Hugging Face's logo

qanastek
/
MedAlpaca-LLaMa2-7B

PEFT
English
medical
llama2
Model card Files Community
MedAlpaca-LLaMa2-7B / trainer_state.json
qanastek's picture
qanastek
Upload 12 files
39ea18c
raw
history blame contribute delete
93.4 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0057142857142858,
"global_step": 8800,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.7431,
"step": 10
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.5523,
"step": 20
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.5362,
"step": 30
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.5844,
"step": 40
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 1.2511,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 1.363,
"step": 60
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 1.3879,
"step": 70
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 1.5116,
"step": 80
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 1.4911,
"step": 90
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 1.2505,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 1.3797,
"step": 110
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 1.3631,
"step": 120
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 1.4894,
"step": 130
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 1.5071,
"step": 140
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 1.1832,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 1.376,
"step": 160
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 1.395,
"step": 170
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 1.4788,
"step": 180
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 1.5353,
"step": 190
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 1.1837,
"step": 200
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 1.3746,
"step": 210
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 1.4004,
"step": 220
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 1.4537,
"step": 230
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 1.5133,
"step": 240
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 1.123,
"step": 250
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 1.333,
"step": 260
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 1.4077,
"step": 270
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 1.5152,
"step": 280
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 1.5137,
"step": 290
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 1.1933,
"step": 300
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 1.4112,
"step": 310
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 1.4278,
"step": 320
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 1.4328,
"step": 330
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 1.5322,
"step": 340
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 1.2536,
"step": 350
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 1.3978,
"step": 360
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 1.422,
"step": 370
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 1.512,
"step": 380
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 1.4934,
"step": 390
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 1.1298,
"step": 400
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 1.3553,
"step": 410
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 1.4127,
"step": 420
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 1.45,
"step": 430
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 1.4496,
"step": 440
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 1.1942,
"step": 450
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 1.3397,
"step": 460
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 1.3896,
"step": 470
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 1.5023,
"step": 480
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 1.5056,
"step": 490
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 1.1338,
"step": 500
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 1.3907,
"step": 510
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 1.4089,
"step": 520
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 1.4939,
"step": 530
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 1.5232,
"step": 540
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 1.2392,
"step": 550
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 1.3408,
"step": 560
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 1.4217,
"step": 570
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 1.4783,
"step": 580
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 1.4852,
"step": 590
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 1.2052,
"step": 600
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 1.3532,
"step": 610
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 1.4059,
"step": 620
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 1.5072,
"step": 630
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 1.4945,
"step": 640
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 1.1679,
"step": 650
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 1.3999,
"step": 660
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 1.3628,
"step": 670
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 1.4646,
"step": 680
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 1.5183,
"step": 690
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 1.1757,
"step": 700
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 1.333,
"step": 710
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 1.3804,
"step": 720
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 1.4815,
"step": 730
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 1.4817,
"step": 740
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 1.2177,
"step": 750
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 1.3734,
"step": 760
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 1.402,
"step": 770
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 1.4573,
"step": 780
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 1.5318,
"step": 790
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 1.2027,
"step": 800
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 1.3658,
"step": 810
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 1.3659,
"step": 820
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 1.5035,
"step": 830
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 1.4851,
"step": 840
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 1.1523,
"step": 850
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 1.3534,
"step": 860
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 1.4259,
"step": 870
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 1.5088,
"step": 880
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 1.505,
"step": 890
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 1.1439,
"step": 900
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 1.3234,
"step": 910
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 1.3883,
"step": 920
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 1.4831,
"step": 930
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 1.4565,
"step": 940
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 1.1611,
"step": 950
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 1.3911,
"step": 960
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 1.3923,
"step": 970
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 1.4645,
"step": 980
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 1.4959,
"step": 990
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 1.1621,
"step": 1000
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 1.3755,
"step": 1010
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 1.36,
"step": 1020
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 1.4925,
"step": 1030
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 1.4878,
"step": 1040
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 1.1575,
"step": 1050
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 1.3751,
"step": 1060
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 1.3347,
"step": 1070
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 1.5014,
"step": 1080
