learning_rate: 1e-05
train_batch_size: 48
eval_batch_size: 32
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: 60
num_epochs: 3.0
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
1.1305	0.1256	25	1.2177	38.0481
1.0545	0.2513	50	1.1792	37.7562
1.0388	0.3769	75	1.1019	36.4981
0.8978	0.5025	100	1.0050	35.2367
0.8493	0.6281	125	0.8985	34.7335
0.7361	0.7538	150	0.7810	32.4991
0.6705	0.8794	175	0.6587	30.7310
0.5678	1.0050	200	0.5721	28.9395
0.5388	1.1307	225	0.5168	26.0174
0.5005	1.2563	250	0.4757	23.0852
0.4401	1.3819	275	0.4482	22.8705
0.4213	1.5075	300	0.4334	21.5755
0.4544	1.6332	325	0.4239	21.2970
0.4271	1.7588	350	0.4166	19.9886
0.4333	1.8844	375	0.4115	19.8510
0.411	2.0101	400	0.4070	19.3444
0.421	2.1357	425	0.4037	19.4216
0.3917	2.2613	450	0.4008	19.4518
0.3744	2.3869	475	0.3985	19.4552
0.4063	2.5126	500	0.3965	18.5862
0.4204	2.6382	525	0.3952	18.5326
0.4129	2.7638	550	0.3942	19.1398
0.3911	2.8894	575	0.3937	19.3847

greenw0lf/exp3-whisper-classified-jasmin

exp3-whisper-classified-jasmin