learning_rate: 1e-05
train_batch_size: 32
eval_batch_size: 8
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_ratio: 0.1
num_epochs: 4

Training Loss	Epoch	Step	Validation Loss	Wer
0.9753	0.1667	5	0.8754	90.8756
0.7748	0.3333	10	0.6836	97.8802
0.7617	0.5	15	0.6131	80.6452
0.5298	0.6667	20	0.5739	86.2673
0.5449	0.8333	25	0.5517	80.7373
0.5278	1.0	30	0.5394	79.5392
0.3472	1.1667	35	0.5322	83.3180
0.4027	1.3333	40	0.5289	81.7512
0.3461	1.5	45	0.5169	80.6452
0.3512	1.6667	50	0.5117	78.7097
0.4236	1.8333	55	0.5033	75.0230
0.3461	2.0	60	0.5070	78.1567
0.2108	2.1667	65	0.5180	82.5806
0.1904	2.3333	70	0.5205	73.1797
0.2409	2.5	75	0.5336	83.9631
0.2021	2.6667	80	0.5325	81.1060
0.2072	2.8333	85	0.5247	76.6820
0.2788	3.0	90	0.5185	75.3917
0.1257	3.1667	95	0.5240	79.0783
0.19	3.3333	100	0.5312	80.0
0.1355	3.5	105	0.5405	75.9447
0.1365	3.6667	110	0.5458	76.4055
0.1459	3.8333	115	0.5485	77.6037
0.0772	4.0	120	0.5481	77.6037

nkkbr/whisper-large-v3-zatoichi-ja-EX-6-TRAIN_2_TO_10_EVAL_1

Whisper Large v3 - Japanese Zatoichi ASR