代码大模型 / AI 编程 - arXivDaily 专题

2402.16896 2024-03-08 cs.CR cs.LG cs.SE 62%

On Trojan Signatures in Large Language Models of Code

Aftab Hussain, Md Rafiqul Islam Rabin, Mohammad Amin Alipour

专题命中其他AI编程：code model(abstract)；分类 cs.SE、cs.LG

Comments This work has been accepted at the International Conference on Learning Representations 2024 Workshop on Secure and Trustworthy Large Language Models, SeT LLM @ ICLR 2024 (Vienna, Austria)

2402.16790 2024-02-27 cs.SE cs.AI 62%

Beyond Self-learned Attention: Mitigating Attention Bias in Transformer-based Models Using Attention Guidance

Jiri Gesi, Iftekhar Ahmed

专题命中其他AI编程：code model(abstract)；分类 cs.SE、cs.AI

2303.03919 2023-12-15 cs.LG cs.CL 62%

Data Portraits: Recording Foundation Model Training Data

Marc Marone, Benjamin Van Durme

专题命中其他AI编程：code model(abstract)；分类 cs.CL、cs.LG

Comments NeurIPS 2023 Datasets and Benchmarks

2310.16803 2023-10-26 cs.CL cs.LG 62%

Language Agnostic Code Embeddings

Saiteja Utpala, Alex Gu, Pin Yu Chen

专题命中其他AI编程：code model(abstract)；分类 cs.CL、cs.LG

2204.11454 2022-11-02 cs.CL cs.SE 62%

Natural Language to Code Translation with Execution

Freda Shi, Daniel Fried, Marjan Ghazvininejad, Luke Zettlemoyer, Sida I. Wang

专题命中其他AI编程：code model(abstract)；分类 cs.SE、cs.CL

Comments EMNLP 2022

2201.05624 2022-06-08 cs.LG cs.AI cs.NA math.NA physics.data-an 62%

Scientific Machine Learning through Physics-Informed Neural Networks: Where we are and What's next

Salvatore Cuomo, Vincenzo Schiano di Cola, Fabio Giampaolo, Gianluigi Rozza, Maziar Raissi, Francesco Piccialli

专题命中其他AI编程：code model(abstract)；分类 cs.AI、cs.LG

2107.07402 2022-01-14 cs.CL cs.LG cs.SD eess.AS 62%

CLSRIL-23: Cross Lingual Speech Representations for Indic Languages

Anirudh Gupta, Harveen Singh Chadha, Priyanshi Shah, Neeraj Chhimwal, Ankur Dhuriya, Rishabh Gaur, Vivek Raghavan

专题命中其他AI编程：code model(abstract)；分类 cs.CL、cs.LG

Comments 7 pages, 2 figures

2111.11435 2021-11-24 cs.SE cs.LG 62%

Precise Learning of Source Code Contextual Semantics via Hierarchical Dependence Structure and Graph Attention Networks

Zhehao Zhao, Bo Yang, Ge Li, Huai Liu, Zhi Jin

专题命中其他AI编程：code model(abstract)；分类 cs.SE、cs.LG

Comments 17pages, published on Journal of Systems and Software

Journal ref Journal of Systems and Software, Volume 184, 2022, 111108, ISSN 0164-1212

2104.09948 2021-04-21 cs.PL cs.SE 62%

A Generative Approach for User-Centered, Collaborative, Domain-Specific Modeling Environments

Philip Zweihoff, Bernhard Steffen

专题命中其他AI编程：code model(abstract)；分类 cs.SE、cs.PL

1807.00973 2018-07-04 cs.AI cs.LG stat.ML 62%

Scalable Structure Learning for Probabilistic Soft Logic

Varun Embar, Dhanya Sridhar, Golnoosh Farnadi, Lise Getoor

专题命中其他AI编程：code model(abstract)；分类 cs.AI、cs.LG

Comments 8 pages, 2 figures

1606.03153 2018-05-30 cs.CL cs.LG 62%

Unsupervised Learning of Word-Sequence Representations from Scratch via Convolutional Tensor Decomposition

Furong Huang, Animashree Anandkumar

专题命中其他AI编程：code model(abstract)；分类 cs.CL、cs.LG

Comments There was an error in section 3, there is a bug in the experiment section. We would like to take it down

1312.0344 2013-12-03 cs.SE cs.PL 62%

An NMF solution for the Flowgraphs case at the TTC 2013

Georg Hinkel, Thomas Goldschmidt, Lucia Happe

专题命中其他AI编程：code model(abstract)；分类 cs.SE、cs.PL

Comments In Proceedings TTC 2013, arXiv:1311.7536

Journal ref EPTCS 135, 2013, pp. 37-42

2607.20950 2026-07-24 cs.LG 新提交 57%

Best-of-Evidence: Best-of-N Selection under Partial Verification

最佳证据：部分验证下的最佳N选择

Cenwei Zhang, Teng Fang, Yuxia Wang, Derek Li, Bryan Dai, Lei You

机构 * IQuest Research（IQuest研究公司）； INSAIT（INSAIT公司）； Technical University of Denmark（丹麦技术大学）

专题命中其他AI编程：code model(abstract)；分类 cs.LG

AI总结研究视觉语言任务中部分验证问题，提出最佳证据（BoE）推理时选择框架，保持候选池固定，用图表示主张并分配预算，形式化部分验证选择，有实用控制器，实验表明其能改进选择并揭示相关限制。

Comments 3 figures, 28 pages

2602.06774 2026-06-18 cs.AI 版本更新 57%

Towards Understanding What State Space Models Learn About Code

理解状态空间模型在代码中学到了什么

Jiali Wu, Abhinav Anand, Shweta Verma, Mira Mezini

机构 * TU Darmstadt（图宾根大学）； Hessian Center for Artificial Intelligence（黑森人工智能中心）； National Research Center for Applied Cybersecurity ATHENE（应用网络安全国家研究中心ATHENE）

