whisper-small-es / trainer_state.json
danielizham's picture
End of training
72efb0f
raw
history blame
25.9 kB
{
"best_metric": 8.212281066472636,
"best_model_checkpoint": "./checkpoint-4000",
"epoch": 20.03,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.800000000000001e-07,
"loss": 1.0543,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.800000000000001e-07,
"loss": 0.8847,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.48e-06,
"loss": 0.54,
"step": 75
},
{
"epoch": 0.02,
"learning_rate": 1.98e-06,
"loss": 0.304,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 2.4800000000000004e-06,
"loss": 0.2861,
"step": 125
},
{
"epoch": 0.03,
"learning_rate": 2.9800000000000003e-06,
"loss": 0.2395,
"step": 150
},
{
"epoch": 0.04,
"learning_rate": 3.48e-06,
"loss": 0.2282,
"step": 175
},
{
"epoch": 0.04,
"learning_rate": 3.980000000000001e-06,
"loss": 0.2209,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 4.48e-06,
"loss": 0.2299,
"step": 225
},
{
"epoch": 1.0,
"learning_rate": 4.980000000000001e-06,
"loss": 0.2867,
"step": 250
},
{
"epoch": 1.01,
"learning_rate": 5.480000000000001e-06,
"loss": 0.3646,
"step": 275
},
{
"epoch": 1.01,
"learning_rate": 5.98e-06,
"loss": 0.3278,
"step": 300
},
{
"epoch": 1.02,
"learning_rate": 6.480000000000001e-06,
"loss": 0.3748,
"step": 325
},
{
"epoch": 1.02,
"learning_rate": 6.98e-06,
"loss": 0.2737,
"step": 350
},
{
"epoch": 1.03,
"learning_rate": 7.48e-06,
"loss": 0.2653,
"step": 375
},
{
"epoch": 1.03,
"learning_rate": 7.980000000000002e-06,
"loss": 0.2596,
"step": 400
},
{
"epoch": 1.04,
"learning_rate": 8.48e-06,
"loss": 0.2665,
"step": 425
},
{
"epoch": 1.04,
"learning_rate": 8.98e-06,
"loss": 0.2598,
"step": 450
},
{
"epoch": 1.05,
"learning_rate": 9.48e-06,
"loss": 0.242,
"step": 475
},
{
"epoch": 2.0,
"learning_rate": 9.980000000000001e-06,
"loss": 0.245,
"step": 500
},
{
"epoch": 2.01,
"learning_rate": 9.946666666666667e-06,
"loss": 0.2256,
"step": 525
},
{
"epoch": 2.01,
"learning_rate": 9.891111111111113e-06,
"loss": 0.209,
"step": 550
},
{
"epoch": 2.02,
"learning_rate": 9.835555555555556e-06,
"loss": 0.2339,
"step": 575
},
{
"epoch": 2.02,
"learning_rate": 9.780000000000001e-06,
"loss": 0.1714,
"step": 600
},
{
"epoch": 2.03,
"learning_rate": 9.724444444444445e-06,
"loss": 0.1657,
"step": 625
},
{
"epoch": 2.03,
"learning_rate": 9.66888888888889e-06,
"loss": 0.1688,
"step": 650
},
{
"epoch": 2.04,
"learning_rate": 9.613333333333335e-06,
"loss": 0.1616,
"step": 675
},
{
"epoch": 2.04,
"learning_rate": 9.557777777777777e-06,
"loss": 0.1541,
"step": 700
},
{
"epoch": 2.05,
"learning_rate": 9.502222222222223e-06,
"loss": 0.1565,
"step": 725
},
{
"epoch": 3.0,
"learning_rate": 9.446666666666667e-06,
"loss": 0.1355,
"step": 750
