dica-llama2-13b-chat-hf / trainer_state.json
nhat117's picture
Upload folder using huggingface_hub
15066a9
raw
history blame
No virus
58.4 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.5298332229514119,
"global_step": 5500,
"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": 1.1938,
"step": 10
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.9654,
"step": 20
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.8956,
"step": 30
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.8471,
"step": 40
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.8339,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.8938,
"step": 60
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.8326,
"step": 70
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.8485,
"step": 80
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.8344,
"step": 90
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.7906,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.8281,
"step": 110
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.7754,
"step": 120
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.756,
"step": 130
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.7897,
"step": 140
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.7877,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7702,
"step": 160
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7708,
"step": 170
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7661,
"step": 180
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7778,
"step": 190
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7886,
"step": 200
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7938,
"step": 210
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7891,
"step": 220
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7624,
"step": 230
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7464,
"step": 240
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7609,
"step": 250
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.7524,
"step": 260
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.8035,
"step": 270
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.7652,
"step": 280
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.729,
"step": 290
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.74,
"step": 300
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.7707,
"step": 310
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.7492,
"step": 320
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.764,
"step": 330
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.8004,
"step": 340
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.7152,
"step": 350
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.7655,
"step": 360
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.7631,
"step": 370
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.7939,
"step": 380
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.7283,
"step": 390
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.7662,
"step": 400
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.763,
"step": 410
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.755,
"step": 420
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.7703,
"step": 430
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.7685,
"step": 440
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.73,
"step": 450
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.745,
"step": 460
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.7501,
"step": 470
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.7431,
"step": 480
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.7346,
"step": 490
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.7409,
"step": 500
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.7505,
"step": 510
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.7504,
"step": 520
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.7602,
"step": 530
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.7239,
"step": 540
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.7386,
"step": 550
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.7223,
"step": 560
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.7629,
"step": 570
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.7385,
"step": 580
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.7367,
"step": 590
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.7453,
"step": 600
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.7534,
"step": 610
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.7314,
"step": 620
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.75,
"step": 630
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.729,
"step": 640
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.7084,
"step": 650
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.7478,
"step": 660
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.7316,
"step": 670
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.74,
"step": 680
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.7033,
"step": 690
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.7491,
"step": 700
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.7395,
"step": 710
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.7168,
"step": 720
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.7537,
"step": 730
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.7021,
"step": 740
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.743,
"step": 750
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.7121,
"step": 760
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.7209,
"step": 770
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.7105,
"step": 780
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.7695,
"step": 790
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.7322,
"step": 800
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.709,
"step": 810
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.7186,
"step": 820
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.7244,
"step": 830
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.6892,
"step": 840
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.7405,
"step": 850
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.7345,
"step": 860
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.724,
"step": 870
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.7511,
"step": 880
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.7498,
"step": 890
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.726,
"step": 900
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.7419,
"step": 910
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.7117,
"step": 920
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.7193,
"step": 930
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.6877,
"step": 940
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.7084,
"step": 950
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.7087,
"step": 960
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.7354,
"step": 970
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.7164,
"step": 980
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.7076,
"step": 990
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.7031,
"step": 1000
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.7186,
"step": 1010
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.7311,
"step": 1020
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.7047,
"step": 1030
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.7054,
"step": 1040
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.7221,
"step": 1050
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.7158,
"step": 1060
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.6933,
"step": 1070
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.6992,
"step": 1080
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.6806,
"step": 1090
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.7053,
"step": 1100
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.7124,
"step": 1110
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.7011,
"step": 1120
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.7036,
"step": 1130
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.7195,
"step": 1140
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.706,
"step": 1150
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.6977,
"step": 1160
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.709,
"step": 1170
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.6983,
"step": 1180
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.7074,
"step": 1190
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.7214,
"step": 1200
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.7162,
"step": 1210
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.7396,
"step": 1220
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.7097,
"step": 1230
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.696,
"step": 1240
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.6924,
"step": 1250
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.7157,
"step": 1260
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.6844,
"step": 1270
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.7076,
"step": 1280
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.7204,
"step": 1290
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.7031,
"step": 1300
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.7215,
"step": 1310
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.7195,
"step": 1320
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.7149,
"step": 1330
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.6728,
"step": 1340
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.7129,
"step": 1350
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.6875,
"step": 1360
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.7116,
