Oriel — AI & technology worth noticing, every day.

huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF

AI model · https://huggingface.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF

First seen by Oriel on .

AI summary

This Hugging Face page provides GGUF variants of an abliteration-based, uncensored Qwen/Qwen3.8-27B model, with conversion details, runtime commands, and warnings about reduced safety filtering.

  • Abliteration removes refusals without TransformerLens and is described as a crude proof-of-concept implementation.
  • Most variants ablate layers 22 through 52, while the UD series ablates layers 17 through 52 and retains the first 15 layers.
  • Targeted weights use nonstandard quantization, including Q8_0 or BF16 conversions, producing special _L filenames and potentially larger files.
  • Ternary variants require the PrismML-Eng/llama.cpp fork, while Ollama and standard llama.cpp usage commands are provided.
  • The page recommends research or controlled use because reduced filtering can produce sensitive content without default safety guarantees.

#gguf #abliteration #quantization #llama.cpp

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Likes

Total likes reported by the source.

  • Hugging Face Models: 2,260, latest observation ; +709 over observed period since

Downloads (30d)

Rolling 30-day downloads. A change in this value is not new downloads during the observed period.

  • Hugging Face Models: 4,253,987, latest observation ; +1,523,665 over observed period since

Aliases

SourceExternal IDURL
Hugging Face Modelshuihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUFhttps://huggingface.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF Read discussion