flammenai/Mahou-1.2-phi-14B
Abliterated vs microsoft/Phi-3-medium-128k-instruct. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 40 of 40 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/81, MLP 40/80, Attention 40/80, lm_head 0/1
1 derivatives checked
- mradermacher/Mahou-1.2-phi-14B-GGUF: Quantization