TRAD: Enhancing LLM Agents with Step-Wise Thought Retrieval and Aligned Decision

Open in new window