learning_rate: 0.0001
train_batch_size: 8
eval_batch_size: 8
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: 1000
num_epochs: 30
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
5.5682	1.9685	500	3.3982	1.0
3.3854	3.9370	1000	2.5579	1.0093
2.4499	5.9055	1500	1.6875	0.8812
1.9414	7.8740	2000	1.4364	0.8007
1.6279	9.8425	2500	1.3482	0.7087
1.4296	11.8110	3000	1.3312	0.6715
1.2397	13.7795	3500	1.3512	0.6631
1.0938	15.7480	4000	1.3674	0.6421
0.9995	17.7165	4500	1.3968	0.6172
0.9201	19.6850	5000	1.3733	0.5996
0.8308	21.6535	5500	1.3849	0.5960
0.7652	23.6220	6000	1.4066	0.5813
0.7195	25.5906	6500	1.4331	0.5893
0.6514	27.5591	7000	1.4573	0.5774
0.6414	29.5276	7500	1.4475	0.5751

Hung1294/ASR-model-1

ASR-model-1