whisper-large-et-ERR2020-v2 / trainer_state.json
agnesluhtaru's picture
End of training
851d27b
raw
history blame
No virus
51 kB
{
"best_metric": 13.86399041811165,
"best_model_checkpoint": "models/whisper-large-et-ERR2020-v2/checkpoint-9000",
"epoch": 1.232,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 2.3000000000000002e-07,
"loss": 1.0838,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 4.800000000000001e-07,
"loss": 0.9859,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 7.2e-07,
"loss": 0.8297,
"step": 75
},
{
"epoch": 0.01,
"learning_rate": 9.7e-07,
"loss": 0.6431,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 1.21e-06,
"loss": 0.5578,
"step": 125
},
{
"epoch": 0.01,
"learning_rate": 1.46e-06,
"loss": 0.5252,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 1.7100000000000004e-06,
"loss": 0.5095,
"step": 175
},
{
"epoch": 0.02,
"learning_rate": 1.9600000000000003e-06,
"loss": 0.5056,
"step": 200
},
{
"epoch": 0.02,
"learning_rate": 2.21e-06,
"loss": 0.478,
"step": 225
},
{
"epoch": 0.03,
"learning_rate": 2.46e-06,
"loss": 0.4534,
"step": 250
},
{
"epoch": 0.03,
"learning_rate": 2.7100000000000003e-06,
"loss": 0.4758,
"step": 275
},
{
"epoch": 0.03,
"learning_rate": 2.96e-06,
"loss": 0.4468,
"step": 300
},
{
"epoch": 0.03,
"learning_rate": 3.21e-06,
"loss": 0.4147,
"step": 325
},
{
"epoch": 0.04,
"learning_rate": 3.46e-06,
"loss": 0.407,
"step": 350
},
{
"epoch": 0.04,
"learning_rate": 3.7100000000000005e-06,
"loss": 0.4065,
"step": 375
},
{
"epoch": 0.04,
"learning_rate": 3.96e-06,
"loss": 0.3924,
"step": 400
},
{
"epoch": 0.04,
"learning_rate": 4.2000000000000004e-06,
"loss": 0.3786,
"step": 425
},
{
"epoch": 0.04,
"learning_rate": 4.450000000000001e-06,
"loss": 0.3851,
"step": 450
},
{
"epoch": 0.05,
"learning_rate": 4.7e-06,
"loss": 0.3962,
"step": 475
},
{
"epoch": 0.05,
"learning_rate": 4.95e-06,
"loss": 0.3683,
"step": 500
},
{
"epoch": 0.05,
"learning_rate": 5.2e-06,
"loss": 0.3327,
"step": 525
},
{
"epoch": 0.06,
"learning_rate": 5.450000000000001e-06,
"loss": 0.3425,
"step": 550
},
{
"epoch": 0.06,
"learning_rate": 5.7e-06,
"loss": 0.3189,
"step": 575
},
{
"epoch": 0.06,
"learning_rate": 5.950000000000001e-06,
"loss": 0.3302,
"step": 600
},
{
"epoch": 0.06,
"learning_rate": 6.200000000000001e-06,
"loss": 0.2902,
"step": 625
},
{
"epoch": 0.07,
"learning_rate": 6.450000000000001e-06,
"loss": 0.2975,
"step": 650
},
{
"epoch": 0.07,
"learning_rate": 6.700000000000001e-06,
"loss": 0.2848,
"step": 675
},
{
"epoch": 0.07,
"learning_rate": 6.95e-06,
"loss": 0.281,
"step": 700
},
{
"epoch": 0.07,
"learning_rate": 7.2000000000000005e-06,
"loss": 0.2638,
"step": 725
},
{
"epoch": 0.07,
"learning_rate": 7.450000000000001e-06,
"loss": 0.2566,
"step": 750
},
{
"epoch": 0.08,
"learning_rate": 7.7e-06,
"loss": 0.256,
"step": 775
},
{
"epoch": 0.08,
"learning_rate": 7.950000000000002e-06,
"loss": 0.266,
"step": 800
},
{
"epoch": 0.08,
"learning_rate": 8.2e-06,
"loss": 0.2562,
"step": 825
},
{
"epoch": 0.09,
"learning_rate": 8.45e-06,
"loss": 0.238,
"step": 850
},
{
"epoch": 0.09,
"learning_rate": 8.700000000000001e-06,
"loss": 0.2394,
"step": 875
},
{
"epoch": 0.09,
"learning_rate": 8.95e-06,
"loss": 0.2342,
"step": 900
},
{
"epoch": 0.09,
"learning_rate": 9.200000000000002e-06,
"loss": 0.2234,
"step": 925
},
{
"epoch": 0.1,
"learning_rate": 9.450000000000001e-06,
"loss": 0.2032,
"step": 950
},
{
"epoch": 0.1,
"learning_rate": 9.7e-06,
"loss": 0.2178,
"step": 975
},
{
"epoch": 0.1,
"learning_rate": 9.950000000000001e-06,
"loss": 0.2158,
"step": 1000
},
{
"epoch": 0.1,
"eval_loss": 0.3204958140850067,
"eval_runtime": 17202.9952,
"eval_samples_per_second": 0.46,
"eval_steps_per_second": 0.46,
"eval_wer": 23.81537642573645,
"step": 1000
},
{
"epoch": 0.1,
"learning_rate": 9.977777777777778e-06,
"loss": 0.2109,
"step": 1025
},
{
"epoch": 0.1,
"learning_rate": 9.950000000000001e-06,
"loss": 0.1975,
"step": 1050
},
{
"epoch": 0.11,
"learning_rate": 9.922222222222222e-06,
"loss": 0.1858,
"step": 1075
},
{
"epoch": 0.11,
"learning_rate": 9.894444444444445e-06,
"loss": 0.1897,
"step": 1100
},
{
"epoch": 0.11,
"learning_rate": 9.866666666666668e-06,
"loss": 0.1747,
"step": 1125
},
{
"epoch": 0.12,
"learning_rate": 9.83888888888889e-06,
"loss": 0.1843,
"step": 1150
},
{
"epoch": 0.12,
"learning_rate": 9.811111111111112e-06,
"loss": 0.1752,
"step": 1175
},
{
"epoch": 0.12,
"learning_rate": 9.783333333333335e-06,
"loss": 0.1741,
"step": 1200
},
{
"epoch": 0.12,
"learning_rate": 9.755555555555556e-06,
"loss": 0.1627,
"step": 1225
},
{
"epoch": 0.12,
"learning_rate": 9.727777777777777e-06,
"loss": 0.1712,
"step": 1250
},
{
"epoch": 0.13,
"learning_rate": 9.7e-06,
"loss": 0.1551,
"step": 1275
},
{
