grayarea/Mistral-Small-3.2-24B-Instruct-2506-Heretic-v1.2-2
Abliterated vs unsloth/Mistral-Small-3.2-24B-Instruct-2506. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 32 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 23/192, Attention 32/256, other 0/4, lm_head 0/1
3 derivatives checked
- mradermacher/Mistral-Small-3.2-24B-Instruct-2506-Heretic-v1.2-2-i1-GGUF: Quantization (fewer layers)
- grayarea/Mistral-Small-3.2-24B-Instruct-2506-Heretic-v1.2-2-GGUF: Same weights, converted (fewer layers)
- mradermacher/Mistral-Small-3.2-24B-Instruct-2506-Heretic-v1.2-2-GGUF: Quantization (fewer layers)