FiditeNemini/Unhinged-Qwen2.5-R1-1M-Uncensored-BF16
Abliterated vs huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 35 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 25/144, Attention 21/336, lm_head 0/1