NovaCorp/Creative_Writing-Heretic-3.2-1B
Abliterated vs Green-Eye/Llama-3.2-1B-Instruct-heretic. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 14 of 16 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/33, MLP 14/48, Attention 14/64
2 derivatives checked
- mradermacher/Creative_Writing-Heretic-3.2-1B-i1-GGUF: Quantization
- mradermacher/Creative_Writing-Heretic-3.2-1B-GGUF: Same weights, converted