paper-with-me

홈 › Papers

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

2024-02-26 · Tianyi Tang, Wenyang Luo, Haoyang Huang, Dongdong Zhang, Xiaolei Wang, Xin Zhao, Furu Wei, Ji-Rong Wen

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.

📄 PDF Abstract BibTeX arXiv:2402.16438

Code (1)

rucaibox/language-specific-neurons 공식 구현 pytorch

Methods 이 논문이 사용한 방법론

Attention 설명 없음
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
BPE Byte Pair Encoding, or BPE, is a subword segmentation algorithm that encodes rare and unknown words as sequences of subword units. The intuition is that various word…
Dropout Dropout is a regularization technique for neural networks that drops a unit (along with connections) at training time with a specified probability $p$ (a common value is…
Dense Connections Dense Connections, or Fully Connected Connections, are a type of layer in a deep neural network that use a linear operation where every input is connected to every output…
Label Smoothing Label Smoothing is a regularization technique that introduces noise for the labels. This accounts for the fact that datasets may have mistakes in them, so maximizing the…
Adam 설명 없음
Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…

Similar Papers 제목 키워드 기반

How do Large Language Models Handle Multilingualism?

2024-02-29 · Yiran Zhao, Wenxuan Zhang, Guizhen Chen, Kenji Kawaguchi 외

Large language models (LLMs) have demonstrated impressive capabilities across diverse languages. This study explores how LLMs handle multilingualism. Based on observed language ratio shifts among layers and the relations…

On the Multilingual Ability of Decoder-based Pre-trained Language Models: Finding and Controlling Language-Specific Neurons

2024-04-03 · Takeshi Kojima, Itsuki Okimura, Yusuke Iwasawa, Hitomi Yanaka 외

Current decoder-based pre-trained language models (PLMs) successfully demonstrate multilingual capabilities. However, it is unclear how these models handle multilingualism. We analyze the neuron-level internal behavior o…

DecoderText Generation

Targeted Interpretable Safety Neuron Enhancement for Multilingual Vision-Language Large Models

2026-04-10 · Enyi Shi, Fei Shen, Shuyi Miao, Linxia Zhu 외 arxiv

With the widespread deployment of vision-language large models (VLLMs), their safety alignment faces dual challenges across languages and modalities. Existing methods model multilingual and multimodal safety separately, …

CRANE: Causal Relevance Analysis of Language-Specific Neurons in Multilingual Large Language Models

2026-01-08 · Yifan Le, Yunliang Li arxiv

Multilingual large language models (LLMs) achieve strong performance across languages, yet how language capabilities are organized at the neuron level remains poorly understood. Prior work has identified language-related…

Isolating Culture Neurons in Multilingual Large Language Models

2025-08-04 · Danial Namazifard, Lukas Galke Poech arxiv

Language and culture are deeply intertwined, yet it has been unclear how and where multilingual large language models encode culture. Here, we build on an established methodology for identifying language-specific neurons…