Goekdeniz-Guelmez/Qwen3-0.6B-gabliterated
Abliterated vs Qwen/Qwen3-0.6B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 21 of 28 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/113, MLP 21/84, Attention 21/112
4 derivatives checked
- mradermacher/Gabliterated-Qwen3-0.6B-i1-GGUF: Quantization
- mradermacher/Gabliterated-Qwen3-0.6B-GGUF: Abliterated
- mradermacher/Qwen3-0.6B-gabliterated-GGUF: Abliterated
- Goekdeniz-Guelmez/Josiefied-Qwen3-0.6B-gabliterated-v1: Abliterated