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: 95
num_epochs: 3.0
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.4895	0.2978	95	0.4338	21.7935
0.4647	0.5956	190	0.3713	17.6066
0.4092	0.8934	285	0.3529	16.3854
0.43	1.1912	380	0.3461	16.8417
0.3641	1.4890	475	0.3381	15.7949
0.2685	1.7868	570	0.3321	15.3588
0.2357	2.0846	665	0.3290	16.7545
0.242	2.3824	760	0.3284	15.7681
0.415	2.6803	855	0.3259	15.6406
0.3297	2.9781	950	0.3254	15.5937

greenw0lf/whisper-ecapa-caps-20h

whisper-ecapa-caps-20h