19_2000_1e-5_hp-base / trainer_state.json
lnxdx's picture
End of training
309cc0a
raw
history blame
7.9 kB
{
"best_metric": 0.33440420560747663,
"best_model_checkpoint": "/content/drive/MyDrive/wav2vec2-large-xlsr-persian-asr-shemo_lnxdx/checkpoint-1700",
"epoch": 12.5,
"eval_steps": 100,
"global_step": 2000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.62,
"learning_rate": 2.0000000000000003e-06,
"loss": 1.9455,
"step": 100
},
{
"epoch": 0.62,
"eval_loss": 1.4246947765350342,
"eval_runtime": 33.9505,
"eval_samples_per_second": 8.365,
"eval_steps_per_second": 1.06,
"eval_wer": 0.4830607476635514,
"step": 100
},
{
"epoch": 1.25,
"learning_rate": 4.000000000000001e-06,
"loss": 1.4824,
"step": 200
},
{
"epoch": 1.25,
"eval_loss": 1.110669493675232,
"eval_runtime": 35.7852,
"eval_samples_per_second": 7.936,
"eval_steps_per_second": 1.006,
"eval_wer": 0.43311915887850466,
"step": 200
},
{
"epoch": 1.88,
"learning_rate": 6e-06,
"loss": 1.2516,
"step": 300
},
{
"epoch": 1.88,
"eval_loss": 0.9141018390655518,
"eval_runtime": 33.4012,
"eval_samples_per_second": 8.503,
"eval_steps_per_second": 1.078,
"eval_wer": 0.4135514018691589,
"step": 300
},
{
"epoch": 2.5,
"learning_rate": 7.980000000000002e-06,
"loss": 1.0859,
"step": 400
},
{
"epoch": 2.5,
"eval_loss": 0.8359942436218262,
"eval_runtime": 33.9773,
"eval_samples_per_second": 8.359,
"eval_steps_per_second": 1.06,
"eval_wer": 0.3974883177570093,
"step": 400
},
{
"epoch": 3.12,
"learning_rate": 9.980000000000001e-06,
"loss": 1.0357,
"step": 500
},
{
"epoch": 3.12,
"eval_loss": 0.8096792101860046,
"eval_runtime": 33.8733,
"eval_samples_per_second": 8.384,
"eval_steps_per_second": 1.063,
"eval_wer": 0.3814252336448598,
"step": 500
},
{
"epoch": 3.75,
"learning_rate": 9.340000000000002e-06,
"loss": 1.0472,
"step": 600
},
{
"epoch": 3.75,
"eval_loss": 0.7550356984138489,
"eval_runtime": 34.119,
"eval_samples_per_second": 8.324,
"eval_steps_per_second": 1.055,
"eval_wer": 0.37529205607476634,
"step": 600
},
{
"epoch": 4.38,
"learning_rate": 8.673333333333334e-06,
"loss": 0.9963,
"step": 700
},
{
"epoch": 4.38,
"eval_loss": 0.7533305287361145,
"eval_runtime": 36.7992,
"eval_samples_per_second": 7.718,
"eval_steps_per_second": 0.978,
"eval_wer": 0.36360981308411217,
"step": 700
},
{
"epoch": 5.0,
"learning_rate": 8.006666666666667e-06,
"loss": 0.9767,
"step": 800
},
{
"epoch": 5.0,
"eval_loss": 0.7423538565635681,
"eval_runtime": 33.6729,
"eval_samples_per_second": 8.434,
"eval_steps_per_second": 1.069,
"eval_wer": 0.3589369158878505,
"step": 800
},
{
"epoch": 5.62,
"learning_rate": 7.340000000000001e-06,
"loss": 0.9667,
"step": 900
},
{
"epoch": 5.62,
"eval_loss": 0.736015796661377,
"eval_runtime": 34.1214,
"eval_samples_per_second": 8.323,
"eval_steps_per_second": 1.055,
"eval_wer": 0.3516355140186916,
"step": 900
},
{
"epoch": 6.25,
"learning_rate": 6.680000000000001e-06,
"loss": 0.9385,
"step": 1000
},
{
"epoch": 6.25,
"eval_loss": 0.7354543209075928,
"eval_runtime": 33.4234,
"eval_samples_per_second": 8.497,
"eval_steps_per_second": 1.077,
"eval_wer": 0.348714953271028,
"step": 1000
},
{
"epoch": 6.88,
