whisper-small_zh-TW / trainer_state.json
ryL's picture
Upload 7 files
884eafb
raw
history blame contribute delete
No virus
36.9 kB
{
"best_metric": 39.35018050541516,
"best_model_checkpoint": "./whisper-small_zh-TW\\checkpoint-3000",
"epoch": 4.25531914893617,
"global_step": 3000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 2.0000000000000002e-07,
"loss": 5.2541,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 4.0000000000000003e-07,
"loss": 4.8117,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 6.000000000000001e-07,
"loss": 4.1744,
"step": 30
},
{
"epoch": 0.06,
"learning_rate": 8.000000000000001e-07,
"loss": 3.4216,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 1.0000000000000002e-06,
"loss": 2.9598,
"step": 50
},
{
"epoch": 0.09,
"learning_rate": 1.2000000000000002e-06,
"loss": 2.4197,
"step": 60
},
{
"epoch": 0.1,
"learning_rate": 1.4000000000000001e-06,
"loss": 2.0626,
"step": 70
},
{
"epoch": 0.11,
"learning_rate": 1.6000000000000001e-06,
"loss": 1.5578,
"step": 80
},
{
"epoch": 0.13,
"learning_rate": 1.8000000000000001e-06,
"loss": 1.0855,
"step": 90
},
{
"epoch": 0.14,
"learning_rate": 2.0000000000000003e-06,
"loss": 0.8664,
"step": 100
},
{
"epoch": 0.16,
"learning_rate": 2.2e-06,
"loss": 0.8396,
"step": 110
},
{
"epoch": 0.17,
"learning_rate": 2.4000000000000003e-06,
"loss": 0.7711,
"step": 120
},
{
"epoch": 0.18,
"learning_rate": 2.6e-06,
"loss": 0.7121,
"step": 130
},
{
"epoch": 0.2,
"learning_rate": 2.8000000000000003e-06,
"loss": 0.656,
"step": 140
},
{
"epoch": 0.21,
"learning_rate": 3e-06,
"loss": 0.6073,
"step": 150
},
{
"epoch": 0.23,
"learning_rate": 3.2000000000000003e-06,
"loss": 0.5268,
"step": 160
},
{
"epoch": 0.24,
"learning_rate": 3.4000000000000005e-06,
"loss": 0.4504,
"step": 170
},
{
"epoch": 0.26,
"learning_rate": 3.6000000000000003e-06,
"loss": 0.4194,
"step": 180
},
{
"epoch": 0.27,
"learning_rate": 3.8000000000000005e-06,
"loss": 0.3585,
"step": 190
},
{
"epoch": 0.28,
"learning_rate": 4.000000000000001e-06,
"loss": 0.3142,
"step": 200
},
{
"epoch": 0.3,
"learning_rate": 4.2000000000000004e-06,
"loss": 0.2462,
"step": 210
},
{
"epoch": 0.31,
"learning_rate": 4.4e-06,
"loss": 0.2657,
"step": 220
},
{
"epoch": 0.33,
"learning_rate": 4.600000000000001e-06,
"loss": 0.242,
"step": 230
},
{
"epoch": 0.34,
"learning_rate": 4.800000000000001e-06,
"loss": 0.2142,
"step": 240
},
{
"epoch": 0.35,
"learning_rate": 5e-06,
"loss": 0.2109,
"step": 250
},
{
"epoch": 0.37,
"learning_rate": 5.2e-06,
"loss": 0.2016,
"step": 260
},
{
"epoch": 0.38,
"learning_rate": 5.400000000000001e-06,
"loss": 0.2224,
"step": 270
},
{
"epoch": 0.4,
"learning_rate": 5.600000000000001e-06,
"loss": 0.2203,
"step": 280
},
{
"epoch": 0.41,
"learning_rate": 5.8e-06,
"loss": 0.1787,
"step": 290
},
{
"epoch": 0.43,
"learning_rate": 6e-06,
"loss": 0.2149,
"step": 300
},
{
"epoch": 0.44,
"learning_rate": 6.200000000000001e-06,
"loss": 0.1736,
"step": 310
},
{
"epoch": 0.45,
"learning_rate": 6.4000000000000006e-06,
"loss": 0.211,
"step": 320
},
{
"epoch": 0.47,
"learning_rate": 6.600000000000001e-06,
"loss": 0.2084,
"step": 330
},
{
"epoch": 0.48,
"learning_rate": 6.800000000000001e-06,
"loss": 0.1979,
"step": 340
},
{
"epoch": 0.5,
"learning_rate": 7e-06,
"loss": 0.2049,
"step": 350
},
{
"epoch": 0.51,
"learning_rate": 7.2000000000000005e-06,
"loss": 0.2031,
"step": 360
},
{
"epoch": 0.52,
"learning_rate": 7.4e-06,
