ccore's picture
Upload folder using huggingface_hub
e126367
raw
history blame contribute delete
No virus
126 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.9267015706806285,
"eval_steps": 500,
"global_step": 6000,
"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
},
{
"epoch": 1.97,
"learning_rate": 0.001,
"loss": 2.3264,
"step": 3005
},
{
"epoch": 1.97,
"learning_rate": 0.001,
"loss": 2.1732,
"step": 3010
},
{
"epoch": 1.97,
"learning_rate": 0.001,
"loss": 2.3932,
"step": 3015
},
{
"epoch": 1.98,
"learning_rate": 0.001,
"loss": 2.168,
"step": 3020
},
{
"epoch": 1.98,
"learning_rate": 0.001,
"loss": 2.2907,
"step": 3025
},
{
"epoch": 1.98,
"learning_rate": 0.001,
"loss": 2.1086,
"step": 3030
},
{
"epoch": 1.99,
"learning_rate": 0.001,
"loss": 2.2868,
"step": 3035
},
{
"epoch": 1.99,
"learning_rate": 0.001,
"loss": 1.9612,
"step": 3040
},
{
"epoch": 1.99,
"learning_rate": 0.001,
"loss": 2.1025,
"step": 3045
},
{
"epoch": 2.0,
"learning_rate": 0.001,
"loss": 2.1001,
"step": 3050
},
{
"epoch": 2.0,
"learning_rate": 0.001,
"loss": 2.0419,
"step": 3055
},
{
"epoch": 2.0,
"learning_rate": 0.001,
"loss": 2.1358,
"step": 3060
},
{
"epoch": 2.01,
"learning_rate": 0.001,
"loss": 1.762,
"step": 3065
},
{
"epoch": 2.01,
"learning_rate": 0.001,
"loss": 1.8736,
"step": 3070
},
{
"epoch": 2.01,
"learning_rate": 0.001,
"loss": 1.7017,
"step": 3075
},
{
"epoch": 2.02,
"learning_rate": 0.001,
"loss": 2.0051,
"step": 3080
},
{
"epoch": 2.02,
"learning_rate": 0.001,
"loss": 1.9062,
"step": 3085
},
{
"epoch": 2.02,
"learning_rate": 0.001,
"loss": 1.8619,
"step": 3090
},
{
"epoch": 2.03,
"learning_rate": 0.001,
"loss": 1.9188,
"step": 3095
},
{
"epoch": 2.03,
"learning_rate": 0.001,
"loss": 1.7303,
"step": 3100
},
{
"epoch": 2.03,
"learning_rate": 0.001,
"loss": 1.846,
"step": 3105
},
{
"epoch": 2.04,
"learning_rate": 0.001,
"loss": 1.8145,
"step": 3110
},
{
"epoch": 2.04,
"learning_rate": 0.001,
"loss": 1.8482,
"step": 3115
},
{
"epoch": 2.04,
"learning_rate": 0.001,
"loss": 1.5815,
"step": 3120
},
{
"epoch": 2.05,
"learning_rate": 0.001,
"loss": 1.5672,
"step": 3125
},
{
"epoch": 2.05,
"learning_rate": 0.001,
"loss": 1.7365,
"step": 3130
},
{
"epoch": 2.05,
"learning_rate": 0.001,
"loss": 1.971,
"step": 3135
},
{
"epoch": 2.05,
"learning_rate": 0.001,
"loss": 1.9756,
"step": 3140
},
{
"epoch": 2.06,
"learning_rate": 0.001,
"loss": 1.8266,
"step": 3145
},
{
"epoch": 2.06,
"learning_rate": 0.001,
"loss": 1.9481,
"step": 3150
},
{
"epoch": 2.06,
"learning_rate": 0.001,
"loss": 1.7729,
"step": 3155
},
{
"epoch": 2.07,
"learning_rate": 0.001,
"loss": 1.7883,
"step": 3160
},
{
"epoch": 2.07,
"learning_rate": 0.001,
"loss": 1.6513,
"step": 3165
},
{
"epoch": 2.07,
"learning_rate": 0.001,
"loss": 1.9425,
"step": 3170
},
{
"epoch": 2.08,
"learning_rate": 0.001,
"loss": 1.7869,
"step": 3175
},
{
"epoch": 2.08,
"learning_rate": 0.001,
"loss": 1.9321,
"step": 3180
},
{
"epoch": 2.08,
"learning_rate": 0.001,
"loss": 2.0544,
"step": 3185
},
{
"epoch": 2.09,
"learning_rate": 0.001,
"loss": 1.9293,
"step": 3190
},
{
"epoch": 2.09,
"learning_rate": 0.001,
"loss": 1.8088,
"step": 3195
},
{
"epoch": 2.09,
"learning_rate": 0.001,
"loss": 1.9035,
"step": 3200
},
{
"epoch": 2.1,
"learning_rate": 0.001,
"loss": 1.8003,
"step": 3205
},
{
"epoch": 2.1,
"learning_rate": 0.001,
"loss": 1.924,
"step": 3210
},
{
"epoch": 2.1,
"learning_rate": 0.001,
"loss": 1.5344,
"step": 3215
},
{
"epoch": 2.11,
"learning_rate": 0.001,
"loss": 1.7529,
"step": 3220
},
{
"epoch": 2.11,
"learning_rate": 0.001,
"loss": 1.9707,
"step": 3225
},
{
"epoch": 2.11,
"learning_rate": 0.001,
"loss": 2.1602,
"step": 3230
},
{
"epoch": 2.12,
"learning_rate": 0.001,
"loss": 2.0995,
"step": 3235
},
{
"epoch": 2.12,
"learning_rate": 0.001,
"loss": 1.7761,
"step": 3240
},
{
"epoch": 2.12,
"learning_rate": 0.001,
"loss": 1.7302,
"step": 3245
},
{
"epoch": 2.13,
"learning_rate": 0.001,
"loss": 2.0175,
"step": 3250
},
{
"epoch": 2.13,
"learning_rate": 0.001,
"loss": 1.9675,
"step": 3255
},
{
"epoch": 2.13,
"learning_rate": 0.001,
"loss": 1.7262,
"step": 3260
},
{
"epoch": 2.14,
"learning_rate": 0.001,
"loss": 1.8225,
"step": 3265
},
{
"epoch": 2.14,
"learning_rate": 0.001,
"loss": 1.9512,
"step": 3270
},
{
"epoch": 2.14,
"learning_rate": 0.001,
"loss": 1.8595,
"step": 3275
},
{
"epoch": 2.15,
"learning_rate": 0.001,
"loss": 1.8585,
"step": 3280
},
{
"epoch": 2.15,
"learning_rate": 0.001,
"loss": 1.8572,
"step": 3285
},
{
"epoch": 2.15,
"learning_rate": 0.001,
"loss": 1.9084,
"step": 3290
},
{
"epoch": 2.16,
"learning_rate": 0.001,
"loss": 1.9047,
"step": 3295
},
{
"epoch": 2.16,
"learning_rate": 0.001,
"loss": 2.0312,
"step": 3300
},
{
"epoch": 2.16,
"learning_rate": 0.001,
"loss": 1.7408,
"step": 3305
},
{
"epoch": 2.17,
"learning_rate": 0.001,
"loss": 1.9193,
"step": 3310
},
{
"epoch": 2.17,
"learning_rate": 0.001,
"loss": 1.8676,
"step": 3315
},
{
"epoch": 2.17,
"learning_rate": 0.001,
"loss": 1.9841,
"step": 3320
},
{
"epoch": 2.18,
"learning_rate": 0.001,
"loss": 2.0652,
"step": 3325
},
{
"epoch": 2.18,
"learning_rate": 0.001,
"loss": 1.7531,
"step": 3330
},
{
"epoch": 2.18,
"learning_rate": 0.001,
"loss": 1.8534,
"step": 3335
},
{
"epoch": 2.19,
"learning_rate": 0.001,
"loss": 1.9164,
"step": 3340
},
{
"epoch": 2.19,
"learning_rate": 0.001,
"loss": 2.1042,
"step": 3345
},
{
"epoch": 2.19,
"learning_rate": 0.001,
"loss": 1.7698,
"step": 3350
},
{
"epoch": 2.2,
"learning_rate": 0.001,
"loss": 1.9511,
"step": 3355
},
{
"epoch": 2.2,
"learning_rate": 0.001,
"loss": 1.7882,
"step": 3360
