LLMEval-Fair
eval200K+ generative QA items across 13 disciplines for robust and fair LLM evaluation (ACL 2026) — the latest in FudanNLP's LLMEval line (Xuanjing Huang), which also includes LLMEval-Med (EMNLP 2025, arXiv:2506.04078), maintained with a live leaderboard.
Paper
Evaluation Details
Questions 200,000
Domains 5
Scoring generative QA graded for robust/fair evaluation
Domains: science, engineering, medicine, law, humanities