huihui-ai/Huihui-Mistral-Small-3.2-24B-Instruct-2506-abliterated
Abliterated vs mistralai/Mistral-Small-3.2-24B-Instruct-2506. 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 0/131, MLP 40/192, Attention 40/256, other 0/4, lm_head 0/1
2 derivatives checked
- maerong3/Huihui-Mistral-Small-3.2-24B-Instruct-2506-abliterated-Q4_K_M-GGUF: Quantization (fewer layers)
- therummage/Huihui-Mistral-Small-3.2-24B-Instruct-2506-abliterated-Q8_0-GGUF: Quantization (fewer layers)
2 other repos hold byte-for-byte identical weight files.