huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated
Abliterated vs Qwen/Qwen3.6-35B-A3B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 40 of 40 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/5, Attention 41/392, MLP 82/399, Norms 0/247, lm_head 0/1, other 0/1
15 derivatives checked
- mradermacher/Huihui-Qwen3.6-35B-A3B-abliterated-GGUF: Quantization
- mradermacher/Huihui-Qwen3.6-35B-A3B-abliterated-i1-GGUF: Quantization
- Abiray/Huihui-Qwen3.6-35B-A3B-abliterated-GGUF: Quantization
- sakamakismile/Huihui-Qwen3.6-35B-A3B-abliterated-NVFP4: Quantization
- coolthor/Huihui-Qwen3.6-35B-A3B-abliterated-FP8-DYNAMIC: Quantization
- THe-Plague/Qwen3.6-35B-A3B-abliterated-NVFP4-MTP: Quantization
- YuYu1015/Huihui-Qwen3.6-35B-A3B-abliterated-NVFP4: Quantization
- rafw007/Qwen3.6-35B-A3B-mlx-claude-coder-abliterated: Quantization
- leonsarmiento/Huihui-Qwen3.6-35B-A3B-abliterated-3bit-XL-mlx: Quantization
- augustine223/Huihui-Qwen3.6-35B-A3B-abliterated-KO-i1-GGUF: Quantization
- YuYu1015/Huihui-Qwen3.6-35B-A3B-abliterated-int4-AutoRound: Quantization
- vanch007/Huihui-Qwen3.6-35B-A3B-abliterated-mlx-4bit: Quantization
- coolthor/Huihui-Qwen3.6-35B-A3B-abliterated-NVFP4-W4A4: Quantization
- vanch007/Huihui-Qwen3.6-35B-A3B-abliterated-mlx-bf16: Partial fine-tune
- leonsarmiento/Huihui-Qwen3.6-35B-A3B-abliterated-6bit-XL-mlx: Quantization
5 other repos hold byte-for-byte identical weight files.