"loss": 0.238,
"step": 370
},
{
"epoch": 0.54,
"learning_rate": 7.600000000000001e-06,
"loss": 0.2313,
"step": 380
},
{
"epoch": 0.55,
"learning_rate": 7.800000000000002e-06,
"loss": 0.1935,
"step": 390
},
{
"epoch": 0.57,
"learning_rate": 8.000000000000001e-06,
"loss": 0.2256,
"step": 400
},
{
"epoch": 0.58,
"learning_rate": 8.2e-06,
"loss": 0.1949,
"step": 410
},
{
"epoch": 0.6,
"learning_rate": 8.400000000000001e-06,
"loss": 0.1904,
"step": 420
},
{
"epoch": 0.61,
"learning_rate": 8.6e-06,
"loss": 0.2153,
"step": 430
},
{
"epoch": 0.62,
"learning_rate": 8.8e-06,
"loss": 0.178,
"step": 440
},
{
"epoch": 0.64,
"learning_rate": 9e-06,
"loss": 0.2113,
"step": 450
},
{
"epoch": 0.65,
"learning_rate": 9.200000000000002e-06,
"loss": 0.191,
"step": 460
},
{
"epoch": 0.67,
"learning_rate": 9.4e-06,
"loss": 0.2046,
"step": 470
},
{
"epoch": 0.68,
"learning_rate": 9.600000000000001e-06,
"loss": 0.2586,
"step": 480
},
{
"epoch": 0.7,
"learning_rate": 9.800000000000001e-06,
"loss": 0.1938,
"step": 490
},
{
"epoch": 0.71,
"learning_rate": 1e-05,
"loss": 0.2502,
"step": 500
},
{
"epoch": 0.72,
"learning_rate": 9.971428571428571e-06,
"loss": 0.2525,
"step": 510
},
{
"epoch": 0.74,
"learning_rate": 9.942857142857145e-06,
"loss": 0.1873,
"step": 520
},
{
"epoch": 0.75,
"learning_rate": 9.914285714285715e-06,
"loss": 0.2259,
"step": 530
},
{
"epoch": 0.77,
"learning_rate": 9.885714285714287e-06,
"loss": 0.1972,
"step": 540
},
{
"epoch": 0.78,
"learning_rate": 9.857142857142859e-06,
"loss": 0.1955,
"step": 550
},
{
"epoch": 0.79,
"learning_rate": 9.828571428571429e-06,
"loss": 0.1988,
"step": 560
},
{
"epoch": 0.81,
"learning_rate": 9.800000000000001e-06,
"loss": 0.2068,
"step": 570
},
{
"epoch": 0.82,
"learning_rate": 9.771428571428571e-06,
"loss": 0.2075,
"step": 580
},
{
"epoch": 0.84,
"learning_rate": 9.742857142857143e-06,
"loss": 0.2048,
"step": 590
},
{
"epoch": 0.85,
"learning_rate": 9.714285714285715e-06,
"loss": 0.2274,
"step": 600
},
{
"epoch": 0.87,
"learning_rate": 9.685714285714285e-06,
"loss": 0.2107,
"step": 610
},
{
"epoch": 0.88,
"learning_rate": 9.657142857142859e-06,
"loss": 0.2081,
"step": 620
},
{
"epoch": 0.89,
"learning_rate": 9.62857142857143e-06,
"loss": 0.2149,
"step": 630
},
{
"epoch": 0.91,
"learning_rate": 9.600000000000001e-06,
"loss": 0.1839,
"step": 640
},
{
"epoch": 0.92,
"learning_rate": 9.571428571428573e-06,
"loss": 0.2318,
"step": 650
},
{
"epoch": 0.94,
"learning_rate": 9.542857142857143e-06,
"loss": 0.1925,
"step": 660
},
{
"epoch": 0.95,
"learning_rate": 9.514285714285715e-06,
"loss": 0.1889,
"step": 670
},
{
"epoch": 0.96,
"learning_rate": 9.485714285714287e-06,
"loss": 0.1861,
"step": 680
},
{
"epoch": 0.98,
"learning_rate": 9.457142857142858e-06,
"loss": 0.267,
"step": 690
},
{
"epoch": 0.99,
"learning_rate": 9.42857142857143e-06,
"loss": 0.1681,
"step": 700
},
{
"epoch": 1.01,
"learning_rate": 9.4e-06,
"loss": 0.1575,
"step": 710
},
{
"epoch": 1.02,
"learning_rate": 9.371428571428572e-06,
"loss": 0.0999,
"step": 720
},
{
"epoch": 1.04,
"learning_rate": 9.342857142857144e-06,
"loss": 0.096,
"step": 730
},
{
"epoch": 1.05,
"learning_rate": 9.314285714285714e-06,
"loss": 0.0937,
"step": 740
},
{
"epoch": 1.06,
"learning_rate": 9.285714285714288e-06,
"loss": 0.0834,
"step": 750
},
{
"epoch": 1.08,
"learning_rate": 9.257142857142858e-06,
"loss": 0.1019,
"step": 760
},
{
"epoch": 1.09,
"learning_rate": 9.22857142857143e-06,
