mlsanityData to 7 Oct 2026

Language-Specific Neurons: The Key to Multilingual Capabilities in Large Language Models

arXiv 2402.16438 ↗First posted 26 Feb 2024cs.CLcited by 124

Not on the map. Why?

Abstract (arXiv)

Large language models (LLMs) demonstrate remarkable multilingual capabilities without being pre-trained on specially curated multilingual parallel corpora. It remains a challenging problem to explain the underlying mechanisms by which LLMs process multilingual texts. In this paper, we delve into the composition of Transformer architectures in LLMs to pinpoint language-specific regions. Specially, we propose a novel detection method, language activation probability entropy (LAPE), to identify language-specific neurons within LLMs. Based on LAPE, we conduct comprehensive experiments on several representative LLMs, such as LLaMA-2, BLOOM, and Mistral. Our findings indicate that LLMs' proficiency in processing a particular language is predominantly due to a small subset of neurons, primarily situated in the models' top and bottom layers. Furthermore, we showcase the feasibility to "steer" the output language of LLMs by selectively activating or deactivating language-specific neurons. Our research provides important evidence to the understanding and exploration of the multilingual capabilities of LLMs.

Key facts. Cited by 124 papers in the map (data through 7 Oct 2026). Builds on 33 of its 49 references that are in the map. First posted on arXiv 26 Feb 2024.

Builds on

33 of 49 references.

References · 33

  1. 2017Attention Is All You Needcited by 59,932
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  3. 2023GPT-4 Technical Reportcited by 18,132
  4. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  5. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,393
  6. 2019Unsupervised Cross-lingual Representation Learning at Scalecited by 3,907
  7. 2022OPT: Open Pre-trained Transformer Language Modelscited by 3,464
  8. 2016Gaussian Error Linear Units (GELUs)cited by 3,431
  9. 2022BLOOM: A 176B-Parameter Open-Access Multilingual Language Modelcited by 1,968
  10. 2020mT5: A massively multilingual pre-trained text-to-text transformercited by 1,813
  11. 2020GLU Variants Improve Transformercited by 1,724
  12. 2019How multilingual is Multilingual BERT?cited by 642
21 more
  1. 2023Finding Neurons in a Haystack: Case Studies with Sparse Probingcited by 328
  2. 2020Understanding the Role of Individual Units in a Deep Neural Networkcited by 254
  3. 2019Cross-Lingual Ability of Multilingual BERT: An Empirical Studycited by 182
  4. 2019Emerging Cross-lingual Structure in Pretrained Language Modelscited by 180
  5. 2020Compositional Explanations of Neuronscited by 172
  6. 2017Learning to Generate Reviews and Discovering Sentimentcited by 147
  7. 2018What Is One Grain of Sand in the Desert? Analyzing Individual Neurons in Deep NLP Modelscited by 124
  8. 2020Multilingual Alignment of Contextual Word Representationscited by 113
  9. 2018Identifying and Controlling Important Neurons in Neural Machine Translationcited by 110
  10. 2023SeaLLMs -- Large Language Models for Southeast Asiacited by 104
  11. 2024Universal Neurons in GPT2 Language Modelscited by 100
  12. 2023Neurons in Large Language Models: Dead, N-gram, Positionalcited by 88
  13. 2021Neuron-level Interpretation of Deep NLP Models: A Surveycited by 81
  14. 2019Improved Zero-shot Neural Machine Translation via Ignoring Spurious Correlationscited by 61
  15. 2023Journey to the Center of the Knowledge Neurons: Discoveries of Language-Independent Knowledge Neurons and Degenerate Knowledge Neuronscited by 56
  16. 2023Towards a Common Understanding of Contributing Factors for Cross-Lingual Transfer in Multilingual Language Models: A Reviewcited by 50
  17. 2023Mitigating Hallucinations and Off-target Machine Translation with Source-Contrastive and Language-Contrastive Decodingcited by 42
  18. 2021When is BERT Multilingual? Isolating Crucial Ingredients for Cross-lingual Transfercited by 41
  19. 2020Cross-lingual Alignment Methods for Multilingual BERT: A Comparative Studycited by 24
  20. 2022Multi Task Learning For Zero Shot Performance Prediction of Multilingual Modelscited by 20
  21. 2023Language Representation Projection: Can We Transfer Factual Knowledge across Languages in Multilingual Language Models?cited by 19

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 39; 2025: 45; 2026: 40.

124 papers.

