learning_rate: 0.0001
train_batch_size: 48
eval_batch_size: 32
seed: 42
optimizer: Use 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: 13
num_epochs: 3.0
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
1.0308	0.2222	10	1.2718	44.0172
0.8973	0.4444	20	0.9979	38.1930
0.6296	0.6667	30	0.7704	28.7661
0.5903	0.8889	40	0.6406	19.7405
0.509	1.1111	50	0.5733	18.4992
0.4961	1.3333	60	0.5398	18.4338
0.4695	1.5556	70	0.5194	17.8178
0.4777	1.7778	80	0.5098	16.7911
0.4889	2.0	90	0.5022	18.1445
0.4709	2.2222	100	0.4964	16.2591
0.4257	2.4444	110	0.4931	16.2778
0.4195	2.6667	120	0.4909	16.2591
0.4289	2.8889	130	0.4903	16.2778

greenw0lf/whisper-nonnat-10h-maxcos-xlsr

whisper-nonnat-10h-maxcos-xlsr