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 1.5137,
"step": 1090
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 1.2194,
"step": 1100
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 1.3787,
"step": 1110
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 1.3776,
"step": 1120
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 1.5055,
"step": 1130
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 1.4581,
"step": 1140
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 1.1616,
"step": 1150
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 1.33,
"step": 1160
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 1.414,
"step": 1170
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 1.5124,
"step": 1180
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 1.5139,
"step": 1190
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 1.1085,
"step": 1200
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 1.355,
"step": 1210
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 1.3597,
"step": 1220
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 1.4835,
"step": 1230
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 1.4739,
"step": 1240
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 1.1206,
"step": 1250
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 1.3581,
"step": 1260
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 1.388,
"step": 1270
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 1.4624,
"step": 1280
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 1.4926,
"step": 1290
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 1.1745,
"step": 1300
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 1.3422,
"step": 1310
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 1.3778,
"step": 1320
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 1.5218,
"step": 1330
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 1.5068,
"step": 1340
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 1.1988,
"step": 1350
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 1.3527,
"step": 1360
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 1.4118,
"step": 1370
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 1.46,
"step": 1380
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 1.5036,
"step": 1390
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 1.2142,
"step": 1400
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 1.3566,
"step": 1410
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 1.4106,
"step": 1420
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 1.4844,
"step": 1430
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 1.469,
"step": 1440
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 1.1471,
"step": 1450
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 1.3605,
"step": 1460
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 1.3821,
"step": 1470
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 1.4964,
"step": 1480
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 1.493,
"step": 1490
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 1.2144,
"step": 1500
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 1.3792,
"step": 1510
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 1.3787,
"step": 1520
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 1.4801,
"step": 1530
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 1.5004,
"step": 1540
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 1.1574,
"step": 1550
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 1.371,
"step": 1560
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 1.3763,
"step": 1570
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 1.4883,
"step": 1580
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 1.5189,
"step": 1590
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 1.0868,
"step": 1600
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 1.3933,
"step": 1610
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 1.4105,
"step": 1620
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 1.4881,
"step": 1630
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 1.4615,
"step": 1640
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 1.0866,
"step": 1650
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 1.3884,
"step": 1660
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 1.3647,
"step": 1670
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 1.5341,
"step": 1680
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 1.5178,
"step": 1690
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 1.1867,
"step": 1700
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 1.3494,
"step": 1710
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 1.3895,
"step": 1720
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 1.4585,
"step": 1730
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 1.4815,
"step": 1740
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 1.1065,
"step": 1750
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 1.3551,
"step": 1760
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 1.388,
"step": 1770
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 1.4888,
"step": 1780
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 1.4693,
"step": 1790
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 1.2007,
"step": 1800
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 1.3711,
"step": 1810
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 1.4145,
"step": 1820
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 1.4982,
"step": 1830
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 1.4874,
"step": 1840
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 1.1387,
"step": 1850
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 1.3761,
"step": 1860
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 1.3737,
"step": 1870
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 1.4931,
"step": 1880
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 1.4763,
"step": 1890
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 1.1276,
"step": 1900
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 1.3674,
"step": 1910
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 1.4133,
"step": 1920
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 1.451,
"step": 1930
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 1.5189,
"step": 1940
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 1.0636,
"step": 1950
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 1.3954,
"step": 1960
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 1.3651,
"step": 1970
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 1.4482,
"step": 1980
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 1.478,
"step": 1990
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 1.0778,
"step": 2000
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 1.3601,
"step": 2010
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 1.4251,
"step": 2020
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 1.4854,
"step": 2030
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 1.4664,
"step": 2040
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 1.141,
"step": 2050
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 1.3472,
"step": 2060
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 1.3278,
"step": 2070
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 1.5566,
"step": 2080
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 1.4889,
"step": 2090
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 1.2219,
"step": 2100
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 1.3655,
"step": 2110
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 1.4287,
"step": 2120
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 1.4809,
"step": 2130
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 1.5159,
"step": 2140
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 1.1543,
"step": 2150
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 1.3774,
"step": 2160
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 1.3806,
"step": 2170
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 1.4786,
"step": 2180
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 1.4945,
