StructRAG: Boosting Knowledge Intensive Reasoning of LLMs via Inference-time Hybrid Information Structurization
In order to help readers gain a deeper understanding of the concepts in the paper, we release an unrefined version of the experimental code, which can generally reflect the core process of the method described in the paper.
The final, organized version of the completed code (StructRAG framework + Router training) will be released after the paper is officially accepted.