},
{
"epoch": 2.2,
"learning_rate": 0.001,
"loss": 2.0131,
"step": 3365
},
{
"epoch": 2.21,
"learning_rate": 0.001,
"loss": 1.8166,
"step": 3370
},
{
"epoch": 2.21,
"learning_rate": 0.001,
"loss": 1.7719,
"step": 3375
},
{
"epoch": 2.21,
"learning_rate": 0.001,
"loss": 1.7812,
"step": 3380
},
{
"epoch": 2.22,
"learning_rate": 0.001,
"loss": 1.8818,
"step": 3385
},
{
"epoch": 2.22,
"learning_rate": 0.001,
"loss": 1.8874,
"step": 3390
},
{
"epoch": 2.22,
"learning_rate": 0.001,
"loss": 1.8422,
"step": 3395
},
{
"epoch": 2.23,
"learning_rate": 0.001,
"loss": 2.1017,
"step": 3400
},
{
"epoch": 2.23,
"learning_rate": 0.001,
"loss": 1.8183,
"step": 3405
},
{
"epoch": 2.23,
"learning_rate": 0.001,
"loss": 1.9675,
"step": 3410
},
{
"epoch": 2.23,
"learning_rate": 0.001,
"loss": 1.7103,
"step": 3415
},
{
"epoch": 2.24,
"learning_rate": 0.001,
"loss": 1.9424,
"step": 3420
},
{
"epoch": 2.24,
"learning_rate": 0.001,
"loss": 1.6799,
"step": 3425
},
{
"epoch": 2.24,
"learning_rate": 0.001,
"loss": 1.6281,
"step": 3430
},
{
"epoch": 2.25,
"learning_rate": 0.001,
"loss": 1.9875,
"step": 3435
},
{
"epoch": 2.25,
"learning_rate": 0.001,
"loss": 1.6356,
"step": 3440
},
{
"epoch": 2.25,
"learning_rate": 0.001,
"loss": 1.8295,
"step": 3445
},
{
"epoch": 2.26,
"learning_rate": 0.001,
"loss": 2.1965,
"step": 3450
},
{
"epoch": 2.26,
"learning_rate": 0.001,
"loss": 1.8972,
"step": 3455
},
{
"epoch": 2.26,
"learning_rate": 0.001,
"loss": 2.1296,
"step": 3460
},
{
"epoch": 2.27,
"learning_rate": 0.001,
"loss": 1.987,
"step": 3465
},
{
"epoch": 2.27,
"learning_rate": 0.001,
"loss": 1.7851,
"step": 3470
},
{
"epoch": 2.27,
"learning_rate": 0.001,
"loss": 1.758,
"step": 3475
},
{
"epoch": 2.28,
"learning_rate": 0.001,
"loss": 1.752,
"step": 3480
},
{
"epoch": 2.28,
"learning_rate": 0.001,
"loss": 1.9093,
"step": 3485
},
{
"epoch": 2.28,
"learning_rate": 0.001,
"loss": 1.9931,
"step": 3490
},
{
"epoch": 2.29,
"learning_rate": 0.001,
"loss": 1.8762,
"step": 3495
},
{
"epoch": 2.29,
"learning_rate": 0.001,
"loss": 1.9769,
"step": 3500
},
{
"epoch": 2.29,
"learning_rate": 0.001,
"loss": 2.0007,
"step": 3505
},
{
"epoch": 2.3,
"learning_rate": 0.001,
"loss": 2.2291,
"step": 3510
},
{
"epoch": 2.3,
"learning_rate": 0.001,
"loss": 2.0776,
"step": 3515
},
{
"epoch": 2.3,
"learning_rate": 0.001,
"loss": 2.2868,
"step": 3520
},
{
"epoch": 2.31,
"learning_rate": 0.001,
"loss": 2.0916,
"step": 3525
},
{
"epoch": 2.31,
"learning_rate": 0.001,
"loss": 2.0035,
"step": 3530
},
{
"epoch": 2.31,
"learning_rate": 0.001,
"loss": 1.9706,
"step": 3535
},
{
"epoch": 2.32,
"learning_rate": 0.001,
"loss": 1.6881,
"step": 3540
},
{
"epoch": 2.32,
"learning_rate": 0.001,
"loss": 2.0781,
"step": 3545
},
{
"epoch": 2.32,
"learning_rate": 0.001,
"loss": 1.9761,
"step": 3550
},
{
"epoch": 2.33,
"learning_rate": 0.001,
"loss": 1.9865,
"step": 3555
},
{
"epoch": 2.33,
"learning_rate": 0.001,
"loss": 1.7636,
"step": 3560
},
{
"epoch": 2.33,
"learning_rate": 0.001,
"loss": 2.1409,
"step": 3565
},
{
"epoch": 2.34,
"learning_rate": 0.001,
"loss": 2.11,
"step": 3570
},
{
"epoch": 2.34,
"learning_rate": 0.001,
"loss": 1.7359,
"step": 3575
},
{
"epoch": 2.34,
"learning_rate": 0.001,
"loss": 1.9619,
"step": 3580
},
{
"epoch": 2.35,
"learning_rate": 0.001,
"loss": 2.0434,
"step": 3585
},
{
"epoch": 2.35,
"learning_rate": 0.001,
"loss": 1.8588,
"step": 3590
},
{
"epoch": 2.35,
"learning_rate": 0.001,
"loss": 2.0681,
"step": 3595
},
{
"epoch": 2.36,
"learning_rate": 0.001,
"loss": 1.8467,
"step": 3600
},
{
"epoch": 2.36,
"learning_rate": 0.001,
"loss": 1.5267,
"step": 3605
},
{
"epoch": 2.36,
"learning_rate": 0.001,
"loss": 1.9642,
"step": 3610
},
{
"epoch": 2.37,
"learning_rate": 0.001,
"loss": 1.9597,
"step": 3615
},
{
"epoch": 2.37,
"learning_rate": 0.001,
"loss": 1.8687,
"step": 3620
},
{
"epoch": 2.37,
"learning_rate": 0.001,
"loss": 2.0584,
"step": 3625
},
{
"epoch": 2.38,
"learning_rate": 0.001,
"loss": 2.0966,
"step": 3630
},
{
"epoch": 2.38,
"learning_rate": 0.001,
"loss": 2.0021,
"step": 3635
},
{
"epoch": 2.38,
"learning_rate": 0.001,
"loss": 1.8767,
"step": 3640
},
{
"epoch": 2.39,
"learning_rate": 0.001,
"loss": 1.713,
"step": 3645
},
{
"epoch": 2.39,
"learning_rate": 0.001,
"loss": 1.5825,
"step": 3650
},
{
"epoch": 2.39,
"learning_rate": 0.001,
"loss": 1.8617,
"step": 3655
},
{
"epoch": 2.4,
"learning_rate": 0.001,
"loss": 2.0615,
"step": 3660
},
{
"epoch": 2.4,
"learning_rate": 0.001,
"loss": 1.6553,
"step": 3665
},
{
"epoch": 2.4,
"learning_rate": 0.001,
"loss": 1.6166,
"step": 3670
},
{
"epoch": 2.41,
"learning_rate": 0.001,
"loss": 1.6974,
"step": 3675
},
{
"epoch": 2.41,
"learning_rate": 0.001,
"loss": 2.0093,
"step": 3680
},
{
"epoch": 2.41,
"learning_rate": 0.001,
"loss": 1.8497,
"step": 3685
},
{
"epoch": 2.41,
"learning_rate": 0.001,
"loss": 1.7226,
"step": 3690
},
{
"epoch": 2.42,
"learning_rate": 0.001,
"loss": 1.668,
"step": 3695
},
{
"epoch": 2.42,
"learning_rate": 0.001,
"loss": 2.115,
"step": 3700
},
{
"epoch": 2.42,
"learning_rate": 0.001,
"loss": 2.145,
"step": 3705
},
{
"epoch": 2.43,
"learning_rate": 0.001,
"loss": 1.7443,
"step": 3710
},
{
"epoch": 2.43,
"learning_rate": 0.001,
"loss": 1.7406,
"step": 3715
},
{
"epoch": 2.43,
"learning_rate": 0.001,
"loss": 1.8569,
"step": 3720
},
{
"epoch": 2.44,
"learning_rate": 0.001,
"loss": 2.0726,
"step": 3725
},
{
"epoch": 2.44,
"learning_rate": 0.001,
"loss": 2.1467,
"step": 3730
},
{
"epoch": 2.44,
"learning_rate": 0.001,
"loss": 2.019,
"step": 3735
},
{
"epoch": 2.45,
"learning_rate": 0.001,
"loss": 2.1104,
