whisper-large-es-cv11 / trainer_state.json
jonatasgrosman's picture
End of training
5f73e5b
raw history blame
No virus
40.6 kB
{
"best_metric": 3.6508096148043854,
"best_model_checkpoint": "data/model/whisper-large-es-cv11-2/checkpoint-5000",
"epoch": 2.56,
"global_step": 8000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.1499999999999999e-08,
"loss": 0.8552,
"step": 25
},
{
"epoch": 0.02,
"learning_rate": 2.35e-08,
"loss": 0.8366,
"step": 50
},
{
"epoch": 0.02,
"learning_rate": 3.6e-08,
"loss": 0.8538,
"step": 75
},
{
"epoch": 0.03,
"learning_rate": 4.85e-08,
"loss": 0.842,
"step": 100
},
{
"epoch": 0.04,
"learning_rate": 6.049999999999999e-08,
"loss": 0.8284,
"step": 125
},
{
"epoch": 0.05,
"learning_rate": 7.299999999999999e-08,
"loss": 0.8163,
"step": 150
},
{
"epoch": 0.06,
"learning_rate": 8.55e-08,
"loss": 0.7738,
"step": 175
},
{
"epoch": 0.06,
"learning_rate": 9.8e-08,
"loss": 0.7009,
"step": 200
},
{
"epoch": 0.07,
"learning_rate": 1.0999999999999999e-07,
"loss": 0.5658,
"step": 225
},
{
"epoch": 0.08,
"learning_rate": 1.225e-07,
"loss": 0.4291,
"step": 250
},
{
"epoch": 0.09,
"learning_rate": 1.35e-07,
"loss": 0.3306,
"step": 275
},
{
"epoch": 0.1,
"learning_rate": 1.475e-07,
"loss": 0.3119,
"step": 300
},
{
"epoch": 0.1,
"learning_rate": 1.6e-07,
"loss": 0.3076,
"step": 325
},
{
"epoch": 0.11,
"learning_rate": 1.725e-07,
"loss": 0.301,
"step": 350
},
{
"epoch": 0.12,
"learning_rate": 1.85e-07,
"loss": 0.2942,
"step": 375
},
{
"epoch": 0.13,
"learning_rate": 1.975e-07,
"loss": 0.279,
"step": 400
},
{
"epoch": 0.14,
"learning_rate": 2.0999999999999997e-07,
"loss": 0.2694,
"step": 425
},
{
"epoch": 0.14,
"learning_rate": 2.225e-07,
"loss": 0.2416,
"step": 450
},
{
"epoch": 0.15,
"learning_rate": 2.3499999999999997e-07,
"loss": 0.2491,
"step": 475
},
{
"epoch": 0.16,
"learning_rate": 2.475e-07,
"loss": 0.2518,
"step": 500
},
{
"epoch": 0.17,
"learning_rate": 2.6e-07,
"loss": 0.2497,
"step": 525
},
{
"epoch": 0.18,
"learning_rate": 2.725e-07,
"loss": 0.2362,
"step": 550
},
{
"epoch": 0.18,
"learning_rate": 2.8499999999999997e-07,
"loss": 0.24,
"step": 575
},
{
"epoch": 0.19,
"learning_rate": 2.9749999999999996e-07,
"loss": 0.2215,
"step": 600
},
{
"epoch": 0.2,
"learning_rate": 3.1e-07,
"loss": 0.2104,
"step": 625
},
{
"epoch": 0.21,
"learning_rate": 3.225e-07,
"loss": 0.2115,
"step": 650
},
{
"epoch": 0.22,
"learning_rate": 3.35e-07,
"loss": 0.2064,
"step": 675
},
{
"epoch": 0.22,
"learning_rate": 3.4749999999999996e-07,
"loss": 0.2026,
"step": 700
},
{
"epoch": 0.23,
"learning_rate": 3.6e-07,
"loss": 0.2048,
"step": 725
},
{
"epoch": 0.24,
"learning_rate": 3.725e-07,
"loss": 0.1946,
"step": 750
},
{
"epoch": 0.25,
"learning_rate": 3.8499999999999997e-07,
"loss": 0.1872,
"step": 775
},
{
"epoch": 0.26,
"learning_rate": 3.975e-07,
"loss": 0.1851,
"step": 800
},
{
"epoch": 0.26,
"learning_rate": 4.0999999999999994e-07,
"loss": 0.1714,
"step": 825
},
{
"epoch": 0.27,
"learning_rate": 4.225e-07,
"loss": 0.1869,
"step": 850
},
{
"epoch": 0.28,
"learning_rate": 4.3499999999999996e-07,
"loss": 0.1847,
"step": 875
},
{
"epoch": 0.29,
"learning_rate": 4.475e-07,
"loss": 0.1931,
"step": 900
},
{
"epoch": 0.3,
"learning_rate": 4.6e-07,
"loss": 0.1776,
"step": 925
},
{
"epoch": 0.3,
"learning_rate": 4.725e-07,
"loss": 0.1803,
"step": 950
},
{
"epoch": 0.31,
"learning_rate": 4.85e-07,
"loss": 0.1721,
"step": 975
},
{
"epoch": 0.32,
"learning_rate": 4.975e-07,
"loss": 0.1837,
"step": 1000
},
{
"epoch": 0.32,
"eval_cer": 1.2487604833694683,
"eval_loss": 0.16694751381874084,
"eval_runtime": 939.4915,
"eval_samples_per_second": 1.064,
"eval_steps_per_second": 0.133,
"eval_wer": 4.244191893794629,
"step": 1000
},
{
"epoch": 0.33,
"learning_rate": 5.1e-07,
"loss": 0.1685,
