paper-with-me

Papers

CodeCircuit: Toward Inferring LLM-Generated Code Correctness via Attribution Graphs

2026-02-06 · Yicheng He, Zheng Zhao, Zhou Kaiyu, Bryan Dai, Jie Fu, Yonghui Yang arxiv

Current paradigms for code verification rely heavily on external mechanisms-such as execution-based unit tests or auxiliary LLM judges-which are often labor-intensive or limited by the judging model's own capabilities. This raises a fundamental, yet unexplored question: Can an LLM's functional correctness be assessed purely from its internal computational structure? Our primary objective is to investigate whether the model's neural dynamics encode internally decodable signals that are predictive of logical validity during code generation. Inspired by mechanistic interpretability, we propose to treat code verification as a mechanistic diagnostic task, mapping the model's explicit algorithmic trajectory into line-level attribution graphs. By decomposing complex residual flows, we aim to identify the structural signatures that distinguish sound reasoning from logical failure within the model's internal circuits. Analysis across Python, C++, and Java confirms that intrinsic correctness signals are robust across diverse syntaxes. Topological features from these internal graphs predict correctness more reliably than surface heuristics and enable targeted causal interventions to fix erroneous logic. These findings establish internal introspection as a decodable property for verifying generated code. Our code is at https:// github.com/bruno686/CodeCircuit.

📄 PDF Abstract BibTeX arXiv:2602.07080

Code (0)

등록된 구현이 없습니다.

Tasks

Code Generation

Similar Papers 제목 키워드 기반

Correctness is not Faithfulness in RAG Attributions

2024-12-23 · Jonas Wallat, Maria Heuss, Maarten de Rijke, Avishek Anand

Retrieving relevant context is a common approach to reduce hallucinations and enhance answer reliability. Explicitly citing source documents allows users to verify generated responses and increases trust. Prior work larg…

RAG

Self-Attribution Bias: When AI Monitors Go Easy on Themselves

2026-03-04 · Dipika Khullar, Jack Hopkins, Rowan Wang, Fabien Roger arxiv

Agentic systems increasingly rely on language models to monitor their own behavior. For example, coding agents may self critique generated code for pull request approval or assess the safety of tool-use actions. We show …

Where Did I Come From? Origin Attribution of AI-Generated Images

2023-09-21 · NeurIPS 2023 11

Image generation techniques have been gaining increasing attention recently, but concerns have been raised about the potential misuse and intellectual property (IP) infringement associated with image generation models. I…

Alteration-free and Model-agnostic Origin Attribution of Generated Images

2023-05-29 · Zhenting Wang, Chen Chen, Yi Zeng, Lingjuan Lyu 외

Recently, there has been a growing attention in image generation models. However, concerns have emerged regarding potential misuse and intellectual property (IP) infringement associated with these models. Therefore, it i…

Image Generation

SAM: The Sensitivity of Attribution Methods to Hyperparameters

2020-03-04 · CVPR 2020 6 · Naman Bansal, Chirag Agarwal, Anh Nguyen

Attribution methods can provide powerful insights into the reasons for a classifier's decision. We argue that a key desideratum of an explanation method is its robustness to input hyperparameters which are often randomly…

Sensitivity