},
{
"epoch": 3.01,
"learning_rate": 9.391111111111111e-06,
"loss": 0.1341,
"step": 775
},
{
"epoch": 3.01,
"learning_rate": 9.335555555555557e-06,
"loss": 0.1286,
"step": 800
},
{
"epoch": 3.02,
"learning_rate": 9.280000000000001e-06,
"loss": 0.1343,
"step": 825
},
{
"epoch": 3.02,
"learning_rate": 9.224444444444445e-06,
"loss": 0.0982,
"step": 850
},
{
"epoch": 3.03,
"learning_rate": 9.168888888888889e-06,
"loss": 0.0957,
"step": 875
},
{
"epoch": 3.03,
"learning_rate": 9.113333333333335e-06,
"loss": 0.1034,
"step": 900
},
{
"epoch": 3.04,
"learning_rate": 9.057777777777779e-06,
"loss": 0.099,
"step": 925
},
{
"epoch": 3.04,
"learning_rate": 9.002222222222223e-06,
"loss": 0.0863,
"step": 950
},
{
"epoch": 4.0,
"learning_rate": 8.946666666666669e-06,
"loss": 0.1075,
"step": 975
},
{
"epoch": 4.01,
"learning_rate": 8.891111111111111e-06,
"loss": 0.1386,
"step": 1000
},
{
"epoch": 4.01,
"eval_loss": 0.24644243717193604,
"eval_runtime": 3384.9122,
"eval_samples_per_second": 4.585,
"eval_steps_per_second": 0.072,
"eval_wer": 9.800036380645496,
"step": 1000
},
{
"epoch": 4.01,
"learning_rate": 8.835555555555557e-06,
"loss": 0.1262,
"step": 1025
},
{
"epoch": 4.02,
"learning_rate": 8.78e-06,
"loss": 0.1261,
"step": 1050
},
{
"epoch": 4.02,
"learning_rate": 8.724444444444445e-06,
"loss": 0.1084,
"step": 1075
},
{
"epoch": 4.03,
"learning_rate": 8.66888888888889e-06,
"loss": 0.1237,
"step": 1100
},
{
"epoch": 4.03,
"learning_rate": 8.613333333333333e-06,
"loss": 0.1078,
"step": 1125
},
{
"epoch": 4.04,
"learning_rate": 8.557777777777778e-06,
"loss": 0.0952,
"step": 1150
},
{
"epoch": 4.04,
"learning_rate": 8.502222222222223e-06,
"loss": 0.0921,
"step": 1175
},
{
"epoch": 4.05,
"learning_rate": 8.446666666666668e-06,
"loss": 0.0994,
"step": 1200
},
{
"epoch": 5.0,
"learning_rate": 8.391111111111112e-06,
"loss": 0.125,
"step": 1225
},
{
"epoch": 5.01,
"learning_rate": 8.335555555555556e-06,
"loss": 0.1309,
"step": 1250
},
{
"epoch": 5.01,
"learning_rate": 8.28e-06,
"loss": 0.1241,
"step": 1275
},
{
"epoch": 5.02,
"learning_rate": 8.224444444444444e-06,
"loss": 0.1299,
"step": 1300
},
{
"epoch": 5.02,
"learning_rate": 8.16888888888889e-06,
"loss": 0.1249,
"step": 1325
},
{
"epoch": 5.03,
"learning_rate": 8.113333333333334e-06,
"loss": 0.1199,
"step": 1350
},
{
"epoch": 5.03,
"learning_rate": 8.057777777777778e-06,
"loss": 0.1405,
"step": 1375
},
{
"epoch": 5.04,
"learning_rate": 8.002222222222222e-06,
"loss": 0.1291,
"step": 1400
},
{
"epoch": 5.04,
"learning_rate": 7.946666666666666e-06,
"loss": 0.1359,
"step": 1425
},
{
"epoch": 5.05,
"learning_rate": 7.891111111111112e-06,
"loss": 0.1306,
"step": 1450
},
{
"epoch": 6.0,
"learning_rate": 7.835555555555556e-06,
"loss": 0.19,
"step": 1475
},
