Imagen de Portada

Anatomy of an Idiom: Tracing Non-Compositionality in Language Models

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autor principal:	Gomes, Andrew
Formato:	Preprint
Publicado:	2025
Materias:	Computation and Language Artificial Intelligence Machine Learning
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Schoenfeld's Anatomy of Mathematical Reasoning by Language Models
por: Li, Ming, et al.
Publicado: (2025)

Tracing Uncertainty in Language Model "Reasoning"
por: Grünefeld, Nils, et al.
Publicado: (2026)

MemTrace: Tracing and Attributing Errors in Large Language Model Memory Systems
por: Deng, Xinle, et al.
Publicado: (2026)

Language over Content: Tracing Cultural Understanding in Multilingual Large Language Models
por: Cho, Seungho, et al.
Publicado: (2025)

Transcoders Trace Visual Grounding and Hallucinations in Vision-Language Models
por: Damianos, Dimitrios, et al.
Publicado: (2026)

Tracing the Representation Geometry of Language Models from Pretraining to Post-training
por: Li, Melody Zixuan, et al.
Publicado: (2025)

The Geometric Anatomy of Capability Acquisition in Transformers
por: Billa, Jayadev
Publicado: (2026)

Generalization v.s. Memorization: Tracing Language Models' Capabilities Back to Pretraining Data
por: Wang, Xinyi, et al.
Publicado: (2024)

Compositional Causal Reasoning Evaluation in Language Models
por: Maasch, Jacqueline R. M. A., et al.
Publicado: (2025)

Understanding Subword Compositionality of Large Language Models
por: Peng, Qiwei, et al.
Publicado: (2025)

The Compositional Architecture of Regret in Large Language Models
por: Cui, Xiangxiang, et al.
Publicado: (2025)

Compositional Steering of Large Language Models with Steering Tokens
por: Radevski, Gorjan, et al.
Publicado: (2026)

Tracing Facts or just Copies? A critical investigation of the Competitions of Mechanisms in Large Language Models
por: Campregher, Dante, et al.
Publicado: (2025)

Non-linear Interventions on Large Language Models
por: Kim, Sangwoo
Publicado: (2026)

On-device System of Compositional Multi-tasking in Large Language Models
por: Bohdal, Ondrej, et al.
Publicado: (2025)

Efficient Compositional Multi-tasking for On-device Large Language Models
por: Bohdal, Ondrej, et al.
Publicado: (2025)

Recursive Concept Evolution for Compositional Reasoning in Large Language Models
por: Chaudhry, Sarim
Publicado: (2026)

Polynomial Composition Activations: Unleashing the Dynamics of Large Language Models
por: Zhuo, Zhijian, et al.
Publicado: (2024)

Non-Markovian Discrete Diffusion with Causal Language Models
por: Zhang, Yangtian, et al.
Publicado: (2025)

Multi-property Steering of Large Language Models with Dynamic Activation Composition
por: Scalena, Daniel, et al.
Publicado: (2024)

Learning to Route for Dynamic Adapter Composition in Continual Learning with Language Models
por: Araujo, Vladimir, et al.
Publicado: (2024)

Exposing Limitations of Language Model Agents in Sequential-Task Compositions on the Web
por: Furuta, Hiroki, et al.
Publicado: (2023)

Re-Tuning: Overcoming the Compositionality Limits of Large Language Models with Recursive Tuning
por: Pasewark, Eric, et al.
Publicado: (2024)

Do Large Language Models Have Compositional Ability? An Investigation into Limitations and Scalability
por: Xu, Zhuoyan, et al.
Publicado: (2024)

Composition of Experts: A Modular Compound AI System Leveraging Large Language Models
por: Jain, Swayambhoo, et al.
Publicado: (2024)

How Abilities in Large Language Models are Affected by Supervised Fine-tuning Data Composition
por: Dong, Guanting, et al.
Publicado: (2023)

Distinct Computations Emerge From Compositional Curricula in In-Context Learning
por: Lee, Jin Hwa, et al.
Publicado: (2025)

Towards Real-world Human Behavior Simulation: Benchmarking Large Language Models on Long-horizon, Cross-scenario, Heterogeneous Behavior Traces
por: Chen, Jiawei, et al.
Publicado: (2026)

The Randomness Floor: Measuring Intrinsic Non-Randomness in Language Model Token Distributions
por: Hryszko, Jarosław
Publicado: (2026)

Large Human Language Models: A Need and the Challenges
por: Soni, Nikita, et al.
Publicado: (2023)

Inconsistent Tokenizations Cause Language Models to be Perplexed by Japanese Grammar
por: Gambardella, Andrew, et al.
Publicado: (2025)

Time Travel in LLMs: Tracing Data Contamination in Large Language Models
por: Golchin, Shahriar, et al.
Publicado: (2023)

Labrador: Exploring the Limits of Masked Language Modeling for Laboratory Data
por: Bellamy, David R., et al.
Publicado: (2023)

ACAR: Adaptive Complexity Routing for Multi-Model Ensembles with Auditable Decision Traces
por: Kumaresan, Ramchand
Publicado: (2026)

Tracing Computation Density in LLMs
por: Kervadec, Corentin, et al.
Publicado: (2026)

Language Models Do Hard Arithmetic Tasks Easily and Hardly Do Easy Arithmetic Tasks
por: Gambardella, Andrew, et al.
Publicado: (2024)

Efficient Non-Parametric Uncertainty Quantification for Black-Box Large Language Models and Decision Planning
por: Tsai, Yao-Hung Hubert, et al.
Publicado: (2024)

Semantic Token Clustering for Efficient Uncertainty Quantification in Large Language Models
por: Cao, Qi, et al.
Publicado: (2026)

EmbedLLM: Learning Compact Representations of Large Language Models
por: Zhuang, Richard, et al.
Publicado: (2024)

Looking to Learn: Token-wise Dynamic Gating for Low-Resource Vision-Language Modelling
por: Ganescu, Bianca-Mihaela, et al.
Publicado: (2025)