"epoch": 0.13,
"learning_rate": 9.672222222222223e-06,
"loss": 0.1675,
"step": 1300
},
{
"epoch": 0.13,
"learning_rate": 9.644444444444444e-06,
"loss": 0.1495,
"step": 1325
},
{
"epoch": 0.14,
"learning_rate": 9.616666666666667e-06,
"loss": 0.1462,
"step": 1350
},
{
"epoch": 0.14,
"learning_rate": 9.58888888888889e-06,
"loss": 0.1503,
"step": 1375
},
{
"epoch": 0.14,
"learning_rate": 9.561111111111113e-06,
"loss": 0.1384,
"step": 1400
},
{
"epoch": 0.14,
"learning_rate": 9.533333333333334e-06,
"loss": 0.1518,
"step": 1425
},
{
"epoch": 0.14,
"learning_rate": 9.505555555555557e-06,
"loss": 0.1313,
"step": 1450
},
{
"epoch": 0.15,
"learning_rate": 9.47777777777778e-06,
"loss": 0.1411,
"step": 1475
},
{
"epoch": 0.15,
"learning_rate": 9.450000000000001e-06,
"loss": 0.1341,
"step": 1500
},
{
"epoch": 0.15,
"learning_rate": 9.422222222222222e-06,
"loss": 0.1297,
"step": 1525
},
{
"epoch": 0.15,
"learning_rate": 9.394444444444445e-06,
"loss": 0.1281,
"step": 1550
},
{
"epoch": 0.16,
"learning_rate": 9.366666666666668e-06,
"loss": 0.1425,
"step": 1575
},
{
"epoch": 0.16,
"learning_rate": 9.338888888888889e-06,
"loss": 0.125,
"step": 1600
},
{
"epoch": 0.16,
"learning_rate": 9.311111111111112e-06,
"loss": 0.123,
"step": 1625
},
{
"epoch": 0.17,
"learning_rate": 9.283333333333335e-06,
"loss": 0.1221,
"step": 1650
},
{
"epoch": 0.17,
"learning_rate": 9.255555555555556e-06,
"loss": 0.1233,
"step": 1675
},
{
"epoch": 0.17,
"learning_rate": 9.227777777777779e-06,
"loss": 0.117,
"step": 1700
},
{
"epoch": 0.17,
"learning_rate": 9.200000000000002e-06,
"loss": 0.1064,
"step": 1725
},
{
"epoch": 0.17,
"learning_rate": 9.172222222222223e-06,
"loss": 0.1223,
"step": 1750
},
{
"epoch": 0.18,
"learning_rate": 9.144444444444444e-06,
"loss": 0.121,
"step": 1775
},
{
"epoch": 0.18,
"learning_rate": 9.116666666666667e-06,
"loss": 0.1117,
"step": 1800
},
{
"epoch": 0.18,
"learning_rate": 9.08888888888889e-06,
"loss": 0.1091,
"step": 1825
},
{
"epoch": 0.18,
"learning_rate": 9.061111111111111e-06,
"loss": 0.1059,
"step": 1850
},
{
"epoch": 0.19,
"learning_rate": 9.033333333333334e-06,
"loss": 0.1106,
"step": 1875
},
{
"epoch": 0.19,
"learning_rate": 9.005555555555557e-06,
"loss": 0.1036,
"step": 1900
},
{
"epoch": 0.19,
"learning_rate": 8.977777777777778e-06,
"loss": 0.1043,
"step": 1925
},
{
"epoch": 0.2,
"learning_rate": 8.95e-06,
"loss": 0.1112,
"step": 1950
},
{
"epoch": 0.2,
"learning_rate": 8.922222222222224e-06,
"loss": 0.1056,
"step": 1975
},
{
"epoch": 0.2,
"learning_rate": 8.894444444444445e-06,
"loss": 0.0897,
"step": 2000
},
{
"epoch": 0.2,
"eval_loss": 0.29614511132240295,
"eval_runtime": 17293.2366,
"eval_samples_per_second": 0.458,
"eval_steps_per_second": 0.458,
"eval_wer": 18.333972647779493,
"step": 2000
},
{
"epoch": 0.2,
"learning_rate": 8.866666666666668e-06,
"loss": 0.1163,
"step": 2025
},
{
"epoch": 0.2,
"learning_rate": 8.838888888888889e-06,
"loss": 0.1011,
"step": 2050
},
{
"epoch": 0.21,
"learning_rate": 8.811111111111112e-06,
"loss": 0.1013,
"step": 2075
},
{
"epoch": 0.21,
"learning_rate": 8.783333333333335e-06,
"loss": 0.0966,
"step": 2100
},
{
"epoch": 0.21,
"learning_rate": 8.755555555555556e-06,
"loss": 0.0933,
"step": 2125
},
{
"epoch": 0.21,
"learning_rate": 8.727777777777779e-06,
"loss": 0.0996,
"step": 2150
},
{
"epoch": 0.22,
"learning_rate": 8.700000000000001e-06,
"loss": 0.0907,
"step": 2175
},
{
"epoch": 0.22,
"learning_rate": 8.672222222222223e-06,
"loss": 0.0958,
"step": 2200
},
{
"epoch": 0.22,
"learning_rate": 8.644444444444445e-06,
"loss": 0.0864,
"step": 2225
},
{
"epoch": 0.23,
"learning_rate": 8.616666666666668e-06,
"loss": 0.0897,
"step": 2250
},
{
"epoch": 0.23,
"learning_rate": 8.58888888888889e-06,
"loss": 0.0915,
"step": 2275
},
{
"epoch": 0.23,
"learning_rate": 8.561111111111112e-06,
"loss": 0.0935,
"step": 2300
},
{
"epoch": 0.23,
"learning_rate": 8.533333333333335e-06,
"loss": 0.0957,
"step": 2325
},
{
"epoch": 0.23,
"learning_rate": 8.505555555555556e-06,
"loss": 0.0972,
"step": 2350
},
{
"epoch": 0.24,
"learning_rate": 8.477777777777778e-06,
"loss": 0.1013,
"step": 2375
},
{
"epoch": 0.24,
"learning_rate": 8.45e-06,
"loss": 0.0937,
"step": 2400
},
{
"epoch": 0.24,
"learning_rate": 8.422222222222223e-06,
"loss": 0.0866,
"step": 2425
},
{
"epoch": 0.24,
"learning_rate": 8.394444444444444e-06,
"loss": 0.0988,
"step": 2450
},
{
"epoch": 0.25,
"learning_rate": 8.366666666666667e-06,
"loss": 0.0876,
"step": 2475
},
{
"epoch": 0.25,
"learning_rate": 8.33888888888889e-06,
"loss": 0.0744,
"step": 2500
},
{
"epoch": 0.25,
"learning_rate": 8.311111111111111e-06,
"loss": 0.0933,
"step": 2525
},
{
"epoch": 0.26,
"learning_rate": 8.283333333333334e-06,
"loss": 0.0789,
"step": 2550
},
{
"epoch": 0.26,
