learning_rate: 3e-05
train_batch_size: 12
eval_batch_size: 6
seed: 42
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 4000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Cer
0.4889	1.9231	100	0.4384	8.0056
0.1955	3.8462	200	0.3781	7.4005
0.0794	5.7692	300	0.4394	8.0521
0.0357	7.6923	400	0.5209	8.2848
0.0267	9.6154	500	0.5358	8.4012
0.0209	11.5385	600	0.5863	8.4943
0.0172	13.4615	700	0.6061	7.6565
0.0105	15.3846	800	0.5980	8.4012
0.0064	17.3077	900	0.6227	8.2848
0.0034	19.2308	1000	0.6307	8.8434
0.0016	21.1538	1100	0.6634	7.7030
0.0008	23.0769	1200	0.6657	7.3540
0.0021	25.0	1300	0.6710	8.7736
0.0002	26.9231	1400	0.6890	8.5176
0.0001	28.8462	1500	0.6949	7.8427
0.0001	30.7692	1600	0.6988	7.7729
0.0001	32.6923	1700	0.7028	7.7496
0.0001	34.6154	1800	0.7064	7.7263
0.0001	36.5385	1900	0.7092	7.5169
0.0001	38.4615	2000	0.7122	7.4703
0.0001	40.3846	2100	0.7153	7.4703
0.0001	42.3077	2200	0.7175	7.3772
0.0	44.2308	2300	0.7198	7.3772
0.0	46.1538	2400	0.7217	7.3772
0.0	48.0769	2500	0.7237	7.3772
0.0	50.0	2600	0.7256	7.3772
0.0	51.9231	2700	0.7271	7.5634
0.0	53.8462	2800	0.7285	7.5634
0.0	55.7692	2900	0.7301	7.5634
0.0	57.6923	3000	0.7316	7.5634
0.0	59.6154	3100	0.7329	7.6332
0.0	61.5385	3200	0.7341	7.4936
0.0	63.4615	3300	0.7350	7.4936
0.0	65.3846	3400	0.7358	7.6332
0.0	67.3077	3500	0.7366	7.6332
0.0	69.2308	3600	0.7373	7.6332
0.0	71.1538	3700	0.7378	7.8427
0.0	73.0769	3800	0.7382	7.9125
0.0	75.0	3900	0.7386	7.8892
0.0	76.9231	4000	0.7385	7.8892

CianKim/whisper-tiny-kor_eng_tiny_ed_op

whisper-tiny-kor_eng_tiny_ed_op