Cover Image

How Is LLM Reasoning Distracted by Irrelevant Context? An Analysis Using a Controlled Benchmark

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Yang, Minglai, Huang, Ethan, Zhang, Liang, Surdeanu, Mihai, Wang, William, Pan, Liangming
Format:	Preprint
Published:	2025
Subjects:	Computation and Language Artificial Intelligence Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

CopySpec: Accelerating LLMs with Speculative Copy-and-Paste Without Compromising Quality
by: Dumitru, Razvan-Gabriel, et al.
Published: (2025)

Data Contamination Quiz: A Tool to Detect and Estimate Contamination in Large Language Models
by: Golchin, Shahriar, et al.
Published: (2023)

Opening the Black Box: A Survey on the Mechanisms of Multi-Step Reasoning in Large Language Models
by: Pan, Liangming, et al.
Published: (2026)

Memorization in In-Context Learning
by: Golchin, Shahriar, et al.
Published: (2024)

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

AlignSAE: Concept-Aligned Sparse Autoencoders
by: Yang, Minglai, et al.
Published: (2025)

Say Less, Mean More: Leveraging Pragmatics in Retrieval-Augmented Generation
by: Riaz, Haris, et al.
Published: (2025)

Understanding Reasoning Ability of Language Models From the Perspective of Reasoning Paths Aggregation
by: Wang, Xinyi, et al.
Published: (2024)

ELLEN: Extremely Lightly Supervised Learning For Efficient Named Entity Recognition
by: Riaz, Haris, et al.
Published: (2024)

Mechanistic Data Attribution: Tracing the Training Origins of Interpretable LLM Units
by: Chen, Jianhui, et al.
Published: (2026)

Investigating the Transferability of Code Repair for Low-Resource Programming Languages
by: Wong, Kyle, et al.
Published: (2024)

Collapse of Irrelevant Representations (CIR) Ensures Robust and Non-Disruptive LLM Unlearning
by: Sondej, Filip, et al.
Published: (2025)

A Lightweight Explainable Guardrail for Prompt Safety
by: Islam, Md Asiful, et al.
Published: (2026)

Do Latent-CoT Models Think Step-by-Step? A Mechanistic Study on Sequential Reasoning Tasks
by: Liang, Jia, et al.
Published: (2026)

Input-Time Scaling: Adding Noise and Irrelevance into Less-Is-More Drastically Improves Reasoning Performance and Efficiency
by: Huang, Rapheal, et al.
Published: (2025)

From Words to Numbers: Your Large Language Model Is Secretly A Capable Regressor When Given In-Context Examples
by: Vacareanu, Robert, et al.
Published: (2024)

Contextual Drag: How Errors in the Context Affect LLM Reasoning
by: Cheng, Yun, et al.
Published: (2026)

Towards Compute-Optimal Many-Shot In-Context Learning
by: Golchin, Shahriar, et al.
Published: (2025)

Improving Faithfulness of Abstractive Summarization by Controlling Confounding Effect of Irrelevant Sentences
by: Ghoshal, Asish, et al.
Published: (2022)

RuleArena: A Benchmark for Rule-Guided Reasoning with LLMs in Real-World Scenarios
by: Zhou, Ruiwen, et al.
Published: (2024)

Enhancing Transformer RNNs with Multiple Temporal Perspectives
by: Dumitru, Razvan-Gabriel, et al.
Published: (2024)

Finding a Wolf in Sheep's Clothing: Combating Adversarial Text-To-Image Prompts with Text Summarization
by: Cooper, Portia, et al.
Published: (2024)

AcademicEval: Live Long-Context LLM Benchmark
by: Zhang, Haozhen, et al.
Published: (2025)

ReasonBENCH: Benchmarking the (In)Stability of LLM Reasoning
by: Potamitis, Nearchos, et al.
Published: (2025)

Change Is the Only Constant: Dynamic LLM Slicing based on Layer Redundancy
by: Dumitru, Razvan-Gabriel, et al.
Published: (2024)

MorphNLI: A Stepwise Approach to Natural Language Inference Using Text Morphing
by: Negru, Vlad Andrei, et al.
Published: (2025)

The Knowledge Alignment Problem: Bridging Human and External Knowledge for Large Language Models
by: Zhang, Shuo, et al.
Published: (2023)

InductionBench: LLMs Fail in the Simplest Complexity Class
by: Hua, Wenyue, et al.
Published: (2025)

When and Where Did it Happen? An Encoder-Decoder Model to Identify Scenario Context
by: Noriega-Atala, Enrique, et al.
Published: (2024)

Pride and Prejudice: LLM Amplifies Self-Bias in Self-Refinement
by: Xu, Wenda, et al.
Published: (2024)

DRAGON: Guard LLM Unlearning in Context via Negative Detection and Reasoning
by: Wang, Yaxuan, et al.
Published: (2025)

Making Retrieval-Augmented Language Models Robust to Irrelevant Context
by: Yoran, Ori, et al.
Published: (2023)

ConciseRL: Conciseness-Guided Reinforcement Learning for Efficient Reasoning Models
by: Dumitru, Razvan-Gabriel, et al.
Published: (2025)

Towards a Mechanistic Understanding of Propositional Logical Reasoning in Large Language Models
by: Chen, Danchun, et al.
Published: (2026)

Benchmarking the Capabilities of Large Language Models in Transportation System Engineering: Accuracy, Consistency, and Reasoning Behaviors
by: Syed, Usman, et al.
Published: (2024)

Positional Failures in Long-Context LLMs: A Blind Spot in Reasoning Benchmarks
by: Zhang, Chuyifei, et al.
Published: (2026)

LogicVista: Multimodal LLM Logical Reasoning Benchmark in Visual Contexts
by: Xiao, Yijia, et al.
Published: (2024)

Interpreting and Controlling LLM Reasoning through Integrated Policy Gradient
by: Li, Changming, et al.
Published: (2026)

CriticBench: Benchmarking LLMs for Critique-Correct Reasoning
by: Lin, Zicheng, et al.
Published: (2024)

Large Language Models Badly Generalize across Option Length, Problem Types, and Irrelevant Noun Replacements
by: Zhao, Guangxiang, et al.
Published: (2025)