DreamFast/qwen3-8b-heretic
Abliterated vs Qwen/Qwen3-8B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 32 of 36 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/145, MLP 28/108, Attention 32/144, lm_head 0/1
1 derivatives checked
- liodon-ai/qwen3-8b-heretic-imatrix-GGUF: Quantization
1 other repos hold byte-for-byte identical weight files.