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
0.1851	14.2857	100	0.1529	3.5330
0.0008	28.5714	200	0.1691	2.4578
0.0002	42.8571	300	0.1754	2.3041
0.0001	57.1429	400	0.1805	1.9969
0.0	71.4286	500	0.1856	2.4578
0.0	85.7143	600	0.1890	2.4578
0.0	100.0	700	0.1915	2.7650
0.0	114.2857	800	0.1932	2.7650
0.0	128.5714	900	0.1956	2.7650
0.0	142.8571	1000	0.1972	2.7650
0.0	157.1429	1100	0.1984	2.1505
0.0	171.4286	1200	0.2001	2.1505
0.0	185.7143	1300	0.2011	2.1505
0.0	200.0	1400	0.2022	2.1505
0.0	214.2857	1500	0.2033	2.1505
0.0	228.5714	1600	0.2035	2.1505
0.0	242.8571	1700	0.2048	2.1505
0.0	257.1429	1800	0.2058	2.1505
0.0	271.4286	1900	0.2064	2.1505
0.0	285.7143	2000	0.2079	2.1505
0.0	300.0	2100	0.2093	2.1505
0.0	314.2857	2200	0.2100	2.1505
0.0	328.5714	2300	0.2102	2.1505
0.0	342.8571	2400	0.2108	2.1505
0.0	357.1429	2500	0.2111	2.1505
0.0	371.4286	2600	0.2116	2.1505
0.0	385.7143	2700	0.2131	1.9969
0.0	400.0	2800	0.2134	1.9969
0.0	414.2857	2900	0.2138	1.9969
0.0	428.5714	3000	0.2144	1.9969
0.0	442.8571	3100	0.2147	1.9969
0.0	457.1429	3200	0.2153	1.9969
0.0	471.4286	3300	0.2159	1.9969
0.0	485.7143	3400	0.2161	1.9969
0.0	500.0	3500	0.2161	1.9969
0.0	514.2857	3600	0.2167	1.9969
0.0	528.5714	3700	0.2168	1.9969
0.0	542.8571	3800	0.2168	1.9969
0.0	557.1429	3900	0.2168	1.9969
0.0	571.4286	4000	0.2171	1.9969

CianKim/whisper-tiny-kor_eng_tiny_ed_tx

whisper-tiny-kor_eng_tiny_ed_tx