"step": 1025
},
{
"epoch": 0.34,
"learning_rate": 5.225e-07,
"loss": 0.1667,
"step": 1050
},
{
"epoch": 0.34,
"learning_rate": 5.35e-07,
"loss": 0.1731,
"step": 1075
},
{
"epoch": 0.35,
"learning_rate": 5.474999999999999e-07,
"loss": 0.1667,
"step": 1100
},
{
"epoch": 0.36,
"learning_rate": 5.6e-07,
"loss": 0.1684,
"step": 1125
},
{
"epoch": 0.37,
"learning_rate": 5.725e-07,
"loss": 0.1713,
"step": 1150
},
{
"epoch": 0.38,
"learning_rate": 5.849999999999999e-07,
"loss": 0.1625,
"step": 1175
},
{
"epoch": 0.38,
"learning_rate": 5.975e-07,
"loss": 0.1578,
"step": 1200
},
{
"epoch": 0.39,
"learning_rate": 6.1e-07,
"loss": 0.1621,
"step": 1225
},
{
"epoch": 0.4,
"learning_rate": 6.225000000000001e-07,
"loss": 0.1652,
"step": 1250
},
{
"epoch": 0.41,
"learning_rate": 6.35e-07,
"loss": 0.159,
"step": 1275
},
{
"epoch": 0.42,
"learning_rate": 6.474999999999999e-07,
"loss": 0.1722,
"step": 1300
},
{
"epoch": 0.42,
"learning_rate": 6.6e-07,
"loss": 0.16,
"step": 1325
},
{
"epoch": 0.43,
"learning_rate": 6.724999999999999e-07,
"loss": 0.1582,
"step": 1350
},
{
"epoch": 0.44,
"learning_rate": 6.85e-07,
"loss": 0.1636,
"step": 1375
},
{
"epoch": 0.45,
"learning_rate": 6.975e-07,
"loss": 0.1555,
"step": 1400
},
{
"epoch": 0.46,
"learning_rate": 7.1e-07,
"loss": 0.1576,
"step": 1425
},
{
"epoch": 0.46,
"learning_rate": 7.225e-07,
"loss": 0.1605,
"step": 1450
},
{
"epoch": 0.47,
"learning_rate": 7.345e-07,
"loss": 0.1614,
"step": 1475
},
{
"epoch": 0.48,
"learning_rate": 7.47e-07,
"loss": 0.1412,
"step": 1500
},
{
"epoch": 0.49,
"learning_rate": 7.594999999999999e-07,
"loss": 0.1451,
"step": 1525
},
{
"epoch": 0.5,
"learning_rate": 7.72e-07,
"loss": 0.1549,
"step": 1550
},
{
"epoch": 0.5,
"learning_rate": 7.845e-07,
"loss": 0.1528,
"step": 1575
},
{
"epoch": 0.51,
"learning_rate": 7.970000000000001e-07,
"loss": 0.1535,
"step": 1600
},
{
"epoch": 0.52,
"learning_rate": 8.094999999999999e-07,
"loss": 0.1578,
"step": 1625
},
{
"epoch": 0.53,
"learning_rate": 8.219999999999999e-07,
"loss": 0.1526,
"step": 1650
},
{
"epoch": 0.54,
"learning_rate": 8.345e-07,
"loss": 0.1493,
"step": 1675
},
{
"epoch": 0.54,
"learning_rate": 8.469999999999999e-07,
"loss": 0.1471,
"step": 1700
},
{
"epoch": 0.55,
"learning_rate": 8.595e-07,
"loss": 0.1359,
"step": 1725
},
{
"epoch": 0.56,
"learning_rate": 8.72e-07,
"loss": 0.1494,
"step": 1750
},
{
"epoch": 0.57,
"learning_rate": 8.845e-07,
"loss": 0.1521,
"step": 1775
},
{
"epoch": 0.58,
"learning_rate": 8.969999999999999e-07,
"loss": 0.158,
"step": 1800
},
{
"epoch": 0.58,
"learning_rate": 9.094999999999999e-07,
"loss": 0.1591,
"step": 1825
},
{
"epoch": 0.59,
"learning_rate": 9.22e-07,
"loss": 0.1408,
"step": 1850
},
{
"epoch": 0.6,
"learning_rate": 9.344999999999999e-07,
"loss": 0.1719,
"step": 1875
},
{
"epoch": 0.61,
"learning_rate": 9.469999999999999e-07,
"loss": 0.1466,
"step": 1900
},
{
"epoch": 0.62,
"learning_rate": 9.594999999999999e-07,
"loss": 0.1502,
"step": 1925
},
{
"epoch": 0.62,
"learning_rate": 9.72e-07,
"loss": 0.1487,
"step": 1950
},
{
"epoch": 0.63,
"learning_rate": 9.845e-07,
"loss": 0.1463,
"step": 1975
},
{
"epoch": 0.64,
"learning_rate": 9.97e-07,
"loss": 0.1343,
"step": 2000
},
{
"epoch": 0.64,
"eval_cer": 1.208423670986067,
"eval_loss": 0.14436140656471252,
"eval_runtime": 941.5744,
"eval_samples_per_second": 1.062,
"eval_steps_per_second": 0.133,
"eval_wer": 4.083274665593885,
"step": 2000
},
{
"epoch": 0.65,
"learning_rate": 9.989444444444443e-07,
"loss": 0.1575,
"step": 2025
},
{
"epoch": 0.66,
"learning_rate": 9.975555555555556e-07,
"loss": 0.1468,
"step": 2050
},
{
"epoch": 0.66,
"learning_rate": 9.961666666666667e-07,
"loss": 0.1389,
"step": 2075
},
{
"epoch": 0.67,
"learning_rate": 9.947777777777778e-07,
"loss": 0.1412,
