learning_rate: 0.001
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
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: 100
num_epochs: 15
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
0.4009	0.8016	200	0.1685	0.1065	0.4153
0.3274	1.6012	400	0.1685	0.1070	0.4150
0.2955	2.4008	600	0.1588	0.1048	0.4077
0.2911	3.2004	800	0.1434	0.1031	0.4035
0.2792	4.0	1000	0.1520	0.1072	0.4126
0.2624	4.8016	1200	0.1533	0.1063	0.4090
0.2481	5.6012	1400	0.1428	0.1028	0.4023
0.2281	6.4008	1600	0.1324	0.1002	0.3923
0.2295	7.2004	1800	0.1354	0.1008	0.3959
0.2211	8.0	2000	0.1331	0.1003	0.3959
0.2134	8.8016	2200	0.1347	0.0992	0.3896
0.2136	9.6012	2400	0.1322	0.0994	0.3896
0.1942	10.4008	2600	0.1378	0.1004	0.3935
0.1916	11.2004	2800	0.1338	0.0986	0.3899
0.1874	12.0	3000	0.1325	0.0969	0.3808
0.1838	12.8016	3200	0.1332	0.0994	0.3923
0.1759	13.6012	3400	0.1307	0.0986	0.3911
0.1721	14.4008	3600	0.1301	0.0976	0.3851

ctaguchi/ssc-bas-mms-model-mix-adapt-max-longcv

ssc-bas-mms-model-mix-adapt-max-longcv