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: 19
num_epochs: 3.0
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
1.4839	0.2969	19	1.2710	46.5930
1.0542	0.5938	38	0.7480	37.3858
0.6692	0.8906	57	0.6132	31.3258
0.5312	1.1875	76	0.5760	25.3501
0.536	1.4844	95	0.5574	23.1864
0.4747	1.7812	114	0.5454	23.9030
0.4825	2.0781	133	0.5389	23.7812
0.4756	2.375	152	0.5341	23.8187
0.4707	2.6719	171	0.5312	23.7110
0.4599	2.9688	190	0.5301	22.9710

greenw0lf/whisper-new-nnat-5h-caps-xlsr

whisper-new-nnat-5h-caps-xlsr