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
0.8757	1.2821	100	0.7532	11.4080
0.5888	2.5641	200	0.6619	10.5578
0.3103	3.8462	300	0.6157	10.7574
0.1136	5.1282	400	0.6487	10.1848
0.0416	6.4103	500	0.6659	10.6446
0.026	7.6923	600	0.6899	11.5989
0.0167	8.9744	700	0.7216	13.0910
0.0099	10.2564	800	0.7148	10.5838
0.0074	11.5385	900	0.7142	10.7400
0.0033	12.8205	1000	0.7085	10.0286
0.0026	14.1026	1100	0.7268	9.7857
0.0032	15.3846	1200	0.7560	10.9569
0.0016	16.6667	1300	0.7510	10.2542
0.0016	17.9487	1400	0.7487	10.2542
0.001	19.2308	1500	0.7514	10.6532
0.0005	20.5128	1600	0.7575	10.3062
0.001	21.7949	1700	0.7643	10.2021
0.0006	23.0769	1800	0.7640	10.2629
0.0004	24.3590	1900	0.7744	10.6793
0.0003	25.6410	2000	0.7789	10.5405
0.0002	26.9231	2100	0.7807	10.7140
0.0002	28.2051	2200	0.7819	10.6446
0.0002	29.4872	2300	0.7833	10.6012
0.0002	30.7692	2400	0.7853	10.5231
0.0002	32.0513	2500	0.7869	10.1414
0.0002	33.3333	2600	0.7884	10.1761
0.0002	34.6154	2700	0.7900	10.2802
0.0002	35.8974	2800	0.7915	10.2542
0.0001	37.1795	2900	0.7929	10.2976
0.0001	38.4615	3000	0.7945	10.2629
0.0001	39.7436	3100	0.7953	10.2629
0.0001	41.0256	3200	0.7965	10.2629
0.0001	42.3077	3300	0.7975	10.3670
0.0001	43.5897	3400	0.7982	10.3930
0.0001	44.8718	3500	0.7990	10.3236
0.0001	46.1538	3600	0.7998	10.5318
0.0001	47.4359	3700	0.8002	10.5318
0.0001	48.7179	3800	0.8007	10.5318
0.0001	50.0	3900	0.8010	10.5318
0.0001	51.2821	4000	0.8011	10.5318

CianKim/whisper-tiny-kor_eng_tiny_ps_pr

whisper-tiny-kor_eng_tiny_ps_pr