megabytes/ZwZ-8B-heretic
Abliterated vs inclusionAI/ZwZ-8B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 29 of 36 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: lm_head 0/1, MLP 29/232, Norms 0/261, Attention 27/252, Embeddings 0/4
1 derivatives checked
- filvyb/ZwZ-8B-heretic-GGUF: Same weights, converted (fewer layers)