Evaluation of Large Language Models in Multi-hop Reasoning Tasks

The author introduces an LLM MHQA evaluation benchmark to address limitations in current benchmarks and highlights the need for trustworthy evaluation methods to assess LLM reasoning ability accurately.