llmfan46/Qwen3.5-35B-A3B-uncensored-heretic
Abliterated vs Qwen/Qwen3.5-35B-A3B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 37 of 40 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: lm_head 0/1, Embeddings 0/4, Attention 37/388, MLP 35/312, Norms 0/241
5 derivatives checked
- mradermacher/Qwen3.5-35B-A3B-uncensored-heretic-i1-GGUF: Quantization
- llmfan46/Qwen3.5-35B-A3B-uncensored-heretic-GGUF: Same weights, converted
- mradermacher/Qwen3.5-35B-A3B-heretic-v1-i1-GGUF: Quantization
- mradermacher/Qwen3.5-35B-A3B-uncensored-heretic-GGUF: Quantization
- mradermacher/Qwen3.5-35B-A3B-heretic-v1-GGUF: Quantization
1 other repos hold byte-for-byte identical weight files.