"step": 2190
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 1.1604,
"step": 2200
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 1.3667,
"step": 2210
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 1.3984,
"step": 2220
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 1.4627,
"step": 2230
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 1.4767,
"step": 2240
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 1.0874,
"step": 2250
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 1.3641,
"step": 2260
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 1.4196,
"step": 2270
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 1.5019,
"step": 2280
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 1.4729,
"step": 2290
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 1.0994,
"step": 2300
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 1.3485,
"step": 2310
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 1.353,
"step": 2320
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 1.4497,
"step": 2330
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 1.4204,
"step": 2340
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 1.1602,
"step": 2350
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 1.3371,
"step": 2360
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 1.404,
"step": 2370
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 1.5122,
"step": 2380
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 1.4756,
"step": 2390
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 1.1223,
"step": 2400
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 1.3523,
"step": 2410
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 1.433,
"step": 2420
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 1.5184,
"step": 2430
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 1.5016,
"step": 2440
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 1.1152,
"step": 2450
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 1.4008,
"step": 2460
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 1.405,
"step": 2470
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 1.4862,
"step": 2480
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 1.48,
"step": 2490
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 1.1708,
"step": 2500
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 1.3488,
"step": 2510
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 1.3866,
"step": 2520
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 1.4876,
"step": 2530
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 1.4691,
"step": 2540
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 1.1237,
"step": 2550
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 1.3791,
"step": 2560
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 1.3704,
"step": 2570
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 1.4423,
"step": 2580
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 1.5363,
"step": 2590
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 1.1361,
"step": 2600
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 1.3744,
"step": 2610
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 1.4153,
"step": 2620
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 1.5511,
"step": 2630
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 1.4789,
"step": 2640
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 1.1828,
"step": 2650
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 1.3552,
"step": 2660
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 1.3887,
"step": 2670
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 1.5207,
"step": 2680
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 1.4766,
"step": 2690
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 1.1612,
"step": 2700
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 1.3412,
"step": 2710
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 1.4014,
"step": 2720
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 1.4732,
"step": 2730
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 1.4645,
"step": 2740
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 1.1439,
"step": 2750
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 1.3987,
"step": 2760
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 1.4204,
"step": 2770
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 1.5018,
"step": 2780
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 1.5062,
"step": 2790
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 1.1205,
"step": 2800
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 1.3577,
"step": 2810
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 1.3845,
"step": 2820
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 1.4913,
"step": 2830
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 1.5165,
"step": 2840
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 1.1937,
"step": 2850
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 1.3885,
"step": 2860
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 1.3824,
"step": 2870
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 1.4714,
"step": 2880
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 1.4905,
"step": 2890
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 1.1824,
"step": 2900
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 1.3445,
"step": 2910
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 1.38,
"step": 2920
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 1.4518,
"step": 2930
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 1.4752,
"step": 2940
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 1.1911,
"step": 2950
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 1.3991,
"step": 2960
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 1.3773,
"step": 2970
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 1.4901,
"step": 2980
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 1.4659,
"step": 2990
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 1.1806,
"step": 3000
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 1.35,
"step": 3010
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 1.4208,
"step": 3020
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 1.4913,
"step": 3030
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 1.4977,
"step": 3040
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 1.2201,
"step": 3050
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 1.3578,
"step": 3060
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 1.3585,
"step": 3070
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 1.4842,
"step": 3080
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 1.461,
"step": 3090
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 1.1151,
"step": 3100
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 1.3658,
"step": 3110
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 1.4156,
"step": 3120
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 1.4507,
"step": 3130
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 1.4868,
"step": 3140
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 1.1919,
"step": 3150
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 1.3572,
"step": 3160
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 1.408,
"step": 3170
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 1.513,
"step": 3180
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 1.4418,
"step": 3190
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 1.1063,
"step": 3200
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 1.3842,
"step": 3210
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 1.4063,
"step": 3220
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 1.5104,
"step": 3230
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 1.4726,
"step": 3240
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 1.1022,
"step": 3250
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 1.3459,
"step": 3260
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 1.4093,
"step": 3270
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 1.4945,
"step": 3280
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 1.5151,
"step": 3290