"loss": 0.0707,
"step": 770
},
{
"epoch": 1.11,
"learning_rate": 9.200000000000002e-06,
"loss": 0.0834,
"step": 780
},
{
"epoch": 1.12,
"learning_rate": 9.171428571428572e-06,
"loss": 0.0866,
"step": 790
},
{
"epoch": 1.13,
"learning_rate": 9.142857142857144e-06,
"loss": 0.1191,
"step": 800
},
{
"epoch": 1.15,
"learning_rate": 9.114285714285716e-06,
"loss": 0.0959,
"step": 810
},
{
"epoch": 1.16,
"learning_rate": 9.085714285714286e-06,
"loss": 0.0876,
"step": 820
},
{
"epoch": 1.18,
"learning_rate": 9.057142857142858e-06,
"loss": 0.1025,
"step": 830
},
{
"epoch": 1.19,
"learning_rate": 9.028571428571428e-06,
"loss": 0.1038,
"step": 840
},
{
"epoch": 1.21,
"learning_rate": 9e-06,
"loss": 0.0922,
"step": 850
},
{
"epoch": 1.22,
"learning_rate": 8.971428571428572e-06,
"loss": 0.0837,
"step": 860
},
{
"epoch": 1.23,
"learning_rate": 8.942857142857142e-06,
"loss": 0.1068,
"step": 870
},
{
"epoch": 1.25,
"learning_rate": 8.914285714285716e-06,
"loss": 0.0699,
"step": 880
},
{
"epoch": 1.26,
"learning_rate": 8.885714285714286e-06,
"loss": 0.0909,
"step": 890
},
{
"epoch": 1.28,
"learning_rate": 8.857142857142858e-06,
"loss": 0.1076,
"step": 900
},
{
"epoch": 1.29,
"learning_rate": 8.82857142857143e-06,
"loss": 0.0674,
"step": 910
},
{
"epoch": 1.3,
"learning_rate": 8.8e-06,
"loss": 0.0831,
"step": 920
},
{
"epoch": 1.32,
"learning_rate": 8.771428571428572e-06,
"loss": 0.0834,
"step": 930
},
{
"epoch": 1.33,
"learning_rate": 8.742857142857144e-06,
"loss": 0.1132,
"step": 940
},
{
"epoch": 1.35,
"learning_rate": 8.714285714285715e-06,
"loss": 0.0942,
"step": 950
},
{
"epoch": 1.36,
"learning_rate": 8.685714285714287e-06,
"loss": 0.0989,
"step": 960
},
{
"epoch": 1.38,
"learning_rate": 8.657142857142858e-06,
"loss": 0.1139,
"step": 970
},
{
"epoch": 1.39,
"learning_rate": 8.628571428571429e-06,
"loss": 0.0827,
"step": 980
},
{
"epoch": 1.4,
"learning_rate": 8.6e-06,
"loss": 0.1014,
"step": 990
},
{
"epoch": 1.42,
"learning_rate": 8.571428571428571e-06,
"loss": 0.1171,
"step": 1000
},
{
"epoch": 1.42,
"eval_loss": 0.19091078639030457,
"eval_runtime": 1642.5978,
"eval_samples_per_second": 2.867,
"eval_steps_per_second": 0.359,
"eval_wer": 42.21703121681885,
"step": 1000
},
{
"epoch": 1.43,
"learning_rate": 8.542857142857145e-06,
"loss": 0.0836,
"step": 1010
},
{
"epoch": 1.45,
"learning_rate": 8.514285714285715e-06,
"loss": 0.0781,
"step": 1020
},
{
"epoch": 1.46,
"learning_rate": 8.485714285714287e-06,
"loss": 0.0944,
"step": 1030
},
{
"epoch": 1.48,
"learning_rate": 8.457142857142859e-06,
"loss": 0.0954,
"step": 1040
},
{
"epoch": 1.49,
"learning_rate": 8.428571428571429e-06,
"loss": 0.0925,
"step": 1050
},
{
"epoch": 1.5,
"learning_rate": 8.400000000000001e-06,
"loss": 0.0811,
"step": 1060
},
{
"epoch": 1.52,
"learning_rate": 8.371428571428573e-06,
"loss": 0.093,
"step": 1070
},
{
"epoch": 1.53,
"learning_rate": 8.342857142857143e-06,
"loss": 0.1137,
"step": 1080
},
{
"epoch": 1.55,
"learning_rate": 8.314285714285715e-06,
"loss": 0.0916,
"step": 1090
},
{
"epoch": 1.56,
"learning_rate": 8.285714285714287e-06,
"loss": 0.0913,
"step": 1100
},
{
"epoch": 1.57,
"learning_rate": 8.257142857142857e-06,
"loss": 0.0839,
"step": 1110
},
{
"epoch": 1.59,
"learning_rate": 8.22857142857143e-06,
"loss": 0.0998,
"step": 1120
},
{
"epoch": 1.6,
"learning_rate": 8.2e-06,
"loss": 0.09,
"step": 1130
},
{
"epoch": 1.62,
"learning_rate": 8.171428571428573e-06,
