3MPER0RR/Qwen2.5-Coder-14B-Instruct-3MPER0RR-abliterated
Abliterated vs Qwen/Qwen2.5-Coder-14B-Instruct. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 34 of 48 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/97, MLP 31/144, Attention 34/336, lm_head 0/1