coder3101/Magidonia-24B-v4.3-heretic-v2
Abliterated vs TheDrummer/Magidonia-24B-v4.3. 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 0/81, MLP 12/120, Attention 32/160, lm_head 0/1
2 derivatives checked
- mradermacher/Magidonia-24B-v4.3-heretic-v2-i1-GGUF: Quantization
- mradermacher/Magidonia-24B-v4.3-heretic-v2-GGUF: Quantization