"loss": 0.1151,
"step": 1140
},
{
"epoch": 1.63,
"learning_rate": 8.142857142857143e-06,
"loss": 0.1007,
"step": 1150
},
{
"epoch": 1.65,
"learning_rate": 8.114285714285715e-06,
"loss": 0.0727,
"step": 1160
},
{
"epoch": 1.66,
"learning_rate": 8.085714285714287e-06,
"loss": 0.0875,
"step": 1170
},
{
"epoch": 1.67,
"learning_rate": 8.057142857142857e-06,
"loss": 0.1126,
"step": 1180
},
{
"epoch": 1.69,
"learning_rate": 8.02857142857143e-06,
"loss": 0.0855,
"step": 1190
},
{
"epoch": 1.7,
"learning_rate": 8.000000000000001e-06,
"loss": 0.0783,
"step": 1200
},
{
"epoch": 1.72,
"learning_rate": 7.971428571428572e-06,
"loss": 0.1063,
"step": 1210
},
{
"epoch": 1.73,
"learning_rate": 7.942857142857144e-06,
"loss": 0.1092,
"step": 1220
},
{
"epoch": 1.74,
"learning_rate": 7.914285714285715e-06,
"loss": 0.0927,
"step": 1230
},
{
"epoch": 1.76,
"learning_rate": 7.885714285714286e-06,
"loss": 0.0916,
"step": 1240
},
{
"epoch": 1.77,
"learning_rate": 7.857142857142858e-06,
"loss": 0.1167,
"step": 1250
},
{
"epoch": 1.79,
"learning_rate": 7.828571428571428e-06,
"loss": 0.0825,
"step": 1260
},
{
"epoch": 1.8,
"learning_rate": 7.800000000000002e-06,
"loss": 0.0906,
"step": 1270
},
{
"epoch": 1.82,
"learning_rate": 7.771428571428572e-06,
"loss": 0.0923,
"step": 1280
},
{
"epoch": 1.83,
"learning_rate": 7.742857142857144e-06,
"loss": 0.1032,
"step": 1290
},
{
"epoch": 1.84,
"learning_rate": 7.714285714285716e-06,
"loss": 0.1005,
"step": 1300
},
{
"epoch": 1.86,
"learning_rate": 7.685714285714286e-06,
"loss": 0.0721,
"step": 1310
},
{
"epoch": 1.87,
"learning_rate": 7.657142857142858e-06,
"loss": 0.0871,
"step": 1320
},
{
"epoch": 1.89,
"learning_rate": 7.62857142857143e-06,
"loss": 0.0963,
"step": 1330
},
{
"epoch": 1.9,
"learning_rate": 7.600000000000001e-06,
"loss": 0.1011,
"step": 1340
},
{
"epoch": 1.91,
"learning_rate": 7.571428571428572e-06,
"loss": 0.0809,
"step": 1350
},
{
"epoch": 1.93,
"learning_rate": 7.542857142857144e-06,
"loss": 0.1048,
"step": 1360
},
{
"epoch": 1.94,
"learning_rate": 7.514285714285715e-06,
"loss": 0.0981,
"step": 1370
},
{
"epoch": 1.96,
"learning_rate": 7.485714285714286e-06,
"loss": 0.0949,
"step": 1380
},
{
"epoch": 1.97,
"learning_rate": 7.457142857142857e-06,
"loss": 0.088,
"step": 1390
},
{
"epoch": 1.99,
"learning_rate": 7.428571428571429e-06,
"loss": 0.0983,
"step": 1400
},
{
"epoch": 2.0,
"learning_rate": 7.4e-06,
"loss": 0.0768,
"step": 1410
},
{
"epoch": 2.01,
"learning_rate": 7.371428571428571e-06,
"loss": 0.043,
"step": 1420
},
{
"epoch": 2.03,
"learning_rate": 7.342857142857144e-06,
"loss": 0.0284,
"step": 1430
},
{
"epoch": 2.04,
"learning_rate": 7.314285714285715e-06,
"loss": 0.032,
"step": 1440
},
{
"epoch": 2.06,
"learning_rate": 7.285714285714286e-06,
"loss": 0.0204,
"step": 1450
},
{
"epoch": 2.07,
"learning_rate": 7.257142857142858e-06,
"loss": 0.0291,
"step": 1460
},
{
"epoch": 2.09,
"learning_rate": 7.2285714285714294e-06,
"loss": 0.0275,
"step": 1470
},
{
"epoch": 2.1,
"learning_rate": 7.2000000000000005e-06,
"loss": 0.0448,
"step": 1480
},
{
"epoch": 2.11,
"learning_rate": 7.1714285714285725e-06,
"loss": 0.0385,
"step": 1490
},
{
"epoch": 2.13,
"learning_rate": 7.1428571428571436e-06,
"loss": 0.0237,
"step": 1500
},
{
"epoch": 2.14,
"learning_rate": 7.114285714285715e-06,
"loss": 0.0217,
"step": 1510
},
{
"epoch": 2.16,
"learning_rate": 7.085714285714286e-06,
"loss": 0.0343,