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 1.1194,
"step": 3300
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 1.3495,
"step": 3310
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 1.348,
"step": 3320
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 1.5025,
"step": 3330
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 1.4858,
"step": 3340
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 1.1149,
"step": 3350
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 1.3526,
"step": 3360
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 1.4106,
"step": 3370
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 1.5304,
"step": 3380
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 1.462,
"step": 3390
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 1.0945,
"step": 3400
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 1.3582,
"step": 3410
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 1.3727,
"step": 3420
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 1.4673,
"step": 3430
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 1.4891,
"step": 3440
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 1.156,
"step": 3450
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 1.3427,
"step": 3460
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 1.3656,
"step": 3470
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 1.4977,
"step": 3480
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 1.4801,
"step": 3490
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 1.0395,
"step": 3500
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 1.3571,
"step": 3510
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 1.3911,
"step": 3520
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 1.4859,
"step": 3530
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 1.4886,
"step": 3540
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 1.1248,
"step": 3550
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 1.3461,
"step": 3560
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 1.4296,
"step": 3570
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 1.5224,
"step": 3580
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 1.4436,
"step": 3590
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 1.1429,
"step": 3600
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 1.4042,
"step": 3610
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 1.3643,
"step": 3620
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 1.4806,
"step": 3630
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 1.4694,
"step": 3640
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 1.1284,
"step": 3650
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 1.3763,
"step": 3660
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 1.3785,
"step": 3670
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 1.5124,
"step": 3680
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 1.4636,
"step": 3690
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 1.1329,
"step": 3700
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 1.3724,
"step": 3710
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 1.3632,
"step": 3720
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 1.4831,
"step": 3730
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 1.436,
"step": 3740
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 1.0665,
"step": 3750
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 1.3238,
"step": 3760
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 1.3803,
"step": 3770
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 1.4619,
"step": 3780
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 1.5251,
"step": 3790
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 1.1907,
"step": 3800
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 1.3207,
"step": 3810
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 1.3872,
"step": 3820
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 1.5255,
"step": 3830
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 1.4855,
"step": 3840
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 1.1058,
"step": 3850
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 1.3872,
"step": 3860
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 1.415,
"step": 3870
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 1.5502,
"step": 3880
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 1.4852,
"step": 3890
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 1.1746,
"step": 3900
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 1.3513,
"step": 3910
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 1.3604,
"step": 3920
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 1.4799,
"step": 3930
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 1.5093,
"step": 3940
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 1.1486,
"step": 3950
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 1.3901,
"step": 3960
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 1.3957,
"step": 3970
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 1.4782,
"step": 3980
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 1.5271,
"step": 3990
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 1.1536,
"step": 4000
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 1.4057,
"step": 4010
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 1.4112,
"step": 4020
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 1.5066,
"step": 4030
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 1.4654,
"step": 4040
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 1.1995,
"step": 4050
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 1.3912,
"step": 4060
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 1.4331,
"step": 4070
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 1.4904,
"step": 4080
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 1.4481,
"step": 4090
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 1.0531,
"step": 4100
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 1.4215,
"step": 4110
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 1.3689,
"step": 4120
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 1.5042,
"step": 4130
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 1.4755,
"step": 4140
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 1.1606,
"step": 4150
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 1.379,
"step": 4160
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 1.3928,
"step": 4170
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 1.4858,
"step": 4180
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 1.4799,
"step": 4190
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 1.1618,
"step": 4200
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 1.3297,
"step": 4210
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 1.3624,
"step": 4220
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 1.5152,
"step": 4230
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 1.4984,
"step": 4240
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 1.1028,
"step": 4250
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 1.3869,
"step": 4260
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 1.4085,
"step": 4270
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 1.4954,
"step": 4280
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 1.5095,
"step": 4290
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 1.1629,
"step": 4300
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 1.364,
"step": 4310
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 1.4175,
"step": 4320
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 1.4608,
"step": 4330
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 1.4981,
"step": 4340
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 1.139,
"step": 4350
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 1.3476,
"step": 4360
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 1.4028,
"step": 4370
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 1.5133,
"step": 4380
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 1.5031,
"step": 4390
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 1.1492,
