The GPQA (Graduate-Level Google-Proof Q&A) benchmark is a challenging dataset designed to assess the capabilities of large language models (LLMs) in handling complex, domain-specific questions in biology, physics, and chemistry. It consists of 448 multiple-choice questions

→ View original post on X — @testingcatalog