Rootkit7/Ternary-Bonsai-4B-abliterated
Abliterated vs prism-ml/Ternary-Bonsai-4B-unpacked. Looks like abliteration (refusal removal): only the output projections (down_proj, embed_tokens, o_proj) changed, in 36 of 36 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: MLP 36/108, Norms 0/145, Attention 36/144, Embeddings 1/1