wav2vec2-xls-r-300m-japanese / trainer_state.json
AndrewMcDowell's picture
End of training
7dfcfaf
raw
history blame
15.8 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 50.0,
"global_step": 10600,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.47,
"learning_rate": 3.6e-06,
"loss": 36.2649,
"step": 100
},
{
"epoch": 0.94,
"learning_rate": 7.35e-06,
"loss": 21.3706,
"step": 200
},
{
"epoch": 1.42,
"learning_rate": 1.1099999999999999e-05,
"loss": 13.7098,
"step": 300
},
{
"epoch": 1.89,
"learning_rate": 1.485e-05,
"loss": 10.7365,
"step": 400
},
{
"epoch": 2.36,
"learning_rate": 1.8599999999999998e-05,
"loss": 7.9817,
"step": 500
},
{
"epoch": 2.83,
"learning_rate": 2.2349999999999998e-05,
"loss": 5.8439,
"step": 600
},
{
"epoch": 3.3,
"learning_rate": 2.6099999999999997e-05,
"loss": 4.7275,
"step": 700
},
{
"epoch": 3.77,
"learning_rate": 2.985e-05,
"loss": 4.3969,
"step": 800
},
{
"epoch": 4.25,
"learning_rate": 3.36e-05,
"loss": 4.1733,
"step": 900
},
{
"epoch": 4.72,
"learning_rate": 3.735e-05,
"loss": 4.0974,
"step": 1000
},
{
"epoch": 4.72,
"eval_loss": 4.017753601074219,
"eval_runtime": 233.3211,
"eval_samples_per_second": 19.141,
"eval_steps_per_second": 2.396,
"eval_wer": 1.9534571492503916,
"step": 1000
},
{
"epoch": 5.19,
"learning_rate": 4.11e-05,
"loss": 4.0623,
"step": 1100
},
{
"epoch": 5.66,
"learning_rate": 4.484999999999999e-05,
"loss": 3.9986,
"step": 1200
},
{
"epoch": 6.13,
"learning_rate": 4.8599999999999995e-05,
"loss": 3.9302,
"step": 1300
},
{
"epoch": 6.6,
"learning_rate": 5.234999999999999e-05,
"loss": 3.7211,
"step": 1400
},
{
"epoch": 7.08,
"learning_rate": 5.6099999999999995e-05,
"loss": 3.162,
"step": 1500
},
{
"epoch": 7.55,
"learning_rate": 5.985e-05,
"loss": 2.6718,
"step": 1600
},
{
"epoch": 8.02,
"learning_rate": 6.359999999999999e-05,
"loss": 2.4628,
"step": 1700
},
{
"epoch": 8.49,
"learning_rate": 6.735e-05,
"loss": 2.2766,
"step": 1800
},
{
"epoch": 8.96,
"learning_rate": 7.11e-05,
"loss": 2.2276,
"step": 1900
},
{
"epoch": 9.43,
"learning_rate": 7.484999999999999e-05,
"loss": 2.1276,
"step": 2000
},
{
"epoch": 9.43,
"eval_loss": 0.930120587348938,
"eval_runtime": 231.4711,
"eval_samples_per_second": 19.294,
"eval_steps_per_second": 2.415,
"eval_wer": 1.2127992839561423,
"step": 2000
},
{
"epoch": 9.91,
"learning_rate": 7.416279069767441e-05,
"loss": 2.0853,
"step": 2100
},
{
"epoch": 10.38,
"learning_rate": 7.329069767441859e-05,
"loss": 2.0115,
"step": 2200
},
{
"epoch": 10.85,
"learning_rate": 7.241860465116278e-05,
"loss": 1.9745,
"step": 2300
},
{
"epoch": 11.32,
"learning_rate": 7.154651162790697e-05,
"loss": 1.9235,
"step": 2400
},
{
"epoch": 11.79,
"learning_rate": 7.067441860465115e-05,
"loss": 1.9008,
"step": 2500
},
{
"epoch": 12.26,
"learning_rate": 6.980232558139535e-05,
"loss": 1.878,
"step": 2600
},
{
"epoch": 12.74,
"learning_rate": 6.893023255813953e-05,
"loss": 1.8435,
"step": 2700
},
{
"epoch": 13.21,
"learning_rate": 6.805813953488372e-05,
