Coherence Collapse: Diagnosing Why Code Agents Fail After Reaching the Right Code

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Kim, Myeongsoo, Wang, Dingmin, Cui, Siwei, Farmahinifarahani, Farima, Zhuo, Terry Yue, Garg, Shweta, Ray, Baishakhi, Mukherjee, Rajdeep, Kumar, Varun
Format:	Preprint
Published:	2026
Subjects:	Software Engineering Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

CodeAssistBench (CAB): Dataset & Benchmarking for Multi-turn Chat-Based Code Assistance
by: Kim, Myeongsoo, et al.
Published: (2025)

CODESTRUCT: Code Agents over Structured Action Spaces
by: Kim, Myeongsoo, et al.
Published: (2026)

Towards Effectively Leveraging Execution Traces for Program Repair with Code LLMs
by: Haque, Mirazul, et al.
Published: (2025)

Cyber-Zero: Training Cybersecurity Agents without Runtime
by: Zhuo, Terry Yue, et al.
Published: (2025)

Training Language Model Agents to Find Vulnerabilities with CTF-Dojo
by: Zhuo, Terry Yue, et al.
Published: (2025)

Chain-of-Code Collapse: Reasoning Failures in LLMs via Adversarial Prompting in Code Generation
by: Roh, Jaechul, et al.
Published: (2025)

On Mitigating Code LLM Hallucinations with API Documentation
by: Jain, Nihal, et al.
Published: (2024)

ICE-Score: Instructing Large Language Models to Evaluate Code
by: Zhuo, Terry Yue
Published: (2023)

CWEval: Outcome-driven Evaluation on Functionality and Security of LLM Code Generation
by: Peng, Jinjun, et al.
Published: (2025)

Dynamic Benchmarking of Reasoning Capabilities in Code Large Language Models Under Data Contamination
by: Chen, Simin, et al.
Published: (2025)

SpecTra: Enhancing the Code Translation Ability of Language Models by Generating Multi-Modal Specifications
by: Nitin, Vikram, et al.
Published: (2024)

CYCLE: Learning to Self-Refine the Code Generation
by: Ding, Yangruibo, et al.
Published: (2024)

EditLord: Learning Code Transformation Rules for Code Editing
by: Li, Weichen, et al.
Published: (2025)

Automated Code Editing with Search-Generate-Modify
by: Liu, Changshu, et al.
Published: (2023)

CodeSense: a Real-World Benchmark and Dataset for Code Semantic Reasoning
by: Roy, Monoshi Kumar, et al.
Published: (2025)

LeDex: Training LLMs to Better Self-Debug and Explain Code
by: Jiang, Nan, et al.
Published: (2024)

Substance Beats Style: Why Beginning Students Fail to Code with LLMs
by: Lucchetti, Francesca, et al.
Published: (2024)

CodeScout: Contextual Problem Statement Enhancement for Software Agents
by: Suri, Manan, et al.
Published: (2026)

Revealing degradation mechanisms in YSZ ceramics through machine learning-guided aging and multiscale characterization
by: Garg, Prachi, et al.
Published: (2025)

Code Quality Analysis of Translations from C to Rust
by: Tadesse, Biruk, et al.
Published: (2026)

Code Reasoning for Software Engineering Tasks: A Survey and A Call to Action
by: Pujar, Saurabh, et al.
Published: (2025)

PrivCode: When Code Generation Meets Differential Privacy
by: Liu, Zheng, et al.
Published: (2025)

SemCoder: Training Code Language Models with Comprehensive Semantics Reasoning
by: Ding, Yangruibo, et al.
Published: (2024)

Why Is Exclusivity in Broadcasting Rights Prevalent and Why Does Simple Regulation Fail?
by: David Martimort, et al.
Published: (2025)

Why Prototypes Collapse: Diagnosing and Preventing Partial Collapse in Prototypical Self-Supervised Learning
by: Arteaga, Gabriel Y., et al.
Published: (2025)

Why Self-Awareness Fails: Coherence, Defense, and the Structure of Repair
by: Jovanovic, Vladisav
Published: (2026)

Cracking the Code of Arctic Sea Ice: Why Models Fail to Predict Its Retreat?
by: Gou, Ruijian, et al.
Published: (2025)

The Observability Gap: Why Output-Level Human Feedback Fails for LLM Coding Agents
by: Wang, Yinghao, et al.
Published: (2026)

CodeSSM: Towards State Space Models for Code Understanding
by: Verma, Shweta, et al.
Published: (2025)

CodeFort: Robust Training for Code Generation Models
by: Zhang, Yuhao, et al.
Published: (2024)

Stop Probing, Start Coding: Why Linear Probes and Sparse Autoencoders Fail at Compositional Generalisation
by: Pacela, Vitória Barin, et al.
Published: (2026)

Robustness, Security, Privacy, Explainability, Efficiency, and Usability of Large Language Models for Code
by: Yang, Zhou, et al.
Published: (2024)

CodeArena: A Collective Evaluation Platform for LLM Code Generation
by: Du, Mingzhe, et al.
Published: (2025)

XFT: Unlocking the Power of Code Instruction Tuning by Simply Merging Upcycled Mixture-of-Experts
by: Ding, Yifeng, et al.
Published: (2024)

Beyond Accuracy: Evaluating Self-Consistency of Code Large Language Models with IdentityChain
by: Min, Marcus J., et al.
Published: (2023)

Coherence manipulation in asymmetry and thermodynamics
by: Kondra, Tulja Varun, et al.
Published: (2023)

Colour Codes Reach Surface Code Performance using Vibe Decoding
by: Koutsioumpas, Stergios, et al.
Published: (2025)

Machine learning–guided optimization of coercive field in Al 1− x Sc x N thin films for nonvolatile memory
by: Shaon Das, et al.
Published: (2024)

Vulnerability Detection with Code Language Models: How Far Are We?
by: Ding, Yangruibo, et al.
Published: (2024)

Chain-of-Thought in Neural Code Generation: From and For Lightweight Language Models
by: Yang, Guang, et al.
Published: (2023)