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
1.0145	0.2778	10	1.4080	46.5274
0.8772	0.5556	20	1.1278	42.1627
0.7095	0.8333	30	0.8493	47.2112
0.552	1.1111	40	0.7166	31.4476
0.477	1.3889	50	0.6493	26.3429
0.4568	1.6667	60	0.6104	25.2424
0.4537	1.9444	70	0.5894	26.2071
0.4134	2.2222	80	0.5793	26.3242
0.4268	2.5	90	0.5733	25.8793
0.4445	2.7778	100	0.5701	25.8605

greenw0lf/whisper-new-nnat-5h-random-seed333

whisper-new-nnat-5h-random-seed333