dEcQEGdeTshDWbOJQcYp/Qwen3-30B-A3B-Instruct-2507-Heretic
Abliterated vs Qwen/Qwen3-30B-A3B-Instruct-2507. 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: Embeddings 0/1, Norms 0/193, MLP 577/2613, Attention 43/192, lm_head 0/1
2 derivatives checked
- mradermacher/Qwen3-30B-A3B-Instruct-2507-Heretic-i1-GGUF: Quantization
- mradermacher/Qwen3-30B-A3B-Instruct-2507-Heretic-GGUF: Quantization