Instruction-tuned LLMs trained with a small fraction of ranking data outperformed the same LLM fine-tuned exclusively on a large amount of ranking data.
Notes on verification
Directly corroborated by the peer-reviewed NeurIPS 2024 paper (Yu et al., RankRAG), arXiv preprint, OpenReview PDF, and official NeurIPS proceedings, all containing identical wording of the claim.
Sources
- unifying context ranking with retrieval-augmented generation in LLMs (dl.acm.org)
- https://neurips.cc/virtual/2024/poster/95135 (neurips.cc)
- https://arxiv.org/abs/2407.02485v1 (arxiv.org)
- https://openreview.net/pdf/e799910ea1c9e2dfb86d87d93e60724fc05e0aab.pdf (openreview.net)
- https://proceedings.neurips.cc/paper_files/paper/2024/file/db93ccb6cf392f352570dd5af0a223d3-Paper-Conference.pdf (proceedings.neurips.cc)