"step": 1370
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.7241,
"step": 1380
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.6876,
"step": 1390
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.7109,
"step": 1400
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.6953,
"step": 1410
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.7023,
"step": 1420
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.7145,
"step": 1430
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.6958,
"step": 1440
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.6825,
"step": 1450
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.6811,
"step": 1460
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.6851,
"step": 1470
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.6866,
"step": 1480
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.7145,
"step": 1490
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.672,
"step": 1500
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.7374,
"step": 1510
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.6459,
"step": 1520
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.6859,
"step": 1530
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.7014,
"step": 1540
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.6974,
"step": 1550
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.6764,
"step": 1560
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.6902,
"step": 1570
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.7007,
"step": 1580
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.6808,
"step": 1590
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.7018,
"step": 1600
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.6991,
"step": 1610
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.687,
"step": 1620
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.6815,
"step": 1630
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.6754,
"step": 1640
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.7208,
"step": 1650
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.6822,
"step": 1660
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.6964,
"step": 1670
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.6762,
"step": 1680
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.6847,
"step": 1690
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.698,
"step": 1700
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.6905,
"step": 1710
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.6765,
"step": 1720
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.662,
"step": 1730
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.6654,
"step": 1740
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.7016,
"step": 1750
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.6975,
"step": 1760
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.6733,
"step": 1770
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.698,
"step": 1780
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.6764,
"step": 1790
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.727,
"step": 1800
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.7089,
"step": 1810
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.6938,
"step": 1820
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.6931,
"step": 1830
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.7009,
"step": 1840
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.723,
"step": 1850
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.6701,
"step": 1860
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.6604,
"step": 1870
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.6837,
"step": 1880
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.6793,
"step": 1890
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.6826,
"step": 1900
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.6722,
"step": 1910
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.6859,
"step": 1920
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.6744,
"step": 1930
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.7037,
"step": 1940
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.6826,
"step": 1950
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.6663,
"step": 1960
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.6986,
"step": 1970
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.6701,
"step": 1980
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.6932,
"step": 1990
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.6902,
"step": 2000
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.6797,
"step": 2010
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.6729,
"step": 2020
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.6662,
"step": 2030
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.6977,
"step": 2040
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.6798,
"step": 2050
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.6549,
"step": 2060
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.7058,
"step": 2070
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.6936,
"step": 2080
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.6549,
"step": 2090
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.6969,
"step": 2100
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.6804,
"step": 2110
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.6737,
"step": 2120
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6796,
"step": 2130
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6777,
"step": 2140
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6447,
"step": 2150
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.689,
"step": 2160
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6698,
"step": 2170
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6826,
"step": 2180
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6801,
"step": 2190
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6362,
"step": 2200
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6868,
"step": 2210
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6581,
"step": 2220
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6759,
"step": 2230
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.6813,
"step": 2240
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.7001,
"step": 2250
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.6569,
"step": 2260
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.6986,
"step": 2270
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.6887,
"step": 2280
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.6711,
"step": 2290
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.666,
"step": 2300
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.6572,
"step": 2310
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.6839,
"step": 2320
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.6466,
"step": 2330
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.6916,
"step": 2340
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.6941,
"step": 2350
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.6797,
"step": 2360
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.6653,
"step": 2370
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.6759,
"step": 2380
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.6639,
"step": 2390
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.6566,
"step": 2400
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.6564,
"step": 2410
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.6667,
"step": 2420
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.7028,
"step": 2430
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.6602,
"step": 2440
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.6432,
"step": 2450
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.659,
"step": 2460
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.6631,
"step": 2470
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.6816,
"step": 2480
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.668,
"step": 2490
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.674,
"step": 2500
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.6973,
"step": 2510
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.6602,
"step": 2520
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.6489,
"step": 2530
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.6589,
"step": 2540
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.6595,
"step": 2550
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.647,
"step": 2560
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.671,
"step": 2570
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.6623,
"step": 2580
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.6985,
"step": 2590
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.684,
"step": 2600
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.6947,
"step": 2610
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.672,
"step": 2620
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.7097,
"step": 2630
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.6685,
"step": 2640
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.6572,
"step": 2650
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.6503,
"step": 2660
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.6454,
"step": 2670
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.6792,
"step": 2680
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.6816,
"step": 2690
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.6713,
"step": 2700
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.652,
"step": 2710
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.6558,
"step": 2720
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.6696,
"step": 2730
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.6561,
"step": 2740
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.6527,
