deeepfake-audio-555
This model is a fine-tuned version of facebook/wav2vec2-base on the audiofolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.4156
- Accuracy: 0.9247
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: 3e-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_ratio: 0.01
- num_epochs: 16
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.6428 | 1.0 | 46 | 0.6271 | 0.7204 |
| 0.4622 | 2.0 | 92 | 0.4054 | 0.8602 |
| 0.3098 | 3.0 | 138 | 0.5667 | 0.8172 |
| 0.2696 | 4.0 | 184 | 0.4179 | 0.8817 |
| 0.2806 | 5.0 | 230 | 0.4129 | 0.8710 |
| 0.2078 | 6.0 | 276 | 0.3541 | 0.9140 |
| 0.1652 | 7.0 | 322 | 0.3338 | 0.9140 |
| 0.0871 | 8.0 | 368 | 0.4072 | 0.9140 |
| 0.1267 | 9.0 | 414 | 0.3649 | 0.9247 |
| 0.0651 | 10.0 | 460 | 0.3436 | 0.9355 |
| 0.0976 | 11.0 | 506 | 0.4163 | 0.9140 |
| 0.0186 | 12.0 | 552 | 0.4164 | 0.9247 |
| 0.0324 | 13.0 | 598 | 0.4156 | 0.9247 |
Framework versions
- Transformers 4.39.3
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2