Sabomako/Qwen3-Omni-30B-A3B-Thinking-heretic
Abliterated vs Qwen/Qwen3-Omni-30B-A3B-Thinking. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 43 of 48 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: other 0/11, MLP 360/1989, Norms 0/439, Attention 43/556, Embeddings 0/4, lm_head 0/1
2 derivatives checked
- mradermacher/Qwen3-Omni-30B-A3B-Thinking-heretic-GGUF: Quantization
- mradermacher/Qwen3-Omni-30B-A3B-Thinking-heretic-i1-GGUF: Quantization