sbussiso/Qwen2.5-7B-abliterated
Abliterated vs Qwen/Qwen2.5-7B-Instruct. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 3 of 28 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/57, MLP 3/84, Attention 3/196, lm_head 0/1