"step": 2100
},
{
"epoch": 0.68,
"learning_rate": 9.933888888888889e-07,
"loss": 0.1589,
"step": 2125
},
{
"epoch": 0.69,
"learning_rate": 9.92e-07,
"loss": 0.141,
"step": 2150
},
{
"epoch": 0.7,
"learning_rate": 9.90611111111111e-07,
"loss": 0.1374,
"step": 2175
},
{
"epoch": 0.7,
"learning_rate": 9.892222222222222e-07,
"loss": 0.1444,
"step": 2200
},
{
"epoch": 0.71,
"learning_rate": 9.878333333333333e-07,
"loss": 0.1583,
"step": 2225
},
{
"epoch": 0.72,
"learning_rate": 9.864444444444444e-07,
"loss": 0.1529,
"step": 2250
},
{
"epoch": 0.73,
"learning_rate": 9.850555555555555e-07,
"loss": 0.1455,
"step": 2275
},
{
"epoch": 0.74,
"learning_rate": 9.836666666666666e-07,
"loss": 0.1498,
"step": 2300
},
{
"epoch": 0.74,
"learning_rate": 9.822777777777777e-07,
"loss": 0.1464,
"step": 2325
},
{
"epoch": 0.75,
"learning_rate": 9.808888888888888e-07,
"loss": 0.1486,
"step": 2350
},
{
"epoch": 0.76,
"learning_rate": 9.795e-07,
"loss": 0.1449,
"step": 2375
},
{
"epoch": 0.77,
"learning_rate": 9.781111111111112e-07,
"loss": 0.1439,
"step": 2400
},
{
"epoch": 0.78,
"learning_rate": 9.767222222222223e-07,
"loss": 0.1454,
"step": 2425
},
{
"epoch": 0.78,
"learning_rate": 9.753333333333334e-07,
"loss": 0.1285,
"step": 2450
},
{
"epoch": 0.79,
"learning_rate": 9.739444444444443e-07,
"loss": 0.1495,
"step": 2475
},
{
"epoch": 0.8,
"learning_rate": 9.725555555555554e-07,
"loss": 0.1487,
"step": 2500
},
{
"epoch": 0.81,
"learning_rate": 9.711666666666665e-07,
"loss": 0.1284,
"step": 2525
},
{
"epoch": 0.82,
"learning_rate": 9.697777777777776e-07,
"loss": 0.14,
"step": 2550
},
{
"epoch": 0.82,
"learning_rate": 9.68388888888889e-07,
"loss": 0.137,
"step": 2575
},
{
"epoch": 0.83,
"learning_rate": 9.67e-07,
"loss": 0.1532,
"step": 2600
},
{
"epoch": 0.84,
"learning_rate": 9.656111111111111e-07,
"loss": 0.1293,
"step": 2625
},
{
"epoch": 0.85,
"learning_rate": 9.642222222222222e-07,
"loss": 0.1386,
"step": 2650
},
{
"epoch": 0.86,
"learning_rate": 9.628333333333333e-07,
"loss": 0.1492,
"step": 2675
},
{
"epoch": 0.86,
"learning_rate": 9.614444444444444e-07,
"loss": 0.1413,
"step": 2700
},
{
"epoch": 0.87,
"learning_rate": 9.600555555555555e-07,
"loss": 0.1394,
"step": 2725
},
{
"epoch": 0.88,
"learning_rate": 9.586666666666666e-07,
"loss": 0.1377,
"step": 2750
},
{
"epoch": 0.89,
"learning_rate": 9.572777777777777e-07,
"loss": 0.1467,
"step": 2775
},
{
"epoch": 0.9,
"learning_rate": 9.558888888888888e-07,
"loss": 0.1341,
"step": 2800
},
{
"epoch": 0.9,
"learning_rate": 9.545e-07,
"loss": 0.1387,
"step": 2825
},
{
"epoch": 0.91,
"learning_rate": 9.531111111111111e-07,
"loss": 0.1358,
"step": 2850
},
{
"epoch": 0.92,
"learning_rate": 9.517222222222222e-07,
"loss": 0.1346,
"step": 2875
},
{
"epoch": 0.93,
"learning_rate": 9.503333333333333e-07,
"loss": 0.1438,
"step": 2900
},
{
"epoch": 0.94,
"learning_rate": 9.489444444444444e-07,
"loss": 0.149,
"step": 2925
},
{
"epoch": 0.94,
"learning_rate": 9.475555555555555e-07,
"loss": 0.138,
"step": 2950
},
{
"epoch": 0.95,
"learning_rate": 9.461666666666666e-07,
"loss": 0.1292,
"step": 2975
},
{
"epoch": 0.96,
"learning_rate": 9.447777777777777e-07,
"loss": 0.1312,
"step": 3000
},
{
"epoch": 0.96,
"eval_cer": 1.1932973663422914,
"eval_loss": 0.13624157011508942,
"eval_runtime": 939.634,
"eval_samples_per_second": 1.064,
"eval_steps_per_second": 0.133,
"eval_wer": 3.9324147641556872,
"step": 3000
},
{
"epoch": 0.97,
"learning_rate": 9.433888888888889e-07,
"loss": 0.1414,
"step": 3025
},
{
"epoch": 0.98,
"learning_rate": 9.419999999999999e-07,
"loss": 0.1446,
"step": 3050
},
{
"epoch": 0.98,
"learning_rate": 9.40611111111111e-07,
"loss": 0.129,
"step": 3075
},
{
"epoch": 0.99,
"learning_rate": 9.392222222222221e-07,
"loss": 0.1378,
"step": 3100
},
