SicariusSicariiStuff/Qwen3.5-0.8B_Abliterated
Abliterated vs Qwen/Qwen3.5-0.8B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 17 of 24 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Attention 17/216, Norms 0/129, Embeddings 0/4, MLP 15/124
5 derivatives checked
- mradermacher/Qwen3.5-0.8B_Abliterated-i1-GGUF: Quantization
- mradermacher/Qwen3.5-0.8B_Abliterated-GGUF: Same weights, converted
- SicariusSicariiStuff/Qwen3.5-0.8B_Abliterated_GGUF: Quantization
- SicariusSicariiStuff/Qwen3.5-0.8B_Abliterated_ARM: Quantization
- jduartedj/MiniCPM-V-4.6-0.8B-Abliterated: Same weights, repacked (fewer layers)