"learning_rate": 8.255555555555557e-06,
"loss": 0.0791,
"step": 2575
},
{
"epoch": 0.26,
"learning_rate": 8.227777777777778e-06,
"loss": 0.0739,
"step": 2600
},
{
"epoch": 0.26,
"learning_rate": 8.2e-06,
"loss": 0.0791,
"step": 2625
},
{
"epoch": 0.27,
"learning_rate": 8.172222222222222e-06,
"loss": 0.0864,
"step": 2650
},
{
"epoch": 0.27,
"learning_rate": 8.144444444444445e-06,
"loss": 0.0771,
"step": 2675
},
{
"epoch": 0.27,
"learning_rate": 8.116666666666666e-06,
"loss": 0.0928,
"step": 2700
},
{
"epoch": 0.27,
"learning_rate": 8.08888888888889e-06,
"loss": 0.0834,
"step": 2725
},
{
"epoch": 0.28,
"learning_rate": 8.061111111111112e-06,
"loss": 0.0715,
"step": 2750
},
{
"epoch": 0.28,
"learning_rate": 8.033333333333335e-06,
"loss": 0.0827,
"step": 2775
},
{
"epoch": 0.28,
"learning_rate": 8.005555555555556e-06,
"loss": 0.0871,
"step": 2800
},
{
"epoch": 0.28,
"learning_rate": 7.977777777777779e-06,
"loss": 0.0678,
"step": 2825
},
{
"epoch": 0.28,
"learning_rate": 7.950000000000002e-06,
"loss": 0.0802,
"step": 2850
},
{
"epoch": 0.29,
"learning_rate": 7.922222222222223e-06,
"loss": 0.083,
"step": 2875
},
{
"epoch": 0.29,
"learning_rate": 7.894444444444444e-06,
"loss": 0.0767,
"step": 2900
},
{
"epoch": 0.29,
"learning_rate": 7.866666666666667e-06,
"loss": 0.0873,
"step": 2925
},
{
"epoch": 0.29,
"learning_rate": 7.83888888888889e-06,
"loss": 0.0749,
"step": 2950
},
{
"epoch": 0.3,
"learning_rate": 7.811111111111111e-06,
"loss": 0.0771,
"step": 2975
},
{
"epoch": 0.3,
"learning_rate": 7.783333333333334e-06,
"loss": 0.0785,
"step": 3000
},
{
"epoch": 0.3,
"eval_loss": 0.2839198112487793,
"eval_runtime": 17462.7006,
"eval_samples_per_second": 0.453,
"eval_steps_per_second": 0.453,
"eval_wer": 17.52295660751051,
"step": 3000
},
{
"epoch": 0.3,
"learning_rate": 7.755555555555557e-06,
"loss": 0.0782,
"step": 3025
},
{
"epoch": 0.3,
"learning_rate": 7.727777777777778e-06,
"loss": 0.0751,
"step": 3050
},
{
"epoch": 0.31,
"learning_rate": 7.7e-06,
"loss": 0.0756,
"step": 3075
},
{
"epoch": 0.31,
"learning_rate": 7.672222222222224e-06,
"loss": 0.0801,
"step": 3100
},
{
"epoch": 0.31,
"learning_rate": 7.644444444444445e-06,
"loss": 0.081,
"step": 3125
},
{
"epoch": 0.32,
"learning_rate": 7.616666666666668e-06,
"loss": 0.0809,
"step": 3150
},
{
"epoch": 0.32,
"learning_rate": 7.588888888888889e-06,
"loss": 0.0768,
"step": 3175
},
{
"epoch": 0.32,
"learning_rate": 7.561111111111111e-06,
"loss": 0.0741,
"step": 3200
},
{
"epoch": 0.32,
"learning_rate": 7.533333333333334e-06,
"loss": 0.0761,
"step": 3225
},
{
"epoch": 0.33,
"learning_rate": 7.505555555555556e-06,
"loss": 0.0717,
"step": 3250
},
{
"epoch": 0.33,
"learning_rate": 7.477777777777779e-06,
"loss": 0.0694,
"step": 3275
},
{
"epoch": 0.33,
"learning_rate": 7.450000000000001e-06,
"loss": 0.0777,
"step": 3300
},
{
"epoch": 0.33,
"learning_rate": 7.422222222222223e-06,
"loss": 0.0698,
"step": 3325
},
{
"epoch": 0.34,
"learning_rate": 7.3944444444444456e-06,
"loss": 0.0831,
"step": 3350
},
{
"epoch": 0.34,
"learning_rate": 7.3666666666666676e-06,
"loss": 0.0715,
"step": 3375
},
{
"epoch": 0.34,
"learning_rate": 7.33888888888889e-06,
"loss": 0.0797,
"step": 3400
},
{
"epoch": 0.34,
"learning_rate": 7.3111111111111125e-06,
"loss": 0.0706,
"step": 3425
},
{
"epoch": 0.34,
"learning_rate": 7.2833333333333345e-06,
"loss": 0.0756,
"step": 3450
},
{
"epoch": 0.35,
"learning_rate": 7.255555555555556e-06,
"loss": 0.0693,
"step": 3475
},
{
"epoch": 0.35,
"learning_rate": 7.227777777777778e-06,
"loss": 0.0754,
"step": 3500
},
{
"epoch": 0.35,
"learning_rate": 7.2000000000000005e-06,
"loss": 0.0733,
"step": 3525
},
{
"epoch": 0.35,
"learning_rate": 7.1722222222222225e-06,
"loss": 0.07,
"step": 3550
},
{
"epoch": 0.36,
"learning_rate": 7.1444444444444446e-06,
"loss": 0.0714,
"step": 3575
},
{
"epoch": 0.36,
"learning_rate": 7.116666666666667e-06,
"loss": 0.0717,
"step": 3600
},
{
"epoch": 0.36,
"learning_rate": 7.0888888888888894e-06,
"loss": 0.0724,
"step": 3625
},
{
"epoch": 0.36,
"learning_rate": 7.061111111111112e-06,
"loss": 0.0727,
"step": 3650
},
{
"epoch": 0.37,
"learning_rate": 7.033333333333334e-06,
"loss": 0.0736,
"step": 3675
},
{
"epoch": 0.37,
"learning_rate": 7.005555555555556e-06,
"loss": 0.0712,
"step": 3700
},
{
"epoch": 0.37,
"learning_rate": 6.977777777777779e-06,
"loss": 0.0724,
"step": 3725
},
{
"epoch": 0.38,
"learning_rate": 6.95e-06,
"loss": 0.0688,
"step": 3750
},
{
"epoch": 0.38,
"learning_rate": 6.922222222222222e-06,
"loss": 0.0603,
"step": 3775
},
{
"epoch": 0.38,
"learning_rate": 6.894444444444444e-06,
"loss": 0.0669,
"step": 3800
},
{
"epoch": 0.38,
"learning_rate": 6.866666666666667e-06,
"loss": 0.068,
"step": 3825
},
{
"epoch": 0.39,
"learning_rate": 6.838888888888889e-06,