"loss": 1.8357,
"step": 2800
},
{
"epoch": 13.68,
"learning_rate": 6.71860465116279e-05,
"loss": 1.8002,
"step": 2900
},
{
"epoch": 14.15,
"learning_rate": 6.631395348837209e-05,
"loss": 1.7622,
"step": 3000
},
{
"epoch": 14.15,
"eval_loss": 0.7103351354598999,
"eval_runtime": 235.0291,
"eval_samples_per_second": 19.002,
"eval_steps_per_second": 2.378,
"eval_wer": 1.5526963526516,
"step": 3000
},
{
"epoch": 14.62,
"learning_rate": 6.544186046511627e-05,
"loss": 1.7696,
"step": 3100
},
{
"epoch": 15.09,
"learning_rate": 6.456976744186045e-05,
"loss": 1.732,
"step": 3200
},
{
"epoch": 15.57,
"learning_rate": 6.369767441860465e-05,
"loss": 1.7099,
"step": 3300
},
{
"epoch": 16.04,
"learning_rate": 6.282558139534884e-05,
"loss": 1.7177,
"step": 3400
},
{
"epoch": 16.51,
"learning_rate": 6.195348837209302e-05,
"loss": 1.6886,
"step": 3500
},
{
"epoch": 16.98,
"learning_rate": 6.10813953488372e-05,
"loss": 1.7001,
"step": 3600
},
{
"epoch": 17.45,
"learning_rate": 6.020930232558139e-05,
"loss": 1.6783,
"step": 3700
},
{
"epoch": 17.92,
"learning_rate": 5.933720930232557e-05,
"loss": 1.6688,
"step": 3800
},
{
"epoch": 18.4,
"learning_rate": 5.8465116279069764e-05,
"loss": 1.6316,
"step": 3900
},
{
"epoch": 18.87,
"learning_rate": 5.759302325581395e-05,
"loss": 1.6397,
"step": 4000
},
{
"epoch": 18.87,
"eval_loss": 0.6728963851928711,
"eval_runtime": 232.2669,
"eval_samples_per_second": 19.228,
"eval_steps_per_second": 2.407,
"eval_wer": 1.4269411501454463,
"step": 4000
},
{
"epoch": 19.34,
"learning_rate": 5.672093023255814e-05,
"loss": 1.6214,
"step": 4100
},
{
"epoch": 19.81,
"learning_rate": 5.5848837209302316e-05,
"loss": 1.6283,
"step": 4200
},
{
"epoch": 20.28,
"learning_rate": 5.49767441860465e-05,
"loss": 1.6226,
"step": 4300
},
{
"epoch": 20.75,
"learning_rate": 5.411337209302325e-05,
"loss": 1.5882,
"step": 4400
},
{
"epoch": 21.23,
"learning_rate": 5.324127906976744e-05,
"loss": 1.6129,
"step": 4500
},
{
"epoch": 21.7,
"learning_rate": 5.2369186046511624e-05,
"loss": 1.5853,
"step": 4600
},
{
"epoch": 22.17,
"learning_rate": 5.1497093023255804e-05,
"loss": 1.553,
"step": 4700
},
{
"epoch": 22.64,
"learning_rate": 5.0625e-05,
"loss": 1.5719,
"step": 4800
},
{
"epoch": 23.11,
"learning_rate": 4.9752906976744184e-05,
"loss": 1.5471,
"step": 4900
},
{
"epoch": 23.58,
"learning_rate": 4.888081395348837e-05,
"loss": 1.5468,
"step": 5000
},
{
"epoch": 23.58,
"eval_loss": 0.6087133884429932,
"eval_runtime": 236.0684,
"eval_samples_per_second": 18.918,
"eval_steps_per_second": 2.368,
"eval_wer": 1.2497202953680913,
"step": 5000
},
{
"epoch": 24.06,
"learning_rate": 4.800872093023255e-05,
"loss": 1.5584,
"step": 5100
},
{
"epoch": 24.53,
"learning_rate": 4.7136627906976736e-05,
"loss": 1.5359,
"step": 5200
},
{
"epoch": 25.0,
"learning_rate": 4.626453488372093e-05,
"loss": 1.5546,
"step": 5300
},
{
"epoch": 25.47,
"learning_rate": 4.5392441860465116e-05,
"loss": 1.5263,
"step": 5400
},
{
"epoch": 25.94,
"learning_rate": 4.45203488372093e-05,
