Metraipolis › archive › dc_4ca0ea4adecd

It Takes Two to Match: Co-Evolving Generative Retriever with Reinforcement Learning — arXiv:2609.00638

Source

arXiv (en)

Original

https://arxiv.org/abs/2609.00638

License (as archived)

CC0-1.0

Archived

2026-10-17T06:41:44.803Z

SHA-256 of the archived text

64fbcc79a810c5c3001b4e511805fb6c7eba36621eed5077721cb2d7354a2051

Stable URL of this page

https://metraipolis.com/documento/dc_4ca0ea4adecd

How to cite

It Takes Two to Match: Co-Evolving Generative Retriever with Reinforcement Learning — arXiv:2609.00638 — arXiv (en), https://arxiv.org/abs/2609.00638 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_4ca0ea4adecd (Metraipolis archive dc_4ca0ea4adecd, sha256 64fbcc79a810c5c3001b4e511805fb6c7eba36621eed5077721cb2d7354a2051)

@misc{dc4ca0ea4adecd,
  title = {It Takes Two to Match: Co-Evolving Generative Retriever with Reinforcement Learning — arXiv:2609.00638},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.00638},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_4ca0ea4adecd; sha256 64fbcc79a810c5c3001b4e511805fb6c7eba36621eed5077721cb2d7354a2051},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.00638. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Retrieval is the first stage of modern search and advertising systems, selecting a candidate set from a large item universe for downstream ranking and auction. Recent work increasingly leverages LLMs to improve retrieval through query expansion, data synthesis, and retrieval-feedback training. However, the generative component is typically used for query-side augmentation, while final matching is still delegated to a downstream retriever. We introduce CoGR, a retrieval framework that instead trains LLMs to directly construct retrieval representations on both query and item sides. Each generator produces a compact set of keywords, which are matched directly through an inverted index, preserving compatibility with existing keyword-based retrieval infrastructure. CoGR uses a two-stage training pipeline. Supervised fine-tuning first establishes an aligned keyword space, after which co-evolving reinforcement learning alternately optimizes the query- and item-side generators with GRPO against the opposite side's frozen index. Both sides optimize the same query-to-item retrieval $F_1$ objective: the query side receives retrieval $F_1$ directly, while the item side receives a counterfactual marginal reward measuring the change in query-side $F_1$ caused by its generated keywords. Across 10 representative sparse, dense, and generative baselines, CoGR achieves the best performance on both an internal APP Marketplace dataset and the public WANDS benchmark, improving $F_1$ over the strongest baseline by $10.9\%$ and $36.1\%$, respectively. Further analysis shows stable co-evolution and increasingly aligned query--item keyword spaces over training.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-02 in cs.CL).
Categorie arXiv: Information Retrieval (cs.IR) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.00638 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-17T06:33:38.887Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).