Metraipolis › archive › dc_dfb3347e5537

MARS: Training Multilingual Search Agents for Retriever-Free Evidence Acquisition — arXiv:2611.09090

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.09090
License (as archived)
CC0-1.0
Archived
2026-11-09T22:35:55.823Z
SHA-256 of the archived text
b517ab63fa03af4749a941ad9b84c4437f1b81e000c4e8b05a341b8d6d12e51b
Stable URL of this page
https://metraipolis.com/documento/dc_dfb3347e5537

How to cite

MARS: Training Multilingual Search Agents for Retriever-Free Evidence Acquisition — arXiv:2611.09090 — arXiv (en), https://arxiv.org/abs/2611.09090 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dfb3347e5537 (Metraipolis archive dc_dfb3347e5537, sha256 b517ab63fa03af4749a941ad9b84c4437f1b81e000c4e8b05a341b8d6d12e51b)
BibTeX
@misc{dcdfb3347e5537,
  title = {MARS: Training Multilingual Search Agents for Retriever-Free Evidence Acquisition — arXiv:2611.09090},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.09090},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dfb3347e5537; sha256 b517ab63fa03af4749a941ad9b84c4437f1b81e000c4e8b05a341b8d6d12e51b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.09090. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Existing multilingual retrieval-augmented generation relies heavily on learned retrievers and similarity-based ranking from multilingual collections, which are vulnerable to cross-lingual retrieval errors and indexing overhead. In this paper, we explore retriever-free multilingual search, where the model directly interacts with raw corpora through executable shell commands. This paradigm shifts the bottleneck to multilingual search action generation, where exact search is sensitive to entity surface forms and outcome-only supervision provides limited guidance for multi-turn search. To address these challenges, we propose MARS, a multilingual search agent in direct corpus interaction (DCI) with reinforcement learning (RL). MARS introduces corpus-native entity grounding to recover reliable lexical anchors from corpus evidence, mitigating failure to retrieve evidence. Meanwhile, we present a redundancy-aware search reward to encourage concise and effective search trajectories, addressing training collapse in RL of multilingual scenarios. Experiments show that \ours~consistently outperforms strong retrieval-based and DCI baselines, while demonstrating robust cross-lingual generalization and substantially reducing redundant corpus interactions.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2611.09090 — 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 2026-11-09T22:31:01.945Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).