"step": 3740
},
{
"epoch": 2.45,
"learning_rate": 0.001,
"loss": 2.1444,
"step": 3745
},
{
"epoch": 2.45,
"learning_rate": 0.001,
"loss": 2.0757,
"step": 3750
},
{
"epoch": 2.46,
"learning_rate": 0.001,
"loss": 1.7169,
"step": 3755
},
{
"epoch": 2.46,
"learning_rate": 0.001,
"loss": 1.7558,
"step": 3760
},
{
"epoch": 2.46,
"learning_rate": 0.001,
"loss": 1.6213,
"step": 3765
},
{
"epoch": 2.47,
"learning_rate": 0.001,
"loss": 1.8306,
"step": 3770
},
{
"epoch": 2.47,
"learning_rate": 0.001,
"loss": 2.015,
"step": 3775
},
{
"epoch": 2.47,
"learning_rate": 0.001,
"loss": 2.0952,
"step": 3780
},
{
"epoch": 2.48,
"learning_rate": 0.001,
"loss": 1.6549,
"step": 3785
},
{
"epoch": 2.48,
"learning_rate": 0.001,
"loss": 1.8965,
"step": 3790
},
{
"epoch": 2.48,
"learning_rate": 0.001,
"loss": 2.0753,
"step": 3795
},
{
"epoch": 2.49,
"learning_rate": 0.001,
"loss": 2.135,
"step": 3800
},
{
"epoch": 2.49,
"learning_rate": 0.001,
"loss": 1.8428,
"step": 3805
},
{
"epoch": 2.49,
"learning_rate": 0.001,
"loss": 2.1602,
"step": 3810
},
{
"epoch": 2.5,
"learning_rate": 0.001,
"loss": 1.8219,
"step": 3815
},
{
"epoch": 2.5,
"learning_rate": 0.001,
"loss": 1.8266,
"step": 3820
},
{
"epoch": 2.5,
"learning_rate": 0.001,
"loss": 1.839,
"step": 3825
},
{
"epoch": 2.51,
"learning_rate": 0.001,
"loss": 1.8746,
"step": 3830
},
{
"epoch": 2.51,
"learning_rate": 0.001,
"loss": 2.1668,
"step": 3835
},
{
"epoch": 2.51,
"learning_rate": 0.001,
"loss": 2.1224,
"step": 3840
},
{
"epoch": 2.52,
"learning_rate": 0.001,
"loss": 2.0099,
"step": 3845
},
{
"epoch": 2.52,
"learning_rate": 0.001,
"loss": 2.2438,
"step": 3850
},
{
"epoch": 2.52,
"learning_rate": 0.001,
"loss": 1.8494,
"step": 3855
},
{
"epoch": 2.53,
"learning_rate": 0.001,
"loss": 1.8063,
"step": 3860
},
{
"epoch": 2.53,
"learning_rate": 0.001,
"loss": 1.7068,
"step": 3865
},
{
"epoch": 2.53,
"learning_rate": 0.001,
"loss": 1.8158,
"step": 3870
},
{
"epoch": 2.54,
"learning_rate": 0.001,
"loss": 1.6616,
"step": 3875
},
{
"epoch": 2.54,
"learning_rate": 0.001,
"loss": 1.9542,
"step": 3880
},
{
"epoch": 2.54,
"learning_rate": 0.001,
"loss": 1.9595,
"step": 3885
},
{
"epoch": 2.55,
"learning_rate": 0.001,
"loss": 2.1507,
"step": 3890
},
{
"epoch": 2.55,
"learning_rate": 0.001,
"loss": 1.883,
"step": 3895
},
{
"epoch": 2.55,
"learning_rate": 0.001,
"loss": 1.6634,
"step": 3900
},
{
"epoch": 2.56,
"learning_rate": 0.001,
"loss": 1.7621,
"step": 3905
},
{
"epoch": 2.56,
"learning_rate": 0.001,
"loss": 1.9474,
"step": 3910
},
{
"epoch": 2.56,
"learning_rate": 0.001,
"loss": 2.094,
"step": 3915
},
{
"epoch": 2.57,
"learning_rate": 0.001,
"loss": 2.0111,
"step": 3920
},
{
"epoch": 2.57,
"learning_rate": 0.001,
"loss": 1.872,
"step": 3925
},
{
"epoch": 2.57,
"learning_rate": 0.001,
"loss": 1.604,
"step": 3930
},
{
"epoch": 2.58,
"learning_rate": 0.001,
"loss": 1.8784,
"step": 3935
},
{
"epoch": 2.58,
"learning_rate": 0.001,
"loss": 1.9175,
"step": 3940
},
{
"epoch": 2.58,
"learning_rate": 0.001,
"loss": 1.8607,
"step": 3945
},
{
"epoch": 2.59,
"learning_rate": 0.001,
"loss": 1.9768,
"step": 3950
},
{
"epoch": 2.59,
"learning_rate": 0.001,
"loss": 1.9017,
"step": 3955
},
{
"epoch": 2.59,
"learning_rate": 0.001,
"loss": 1.6314,
"step": 3960
},
{
"epoch": 2.59,
"learning_rate": 0.001,
"loss": 1.8859,
"step": 3965
},
{
"epoch": 2.6,
"learning_rate": 0.001,
"loss": 2.0082,
"step": 3970
},
{
"epoch": 2.6,
"learning_rate": 0.001,
"loss": 1.7938,
"step": 3975
},
{
"epoch": 2.6,
"learning_rate": 0.001,
"loss": 2.0765,
"step": 3980
},
{
"epoch": 2.61,
"learning_rate": 0.001,
"loss": 2.0694,
"step": 3985
},
{
"epoch": 2.61,
"learning_rate": 0.001,
"loss": 1.8571,
"step": 3990
},
{
"epoch": 2.61,
"learning_rate": 0.001,
"loss": 1.6691,
"step": 3995
},
{
"epoch": 2.62,
"learning_rate": 0.001,
"loss": 2.1105,
"step": 4000
},
{
"epoch": 2.62,
"learning_rate": 0.001,
"loss": 1.9738,
"step": 4005
},
{
"epoch": 2.62,
"learning_rate": 0.001,
"loss": 1.8698,
"step": 4010
},
{
"epoch": 2.63,
"learning_rate": 0.001,
"loss": 2.0725,
"step": 4015
},
{
"epoch": 2.63,
"learning_rate": 0.001,
"loss": 1.8761,
"step": 4020
},
{
"epoch": 2.63,
"learning_rate": 0.001,
"loss": 2.0235,
"step": 4025
},
{
"epoch": 2.64,
"learning_rate": 0.001,
"loss": 2.2568,
"step": 4030
},
{
"epoch": 2.64,
"learning_rate": 0.001,
"loss": 1.901,
"step": 4035
},
{
"epoch": 2.64,
"learning_rate": 0.001,
"loss": 1.7577,
"step": 4040
},
{
"epoch": 2.65,
"learning_rate": 0.001,
"loss": 1.9142,
"step": 4045
},
{
"epoch": 2.65,
"learning_rate": 0.001,
"loss": 2.134,
"step": 4050
},
{
"epoch": 2.65,
"learning_rate": 0.001,
"loss": 2.0052,
"step": 4055
},
{
"epoch": 2.66,
"learning_rate": 0.001,
"loss": 2.0817,
"step": 4060
},
{
"epoch": 2.66,
"learning_rate": 0.001,
"loss": 2.1456,
"step": 4065
},
{
"epoch": 2.66,
"learning_rate": 0.001,
"loss": 2.0291,
"step": 4070
},
{
"epoch": 2.67,
"learning_rate": 0.001,
"loss": 1.9683,
"step": 4075
},
{
"epoch": 2.67,
"learning_rate": 0.001,
"loss": 1.6709,
"step": 4080
},
{
"epoch": 2.67,
"learning_rate": 0.001,
"loss": 2.0227,
"step": 4085
},
{
"epoch": 2.68,
"learning_rate": 0.001,
"loss": 2.0289,
"step": 4090
},
{
"epoch": 2.68,
"learning_rate": 0.001,
"loss": 1.8361,
"step": 4095
},
{
"epoch": 2.68,
"learning_rate": 0.001,
"loss": 1.8289,
"step": 4100
},
{
"epoch": 2.69,
"learning_rate": 0.001,
"loss": 2.0069,
"step": 4105
},
{
"epoch": 2.69,
"learning_rate": 0.001,
"loss": 2.1599,
"step": 4110
},
{
"epoch": 2.69,
"learning_rate": 0.001,
"loss": 2.1124,
"step": 4115
},
{
"epoch": 2.7,
"learning_rate": 0.001,
