jordanwoodson/Qwen3.5-2B-heretic
Abliterated vs Qwen/Qwen3.5-2B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 24 of 24 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Attention 24/264, Norms 0/177, Embeddings 0/4, MLP 6/172
2 derivatives checked
- Abiray/Qwen3.5-2B-heretic-GGUF: Quantization
- jordanwoodson/Qwen3.5-2B-heretic-GGUF: Same weights, converted