{
"epoch": 1.0,
"learning_rate": 9.378333333333332e-07,
"loss": 0.1277,
"step": 3125
},
{
"epoch": 1.01,
"learning_rate": 9.364444444444443e-07,
"loss": 0.1301,
"step": 3150
},
{
"epoch": 1.02,
"learning_rate": 9.350555555555554e-07,
"loss": 0.1193,
"step": 3175
},
{
"epoch": 1.02,
"learning_rate": 9.336666666666666e-07,
"loss": 0.1223,
"step": 3200
},
{
"epoch": 1.03,
"learning_rate": 9.322777777777778e-07,
"loss": 0.1233,
"step": 3225
},
{
"epoch": 1.04,
"learning_rate": 9.308888888888889e-07,
"loss": 0.1176,
"step": 3250
},
{
"epoch": 1.05,
"learning_rate": 9.295e-07,
"loss": 0.1145,
"step": 3275
},
{
"epoch": 1.06,
"learning_rate": 9.281111111111111e-07,
"loss": 0.1225,
"step": 3300
},
{
"epoch": 1.06,
"learning_rate": 9.267222222222222e-07,
"loss": 0.1251,
"step": 3325
},
{
"epoch": 1.07,
"learning_rate": 9.253333333333333e-07,
"loss": 0.1257,
"step": 3350
},
{
"epoch": 1.08,
"learning_rate": 9.239444444444445e-07,
"loss": 0.1075,
"step": 3375
},
{
"epoch": 1.09,
"learning_rate": 9.225555555555556e-07,
"loss": 0.122,
"step": 3400
},
{
"epoch": 1.1,
"learning_rate": 9.211666666666667e-07,
"loss": 0.1245,
"step": 3425
},
{
"epoch": 1.1,
"learning_rate": 9.197777777777778e-07,
"loss": 0.122,
"step": 3450
},
{
"epoch": 1.11,
"learning_rate": 9.183888888888889e-07,
"loss": 0.1219,
"step": 3475
},
{
"epoch": 1.12,
"learning_rate": 9.17e-07,
"loss": 0.1053,
"step": 3500
},
{
"epoch": 1.13,
"learning_rate": 9.156111111111111e-07,
"loss": 0.1197,
"step": 3525
},
{
"epoch": 1.14,
"learning_rate": 9.142222222222223e-07,
"loss": 0.1162,
"step": 3550
},
{
"epoch": 1.14,
"learning_rate": 9.128333333333334e-07,
"loss": 0.1185,
"step": 3575
},
{
"epoch": 1.15,
"learning_rate": 9.114444444444444e-07,
"loss": 0.1148,
"step": 3600
},
{
"epoch": 1.16,
"learning_rate": 9.100555555555555e-07,
"loss": 0.1117,
"step": 3625
},
{
"epoch": 1.17,
"learning_rate": 9.086666666666666e-07,
"loss": 0.1211,
"step": 3650
},
{
"epoch": 1.18,
"learning_rate": 9.072777777777777e-07,
"loss": 0.1055,
"step": 3675
},
{
"epoch": 1.18,
"learning_rate": 9.058888888888888e-07,
"loss": 0.1113,
"step": 3700
},
{
"epoch": 1.19,
"learning_rate": 9.045e-07,
"loss": 0.1125,
"step": 3725
},
{
"epoch": 1.2,
"learning_rate": 9.031111111111111e-07,
"loss": 0.1146,
"step": 3750
},
{
"epoch": 1.21,
"learning_rate": 9.017222222222222e-07,
"loss": 0.1293,
"step": 3775
},
{
"epoch": 1.22,
"learning_rate": 9.003333333333333e-07,
"loss": 0.1306,
"step": 3800
},
{
"epoch": 1.22,
"learning_rate": 8.989444444444444e-07,
"loss": 0.107,
"step": 3825
},
{
"epoch": 1.23,
"learning_rate": 8.975555555555555e-07,
"loss": 0.1194,
"step": 3850
},
{
"epoch": 1.24,
"learning_rate": 8.961666666666666e-07,
"loss": 0.112,
"step": 3875
},
{
"epoch": 1.25,
"learning_rate": 8.947777777777778e-07,
"loss": 0.1173,
"step": 3900
},
{
"epoch": 1.26,
"learning_rate": 8.933888888888889e-07,
"loss": 0.1193,
"step": 3925
},
{
"epoch": 1.26,
"learning_rate": 8.92e-07,
"loss": 0.1148,
"step": 3950
},
{
"epoch": 1.27,
"learning_rate": 8.906111111111111e-07,
"loss": 0.1154,
"step": 3975
},
{
"epoch": 1.28,
"learning_rate": 8.892222222222222e-07,
"loss": 0.1206,
"step": 4000
},
{
"epoch": 1.28,
"eval_cer": 1.174809660666566,
"eval_loss": 0.13329146802425385,
"eval_runtime": 941.0652,
"eval_samples_per_second": 1.063,
"eval_steps_per_second": 0.133,
"eval_wer": 3.8519561500553157,
"step": 4000
},
{
"epoch": 1.29,
"learning_rate": 8.878333333333333e-07,
"loss": 0.1156,
"step": 4025
},
{
"epoch": 1.3,
"learning_rate": 8.864444444444444e-07,
"loss": 0.1223,
"step": 4050
},
{
"epoch": 1.3,
"learning_rate": 8.850555555555556e-07,
"loss": 0.1176,
"step": 4075
},
{
"epoch": 1.31,
"learning_rate": 8.836666666666667e-07,
"loss": 0.1134,
"step": 4100
},
{
"epoch": 1.32,
