whisper-small-nomimos

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0000
• Wer: 51.9149

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0004
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 132
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.3807	1.0929	100	0.3220	77.8723
0.5819	2.1858	200	0.2437	122.1277
0.2781	3.2787	300	0.1406	60.6383
0.1925	4.3716	400	0.1137	14.8936
0.1511	5.4645	500	0.0955	16.3830
0.1451	6.5574	600	0.0570	13.1915
0.1307	7.6503	700	0.0830	15.3191
0.1095	8.7432	800	0.0592	12.3404
0.1087	9.8361	900	0.0437	11.0638
0.0809	10.9290	1000	0.0519	11.9149
0.0734	12.0219	1100	0.0638	18.7234
0.0596	13.1148	1200	0.0487	8.2979
0.0558	14.2077	1300	0.0241	11.2766
0.0482	15.3005	1400	0.0256	9.5745
0.0383	16.3934	1500	0.0308	7.0213
0.0274	17.4863	1600	0.0258	103.4043
0.0206	18.5792	1700	0.0125	6.1702
0.0193	19.6721	1800	0.0166	4.8936
0.0096	20.7650	1900	0.0024	53.8298
0.0042	21.8579	2000	0.0002	54.0426
0.0009	22.9508	2100	0.0001	54.0426
0.0007	24.0437	2200	0.0005	52.5532
0.0004	25.1366	2300	0.0000	51.9149
0.0001	26.2295	2400	0.0000	52.3404
0.0001	27.3224	2500	0.0000	51.9149
0.0	28.4153	2600	0.0000	51.9149
0.0	29.5082	2700	0.0000	51.9149

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.0
• Datasets 2.21.0
• Tokenizers 0.19.1