jensg/whisper-small-dv

Whisper Small Dv - Sanchit Gandhi

This model is a fine-tuned version of openai/whisper-small on the Common Voice 13 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3085
• Wer Ortho: 58.3606
• Wer: 11.6320

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• training_steps: 4000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
0.1119	1.63	500	0.1645	60.8817	12.9082
0.0457	3.26	1000	0.1677	59.3286	11.7450
0.0317	4.89	1500	0.1903	58.3258	11.3798
0.012	6.51	2000	0.2292	58.0403	11.7085
0.007	8.14	2500	0.2595	56.9538	11.0982
0.0061	9.77	3000	0.2606	56.6404	10.8878
0.0052	11.4	3500	0.2737	56.8911	11.1799
0.0033	13.03	4000	0.3085	58.3606	11.6320

Framework versions

• Transformers 4.32.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.13.1
• Tokenizers 0.13.3