
"learning_rate": 0.001,
"loss": 1.9508,
"step": 2710
},
{
"epoch": 1.78,
"learning_rate": 0.001,
"loss": 2.4736,
"step": 2715
},
{
"epoch": 1.78,
"learning_rate": 0.001,
"loss": 2.0145,
"step": 2720
},
{
"epoch": 1.78,
"learning_rate": 0.001,
"loss": 2.1276,
"step": 2725
},
{
"epoch": 1.79,
"learning_rate": 0.001,
"loss": 2.1232,
"step": 2730
},
{
"epoch": 1.79,
"learning_rate": 0.001,
"loss": 2.3248,
"step": 2735
},
{
"epoch": 1.79,
"learning_rate": 0.001,
"loss": 2.0148,
"step": 2740
},
{
"epoch": 1.8,
"learning_rate": 0.001,
"loss": 1.974,
"step": 2745
},
{
"epoch": 1.8,
"learning_rate": 0.001,
"loss": 2.2313,
"step": 2750
},
{
"epoch": 1.8,
"learning_rate": 0.001,
"loss": 2.0893,
"step": 2755
},
{
"epoch": 1.81,
"learning_rate": 0.001,
"loss": 2.1423,
"step": 2760
},
{
"epoch": 1.81,
"learning_rate": 0.001,
"loss": 2.2085,
"step": 2765
},
{
"epoch": 1.81,
"learning_rate": 0.001,
"loss": 2.2314,
"step": 2770
},
{
"epoch": 1.82,
"learning_rate": 0.001,
"loss": 2.1122,
"step": 2775
},
{
"epoch": 1.82,
"learning_rate": 0.001,
"loss": 2.0739,
"step": 2780
},
{
"epoch": 1.82,
"learning_rate": 0.001,
"loss": 2.2233,
"step": 2785
},
{
"epoch": 1.83,
"learning_rate": 0.001,
"loss": 2.2696,
"step": 2790
},
{
"epoch": 1.83,
"learning_rate": 0.001,
"loss": 1.9375,
"step": 2795
},
{
"epoch": 1.83,
"learning_rate": 0.001,
"loss": 2.2126,
"step": 2800
},
{
"epoch": 1.84,
"learning_rate": 0.001,
"loss": 2.2643,
"step": 2805
},
{
"epoch": 1.84,
"learning_rate": 0.001,
"loss": 1.881,
"step": 2810
},
{
"epoch": 1.84,
"learning_rate": 0.001,
"loss": 2.427,
"step": 2815
},
{
"epoch": 1.85,
"learning_rate": 0.001,
"loss": 2.419,
"step": 2820
},
{
"epoch": 1.85,
"learning_rate": 0.001,
"loss": 2.2486,
"step": 2825
},
{
"epoch": 1.85,
"learning_rate": 0.001,
"loss": 2.6872,
"step": 2830
},
{
"epoch": 1.86,
"learning_rate": 0.001,
"loss": 2.0539,
"step": 2835
},
{
"epoch": 1.86,
"learning_rate": 0.001,
"loss": 2.4746,
"step": 2840
},
{
"epoch": 1.86,
"learning_rate": 0.001,
"loss": 2.2573,
"step": 2845
},
{
"epoch": 1.87,
"learning_rate": 0.001,
"loss": 2.0748,
"step": 2850
},
{
"epoch": 1.87,
"learning_rate": 0.001,
"loss": 2.2958,
"step": 2855
},
{
"epoch": 1.87,
"learning_rate": 0.001,
"loss": 1.8544,
"step": 2860
},
{
"epoch": 1.88,
"learning_rate": 0.001,
"loss": 2.4232,
"step": 2865
},
{
"epoch": 1.88,
"learning_rate": 0.001,
"loss": 2.3381,
"step": 2870
},
{
"epoch": 1.88,
"learning_rate": 0.001,
"loss": 2.276,
"step": 2875
},
{
"epoch": 1.88,
"learning_rate": 0.001,
"loss": 2.3278,
"step": 2880
},
{
"epoch": 1.89,
"learning_rate": 0.001,
"loss": 2.3119,
"step": 2885
},
{
"epoch": 1.89,
"learning_rate": 0.001,
"loss": 2.2123,
"step": 2890
},
{
"epoch": 1.89,
"learning_rate": 0.001,
"loss": 2.1596,
"step": 2895
},
{
"epoch": 1.9,
"learning_rate": 0.001,
"loss": 2.04,
"step": 2900
},
{
"epoch": 1.9,
"learning_rate": 0.001,
"loss": 2.3909,
"step": 2905
},
{
"epoch": 1.9,
"learning_rate": 0.001,
"loss": 2.1027,
"step": 2910
},
{
"epoch": 1.91,
"learning_rate": 0.001,
"loss": 2.1365,
"step": 2915
},
{
"epoch": 1.91,
"learning_rate": 0.001,
"loss": 2.2151,
"step": 2920
},
{
"epoch": 1.91,
"learning_rate": 0.001,
"loss": 2.2469,
"step": 2925
},
{
"epoch": 1.92,
"learning_rate": 0.001,
"loss": 2.1939,
"step": 2930
},
{
"epoch": 1.92,
"learning_rate": 0.001,
"loss": 2.0386,
"step": 2935
},
{
"epoch": 1.92,
"learning_rate": 0.001,
"loss": 2.094,
"step": 2940
},
{
"epoch": 1.93,
"learning_rate": 0.001,
"loss": 2.3225,
"step": 2945
},
{
"epoch": 1.93,
"learning_rate": 0.001,
"loss": 2.1019,
"step": 2950
},
{
"epoch": 1.93,
"learning_rate": 0.001,
"loss": 2.3213,
"step": 2955
},
{
"epoch": 1.94,
"learning_rate": 0.001,
"loss": 2.4173,
"step": 2960
},
{
"epoch": 1.94,
"learning_rate": 0.001,
"loss": 2.1919,
"step": 2965
},
{
"epoch": 1.94,
"learning_rate": 0.001,
"loss": 2.1733,
"step": 2970
},
{
"epoch": 1.95,
"learning_rate": 0.001,
"loss": 2.3576,
"step": 2975
},
{
"epoch": 1.95,
"learning_rate": 0.001,
"loss": 2.0042,
"step": 2980
},
{
"epoch": 1.95,
"learning_rate": 0.001,
"loss": 2.0977,
"step": 2985
},
{
"epoch": 1.96,
"learning_rate": 0.001,
"loss": 2.2674,
"step": 2990
},
{
"epoch": 1.96,
"learning_rate": 0.001,
"loss": 1.9316,
"step": 2995
},
{
"epoch": 1.96,
"learning_rate": 0.001,
"loss": 2.2472,
"step": 3000
}
],
"logging_steps": 5,
"max_steps": 7640,
"num_train_epochs": 5,
"save_steps": 1000,
"total_flos": 350355456000000.0,
"trial_name": null,
"trial_params": null
}