Browsing: diverse prompts

Researchers have developed a machine learning technique to improve red-teaming for large language models, making them safer by generating diverse prompts that trigger a…