"learning_rate": 6.013333333333335e-06,
"loss": 0.9805,
"step": 1100
},
{
"epoch": 6.88,
"eval_loss": 0.7237350940704346,
"eval_runtime": 36.5425,
"eval_samples_per_second": 7.772,
"eval_steps_per_second": 0.985,
"eval_wer": 0.3463785046728972,
"step": 1100
},
{
"epoch": 7.5,
"learning_rate": 5.346666666666667e-06,
"loss": 0.976,
"step": 1200
},
{
"epoch": 7.5,
"eval_loss": 0.7077719569206238,
"eval_runtime": 33.9188,
"eval_samples_per_second": 8.373,
"eval_steps_per_second": 1.061,
"eval_wer": 0.3455023364485981,
"step": 1200
},
{
"epoch": 8.12,
"learning_rate": 4.680000000000001e-06,
"loss": 0.88,
"step": 1300
},
{
"epoch": 8.12,
"eval_loss": 0.7229148745536804,
"eval_runtime": 33.6191,
"eval_samples_per_second": 8.448,
"eval_steps_per_second": 1.071,
"eval_wer": 0.34375,
"step": 1300
},
{
"epoch": 8.75,
"learning_rate": 4.013333333333334e-06,
"loss": 0.9421,
"step": 1400
},
{
"epoch": 8.75,
"eval_loss": 0.7180063128471375,
"eval_runtime": 33.4233,
"eval_samples_per_second": 8.497,
"eval_steps_per_second": 1.077,
"eval_wer": 0.3431658878504673,
"step": 1400
},
{
"epoch": 9.38,
"learning_rate": 3.346666666666667e-06,
"loss": 0.9584,
"step": 1500
},
{
"epoch": 9.38,
"eval_loss": 0.705901563167572,
"eval_runtime": 33.1792,
"eval_samples_per_second": 8.56,
"eval_steps_per_second": 1.085,
"eval_wer": 0.3364485981308411,
"step": 1500
},
{
"epoch": 10.0,
"learning_rate": 2.68e-06,
"loss": 0.88,
"step": 1600
},
{
"epoch": 10.0,
"eval_loss": 0.7106391191482544,
"eval_runtime": 35.2342,
"eval_samples_per_second": 8.06,
"eval_steps_per_second": 1.022,
"eval_wer": 0.3364485981308411,
"step": 1600
},
{
"epoch": 10.62,
"learning_rate": 2.0133333333333337e-06,
"loss": 0.9113,
"step": 1700
},
{
"epoch": 10.62,
"eval_loss": 0.7124722599983215,
"eval_runtime": 34.7063,
"eval_samples_per_second": 8.183,
"eval_steps_per_second": 1.037,
"eval_wer": 0.33440420560747663,
"step": 1700
},
{
"epoch": 11.25,
"learning_rate": 1.3466666666666668e-06,
"loss": 0.912,
"step": 1800
},
{
"epoch": 11.25,
"eval_loss": 0.7090508937835693,
"eval_runtime": 33.4477,
"eval_samples_per_second": 8.491,
"eval_steps_per_second": 1.076,
"eval_wer": 0.3352803738317757,
"step": 1800
},
{
"epoch": 11.88,
"learning_rate": 6.800000000000001e-07,
"loss": 0.9607,
"step": 1900
},
{
"epoch": 11.88,
"eval_loss": 0.7066251635551453,
"eval_runtime": 33.4966,
"eval_samples_per_second": 8.478,
"eval_steps_per_second": 1.075,
"eval_wer": 0.33440420560747663,
"step": 1900
},
{
"epoch": 12.5,
"learning_rate": 1.3333333333333334e-08,
"loss": 0.8974,
"step": 2000
},
{
"epoch": 12.5,
"eval_loss": 0.7064409852027893,
"eval_runtime": 33.5058,
"eval_samples_per_second": 8.476,
"eval_steps_per_second": 1.074,
"eval_wer": 0.33440420560747663,
"step": 2000
},
{
"epoch": 12.5,
"step": 2000,
"total_flos": 9.605345431463184e+18,
"train_loss": 1.0512529983520509,
"train_runtime": 9036.2889,
"train_samples_per_second": 3.541,
"train_steps_per_second": 0.221
}
],
"logging_steps": 100,
"max_steps": 2000,
"num_train_epochs": 13,
"save_steps": 100,
"total_flos": 9.605345431463184e+18,
"trial_name": null,
"trial_params": null
}