<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
dsn_afrispeech
This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.7122
- Wer: 40.2191
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: 32
- eval_batch_size: 32
- 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: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.9597 | 0.4 | 500 | 0.9808 | 52.4616 |
| 0.8161 | 0.81 | 1000 | 0.8366 | 43.4443 |
| 0.6999 | 1.21 | 1500 | 0.7818 | 43.4467 |
| 0.6235 | 1.61 | 2000 | 0.7511 | 40.5909 |
| 0.5772 | 2.02 | 2500 | 0.7301 | 42.0866 |
| 0.5506 | 2.42 | 3000 | 0.7210 | 41.3731 |
| 0.5525 | 2.82 | 3500 | 0.7144 | 40.6223 |
| 0.5081 | 3.23 | 4000 | 0.7122 | 40.2191 |
Framework versions
- Transformers 4.29.0.dev0
- Pytorch 2.0.0
- Datasets 2.11.0
- Tokenizers 0.13.3
