whisper-base-speech-commands
This model is a fine-tuned version of openai/whisper-base on the speech_commands dataset. It achieves the following results on the evaluation set:
- Loss: 1.1307
- Accuracy: 0.8067
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: 96
- eval_batch_size: 96
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.2604 | 1.0 | 412 | 1.0617 | 0.7977 |
| 0.1168 | 2.0 | 824 | 1.0024 | 0.8017 |
| 0.1527 | 3.0 | 1236 | 0.9757 | 0.8022 |
| 0.0637 | 4.0 | 1648 | 1.0066 | 0.8004 |
| 0.0631 | 5.0 | 2060 | 1.0504 | 0.8053 |
| 0.0554 | 6.0 | 2472 | 1.1307 | 0.8067 |
| 0.1075 | 7.0 | 2884 | 1.1664 | 0.8017 |
| 0.021 | 8.0 | 3296 | 1.4746 | 0.8044 |
| 0.0144 | 9.0 | 3708 | 1.3729 | 0.8044 |
| 0.0158 | 10.0 | 4120 | 1.3561 | 0.8040 |
| 0.0504 | 11.0 | 4532 | 1.3289 | 0.8053 |
Framework versions
- Transformers 4.43.3
- Pytorch 2.2.2+cu121
- Datasets 2.18.0
- Tokenizers 0.19.1