canyrtcn/Gemma_E4B_Kizagan_Abliterated
Abliterated vs AlicanKiraz0/Kizagan-E4B-Turkish-Reasoning-Model. Looks like abliteration (refusal removal): only the output projections (o_proj) changed, in 21 of 42 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: MLP 0/606, Norms 0/502, other 0/261, Attention 21/755, Embeddings 0/6
2 derivatives checked
- mradermacher/Gemma_E4B_Kizagan_Abliterated-GGUF: Same weights, converted (fewer layers)
- canyrtcn/Gemma_E4B_Kizagan_Abliterated-GGUF: Same weights, converted (fewer layers)