"loss": 0.0657,
"step": 3850
},
{
"epoch": 0.39,
"learning_rate": 6.811111111111111e-06,
"loss": 0.0685,
"step": 3875
},
{
"epoch": 0.39,
"learning_rate": 6.783333333333334e-06,
"loss": 0.0588,
"step": 3900
},
{
"epoch": 0.39,
"learning_rate": 6.755555555555556e-06,
"loss": 0.0688,
"step": 3925
},
{
"epoch": 0.4,
"learning_rate": 6.727777777777778e-06,
"loss": 0.0699,
"step": 3950
},
{
"epoch": 0.4,
"learning_rate": 6.700000000000001e-06,
"loss": 0.0727,
"step": 3975
},
{
"epoch": 0.4,
"learning_rate": 6.672222222222223e-06,
"loss": 0.0653,
"step": 4000
},
{
"epoch": 0.4,
"eval_loss": 0.28471437096595764,
"eval_runtime": 17373.7787,
"eval_samples_per_second": 0.455,
"eval_steps_per_second": 0.455,
"eval_wer": 17.875231914577153,
"step": 4000
},
{
"epoch": 0.4,
"learning_rate": 6.644444444444444e-06,
"loss": 0.0649,
"step": 4025
},
{
"epoch": 0.41,
"learning_rate": 6.616666666666667e-06,
"loss": 0.0799,
"step": 4050
},
{
"epoch": 0.41,
"learning_rate": 6.588888888888889e-06,
"loss": 0.0721,
"step": 4075
},
{
"epoch": 0.41,
"learning_rate": 6.561111111111111e-06,
"loss": 0.0687,
"step": 4100
},
{
"epoch": 0.41,
"learning_rate": 6.533333333333334e-06,
"loss": 0.0679,
"step": 4125
},
{
"epoch": 0.41,
"learning_rate": 6.505555555555556e-06,
"loss": 0.0743,
"step": 4150
},
{
"epoch": 0.42,
"learning_rate": 6.477777777777778e-06,
"loss": 0.0674,
"step": 4175
},
{
"epoch": 0.42,
"learning_rate": 6.450000000000001e-06,
"loss": 0.061,
"step": 4200
},
{
"epoch": 0.42,
"learning_rate": 6.422222222222223e-06,
"loss": 0.0684,
"step": 4225
},
{
"epoch": 0.42,
"learning_rate": 6.394444444444445e-06,
"loss": 0.0714,
"step": 4250
},
{
"epoch": 0.43,
"learning_rate": 6.366666666666668e-06,
"loss": 0.06,
"step": 4275
},
{
"epoch": 0.43,
"learning_rate": 6.338888888888889e-06,
"loss": 0.0711,
"step": 4300
},
{
"epoch": 0.43,
"learning_rate": 6.311111111111111e-06,
"loss": 0.0618,
"step": 4325
},
{
"epoch": 0.43,
"learning_rate": 6.283333333333334e-06,
"loss": 0.08,
"step": 4350
},
{
"epoch": 0.44,
"learning_rate": 6.255555555555556e-06,
"loss": 0.0659,
"step": 4375
},
{
"epoch": 0.44,
"learning_rate": 6.227777777777778e-06,
"loss": 0.0681,
"step": 4400
},
{
"epoch": 0.44,
"learning_rate": 6.200000000000001e-06,
"loss": 0.0776,
"step": 4425
},
{
"epoch": 0.45,
"learning_rate": 6.172222222222223e-06,
"loss": 0.0611,
"step": 4450
},
{
"epoch": 0.45,
"learning_rate": 6.144444444444445e-06,
"loss": 0.0675,
"step": 4475
},
{
"epoch": 0.45,
"learning_rate": 6.116666666666668e-06,
"loss": 0.0621,
"step": 4500
},
{
"epoch": 0.45,
"learning_rate": 6.08888888888889e-06,
"loss": 0.0697,
"step": 4525
},
{
"epoch": 0.46,
"learning_rate": 6.061111111111112e-06,
"loss": 0.0654,
"step": 4550
},
{
"epoch": 0.46,
"learning_rate": 6.033333333333335e-06,
"loss": 0.0588,
"step": 4575
},
{
"epoch": 0.46,
"learning_rate": 6.005555555555556e-06,
"loss": 0.0728,
"step": 4600
},
{
"epoch": 0.46,
"learning_rate": 5.977777777777778e-06,
"loss": 0.0628,
"step": 4625
},
{
"epoch": 0.47,
"learning_rate": 5.950000000000001e-06,
"loss": 0.06,
"step": 4650
},
{
"epoch": 0.47,
"learning_rate": 5.922222222222223e-06,
"loss": 0.0671,
"step": 4675
},
{
"epoch": 0.47,
"learning_rate": 5.894444444444445e-06,
"loss": 0.062,
"step": 4700
},
{
"epoch": 0.47,
"learning_rate": 5.8666666666666675e-06,
"loss": 0.0619,
"step": 4725
},
{
"epoch": 0.47,
"learning_rate": 5.84e-06,
"loss": 0.0653,
"step": 4750
},
{
"epoch": 0.48,
"learning_rate": 5.812222222222223e-06,
"loss": 0.0729,
"step": 4775
},
{
"epoch": 0.48,
"learning_rate": 5.784444444444445e-06,
"loss": 0.0626,
"step": 4800
},
{
"epoch": 0.48,
"learning_rate": 5.756666666666667e-06,
"loss": 0.0587,
"step": 4825
},
{
"epoch": 0.48,
"learning_rate": 5.72888888888889e-06,
"loss": 0.0581,
"step": 4850
},
{
"epoch": 0.49,
"learning_rate": 5.701111111111112e-06,
"loss": 0.0617,
"step": 4875
},
{
"epoch": 0.49,
"learning_rate": 5.673333333333334e-06,
"loss": 0.0664,
"step": 4900
},
{
"epoch": 0.49,
"learning_rate": 5.645555555555557e-06,
"loss": 0.0725,
"step": 4925
},
{
"epoch": 0.49,
"learning_rate": 5.617777777777779e-06,
"loss": 0.0685,
"step": 4950
},
{
"epoch": 0.5,
"learning_rate": 5.590000000000001e-06,
"loss": 0.0612,
"step": 4975
},
{
"epoch": 0.5,
"learning_rate": 5.562222222222222e-06,
"loss": 0.0541,
"step": 5000
},
{
"epoch": 0.5,
"eval_loss": 0.290607213973999,
"eval_runtime": 17407.0825,
"eval_samples_per_second": 0.455,
"eval_steps_per_second": 0.455,
"eval_wer": 15.264480472205477,
"step": 5000
},
{
"epoch": 0.5,
"learning_rate": 5.534444444444445e-06,
"loss": 0.0621,
"step": 5025
},
{
"epoch": 0.51,
"learning_rate": 5.506666666666667e-06,
"loss": 0.0652,
"step": 5050
},
{
"epoch": 0.51,