"learning_rate": 8.822777777777778e-07,
"loss": 0.1195,
"step": 4125
},
{
"epoch": 1.33,
"learning_rate": 8.808888888888889e-07,
"loss": 0.1131,
"step": 4150
},
{
"epoch": 1.34,
"learning_rate": 8.794999999999999e-07,
"loss": 0.1251,
"step": 4175
},
{
"epoch": 1.34,
"learning_rate": 8.78111111111111e-07,
"loss": 0.1102,
"step": 4200
},
{
"epoch": 1.35,
"learning_rate": 8.767222222222221e-07,
"loss": 0.1223,
"step": 4225
},
{
"epoch": 1.36,
"learning_rate": 8.753333333333332e-07,
"loss": 0.109,
"step": 4250
},
{
"epoch": 1.37,
"learning_rate": 8.739444444444444e-07,
"loss": 0.1226,
"step": 4275
},
{
"epoch": 1.38,
"learning_rate": 8.725555555555555e-07,
"loss": 0.1201,
"step": 4300
},
{
"epoch": 1.38,
"learning_rate": 8.711666666666666e-07,
"loss": 0.1126,
"step": 4325
},
{
"epoch": 1.39,
"learning_rate": 8.697777777777777e-07,
"loss": 0.113,
"step": 4350
},
{
"epoch": 1.4,
"learning_rate": 8.683888888888888e-07,
"loss": 0.1132,
"step": 4375
},
{
"epoch": 1.41,
"learning_rate": 8.669999999999999e-07,
"loss": 0.103,
"step": 4400
},
{
"epoch": 1.42,
"learning_rate": 8.656111111111111e-07,
"loss": 0.1235,
"step": 4425
},
{
"epoch": 1.42,
"learning_rate": 8.642222222222222e-07,
"loss": 0.1189,
"step": 4450
},
{
"epoch": 1.43,
"learning_rate": 8.628333333333333e-07,
"loss": 0.1303,
"step": 4475
},
{
"epoch": 1.44,
"learning_rate": 8.614444444444444e-07,
"loss": 0.1121,
"step": 4500
},
{
"epoch": 1.45,
"learning_rate": 8.600555555555555e-07,
"loss": 0.1172,
"step": 4525
},
{
"epoch": 1.46,
"learning_rate": 8.586666666666666e-07,
"loss": 0.1214,
"step": 4550
},
{
"epoch": 1.46,
"learning_rate": 8.572777777777777e-07,
"loss": 0.1172,
"step": 4575
},
{
"epoch": 1.47,
"learning_rate": 8.558888888888889e-07,
"loss": 0.1124,
"step": 4600
},
{
"epoch": 1.48,
"learning_rate": 8.545e-07,
"loss": 0.116,
"step": 4625
},
{
"epoch": 1.49,
"learning_rate": 8.531111111111111e-07,
"loss": 0.1147,
"step": 4650
},
{
"epoch": 1.5,
"learning_rate": 8.517222222222222e-07,
"loss": 0.1209,
"step": 4675
},
{
"epoch": 1.5,
"learning_rate": 8.503333333333333e-07,
"loss": 0.1176,
"step": 4700
},
{
"epoch": 1.51,
"learning_rate": 8.489444444444443e-07,
"loss": 0.1139,
"step": 4725
},
{
"epoch": 1.52,
"learning_rate": 8.475555555555554e-07,
"loss": 0.1031,
"step": 4750
},
{
"epoch": 1.53,
"learning_rate": 8.461666666666665e-07,
"loss": 0.1179,
"step": 4775
},
{
"epoch": 1.54,
"learning_rate": 8.447777777777777e-07,
"loss": 0.1177,
"step": 4800
},
{
"epoch": 1.54,
"learning_rate": 8.433888888888888e-07,
"loss": 0.1054,
"step": 4825
},
{
"epoch": 1.55,
"learning_rate": 8.419999999999999e-07,
"loss": 0.1184,
"step": 4850
},
{
"epoch": 1.56,
"learning_rate": 8.40611111111111e-07,
"loss": 0.1149,
"step": 4875
},
{
"epoch": 1.57,
"learning_rate": 8.392222222222221e-07,
"loss": 0.1085,
"step": 4900
},
{
"epoch": 1.58,
"learning_rate": 8.378333333333332e-07,
"loss": 0.1158,
"step": 4925
},
{
"epoch": 1.58,
"learning_rate": 8.364444444444443e-07,
"loss": 0.1168,
"step": 4950
},
{
"epoch": 1.59,
"learning_rate": 8.350555555555556e-07,
"loss": 0.1085,
"step": 4975
},
{
"epoch": 1.6,
"learning_rate": 8.336666666666667e-07,
"loss": 0.1143,
"step": 5000
},
{
"epoch": 1.6,
"eval_cer": 1.0571606245483118,
"eval_loss": 0.13207247853279114,
"eval_runtime": 940.9316,
"eval_samples_per_second": 1.063,
"eval_steps_per_second": 0.133,
"eval_wer": 3.6508096148043854,
"step": 5000
},
{
"epoch": 1.61,
"learning_rate": 8.322777777777778e-07,
"loss": 0.1176,
"step": 5025
},
{
"epoch": 1.62,
"learning_rate": 8.308888888888889e-07,
"loss": 0.1303,
"step": 5050
},
{
"epoch": 1.62,
"learning_rate": 8.295e-07,
"loss": 0.1226,
"step": 5075
},
{
"epoch": 1.63,
"learning_rate": 8.281111111111111e-07,
"loss": 0.1189,
"step": 5100