"loss": 1.8739,
"step": 4120
},
{
"epoch": 2.7,
"learning_rate": 0.001,
"loss": 1.7629,
"step": 4125
},
{
"epoch": 2.7,
"learning_rate": 0.001,
"loss": 1.9235,
"step": 4130
},
{
"epoch": 2.71,
"learning_rate": 0.001,
"loss": 2.0229,
"step": 4135
},
{
"epoch": 2.71,
"learning_rate": 0.001,
"loss": 1.9561,
"step": 4140
},
{
"epoch": 2.71,
"learning_rate": 0.001,
"loss": 1.6533,
"step": 4145
},
{
"epoch": 2.72,
"learning_rate": 0.001,
"loss": 2.1504,
"step": 4150
},
{
"epoch": 2.72,
"learning_rate": 0.001,
"loss": 2.061,
"step": 4155
},
{
"epoch": 2.72,
"learning_rate": 0.001,
"loss": 1.9832,
"step": 4160
},
{
"epoch": 2.73,
"learning_rate": 0.001,
"loss": 1.9235,
"step": 4165
},
{
"epoch": 2.73,
"learning_rate": 0.001,
"loss": 2.0711,
"step": 4170
},
{
"epoch": 2.73,
"learning_rate": 0.001,
"loss": 1.8579,
"step": 4175
},
{
"epoch": 2.74,
"learning_rate": 0.001,
"loss": 2.2342,
"step": 4180
},
{
"epoch": 2.74,
"learning_rate": 0.001,
"loss": 2.2196,
"step": 4185
},
{
"epoch": 2.74,
"learning_rate": 0.001,
"loss": 1.7639,
"step": 4190
},
{
"epoch": 2.75,
"learning_rate": 0.001,
"loss": 2.1289,
"step": 4195
},
{
"epoch": 2.75,
"learning_rate": 0.001,
"loss": 2.0845,
"step": 4200
},
{
"epoch": 2.75,
"learning_rate": 0.001,
"loss": 2.1311,
"step": 4205
},
{
"epoch": 2.76,
"learning_rate": 0.001,
"loss": 1.5461,
"step": 4210
},
{
"epoch": 2.76,
"learning_rate": 0.001,
"loss": 2.0942,
"step": 4215
},
{
"epoch": 2.76,
"learning_rate": 0.001,
"loss": 2.0854,
"step": 4220
},
{
"epoch": 2.77,
"learning_rate": 0.001,
"loss": 2.0704,
"step": 4225
},
{
"epoch": 2.77,
"learning_rate": 0.001,
"loss": 1.9709,
"step": 4230
},
{
"epoch": 2.77,
"learning_rate": 0.001,
"loss": 1.82,
"step": 4235
},
{
"epoch": 2.77,
"learning_rate": 0.001,
"loss": 2.0901,
"step": 4240
},
{
"epoch": 2.78,
"learning_rate": 0.001,
"loss": 2.0107,
"step": 4245
},
{
"epoch": 2.78,
"learning_rate": 0.001,
"loss": 2.0632,
"step": 4250
},
{
"epoch": 2.78,
"learning_rate": 0.001,
"loss": 1.9023,
"step": 4255
},
{
"epoch": 2.79,
"learning_rate": 0.001,
"loss": 2.0827,
"step": 4260
},
{
"epoch": 2.79,
"learning_rate": 0.001,
"loss": 2.0453,
"step": 4265
},
{
"epoch": 2.79,
"learning_rate": 0.001,
"loss": 2.0852,
"step": 4270
},
{
"epoch": 2.8,
"learning_rate": 0.001,
"loss": 2.2534,
"step": 4275
},
{
"epoch": 2.8,
"learning_rate": 0.001,
"loss": 1.8449,
"step": 4280
},
{
"epoch": 2.8,
"learning_rate": 0.001,
"loss": 1.8617,
"step": 4285
},
{
"epoch": 2.81,
"learning_rate": 0.001,
"loss": 2.1616,
"step": 4290
},
{
"epoch": 2.81,
"learning_rate": 0.001,
"loss": 1.99,
"step": 4295
},
{
"epoch": 2.81,
"learning_rate": 0.001,
"loss": 1.9886,
"step": 4300
},
{
"epoch": 2.82,
"learning_rate": 0.001,
"loss": 2.067,
"step": 4305
},
{
"epoch": 2.82,
"learning_rate": 0.001,
"loss": 1.8923,
"step": 4310
},
{
"epoch": 2.82,
"learning_rate": 0.001,
"loss": 2.2929,
"step": 4315
},
{
"epoch": 2.83,
"learning_rate": 0.001,
"loss": 2.2061,
"step": 4320
},
{
"epoch": 2.83,
"learning_rate": 0.001,
"loss": 2.3522,
"step": 4325
},
{
"epoch": 2.83,
"learning_rate": 0.001,
"loss": 1.9623,
"step": 4330
},
{
"epoch": 2.84,
"learning_rate": 0.001,
"loss": 1.7255,
"step": 4335
},
{
"epoch": 2.84,
"learning_rate": 0.001,
"loss": 2.2199,
"step": 4340
},
{
"epoch": 2.84,
"learning_rate": 0.001,
"loss": 1.9392,
"step": 4345
},
{
"epoch": 2.85,
"learning_rate": 0.001,
"loss": 2.0237,
"step": 4350
},
{
"epoch": 2.85,
"learning_rate": 0.001,
"loss": 1.9724,
"step": 4355
},
{
"epoch": 2.85,
"learning_rate": 0.001,
"loss": 2.1273,
"step": 4360
},
{
"epoch": 2.86,
"learning_rate": 0.001,
"loss": 2.0275,
"step": 4365
},
{
"epoch": 2.86,
"learning_rate": 0.001,
"loss": 2.1322,
"step": 4370
},
{
"epoch": 2.86,
"learning_rate": 0.001,
"loss": 2.1448,
"step": 4375
},
{
"epoch": 2.87,
"learning_rate": 0.001,
"loss": 2.1961,
"step": 4380
},
{
"epoch": 2.87,
"learning_rate": 0.001,
"loss": 1.9851,
"step": 4385
},
{
"epoch": 2.87,
"learning_rate": 0.001,
"loss": 2.2935,
"step": 4390
},
{
"epoch": 2.88,
"learning_rate": 0.001,
"loss": 1.9986,
"step": 4395
},
{
"epoch": 2.88,
"learning_rate": 0.001,
"loss": 2.1898,
"step": 4400
},
{
"epoch": 2.88,
"learning_rate": 0.001,
"loss": 1.9743,
"step": 4405
},
{
"epoch": 2.89,
"learning_rate": 0.001,
"loss": 1.9389,
"step": 4410
},
{
"epoch": 2.89,
"learning_rate": 0.001,
"loss": 1.9267,
"step": 4415
},
{
"epoch": 2.89,
"learning_rate": 0.001,
"loss": 2.0713,
"step": 4420
},
{
"epoch": 2.9,
"learning_rate": 0.001,
"loss": 2.2993,
"step": 4425
},
{
"epoch": 2.9,
"learning_rate": 0.001,
"loss": 1.6093,
"step": 4430
},
{
"epoch": 2.9,
"learning_rate": 0.001,
"loss": 2.2791,
"step": 4435
},
{
"epoch": 2.91,
"learning_rate": 0.001,
"loss": 1.8745,
"step": 4440
},
{
"epoch": 2.91,
"learning_rate": 0.001,
"loss": 2.1259,
"step": 4445
},
{
"epoch": 2.91,
"learning_rate": 0.001,
"loss": 2.1719,
"step": 4450
},
{
"epoch": 2.92,
"learning_rate": 0.001,
"loss": 1.9552,
"step": 4455
},
{
"epoch": 2.92,
"learning_rate": 0.001,
"loss": 1.7201,
"step": 4460
},
{
"epoch": 2.92,
"learning_rate": 0.001,
"loss": 1.6175,
"step": 4465
},
{
"epoch": 2.93,
"learning_rate": 0.001,
"loss": 2.0599,
"step": 4470
},
{
"epoch": 2.93,
"learning_rate": 0.001,
"loss": 2.1124,
"step": 4475
},
{
"epoch": 2.93,
"learning_rate": 0.001,
"loss": 2.2331,
"step": 4480
},
{
"epoch": 2.94,
"learning_rate": 0.001,
"loss": 2.1158,
"step": 4485
},
{
"epoch": 2.94,
"learning_rate": 0.001,
"loss": 2.1609,
"step": 4490
},
{
"epoch": 2.94,
"learning_rate": 0.001,
"loss": 2.0898,
"step": 4495
},
{
"epoch": 2.95,