{
"epoch": 6.01,
"learning_rate": 7.78e-06,
"loss": 0.2177,
"step": 1500
},
{
"epoch": 6.01,
"learning_rate": 7.724444444444446e-06,
"loss": 0.1542,
"step": 1525
},
{
"epoch": 6.02,
"learning_rate": 7.66888888888889e-06,
"loss": 0.0816,
"step": 1550
},
{
"epoch": 6.02,
"learning_rate": 7.613333333333334e-06,
"loss": 0.0775,
"step": 1575
},
{
"epoch": 6.03,
"learning_rate": 7.557777777777779e-06,
"loss": 0.0816,
"step": 1600
},
{
"epoch": 6.03,
"learning_rate": 7.502222222222223e-06,
"loss": 0.0829,
"step": 1625
},
{
"epoch": 6.04,
"learning_rate": 7.446666666666668e-06,
"loss": 0.0837,
"step": 1650
},
{
"epoch": 6.04,
"learning_rate": 7.3911111111111125e-06,
"loss": 0.0863,
"step": 1675
},
{
"epoch": 7.0,
"learning_rate": 7.335555555555556e-06,
"loss": 0.0933,
"step": 1700
},
{
"epoch": 7.01,
"learning_rate": 7.280000000000001e-06,
"loss": 0.0797,
"step": 1725
},
{
"epoch": 7.01,
"learning_rate": 7.224444444444445e-06,
"loss": 0.0638,
"step": 1750
},
{
"epoch": 7.02,
"learning_rate": 7.1688888888888895e-06,
"loss": 0.0693,
"step": 1775
},
{
"epoch": 7.02,
"learning_rate": 7.113333333333334e-06,
"loss": 0.0553,
"step": 1800
},
{
"epoch": 7.03,
"learning_rate": 7.057777777777778e-06,
"loss": 0.0637,
"step": 1825
},
{
"epoch": 7.03,
"learning_rate": 7.0022222222222225e-06,
"loss": 0.0556,
"step": 1850
},
{
"epoch": 7.04,
"learning_rate": 6.946666666666667e-06,
"loss": 0.0514,
"step": 1875
},
{
"epoch": 7.04,
"learning_rate": 6.891111111111111e-06,
"loss": 0.0454,
"step": 1900
},
{
"epoch": 7.05,
"learning_rate": 6.835555555555556e-06,
"loss": 0.0505,
"step": 1925
},
{
"epoch": 8.0,
"learning_rate": 6.780000000000001e-06,
"loss": 0.0866,
"step": 1950
},
{
"epoch": 8.01,
"learning_rate": 6.724444444444444e-06,
"loss": 0.128,
"step": 1975
},
{
"epoch": 8.01,
"learning_rate": 6.668888888888889e-06,
"loss": 0.1098,
"step": 2000
},
{
"epoch": 8.01,
"eval_loss": 0.2271677553653717,
"eval_runtime": 3129.8829,
"eval_samples_per_second": 4.959,
"eval_steps_per_second": 0.078,
"eval_wer": 8.622862637077075,
"step": 2000
},
{
"epoch": 8.02,
"learning_rate": 6.613333333333334e-06,
"loss": 0.1169,
"step": 2025
},
{
"epoch": 8.02,
"learning_rate": 6.557777777777778e-06,
"loss": 0.1163,
"step": 2050
},
{
"epoch": 8.03,
"learning_rate": 6.502222222222223e-06,
"loss": 0.1148,
"step": 2075
},
{
"epoch": 8.03,
"learning_rate": 6.446666666666668e-06,
"loss": 0.1126,
"step": 2100
},
{
"epoch": 8.04,
"learning_rate": 6.391111111111111e-06,
"loss": 0.118,
"step": 2125
},
{
"epoch": 8.04,
"learning_rate": 6.335555555555556e-06,
"loss": 0.1091,
"step": 2150
},
{
"epoch": 8.05,
"learning_rate": 6.280000000000001e-06,
"loss": 0.1079,
"step": 2175
},
{
"epoch": 9.0,
"learning_rate": 6.224444444444445e-06,
"loss": 0.0869,