"step": 4400
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 1.3831,
"step": 4410
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 1.4053,
"step": 4420
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 1.491,
"step": 4430
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 1.4898,
"step": 4440
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 1.1842,
"step": 4450
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 1.3801,
"step": 4460
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 1.3624,
"step": 4470
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 1.5033,
"step": 4480
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 1.4971,
"step": 4490
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 1.1225,
"step": 4500
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 1.3766,
"step": 4510
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 1.4046,
"step": 4520
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 1.516,
"step": 4530
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 1.5033,
"step": 4540
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 1.1002,
"step": 4550
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 1.3724,
"step": 4560
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 1.3755,
"step": 4570
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 1.5143,
"step": 4580
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 1.5542,
"step": 4590
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 1.1288,
"step": 4600
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 1.3631,
"step": 4610
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 1.4227,
"step": 4620
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 1.4712,
"step": 4630
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 1.4922,
"step": 4640
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 1.2433,
"step": 4650
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 1.3936,
"step": 4660
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 1.453,
"step": 4670
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 1.5069,
"step": 4680
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 1.5577,
"step": 4690
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 1.174,
"step": 4700
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 1.3459,
"step": 4710
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 1.3971,
"step": 4720
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 1.535,
"step": 4730
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 1.5099,
"step": 4740
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 1.0992,
"step": 4750
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 1.3479,
"step": 4760
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 1.3866,
"step": 4770
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 1.4732,
"step": 4780
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 1.5125,
"step": 4790
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 1.0968,
"step": 4800
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 1.3876,
"step": 4810
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 1.4057,
"step": 4820
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 1.5399,
"step": 4830
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 1.4995,
"step": 4840
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 1.1411,
"step": 4850
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 1.4131,
"step": 4860
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 1.4054,
"step": 4870
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 1.4959,
"step": 4880
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 1.5191,
"step": 4890
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 1.1974,
"step": 4900
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 1.4629,
"step": 4910
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 1.6002,
"step": 4920
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 1.6047,
"step": 4930
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 1.6153,
"step": 4940
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 1.2109,
"step": 4950
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 1.4794,
"step": 4960
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 1.5028,
"step": 4970
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 1.6208,
"step": 4980
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 1.6215,
"step": 4990
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 1.2066,
"step": 5000
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 1.592,
"step": 5010
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 1.5682,
"step": 5020
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 1.6461,
"step": 5030
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 1.6425,
"step": 5040
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 1.2823,
"step": 5050
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 1.5915,
"step": 5060
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 1.5673,
"step": 5070
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 1.724,
"step": 5080
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 1.7876,
"step": 5090
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 1.3042,
"step": 5100
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 1.8139,
"step": 5110
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 1.8781,
"step": 5120
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 2.0134,
"step": 5130
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 1.9675,
"step": 5140
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 1.3781,
"step": 5150
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 1.8037,
"step": 5160
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 1.9102,
"step": 5170
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 2.002,
"step": 5180
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 1.9065,
"step": 5190
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 1.4923,
"step": 5200
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 1.6716,
"step": 5210
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 1.664,
"step": 5220
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 1.6999,
"step": 5230
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 1.7727,
"step": 5240
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 1.4808,
"step": 5250
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 1.6987,
"step": 5260
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 1.8903,
"step": 5270
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 2.0523,
"step": 5280
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 1.9036,
"step": 5290
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 1.536,
"step": 5300
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 1.8511,
"step": 5310
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 1.8459,
"step": 5320
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 1.9724,
"step": 5330
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 1.9299,
"step": 5340
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 1.6449,
"step": 5350
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 1.7879,
"step": 5360
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 1.7839,
"step": 5370
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 1.8283,
"step": 5380
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 1.8055,
"step": 5390
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 1.4122,
"step": 5400
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 1.6398,
"step": 5410
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 1.6537,
"step": 5420
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 1.6973,
"step": 5430
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 1.8209,
"step": 5440
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 1.6343,
"step": 5450
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 2.0785,
"step": 5460
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 4.1734,
"step": 5470
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 6.7746,
"step": 5480
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 5.0767,
"step": 5490
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 3.3103,
"step": 5500