"learning_rate": 0.001,
"loss": 1.7871,
"step": 4500
},
{
"epoch": 2.95,
"learning_rate": 0.001,
"loss": 1.9766,
"step": 4505
},
{
"epoch": 2.95,
"learning_rate": 0.001,
"loss": 1.9859,
"step": 4510
},
{
"epoch": 2.95,
"learning_rate": 0.001,
"loss": 1.8897,
"step": 4515
},
{
"epoch": 2.96,
"learning_rate": 0.001,
"loss": 2.1179,
"step": 4520
},
{
"epoch": 2.96,
"learning_rate": 0.001,
"loss": 1.8307,
"step": 4525
},
{
"epoch": 2.96,
"learning_rate": 0.001,
"loss": 2.0669,
"step": 4530
},
{
"epoch": 2.97,
"learning_rate": 0.001,
"loss": 2.2398,
"step": 4535
},
{
"epoch": 2.97,
"learning_rate": 0.001,
"loss": 2.1574,
"step": 4540
},
{
"epoch": 2.97,
"learning_rate": 0.001,
"loss": 2.0868,
"step": 4545
},
{
"epoch": 2.98,
"learning_rate": 0.001,
"loss": 2.0039,
"step": 4550
},
{
"epoch": 2.98,
"learning_rate": 0.001,
"loss": 2.0899,
"step": 4555
},
{
"epoch": 2.98,
"learning_rate": 0.001,
"loss": 2.0335,
"step": 4560
},
{
"epoch": 2.99,
"learning_rate": 0.001,
"loss": 1.8509,
"step": 4565
},
{
"epoch": 2.99,
"learning_rate": 0.001,
"loss": 2.2191,
"step": 4570
},
{
"epoch": 2.99,
"learning_rate": 0.001,
"loss": 2.158,
"step": 4575
},
{
"epoch": 3.0,
"learning_rate": 0.001,
"loss": 1.8958,
"step": 4580
},
{
"epoch": 3.0,
"learning_rate": 0.001,
"loss": 1.9311,
"step": 4585
},
{
"epoch": 3.0,
"learning_rate": 0.001,
"loss": 1.6468,
"step": 4590
},
{
"epoch": 3.01,
"learning_rate": 0.001,
"loss": 1.4211,
"step": 4595
},
{
"epoch": 3.01,
"learning_rate": 0.001,
"loss": 1.5942,
"step": 4600
},
{
"epoch": 3.01,
"learning_rate": 0.001,
"loss": 1.618,
"step": 4605
},
{
"epoch": 3.02,
"learning_rate": 0.001,
"loss": 1.4268,
"step": 4610
},
{
"epoch": 3.02,
"learning_rate": 0.001,
"loss": 1.8181,
"step": 4615
},
{
"epoch": 3.02,
"learning_rate": 0.001,
"loss": 1.7277,
"step": 4620
},
{
"epoch": 3.03,
"learning_rate": 0.001,
"loss": 1.6623,
"step": 4625
},
{
"epoch": 3.03,
"learning_rate": 0.001,
"loss": 1.7616,
"step": 4630
},
{
"epoch": 3.03,
"learning_rate": 0.001,
"loss": 1.4593,
"step": 4635
},
{
"epoch": 3.04,
"learning_rate": 0.001,
"loss": 1.555,
"step": 4640
},
{
"epoch": 3.04,
"learning_rate": 0.001,
"loss": 1.4423,
"step": 4645
},
{
"epoch": 3.04,
"learning_rate": 0.001,
"loss": 1.7845,
"step": 4650
},
{
"epoch": 3.05,
"learning_rate": 0.001,
"loss": 1.6926,
"step": 4655
},
{
"epoch": 3.05,
"learning_rate": 0.001,
"loss": 1.5921,
"step": 4660
},
{
"epoch": 3.05,
"learning_rate": 0.001,
"loss": 1.4306,
"step": 4665
},
{
"epoch": 3.06,
"learning_rate": 0.001,
"loss": 1.6041,
"step": 4670
},
{
"epoch": 3.06,
"learning_rate": 0.001,
"loss": 1.9242,
"step": 4675
},
{
"epoch": 3.06,
"learning_rate": 0.001,
"loss": 1.7631,
"step": 4680
},
{
"epoch": 3.07,
"learning_rate": 0.001,
"loss": 1.8995,
"step": 4685
},
{
"epoch": 3.07,
"learning_rate": 0.001,
"loss": 1.5928,
"step": 4690
},
{
"epoch": 3.07,
"learning_rate": 0.001,
"loss": 1.6116,
"step": 4695
},
{
"epoch": 3.08,
"learning_rate": 0.001,
"loss": 2.1657,
"step": 4700
},
{
"epoch": 3.08,
"learning_rate": 0.001,
"loss": 1.7712,
"step": 4705
},
{
"epoch": 3.08,
"learning_rate": 0.001,
"loss": 1.8519,
"step": 4710
},
{
"epoch": 3.09,
"learning_rate": 0.001,
"loss": 1.533,
"step": 4715
},
{
"epoch": 3.09,
"learning_rate": 0.001,
"loss": 1.7871,
"step": 4720
},
{
"epoch": 3.09,
"learning_rate": 0.001,
"loss": 1.7858,
"step": 4725
},
{
"epoch": 3.1,
"learning_rate": 0.001,
"loss": 1.6107,
"step": 4730
},
{
"epoch": 3.1,
"learning_rate": 0.001,
"loss": 1.782,
"step": 4735
},
{
"epoch": 3.1,
"learning_rate": 0.001,
"loss": 1.5549,
"step": 4740
},
{
"epoch": 3.11,
"learning_rate": 0.001,
"loss": 1.6906,
"step": 4745
},
{
"epoch": 3.11,
"learning_rate": 0.001,
"loss": 1.8096,
"step": 4750
},
{
"epoch": 3.11,
"learning_rate": 0.001,
"loss": 1.8374,
"step": 4755
},
{
"epoch": 3.12,
"learning_rate": 0.001,
"loss": 1.8821,
"step": 4760
},
{
"epoch": 3.12,
"learning_rate": 0.001,
"loss": 1.7064,
"step": 4765
},
{
"epoch": 3.12,
"learning_rate": 0.001,
"loss": 1.9045,
"step": 4770
},
{
"epoch": 3.12,
"learning_rate": 0.001,
"loss": 1.5462,
"step": 4775
},
{
"epoch": 3.13,
"learning_rate": 0.001,
"loss": 1.857,
"step": 4780
},
{
"epoch": 3.13,
"learning_rate": 0.001,
"loss": 1.7946,
"step": 4785
},
{
"epoch": 3.13,
"learning_rate": 0.001,
"loss": 1.5898,
"step": 4790
},
{
"epoch": 3.14,
"learning_rate": 0.001,
"loss": 1.7049,
"step": 4795
},
{
"epoch": 3.14,
"learning_rate": 0.001,
"loss": 1.9023,
"step": 4800
},
{
"epoch": 3.14,
"learning_rate": 0.001,
"loss": 1.6363,
"step": 4805
},
{
"epoch": 3.15,
"learning_rate": 0.001,
"loss": 1.6567,
"step": 4810
},
{
"epoch": 3.15,
"learning_rate": 0.001,
"loss": 1.978,
"step": 4815
},
{
"epoch": 3.15,
"learning_rate": 0.001,
"loss": 1.6774,
"step": 4820
},
{
"epoch": 3.16,
"learning_rate": 0.001,
"loss": 1.832,
"step": 4825
},
{
"epoch": 3.16,
"learning_rate": 0.001,
"loss": 1.8501,
"step": 4830
},
{
"epoch": 3.16,
"learning_rate": 0.001,
"loss": 1.6633,
"step": 4835
},
{
"epoch": 3.17,
"learning_rate": 0.001,
"loss": 1.455,
"step": 4840
},
{
"epoch": 3.17,
"learning_rate": 0.001,
"loss": 1.7763,
"step": 4845
},
{
"epoch": 3.17,
"learning_rate": 0.001,
"loss": 1.7301,
"step": 4850
},
{
"epoch": 3.18,
"learning_rate": 0.001,
"loss": 1.7956,
"step": 4855
},
{
"epoch": 3.18,
"learning_rate": 0.001,
"loss": 1.7352,
"step": 4860
},
{
"epoch": 3.18,
"learning_rate": 0.001,
"loss": 1.6547,
"step": 4865
},
{
"epoch": 3.19,
"learning_rate": 0.001,
"loss": 1.9271,
"step": 4870
},
{
"epoch": 3.19,
"learning_rate": 0.001,
"loss": 1.68,
"step": 4875
