{"paper":{"title":"RankZephyr: Effective and Robust Zero-Shot Listwise Reranking is a Breeze!","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"An open-source LLM for listwise zero-shot reranking matches or surpasses GPT-4 on multiple retrieval benchmarks.","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Jimmy Lin, Ronak Pradeep, Sahel Sharifymoghaddam","submitted_at":"2023-12-05T12:39:00Z","abstract_excerpt":"In information retrieval, proprietary large language models (LLMs) such as GPT-4 and open-source counterparts such as LLaMA and Vicuna have played a vital role in reranking. However, the gap between open-source and closed models persists, with reliance on proprietary, non-transparent models constraining reproducibility. Addressing this gap, we introduce RankZephyr, a state-of-the-art, open-source LLM for listwise zero-shot reranking. RankZephyr not only bridges the effectiveness gap with GPT-4 but in some cases surpasses the proprietary model. Our comprehensive evaluations across several datas"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"RankZephyr not only bridges the effectiveness gap with GPT-4 but in some cases surpasses the proprietary model.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the NovelEval test set truly contains only queries and passages created after the model's training cutoff and that no leakage occurred during fine-tuning or evaluation.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"RankZephyr is a new open-source LLM that closes the effectiveness gap with GPT-4 for zero-shot listwise reranking while showing robustness to input ordering and document count.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"An open-source LLM for listwise zero-shot reranking matches or surpasses GPT-4 on multiple retrieval benchmarks.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"f473119540cb2855e43226b0423a89f1c9c9e6a26b336dddd63688b2e7f416f4"},"source":{"id":"2312.02724","kind":"arxiv","version":1},"verdict":{"id":"60b5ed54-f5fe-4bea-8172-5d0e845236b7","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T23:36:55.601641Z","strongest_claim":"RankZephyr not only bridges the effectiveness gap with GPT-4 but in some cases surpasses the proprietary model.","one_line_summary":"RankZephyr is a new open-source LLM that closes the effectiveness gap with GPT-4 for zero-shot listwise reranking while showing robustness to input ordering and document count.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the NovelEval test set truly contains only queries and passages created after the model's training cutoff and that no leakage occurred during fine-tuning or evaluation.","pith_extraction_headline":"An open-source LLM for listwise zero-shot reranking matches or surpasses GPT-4 on multiple retrieval benchmarks."},"references":{"count":42,"sample":[{"doi":"","year":2016,"title":"MS MARCO: A Human Generated MAchine Reading COmprehension Dataset","work_id":"78d498ce-11db-4f88-8eb0-40e0f86af615","ref_index":1,"cited_arxiv_id":"1611.09268","is_internal_anchor":true},{"doi":"","year":2022,"title":"Luiz Bonifacio, Hugo Abonizio, Marzieh Fadaee, and Rodrigo Nogueira. 2022. InPars : Unsupervised dataset generation for information retrieval. In Proceedings of the 45th International ACM SIGIR Confer","work_id":"f57f0193-e00a-485d-9e96-2256c1f2509e","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Leonid Boytsov, Preksha Patel, Vivek Sourabh, Riddhi Nisar, Sayani Kundu, Ramya Ramanathan, and Eric Nyberg. 2023. InPars-Light : Cost-effective unsupervised training of efficient rankers. arXiv:2301.","work_id":"c983ce83-cafd-4482-a355-da0aa94208c0","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2010,"title":"Barla Cambazoglu, Hugo Zaragoza, Olivier Chapelle, Jiang Chen, Ciya Liao, Zhaohui Zheng, and Jon Degenhardt","work_id":"ed02f770-7aac-4f93-83a0-1b54059ea148","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Nick Craswell, Bhaskar Mitra, Emine Yilmaz, and Daniel Campos. 2020. Overview of the TREC 2020 deep learning track. In Proceedings of the Twenty-Ninth Text REtrieval Conference Proceedings (TREC 2020)","work_id":"d1e2aca6-6446-49d3-8c67-f02af05fb24d","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":42,"snapshot_sha256":"4bc1177d4cbd6c3944f4c5932af6026b65d9b618c93f50832041417b79a2984a","internal_anchors":5},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}