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 4.5979,
"step": 5510
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 4.3999,
"step": 5520
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 4.0496,
"step": 5530
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 4.2599,
"step": 5540
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 3.6741,
"step": 5550
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 4.7093,
"step": 5560
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 4.016,
"step": 5570
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 4.2352,
"step": 5580
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 3.8359,
"step": 5590
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 2.7654,
"step": 5600
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 4.6375,
"step": 5610
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 4.5832,
"step": 5620
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 3.7359,
"step": 5630
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 3.3395,
"step": 5640
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 2.6695,
"step": 5650
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 4.6645,
"step": 5660
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 4.7006,
"step": 5670
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 4.377,
"step": 5680
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 3.8076,
"step": 5690
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 2.5118,
"step": 5700
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 3.2971,
"step": 5710
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 3.4954,
"step": 5720
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 4.1866,
"step": 5730
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 4.3862,
"step": 5740
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 2.9591,
"step": 5750
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 4.2781,
"step": 5760
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 3.9847,
"step": 5770
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 3.802,
"step": 5780
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 2.9336,
"step": 5790
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 2.183,
"step": 5800
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 3.3075,
"step": 5810
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 3.453,
"step": 5820
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 3.3029,
"step": 5830
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 2.809,
"step": 5840
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 1.8697,
"step": 5850
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 2.76,
"step": 5860
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 2.9094,
"step": 5870
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 2.7784,
"step": 5880
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 2.5071,
"step": 5890
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 1.8809,
"step": 5900
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 2.459,
"step": 5910
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 2.8066,
"step": 5920
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 4.1141,
"step": 5930
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 4.0347,
"step": 5940
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 3.0017,
"step": 5950
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 3.2676,
"step": 5960
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 3.1042,
"step": 5970
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 3.2889,
"step": 5980
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 3.3139,
"step": 5990
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 2.4935,
"step": 6000
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 4.1095,
"step": 6010
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 3.6322,
"step": 6020
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 3.3309,
"step": 6030
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 3.0428,
"step": 6040
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 2.4984,
"step": 6050
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 3.5314,
"step": 6060
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 3.6307,
"step": 6070
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 3.2909,
"step": 6080
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 2.8702,
"step": 6090
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 2.2269,
"step": 6100
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 3.3475,
"step": 6110
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 4.0399,
"step": 6120
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 5.27,
"step": 6130
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 4.9144,
"step": 6140
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 3.0834,
"step": 6150
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 3.7826,
"step": 6160
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 3.4911,
"step": 6170
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 3.2003,
"step": 6180
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 2.9853,
"step": 6190
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 2.1288,
"step": 6200
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 3.1077,
"step": 6210
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 3.065,
"step": 6220
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 2.753,
"step": 6230
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 2.6375,
"step": 6240
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 2.1715,
"step": 6250
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 3.2288,
"step": 6260
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 3.2531,
"step": 6270
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 3.1586,
"step": 6280
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 3.3152,
"step": 6290
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 2.4145,
"step": 6300
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 4.0466,
"step": 6310
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 4.0243,
"step": 6320
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 3.522,
"step": 6330
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 3.106,
"step": 6340
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 2.358,
"step": 6350
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 4.1717,
"step": 6360
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 4.3043,
"step": 6370
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 4.1694,
"step": 6380
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 3.9152,
"step": 6390
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 2.806,
"step": 6400
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 5.7838,
"step": 6410
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 5.6406,
"step": 6420
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 5.0501,
"step": 6430
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 4.2304,
"step": 6440
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 2.7792,
"step": 6450
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 4.368,
"step": 6460
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 3.9682,
"step": 6470
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 3.7431,
"step": 6480
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 3.6235,
"step": 6490
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 2.5896,
"step": 6500
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 3.8402,
"step": 6510
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 3.4356,
"step": 6520
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 3.0214,
"step": 6530
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 2.7369,
"step": 6540
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 2.0243,
"step": 6550
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 3.0303,
"step": 6560
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 2.9176,
"step": 6570
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 2.8165,
"step": 6580
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 2.711,
"step": 6590
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 2.1248,
"step": 6600
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 3.3661,