"step": 2750
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.6705,
"step": 2760
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.6734,
"step": 2770
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.644,
"step": 2780
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.6615,
"step": 2790
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.683,
"step": 2800
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.658,
"step": 2810
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.651,
"step": 2820
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.6695,
"step": 2830
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.6792,
"step": 2840
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.6694,
"step": 2850
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6626,
"step": 2860
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6794,
"step": 2870
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6948,
"step": 2880
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6449,
"step": 2890
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6466,
"step": 2900
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6588,
"step": 2910
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6452,
"step": 2920
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6818,
"step": 2930
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6749,
"step": 2940
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6458,
"step": 2950
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6842,
"step": 2960
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.652,
"step": 2970
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6633,
"step": 2980
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6717,
"step": 2990
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6573,
"step": 3000
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6254,
"step": 3010
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6701,
"step": 3020
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6781,
"step": 3030
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6534,
"step": 3040
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6925,
"step": 3050
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6727,
"step": 3060
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.6548,
"step": 3070
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.6666,
"step": 3080
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.6757,
"step": 3090
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.662,
"step": 3100
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.6293,
"step": 3110
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.6786,
"step": 3120
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.6802,
"step": 3130
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.6698,
"step": 3140
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.678,
"step": 3150
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.646,
"step": 3160
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6518,
"step": 3170
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6808,
"step": 3180
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6446,
"step": 3190
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6747,
"step": 3200
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6614,
"step": 3210
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6425,
"step": 3220
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6641,
"step": 3230
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6637,
"step": 3240
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6187,
"step": 3250
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6592,
"step": 3260
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.624,
"step": 3270
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.6945,
"step": 3280
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.6901,
"step": 3290
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.6502,
"step": 3300
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.685,
"step": 3310
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.6845,
"step": 3320
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.6993,
"step": 3330
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.6537,
"step": 3340
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.665,
"step": 3350
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.6327,
"step": 3360
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.6417,
"step": 3370
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.6702,
"step": 3380
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.6551,
"step": 3390
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.6807,
"step": 3400
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.6569,
"step": 3410
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.6455,
"step": 3420
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.6242,
"step": 3430
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.6415,
"step": 3440
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.6395,
"step": 3450
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.6561,
"step": 3460
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.6292,
"step": 3470
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6347,
"step": 3480
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6301,
"step": 3490
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6155,
"step": 3500
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6512,
"step": 3510
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6486,
"step": 3520
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6402,
"step": 3530
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6545,
"step": 3540
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6461,
"step": 3550
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6368,
"step": 3560
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6517,
"step": 3570
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6372,
"step": 3580
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.6551,
"step": 3590
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.6424,
"step": 3600
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.627,
"step": 3610
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.6739,
"step": 3620
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.6344,
"step": 3630
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.6328,
"step": 3640
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.6435,
"step": 3650
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.6822,
"step": 3660
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.6217,
"step": 3670
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.618,
"step": 3680
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6594,
"step": 3690
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6587,
"step": 3700
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6671,
"step": 3710
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6364,
"step": 3720
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6266,
"step": 3730
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6544,
"step": 3740
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6445,
"step": 3750
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6166,
"step": 3760
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6462,
"step": 3770
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6353,
"step": 3780
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6381,
"step": 3790
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6474,
"step": 3800
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.641,
"step": 3810
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6506,
"step": 3820
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6301,
"step": 3830
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6408,
"step": 3840
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.65,
"step": 3850
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6383,
"step": 3860
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6274,
"step": 3870
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6417,
"step": 3880
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6476,
"step": 3890
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.6334,
"step": 3900
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.6893,
"step": 3910
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.6548,
"step": 3920
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.6401,
"step": 3930
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.5962,
"step": 3940
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.6673,
"step": 3950
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.6749,
"step": 3960
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.6276,
"step": 3970
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.6652,
"step": 3980
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.6337,
"step": 3990
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6439,
"step": 4000
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6533,
"step": 4010
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6455,
"step": 4020
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6745,
"step": 4030
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6481,
"step": 4040
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6416,
"step": 4050
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6227,
"step": 4060
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6307,
"step": 4070
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6634,
"step": 4080
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6468,
"step": 4090
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6532,
"step": 4100
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 0.6223,
"step": 4110
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 0.6036,
"step": 4120
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 0.6236,
"step": 4130