},
{
"epoch": 1.64,
"learning_rate": 8.267222222222222e-07,
"loss": 0.118,
"step": 5125
},
{
"epoch": 1.65,
"learning_rate": 8.253333333333334e-07,
"loss": 0.1264,
"step": 5150
},
{
"epoch": 1.66,
"learning_rate": 8.239444444444445e-07,
"loss": 0.1162,
"step": 5175
},
{
"epoch": 1.66,
"learning_rate": 8.225555555555556e-07,
"loss": 0.1081,
"step": 5200
},
{
"epoch": 1.67,
"learning_rate": 8.211666666666667e-07,
"loss": 0.1205,
"step": 5225
},
{
"epoch": 1.68,
"learning_rate": 8.197777777777778e-07,
"loss": 0.1191,
"step": 5250
},
{
"epoch": 1.69,
"learning_rate": 8.183888888888889e-07,
"loss": 0.1166,
"step": 5275
},
{
"epoch": 1.7,
"learning_rate": 8.169999999999999e-07,
"loss": 0.1173,
"step": 5300
},
{
"epoch": 1.7,
"learning_rate": 8.156111111111111e-07,
"loss": 0.1138,
"step": 5325
},
{
"epoch": 1.71,
"learning_rate": 8.142222222222222e-07,
"loss": 0.1261,
"step": 5350
},
{
"epoch": 1.72,
"learning_rate": 8.128333333333333e-07,
"loss": 0.1213,
"step": 5375
},
{
"epoch": 1.73,
"learning_rate": 8.114444444444444e-07,
"loss": 0.1097,
"step": 5400
},
{
"epoch": 1.74,
"learning_rate": 8.100555555555555e-07,
"loss": 0.1169,
"step": 5425
},
{
"epoch": 1.74,
"learning_rate": 8.086666666666666e-07,
"loss": 0.1145,
"step": 5450
},
{
"epoch": 1.75,
"learning_rate": 8.072777777777777e-07,
"loss": 0.1165,
"step": 5475
},
{
"epoch": 1.76,
"learning_rate": 8.058888888888889e-07,
"loss": 0.122,
"step": 5500
},
{
"epoch": 1.77,
"learning_rate": 8.045e-07,
"loss": 0.1215,
"step": 5525
},
{
"epoch": 1.78,
"learning_rate": 8.031111111111111e-07,
"loss": 0.1172,
"step": 5550
},
{
"epoch": 1.78,
"learning_rate": 8.017777777777778e-07,
"loss": 0.1072,
"step": 5575
},
{
"epoch": 1.79,
"learning_rate": 8.003888888888889e-07,
"loss": 0.1198,
"step": 5600
},
{
"epoch": 1.8,
"learning_rate": 7.99e-07,
"loss": 0.1232,
"step": 5625
},
{
"epoch": 1.81,
"learning_rate": 7.976111111111111e-07,
"loss": 0.1184,
"step": 5650
},
{
"epoch": 1.82,
"learning_rate": 7.962222222222222e-07,
"loss": 0.1163,
"step": 5675
},
{
"epoch": 1.82,
"learning_rate": 7.948333333333333e-07,
"loss": 0.1181,
"step": 5700
},
{
"epoch": 1.83,
"learning_rate": 7.934444444444444e-07,
"loss": 0.1135,
"step": 5725
},
{
"epoch": 1.84,
"learning_rate": 7.920555555555555e-07,
"loss": 0.1195,
"step": 5750
},
{
"epoch": 1.85,
"learning_rate": 7.906666666666666e-07,
"loss": 0.1118,
"step": 5775
},
{
"epoch": 1.86,
"learning_rate": 7.892777777777777e-07,
"loss": 0.1131,
"step": 5800
},
{
"epoch": 1.86,
"learning_rate": 7.878888888888888e-07,
"loss": 0.1154,
"step": 5825
},
{
"epoch": 1.87,
"learning_rate": 7.864999999999999e-07,
"loss": 0.1136,
"step": 5850
},
{
"epoch": 1.88,
"learning_rate": 7.851111111111111e-07,
"loss": 0.1097,
"step": 5875
},
{
"epoch": 1.89,
"learning_rate": 7.837222222222222e-07,
"loss": 0.113,
"step": 5900
},
{
"epoch": 1.9,
"learning_rate": 7.823333333333333e-07,
"loss": 0.1179,
"step": 5925
},
{
"epoch": 1.9,
"learning_rate": 7.809444444444444e-07,
"loss": 0.1118,
"step": 5950
},
{
"epoch": 1.91,
"learning_rate": 7.795555555555555e-07,
"loss": 0.1151,
"step": 5975
},
{
"epoch": 1.92,
"learning_rate": 7.781666666666666e-07,
"loss": 0.1202,
"step": 6000
},
{
"epoch": 1.92,
"eval_cer": 1.1311114472512143,
"eval_loss": 0.12914104759693146,
"eval_runtime": 942.1178,
"eval_samples_per_second": 1.061,
"eval_steps_per_second": 0.133,
"eval_wer": 3.8016695162425824,
"step": 6000
},
{
"epoch": 1.93,
"learning_rate": 7.767777777777777e-07,
"loss": 0.1144,
"step": 6025
},
{
"epoch": 1.94,
"learning_rate": 7.753888888888889e-07,
"loss": 0.1077,
"step": 6050
},
{
"epoch": 1.94,
"learning_rate": 7.74e-07,
"loss": 0.1064,
"step": 6075
},
{
"epoch": 1.95,
"learning_rate": 7.726111111111111e-07,
"loss": 0.1191,
"step": 6100