"step": 1520
},
{
"epoch": 2.17,
"learning_rate": 7.057142857142858e-06,
"loss": 0.0328,
"step": 1530
},
{
"epoch": 2.18,
"learning_rate": 7.028571428571429e-06,
"loss": 0.0294,
"step": 1540
},
{
"epoch": 2.2,
"learning_rate": 7e-06,
"loss": 0.031,
"step": 1550
},
{
"epoch": 2.21,
"learning_rate": 6.971428571428573e-06,
"loss": 0.0224,
"step": 1560
},
{
"epoch": 2.23,
"learning_rate": 6.942857142857144e-06,
"loss": 0.0326,
"step": 1570
},
{
"epoch": 2.24,
"learning_rate": 6.914285714285715e-06,
"loss": 0.0331,
"step": 1580
},
{
"epoch": 2.26,
"learning_rate": 6.885714285714287e-06,
"loss": 0.0336,
"step": 1590
},
{
"epoch": 2.27,
"learning_rate": 6.857142857142858e-06,
"loss": 0.0313,
"step": 1600
},
{
"epoch": 2.28,
"learning_rate": 6.828571428571429e-06,
"loss": 0.0223,
"step": 1610
},
{
"epoch": 2.3,
"learning_rate": 6.800000000000001e-06,
"loss": 0.034,
"step": 1620
},
{
"epoch": 2.31,
"learning_rate": 6.771428571428572e-06,
"loss": 0.03,
"step": 1630
},
{
"epoch": 2.33,
"learning_rate": 6.742857142857143e-06,
"loss": 0.0283,
"step": 1640
},
{
"epoch": 2.34,
"learning_rate": 6.714285714285714e-06,
"loss": 0.0446,
"step": 1650
},
{
"epoch": 2.35,
"learning_rate": 6.685714285714286e-06,
"loss": 0.0354,
"step": 1660
},
{
"epoch": 2.37,
"learning_rate": 6.657142857142857e-06,
"loss": 0.0267,
"step": 1670
},
{
"epoch": 2.38,
"learning_rate": 6.628571428571428e-06,
"loss": 0.0351,
"step": 1680
},
{
"epoch": 2.4,
"learning_rate": 6.600000000000001e-06,
"loss": 0.0271,
"step": 1690
},
{
"epoch": 2.41,
"learning_rate": 6.571428571428572e-06,
"loss": 0.0291,
"step": 1700
},
{
"epoch": 2.43,
"learning_rate": 6.542857142857143e-06,
"loss": 0.0256,
"step": 1710
},
{
"epoch": 2.44,
"learning_rate": 6.514285714285715e-06,
"loss": 0.0219,
"step": 1720
},
{
"epoch": 2.45,
"learning_rate": 6.485714285714286e-06,
"loss": 0.0232,
"step": 1730
},
{
"epoch": 2.47,
"learning_rate": 6.4571428571428575e-06,
"loss": 0.0271,
"step": 1740
},
{
"epoch": 2.48,
"learning_rate": 6.4285714285714295e-06,
"loss": 0.033,
"step": 1750
},
{
"epoch": 2.5,
"learning_rate": 6.4000000000000006e-06,
"loss": 0.0454,
"step": 1760
},
{
"epoch": 2.51,
"learning_rate": 6.371428571428572e-06,
"loss": 0.0348,
"step": 1770
},
{
"epoch": 2.52,
"learning_rate": 6.342857142857143e-06,
"loss": 0.0217,
"step": 1780
},
{
"epoch": 2.54,
"learning_rate": 6.314285714285715e-06,
"loss": 0.0274,
"step": 1790
},
{
"epoch": 2.55,
"learning_rate": 6.285714285714286e-06,
"loss": 0.0202,
"step": 1800
},
{
"epoch": 2.57,
"learning_rate": 6.257142857142857e-06,
"loss": 0.0369,
"step": 1810
},
{
"epoch": 2.58,
"learning_rate": 6.22857142857143e-06,
"loss": 0.0284,
"step": 1820
},
{
"epoch": 2.6,
"learning_rate": 6.200000000000001e-06,
"loss": 0.0415,
"step": 1830
},
{
"epoch": 2.61,
"learning_rate": 6.171428571428572e-06,
"loss": 0.033,
"step": 1840
},
{
"epoch": 2.62,
"learning_rate": 6.142857142857144e-06,
"loss": 0.0316,
"step": 1850
},
{
"epoch": 2.64,
"learning_rate": 6.114285714285715e-06,
"loss": 0.0405,
"step": 1860
},
{
"epoch": 2.65,
"learning_rate": 6.085714285714286e-06,
"loss": 0.0262,
"step": 1870
},
{
"epoch": 2.67,
"learning_rate": 6.057142857142858e-06,
"loss": 0.0428,
"step": 1880
},
{
"epoch": 2.68,
"learning_rate": 6.028571428571429e-06,
"loss": 0.0273,
"step": 1890
},
{
"epoch": 2.7,
"learning_rate": 6e-06,
"loss": 0.0257,
"step": 1900
},
{
"epoch": 2.71,
"learning_rate": 5.971428571428572e-06,
