meta-llama/PurpleLlama
A suite of tools, benchmarks, and safeguards from Meta for evaluating and improving the cybersecurity risks and safety of open generative AI models.
Not currently ranked — collecting fresh signals.
star history
Purple Llama is an umbrella project offering tools and evaluations for responsible generative AI development. It includes CyberSec Eval, a benchmark for assessing cybersecurity risks in LLMs, and Llama Guard, an input/output safeguard model. The project adopts a purple-team approach combining offensive (red) and defensive (blue) security postures to comprehensively evaluate and mitigate risks in AI systems.
Frequently asked
- What is meta-llama/PurpleLlama?
- A suite of tools, benchmarks, and safeguards from Meta for evaluating and improving the cybersecurity risks and safety of open generative AI models.
- Is PurpleLlama open source?
- Yes — meta-llama/PurpleLlama is an open-source project tracked on heatdrop.
- What language is PurpleLlama written in?
- meta-llama/PurpleLlama is primarily written in Python.
- How popular is PurpleLlama?
- meta-llama/PurpleLlama has 4.3k stars on GitHub.
- Where can I find PurpleLlama?
- meta-llama/PurpleLlama is on GitHub at https://github.com/meta-llama/PurpleLlama.