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 0.6307,
"step": 4140
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 0.6681,
"step": 4150
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 0.6631,
"step": 4160
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 0.6394,
"step": 4170
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 0.6262,
"step": 4180
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 0.6236,
"step": 4190
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 0.6047,
"step": 4200
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6444,
"step": 4210
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6514,
"step": 4220
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6529,
"step": 4230
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6567,
"step": 4240
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6536,
"step": 4250
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6426,
"step": 4260
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6396,
"step": 4270
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6398,
"step": 4280
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6604,
"step": 4290
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6288,
"step": 4300
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.6277,
"step": 4310
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.6434,
"step": 4320
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.6302,
"step": 4330
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.6415,
"step": 4340
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.6332,
"step": 4350
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.6346,
"step": 4360
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.6435,
"step": 4370
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.6205,
"step": 4380
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.6364,
"step": 4390
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.6464,
"step": 4400
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.6111,
"step": 4410
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6425,
"step": 4420
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6523,
"step": 4430
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6396,
"step": 4440
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6391,
"step": 4450
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6433,
"step": 4460
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6625,
"step": 4470
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6274,
"step": 4480
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6719,
"step": 4490
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6659,
"step": 4500
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6454,
"step": 4510
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.6417,
"step": 4520
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.6159,
"step": 4530
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.6186,
"step": 4540
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.6328,
"step": 4550
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.6494,
"step": 4560
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.6237,
"step": 4570
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.6427,
"step": 4580
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.642,
"step": 4590
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.6275,
"step": 4600
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.637,
"step": 4610
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6416,
"step": 4620
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6332,
"step": 4630
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6515,
"step": 4640
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6551,
"step": 4650
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6643,
"step": 4660
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6317,
"step": 4670
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6392,
"step": 4680
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6447,
"step": 4690
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6181,
"step": 4700
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6369,
"step": 4710
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6565,
"step": 4720
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6349,
"step": 4730
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6385,
"step": 4740
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.5835,
"step": 4750
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6195,
"step": 4760
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6286,
"step": 4770
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6153,
"step": 4780
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6517,
"step": 4790
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6004,
"step": 4800
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6299,
"step": 4810
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6369,
"step": 4820
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6403,
"step": 4830
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6443,
"step": 4840
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6139,
"step": 4850
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6228,
"step": 4860
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6212,
"step": 4870
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6039,
"step": 4880
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6269,
"step": 4890
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6302,
"step": 4900
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6392,
"step": 4910
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6261,
"step": 4920
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6201,
"step": 4930
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.6393,
"step": 4940
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.6174,
"step": 4950
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.6158,
"step": 4960
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.6317,
"step": 4970
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.6233,
"step": 4980
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.6733,
"step": 4990
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.6473,
"step": 5000
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.6512,
"step": 5010
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.6293,
"step": 5020
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.6291,
"step": 5030
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.6375,
"step": 5040
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.6087,
"step": 5050
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.6297,
"step": 5060
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.6124,
"step": 5070
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.6128,
"step": 5080
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.644,
"step": 5090
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.613,
"step": 5100
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.601,
"step": 5110
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.6441,
"step": 5120
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.6187,
"step": 5130
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6355,
"step": 5140
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6671,
"step": 5150
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6412,
"step": 5160
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6154,
"step": 5170
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6195,
"step": 5180
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6174,
"step": 5190
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6402,
"step": 5200
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6188,
"step": 5210
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6385,
"step": 5220
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6402,
"step": 5230
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6057,
"step": 5240
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.6428,
"step": 5250
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.6161,
"step": 5260
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.6552,
"step": 5270
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.6134,
"step": 5280
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.6543,
"step": 5290
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.6499,
"step": 5300
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.6544,
"step": 5310
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.6269,
"step": 5320
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.614,
"step": 5330
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.6233,
"step": 5340
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.612,
"step": 5350
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.6396,
"step": 5360
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.5731,
"step": 5370
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.6178,
"step": 5380
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.6245,
"step": 5390
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.6131,
"step": 5400
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.642,
"step": 5410
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.6237,
"step": 5420
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.5984,
"step": 5430
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.6262,
"step": 5440
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.6288,
"step": 5450
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.6178,
"step": 5460
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.6479,
"step": 5470
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.5907,
"step": 5480
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.6099,
"step": 5490
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.6218,
"step": 5500
}
],
"max_steps": 62280,
"num_train_epochs": 6,
"total_flos": 7.776659814923059e+17,
"trial_name": null,
"trial_params": null
}