auto-redteam
Open-source project that red-teams language models by using one LLM to automatically generate jailbreak attempts against another, such as GPT-4 against GPT-3.5.
Open sourceResearch & Education
Visit website(opens in a new tab)
Open-source project that red-teams language models by using one LLM to automatically generate jailbreak attempts against another, such as GPT-4 against GPT-3.5.