Gustking/wav2vec2-large-xlsr-deepfake-audio-classification
Full fine-tune vs alefiury/wav2vec2-large-xlsr-53-gender-recognition-librispeech. Fine-tune: 100% of the weights changed, across 24 of 24 layers.
Tensors changed: other 20/20, Norms 114/114, Attention 192/192, MLP 96/96, Embeddings 2/2
2 derivatives checked
- garystafford/wav2vec2-deepfake-voice-detector: Partial fine-tune
- zodumair/wav2vec2-deepfake-voice-detector: Partial fine-tune
3 other repos hold byte-for-byte identical weight files.