"loss": 0.0301,
"step": 1910
},
{
"epoch": 2.72,
"learning_rate": 5.942857142857143e-06,
"loss": 0.0314,
"step": 1920
},
{
"epoch": 2.74,
"learning_rate": 5.914285714285714e-06,
"loss": 0.0321,
"step": 1930
},
{
"epoch": 2.75,
"learning_rate": 5.885714285714285e-06,
"loss": 0.0284,
"step": 1940
},
{
"epoch": 2.77,
"learning_rate": 5.857142857142858e-06,
"loss": 0.0218,
"step": 1950
},
{
"epoch": 2.78,
"learning_rate": 5.828571428571429e-06,
"loss": 0.0363,
"step": 1960
},
{
"epoch": 2.79,
"learning_rate": 5.8e-06,
"loss": 0.0335,
"step": 1970
},
{
"epoch": 2.81,
"learning_rate": 5.771428571428572e-06,
"loss": 0.0246,
"step": 1980
},
{
"epoch": 2.82,
"learning_rate": 5.742857142857143e-06,
"loss": 0.0278,
"step": 1990
},
{
"epoch": 2.84,
"learning_rate": 5.7142857142857145e-06,
"loss": 0.0298,
"step": 2000
},
{
"epoch": 2.84,
"eval_loss": 0.1898459643125534,
"eval_runtime": 1563.6489,
"eval_samples_per_second": 3.012,
"eval_steps_per_second": 0.377,
"eval_wer": 40.22085368443406,
"step": 2000
},
{
"epoch": 2.85,
"learning_rate": 5.6857142857142865e-06,
"loss": 0.0283,
"step": 2010
},
{
"epoch": 2.87,
"learning_rate": 5.6571428571428576e-06,
"loss": 0.033,
"step": 2020
},
{
"epoch": 2.88,
"learning_rate": 5.628571428571429e-06,
"loss": 0.0257,
"step": 2030
},
{
"epoch": 2.89,
"learning_rate": 5.600000000000001e-06,
"loss": 0.0268,
"step": 2040
},
{
"epoch": 2.91,
"learning_rate": 5.571428571428572e-06,
"loss": 0.0312,
"step": 2050
},
{
"epoch": 2.92,
"learning_rate": 5.542857142857143e-06,
"loss": 0.0301,
"step": 2060
},
{
"epoch": 2.94,
"learning_rate": 5.514285714285714e-06,
"loss": 0.0218,
"step": 2070
},
{
"epoch": 2.95,
"learning_rate": 5.485714285714287e-06,
"loss": 0.026,
"step": 2080
},
{
"epoch": 2.96,
"learning_rate": 5.457142857142858e-06,
"loss": 0.0214,
"step": 2090
},
{
"epoch": 2.98,
"learning_rate": 5.428571428571429e-06,
"loss": 0.0221,
"step": 2100
},
{
"epoch": 2.99,
"learning_rate": 5.400000000000001e-06,
"loss": 0.0363,
"step": 2110
},
{
"epoch": 3.01,
"learning_rate": 5.371428571428572e-06,
"loss": 0.017,
"step": 2120
},
{
"epoch": 3.02,
"learning_rate": 5.342857142857143e-06,
"loss": 0.0143,
"step": 2130
},
{
"epoch": 3.04,
"learning_rate": 5.314285714285715e-06,
"loss": 0.0128,
"step": 2140
},
{
"epoch": 3.05,
"learning_rate": 5.285714285714286e-06,
"loss": 0.0091,
"step": 2150
},
{
"epoch": 3.06,
"learning_rate": 5.257142857142857e-06,
"loss": 0.017,
"step": 2160
},
{
"epoch": 3.08,
"learning_rate": 5.22857142857143e-06,
"loss": 0.0138,
"step": 2170
},
{
"epoch": 3.09,
"learning_rate": 5.2e-06,
"loss": 0.0053,
"step": 2180
},
{
"epoch": 3.11,
"learning_rate": 5.171428571428571e-06,
"loss": 0.0096,
"step": 2190
},
{
"epoch": 3.12,
"learning_rate": 5.142857142857142e-06,
"loss": 0.0116,
"step": 2200
},
{
"epoch": 3.13,
"learning_rate": 5.114285714285715e-06,
"loss": 0.0072,
"step": 2210
},
{
"epoch": 3.15,
"learning_rate": 5.085714285714286e-06,
"loss": 0.0086,
"step": 2220
},
{
"epoch": 3.16,
"learning_rate": 5.057142857142857e-06,
"loss": 0.0078,
"step": 2230
},
{
"epoch": 3.18,
"learning_rate": 5.028571428571429e-06,
"loss": 0.0138,
"step": 2240
},
{
"epoch": 3.19,
"learning_rate": 5e-06,
"loss": 0.0114,
"step": 2250
},
{
"epoch": 3.21,
"learning_rate": 4.971428571428572e-06,
"loss": 0.0152,
"step": 2260
},
{
"epoch": 3.22,
"learning_rate": 4.9428571428571435e-06,
"loss": 0.0129,
"step": 2270