"loss": 1.5168,
"step": 5500
},
{
"epoch": 26.42,
"learning_rate": 4.364825581395348e-05,
"loss": 1.489,
"step": 5600
},
{
"epoch": 26.89,
"learning_rate": 4.277616279069767e-05,
"loss": 1.5108,
"step": 5700
},
{
"epoch": 27.36,
"learning_rate": 4.1904069767441855e-05,
"loss": 1.4882,
"step": 5800
},
{
"epoch": 27.83,
"learning_rate": 4.103197674418605e-05,
"loss": 1.5027,
"step": 5900
},
{
"epoch": 28.3,
"learning_rate": 4.015988372093023e-05,
"loss": 1.4885,
"step": 6000
},
{
"epoch": 28.3,
"eval_loss": 0.5786022543907166,
"eval_runtime": 230.8558,
"eval_samples_per_second": 19.345,
"eval_steps_per_second": 2.421,
"eval_wer": 1.3222197359588275,
"step": 6000
},
{
"epoch": 28.77,
"learning_rate": 3.9287790697674414e-05,
"loss": 1.4804,
"step": 6100
},
{
"epoch": 29.25,
"learning_rate": 3.84156976744186e-05,
"loss": 1.5115,
"step": 6200
},
{
"epoch": 29.72,
"learning_rate": 3.754360465116279e-05,
"loss": 1.4561,
"step": 6300
},
{
"epoch": 30.19,
"learning_rate": 3.6671511627906973e-05,
"loss": 1.4786,
"step": 6400
},
{
"epoch": 30.66,
"learning_rate": 3.579941860465116e-05,
"loss": 1.4522,
"step": 6500
},
{
"epoch": 31.13,
"learning_rate": 3.4927325581395346e-05,
"loss": 1.4499,
"step": 6600
},
{
"epoch": 31.6,
"learning_rate": 3.406395348837209e-05,
"loss": 1.4398,
"step": 6700
},
{
"epoch": 32.08,
"learning_rate": 3.3191860465116275e-05,
"loss": 1.4364,
"step": 6800
},
{
"epoch": 32.55,
"learning_rate": 3.231976744186046e-05,
"loss": 1.4247,
"step": 6900
},
{
"epoch": 33.02,
"learning_rate": 3.144767441860465e-05,
"loss": 1.451,
"step": 7000
},
{
"epoch": 33.02,
"eval_loss": 0.5726000070571899,
"eval_runtime": 233.1518,
"eval_samples_per_second": 19.155,
"eval_steps_per_second": 2.398,
"eval_wer": 1.3768180801074066,
"step": 7000
},
{
"epoch": 33.49,
"learning_rate": 3.0575581395348834e-05,
"loss": 1.4259,
"step": 7100
},
{
"epoch": 33.96,
"learning_rate": 2.970348837209302e-05,
"loss": 1.4252,
"step": 7200
},
{
"epoch": 34.43,
"learning_rate": 2.8831395348837207e-05,
"loss": 1.4214,
"step": 7300
},
{
"epoch": 34.91,
"learning_rate": 2.795930232558139e-05,
"loss": 1.4222,
"step": 7400
},
{
"epoch": 35.38,
"learning_rate": 2.708720930232558e-05,
"loss": 1.3976,
"step": 7500
},
{
"epoch": 35.85,
"learning_rate": 2.6215116279069766e-05,
"loss": 1.4026,
"step": 7600
},
{
"epoch": 36.32,
"learning_rate": 2.5343023255813953e-05,
"loss": 1.4031,
"step": 7700
},
{
"epoch": 36.79,
"learning_rate": 2.447093023255814e-05,
"loss": 1.406,
"step": 7800
},
{
"epoch": 37.26,
"learning_rate": 2.3598837209302322e-05,
"loss": 1.4022,
"step": 7900
},
{
"epoch": 37.74,
"learning_rate": 2.2726744186046512e-05,
"loss": 1.3912,
"step": 8000
},
{
"epoch": 37.74,
"eval_loss": 0.551810622215271,
"eval_runtime": 230.6724,
"eval_samples_per_second": 19.361,
"eval_steps_per_second": 2.423,
"eval_wer": 1.2497202953680913,
"step": 8000
},
{
"epoch": 38.21,
"learning_rate": 2.1854651162790695e-05,
"loss": 1.4014,
"step": 8100
},
{