"step": 2200
},
{
"epoch": 9.01,
"learning_rate": 6.16888888888889e-06,
"loss": 0.0694,
"step": 2225
},
{
"epoch": 9.01,
"learning_rate": 6.113333333333333e-06,
"loss": 0.0647,
"step": 2250
},
{
"epoch": 9.02,
"learning_rate": 6.057777777777778e-06,
"loss": 0.0642,
"step": 2275
},
{
"epoch": 9.02,
"learning_rate": 6.002222222222223e-06,
"loss": 0.0697,
"step": 2300
},
{
"epoch": 9.03,
"learning_rate": 5.946666666666668e-06,
"loss": 0.0621,
"step": 2325
},
{
"epoch": 9.03,
"learning_rate": 5.891111111111112e-06,
"loss": 0.0672,
"step": 2350
},
{
"epoch": 9.04,
"learning_rate": 5.8355555555555565e-06,
"loss": 0.0676,
"step": 2375
},
{
"epoch": 9.04,
"learning_rate": 5.78e-06,
"loss": 0.0626,
"step": 2400
},
{
"epoch": 9.05,
"learning_rate": 5.724444444444445e-06,
"loss": 0.0627,
"step": 2425
},
{
"epoch": 10.01,
"learning_rate": 5.6688888888888895e-06,
"loss": 0.0655,
"step": 2450
},
{
"epoch": 10.01,
"learning_rate": 5.613333333333334e-06,
"loss": 0.0619,
"step": 2475
},
{
"epoch": 10.02,
"learning_rate": 5.557777777777778e-06,
"loss": 0.0626,
"step": 2500
},
{
"epoch": 10.02,
"learning_rate": 5.5022222222222224e-06,
"loss": 0.0617,
"step": 2525
},
{
"epoch": 10.03,
"learning_rate": 5.4466666666666665e-06,
"loss": 0.0423,
"step": 2550
},
{
"epoch": 10.03,
"learning_rate": 5.391111111111111e-06,
"loss": 0.0445,
"step": 2575
},
{
"epoch": 10.04,
"learning_rate": 5.335555555555556e-06,
"loss": 0.0492,
"step": 2600
},
{
"epoch": 10.04,
"learning_rate": 5.28e-06,
"loss": 0.0425,
"step": 2625
},
{
"epoch": 10.04,
"learning_rate": 5.224444444444445e-06,
"loss": 0.0378,
"step": 2650
},
{
"epoch": 11.0,
"learning_rate": 5.168888888888889e-06,
"loss": 0.0766,
"step": 2675
},
{
"epoch": 11.01,
"learning_rate": 5.113333333333333e-06,
"loss": 0.12,
"step": 2700
},
{
"epoch": 11.01,
"learning_rate": 5.057777777777778e-06,
"loss": 0.1124,
"step": 2725
},
{
"epoch": 11.02,
"learning_rate": 5.002222222222223e-06,
"loss": 0.0621,
"step": 2750
},
{
"epoch": 11.02,
"learning_rate": 4.946666666666667e-06,
"loss": 0.0397,
"step": 2775
},
{
"epoch": 11.03,
"learning_rate": 4.891111111111111e-06,
"loss": 0.0417,
"step": 2800
},
{
"epoch": 11.03,
"learning_rate": 4.835555555555556e-06,
"loss": 0.0452,
"step": 2825
},
{
"epoch": 11.04,
"learning_rate": 4.78e-06,
"loss": 0.0419,
"step": 2850
},
{
"epoch": 11.04,
"learning_rate": 4.724444444444445e-06,
"loss": 0.0468,
"step": 2875
},
{
"epoch": 11.05,
"learning_rate": 4.66888888888889e-06,
"loss": 0.0477,
"step": 2900
},
{
"epoch": 12.0,
"learning_rate": 4.613333333333334e-06,
"loss": 0.073,
"step": 2925
},
{
"epoch": 12.01,
"learning_rate": 4.557777777777778e-06,
"loss": 0.0743,
"step": 2950
},
{
"epoch": 12.01,
"learning_rate": 4.502222222222223e-06,
