whisper-small-CV_Fleurs_AMMI_ALFFA-sw-200hrs-v1

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: 0.4605
• Wer: 0.1870
• Cer: 0.0746

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use adamw_hf with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
1.3655	1.0	8064	0.4347	0.3461	0.1533
0.4274	2.0	16128	0.3388	0.1987	0.0715
0.2724	3.0	24192	0.3215	0.1789	0.0658
0.1953	4.0	32256	0.3383	0.1737	0.0632
0.1563	5.0	40320	0.3501	0.1856	0.0728
0.1368	6.0	48384	0.3699	0.1954	0.0870
0.127	7.0	56448	0.3858	0.1844	0.0711
0.1236	8.0	64512	0.4001	0.1905	0.0786
0.1216	9.0	72576	0.4121	0.1982	0.0809
0.1204	10.0	80640	0.4407	0.2144	0.0916
0.1123	11.0	88704	0.4334	0.1946	0.0798
0.0949	12.0	96768	0.4478	0.1869	0.0754
0.0808	13.0	104832	0.4396	0.1862	0.0762
0.0702	14.0	112896	0.4605	0.1870	0.0746

Framework versions

• Transformers 4.46.1
• Pytorch 2.1.0+cu118
• Datasets 3.1.0
• Tokenizers 0.20.1