"learning_rate": 5.478888888888889e-06,
"loss": 0.0546,
"step": 5075
},
{
"epoch": 0.51,
"learning_rate": 5.451111111111112e-06,
"loss": 0.0532,
"step": 5100
},
{
"epoch": 0.51,
"learning_rate": 5.423333333333334e-06,
"loss": 0.0606,
"step": 5125
},
{
"epoch": 0.52,
"learning_rate": 5.3955555555555565e-06,
"loss": 0.0652,
"step": 5150
},
{
"epoch": 0.52,
"learning_rate": 5.3677777777777785e-06,
"loss": 0.0658,
"step": 5175
},
{
"epoch": 0.52,
"learning_rate": 5.3400000000000005e-06,
"loss": 0.063,
"step": 5200
},
{
"epoch": 0.52,
"learning_rate": 5.312222222222223e-06,
"loss": 0.0659,
"step": 5225
},
{
"epoch": 0.53,
"learning_rate": 5.2844444444444454e-06,
"loss": 0.0684,
"step": 5250
},
{
"epoch": 0.53,
"learning_rate": 5.256666666666667e-06,
"loss": 0.0536,
"step": 5275
},
{
"epoch": 0.53,
"learning_rate": 5.228888888888889e-06,
"loss": 0.0639,
"step": 5300
},
{
"epoch": 0.53,
"learning_rate": 5.2011111111111115e-06,
"loss": 0.0586,
"step": 5325
},
{
"epoch": 0.54,
"learning_rate": 5.1733333333333335e-06,
"loss": 0.0666,
"step": 5350
},
{
"epoch": 0.54,
"learning_rate": 5.1455555555555555e-06,
"loss": 0.061,
"step": 5375
},
{
"epoch": 0.54,
"learning_rate": 5.117777777777778e-06,
"loss": 0.0702,
"step": 5400
},
{
"epoch": 0.54,
"learning_rate": 5.09e-06,
"loss": 0.0569,
"step": 5425
},
{
"epoch": 0.55,
"learning_rate": 5.062222222222222e-06,
"loss": 0.0608,
"step": 5450
},
{
"epoch": 0.55,
"learning_rate": 5.034444444444445e-06,
"loss": 0.0559,
"step": 5475
},
{
"epoch": 0.55,
"learning_rate": 5.006666666666667e-06,
"loss": 0.0577,
"step": 5500
},
{
"epoch": 0.55,
"learning_rate": 4.978888888888889e-06,
"loss": 0.0595,
"step": 5525
},
{
"epoch": 0.56,
"learning_rate": 4.951111111111111e-06,
"loss": 0.0584,
"step": 5550
},
{
"epoch": 0.56,
"learning_rate": 4.923333333333334e-06,
"loss": 0.0646,
"step": 5575
},
{
"epoch": 0.56,
"learning_rate": 4.895555555555556e-06,
"loss": 0.056,
"step": 5600
},
{
"epoch": 0.56,
"learning_rate": 4.867777777777778e-06,
"loss": 0.0628,
"step": 5625
},
{
"epoch": 0.56,
"learning_rate": 4.84e-06,
"loss": 0.0627,
"step": 5650
},
{
"epoch": 0.57,
"learning_rate": 4.812222222222222e-06,
"loss": 0.0627,
"step": 5675
},
{
"epoch": 0.57,
"learning_rate": 4.784444444444445e-06,
"loss": 0.0558,
"step": 5700
},
{
"epoch": 0.57,
"learning_rate": 4.756666666666667e-06,
"loss": 0.0603,
"step": 5725
},
{
"epoch": 0.57,
"learning_rate": 4.728888888888889e-06,
"loss": 0.0678,
"step": 5750
},
{
"epoch": 0.58,
"learning_rate": 4.701111111111111e-06,
"loss": 0.0576,
"step": 5775
},
{
"epoch": 0.58,
"learning_rate": 4.673333333333333e-06,
"loss": 0.0618,
"step": 5800
},
{
"epoch": 0.58,
"learning_rate": 4.645555555555556e-06,
"loss": 0.0646,
"step": 5825
},
{
"epoch": 0.58,
"learning_rate": 4.617777777777778e-06,
"loss": 0.0588,
"step": 5850
},
{
"epoch": 0.59,
"learning_rate": 4.590000000000001e-06,
"loss": 0.0588,
"step": 5875
},
{
"epoch": 0.59,
"learning_rate": 4.562222222222222e-06,
"loss": 0.0613,
"step": 5900
},
{
"epoch": 0.59,
"learning_rate": 4.534444444444445e-06,
"loss": 0.0607,
"step": 5925
},
{
"epoch": 0.59,
"learning_rate": 4.506666666666667e-06,
"loss": 0.0623,
"step": 5950
},
{
"epoch": 0.6,
"learning_rate": 4.478888888888889e-06,
"loss": 0.0544,
"step": 5975
},
{
"epoch": 0.6,
"learning_rate": 4.451111111111112e-06,
"loss": 0.0566,
"step": 6000
},
{
"epoch": 0.6,
"eval_loss": 0.2844848334789276,
"eval_runtime": 17537.6663,
"eval_samples_per_second": 0.451,
"eval_steps_per_second": 0.451,
"eval_wer": 15.208116423074816,
"step": 6000
},
{
"epoch": 0.6,
"learning_rate": 4.423333333333334e-06,
"loss": 0.0531,
"step": 6025
},
{
"epoch": 0.6,
"learning_rate": 4.395555555555556e-06,
"loss": 0.0522,
"step": 6050
},
{
"epoch": 0.61,
"learning_rate": 4.367777777777778e-06,
"loss": 0.0715,
"step": 6075
},
{
"epoch": 0.61,
"learning_rate": 4.34e-06,
"loss": 0.057,
"step": 6100
},
{
"epoch": 0.61,
"learning_rate": 4.312222222222223e-06,
"loss": 0.0635,
"step": 6125
},
{
"epoch": 0.61,
"learning_rate": 4.284444444444445e-06,
"loss": 0.0596,
"step": 6150
},
{
"epoch": 0.62,
"learning_rate": 4.256666666666668e-06,
"loss": 0.0669,
"step": 6175
},
{
"epoch": 0.62,
"learning_rate": 4.228888888888889e-06,
"loss": 0.0529,
"step": 6200
},
{
"epoch": 0.62,
"learning_rate": 4.201111111111112e-06,
"loss": 0.0632,
"step": 6225
},
{
"epoch": 0.62,
"learning_rate": 4.173333333333334e-06,
"loss": 0.0604,
"step": 6250
},
{
"epoch": 0.63,
"learning_rate": 4.145555555555556e-06,
"loss": 0.0578,
"step": 6275
},
{
"epoch": 0.63,
"learning_rate": 4.117777777777779e-06,
"loss": 0.052,
"step": 6300
},
{
"epoch": 0.63,
"learning_rate": 4.09e-06,
"loss": 0.0588,
"step": 6325
},
{
"epoch": 0.64,
"learning_rate": 4.062222222222223e-06,
