llmfan46/Qwen3.5-9B-Nikusui-v1-Uncensored-Heretic-Native-MTP-Preserved
Abliterated vs extraltodeus/Qwen3.5-9B-Nikusui-v1. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 32 of 32 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: lm_head 0/1, Embeddings 0/5, MLP 24/211, Attention 32/336, other 0/1, Norms 0/221
4 derivatives checked
- llmfan46/Qwen3.5-9B-Nikusui-v1-Uncensored-Heretic-Native-MTP-Preserved-GGUF: Same weights, converted
- mradermacher/Qwen3.5-9B-Nikusui-v1-Uncensored-Heretic-Native-MTP-Preserved-GGUF: Same weights, converted
- mradermacher/Qwen3.5-9B-Nikusui-v1-Uncensored-Heretic-Native-MTP-Preserved-i1-GGUF: Quantization
- nuofang/Qwen3.5-9B-Nikusui-v1-Uncensored-Heretic-Native-MTP-Preserved-GGUF: Not comparable yet