},
{
"epoch": 3.19,
"learning_rate": 0.001,
"loss": 1.9004,
"step": 4880
},
{
"epoch": 3.2,
"learning_rate": 0.001,
"loss": 1.942,
"step": 4885
},
{
"epoch": 3.2,
"learning_rate": 0.001,
"loss": 1.6161,
"step": 4890
},
{
"epoch": 3.2,
"learning_rate": 0.001,
"loss": 1.6055,
"step": 4895
},
{
"epoch": 3.21,
"learning_rate": 0.001,
"loss": 1.7992,
"step": 4900
},
{
"epoch": 3.21,
"learning_rate": 0.001,
"loss": 1.7103,
"step": 4905
},
{
"epoch": 3.21,
"learning_rate": 0.001,
"loss": 1.7935,
"step": 4910
},
{
"epoch": 3.22,
"learning_rate": 0.001,
"loss": 1.8277,
"step": 4915
},
{
"epoch": 3.22,
"learning_rate": 0.001,
"loss": 1.6741,
"step": 4920
},
{
"epoch": 3.22,
"learning_rate": 0.001,
"loss": 1.9507,
"step": 4925
},
{
"epoch": 3.23,
"learning_rate": 0.001,
"loss": 1.7606,
"step": 4930
},
{
"epoch": 3.23,
"learning_rate": 0.001,
"loss": 1.7644,
"step": 4935
},
{
"epoch": 3.23,
"learning_rate": 0.001,
"loss": 1.8093,
"step": 4940
},
{
"epoch": 3.24,
"learning_rate": 0.001,
"loss": 1.9758,
"step": 4945
},
{
"epoch": 3.24,
"learning_rate": 0.001,
"loss": 1.8184,
"step": 4950
},
{
"epoch": 3.24,
"learning_rate": 0.001,
"loss": 1.807,
"step": 4955
},
{
"epoch": 3.25,
"learning_rate": 0.001,
"loss": 1.6004,
"step": 4960
},
{
"epoch": 3.25,
"learning_rate": 0.001,
"loss": 1.8949,
"step": 4965
},
{
"epoch": 3.25,
"learning_rate": 0.001,
"loss": 1.3879,
"step": 4970
},
{
"epoch": 3.26,
"learning_rate": 0.001,
"loss": 1.8248,
"step": 4975
},
{
"epoch": 3.26,
"learning_rate": 0.001,
"loss": 1.8331,
"step": 4980
},
{
"epoch": 3.26,
"learning_rate": 0.001,
"loss": 1.8797,
"step": 4985
},
{
"epoch": 3.27,
"learning_rate": 0.001,
"loss": 1.6555,
"step": 4990
},
{
"epoch": 3.27,
"learning_rate": 0.001,
"loss": 1.45,
"step": 4995
},
{
"epoch": 3.27,
"learning_rate": 0.001,
"loss": 1.8153,
"step": 5000
},
{
"epoch": 3.28,
"learning_rate": 0.001,
"loss": 1.7293,
"step": 5005
},
{
"epoch": 3.28,
"learning_rate": 0.001,
"loss": 1.689,
"step": 5010
},
{
"epoch": 3.28,
"learning_rate": 0.001,
"loss": 1.7214,
"step": 5015
},
{
"epoch": 3.29,
"learning_rate": 0.001,
"loss": 1.7008,
"step": 5020
},
{
"epoch": 3.29,
"learning_rate": 0.001,
"loss": 1.8109,
"step": 5025
},
{
"epoch": 3.29,
"learning_rate": 0.001,
"loss": 1.8343,
"step": 5030
},
{
"epoch": 3.3,
"learning_rate": 0.001,
"loss": 1.7734,
"step": 5035
},
{
"epoch": 3.3,
"learning_rate": 0.001,
"loss": 1.7727,
"step": 5040
},
{
"epoch": 3.3,
"learning_rate": 0.001,
"loss": 1.9285,
"step": 5045
},
{
"epoch": 3.3,
"learning_rate": 0.001,
"loss": 1.9512,
"step": 5050
},
{
"epoch": 3.31,
"learning_rate": 0.001,
"loss": 1.7036,
"step": 5055
},
{
"epoch": 3.31,
"learning_rate": 0.001,
"loss": 1.744,
"step": 5060
},
{
"epoch": 3.31,
"learning_rate": 0.001,
"loss": 1.7296,
"step": 5065
},
{
"epoch": 3.32,
"learning_rate": 0.001,
"loss": 1.8881,
"step": 5070
},
{
"epoch": 3.32,
"learning_rate": 0.001,
"loss": 1.7181,
"step": 5075
},
{
"epoch": 3.32,
"learning_rate": 0.001,
"loss": 1.9565,
"step": 5080
},
{
"epoch": 3.33,
"learning_rate": 0.001,
"loss": 1.8996,
"step": 5085
},
{
"epoch": 3.33,
"learning_rate": 0.001,
"loss": 1.9458,
"step": 5090
},
{
"epoch": 3.33,
"learning_rate": 0.001,
"loss": 1.6911,
"step": 5095
},
{
"epoch": 3.34,
"learning_rate": 0.001,
"loss": 1.6292,
"step": 5100
},
{
"epoch": 3.34,
"learning_rate": 0.001,
"loss": 1.8748,
"step": 5105
},
{
"epoch": 3.34,
"learning_rate": 0.001,
"loss": 1.8222,
"step": 5110
},
{
"epoch": 3.35,
"learning_rate": 0.001,
"loss": 1.6538,
"step": 5115
},
{
"epoch": 3.35,
"learning_rate": 0.001,
"loss": 1.7022,
"step": 5120
},
{
"epoch": 3.35,
"learning_rate": 0.001,
"loss": 1.6222,
"step": 5125
},
{
"epoch": 3.36,
"learning_rate": 0.001,
"loss": 1.7441,
"step": 5130
},
{
"epoch": 3.36,
"learning_rate": 0.001,
"loss": 1.6571,
"step": 5135
},
{
"epoch": 3.36,
"learning_rate": 0.001,
"loss": 1.9465,
"step": 5140
},
{
"epoch": 3.37,
"learning_rate": 0.001,
"loss": 1.4245,
"step": 5145
},
{
"epoch": 3.37,
"learning_rate": 0.001,
"loss": 1.7053,
"step": 5150
},
{
"epoch": 3.37,
"learning_rate": 0.001,
"loss": 1.522,
"step": 5155
},
{
"epoch": 3.38,
"learning_rate": 0.001,
"loss": 1.9395,
"step": 5160
},
{
"epoch": 3.38,
"learning_rate": 0.001,
"loss": 1.9675,
"step": 5165
},
{
"epoch": 3.38,
"learning_rate": 0.001,
"loss": 1.6709,
"step": 5170
},
{
"epoch": 3.39,
"learning_rate": 0.001,
"loss": 1.8034,
"step": 5175
},
{
"epoch": 3.39,
"learning_rate": 0.001,
"loss": 1.7327,
"step": 5180
},
{
"epoch": 3.39,
"learning_rate": 0.001,
"loss": 1.9149,
"step": 5185
},
{
"epoch": 3.4,
"learning_rate": 0.001,
"loss": 1.5989,
"step": 5190
},
{
"epoch": 3.4,
"learning_rate": 0.001,
"loss": 2.0438,
"step": 5195
},
{
"epoch": 3.4,
"learning_rate": 0.001,
"loss": 1.7267,
"step": 5200
},
{
"epoch": 3.41,
"learning_rate": 0.001,
"loss": 1.6355,
"step": 5205
},
{
"epoch": 3.41,
"learning_rate": 0.001,
"loss": 1.8696,
"step": 5210
},
{
"epoch": 3.41,
"learning_rate": 0.001,
"loss": 1.7148,
"step": 5215
},
{
"epoch": 3.42,
"learning_rate": 0.001,
"loss": 1.8159,
"step": 5220
},
{
"epoch": 3.42,
"learning_rate": 0.001,
"loss": 1.7778,
"step": 5225
},
{
"epoch": 3.42,
"learning_rate": 0.001,
"loss": 1.5795,
"step": 5230
},
{
"epoch": 3.43,
"learning_rate": 0.001,
"loss": 1.8921,
"step": 5235
},
{
"epoch": 3.43,
"learning_rate": 0.001,
"loss": 1.8977,
"step": 5240
},
{
"epoch": 3.43,
"learning_rate": 0.001,
"loss": 1.742,
"step": 5245
},
{
"epoch": 3.44,
"learning_rate": 0.001,
"loss": 1.7229,
"step": 5250
},
{
"epoch": 3.44,
"learning_rate": 0.001,
"loss": 1.7709,
