coder3101/Qwen3.5-0.8B-heretic
Abliterated vs Qwen/Qwen3.5-0.8B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 18 of 24 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Attention 16/216, Norms 0/129, Embeddings 0/4, MLP 18/124
4 derivatives checked
- FadedRedStar/Qwen3.5-0.8B-heretic-imatrix-GGUF: Same weights, converted (fewer layers)
- FadedRedStar/Qwen3.5-0.8B-heretic-GGUF: Same weights, converted (fewer layers)
- Alexs28/Qwen3.5-0.8B-heretic-250-v1-GGUF: Quantization
- ENOSYS/Qwen3.5-0.8B-heretic-250-v1-GGUF: Quantization