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: 10
num_epochs: 3.0
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
1.0023	0.1515	5	1.3400	44.2785
1.0472	0.3030	10	1.2532	43.4478
0.8951	0.4545	15	1.1128	40.1157
0.979	0.6061	20	0.9752	37.5770
0.8326	0.7576	25	0.8511	39.0610
0.701	0.9091	30	0.7512	34.6183
0.5083	1.0606	35	0.6850	26.2087
0.6136	1.2121	40	0.6448	21.7286
0.6453	1.3636	45	0.6096	19.1525
0.4892	1.5152	50	0.5818	18.2005
0.5009	1.6667	55	0.5640	17.8831
0.3788	1.8182	60	0.5514	17.8085
0.5336	1.9697	65	0.5423	17.5005
0.5214	2.1212	70	0.5354	17.3698
0.4278	2.2727	75	0.5304	18.4618
0.416	2.4242	80	0.5259	18.4712
0.5102	2.5758	85	0.5233	18.3872
0.4586	2.7273	90	0.5217	18.3872
0.4488	2.8788	95	0.5208	18.3778

greenw0lf/whisper-nonnat-5h-random-seed333

whisper-nonnat-5h-random-seed333