deeepfake-audio-Recognition-ttoo

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: 0.2288
• Accuracy: 0.9545

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
0.6417	1.0	33	0.5774	0.7879
0.4818	2.0	66	0.3792	0.8485
0.2756	3.0	99	0.3066	0.8788
0.3106	4.0	132	0.1951	0.9545
0.2138	5.0	165	0.2078	0.9394
0.0988	6.0	198	0.3227	0.9091
0.1043	7.0	231	0.2893	0.9394
0.0808	8.0	264	0.2177	0.9545
0.1312	9.0	297	0.2846	0.9091
0.0667	10.0	330	0.1955	0.9545
0.0513	11.0	363	0.2553	0.9545
0.0217	12.0	396	0.1708	0.9545
0.0136	13.0	429	0.1641	0.9545
0.0236	14.0	462	0.2203	0.9545
0.0097	15.0	495	0.2253	0.9545
0.003	16.0	528	0.2288	0.9545

Framework versions

• Transformers 4.39.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2