whisper-large-khmer / trainer_state.json
steja's picture
End of training
21e41b1
{
"best_metric": 51.168264538722084,
"best_model_checkpoint": "./whisper-large-khmer/checkpoint-1000",
"epoch": 100.0,
"global_step": 1000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 2.5,
"learning_rate": 6.393768004764144e-06,
"loss": 1.6773,
"step": 25
},
{
"epoch": 5.0,
"learning_rate": 8.217263382430936e-06,
"loss": 1.1775,
"step": 50
},
{
"epoch": 7.5,
"learning_rate": 9.194245453686277e-06,
"loss": 0.5627,
"step": 75
},
{
"epoch": 10.0,
"learning_rate": 9.865639267998493e-06,
"loss": 0.2161,
"step": 100
},
{
"epoch": 12.5,
"learning_rate": 9.800000000000001e-06,
"loss": 0.0863,
"step": 125
},
{
"epoch": 15.0,
"learning_rate": 9.522222222222223e-06,
"loss": 0.0353,
"step": 150
},
{
"epoch": 17.5,
"learning_rate": 9.244444444444445e-06,
"loss": 0.0137,
"step": 175
},
{
"epoch": 20.0,
"learning_rate": 8.966666666666667e-06,
"loss": 0.0076,
"step": 200
},
{
"epoch": 22.5,
"learning_rate": 8.68888888888889e-06,
"loss": 0.004,
"step": 225
},
{
"epoch": 25.0,
"learning_rate": 8.411111111111112e-06,
"loss": 0.0018,
"step": 250
},
{
"epoch": 27.5,
"learning_rate": 8.133333333333334e-06,
"loss": 0.0022,
"step": 275
},
{
"epoch": 30.0,
"learning_rate": 7.855555555555556e-06,
"loss": 0.0028,
"step": 300
},
{
"epoch": 32.5,
"learning_rate": 7.5777777777777785e-06,
"loss": 0.0021,
"step": 325
},
{
"epoch": 35.0,
"learning_rate": 7.3e-06,
"loss": 0.0022,
"step": 350
},
{
"epoch": 37.5,
"learning_rate": 7.022222222222222e-06,
"loss": 0.0028,
"step": 375
},
{
"epoch": 40.0,
"learning_rate": 6.744444444444444e-06,
"loss": 0.0022,
"step": 400
},
{
"epoch": 42.5,
"learning_rate": 6.466666666666667e-06,
"loss": 0.0012,
"step": 425
},
{
"epoch": 45.0,
"learning_rate": 6.18888888888889e-06,
"loss": 0.0005,
"step": 450
},
{
"epoch": 47.5,
"learning_rate": 5.911111111111112e-06,
"loss": 0.0003,
"step": 475
},
{
"epoch": 50.0,
"learning_rate": 5.633333333333334e-06,
"loss": 0.0002,
"step": 500
},
{
"epoch": 50.0,
"eval_loss": 0.548828125,
"eval_runtime": 569.4356,
"eval_samples_per_second": 1.261,
"eval_steps_per_second": 0.021,
"eval_wer": 51.532809001664845,
"step": 500
},
{
"epoch": 52.5,
"learning_rate": 5.355555555555556e-06,
"loss": 0.0002,
"step": 525
},
{
"epoch": 55.0,
"learning_rate": 5.077777777777778e-06,
"loss": 0.0001,
"step": 550
},
{
"epoch": 57.5,
"learning_rate": 4.800000000000001e-06,
"loss": 0.0001,
"step": 575
},
{
"epoch": 60.0,
"learning_rate": 4.5222222222222225e-06,
"loss": 0.0001,
"step": 600
},
{
"epoch": 62.5,
"learning_rate": 4.244444444444445e-06,
"loss": 0.0001,
"step": 625
},
{
"epoch": 65.0,
"learning_rate": 3.966666666666667e-06,
"loss": 0.0001,
"step": 650
},
{
"epoch": 67.5,
"learning_rate": 3.688888888888889e-06,
"loss": 0.0001,
"step": 675
},
{
"epoch": 70.0,
"learning_rate": 3.4111111111111113e-06,
"loss": 0.0001,
"step": 700
},
{
"epoch": 72.5,
"learning_rate": 3.133333333333334e-06,
"loss": 0.0001,
"step": 725
},
{
"epoch": 75.0,
"learning_rate": 2.855555555555556e-06,
"loss": 0.0001,
"step": 750
},
{
"epoch": 77.5,
"learning_rate": 2.577777777777778e-06,
"loss": 0.0001,
"step": 775
},
{
"epoch": 80.0,
"learning_rate": 2.3000000000000004e-06,
"loss": 0.0001,
"step": 800
},
{
"epoch": 82.5,
"learning_rate": 2.0222222222222223e-06,
"loss": 0.0001,
"step": 825
},
{
"epoch": 85.0,
"learning_rate": 1.7444444444444448e-06,
"loss": 0.0001,
"step": 850
},
{
"epoch": 87.5,
"learning_rate": 1.4666666666666669e-06,
"loss": 0.0001,
"step": 875
},
{
"epoch": 90.0,
"learning_rate": 1.188888888888889e-06,
"loss": 0.0001,
"step": 900
},
{
"epoch": 92.5,
"learning_rate": 9.111111111111113e-07,
"loss": 0.0001,
"step": 925
},
{
"epoch": 95.0,
"learning_rate": 6.333333333333334e-07,
"loss": 0.0001,
"step": 950
},
{
"epoch": 97.5,
"learning_rate": 3.555555555555556e-07,
"loss": 0.0001,
"step": 975
},
{
"epoch": 100.0,
"learning_rate": 7.777777777777778e-08,
"loss": 0.0001,
"step": 1000
},
{
"epoch": 100.0,
"eval_loss": 0.56591796875,
"eval_runtime": 568.6447,
"eval_samples_per_second": 1.263,
"eval_steps_per_second": 0.021,
"eval_wer": 51.168264538722084,
"step": 1000
},
{
"epoch": 100.0,
"step": 1000,
"total_flos": 1.2993806865463771e+20,
"train_loss": 0.0950160476192832,
"train_runtime": 9966.5948,
"train_samples_per_second": 6.421,
"train_steps_per_second": 0.1
}
],
"max_steps": 1000,
"num_train_epochs": 100,
"total_flos": 1.2993806865463771e+20,
"trial_name": null,
"trial_params": null
}