{"ID":2879125,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2508.16998","arxiv_id":"2508.16998","title":"DeAR: Dual-Stage Document Reranking with Reasoning Agents via LLM Distillation","abstract":"Large Language Models (LLMs) have transformed listwise document reranking by enabling global reasoning over candidate sets, yet single models often struggle to balance fine-grained relevance scoring with holistic cross-document analysis. We propose \\textbf{De}ep\\textbf{A}gent\\textbf{R}ank (\\textbf{\\DeAR}), an open-source framework that decouples these tasks through a dual-stage approach, achieving superior accuracy and interpretability. In \\emph{Stage 1}, we distill token-level relevance signals from a frozen 13B LLaMA teacher into a compact \\{3, 8\\}B student model using a hybrid of cross-entropy, RankNet, and KL divergence losses, ensuring robust pointwise scoring. In \\emph{Stage 2}, we attach a second LoRA adapter and fine-tune on 20K GPT-4o-generated chain-of-thought permutations, enabling listwise reasoning with natural-language justifications. Evaluated on TREC-DL19/20, eight BEIR datasets, and NovelEval-2306, \\DeAR surpasses open-source baselines by +5.1 nDCG@5 on DL20 and achieves 90.97 nDCG@10 on NovelEval, outperforming GPT-4 by +3.09. Without fine-tuning on Wikipedia, DeAR also excels in open-domain QA, achieving 54.29 Top-1 accuracy on Natural Questions, surpassing baselines like MonoT5, UPR, and RankGPT. Ablations confirm that dual-loss distillation ensures stable calibration, making \\DeAR a highly effective and interpretable solution for modern reranking systems.\\footnote{Dataset and code available at https://github.com/DataScienceUIBK/DeAR-Reranking.}.","short_abstract":"Large Language Models (LLMs) have transformed listwise document reranking by enabling global reasoning over candidate sets, yet single models often struggle to balance fine-grained relevance scoring with holistic cross-document analysis. We propose \\textbf{De}ep\\textbf{A}gent\\textbf{R}ank (\\textbf{\\DeAR}), an open-sour...","url_abs":"https://arxiv.org/abs/2508.16998","url_pdf":"https://arxiv.org/pdf/2508.16998v1","authors":"[\"Abdelrahman Abdallah\",\"Jamshid Mozafari\",\"Bhawna Piryani\",\"Adam Jatowt\"]","published":"2025-08-23T11:46:08Z","proceeding":"cs.CL","tasks":"[\"cs.CL\",\"cs.IR\"]","methods":"[\"Large Language Model\",\"Language Model\",\"LoRA\"]","has_code":false,"code_links":[{"ID":610555,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2879125,"paper_url":"https://arxiv.org/abs/2508.16998","paper_title":"DeAR: Dual-Stage Document Reranking with Reasoning Agents via LLM Distillation","repo_url":"https://github.com/DataScienceUIBK/DeAR-Reranking","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}