},
{
"epoch": 3.23,
"learning_rate": 4.9142857142857145e-06,
"loss": 0.0093,
"step": 2280
},
{
"epoch": 3.25,
"learning_rate": 4.885714285714286e-06,
"loss": 0.0087,
"step": 2290
},
{
"epoch": 3.26,
"learning_rate": 4.857142857142858e-06,
"loss": 0.0085,
"step": 2300
},
{
"epoch": 3.28,
"learning_rate": 4.8285714285714295e-06,
"loss": 0.0121,
"step": 2310
},
{
"epoch": 3.29,
"learning_rate": 4.800000000000001e-06,
"loss": 0.01,
"step": 2320
},
{
"epoch": 3.3,
"learning_rate": 4.771428571428572e-06,
"loss": 0.0091,
"step": 2330
},
{
"epoch": 3.32,
"learning_rate": 4.742857142857144e-06,
"loss": 0.0122,
"step": 2340
},
{
"epoch": 3.33,
"learning_rate": 4.714285714285715e-06,
"loss": 0.0078,
"step": 2350
},
{
"epoch": 3.35,
"learning_rate": 4.685714285714286e-06,
"loss": 0.0115,
"step": 2360
},
{
"epoch": 3.36,
"learning_rate": 4.657142857142857e-06,
"loss": 0.0107,
"step": 2370
},
{
"epoch": 3.38,
"learning_rate": 4.628571428571429e-06,
"loss": 0.0095,
"step": 2380
},
{
"epoch": 3.39,
"learning_rate": 4.600000000000001e-06,
"loss": 0.0098,
"step": 2390
},
{
"epoch": 3.4,
"learning_rate": 4.571428571428572e-06,
"loss": 0.0125,
"step": 2400
},
{
"epoch": 3.42,
"learning_rate": 4.542857142857143e-06,
"loss": 0.0075,
"step": 2410
},
{
"epoch": 3.43,
"learning_rate": 4.514285714285714e-06,
"loss": 0.0135,
"step": 2420
},
{
"epoch": 3.45,
"learning_rate": 4.485714285714286e-06,
"loss": 0.0075,
"step": 2430
},
{
"epoch": 3.46,
"learning_rate": 4.457142857142858e-06,
"loss": 0.012,
"step": 2440
},
{
"epoch": 3.48,
"learning_rate": 4.428571428571429e-06,
"loss": 0.0185,
"step": 2450
},
{
"epoch": 3.49,
"learning_rate": 4.4e-06,
"loss": 0.0062,
"step": 2460
},
{
"epoch": 3.5,
"learning_rate": 4.371428571428572e-06,
"loss": 0.01,
"step": 2470
},
{
"epoch": 3.52,
"learning_rate": 4.342857142857143e-06,
"loss": 0.0181,
"step": 2480
},
{
"epoch": 3.53,
"learning_rate": 4.314285714285714e-06,
"loss": 0.0111,
"step": 2490
},
{
"epoch": 3.55,
"learning_rate": 4.2857142857142855e-06,
"loss": 0.0058,
"step": 2500
},
{
"epoch": 3.56,
"learning_rate": 4.257142857142857e-06,
"loss": 0.0079,
"step": 2510
},
{
"epoch": 3.57,
"learning_rate": 4.228571428571429e-06,
"loss": 0.0111,
"step": 2520
},
{
"epoch": 3.59,
"learning_rate": 4.2000000000000004e-06,
"loss": 0.011,
"step": 2530
},
{
"epoch": 3.6,
"learning_rate": 4.1714285714285715e-06,
"loss": 0.0129,
"step": 2540
},
{
"epoch": 3.62,
"learning_rate": 4.1428571428571435e-06,
"loss": 0.0113,
"step": 2550
},
{
"epoch": 3.63,
"learning_rate": 4.114285714285715e-06,
"loss": 0.0102,
"step": 2560
},
{
"epoch": 3.65,
"learning_rate": 4.0857142857142865e-06,
"loss": 0.0096,
"step": 2570
},
{
"epoch": 3.66,
"learning_rate": 4.057142857142858e-06,
"loss": 0.0106,
"step": 2580
},
{
"epoch": 3.67,
"learning_rate": 4.028571428571429e-06,
"loss": 0.0077,
"step": 2590
},
{
"epoch": 3.69,
"learning_rate": 4.000000000000001e-06,
"loss": 0.011,
"step": 2600
},
{
"epoch": 3.7,
"learning_rate": 3.971428571428572e-06,
"loss": 0.0089,
"step": 2610
},
{
"epoch": 3.72,
"learning_rate": 3.942857142857143e-06,
"loss": 0.0097,
"step": 2620
},
{
"epoch": 3.73,
"learning_rate": 3.914285714285714e-06,
"loss": 0.0101,
"step": 2630
},
{
"epoch": 3.74,
"learning_rate": 3.885714285714286e-06,
"loss": 0.012,
"step": 2640
},
{
"epoch": 3.76,
"learning_rate": 3.857142857142858e-06,
"loss": 0.0078,
"step": 2650
},
{
"epoch": 3.77,
