Heretic github.com ↗
An open-source tool for automatically removing refusal behavior from language models.
ToolAI & MLabliterationllmcensorship-removalopen-weights
Recommended in 1 thread
Memdoor: A team of 5 Qwen3.8-27B models beats Fable 5 on LiveCodeBench Hardgithub.com ·
Makes DIY abliteration easy, so HuggingFace takedowns would matter little.
Similar tools
Also in AI & ML
- 11 threads
- 7 threads
- DeepSeek6 threads
A family of large language models offered via a low-cost API.
- 6 threads
- 3 threads
- Claude3 threads
Anthropic's AI model and command-line assistant.
- 3 threads
- 3 threads
- 3 threads
- 3 threads
- Qwen3 threads
Alibaba's family of open-weight large language models.
- oh-my-pi3 threads
A setup built around the pi coding agent.
Link from a commenter, not checked or endorsed by us: ↗ https://github.com/p-e-w/heretic