learning_rate: 1e-05
train_batch_size: 16
eval_batch_size: 16
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: 1000
num_epochs: 15
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.539	0.4941	500	0.6416	35.1020
0.3446	0.9881	1000	0.4576	24.7569
0.242	1.4822	1500	0.3910	21.1218
0.2412	1.9763	2000	0.3525	19.9057
0.1419	2.4704	2500	0.3544	18.5183
0.1282	2.9644	3000	0.3399	17.3549
0.0793	3.4585	3500	0.3527	17.9382
0.0837	3.9526	4000	0.3547	17.3417
0.0438	4.4466	4500	0.3792	16.9693
0.0409	4.9407	5000	0.3831	17.0319
0.0206	5.4348	5500	0.4115	17.2923
0.0219	5.9289	6000	0.4062	16.5870
0.0089	6.4229	6500	0.4445	16.9528
0.0089	6.9170	7000	0.4532	16.8540
0.0061	7.4111	7500	0.4590	16.5804
0.0077	7.9051	8000	0.4735	16.3366
0.0033	8.3992	8500	0.4813	16.5178
0.0028	8.8933	9000	0.4968	16.8144
0.0016	9.3874	9500	0.5113	16.5013
0.0026	9.8814	10000	0.5040	16.6892
0.001	10.3755	10500	0.5144	16.2377
0.0008	10.8696	11000	0.5250	15.9740
0.0006	11.3636	11500	0.5317	16.0202
0.0005	11.8577	12000	0.5387	16.0235
0.0004	12.3518	12500	0.5498	16.0960
0.0004	12.8458	13000	0.5582	16.0399
0.0003	13.3399	13500	0.5577	15.9543
0.0003	13.8340	14000	0.5590	15.9477
0.0002	14.3281	14500	0.5609	15.9938
0.0003	14.8221	15000	0.5636	16.0465

samil24/whisper-small-kurmanji-v10

whisper-small-kurmanji-v10