whisper_small-fa_v02

This model is a fine-tuned version of openai/whisper-small on the mozilla-foundation/common_voice_11_0 fa dataset. We also did data augmentation using audiomentations library. It achieves the following results on the evaluation set:

• Loss: 0.2291
• Wer: 30.3423

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

You can Find the notebooks here.

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Step	Training Loss	Validation Loss	Wer
500	1.770700	0.476709	52.29181
1000	0.762300	0.368512	41.83410
1500	0.645000	0.323680	37.57881
2000	0.601900	0.297370	36.43209
2500	0.529700	0.276422	33.52608
3000	0.523200	0.260825	31.94485
3500	0.488400	0.249957	33.11771
4000	0.464800	0.241462	30.34238
4500	0.440500	0.233215	31.04969
5000	0.440500	0.229116	30.73605

Framework versions

• Transformers 4.26.0
• Pytorch 2.0.1+cu117
• Datasets 2.8.0
• Tokenizers 0.13.3