thu-coai/Safety-Prompts
Chinese safety prompts for evaluating and improving the safety of LLMs. 中文安全prompts，用于评估和提升大模型的安全性。
This project provides a comprehensive collection of Chinese safety prompts and corresponding AI model responses. It helps developers and researchers train and fine-tune large language models to be safer and more aligned with human values by providing a dataset of challenging, unsafe prompts. The output is a more robust and ethically sound AI model capable of handling diverse and sensitive user inputs appropriately.
1,135 stars. No commits in the last 6 months.
Use this if you are a large language model developer or researcher aiming to improve your Chinese LLM's safety features and alignment with human values during training or fine-tuning.
Not ideal if you primarily need to evaluate an LLM's safety; for that, refer to the recommended SafetyBench or ShieldLM platforms, which offer more dedicated evaluation capabilities.
Stars
1,135
Forks
88
Language
—
License
Apache-2.0
Category
Last pushed
Feb 27, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/prompt-engineering/thu-coai/Safety-Prompts"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related tools
ahmadbuilds/multi-agent-hr-assistant
An autonomous, multi-agent HR service desk. It uses a supervisor architecture to route employee...
x-hannibal/open-webui-easymage
Multi-engine image generation filter for Open WebUI. Features automated prompt enhancement,...
recordrnase224/brix-protocol
Build runtime reliability guards for LLM pipelines with one wrap() call and configurable...
wisterx-spec/agent-rails
Description：Opinionated workflow framework for AI-assisted development — rules, skills &...
Dewensong/email-marketing-skill
Reusable email marketing skill with local-first setup, knowledge-driven drafting, SMTP/IMAP...