"loss": 0.0617,
"step": 2975
},
{
"epoch": 12.02,
"learning_rate": 4.446666666666667e-06,
"loss": 0.028,
"step": 3000
},
{
"epoch": 12.02,
"eval_loss": 0.25769394636154175,
"eval_runtime": 3033.1756,
"eval_samples_per_second": 5.117,
"eval_steps_per_second": 0.08,
"eval_wer": 8.695623928070265,
"step": 3000
},
{
"epoch": 12.02,
"learning_rate": 4.391111111111112e-06,
"loss": 0.0242,
"step": 3025
},
{
"epoch": 12.03,
"learning_rate": 4.3355555555555565e-06,
"loss": 0.0257,
"step": 3050
},
{
"epoch": 12.03,
"learning_rate": 4.2800000000000005e-06,
"loss": 0.0264,
"step": 3075
},
{
"epoch": 12.04,
"learning_rate": 4.2244444444444446e-06,
"loss": 0.0253,
"step": 3100
},
{
"epoch": 12.04,
"learning_rate": 4.168888888888889e-06,
"loss": 0.0274,
"step": 3125
},
{
"epoch": 12.05,
"learning_rate": 4.1133333333333335e-06,
"loss": 0.0295,
"step": 3150
},
{
"epoch": 13.0,
"learning_rate": 4.057777777777778e-06,
"loss": 0.0433,
"step": 3175
},
{
"epoch": 13.01,
"learning_rate": 4.002222222222222e-06,
"loss": 0.031,
"step": 3200
},
{
"epoch": 13.01,
"learning_rate": 3.946666666666667e-06,
"loss": 0.0389,
"step": 3225
},
{
"epoch": 13.02,
"learning_rate": 3.891111111111111e-06,
"loss": 0.0287,
"step": 3250
},
{
"epoch": 13.02,
"learning_rate": 3.835555555555555e-06,
"loss": 0.0324,
"step": 3275
},
{
"epoch": 13.03,
"learning_rate": 3.7800000000000002e-06,
"loss": 0.0293,
"step": 3300
},
{
"epoch": 13.03,
"learning_rate": 3.724444444444445e-06,
"loss": 0.0281,
"step": 3325
},
{
"epoch": 13.04,
"learning_rate": 3.668888888888889e-06,
"loss": 0.0218,
"step": 3350
},
{
"epoch": 13.04,
"learning_rate": 3.6133333333333336e-06,
"loss": 0.0249,
"step": 3375
},
{
"epoch": 14.0,
"learning_rate": 3.5577777777777785e-06,
"loss": 0.0314,
"step": 3400
},
{
"epoch": 14.01,
"learning_rate": 3.5022222222222225e-06,
"loss": 0.0457,
"step": 3425
},
{
"epoch": 14.01,
"learning_rate": 3.446666666666667e-06,
"loss": 0.0354,
"step": 3450
},
{
"epoch": 14.02,
"learning_rate": 3.391111111111111e-06,
"loss": 0.0405,
"step": 3475
},
{
"epoch": 14.02,
"learning_rate": 3.335555555555556e-06,
"loss": 0.0354,
"step": 3500
},
{
"epoch": 14.03,
"learning_rate": 3.2800000000000004e-06,
"loss": 0.0395,
"step": 3525
},
{
"epoch": 14.03,
"learning_rate": 3.2244444444444444e-06,
"loss": 0.0367,
"step": 3550
},
{
"epoch": 14.04,
"learning_rate": 3.1688888888888893e-06,
"loss": 0.0389,
"step": 3575
},
{
"epoch": 14.04,
"learning_rate": 3.1133333333333337e-06,
"loss": 0.037,
"step": 3600
},
{
"epoch": 14.05,
"learning_rate": 3.0577777777777778e-06,
"loss": 0.0368,
"step": 3625
},
{
"epoch": 15.0,
"learning_rate": 3.0022222222222227e-06,
"loss": 0.0517,
"step": 3650
},
{
"epoch": 15.01,
"learning_rate": 2.946666666666667e-06,
"loss": 0.0638,
