gorbatjovy/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-heretic
Abliterated vs DavidAU/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1. Looks like abliteration (refusal removal): only the output projections (down_proj, embed_tokens, o_proj, out_proj) changed, in 51 of 64 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Norms 0/325, Attention 51/560, MLP 51/307, Embeddings 1/5, lm_head 0/1, other 0/1