coder3101/Ministral-3-8B-Reasoning-2512-heretic
Abliterated vs mistralai/Ministral-3-8B-Reasoning-2512. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 29 of 34 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/119, MLP 29/174, Attention 5/232, other 0/4, lm_head 0/1
3 derivatives checked
- coder3101/Ministral-3-8B-Reasoning-2512-heretic-GGUF: Same weights, converted (fewer layers)
- FadedRedStar/Ministral-3-8B-Reasoning-2512-heretic-imatrix-GGUF: Quantization (fewer layers)
- FadedRedStar/Ministral-3-8B-Reasoning-2512-heretic-GGUF: Quantization (fewer layers)