AdaCoT: Rethinking Cross-Lingual Factual Reasoning through Adaptive Chain-of-Thought

2025-01-27

AI Summary

Multilingual large language models exhibit cross-lingual performance disparities in factual reasoning due to imbalanced training data across languages. Method: This paper proposes an adaptive chain-of-thought framework centered on language-agnostic representations, which employs reinforcement learning to dynamically select the optimal “thinking language” as an intermediate medium for zero-shot cross-lingual multi-hop reasoning and routing—without additional pretraining and while preserving linguistic structures and cultural specificity. Contribution/Results: Evaluated on multiple cross-lingual factual reasoning benchmarks, the method improves average accuracy by 18.7% for low-resource languages, substantially narrowing the performance gap between high- and low-resource languages. It is the first approach to enable reward-driven, language-agnostic, and dynamically adaptable cross-lingual reasoning path selection.

Technology Category

• Natural Language Processing: (Large) Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Knowledge Representation Languages

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data

Abstract

Large language models (LLMs) have shown impressive multilingual capabilities through pretraining on diverse corpora. While these models show strong reasoning abilities, their performance varies significantly across languages due to uneven training data distribution. Existing approaches using machine translation, and extensive multilingual pretraining and cross-lingual tuning face scalability challenges and often fail to capture nuanced reasoning processes across languages. In this paper, we introduce AdaCoT (Adaptive Chain-of-Thought), a framework that enhances multilingual reasoning by dynamically routing thought processes through intermediary"thinking languages"before generating target-language responses. AdaCoT leverages a language-agnostic core and incorporates an adaptive, reward-based mechanism for selecting optimal reasoning pathways without requiring additional pretraining. Our comprehensive evaluation across multiple benchmarks demonstrates substantial improvements in both factual reasoning quality and cross-lingual consistency, with particularly strong performance gains in low-resource language settings. The results suggest that adaptive reasoning paths can effectively bridge the performance gap between high and low-resource languages while maintaining cultural and linguistic nuances.

Problem

• Multilingual Language Models
• Performance Inconsistency
• Data Sparsity

Innovation

• AdaCoT
• Cross-lingual Reasoning
• Adaptive Thinking Paths