"step": 6610
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 3.2783,
"step": 6620
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 3.0595,
"step": 6630
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 2.9547,
"step": 6640
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 2.3153,
"step": 6650
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 3.3942,
"step": 6660
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 3.2098,
"step": 6670
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 3.134,
"step": 6680
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 3.0744,
"step": 6690
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 2.702,
"step": 6700
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 3.6413,
"step": 6710
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 3.7017,
"step": 6720
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 3.601,
"step": 6730
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 3.2477,
"step": 6740
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 2.5295,
"step": 6750
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 3.6857,
"step": 6760
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 3.6193,
"step": 6770
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 3.4057,
"step": 6780
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 3.1279,
"step": 6790
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 2.3629,
"step": 6800
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 3.3246,
"step": 6810
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 3.1549,
"step": 6820
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 2.9732,
"step": 6830
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 2.7179,
"step": 6840
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 2.1229,
"step": 6850
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 3.1299,
"step": 6860
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 3.1563,
"step": 6870
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 2.8985,
"step": 6880
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 2.6909,
"step": 6890
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 2.005,
"step": 6900
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 3.2559,
"step": 6910
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 3.2412,
"step": 6920
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 2.8569,
"step": 6930
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 2.5241,
"step": 6940
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 1.837,
"step": 6950
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 2.5914,
"step": 6960
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 2.4782,
"step": 6970
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 2.4614,
"step": 6980
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 2.4431,
"step": 6990
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 1.8716,
"step": 7000
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 2.745,
"step": 7010
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 2.8138,
"step": 7020
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 2.6609,
"step": 7030
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 2.5277,
"step": 7040
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 2.0266,
"step": 7050
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 2.8222,
"step": 7060
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 2.7629,
"step": 7070
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 2.6332,
"step": 7080
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 2.5151,
"step": 7090
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 1.9864,
"step": 7100
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 2.6369,
"step": 7110
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 2.5675,
"step": 7120
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 2.4732,
"step": 7130
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 2.426,
"step": 7140
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 1.9873,
"step": 7150
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 2.6666,
"step": 7160
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 2.5453,
"step": 7170
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 2.4866,
"step": 7180
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 2.4204,
"step": 7190
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 1.9865,
"step": 7200
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 2.5969,
"step": 7210
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 2.5248,
"step": 7220
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 2.4691,
"step": 7230
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 2.4653,
"step": 7240
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 1.9409,
"step": 7250
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 2.4511,
"step": 7260
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 2.5139,
"step": 7270
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 2.4236,
"step": 7280
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 2.3505,
"step": 7290
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 1.9038,
"step": 7300
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 2.4344,
"step": 7310
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 2.3629,
"step": 7320
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 2.2656,
"step": 7330
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 2.1788,
"step": 7340
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 1.54,
"step": 7350
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 2.1208,
"step": 7360
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 2.1735,
"step": 7370
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 2.2392,
"step": 7380
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 2.2141,
"step": 7390
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 1.7179,
"step": 7400
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 2.2849,
"step": 7410
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 2.277,
"step": 7420
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 2.27,
"step": 7430
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 2.2229,
"step": 7440
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 1.712,
"step": 7450
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 2.1984,
"step": 7460
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 2.2542,
"step": 7470
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 2.3139,
"step": 7480
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 2.2685,
"step": 7490
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 1.6901,
"step": 7500
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 2.1149,
"step": 7510
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 2.0516,
"step": 7520
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 2.0119,
"step": 7530
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 2.0099,
"step": 7540
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 1.5823,
"step": 7550
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 1.9658,
"step": 7560
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 1.9999,
"step": 7570
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 2.0033,
"step": 7580
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 1.9746,
"step": 7590
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 1.5731,
"step": 7600
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 1.9657,
"step": 7610
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 1.9967,
"step": 7620
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 1.9404,
"step": 7630
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 1.8483,
"step": 7640
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 1.4618,
"step": 7650
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 1.7986,
"step": 7660
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 1.8033,
"step": 7670
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 1.8471,
"step": 7680
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 1.8007,
"step": 7690
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 1.3605,
"step": 7700
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 1.6258,
"step": 7710
