learning_rate: 1e-05
train_batch_size: 4
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 4
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: constant_with_warmup
lr_scheduler_warmup_steps: 500
num_epochs: 8
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
2.5174	1.0	92	2.3611	0.2754
2.5662	2.0	184	2.3611	0.2754
2.4968	3.0	276	2.3604	0.2754
2.5793	4.0	368	2.3561	0.2754
2.5702	5.0	460	2.3407	0.2754
2.5302	6.0	552	2.2996	0.2756
2.5017	7.0	644	2.2359	0.2739
2.4418	8.0	736	2.1597	0.2737

SPL123/whisper-hf-adalora

whisper-hf-adalora