Copertina

Do LLMs Need to Think in One Language? Correlation between Latent Language and Task Performance

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Ozaki, Shintaro, Hiraoka, Tatsuya, Otake, Hiroto, Ouchi, Hiroki, Isonuma, Masaru, Heinzerling, Benjamin, Inui, Kentaro, Watanabe, Taro, Miyao, Yusuke, Oseki, Yohei, Takagi, Yu
Natura:	Preprint
Pubblicazione:	2025
Soggetti:	Computation and Language
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

How a Bilingual LM Becomes Bilingual: Tracing Internal Representations with Sparse Autoencoders
di: Inaba, Tatsuro, et al.
Pubblicazione: (2025)

Exclusive Unlearning
di: Sasaki, Mutsumi, et al.
Pubblicazione: (2026)

Repetition Neurons: How Do Language Models Produce Repetitions?
di: Hiraoka, Tatsuya, et al.
Pubblicazione: (2024)

Monotonic Representation of Numeric Properties in Language Models
di: Heinzerling, Benjamin, et al.
Pubblicazione: (2024)

Spelling-out is not Straightforward: LLMs' Capability of Tokenization from Token to Characters
di: Hiraoka, Tatsuya, et al.
Pubblicazione: (2025)

The Geometry of Numerical Reasoning: Language Models Compare Numeric Properties in Linear Subspaces
di: El-Shangiti, Ahmed Oumar, et al.
Pubblicazione: (2024)

Massive Supervised Fine-tuning Experiments Reveal How Data, Layer, and Training Factors Shape LLM Alignment Quality
di: Harada, Yuto, et al.
Pubblicazione: (2025)

Can Language Models Induce Grammatical Knowledge from Indirect Evidence?
di: Oba, Miyu, et al.
Pubblicazione: (2024)

Representational Analysis of Binding in Language Models
di: Dai, Qin, et al.
Pubblicazione: (2024)

Cell-Based Representation of Relational Binding in Language Models
di: Dai, Qin, et al.
Pubblicazione: (2026)

Large Language Models Are Human-Like Internally
di: Kuribayashi, Tatsuki, et al.
Pubblicazione: (2025)

TopK Language Models
di: Takahashi, Ryosuke, et al.
Pubblicazione: (2025)

Unlearning Traces the Influential Training Data of Language Models
di: Isonuma, Masaru, et al.
Pubblicazione: (2024)

Understanding and Controlling Repetition Neurons and Induction Heads in In-Context Learning
di: Doan, Nhi Hoai, et al.
Pubblicazione: (2025)

RECALL: Library-Like Behavior In Language Models is Enhanced by Self-Referencing Causal Cycles
di: Nwadike, Munachiso, et al.
Pubblicazione: (2025)

Understanding Fact Recall in Language Models: Why Two-Stage Training Encourages Memorization but Mixed Training Teaches Knowledge
di: Zhang, Ying, et al.
Pubblicazione: (2025)

Diagnosing Vision Language Models' Perception by Leveraging Human Methods for Color Vision Deficiencies
di: Hayashi, Kazuki, et al.
Pubblicazione: (2025)

On Entity Identification in Language Models
di: Sakata, Masaki, et al.
Pubblicazione: (2025)

BQA: Body Language Question Answering Dataset for Video Large Language Models
di: Ozaki, Shintaro, et al.
Pubblicazione: (2024)

Linear Representations of Hierarchical Concepts in Language Models
di: Sakata, Masaki, et al.
Pubblicazione: (2026)

Instability in Downstream Task Performance During LLM Pretraining
di: Nishida, Yuto, et al.
Pubblicazione: (2025)

Comprehensive Evaluation of Large Language Models for Topic Modeling
di: Doi, Tomoki, et al.
Pubblicazione: (2024)

Investigating Training and Generalization in Faithful Self-Explanations of Large Language Models
di: Doi, Tomoki, et al.
Pubblicazione: (2025)

Towards Cross-Lingual Explanation of Artwork in Large-scale Vision Language Models
di: Ozaki, Shintaro, et al.
Pubblicazione: (2024)

The Curse of Popularity: Popular Entities have Catastrophic Side Effects when Deleting Knowledge from Language Models
di: Takahashi, Ryosuke, et al.
Pubblicazione: (2024)

Hidden Failures in Robustness: Why Supervised Uncertainty Quantification Needs Better Evaluation
di: Stacey, Joe, et al.
Pubblicazione: (2026)

What's New in My Data? Novelty Exploration via Contrastive Generation
di: Isonuma, Masaru, et al.
Pubblicazione: (2024)

Psychometric Predictive Power of Large Language Models
di: Kuribayashi, Tatsuki, et al.
Pubblicazione: (2023)

Is Structure Dependence Shaped for Efficient Communication?: A Case Study on Coordination
di: Kajikawa, Kohei, et al.
Pubblicazione: (2024)

Weight-based Analysis of Detokenization in Language Models: Understanding the First Stage of Inference Without Inference
di: Kamoda, Go, et al.
Pubblicazione: (2025)

Language Acquisition Device in Large Language Models
di: Mita, Masato, et al.
Pubblicazione: (2026)

UniDetox: Universal Detoxification of Large Language Models via Dataset Distillation
di: Lu, Huimin, et al.
Pubblicazione: (2025)

Developmentally-plausible Working Memory Shapes a Critical Period for Language Acquisition
di: Mita, Masato, et al.
Pubblicazione: (2025)

Tree-Planted Transformers: Unidirectional Transformer Language Models with Implicit Syntactic Supervision
di: Yoshida, Ryo, et al.
Pubblicazione: (2024)

The Imperfective Paradox in Large Language Models
di: Ma, Bolei, et al.
Pubblicazione: (2026)

AdTEC: A Unified Benchmark for Evaluating Text Quality in Search Engine Advertising
di: Zhang, Peinan, et al.
Pubblicazione: (2024)

Sycophancy Hides Linearly in the Attention Heads
di: Genadi, Rifo, et al.
Pubblicazione: (2026)

Can Language Models Handle a Non-Gregorian Calendar? The Case of the Japanese wareki
di: Sasaki, Mutsumi, et al.
Pubblicazione: (2025)

Constructing Multilingual Visual-Text Datasets Revealing Visual Multilingual Ability of Vision Language Models
di: Atuhurra, Jesse, et al.
Pubblicazione: (2024)

Knowledge of Pretrained Language Models on Surface Information of Tokens
di: Hiraoka, Tatsuya, et al.
Pubblicazione: (2024)