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 1.733,
"step": 7720
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 1.812,
"step": 7730
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 1.7575,
"step": 7740
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 1.3581,
"step": 7750
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 1.6633,
"step": 7760
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 1.6169,
"step": 7770
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 1.7135,
"step": 7780
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 1.7109,
"step": 7790
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 1.407,
"step": 7800
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 1.641,
"step": 7810
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 1.6593,
"step": 7820
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 1.7338,
"step": 7830
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 1.7325,
"step": 7840
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 1.3272,
"step": 7850
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 1.6207,
"step": 7860
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 1.6328,
"step": 7870
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 1.6865,
"step": 7880
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 1.6119,
"step": 7890
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 1.3014,
"step": 7900
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 1.5949,
"step": 7910
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 1.5938,
"step": 7920
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 1.6891,
"step": 7930
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 1.6699,
"step": 7940
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 1.3358,
"step": 7950
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 1.5619,
"step": 7960
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 1.5884,
"step": 7970
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 1.634,
"step": 7980
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 1.6221,
"step": 7990
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 1.3082,
"step": 8000
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 1.5455,
"step": 8010
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 1.5425,
"step": 8020
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 1.6359,
"step": 8030
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 1.6494,
"step": 8040
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 1.2916,
"step": 8050
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 1.4993,
"step": 8060
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 1.5583,
"step": 8070
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 1.5923,
"step": 8080
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 1.6308,
"step": 8090
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 1.3073,
"step": 8100
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 1.5363,
"step": 8110
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 1.6257,
"step": 8120
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 1.6471,
"step": 8130
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 1.6669,
"step": 8140
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 1.3195,
"step": 8150
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 1.5359,
"step": 8160
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 1.5833,
"step": 8170
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 1.6571,
"step": 8180
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 1.6456,
"step": 8190
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 1.3148,
"step": 8200
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 1.5396,
"step": 8210
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 1.5219,
"step": 8220
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 1.6478,
"step": 8230
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 1.6146,
"step": 8240
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 1.284,
"step": 8250
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 1.5497,
"step": 8260
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 1.5858,
"step": 8270
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 1.6964,
"step": 8280
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 1.6518,
"step": 8290
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 1.4095,
"step": 8300
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 1.6811,
"step": 8310
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 1.7725,
"step": 8320
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 2.9559,
"step": 8330
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 4.9749,
"step": 8340
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 4.138,
"step": 8350
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 6.1766,
"step": 8360
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 5.5577,
"step": 8370
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 4.946,
"step": 8380
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 4.2825,
"step": 8390
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 3.2138,
"step": 8400
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 4.8699,
"step": 8410
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 4.3577,
"step": 8420
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 4.5203,
"step": 8430
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 4.2161,
"step": 8440
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 2.6815,
"step": 8450
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 3.572,
"step": 8460
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 3.7959,
"step": 8470
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 4.3023,
"step": 8480
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 4.6155,
"step": 8490
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 3.4514,
"step": 8500
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 4.8087,
"step": 8510
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 4.5272,
"step": 8520
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 4.1084,
"step": 8530
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 3.6305,
"step": 8540
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 2.5536,
"step": 8550
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 4.4515,
"step": 8560
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 3.4525,
"step": 8570
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 3.5976,
"step": 8580
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 3.4597,
"step": 8590
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 3.1653,
"step": 8600
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 4.5532,
"step": 8610
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 4.1625,
"step": 8620
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 4.3235,
"step": 8630
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 4.1094,
"step": 8640
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 3.1124,
"step": 8650
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 4.5213,
"step": 8660
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 4.492,
"step": 8670
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 4.3158,
"step": 8680
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 3.9548,
"step": 8690
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 3.7034,
"step": 8700
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 3.9088,
"step": 8710
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 3.9799,
"step": 8720
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 4.1253,
"step": 8730
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 3.9726,
"step": 8740
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 3.958,
"step": 8750
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 4.1978,
"step": 8760
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 4.9392,
"step": 8770
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 5.4691,
"step": 8780
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 4.2608,
"step": 8790
},
{
"epoch": 1.01,
"learning_rate": 0.0002,
"loss": 3.4531,
"step": 8800
}
],
"max_steps": 10000,
"num_train_epochs": 2,
"total_flos": 1.1443961596526592e+18,
"trial_name": null,
"trial_params": null
}