"loss": 0.0662,
"step": 6350
},
{
"epoch": 0.64,
"learning_rate": 4.034444444444445e-06,
"loss": 0.0623,
"step": 6375
},
{
"epoch": 0.64,
"learning_rate": 4.006666666666667e-06,
"loss": 0.0635,
"step": 6400
},
{
"epoch": 0.64,
"learning_rate": 3.9788888888888896e-06,
"loss": 0.0583,
"step": 6425
},
{
"epoch": 0.65,
"learning_rate": 3.951111111111112e-06,
"loss": 0.052,
"step": 6450
},
{
"epoch": 0.65,
"learning_rate": 3.923333333333334e-06,
"loss": 0.0597,
"step": 6475
},
{
"epoch": 0.65,
"learning_rate": 3.895555555555556e-06,
"loss": 0.0535,
"step": 6500
},
{
"epoch": 0.65,
"learning_rate": 3.8677777777777785e-06,
"loss": 0.0569,
"step": 6525
},
{
"epoch": 0.66,
"learning_rate": 3.8400000000000005e-06,
"loss": 0.0623,
"step": 6550
},
{
"epoch": 0.66,
"learning_rate": 3.8122222222222225e-06,
"loss": 0.0611,
"step": 6575
},
{
"epoch": 0.66,
"learning_rate": 3.784444444444445e-06,
"loss": 0.0591,
"step": 6600
},
{
"epoch": 0.66,
"learning_rate": 3.756666666666667e-06,
"loss": 0.063,
"step": 6625
},
{
"epoch": 0.67,
"learning_rate": 3.728888888888889e-06,
"loss": 0.0539,
"step": 6650
},
{
"epoch": 0.67,
"learning_rate": 3.7011111111111114e-06,
"loss": 0.0552,
"step": 6675
},
{
"epoch": 0.67,
"learning_rate": 3.673333333333334e-06,
"loss": 0.0627,
"step": 6700
},
{
"epoch": 0.67,
"learning_rate": 3.645555555555556e-06,
"loss": 0.061,
"step": 6725
},
{
"epoch": 0.68,
"learning_rate": 3.617777777777778e-06,
"loss": 0.0511,
"step": 6750
},
{
"epoch": 0.68,
"learning_rate": 3.5900000000000004e-06,
"loss": 0.0559,
"step": 6775
},
{
"epoch": 0.68,
"learning_rate": 3.5622222222222224e-06,
"loss": 0.054,
"step": 6800
},
{
"epoch": 0.68,
"learning_rate": 3.534444444444445e-06,
"loss": 0.0531,
"step": 6825
},
{
"epoch": 0.69,
"learning_rate": 3.5066666666666673e-06,
"loss": 0.0559,
"step": 6850
},
{
"epoch": 0.69,
"learning_rate": 3.4788888888888893e-06,
"loss": 0.0554,
"step": 6875
},
{
"epoch": 0.69,
"learning_rate": 3.4511111111111113e-06,
"loss": 0.0556,
"step": 6900
},
{
"epoch": 0.69,
"learning_rate": 3.4233333333333333e-06,
"loss": 0.0585,
"step": 6925
},
{
"epoch": 0.69,
"learning_rate": 3.3955555555555558e-06,
"loss": 0.0608,
"step": 6950
},
{
"epoch": 0.7,
"learning_rate": 3.367777777777778e-06,
"loss": 0.0624,
"step": 6975
},
{
"epoch": 0.7,
"learning_rate": 3.3400000000000006e-06,
"loss": 0.051,
"step": 7000
},
{
"epoch": 0.7,
"eval_loss": 0.2888207733631134,
"eval_runtime": 17509.83,
"eval_samples_per_second": 0.452,
"eval_steps_per_second": 0.452,
"eval_wer": 14.46677261020346,
"step": 7000
},
{
"epoch": 0.7,
"learning_rate": 3.3122222222222222e-06,
"loss": 0.0553,
"step": 7025
},
{
"epoch": 0.7,
"learning_rate": 3.2844444444444447e-06,
"loss": 0.0503,
"step": 7050
},
{
"epoch": 0.71,
"learning_rate": 3.2566666666666667e-06,
"loss": 0.0522,
"step": 7075
},
{
"epoch": 0.71,
"learning_rate": 3.228888888888889e-06,
"loss": 0.0493,
"step": 7100
},
{
"epoch": 0.71,
"learning_rate": 3.2011111111111116e-06,
"loss": 0.0603,
"step": 7125
},
{
"epoch": 0.71,
"learning_rate": 3.173333333333334e-06,
"loss": 0.055,
"step": 7150
},
{
"epoch": 0.72,
"learning_rate": 3.1455555555555556e-06,
"loss": 0.0563,
"step": 7175
},
{
"epoch": 0.72,
"learning_rate": 3.117777777777778e-06,
"loss": 0.0549,
"step": 7200
},
{
"epoch": 0.72,
"learning_rate": 3.09e-06,
"loss": 0.0617,
"step": 7225
},
{
"epoch": 0.72,
"learning_rate": 3.0622222222222225e-06,
"loss": 0.0558,
"step": 7250
},
{
"epoch": 0.73,
"learning_rate": 3.034444444444445e-06,
"loss": 0.0548,
"step": 7275
},
{
"epoch": 0.73,
"learning_rate": 3.0066666666666674e-06,
"loss": 0.0546,
"step": 7300
},
{
"epoch": 0.73,
"learning_rate": 2.978888888888889e-06,
"loss": 0.0573,
"step": 7325
},
{
"epoch": 0.73,
"learning_rate": 2.9511111111111114e-06,
"loss": 0.0514,
"step": 7350
},
{
"epoch": 0.74,
"learning_rate": 2.9233333333333334e-06,
"loss": 0.057,
"step": 7375
},
{
"epoch": 0.74,
"learning_rate": 2.895555555555556e-06,
"loss": 0.0567,
"step": 7400
},
{
"epoch": 0.74,
"learning_rate": 2.8677777777777783e-06,
"loss": 0.0528,
"step": 7425
},
{
"epoch": 0.74,
"learning_rate": 2.84e-06,
"loss": 0.0512,
"step": 7450
},
{
"epoch": 0.75,
"learning_rate": 2.8122222222222224e-06,
"loss": 0.0473,
"step": 7475
},
{
"epoch": 0.75,
"learning_rate": 2.784444444444445e-06,
"loss": 0.0617,
"step": 7500
},
{
"epoch": 0.75,
"learning_rate": 2.756666666666667e-06,
"loss": 0.0495,
"step": 7525
},
{
"epoch": 0.76,
"learning_rate": 2.7288888888888893e-06,
"loss": 0.0638,
"step": 7550
},
{
"epoch": 0.76,
"learning_rate": 2.7011111111111117e-06,
"loss": 0.049,
"step": 7575
},
{
"epoch": 0.76,
"learning_rate": 2.6733333333333333e-06,
"loss": 0.0571,
"step": 7600
},
{
"epoch": 0.76,
"learning_rate": 2.6455555555555557e-06,
