philbert440/Qwen3.8-27B-Uncensored-Cyber
Abliterated vs philbert440/Qwen3.8-27B-Uncensored-Aggressive. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 60 of 64 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: lm_head 0/1, Embeddings 0/5, Norms 0/325, Attention 60/560, MLP 60/307, other 0/1
14 derivatives checked
- philbert440/Qwen3.8-27B-Uncensored-Cyber-GGUF: Not comparable yet
- cyjin-yl/Qwen3.8-27B-Uncensored-Cyber-agentic-imatrix-GGUF: Quantization
- paubau/Qwen3.8-27B-Uncensored-Cyber-agentic-imatrix-GGUF: Quantization
- philbert440/Qwen3.8-27B-Uncensored-Cyber-W4A16-AWQ: Quantization
- philbert440/Qwen3.8-27B-Uncensored-Cyber-NVFP4: Quantization
- bjivanovich/Qwen3.8-27B-Uncensored-Cyber-ATX-MTP-GGUF: Quantization
- Narium/Qwen3.8-27B-Uncensored-Cyber-agentic-imatrix-GGUF: Quantization
- 0xSojalSec/Qwen3.8-27B-Uncensored-Cyber-agent: Quantization
- melihovoleg173/Qwen3.8-27B-Uncensored-Cyber-agentic-imatrix-GGUF: Quantization
- philbert440/Qwen3.8-27B-Uncensored-Cyber-MLX-4bit: Quantization
- Astark-40/Qwen3.8-27B-Uncensored-Cyber-GGUF: Not comparable yet
- rwcii/Qwen3.8-27B-Uncensored-Cyber-MLX-8bit: Quantization
- Aydge/Qwen3.8-27B-Uncensored-Cyber-GGUF: Not comparable yet
- rwcii/Qwen3.8-27B-Uncensored-Cyber-MLX-4bit: Quantization