metadata
base_model: microsoft/wavlm-base
tags:
- audio-classification
- deepfake
- audio-spoof
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: wavlm-base-960h-asv19-deepfake
results: []
wavlm-base-960h-asv19-deepfake
This model is a fine-tuned version of microsoft/wavlm-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0332
- Accuracy: 0.9950
- Far: 0.0416
- Frr: 0.0009
- Eer: 0.0212
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-06
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 4
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2.0
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Far | Frr | Eer |
|---|---|---|---|---|---|---|---|
| 0.2408 | 0.39 | 2500 | 0.0574 | 0.9889 | 0.0863 | 0.0025 | 0.0444 |
| 0.0372 | 0.79 | 5000 | 0.0524 | 0.9901 | 0.0914 | 0.0006 | 0.0460 |
| 0.0231 | 1.18 | 7500 | 0.0539 | 0.9912 | 0.0824 | 0.0004 | 0.0414 |
| 0.0213 | 1.58 | 10000 | 0.0301 | 0.9951 | 0.0361 | 0.0013 | 0.0187 |
| 0.0191 | 1.97 | 12500 | 0.0332 | 0.9950 | 0.0416 | 0.0009 | 0.0212 |
Framework versions
- Transformers 4.38.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.16.2.dev0
- Tokenizers 0.15.2