"epoch": 38.68,
"learning_rate": 2.098255813953488e-05,
"loss": 1.3701,
"step": 8200
},
{
"epoch": 39.15,
"learning_rate": 2.0110465116279068e-05,
"loss": 1.3704,
"step": 8300
},
{
"epoch": 39.62,
"learning_rate": 1.9238372093023254e-05,
"loss": 1.3746,
"step": 8400
},
{
"epoch": 40.09,
"learning_rate": 1.836627906976744e-05,
"loss": 1.3738,
"step": 8500
},
{
"epoch": 40.57,
"learning_rate": 1.7494186046511627e-05,
"loss": 1.3657,
"step": 8600
},
{
"epoch": 41.04,
"learning_rate": 1.6622093023255814e-05,
"loss": 1.3782,
"step": 8700
},
{
"epoch": 41.51,
"learning_rate": 1.5749999999999997e-05,
"loss": 1.3504,
"step": 8800
},
{
"epoch": 41.98,
"learning_rate": 1.4886627906976742e-05,
"loss": 1.3802,
"step": 8900
},
{
"epoch": 42.45,
"learning_rate": 1.401453488372093e-05,
"loss": 1.3617,
"step": 9000
},
{
"epoch": 42.45,
"eval_loss": 0.5352301597595215,
"eval_runtime": 231.9967,
"eval_samples_per_second": 19.25,
"eval_steps_per_second": 2.41,
"eval_wer": 1.269411501454464,
"step": 9000
},
{
"epoch": 42.92,
"learning_rate": 1.3142441860465116e-05,
"loss": 1.357,
"step": 9100
},
{
"epoch": 43.4,
"learning_rate": 1.2270348837209301e-05,
"loss": 1.332,
"step": 9200
},
{
"epoch": 43.87,
"learning_rate": 1.1398255813953488e-05,
"loss": 1.3447,
"step": 9300
},
{
"epoch": 44.34,
"learning_rate": 1.0526162790697674e-05,
"loss": 1.3431,
"step": 9400
},
{
"epoch": 44.81,
"learning_rate": 9.65406976744186e-06,
"loss": 1.3337,
"step": 9500
},
{
"epoch": 45.28,
"learning_rate": 8.781976744186047e-06,
"loss": 1.3505,
"step": 9600
},
{
"epoch": 45.75,
"learning_rate": 7.909883720930232e-06,
"loss": 1.3325,
"step": 9700
},
{
"epoch": 46.23,
"learning_rate": 7.037790697674418e-06,
"loss": 1.3522,
"step": 9800
},
{
"epoch": 46.7,
"learning_rate": 6.165697674418604e-06,
"loss": 1.3287,
"step": 9900
},
{
"epoch": 47.17,
"learning_rate": 5.293604651162791e-06,
"loss": 1.3113,
"step": 10000
},
{
"epoch": 47.17,
"eval_loss": 0.5227699279785156,
"eval_runtime": 235.0378,
"eval_samples_per_second": 19.001,
"eval_steps_per_second": 2.378,
"eval_wer": 1.2781382859700157,
"step": 10000
},
{
"epoch": 47.64,
"learning_rate": 4.4215116279069765e-06,
"loss": 1.3283,
"step": 10100
},
{
"epoch": 48.11,
"learning_rate": 3.5494186046511625e-06,
"loss": 1.3183,
"step": 10200
},
{
"epoch": 48.58,
"learning_rate": 2.6773255813953485e-06,
"loss": 1.3157,
"step": 10300
},
{
"epoch": 49.06,
"learning_rate": 1.8052325581395345e-06,
"loss": 1.3313,
"step": 10400
},
{
"epoch": 49.53,
"learning_rate": 9.331395348837209e-07,
"loss": 1.3258,
"step": 10500
},
{
"epoch": 50.0,
"learning_rate": 6.104651162790697e-08,
"loss": 1.3432,
"step": 10600
},
{
"epoch": 50.0,
"step": 10600,
"total_flos": 7.86273636478683e+19,
"train_loss": 2.5914736377068284,
"train_runtime": 29653.6134,
"train_samples_per_second": 17.118,
"train_steps_per_second": 0.357
}
],
"max_steps": 10600,
"num_train_epochs": 50,
"total_flos": 7.86273636478683e+19,
"trial_name": null,
"trial_params": null
}