lodrick-the-lafted/Kudzerk-8B
Abliterated vs lodrick-the-lafted/Kudzu-8B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 17 of 32 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/65, MLP 17/96, Attention 17/128
3 derivatives checked
- mradermacher/Kudzerk-8B-i1-GGUF: Quantization
- featherless-ai-quants/lodrick-the-lafted-Kudzerk-8B-GGUF: Quantization
- mradermacher/Kudzerk-8B-GGUF: Same weights, converted