"learning_rate": 3.828571428571429e-06,
"loss": 0.0089,
"step": 2660
},
{
"epoch": 3.79,
"learning_rate": 3.8000000000000005e-06,
"loss": 0.015,
"step": 2670
},
{
"epoch": 3.8,
"learning_rate": 3.771428571428572e-06,
"loss": 0.0112,
"step": 2680
},
{
"epoch": 3.82,
"learning_rate": 3.742857142857143e-06,
"loss": 0.0099,
"step": 2690
},
{
"epoch": 3.83,
"learning_rate": 3.7142857142857146e-06,
"loss": 0.0061,
"step": 2700
},
{
"epoch": 3.84,
"learning_rate": 3.6857142857142857e-06,
"loss": 0.0058,
"step": 2710
},
{
"epoch": 3.86,
"learning_rate": 3.6571428571428576e-06,
"loss": 0.0115,
"step": 2720
},
{
"epoch": 3.87,
"learning_rate": 3.628571428571429e-06,
"loss": 0.0116,
"step": 2730
},
{
"epoch": 3.89,
"learning_rate": 3.6000000000000003e-06,
"loss": 0.0186,
"step": 2740
},
{
"epoch": 3.9,
"learning_rate": 3.5714285714285718e-06,
"loss": 0.0121,
"step": 2750
},
{
"epoch": 3.91,
"learning_rate": 3.542857142857143e-06,
"loss": 0.0098,
"step": 2760
},
{
"epoch": 3.93,
"learning_rate": 3.5142857142857144e-06,
"loss": 0.0089,
"step": 2770
},
{
"epoch": 3.94,
"learning_rate": 3.4857142857142863e-06,
"loss": 0.0122,
"step": 2780
},
{
"epoch": 3.96,
"learning_rate": 3.4571428571428574e-06,
"loss": 0.0111,
"step": 2790
},
{
"epoch": 3.97,
"learning_rate": 3.428571428571429e-06,
"loss": 0.008,
"step": 2800
},
{
"epoch": 3.99,
"learning_rate": 3.4000000000000005e-06,
"loss": 0.0095,
"step": 2810
},
{
"epoch": 4.0,
"learning_rate": 3.3714285714285716e-06,
"loss": 0.0114,
"step": 2820
},
{
"epoch": 4.01,
"learning_rate": 3.342857142857143e-06,
"loss": 0.0033,
"step": 2830
},
{
"epoch": 4.03,
"learning_rate": 3.314285714285714e-06,
"loss": 0.0024,
"step": 2840
},
{
"epoch": 4.04,
"learning_rate": 3.285714285714286e-06,
"loss": 0.004,
"step": 2850
},
{
"epoch": 4.06,
"learning_rate": 3.2571428571428577e-06,
"loss": 0.0047,
"step": 2860
},
{
"epoch": 4.07,
"learning_rate": 3.2285714285714288e-06,
"loss": 0.0046,
"step": 2870
},
{
"epoch": 4.09,
"learning_rate": 3.2000000000000003e-06,
"loss": 0.0062,
"step": 2880
},
{
"epoch": 4.1,
"learning_rate": 3.1714285714285714e-06,
"loss": 0.0051,
"step": 2890
},
{
"epoch": 4.11,
"learning_rate": 3.142857142857143e-06,
"loss": 0.003,
"step": 2900
},
{
"epoch": 4.13,
"learning_rate": 3.114285714285715e-06,
"loss": 0.0036,
"step": 2910
},
{
"epoch": 4.14,
"learning_rate": 3.085714285714286e-06,
"loss": 0.0052,
"step": 2920
},
{
"epoch": 4.16,
"learning_rate": 3.0571428571428575e-06,
"loss": 0.0048,
"step": 2930
},
{
"epoch": 4.17,
"learning_rate": 3.028571428571429e-06,
"loss": 0.0036,
"step": 2940
},
{
"epoch": 4.18,
"learning_rate": 3e-06,
"loss": 0.004,
"step": 2950
},
{
"epoch": 4.2,
"learning_rate": 2.9714285714285716e-06,
"loss": 0.003,
"step": 2960
},
{
"epoch": 4.21,
"learning_rate": 2.9428571428571427e-06,
"loss": 0.0026,
"step": 2970
},
{
"epoch": 4.23,
"learning_rate": 2.9142857142857146e-06,
"loss": 0.0046,
"step": 2980
},
{
"epoch": 4.24,
"learning_rate": 2.885714285714286e-06,
"loss": 0.0031,
"step": 2990
},
{
"epoch": 4.26,
"learning_rate": 2.8571428571428573e-06,
"loss": 0.0036,
"step": 3000
},
{
"epoch": 4.26,
"eval_loss": 0.19506710767745972,
"eval_runtime": 1567.6008,
"eval_samples_per_second": 3.004,
"eval_steps_per_second": 0.376,
"eval_wer": 39.35018050541516,
"step": 3000
}
],
"max_steps": 4000,
"num_train_epochs": 6,
"total_flos": 1.384863619055616e+19,
"trial_name": null,
"trial_params": null
}