Naphula/L3-8B-Stheno-v3.2-MPOA
Abliterated vs Sao10K/L3-8B-Stheno-v3.2. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 32 of 32 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/65, MLP 32/96, Attention 32/128
2 derivatives checked
- mradermacher/L3-8B-Stheno-v3.2-MPOA-i1-GGUF: Quantization
- mradermacher/L3-8B-Stheno-v3.2-MPOA-GGUF: Same weights, converted