"step": 3675
},
{
"epoch": 15.01,
"learning_rate": 2.891111111111111e-06,
"loss": 0.0588,
"step": 3700
},
{
"epoch": 15.02,
"learning_rate": 2.835555555555556e-06,
"loss": 0.0636,
"step": 3725
},
{
"epoch": 15.02,
"learning_rate": 2.7800000000000005e-06,
"loss": 0.062,
"step": 3750
},
{
"epoch": 15.03,
"learning_rate": 2.7244444444444445e-06,
"loss": 0.0596,
"step": 3775
},
{
"epoch": 15.03,
"learning_rate": 2.6688888888888894e-06,
"loss": 0.075,
"step": 3800
},
{
"epoch": 15.04,
"learning_rate": 2.6133333333333334e-06,
"loss": 0.0694,
"step": 3825
},
{
"epoch": 15.04,
"learning_rate": 2.557777777777778e-06,
"loss": 0.0686,
"step": 3850
},
{
"epoch": 15.05,
"learning_rate": 2.5022222222222224e-06,
"loss": 0.0682,
"step": 3875
},
{
"epoch": 16.0,
"learning_rate": 2.446666666666667e-06,
"loss": 0.096,
"step": 3900
},
{
"epoch": 16.01,
"learning_rate": 2.3911111111111113e-06,
"loss": 0.1025,
"step": 3925
},
{
"epoch": 16.01,
"learning_rate": 2.3355555555555557e-06,
"loss": 0.103,
"step": 3950
},
{
"epoch": 16.02,
"learning_rate": 2.28e-06,
"loss": 0.1102,
"step": 3975
},
{
"epoch": 16.02,
"learning_rate": 2.2244444444444447e-06,
"loss": 0.1083,
"step": 4000
},
{
"epoch": 16.02,
"eval_loss": 0.22095343470573425,
"eval_runtime": 3161.6276,
"eval_samples_per_second": 4.909,
"eval_steps_per_second": 0.077,
"eval_wer": 8.212281066472636,
"step": 4000
},
{
"epoch": 16.03,
"learning_rate": 2.168888888888889e-06,
"loss": 0.1083,
"step": 4025
},
{
"epoch": 16.03,
"learning_rate": 2.1133333333333336e-06,
"loss": 0.1099,
"step": 4050
},
{
"epoch": 16.04,
"learning_rate": 2.057777777777778e-06,
"loss": 0.0948,
"step": 4075
},
{
"epoch": 16.04,
"learning_rate": 2.0022222222222225e-06,
"loss": 0.0966,
"step": 4100
},
{
"epoch": 17.0,
"learning_rate": 1.9466666666666665e-06,
"loss": 0.094,
"step": 4125
},
{
"epoch": 17.01,
"learning_rate": 1.8911111111111114e-06,
"loss": 0.0338,
"step": 4150
},
{
"epoch": 17.01,
"learning_rate": 1.8355555555555557e-06,
"loss": 0.0257,
"step": 4175
},
{
"epoch": 17.02,
"learning_rate": 1.7800000000000001e-06,
"loss": 0.0258,
"step": 4200
},
{
"epoch": 17.02,
"learning_rate": 1.7244444444444448e-06,
"loss": 0.0241,
"step": 4225
},
{
"epoch": 17.03,
"learning_rate": 1.668888888888889e-06,
"loss": 0.0246,
"step": 4250
},
{
"epoch": 17.03,
"learning_rate": 1.6133333333333335e-06,
"loss": 0.0216,
"step": 4275
},
{
"epoch": 17.04,
"learning_rate": 1.5577777777777777e-06,
"loss": 0.0252,
"step": 4300
},
{
"epoch": 17.04,
"learning_rate": 1.5022222222222224e-06,
"loss": 0.0239,
"step": 4325
},
{
"epoch": 17.05,
"learning_rate": 1.4466666666666669e-06,
"loss": 0.024,
"step": 4350
},
{
"epoch": 18.0,
"learning_rate": 1.3911111111111111e-06,
"loss": 0.0424,
"step": 4375
},
{
"epoch": 18.01,
