learning_rate: 3.75e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
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
training_steps: 10000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.2527	0.6894	1000	0.3620	28.3522
0.1164	1.3785	2000	0.3459	26.2727
0.0625	2.0676	3000	0.3380	25.2712
0.0642	2.7570	4000	0.3373	24.5218
0.0331	3.4461	5000	0.3440	23.9973
0.0133	4.1351	6000	0.3445	22.7795
0.0116	4.8245	7000	0.3537	22.8306
0.0041	5.5136	8000	0.3623	22.4202
0.002	6.2027	9000	0.3632	21.7355
0.0016	6.8921	10000	0.3683	21.5362

volkan-aslan/whisper-small-tr-v3

Whisper Small TR V3 - Volkan ASLAN