learning_rate: 1e-05
train_batch_size: 16
eval_batch_size: 8
seed: 42
optimizer: Use OptimizerNames.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: 500
training_steps: 5000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Cer
1.1469	0.4496	500	1.1207	42.4127
0.3556	0.8993	1000	0.3324	10.8338
0.1558	1.3489	1500	0.2237	8.6272
0.1591	1.7986	2000	0.2078	6.7553
0.1045	2.2482	2500	0.1975	6.9542
0.124	2.6978	3000	0.1833	6.4297
0.093	3.1475	3500	0.1849	6.8638
0.0764	3.5971	4000	0.1844	7.0537
0.0661	4.0468	4500	0.1851	6.5383
0.0536	4.4964	5000	0.1873	6.6016

zacdan4801/whisper-base-whisper-ds-f4

whisper-base-whisper-ds-f4