"loss": 0.0546,
"step": 7625
},
{
"epoch": 0.77,
"learning_rate": 2.617777777777778e-06,
"loss": 0.0517,
"step": 7650
},
{
"epoch": 0.77,
"learning_rate": 2.59e-06,
"loss": 0.0508,
"step": 7675
},
{
"epoch": 1.0,
"learning_rate": 2.5622222222222226e-06,
"loss": 0.0578,
"step": 7700
},
{
"epoch": 1.0,
"learning_rate": 2.534444444444445e-06,
"loss": 0.0555,
"step": 7725
},
{
"epoch": 1.01,
"learning_rate": 2.5066666666666667e-06,
"loss": 0.0572,
"step": 7750
},
{
"epoch": 1.01,
"learning_rate": 2.478888888888889e-06,
"loss": 0.0538,
"step": 7775
},
{
"epoch": 1.01,
"learning_rate": 2.451111111111111e-06,
"loss": 0.0569,
"step": 7800
},
{
"epoch": 1.01,
"learning_rate": 2.4233333333333336e-06,
"loss": 0.0466,
"step": 7825
},
{
"epoch": 1.02,
"learning_rate": 2.3955555555555556e-06,
"loss": 0.0497,
"step": 7850
},
{
"epoch": 1.02,
"learning_rate": 2.367777777777778e-06,
"loss": 0.0491,
"step": 7875
},
{
"epoch": 1.02,
"learning_rate": 2.3400000000000005e-06,
"loss": 0.0583,
"step": 7900
},
{
"epoch": 1.02,
"learning_rate": 2.3122222222222225e-06,
"loss": 0.0497,
"step": 7925
},
{
"epoch": 1.03,
"learning_rate": 2.2844444444444445e-06,
"loss": 0.0591,
"step": 7950
},
{
"epoch": 1.03,
"learning_rate": 2.2566666666666665e-06,
"loss": 0.0499,
"step": 7975
},
{
"epoch": 1.03,
"learning_rate": 2.228888888888889e-06,
"loss": 0.049,
"step": 8000
},
{
"epoch": 1.03,
"eval_loss": 0.2927439510822296,
"eval_runtime": 17875.459,
"eval_samples_per_second": 0.443,
"eval_steps_per_second": 0.443,
"eval_wer": 15.313016181179103,
"step": 8000
},
{
"epoch": 1.03,
"learning_rate": 2.2011111111111114e-06,
"loss": 0.0558,
"step": 8025
},
{
"epoch": 1.04,
"learning_rate": 2.1733333333333334e-06,
"loss": 0.0514,
"step": 8050
},
{
"epoch": 1.04,
"learning_rate": 2.145555555555556e-06,
"loss": 0.0548,
"step": 8075
},
{
"epoch": 1.04,
"learning_rate": 2.117777777777778e-06,
"loss": 0.0475,
"step": 8100
},
{
"epoch": 1.04,
"learning_rate": 2.09e-06,
"loss": 0.047,
"step": 8125
},
{
"epoch": 1.05,
"learning_rate": 2.0622222222222223e-06,
"loss": 0.0471,
"step": 8150
},
{
"epoch": 1.05,
"learning_rate": 2.0344444444444448e-06,
"loss": 0.0511,
"step": 8175
},
{
"epoch": 1.05,
"learning_rate": 2.006666666666667e-06,
"loss": 0.0444,
"step": 8200
},
{
"epoch": 1.05,
"learning_rate": 1.9788888888888892e-06,
"loss": 0.0545,
"step": 8225
},
{
"epoch": 1.06,
"learning_rate": 1.9511111111111113e-06,
"loss": 0.0513,
"step": 8250
},
{
"epoch": 1.06,
"learning_rate": 1.9233333333333333e-06,
"loss": 0.0448,
"step": 8275
},
{
"epoch": 1.06,
"learning_rate": 1.8955555555555557e-06,
"loss": 0.0421,
"step": 8300
},
{
"epoch": 1.06,
"learning_rate": 1.8677777777777777e-06,
"loss": 0.0441,
"step": 8325
},
{
"epoch": 1.07,
"learning_rate": 1.8400000000000002e-06,
"loss": 0.0429,
"step": 8350
},
{
"epoch": 1.07,
"learning_rate": 1.8122222222222224e-06,
"loss": 0.0475,
"step": 8375
},
{
"epoch": 1.07,
"learning_rate": 1.7844444444444444e-06,
"loss": 0.0471,
"step": 8400
},
{
"epoch": 1.07,
"learning_rate": 1.7566666666666669e-06,
"loss": 0.0446,
"step": 8425
},
{
"epoch": 1.08,
"learning_rate": 1.728888888888889e-06,
"loss": 0.0452,
"step": 8450
},
{
"epoch": 1.08,
"learning_rate": 1.7011111111111111e-06,
"loss": 0.0487,
"step": 8475
},
{
"epoch": 1.08,
"learning_rate": 1.6733333333333335e-06,
"loss": 0.0426,
"step": 8500
},
{
"epoch": 1.08,
"learning_rate": 1.6455555555555558e-06,
"loss": 0.0449,
"step": 8525
},
{
"epoch": 1.09,
"learning_rate": 1.6177777777777778e-06,
"loss": 0.0462,
"step": 8550
},
{
"epoch": 1.09,
"learning_rate": 1.5900000000000002e-06,
"loss": 0.0458,
"step": 8575
},
{
"epoch": 1.09,
"learning_rate": 1.5622222222222225e-06,
"loss": 0.0442,
"step": 8600
},
{
"epoch": 1.09,
"learning_rate": 1.5344444444444445e-06,
"loss": 0.0431,
"step": 8625
},
{
"epoch": 1.1,
"learning_rate": 1.506666666666667e-06,
"loss": 0.0494,
"step": 8650
},
{
"epoch": 1.1,
"learning_rate": 1.478888888888889e-06,
"loss": 0.0357,
"step": 8675
},
{
"epoch": 1.1,
"learning_rate": 1.4511111111111112e-06,
"loss": 0.0438,
"step": 8700
},
{
"epoch": 1.1,
"learning_rate": 1.4233333333333336e-06,
"loss": 0.0426,
"step": 8725
},
{
"epoch": 1.11,
"learning_rate": 1.3955555555555556e-06,
"loss": 0.0462,
"step": 8750
},
{
"epoch": 1.11,
"learning_rate": 1.3677777777777779e-06,
"loss": 0.0392,
"step": 8775
},
{
"epoch": 1.11,
"learning_rate": 1.34e-06,
"loss": 0.0418,
"step": 8800
},
{
"epoch": 1.11,
"learning_rate": 1.3122222222222223e-06,
"loss": 0.0438,
"step": 8825
},
{
"epoch": 1.12,
"learning_rate": 1.2844444444444445e-06,
"loss": 0.0427,
"step": 8850
},
{
"epoch": 1.12,
"learning_rate": 1.2566666666666668e-06,
"loss": 0.0417,
"step": 8875
},
{
"epoch": 1.12,
