learning_rate: 0.0001
train_batch_size: 4
eval_batch_size: 2
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 32
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: 500
training_steps: 5000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.0846	2.7056	500	0.2339	19.7460
0.0383	5.4071	1000	0.1777	14.1061
0.0223	8.1085	1500	0.1673	13.3342
0.0127	10.8141	2000	0.1469	10.5204
0.0041	13.5156	2500	0.1398	9.0388
0.0025	16.2171	3000	0.1240	9.4746
0.0	18.9227	3500	0.0997	7.8063
0.0	21.6242	4000	0.0997	7.6693
0.0	24.3256	4500	0.0998	7.4452
0.0	27.0271	5000	0.0998	7.4577

0x456665/whisper-small-medical

whisper-small-medical