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
1.2196	0.2941	10	1.4059	46.9864
1.0986	0.5882	20	1.1242	41.5398
0.8356	0.8824	30	0.8545	38.7346
0.7395	1.1765	40	0.7218	26.9424
0.6417	1.4706	50	0.6582	25.8840
0.5821	1.7647	60	0.6178	24.4603
0.5898	2.0588	70	0.5927	24.7225
0.5963	2.3529	80	0.5801	24.2776
0.5893	2.6471	90	0.5737	23.2380
0.6056	2.9412	100	0.5708	23.0881

greenw0lf/whisper-new-nnat-5h-meta-seed42

whisper-new-nnat-5h-meta-seed42