},
{
"epoch": 1.96,
"learning_rate": 7.712222222222222e-07,
"loss": 0.1176,
"step": 6125
},
{
"epoch": 1.97,
"learning_rate": 7.698333333333333e-07,
"loss": 0.112,
"step": 6150
},
{
"epoch": 1.98,
"learning_rate": 7.684444444444444e-07,
"loss": 0.1103,
"step": 6175
},
{
"epoch": 1.98,
"learning_rate": 7.670555555555555e-07,
"loss": 0.1189,
"step": 6200
},
{
"epoch": 1.99,
"learning_rate": 7.656666666666667e-07,
"loss": 0.1124,
"step": 6225
},
{
"epoch": 2.0,
"learning_rate": 7.642777777777777e-07,
"loss": 0.1103,
"step": 6250
},
{
"epoch": 2.01,
"learning_rate": 7.628888888888888e-07,
"loss": 0.0959,
"step": 6275
},
{
"epoch": 2.02,
"learning_rate": 7.614999999999999e-07,
"loss": 0.0909,
"step": 6300
},
{
"epoch": 2.02,
"learning_rate": 7.60111111111111e-07,
"loss": 0.0894,
"step": 6325
},
{
"epoch": 2.03,
"learning_rate": 7.587222222222221e-07,
"loss": 0.0895,
"step": 6350
},
{
"epoch": 2.04,
"learning_rate": 7.573333333333332e-07,
"loss": 0.0957,
"step": 6375
},
{
"epoch": 2.05,
"learning_rate": 7.559444444444443e-07,
"loss": 0.0949,
"step": 6400
},
{
"epoch": 2.06,
"learning_rate": 7.545555555555555e-07,
"loss": 0.0966,
"step": 6425
},
{
"epoch": 2.06,
"learning_rate": 7.531666666666666e-07,
"loss": 0.0903,
"step": 6450
},
{
"epoch": 2.07,
"learning_rate": 7.517777777777777e-07,
"loss": 0.0943,
"step": 6475
},
{
"epoch": 2.08,
"learning_rate": 7.503888888888888e-07,
"loss": 0.0923,
"step": 6500
},
{
"epoch": 2.09,
"learning_rate": 7.489999999999999e-07,
"loss": 0.0958,
"step": 6525
},
{
"epoch": 2.1,
"learning_rate": 7.47611111111111e-07,
"loss": 0.0867,
"step": 6550
},
{
"epoch": 2.1,
"learning_rate": 7.462222222222221e-07,
"loss": 0.0901,
"step": 6575
},
{
"epoch": 2.11,
"learning_rate": 7.448333333333334e-07,
"loss": 0.0947,
"step": 6600
},
{
"epoch": 2.12,
"learning_rate": 7.434444444444445e-07,
"loss": 0.0897,
"step": 6625
},
{
"epoch": 2.13,
"learning_rate": 7.420555555555556e-07,
"loss": 0.0908,
"step": 6650
},
{
"epoch": 2.14,
"learning_rate": 7.406666666666667e-07,
"loss": 0.0945,
"step": 6675
},
{
"epoch": 2.14,
"learning_rate": 7.392777777777778e-07,
"loss": 0.0867,
"step": 6700
},
{
"epoch": 2.15,
"learning_rate": 7.378888888888889e-07,
"loss": 0.0979,
"step": 6725
},
{
"epoch": 2.16,
"learning_rate": 7.365e-07,
"loss": 0.098,
"step": 6750
},
{
"epoch": 2.17,
"learning_rate": 7.351111111111112e-07,
"loss": 0.0915,
"step": 6775
},
{
"epoch": 2.18,
"learning_rate": 7.337222222222223e-07,
"loss": 0.092,
"step": 6800
},
{
"epoch": 2.18,
"learning_rate": 7.323333333333333e-07,
"loss": 0.0888,
"step": 6825
},
{
"epoch": 2.19,
"learning_rate": 7.309444444444444e-07,
"loss": 0.0872,
"step": 6850
},
{
"epoch": 2.2,
"learning_rate": 7.295555555555555e-07,
"loss": 0.0892,
"step": 6875
},
{
"epoch": 2.21,
"learning_rate": 7.281666666666666e-07,
"loss": 0.0808,
"step": 6900
},
{
"epoch": 2.22,
"learning_rate": 7.267777777777777e-07,
"loss": 0.0965,
"step": 6925
},
{
"epoch": 2.22,
"learning_rate": 7.253888888888889e-07,
"loss": 0.0879,
"step": 6950
},
{
"epoch": 2.23,
"learning_rate": 7.24e-07,
"loss": 0.0915,
"step": 6975
},
{
"epoch": 2.24,
"learning_rate": 7.226111111111111e-07,
"loss": 0.0856,
"step": 7000
},
{
"epoch": 2.24,
"eval_cer": 1.0840518328039128,
"eval_loss": 0.13248993456363678,
"eval_runtime": 939.5436,
"eval_samples_per_second": 1.064,
"eval_steps_per_second": 0.133,
"eval_wer": 3.7010962486171173,
"step": 7000
},
{
"epoch": 2.25,
"learning_rate": 7.212222222222222e-07,
"loss": 0.0921,
"step": 7025
},
{
"epoch": 2.26,
"learning_rate": 7.198333333333333e-07,
"loss": 0.0928,
"step": 7050
},
{
"epoch": 2.26,
"learning_rate": 7.184444444444444e-07,
"loss": 0.0895,
"step": 7075
},
{
"epoch": 2.27,
"learning_rate": 7.170555555555555e-07,
"loss": 0.0949,
