deepfake-spanish-wav2vec2-linear-augmented

This model is a fine-tuned version of Gustking/wav2vec2-large-xlsr-deepfake-audio-classification on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4476
• Accuracy: 0.9090
• F1: 0.9083
• Precision: 0.9230
• Recall: 0.9090
• Roc Auc: 0.9994
• Eer: 0.0048

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: 32
• eval_batch_size: 48
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 75
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall	Roc Auc	Eer
2.1469	0.2133	100	0.4042	0.8847	0.8832	0.9063	0.8847	0.9965	0.0285
2.1469	0.4267	200	0.3700	0.8874	0.8860	0.9081	0.8874	0.9984	0.0163
2.1469	0.64	300	0.5706	0.8837	0.8821	0.9056	0.8837	0.9974	0.0123
2.1469	0.8533	400	0.3749	0.9128	0.9121	0.9257	0.9128	0.9995	0.0059
0.6001	1.0661	500	0.3203	0.9232	0.9227	0.9334	0.9232	0.9996	0.0043
0.6001	1.2795	600	0.3738	0.9168	0.9162	0.9286	0.9168	0.9994	0.0048
0.6001	1.4928	700	0.3272	0.9304	0.9300	0.9389	0.9304	0.9976	0.0099
0.6001	1.7061	800	0.4476	0.9090	0.9083	0.9230	0.9090	0.9994	0.0048

Framework versions

• Transformers 5.2.0
• Pytorch 2.10.0+cu130
• Datasets 4.6.1
• Tokenizers 0.22.2