whisper-lv-small-augmented / trainer_state.json
Moreno La Quatra
End of training
abfaea6
raw
history blame
13.9 kB
{
"best_metric": 33.282732447817835,
"best_model_checkpoint": "whisper-lv-small-augmented/checkpoint-4000",
"epoch": 83.33333333333333,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.83,
"learning_rate": 4.6e-06,
"loss": 2.9758,
"step": 50
},
{
"epoch": 1.67,
"learning_rate": 9.600000000000001e-06,
"loss": 1.2165,
"step": 100
},
{
"epoch": 2.5,
"learning_rate": 1.4599999999999999e-05,
"loss": 0.7096,
"step": 150
},
{
"epoch": 3.33,
"learning_rate": 1.9600000000000002e-05,
"loss": 0.4438,
"step": 200
},
{
"epoch": 4.17,
"learning_rate": 2.46e-05,
"loss": 0.3072,
"step": 250
},
{
"epoch": 5.0,
"learning_rate": 2.96e-05,
"loss": 0.1973,
"step": 300
},
{
"epoch": 5.83,
"learning_rate": 3.46e-05,
"loss": 0.1346,
"step": 350
},
{
"epoch": 6.67,
"learning_rate": 3.960000000000001e-05,
"loss": 0.107,
"step": 400
},
{
"epoch": 7.5,
"learning_rate": 4.46e-05,
"loss": 0.0858,
"step": 450
},
{
"epoch": 8.33,
"learning_rate": 4.96e-05,
"loss": 0.0758,
"step": 500
},
{
"epoch": 9.17,
"learning_rate": 4.948888888888889e-05,
"loss": 0.0645,
"step": 550
},
{
"epoch": 10.0,
"learning_rate": 4.8933333333333335e-05,
"loss": 0.0608,
"step": 600
},
{
"epoch": 10.83,
"learning_rate": 4.837777777777778e-05,
"loss": 0.0443,
"step": 650
},
{
"epoch": 11.67,
"learning_rate": 4.782222222222222e-05,
"loss": 0.0415,
"step": 700
},
{
"epoch": 12.5,
"learning_rate": 4.726666666666667e-05,
"loss": 0.0378,
"step": 750
},
{
"epoch": 13.33,
"learning_rate": 4.671111111111111e-05,
"loss": 0.0312,
"step": 800
},
{
"epoch": 14.17,
"learning_rate": 4.615555555555556e-05,
"loss": 0.0278,
"step": 850
},
{
"epoch": 15.0,
"learning_rate": 4.5600000000000004e-05,
"loss": 0.0266,
"step": 900
},
{
"epoch": 15.83,
"learning_rate": 4.504444444444445e-05,
"loss": 0.0241,
"step": 950
},
{
"epoch": 16.67,
"learning_rate": 4.448888888888889e-05,
"loss": 0.0219,
"step": 1000
},
{
"epoch": 16.67,
"eval_loss": 0.6128966808319092,
"eval_runtime": 463.1108,
"eval_samples_per_second": 4.789,
"eval_steps_per_second": 0.151,
"eval_wer": 41.15749525616698,
"step": 1000
},
{
"epoch": 17.5,
"learning_rate": 4.3933333333333335e-05,
"loss": 0.0199,
"step": 1050
},
{
"epoch": 18.33,
"learning_rate": 4.337777777777778e-05,
"loss": 0.0187,
"step": 1100
},
{
"epoch": 19.17,
"learning_rate": 4.282222222222222e-05,
"loss": 0.015,
"step": 1150
},
{
"epoch": 20.0,
"learning_rate": 4.226666666666667e-05,
"loss": 0.0139,
"step": 1200
},
{
"epoch": 20.83,
"learning_rate": 4.171111111111111e-05,
"loss": 0.0123,
"step": 1250
},
{
"epoch": 21.67,
"learning_rate": 4.115555555555556e-05,
"loss": 0.0114,
"step": 1300
},
{
"epoch": 22.5,
"learning_rate": 4.0600000000000004e-05,
"loss": 0.0103,
"step": 1350
},
{
"epoch": 23.33,
"learning_rate": 4.004444444444445e-05,
"loss": 0.0101,
"step": 1400
},
{
"epoch": 24.17,
"learning_rate": 3.948888888888889e-05,
"loss": 0.0097,
"step": 1450
},
{
"epoch": 25.0,
"learning_rate": 3.8933333333333336e-05,
"loss": 0.0092,
"step": 1500
},
{
"epoch": 25.83,
"learning_rate": 3.837777777777778e-05,
"loss": 0.0087,
"step": 1550
},
{
"epoch": 26.67,
"learning_rate": 3.782222222222222e-05,
"loss": 0.0073,
"step": 1600
},
{
"epoch": 27.5,
"learning_rate": 3.726666666666667e-05,
"loss": 0.0065,
"step": 1650
},
{
"epoch": 28.33,
"learning_rate": 3.671111111111111e-05,
"loss": 0.0076,
"step": 1700
},
{
"epoch": 29.17,
