seven safety

New framework creates multilingual LLM safety and reasoning evaluations

Researchers have developed a new framework for creating multilingual evaluations of large language models (LLMs) that aims to preserve the meaning of benchmarks across different languages. This reasoning-guided translat…