Sort:

By area:

  1. 2024How do Large Language Models Handle Multilingualism?cited by 148
  2. 2024Llama Scope: Extracting Millions of Features from Llama-3.1-8B with Sparse Autoencoderscited by 133
  3. 2024A Primer on the Inner Workings of Transformer-based Language Modelscited by 83
  4. 2024The Semantic Hub Hypothesis: Language Models Share Semantic Representations Across Languages and Modalitiescited by 69
  5. 2024Explainable and Interpretable Multimodal Large Language Models: A Comprehensive Surveycited by 52
  6. 2024Knowledge Mechanisms in Large Language Models: A Survey and Perspectivecited by 50
  7. 2024Separating Tongue from Thought: Activation Patching Reveals Language-Agnostic Concept Representations in Transformerscited by 42
  8. 2025SciReplicate-Bench: Benchmarking LLMs in Agent-driven Algorithmic Reproduction from Research Paperscited by 42
  9. 2024Talking Heads: Understanding Inter-layer Communication in Transformer Language Modelscited by 39
  10. 2025A Survey on Mechanistic Interpretability for Multi-Modal Foundation Modelscited by 34
  11. 2024Sharing Matters: Analysing Neurons Across Languages and Tasks in LLMscited by 33
  12. 2024The Same But Different: Structural Similarities and Differences in Multilingual Language Modelingcited by 27
38 more
  1. 2024MMNeuron: Discovering Neuron-Level Domain-Specific Interpretation in Multimodal Large Language Modelcited by 26
  2. 2024Opportunities and Challenges of Large Language Models for Low-Resource Languages in Humanities Researchcited by 26
  3. 2025Multilingual Routing in Mixture-of-Expertscited by 25
  4. 2024Style-Specific Neurons for Steering LLMs in Text Style Transfercited by 24
  5. 2025RepreGuard: Detecting LLM-Generated Text by Revealing Hidden Representation Patternscited by 24
  6. 2024Towards Understanding Multi-Task Learning (Generalization) of LLMs via Detecting and Exploring Task-Specific Neuronscited by 23
  7. 2024How Can We Effectively Expand the Vocabulary of LLMs with 0.01GB of Target Language Text?cited by 22
  8. 2024Configurable Foundation Models: Building LLMs from a Modular Perspectivecited by 22
  9. 2026Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Modelscited by 22
  10. 2024Beneath the Surface of Consistency: Exploring Cross-lingual Knowledge Representation Sharing in LLMscited by 20
  11. 2024Layer Swapping for Zero-Shot Cross-Lingual Transfer in Large Language Modelscited by 20
  12. 2025Language-Specific Latent Process Hinders Cross-Lingual Performancecited by 20
  13. 2025Paths Not Taken: Understanding and Mending the Multilingual Factual Recall Pipelinecited by 20
  14. 2025Localized Cultural Knowledge is Conserved and Controllable in Large Language Modelscited by 19
  15. 2024Multilingual Knowledge Editing with Language-Agnostic Factual Neuronscited by 18
  16. 2024Mitigating the Language Mismatch and Repetition Issues in LLM-based Machine Translation via Model Editingcited by 18
  17. 2024Multilingual Large Language Models: A Systematic Surveycited by 18
  18. 2025AudioLens: A Closer Look at Auditory Attribute Perception of Large Audio-Language Modelscited by 18
  19. 2024Efficiently Democratizing Medical LLMs for 50 Languages via a Mixture of Language Family Expertscited by 17
  20. 2025SLAM: Towards Efficient Multilingual Reasoning via Selective Language Alignmentcited by 17
  21. 2025Language Arithmetics: Towards Systematic Language Neuron Identification and Manipulationcited by 17
  22. 2025POWSM: A Phonetic Open Whisper-Style Speech Foundation Modelcited by 14
  23. 2024Knowledge Localization: Mission Not Accomplished? Enter Query Localization!cited by 13
  24. 2025Understanding How Value Neurons Shape the Generation of Specified Values in LLMscited by 13
  25. 2024Mitigating Copy Bias in In-Context Learning through Neuron Pruningcited by 11
  26. 2026CLaS-Bench: A Cross-Lingual Alignment and Steering Benchmarkcited by 11
  27. 2024Math Neurosurgery: Isolating Language Models' Math Reasoning Abilities Using Only Forward Passescited by 10
  28. 2026DNA: Uncovering Universal Latent Forgery Knowledgecited by 10
  29. 2024FuxiTranyu: A Multilingual Large Language Model Trained with Balanced Datacited by 9
  30. 2024Neuron-based Personality Trait Induction in Large Language Modelscited by 9
  31. 2026Align Once, Benefit Multilingually: Enforcing Multilingual Consistency for LLM Safety Alignmentcited by 9
  32. 2024Unveiling Language Competence Neurons: A Psycholinguistic Approach to Model Interpretabilitycited by 8
  33. 2025How Does Alignment Enhance LLMs' Multilingual Capabilities? A Language Neurons Perspectivecited by 8
  34. 2025Where Culture Fades: Revealing the Cultural Gap in Text-to-Image Generationcited by 8
  35. 2024MINER: Mining the Underlying Pattern of Modality-Specific Neurons in Multimodal Large Language Modelscited by 7
  36. 2025How Syntax Specialization Emerges in Language Modelscited by 7
  37. 2025The Achilles' Heel of LLMs: How Altering a Handful of Neurons Can Cripple Language Abilitiescited by 7
  38. 2025CLINIC: Evaluating Multilingual Trustworthiness in Language Models for Healthcarecited by 7

50 most cited of 124 All 124 (JSON)