wav2vec2-base-960h-heart-sounds
This model is a fine-tuned version of facebook/wav2vec2-base-960h on the audiofolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.3595
- Accuracy: 0.8674
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
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 8
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.9791 | 1.0 | 83 | 0.9290 | 0.5442 |
| 0.6532 | 2.0 | 166 | 0.5495 | 0.8186 |
| 0.5202 | 3.0 | 249 | 0.4569 | 0.8216 |
| 0.4421 | 4.0 | 332 | 0.4378 | 0.8399 |
| 0.4144 | 5.0 | 415 | 0.3853 | 0.8765 |
| 0.4213 | 6.0 | 498 | 0.3835 | 0.8537 |
| 0.3819 | 7.0 | 581 | 0.3647 | 0.8674 |
| 0.3994 | 7.9119 | 656 | 0.3595 | 0.8674 |
Framework versions
- Transformers 4.49.0
- Pytorch 2.0.1+cu118
- Datasets 3.3.2
- Tokenizers 0.21.0