Beyond Benchmarks: LLM Evaluation with an Anthropomorphic and Lifecycle-oriented Roadmap

Jun Wang, Ninglun Gu, Kailai Zhang +9

Despite their rapid advancement, large language models (LLMs) suffer from a critical disconnect between benchmark scores and real-world utility. Current evaluation remains fragment…