learning_rate: 0.0001
train_batch_size: 16
eval_batch_size: 8
seed: 42
optimizer: Use adamw_torch 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: 200
training_steps: 5000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.529	2.8090	500	0.7669	47.3748
0.1183	5.6180	1000	0.8567	46.0405
0.0635	8.4270	1500	0.9692	46.4737
0.0356	11.2360	2000	0.9787	41.8298
0.0127	14.0449	2500	1.0125	40.4263
0.0072	16.8539	3000	1.0046	36.3369
0.0027	19.6629	3500	1.0095	34.6560
0.0017	22.4719	4000	1.0172	34.5521
0.0011	25.2809	4500	1.0271	34.0149
0.0007	28.0899	5000	1.0300	34.0669

aslinguist/whisper-sekolik-atayal

whisper-coastal-atayal