"step": 5255
},
{
"epoch": 3.44,
"learning_rate": 0.001,
"loss": 1.9631,
"step": 5260
},
{
"epoch": 3.45,
"learning_rate": 0.001,
"loss": 1.7869,
"step": 5265
},
{
"epoch": 3.45,
"learning_rate": 0.001,
"loss": 1.6913,
"step": 5270
},
{
"epoch": 3.45,
"learning_rate": 0.001,
"loss": 1.7608,
"step": 5275
},
{
"epoch": 3.46,
"learning_rate": 0.001,
"loss": 1.7249,
"step": 5280
},
{
"epoch": 3.46,
"learning_rate": 0.001,
"loss": 1.6338,
"step": 5285
},
{
"epoch": 3.46,
"learning_rate": 0.001,
"loss": 1.8388,
"step": 5290
},
{
"epoch": 3.47,
"learning_rate": 0.001,
"loss": 2.0292,
"step": 5295
},
{
"epoch": 3.47,
"learning_rate": 0.001,
"loss": 1.7162,
"step": 5300
},
{
"epoch": 3.47,
"learning_rate": 0.001,
"loss": 1.7305,
"step": 5305
},
{
"epoch": 3.48,
"learning_rate": 0.001,
"loss": 1.8105,
"step": 5310
},
{
"epoch": 3.48,
"learning_rate": 0.001,
"loss": 1.7147,
"step": 5315
},
{
"epoch": 3.48,
"learning_rate": 0.001,
"loss": 1.7179,
"step": 5320
},
{
"epoch": 3.48,
"learning_rate": 0.001,
"loss": 1.8727,
"step": 5325
},
{
"epoch": 3.49,
"learning_rate": 0.001,
"loss": 1.8479,
"step": 5330
},
{
"epoch": 3.49,
"learning_rate": 0.001,
"loss": 1.5753,
"step": 5335
},
{
"epoch": 3.49,
"learning_rate": 0.001,
"loss": 1.8772,
"step": 5340
},
{
"epoch": 3.5,
"learning_rate": 0.001,
"loss": 1.4923,
"step": 5345
},
{
"epoch": 3.5,
"learning_rate": 0.001,
"loss": 1.5355,
"step": 5350
},
{
"epoch": 3.5,
"learning_rate": 0.001,
"loss": 1.8024,
"step": 5355
},
{
"epoch": 3.51,
"learning_rate": 0.001,
"loss": 1.9238,
"step": 5360
},
{
"epoch": 3.51,
"learning_rate": 0.001,
"loss": 1.8954,
"step": 5365
},
{
"epoch": 3.51,
"learning_rate": 0.001,
"loss": 1.5446,
"step": 5370
},
{
"epoch": 3.52,
"learning_rate": 0.001,
"loss": 1.8563,
"step": 5375
},
{
"epoch": 3.52,
"learning_rate": 0.001,
"loss": 1.6374,
"step": 5380
},
{
"epoch": 3.52,
"learning_rate": 0.001,
"loss": 1.9018,
"step": 5385
},
{
"epoch": 3.53,
"learning_rate": 0.001,
"loss": 1.6509,
"step": 5390
},
{
"epoch": 3.53,
"learning_rate": 0.001,
"loss": 1.5938,
"step": 5395
},
{
"epoch": 3.53,
"learning_rate": 0.001,
"loss": 1.6419,
"step": 5400
},
{
"epoch": 3.54,
"learning_rate": 0.001,
"loss": 1.7466,
"step": 5405
},
{
"epoch": 3.54,
"learning_rate": 0.001,
"loss": 1.6749,
"step": 5410
},
{
"epoch": 3.54,
"learning_rate": 0.001,
"loss": 2.1465,
"step": 5415
},
{
"epoch": 3.55,
"learning_rate": 0.001,
"loss": 1.6034,
"step": 5420
},
{
"epoch": 3.55,
"learning_rate": 0.001,
"loss": 1.9163,
"step": 5425
},
{
"epoch": 3.55,
"learning_rate": 0.001,
"loss": 1.7719,
"step": 5430
},
{
"epoch": 3.56,
"learning_rate": 0.001,
"loss": 1.9431,
"step": 5435
},
{
"epoch": 3.56,
"learning_rate": 0.001,
"loss": 1.4889,
"step": 5440
},
{
"epoch": 3.56,
"learning_rate": 0.001,
"loss": 1.8383,
"step": 5445
},
{
"epoch": 3.57,
"learning_rate": 0.001,
"loss": 1.6336,
"step": 5450
},
{
"epoch": 3.57,
"learning_rate": 0.001,
"loss": 1.6993,
"step": 5455
},
{
"epoch": 3.57,
"learning_rate": 0.001,
"loss": 1.9381,
"step": 5460
},
{
"epoch": 3.58,
"learning_rate": 0.001,
"loss": 1.9026,
"step": 5465
},
{
"epoch": 3.58,
"learning_rate": 0.001,
"loss": 2.1256,
"step": 5470
},
{
"epoch": 3.58,
"learning_rate": 0.001,
"loss": 1.7473,
"step": 5475
},
{
"epoch": 3.59,
"learning_rate": 0.001,
"loss": 1.7186,
"step": 5480
},
{
"epoch": 3.59,
"learning_rate": 0.001,
"loss": 1.5797,
"step": 5485
},
{
"epoch": 3.59,
"learning_rate": 0.001,
"loss": 1.8156,
"step": 5490
},
{
"epoch": 3.6,
"learning_rate": 0.001,
"loss": 1.5645,
"step": 5495
},
{
"epoch": 3.6,
"learning_rate": 0.001,
"loss": 1.773,
"step": 5500
},
{
"epoch": 3.6,
"learning_rate": 0.001,
"loss": 2.0786,
"step": 5505
},
{
"epoch": 3.61,
"learning_rate": 0.001,
"loss": 1.7005,
"step": 5510
},
{
"epoch": 3.61,
"learning_rate": 0.001,
"loss": 1.8878,
"step": 5515
},
{
"epoch": 3.61,
"learning_rate": 0.001,
"loss": 1.9137,
"step": 5520
},
{
"epoch": 3.62,
"learning_rate": 0.001,
"loss": 1.5241,
"step": 5525
},
{
"epoch": 3.62,
"learning_rate": 0.001,
"loss": 1.6922,
"step": 5530
},
{
"epoch": 3.62,
"learning_rate": 0.001,
"loss": 1.8958,
"step": 5535
},
{
"epoch": 3.63,
"learning_rate": 0.001,
"loss": 2.1021,
"step": 5540
},
{
"epoch": 3.63,
"learning_rate": 0.001,
"loss": 1.673,
"step": 5545
},
{
"epoch": 3.63,
"learning_rate": 0.001,
"loss": 1.9568,
"step": 5550
},
{
"epoch": 3.64,
"learning_rate": 0.001,
"loss": 1.8155,
"step": 5555
},
{
"epoch": 3.64,
"learning_rate": 0.001,
"loss": 1.8575,
"step": 5560
},
{
"epoch": 3.64,
"learning_rate": 0.001,
"loss": 1.8335,
"step": 5565
},
{
"epoch": 3.65,
"learning_rate": 0.001,
"loss": 1.7428,
"step": 5570
},
{
"epoch": 3.65,
"learning_rate": 0.001,
"loss": 1.8624,
"step": 5575
},
{
"epoch": 3.65,
"learning_rate": 0.001,
"loss": 1.9491,
"step": 5580
},
{
"epoch": 3.66,
"learning_rate": 0.001,
"loss": 1.7798,
"step": 5585
},
{
"epoch": 3.66,
"learning_rate": 0.001,
"loss": 1.8362,
"step": 5590
},
{
"epoch": 3.66,
"learning_rate": 0.001,
"loss": 1.7843,
"step": 5595
},
{
"epoch": 3.66,
"learning_rate": 0.001,
"loss": 1.8491,
"step": 5600
},
{
"epoch": 3.67,
"learning_rate": 0.001,
"loss": 1.6601,
"step": 5605
},
{
"epoch": 3.67,
"learning_rate": 0.001,
"loss": 1.8691,
"step": 5610
},
{
"epoch": 3.67,
"learning_rate": 0.001,
"loss": 2.0051,
"step": 5615
},
{
"epoch": 3.68,
"learning_rate": 0.001,
"loss": 1.9965,
"step": 5620
},
{
"epoch": 3.68,
"learning_rate": 0.001,
"loss": 1.847,
"step": 5625
},
{
"epoch": 3.68,
"learning_rate": 0.001,
"loss": 1.8804,
"step": 5630
},
{
"epoch": 3.69,
"learning_rate": 0.001,
