learning_rate: 2e-05
train_batch_size: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 8
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 50
num_epochs: 30
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
2.8296	2.7027	100	1.8980	1.2775
1.4551	5.4054	200	1.8875	1.0809
0.6801	8.1081	300	1.9566	1.0432
0.4784	10.8108	400	2.0349	0.9721
0.2899	13.5135	500	2.1027	0.9847
0.1860	16.2162	600	2.1898	0.9777
0.1292	18.9189	700	2.2167	1.0126
0.0690	21.6216	800	2.2689	1.0558
0.0401	24.3243	900	2.2870	1.0098
0.0350	27.0270	1000	2.3181	1.0349
0.0183	29.7297	1100	2.3307	1.2706