Reproduce “TourRank: Utilizing Large Language Models for Documents Ranking with a Tournament-Inspired Strategy” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yiqun Chen, Qi Liu, Yi Zhang, Weiwei Sun, Xinyu Ma, Wei Yang, Daiting Shi, Jiaxin Mao and 1 more
TourRank introduces a tournament-inspired, multi-stage strategy for zero-shot document ranking using Large Language Models, achieving state-of-the-art performance on TREC DL and BEIR benchmarks. The method effectively mitigates positional bias and context window limitations, demonstrating high robustness to initial document order while maintaining computational efficiency.
Key insights
- A tournament-inspired framework effectively enables LLMs to perform global ranking on large document sets by orchestrating multiple local comparisons within their context window.
- Running multiple parallel tournament rounds and aggregating results via a points system significantly improves ranking robustness against initial document order and LLM output variability.
- Strategic parallel processing and progressive filtering offer a favorable balance between ranking effectiveness and computational efficiency for LLM-based re-ranking.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.