arXiv cs.CL· Rui Qi, Yunlong Liang, Yufeng Chen, Jinan Xu, Fandong Meng, Kaiyu Huang·· 11 小时前AI 评分31
MARS：训练无需检索器的多语言搜索智能体
MARS: Training Multilingual Search Agents for Retriever-Free Evidence Acquisition
AI 导读
MARS 是一个通过强化学习训练的多语言搜索智能体，采用直接语料交互（DCI）范式，让模型用可执行 shell 命令直接操作原始语料，无需学习型检索器。它引入语料原生实体锚定来恢复可靠的词汇锚点，并用冗余感知搜索奖励鼓励简洁有效的搜索轨迹。实验显示 MARS 持续优于强检索基线和 DCI 基线，具备稳健的跨语言泛化能力并大幅减少冗余语料交互。
正文 · 原文
Abstract: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.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2611.09090 [cs.CL] |
| (or arXiv:2611.09090v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.09090 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Rui Qi [view email]
[v1]
Fri, 6 Nov 2026 07:20:03 UTC (1,965 KB)
来源：arXiv cs.CL · arxiv.org