"learning_rate": 3.615555555555556e-05,
"loss": 0.0069,
"step": 1750
},
{
"epoch": 30.0,
"learning_rate": 3.56e-05,
"loss": 0.0057,
"step": 1800
},
{
"epoch": 30.83,
"learning_rate": 3.504444444444445e-05,
"loss": 0.0055,
"step": 1850
},
{
"epoch": 31.67,
"learning_rate": 3.448888888888889e-05,
"loss": 0.0052,
"step": 1900
},
{
"epoch": 32.5,
"learning_rate": 3.3933333333333336e-05,
"loss": 0.0039,
"step": 1950
},
{
"epoch": 33.33,
"learning_rate": 3.337777777777778e-05,
"loss": 0.0029,
"step": 2000
},
{
"epoch": 33.33,
"eval_loss": 0.5974836349487305,
"eval_runtime": 428.1165,
"eval_samples_per_second": 5.181,
"eval_steps_per_second": 0.164,
"eval_wer": 36.14800759013283,
"step": 2000
},
{
"epoch": 34.17,
"learning_rate": 3.2822222222222223e-05,
"loss": 0.0035,
"step": 2050
},
{
"epoch": 35.0,
"learning_rate": 3.226666666666667e-05,
"loss": 0.0029,
"step": 2100
},
{
"epoch": 35.83,
"learning_rate": 3.171111111111111e-05,
"loss": 0.003,
"step": 2150
},
{
"epoch": 36.67,
"learning_rate": 3.1155555555555555e-05,
"loss": 0.0037,
"step": 2200
},
{
"epoch": 37.5,
"learning_rate": 3.06e-05,
"loss": 0.0035,
"step": 2250
},
{
"epoch": 38.33,
"learning_rate": 3.004444444444445e-05,
"loss": 0.0029,
"step": 2300
},
{
"epoch": 39.17,
"learning_rate": 2.948888888888889e-05,
"loss": 0.003,
"step": 2350
},
{
"epoch": 40.0,
"learning_rate": 2.8933333333333333e-05,
"loss": 0.0041,
"step": 2400
},
{
"epoch": 40.83,
"learning_rate": 2.837777777777778e-05,
"loss": 0.0034,
"step": 2450
},
{
"epoch": 41.67,
"learning_rate": 2.782222222222222e-05,
"loss": 0.0026,
"step": 2500
},
{
"epoch": 42.5,
"learning_rate": 2.7266666666666668e-05,
"loss": 0.0031,
"step": 2550
},
{
"epoch": 43.33,
"learning_rate": 2.6711111111111115e-05,
"loss": 0.002,
"step": 2600
},
{
"epoch": 44.17,
"learning_rate": 2.6155555555555555e-05,
"loss": 0.0009,
"step": 2650
},
{
"epoch": 45.0,
"learning_rate": 2.5600000000000002e-05,
"loss": 0.002,
"step": 2700
},
{
"epoch": 45.83,
"learning_rate": 2.504444444444445e-05,
"loss": 0.0016,
"step": 2750
},
{
"epoch": 46.67,
"learning_rate": 2.448888888888889e-05,
"loss": 0.0009,
"step": 2800
},
{
"epoch": 47.5,
"learning_rate": 2.3933333333333337e-05,
"loss": 0.001,
"step": 2850
},
{
"epoch": 48.33,
"learning_rate": 2.337777777777778e-05,
"loss": 0.0012,
"step": 2900
},
{
"epoch": 49.17,
"learning_rate": 2.282222222222222e-05,
"loss": 0.0012,
"step": 2950
},
{
"epoch": 50.0,
"learning_rate": 2.2266666666666668e-05,
"loss": 0.0003,
"step": 3000
},
{
"epoch": 50.0,
"eval_loss": 0.5625514388084412,
"eval_runtime": 427.1642,
"eval_samples_per_second": 5.192,
"eval_steps_per_second": 0.164,
"eval_wer": 33.68121442125237,
"step": 3000
},
{
"epoch": 50.83,
"learning_rate": 2.1711111111111112e-05,
"loss": 0.0011,
"step": 3050
},
{
"epoch": 51.67,
"learning_rate": 2.1155555555555556e-05,
"loss": 0.0012,
"step": 3100
},
{
"epoch": 52.5,
"learning_rate": 2.06e-05,
"loss": 0.0007,
"step": 3150
},
{
"epoch": 53.33,
"learning_rate": 2.0044444444444446e-05,
"loss": 0.0009,
"step": 3200
},
{
"epoch": 54.17,
"learning_rate": 1.948888888888889e-05,
"loss": 0.0005,
"step": 3250
},
{
"epoch": 55.0,
"learning_rate": 1.8933333333333334e-05,
"loss": 0.0003,
"step": 3300
},
{
"epoch": 55.83,
"learning_rate": 1.837777777777778e-05,
"loss": 0.0003,
"step": 3350
},
{
"epoch": 56.67,
"learning_rate": 1.7822222222222225e-05,
"loss": 0.0004,
"step": 3400
},
{
"epoch": 57.5,
"learning_rate": 1.726666666666667e-05,
"loss": 0.0006,
"step": 3450
},
{
"epoch": 58.33,
