huihui-ai/Huihui-Qwen3.6-27B-abliterated
Abliterated vs Qwen/Qwen3.6-27B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 64 of 64 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/5, MLP 65/307, Attention 65/560, lm_head 0/1, Norms 0/325, other 0/1
13 derivatives checked
- mradermacher/Huihui-Qwen3.6-27B-abliterated-GGUF: Quantization
- mradermacher/Huihui-Qwen3.6-27B-abliterated-i1-GGUF: Quantization
- sakamakismile/Huihui-Qwen3.6-27B-abliterated-NVFP4: Quantization
- lhca521/Huihui-Qwen3.6-27B-abliterated-GGUF: Quantization
- sakamakismile/Huihui-Qwen3.6-27B-abliterated-NVFP4-TEXT-MTP: Quantization
- hotdogs/qwen27b-abliterated-Fable-MTP: LoRA-style
- tomvaillant/qwen3.6-27b-abliterated-journalist-GGUF: Quantization
- YuYu1015/Huihui-Qwen3.6-27B-abliterated-UD-Q4_K_XL-MTP-GGUF: Quantization
- alexxorm/Huihui-Qwen3.6-27B-abliterated-AWQ: Quantization
- divinetribe/Qwen3.6-27B-abliterated-4bit-mlx: Quantization
- NINI26454/Huihui-Qwen3.6-27B-abliterated-MTP-Q3_K-GGUF: Quantization
- YuYu1015/Huihui-Qwen3.6-27B-abliterated-int4-AutoRound: Quantization
- nbeerbower/Huihui-Qwen3.6-27B-abliterated-Athanorlite-ORPO: LoRA-style
5 other repos hold byte-for-byte identical weight files.