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: 59
num_epochs: 3.0
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.6635	0.2965	59	0.5610	31.2141
0.3267	0.5930	118	0.4449	25.3430
0.4946	0.8894	177	0.4309	25.0411
0.3369	1.1859	236	0.4154	24.1051
0.4091	1.4824	295	0.4108	23.3469
0.2441	1.7789	354	0.4074	22.6323
0.3665	2.0754	413	0.4044	22.4276
0.2295	2.3719	472	0.4054	21.9814
0.2318	2.6683	531	0.4043	22.3136
0.3093	2.9648	590	0.4044	22.6725

greenw0lf/whisper-ecapa-maxcos-5h

whisper-ecapa-maxcos-5h