"step": 7100
},
{
"epoch": 2.28,
"learning_rate": 7.156666666666667e-07,
"loss": 0.09,
"step": 7125
},
{
"epoch": 2.29,
"learning_rate": 7.142777777777778e-07,
"loss": 0.0794,
"step": 7150
},
{
"epoch": 2.3,
"learning_rate": 7.128888888888889e-07,
"loss": 0.0945,
"step": 7175
},
{
"epoch": 2.3,
"learning_rate": 7.115e-07,
"loss": 0.0906,
"step": 7200
},
{
"epoch": 2.31,
"learning_rate": 7.101111111111111e-07,
"loss": 0.0972,
"step": 7225
},
{
"epoch": 2.32,
"learning_rate": 7.087222222222222e-07,
"loss": 0.0922,
"step": 7250
},
{
"epoch": 2.33,
"learning_rate": 7.073333333333333e-07,
"loss": 0.091,
"step": 7275
},
{
"epoch": 2.34,
"learning_rate": 7.059444444444445e-07,
"loss": 0.0981,
"step": 7300
},
{
"epoch": 2.34,
"learning_rate": 7.045555555555556e-07,
"loss": 0.0913,
"step": 7325
},
{
"epoch": 2.35,
"learning_rate": 7.031666666666667e-07,
"loss": 0.0955,
"step": 7350
},
{
"epoch": 2.36,
"learning_rate": 7.017777777777777e-07,
"loss": 0.0918,
"step": 7375
},
{
"epoch": 2.37,
"learning_rate": 7.003888888888888e-07,
"loss": 0.0991,
"step": 7400
},
{
"epoch": 2.38,
"learning_rate": 6.989999999999999e-07,
"loss": 0.0926,
"step": 7425
},
{
"epoch": 2.38,
"learning_rate": 6.97611111111111e-07,
"loss": 0.0999,
"step": 7450
},
{
"epoch": 2.39,
"learning_rate": 6.962222222222222e-07,
"loss": 0.0857,
"step": 7475
},
{
"epoch": 2.4,
"learning_rate": 6.948333333333333e-07,
"loss": 0.0948,
"step": 7500
},
{
"epoch": 2.41,
"learning_rate": 6.934444444444444e-07,
"loss": 0.0961,
"step": 7525
},
{
"epoch": 2.42,
"learning_rate": 6.920555555555555e-07,
"loss": 0.0926,
"step": 7550
},
{
"epoch": 2.42,
"learning_rate": 6.906666666666666e-07,
"loss": 0.0936,
"step": 7575
},
{
"epoch": 2.43,
"learning_rate": 6.892777777777777e-07,
"loss": 0.089,
"step": 7600
},
{
"epoch": 2.44,
"learning_rate": 6.879444444444444e-07,
"loss": 0.0931,
"step": 7625
},
{
"epoch": 2.45,
"learning_rate": 6.865555555555555e-07,
"loss": 0.0931,
"step": 7650
},
{
"epoch": 2.46,
"learning_rate": 6.851666666666667e-07,
"loss": 0.0924,
"step": 7675
},
{
"epoch": 2.46,
"learning_rate": 6.837777777777778e-07,
"loss": 0.0904,
"step": 7700
},
{
"epoch": 2.47,
"learning_rate": 6.823888888888889e-07,
"loss": 0.1014,
"step": 7725
},
{
"epoch": 2.48,
"learning_rate": 6.81e-07,
"loss": 0.0919,
"step": 7750
},
{
"epoch": 2.49,
"learning_rate": 6.796111111111111e-07,
"loss": 0.0939,
"step": 7775
},
{
"epoch": 2.5,
"learning_rate": 6.782222222222221e-07,
"loss": 0.0941,
"step": 7800
},
{
"epoch": 2.5,
"learning_rate": 6.768333333333332e-07,
"loss": 0.0847,
"step": 7825
},
{
"epoch": 2.51,
"learning_rate": 6.754444444444443e-07,
"loss": 0.0895,
"step": 7850
},
{
"epoch": 2.52,
"learning_rate": 6.740555555555555e-07,
"loss": 0.0973,
"step": 7875
},
{
"epoch": 2.53,
"learning_rate": 6.726666666666666e-07,
"loss": 0.0916,
"step": 7900
},
{
"epoch": 2.54,
"learning_rate": 6.712777777777777e-07,
"loss": 0.0934,
"step": 7925
},
{
"epoch": 2.54,
"learning_rate": 6.698888888888888e-07,
"loss": 0.0956,
"step": 7950
},
{
"epoch": 2.55,
"learning_rate": 6.684999999999999e-07,
"loss": 0.0901,
"step": 7975
},
{
"epoch": 2.56,
"learning_rate": 6.67111111111111e-07,
"loss": 0.1005,
"step": 8000
},
{
"epoch": 2.56,
"eval_cer": 1.0554799240323367,
"eval_loss": 0.13198651373386383,
"eval_runtime": 941.8725,
"eval_samples_per_second": 1.062,
"eval_steps_per_second": 0.133,
"eval_wer": 3.7010962486171173,
"step": 8000
},
{
"epoch": 2.56,
"step": 8000,
"total_flos": 5.435275349827584e+20,
"train_loss": 0.1500811086446047,
"train_runtime": 62196.6088,
"train_samples_per_second": 10.29,
"train_steps_per_second": 0.322
}
],
"max_steps": 20000,
"num_train_epochs": 7,
"total_flos": 5.435275349827584e+20,
"trial_name": null,
"trial_params": null
}