mlsanityData to 7 Oct 2026

CALM: Unleashing the Cross-Lingual Self-Aligning Ability of Language Model Question Answering

arXiv 2501.18457 ↗First posted 30 Jan 2025cs.CLcited by 10

On the map: LLM alignment / RLHF / preference optimisation → LLM preference optimization methods

Abstract (arXiv)

Large Language Models (LLMs) are pretrained on extensive multilingual corpora to acquire both language-specific cultural knowledge and general knowledge. Ideally, while LLMs should provide consistent responses to culture-independent questions across languages, we observe significant performance disparities. To address this, we explore the Cross-Lingual Self-Aligning ability of Language Models (CALM) to align knowledge across languages. Specifically, for a given question, we sample multiple responses across different languages and select the most self-consistent response as the target, leaving the remaining responses as negative examples. We then employ direct preference optimization (DPO) to align the model's knowledge across different languages. Evaluations on the MEDQA and X-CSQA datasets demonstrate CALM's effectiveness in enhancing cross-lingual knowledge question answering, both in zero-shot and retrieval-augmented settings. We also found that increasing the number of languages involved in CALM training leads to higher accuracy and consistency. We offer a qualitative analysis of how cross-lingual consistency can enhance knowledge alignment and explore the method's generalizability.

Key facts. Cited by 10 papers in the map (data through 7 Oct 2026). Placed in LLM preference optimization methods, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 22 of its 29 references that are in the map. First posted on arXiv 30 Jan 2025.

Builds on

22 of 29 references.

Closest ancestors · 1

Same direction, newest first

  1. 2024MAPO: Advancing Multilingual Reasoning through Multilingual Alignment-as-Preference Optimization12 Jan

Other references · 21

  1. 2024The Llama 3 Herd of Modelscited by 14,192
  2. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  4. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  5. 2022Large Language Models are Zero-Shot Reasonerscited by 4,323
  6. 2023Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflectioncited by 1,800
  7. 2024LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Modelscited by 1,442
  8. 2020What Disease does this Patient Have? A Large-scale Open Domain Question Answering Dataset from Medical Examscited by 1,324
  9. 2023Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMscited by 946
  10. 2022Large Language Models Can Self-Improvecited by 656
  11. 2023Cross-Lingual Consistency of Factual Knowledge in Multilingual Language Modelscited by 117
  12. 2023SeaEval for Multilingual Foundation Models: From Cross-Lingual Alignment to Cultural Reasoningcited by 94
9 more
  1. 2024A Survey on Multilingual Large Language Models: Corpora, Alignment, and Biascited by 82
  2. 2021Common Sense Beyond English: Evaluating and Improving Multilingual Language Models for Commonsense Reasoningcited by 64
  3. 2024Aligning LLMs with Individual Preferences via Interactioncited by 63
  4. 2023Breaking Language Barriers in Multilingual Mathematical Reasoning: Insights and Observationscited by 58
  5. 2023Do LVLMs Understand Charts? Analyzing and Correcting Factual Errors in Chart Captioningcited by 47
  6. 2024Multilingual Pretraining and Instruction Tuning Improve Cross-Lingual Knowledge Alignment, But Only Shallowlycited by 27
  7. 2024CrossIn: An Efficient Instruction Tuning Approach for Cross-Lingual Knowledge Alignmentcited by 10
  8. 20241+1>2: Can Large Language Models Serve as Cross-Lingual Knowledge Aggregators?cited by 9
  9. 2024PropaInsight: Toward Deeper Understanding of Propaganda in Terms of Techniques, Appeals, and Intentcited by 9

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 6; 2026: 4.

10 papers.

  1. 2025MMBoundary: Advancing MLLM Knowledge Boundary Awareness through Reasoning Step Confidence Calibrationcited by 11
  2. 2025Reasoning Path Divergence: A New Metric and Curation Strategy to Unlock LLM Diverse Thinkingcited by 10
  3. 2025Diversity-Enhanced Reasoning for Subjective Questionscited by 9
  4. 2025Unveiling the Lack of LVLM Robustness to Fundamental Visual Variations: Why and Path Forwardcited by 7
  5. 2025Traveling Across Languages: Benchmarking Cross-Lingual Consistency in Multimodal LLMscited by 3
  6. 2026Post-Training Language Models for Crosslingual Consistencycited by 3
  7. 2026When Languages Disagree: Self-Evolving Multilingual LLM Judgescited by 2
  8. 2026Structuring the Space of Sociotechnical Alignmentcited by 2
  9. 2025Unlocking LLM Safeguards for Low-Resource Languages via Reasoning and Alignment with Minimal Training Datacited by 1
  10. 2026A Systematic Evaluation of Cross-Lingual Consistency Enhancement Methods in Multilingual Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2021Investigating Post-pretraining Representation Alignment for Cross-Lingual Question Answering
  2. 2026Cross-Lingual Consensus: Aligning Multilingual Cultural Knowledge via Multilingual Self-Consistency
  3. 2024The Power of Question Translation Training in Multilingual Reasoning: Broadened Scope and Deepened Insights
  4. 2026MSQA: A Natively Sourced Multilingual and Multicultural SimpleQA Benchmark
  5. 2023On the Calibration of Multilingual Question Answering LLMs
  6. 2024Don't Just Say "I don't know"! Self-aligning Large Language Models for Responding to Unknown Questions with Explanations
  7. 2025Evaluating Large Language Model with Knowledge Oriented Language Specific Simple Question Answering
  8. 2024Self-Pluralising Culture Alignment for Large Language Models