wavlm-base-960h-itw-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: 0.0593
• Accuracy: 0.9896
• FAR: 0.0080
• FRR: 0.0144
• EER: 0.0112

Model description

Quick Use

  device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

  config = AutoConfig.from_pretrained("abhishtagatya/wavlm-base-960h-itw-deepfake")
  feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained("abhishtagatya/wavlm-base-960h-itw-deepfake")

  model = WavLMForSequenceClassification.from_pretrained("abhishtagatya/wavlm-base-960h-itw-deepfake", config=config).to(device)

  # Your Logic Here

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
0.3205	0.39	2500	0.1223	0.9699	0.0343	0.0229	0.0286
0.0752	0.79	5000	0.0822	0.9843	0.0145	0.0178	0.0161
0.0666	1.18	7500	0.0825	0.9849	0.0158	0.0140	0.0149
0.0569	1.57	10000	0.0674	0.9884	0.0103	0.0140	0.0121
0.0567	1.97	12500	0.0593	0.9896	0.0080	0.0144	0.0112

Framework versions

• Transformers 4.38.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.16.2.dev0
• Tokenizers 0.15.1