whisper-small-pissa-afrispeech-general
This model is a fine-tuned version of openai/whisper-small on the afrispeech-200 dataset. It achieves the following results on the evaluation set:
- Loss: 0.6409
- Wer: 0.175
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: 0.0001
- 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: linear
- lr_scheduler_warmup_steps: 1349
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.5217 | 1.0 | 1350 | 0.6052 | 0.187 |
| 0.4299 | 2.0 | 2700 | 0.5552 | 0.179 |
| 0.2937 | 3.0 | 4050 | 0.5484 | 0.175 |
| 0.2074 | 4.0 | 5400 | 0.5644 | 0.172 |
| 0.1376 | 5.0 | 6750 | 0.5928 | 0.177 |
| 0.094 | 6.0 | 8100 | 0.6126 | 0.176 |
| 0.0649 | 7.0 | 9450 | 0.6409 | 0.175 |
Framework versions
- Transformers 4.44.0
- Pytorch 2.10.0+cu128
- Datasets 2.19.0
- Tokenizers 0.19.1