VagueBench

New DialogGRPO Method Enhances LLM Clarification Skills

Researchers have introduced a new benchmark and training method called DialogGRPO to improve how large language models handle incomplete user queries. Current LLMs often fail to ask clarifying questions, with fewer than…