rahulshendre/deepfake-detector-model-v1
Full fine-tune (fewer layers) vs google/siglip2-base-patch16-512. rahulshendre/deepfake-detector-model-v1 matches part of google/siglip2-base-patch16-512 (207 of its 408 tensors; the rest of google/siglip2-base-patch16-512 has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 12 of 12 layers.
Tensors changed: Embeddings 2/2, Norms 52/52, MLP 52/52, Attention 100/100, other 1/1
2 other repos hold byte-for-byte identical weight files.