Pluto-AI-Labs/Hades-4B-Abliterated
Abliterated vs Qwen/Qwen3-4B-Instruct-2507. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 26 of 36 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/145, MLP 26/108, Attention 26/144
3 derivatives checked
- mradermacher/Hades-4B-Abliterated-i1-GGUF: Quantization
- mradermacher/Hades-4B-Abliterated-GGUF: Same weights, converted
- Pluto-AI-Labs/Hades-4B-GGUF: Same weights, converted