aiqbench analysis
This research introduces LcRL, a reinforcement learning framework designed to mitigate knowledge bias and conflict in multilingual retrieval-augmented generation. By employing language-coupled Group Relative Policy Optimization and an anti-consistency penalty, the method improves how models integrate external information across diverse linguistic contexts without relying on a single-turn optimization strategy. The approach demonstrates robust performance even under constrained training data conditions, offering practical solutions for handling large-scale multilingual collections.
Language-Coupled Reinforcement Learning for Multilingual Retrieval-Augmented Generation
arXiv:2601.14896v3 Announce Type: replace Abstract: Multilingual retrieval-augmented generation (MRAG) requires models to effectively acquire and integrate beneficial external knowledge from multilingual collections. However, most existing studies employ a unitive process where queries of equivalent semantics across different languages are processed through a single-turn retrieval and subsequent optimization. Such a ``one-size-fits-all'' strategy is often suboptimal in multilingual settings, as the models occur to knowledge bias and conflict during the interaction with the search engine. To alleviate the issues, we propose LcRL, a multilingual search-augmented reinforcement learning framework that integrates a language-coupled Group Relative Policy Optimization into the policy and reward models. We adopt the language-coupled group sampling in the rollout module to reduce knowledge bias, and regularize an auxiliary anti-consistency penalty in the reward models to mitigate the knowledge conflict. Experimental results demonstrate that LcRL not only achieves competitive performance but is also appropriate for various practical scenarios such as constrained training data and retrieval over collections encompassing a large number of languages. Our code is available at https://github.com/Cherry-qwq/LcRL-Open.
Abstract:Multilingual retrieval-augmented generation (MRAG) requires models to effectively acquire and integrate beneficial external knowledge from multilingual collections. However, most existing studies employ a unitive process where queries of equivalent semantics across different languages are processed through a single-turn retrieval and subsequent optimization. Such a ``one-size-fits-all'' strategy is often suboptimal in multilingual settings, as the models occur to knowledge bias and conflict during the interaction with the search engine. To alleviate the issues, we propose LcRL, a multilingual search-augmented reinforcement learning framework that integrates a language-coupled Group Relative Policy Optimization into the policy and reward models. We adopt the language-coupled group sampling in the rollout module to reduce knowledge bias, and regularize an auxiliary anti-consistency penalty in the reward models to mitigate the knowledge conflict. Experimental results demonstrate that LcRL not only achieves competitive performance but is also appropriate for various practical scenarios such as constrained training data and retrieval over collections encompassing a large number of languages. Our code is available at this https URL.
Computation and Language (cs.CL)
arXiv:2601.14896 [cs.CL]
(or arXiv:2601.14896v3 [cs.CL] for this version)
Full text under the publisher's open licence · 177 words
Related coverage
Introducing Instella-MoE: A State-of-the-Art Fully Open Mixture-of-Experts Language Model
AMD · Jul 24, 2026
Granite 4.1 LLMs: How They’re Built
Hugging Face · Apr 29, 2026
VISOR: Agentic Visual Retrieval-Augmented Generation via Iterative Search and Over-horizon Reasoning
arXiv · Aug 17, 2026
VerTox: Verifiable Reward-Guided Corpus Poisoning Against Neural Ranking Models
arXiv · Sep 2, 2026