"loss": 1.8575,
"step": 5635
},
{
"epoch": 3.69,
"learning_rate": 0.001,
"loss": 1.4412,
"step": 5640
},
{
"epoch": 3.69,
"learning_rate": 0.001,
"loss": 1.7574,
"step": 5645
},
{
"epoch": 3.7,
"learning_rate": 0.001,
"loss": 1.7796,
"step": 5650
},
{
"epoch": 3.7,
"learning_rate": 0.001,
"loss": 1.7425,
"step": 5655
},
{
"epoch": 3.7,
"learning_rate": 0.001,
"loss": 2.1566,
"step": 5660
},
{
"epoch": 3.71,
"learning_rate": 0.001,
"loss": 2.03,
"step": 5665
},
{
"epoch": 3.71,
"learning_rate": 0.001,
"loss": 1.693,
"step": 5670
},
{
"epoch": 3.71,
"learning_rate": 0.001,
"loss": 1.8462,
"step": 5675
},
{
"epoch": 3.72,
"learning_rate": 0.001,
"loss": 2.0286,
"step": 5680
},
{
"epoch": 3.72,
"learning_rate": 0.001,
"loss": 1.6423,
"step": 5685
},
{
"epoch": 3.72,
"learning_rate": 0.001,
"loss": 1.6449,
"step": 5690
},
{
"epoch": 3.73,
"learning_rate": 0.001,
"loss": 1.8167,
"step": 5695
},
{
"epoch": 3.73,
"learning_rate": 0.001,
"loss": 1.6899,
"step": 5700
},
{
"epoch": 3.73,
"learning_rate": 0.001,
"loss": 1.9542,
"step": 5705
},
{
"epoch": 3.74,
"learning_rate": 0.001,
"loss": 1.5409,
"step": 5710
},
{
"epoch": 3.74,
"learning_rate": 0.001,
"loss": 1.9445,
"step": 5715
},
{
"epoch": 3.74,
"learning_rate": 0.001,
"loss": 1.5969,
"step": 5720
},
{
"epoch": 3.75,
"learning_rate": 0.001,
"loss": 1.8589,
"step": 5725
},
{
"epoch": 3.75,
"learning_rate": 0.001,
"loss": 2.1791,
"step": 5730
},
{
"epoch": 3.75,
"learning_rate": 0.001,
"loss": 2.126,
"step": 5735
},
{
"epoch": 3.76,
"learning_rate": 0.001,
"loss": 1.6072,
"step": 5740
},
{
"epoch": 3.76,
"learning_rate": 0.001,
"loss": 1.8024,
"step": 5745
},
{
"epoch": 3.76,
"learning_rate": 0.001,
"loss": 1.9641,
"step": 5750
},
{
"epoch": 3.77,
"learning_rate": 0.001,
"loss": 1.7395,
"step": 5755
},
{
"epoch": 3.77,
"learning_rate": 0.001,
"loss": 1.8684,
"step": 5760
},
{
"epoch": 3.77,
"learning_rate": 0.001,
"loss": 1.6303,
"step": 5765
},
{
"epoch": 3.78,
"learning_rate": 0.001,
"loss": 1.7802,
"step": 5770
},
{
"epoch": 3.78,
"learning_rate": 0.001,
"loss": 2.0401,
"step": 5775
},
{
"epoch": 3.78,
"learning_rate": 0.001,
"loss": 1.9099,
"step": 5780
},
{
"epoch": 3.79,
"learning_rate": 0.001,
"loss": 2.0348,
"step": 5785
},
{
"epoch": 3.79,
"learning_rate": 0.001,
"loss": 1.8504,
"step": 5790
},
{
"epoch": 3.79,
"learning_rate": 0.001,
"loss": 1.9954,
"step": 5795
},
{
"epoch": 3.8,
"learning_rate": 0.001,
"loss": 1.5903,
"step": 5800
},
{
"epoch": 3.8,
"learning_rate": 0.001,
"loss": 1.922,
"step": 5805
},
{
"epoch": 3.8,
"learning_rate": 0.001,
"loss": 1.8154,
"step": 5810
},
{
"epoch": 3.81,
"learning_rate": 0.001,
"loss": 1.7238,
"step": 5815
},
{
"epoch": 3.81,
"learning_rate": 0.001,
"loss": 1.6914,
"step": 5820
},
{
"epoch": 3.81,
"learning_rate": 0.001,
"loss": 1.6808,
"step": 5825
},
{
"epoch": 3.82,
"learning_rate": 0.001,
"loss": 1.8802,
"step": 5830
},
{
"epoch": 3.82,
"learning_rate": 0.001,
"loss": 2.0595,
"step": 5835
},
{
"epoch": 3.82,
"learning_rate": 0.001,
"loss": 2.0912,
"step": 5840
},
{
"epoch": 3.83,
"learning_rate": 0.001,
"loss": 1.9024,
"step": 5845
},
{
"epoch": 3.83,
"learning_rate": 0.001,
"loss": 1.6901,
"step": 5850
},
{
"epoch": 3.83,
"learning_rate": 0.001,
"loss": 1.9322,
"step": 5855
},
{
"epoch": 3.84,
"learning_rate": 0.001,
"loss": 1.715,
"step": 5860
},
{
"epoch": 3.84,
"learning_rate": 0.001,
"loss": 1.9352,
"step": 5865
},
{
"epoch": 3.84,
"learning_rate": 0.001,
"loss": 2.0361,
"step": 5870
},
{
"epoch": 3.84,
"learning_rate": 0.001,
"loss": 1.632,
"step": 5875
},
{
"epoch": 3.85,
"learning_rate": 0.001,
"loss": 1.6774,
"step": 5880
},
{
"epoch": 3.85,
"learning_rate": 0.001,
"loss": 2.1169,
"step": 5885
},
{
"epoch": 3.85,
"learning_rate": 0.001,
"loss": 1.6391,
"step": 5890
},
{
"epoch": 3.86,
"learning_rate": 0.001,
"loss": 2.0607,
"step": 5895
},
{
"epoch": 3.86,
"learning_rate": 0.001,
"loss": 1.7642,
"step": 5900
},
{
"epoch": 3.86,
"learning_rate": 0.001,
"loss": 1.8561,
"step": 5905
},
{
"epoch": 3.87,
"learning_rate": 0.001,
"loss": 1.7327,
"step": 5910
},
{
"epoch": 3.87,
"learning_rate": 0.001,
"loss": 1.8818,
"step": 5915
},
{
"epoch": 3.87,
"learning_rate": 0.001,
"loss": 1.7906,
"step": 5920
},
{
"epoch": 3.88,
"learning_rate": 0.001,
"loss": 1.7148,
"step": 5925
},
{
"epoch": 3.88,
"learning_rate": 0.001,
"loss": 1.8405,
"step": 5930
},
{
"epoch": 3.88,
"learning_rate": 0.001,
"loss": 1.9279,
"step": 5935
},
{
"epoch": 3.89,
"learning_rate": 0.001,
"loss": 1.8585,
"step": 5940
},
{
"epoch": 3.89,
"learning_rate": 0.001,
"loss": 2.1013,
"step": 5945
},
{
"epoch": 3.89,
"learning_rate": 0.001,
"loss": 1.8939,
"step": 5950
},
{
"epoch": 3.9,
"learning_rate": 0.001,
"loss": 1.8666,
"step": 5955
},
{
"epoch": 3.9,
"learning_rate": 0.001,
"loss": 2.0714,
"step": 5960
},
{
"epoch": 3.9,
"learning_rate": 0.001,
"loss": 1.8829,
"step": 5965
},
{
"epoch": 3.91,
"learning_rate": 0.001,
"loss": 1.8614,
"step": 5970
},
{
"epoch": 3.91,
"learning_rate": 0.001,
"loss": 1.9029,
"step": 5975
},
{
"epoch": 3.91,
"learning_rate": 0.001,
"loss": 1.9093,
"step": 5980
},
{
"epoch": 3.92,
"learning_rate": 0.001,
"loss": 1.8506,
"step": 5985
},
{
"epoch": 3.92,
"learning_rate": 0.001,
"loss": 2.0245,
"step": 5990
},
{
"epoch": 3.92,
"learning_rate": 0.001,
"loss": 2.075,
"step": 5995
},
{
"epoch": 3.93,
"learning_rate": 0.001,
"loss": 1.6811,
"step": 6000
}
],
"logging_steps": 5,
"max_steps": 7640,
"num_train_epochs": 5,
"save_steps": 1000,
"total_flos": 700710912000000.0,
"trial_name": null,
"trial_params": null
}