"learning_rate": 1.6711111111111112e-05,
"loss": 0.0014,
"step": 3500
},
{
"epoch": 59.17,
"learning_rate": 1.6155555555555556e-05,
"loss": 0.0006,
"step": 3550
},
{
"epoch": 60.0,
"learning_rate": 1.56e-05,
"loss": 0.0004,
"step": 3600
},
{
"epoch": 60.83,
"learning_rate": 1.5044444444444445e-05,
"loss": 0.0004,
"step": 3650
},
{
"epoch": 61.67,
"learning_rate": 1.448888888888889e-05,
"loss": 0.0004,
"step": 3700
},
{
"epoch": 62.5,
"learning_rate": 1.3933333333333334e-05,
"loss": 0.0002,
"step": 3750
},
{
"epoch": 63.33,
"learning_rate": 1.3377777777777778e-05,
"loss": 0.0004,
"step": 3800
},
{
"epoch": 64.17,
"learning_rate": 1.2822222222222222e-05,
"loss": 0.0002,
"step": 3850
},
{
"epoch": 65.0,
"learning_rate": 1.2266666666666667e-05,
"loss": 0.0002,
"step": 3900
},
{
"epoch": 65.83,
"learning_rate": 1.1711111111111111e-05,
"loss": 0.0004,
"step": 3950
},
{
"epoch": 66.67,
"learning_rate": 1.1155555555555556e-05,
"loss": 0.0004,
"step": 4000
},
{
"epoch": 66.67,
"eval_loss": 0.560806930065155,
"eval_runtime": 426.638,
"eval_samples_per_second": 5.199,
"eval_steps_per_second": 0.164,
"eval_wer": 33.282732447817835,
"step": 4000
},
{
"epoch": 67.5,
"learning_rate": 1.06e-05,
"loss": 0.0002,
"step": 4050
},
{
"epoch": 68.33,
"learning_rate": 1.0044444444444446e-05,
"loss": 0.0002,
"step": 4100
},
{
"epoch": 69.17,
"learning_rate": 9.48888888888889e-06,
"loss": 0.0003,
"step": 4150
},
{
"epoch": 70.0,
"learning_rate": 8.933333333333333e-06,
"loss": 0.0003,
"step": 4200
},
{
"epoch": 70.83,
"learning_rate": 8.377777777777779e-06,
"loss": 0.0002,
"step": 4250
},
{
"epoch": 71.67,
"learning_rate": 7.822222222222222e-06,
"loss": 0.0004,
"step": 4300
},
{
"epoch": 72.5,
"learning_rate": 7.266666666666668e-06,
"loss": 0.0002,
"step": 4350
},
{
"epoch": 73.33,
"learning_rate": 6.711111111111111e-06,
"loss": 0.0001,
"step": 4400
},
{
"epoch": 74.17,
"learning_rate": 6.155555555555556e-06,
"loss": 0.0001,
"step": 4450
},
{
"epoch": 75.0,
"learning_rate": 5.600000000000001e-06,
"loss": 0.0001,
"step": 4500
},
{
"epoch": 75.83,
"learning_rate": 5.044444444444444e-06,
"loss": 0.0001,
"step": 4550
},
{
"epoch": 76.67,
"learning_rate": 4.488888888888889e-06,
"loss": 0.0001,
"step": 4600
},
{
"epoch": 77.5,
"learning_rate": 3.9333333333333335e-06,
"loss": 0.0001,
"step": 4650
},
{
"epoch": 78.33,
"learning_rate": 3.3777777777777777e-06,
"loss": 0.0003,
"step": 4700
},
{
"epoch": 79.17,
"learning_rate": 2.8222222222222223e-06,
"loss": 0.0002,
"step": 4750
},
{
"epoch": 80.0,
"learning_rate": 2.266666666666667e-06,
"loss": 0.0001,
"step": 4800
},
{
"epoch": 80.83,
"learning_rate": 1.7111111111111112e-06,
"loss": 0.0001,
"step": 4850
},
{
"epoch": 81.67,
"learning_rate": 1.1555555555555556e-06,
"loss": 0.0002,
"step": 4900
},
{
"epoch": 82.5,
"learning_rate": 6.000000000000001e-07,
"loss": 0.0001,
"step": 4950
},
{
"epoch": 83.33,
"learning_rate": 4.444444444444445e-08,
"loss": 0.0001,
"step": 5000
},
{
"epoch": 83.33,
"eval_loss": 0.5715191960334778,
"eval_runtime": 429.9918,
"eval_samples_per_second": 5.158,
"eval_steps_per_second": 0.163,
"eval_wer": 33.804554079696395,
"step": 5000
},
{
"epoch": 83.33,
"step": 5000,
"total_flos": 9.148503492624384e+19,
"train_loss": 0.06885672171264887,
"train_runtime": 14920.221,
"train_samples_per_second": 21.447,
"train_steps_per_second": 0.335
}
],
"max_steps": 5000,
"num_train_epochs": 84,
"total_flos": 9.148503492624384e+19,
"trial_name": null,
"trial_params": null
}