learning_rate: 0.0001
train_batch_size: 64
eval_batch_size: 64
seed: 42
optimizer: Use OptimizerNames.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: 125
training_steps: 1250
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.4992	0.7184	125	0.4896	40.4581
0.3946	1.4368	250	0.4146	35.1031
0.3378	2.1552	375	0.3881	33.6112
0.3296	2.8736	500	0.3726	32.4134
0.2977	3.5920	625	0.3679	31.4118
0.2635	4.3103	750	0.3707	30.8829
0.2598	5.0287	875	0.3666	30.4486
0.2373	5.7471	1000	0.3688	30.5292
0.2135	6.4655	1125	0.3743	30.3226
0.2029	7.1839	1250	0.3737	30.1369

attoucheaziz/whisper-algerian-darja-stage1

whisper-algerian-darja-stage1