"learning_rate": 1.3355555555555558e-06,
"loss": 0.0826,
"step": 4400
},
{
"epoch": 18.01,
"learning_rate": 1.28e-06,
"loss": 0.0711,
"step": 4425
},
{
"epoch": 18.02,
"learning_rate": 1.2244444444444445e-06,
"loss": 0.0866,
"step": 4450
},
{
"epoch": 18.02,
"learning_rate": 1.168888888888889e-06,
"loss": 0.0783,
"step": 4475
},
{
"epoch": 18.03,
"learning_rate": 1.1133333333333334e-06,
"loss": 0.0822,
"step": 4500
},
{
"epoch": 18.03,
"learning_rate": 1.0577777777777779e-06,
"loss": 0.0853,
"step": 4525
},
{
"epoch": 18.04,
"learning_rate": 1.0022222222222223e-06,
"loss": 0.0806,
"step": 4550
},
{
"epoch": 18.04,
"learning_rate": 9.466666666666667e-07,
"loss": 0.0697,
"step": 4575
},
{
"epoch": 18.05,
"learning_rate": 8.911111111111112e-07,
"loss": 0.0781,
"step": 4600
},
{
"epoch": 19.0,
"learning_rate": 8.355555555555556e-07,
"loss": 0.0437,
"step": 4625
},
{
"epoch": 19.01,
"learning_rate": 7.8e-07,
"loss": 0.0221,
"step": 4650
},
{
"epoch": 19.01,
"learning_rate": 7.244444444444446e-07,
"loss": 0.0193,
"step": 4675
},
{
"epoch": 19.02,
"learning_rate": 6.68888888888889e-07,
"loss": 0.0189,
"step": 4700
},
{
"epoch": 19.02,
"learning_rate": 6.133333333333333e-07,
"loss": 0.02,
"step": 4725
},
{
"epoch": 19.03,
"learning_rate": 5.577777777777779e-07,
"loss": 0.0174,
"step": 4750
},
{
"epoch": 19.03,
"learning_rate": 5.022222222222222e-07,
"loss": 0.0191,
"step": 4775
},
{
"epoch": 19.04,
"learning_rate": 4.466666666666667e-07,
"loss": 0.0192,
"step": 4800
},
{
"epoch": 19.04,
"learning_rate": 3.9111111111111115e-07,
"loss": 0.0183,
"step": 4825
},
{
"epoch": 19.05,
"learning_rate": 3.3555555555555556e-07,
"loss": 0.0184,
"step": 4850
},
{
"epoch": 20.0,
"learning_rate": 2.8e-07,
"loss": 0.0261,
"step": 4875
},
{
"epoch": 20.01,
"learning_rate": 2.2444444444444445e-07,
"loss": 0.0187,
"step": 4900
},
{
"epoch": 20.02,
"learning_rate": 1.6888888888888888e-07,
"loss": 0.0238,
"step": 4925
},
{
"epoch": 20.02,
"learning_rate": 1.1333333333333336e-07,
"loss": 0.0177,
"step": 4950
},
{
"epoch": 20.02,
"learning_rate": 5.777777777777778e-08,
"loss": 0.021,
"step": 4975
},
{
"epoch": 20.03,
"learning_rate": 2.2222222222222225e-09,
"loss": 0.0189,
"step": 5000
},
{
"epoch": 20.03,
"eval_loss": 0.25196951627731323,
"eval_runtime": 3402.3954,
"eval_samples_per_second": 4.561,
"eval_steps_per_second": 0.071,
"eval_wer": 8.44550699028117,
"step": 5000
},
{
"epoch": 20.03,
"step": 5000,
"total_flos": 1.846946562048e+20,
"train_loss": 0.10428944413661957,
"train_runtime": 162214.2802,
"train_samples_per_second": 3.945,
"train_steps_per_second": 0.031
}
],
"max_steps": 5000,
"num_train_epochs": 9223372036854775807,
"total_flos": 1.846946562048e+20,
"trial_name": null,
"trial_params": null
}