Evaluation Methods

LLM-Agent Research Coalesces Around Generalizability, RLVR, and Grounded Retrieval

New surveys, benchmarks and modular methods chart a path to more reliable agents.