"learning_rate": 1.228888888888889e-06,
"loss": 0.0378,
"step": 8900
},
{
"epoch": 1.12,
"learning_rate": 1.2011111111111112e-06,
"loss": 0.0374,
"step": 8925
},
{
"epoch": 1.13,
"learning_rate": 1.1733333333333335e-06,
"loss": 0.0433,
"step": 8950
},
{
"epoch": 1.13,
"learning_rate": 1.1455555555555557e-06,
"loss": 0.0441,
"step": 8975
},
{
"epoch": 1.13,
"learning_rate": 1.117777777777778e-06,
"loss": 0.044,
"step": 9000
},
{
"epoch": 1.13,
"eval_loss": 0.29153117537498474,
"eval_runtime": 17455.8976,
"eval_samples_per_second": 0.453,
"eval_steps_per_second": 0.453,
"eval_wer": 13.86399041811165,
"step": 9000
},
{
"epoch": 1.13,
"learning_rate": 1.0900000000000002e-06,
"loss": 0.0415,
"step": 9025
},
{
"epoch": 1.14,
"learning_rate": 1.0622222222222222e-06,
"loss": 0.0415,
"step": 9050
},
{
"epoch": 1.14,
"learning_rate": 1.0344444444444446e-06,
"loss": 0.0366,
"step": 9075
},
{
"epoch": 1.14,
"learning_rate": 1.0066666666666668e-06,
"loss": 0.0398,
"step": 9100
},
{
"epoch": 1.14,
"learning_rate": 9.788888888888889e-07,
"loss": 0.038,
"step": 9125
},
{
"epoch": 1.15,
"learning_rate": 9.511111111111111e-07,
"loss": 0.0456,
"step": 9150
},
{
"epoch": 1.15,
"learning_rate": 9.233333333333334e-07,
"loss": 0.0439,
"step": 9175
},
{
"epoch": 1.15,
"learning_rate": 8.955555555555557e-07,
"loss": 0.0403,
"step": 9200
},
{
"epoch": 1.15,
"learning_rate": 8.677777777777778e-07,
"loss": 0.0388,
"step": 9225
},
{
"epoch": 1.16,
"learning_rate": 8.400000000000001e-07,
"loss": 0.0493,
"step": 9250
},
{
"epoch": 1.16,
"learning_rate": 8.122222222222223e-07,
"loss": 0.0355,
"step": 9275
},
{
"epoch": 1.16,
"learning_rate": 7.844444444444445e-07,
"loss": 0.0407,
"step": 9300
},
{
"epoch": 1.16,
"learning_rate": 7.566666666666667e-07,
"loss": 0.0344,
"step": 9325
},
{
"epoch": 1.17,
"learning_rate": 7.28888888888889e-07,
"loss": 0.0379,
"step": 9350
},
{
"epoch": 1.17,
"learning_rate": 7.011111111111112e-07,
"loss": 0.0387,
"step": 9375
},
{
"epoch": 1.17,
"learning_rate": 6.733333333333334e-07,
"loss": 0.0351,
"step": 9400
},
{
"epoch": 1.17,
"learning_rate": 6.455555555555556e-07,
"loss": 0.0399,
"step": 9425
},
{
"epoch": 1.18,
"learning_rate": 6.177777777777778e-07,
"loss": 0.0399,
"step": 9450
},
{
"epoch": 1.18,
"learning_rate": 5.900000000000001e-07,
"loss": 0.0369,
"step": 9475
},
{
"epoch": 1.18,
"learning_rate": 5.622222222222223e-07,
"loss": 0.0419,
"step": 9500
},
{
"epoch": 1.18,
"learning_rate": 5.344444444444445e-07,
"loss": 0.0355,
"step": 9525
},
{
"epoch": 1.19,
"learning_rate": 5.066666666666667e-07,
"loss": 0.0399,
"step": 9550
},
{
"epoch": 1.19,
"learning_rate": 4.78888888888889e-07,
"loss": 0.0369,
"step": 9575
},
{
"epoch": 1.19,
"learning_rate": 4.511111111111111e-07,
"loss": 0.0377,
"step": 9600
},
{
"epoch": 1.19,
"learning_rate": 4.233333333333334e-07,
"loss": 0.0376,
"step": 9625
},
{
"epoch": 1.2,
"learning_rate": 3.9555555555555557e-07,
"loss": 0.0351,
"step": 9650
},
{
"epoch": 1.2,
"learning_rate": 3.677777777777778e-07,
"loss": 0.0346,
"step": 9675
},
{
"epoch": 1.2,
"learning_rate": 3.4000000000000003e-07,
"loss": 0.0406,
"step": 9700
},
{
"epoch": 1.2,
"learning_rate": 3.1333333333333333e-07,
"loss": 0.0364,
"step": 9725
},
{
"epoch": 1.21,
"learning_rate": 2.8555555555555556e-07,
"loss": 0.043,
"step": 9750
},
{
"epoch": 1.21,
"learning_rate": 2.577777777777778e-07,
"loss": 0.0361,
"step": 9775
},
{
"epoch": 1.21,
"learning_rate": 2.3000000000000002e-07,
"loss": 0.041,
"step": 9800
},
{
"epoch": 1.21,
"learning_rate": 2.0222222222222222e-07,
"loss": 0.0332,
"step": 9825
},
{
"epoch": 1.22,
"learning_rate": 1.7444444444444445e-07,
"loss": 0.0336,
"step": 9850
},
{
"epoch": 1.22,
"learning_rate": 1.4666666666666668e-07,
"loss": 0.0331,
"step": 9875
},
{
"epoch": 1.22,
"learning_rate": 1.1888888888888889e-07,
"loss": 0.0406,
"step": 9900
},
{
"epoch": 1.22,
"learning_rate": 9.111111111111113e-08,
"loss": 0.0324,
"step": 9925
},
{
"epoch": 1.23,
"learning_rate": 6.333333333333333e-08,
"loss": 0.0332,
"step": 9950
},
{
"epoch": 1.23,
"learning_rate": 3.555555555555556e-08,
"loss": 0.039,
"step": 9975
},
{
"epoch": 1.23,
"learning_rate": 7.777777777777778e-09,
"loss": 0.0379,
"step": 10000
},
{
"epoch": 1.23,
"eval_loss": 0.2912769317626953,
"eval_runtime": 19086.3126,
"eval_samples_per_second": 0.415,
"eval_steps_per_second": 0.415,
"eval_wer": 16.5772931165405,
"step": 10000
},
{
"epoch": 1.23,
"step": 10000,
"total_flos": 6.794550669053952e+20,
"train_loss": 0.008363822835683822,
"train_runtime": 112568.9236,
"train_samples_per_second": 2.843,
"train_steps_per_second": 0.089
}
],
"max_steps": 10000,
"num_train_epochs": 9223372036854775807,
"total_flos": 6.794550669053952e+20,
"trial_name": null,
"trial_params": null
}