huihui-ai/Huihui-Qwen-AgentWorld-35B-A3B-abliterated
Abliterated vs Qwen/Qwen-AgentWorld-35B-A3B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 39 of 40 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: MLP 78/280, Attention 39/280, Norms 0/131, Embeddings 0/1, lm_head 0/1
6 derivatives checked
- iamhsouna/Huihui-Qwen-AgentWorld-35B-A3B-abliterated-GGUF: Quantization
- CrocoCpp/Huihui-Qwen-AgentWorld-35B-A3B-abliterated-Q8_0-GGUF: Quantization
- hotdogs/Qwen35B-Agent-R2-Abliterated: LoRA-style
- hotdogs/qwen35b-a3b-fable-sft-abliterated: LoRA-style
- Gwakweena/Huihui-Qwen-AgentWorld-35B-A3B-abliterated-Q4_K_M-GGUF: Quantization
- coolthor/Huihui-Qwen-AgentWorld-35B-A3B-abliterated-UD-Q3_K_M-GGUF: Quantization