专题命中其他AI编程：code model(abstract)；分类 cs.AI

AI总结本文首次系统分析状态空间模型（SSM）在代码理解中的学习机制，发现SSM在预训练时比Transformer更有效捕获语法和语义结构，但微调时会遗忘某些关系，并提出SSM-Interpret框架和架构改进，将NLCodeSearch的MRR提升高达6。

2606.16229 2026-06-16 cs.PL 新提交 57%

Synthesizing Best Abstract Transformers via Parallel Bit-Vector Optimization

通过并行位向量优化合成最佳抽象转换器

Weiqi Wang, Peisen Yao, Hanrui Zuo, Yuan Li, Hongfei Fu, Kui Ren

专题命中其他AI编程：code model(abstract)；分类 cs.PL

AI总结提出Spear并行合成框架，利用目标间位独立性加速最优抽象转换器合成，在二进制分析基准上优于现有OMT求解器。

2606.13918 2026-06-15 cs.SE cs.CR 新提交 57%

Bayesian-Calibrated Detection of Hallucinated Package Imports in AI-Assisted Code

AI辅助代码中幻觉包导入的贝叶斯校准检测

Lom M. Hillah, Jean-Marc Richard, Ryan Hasnaoui

专题命中其他AI编程：code model(abstract)；分类 cs.SE

AI总结提出贝叶斯校准层，基于三元认知分类法输出Beta后验概率，并利用PyPI元数据检测注册但可疑的包，优于二元基线。

Comments 23 pages, 2 figures, 5 tables

2606.12364 2026-06-11 cs.LG 新提交 57%

On Subquadratic Architectures: From Applications to Principles

关于次二次架构：从应用到原理

Anamaria-Roberta Hartl, Levente Zólyomi, David Stap, Pieter-Jan Hoedt, Niklas Schmidinger, Lukas Hauzenberger, Sebastian Böck, Günter Klambauer, Sepp Hochreiter

机构 * ELLIS Unit Linz, LIT AI Lab, Institute for Machine Learning, Johannes Kepler University Linz（林茨ELLIS单元、LIT AI实验室、机器学习研究所、约翰内斯·开普勒大学林茨）； NVIDIA（英伟达）

专题命中其他AI编程：code model(abstract)；分类 cs.LG

AI总结本文比较了xLSTM、Mamba-2和Gated DeltaNet三种次二次架构，发现xLSTM在代码预训练、蒸馏和时间序列预训练中表现最佳，其优势源于灵活稳定的门控记忆校正机制。

2504.16584 2026-05-14 cs.CR cs.AI 57%

Case Study: Fine-tuning Small Language Models for Accurate and Private CWE Detection in Python Code

案例研究：微调小型语言模型以在Python代码中实现准确且隐私的CWE检测

Md. Azizul Hakim Bappy, Hossen A Mustafa, Prottoy Saha, Rajinus Salehat

机构 * Institute of Information and Communication Technology, Bangladesh University of Engineering Technology（孟加拉工程科技大学信息与通信技术研究所）； Hajee Mohammad Danesh Science and Technology University（海杰莫哈默德丹什科学与技术大学）

专题命中其他AI编程：code model(abstract)；分类 cs.AI

AI总结本文研究了微调小型语言模型用于在Python代码中准确检测CWEs的可能性，通过半监督方法生成数据并人工审核，最终在测试集上实现了99%的准确率和99.04%的F1分数，展示了其在隐私保护下的高效安全分析潜力。

Comments 11 pages, 2 figures, 3 tables. Dataset available at https://huggingface.co/datasets/floxihunter/synthetic_python_cwe. Model available at https://huggingface.co/floxihunter/codegen-mono-CWEdetect. Keywords: Small Language Models (SLMs), Vulnerability Detection, CWE, Fine-tuning, Python Security, Privacy-Preserving Code Analysis

2604.06095 2026-04-08 cs.CR cs.AI 57%

LLM4CodeRE: Generative AI for Code Decompilation Analysis and Reverse Engineering

LLM4CodeRE: 生成式AI用于代码反编译分析与逆向工程

Hamed Jelodar, Samita Bai, Tochukwu Emmanuel Nwankwo, Parisa Hamedi, Mohammad Meymani, Roozbeh Razavi-Far, Ali A. Ghorbani

专题命中其他AI编程：code model(abstract)；分类 cs.AI

AI总结本文提出LLM4CodeRE框架，通过双向代码逆向工程实现反编译与源码翻译，采用多适配器和序列到序列统一方法提升任务适应性，实验显示其在反编译任务中表现优异。

2604.02278 2026-04-03 cs.SE 57%

LLMs as Idiomatic Decompilers: Recovering High-Level Code from x86-64 Assembly for Dart

LLMs作为idiomatic反编译器：从x86-64汇编语言恢复Dart的高级代码

Raafat Abualazm, Ayman Abo Elhassan

专题命中其他AI编程：code model(abstract)；分类 cs.SE

AI总结本文研究了小型专用LLM作为idiomatic反编译器在Dart等现代语言中的应用，通过合成同语言示例和相关语言示例的对比，展示了专用模型在生成可读且符合 idiomatic 的Dart代码方面的有效性。

Comments 5 pages, 1 figure, 3 tables. Accepted at SANER 2026 ERA Track

2601.15892 2026-01-26 cs.CL 57%

Stable-DiffCoder: Pushing the Frontier of Code Diffusion Large Language Model

Stable-DiffCoder：推动代码扩散大语言模型的前沿

Chenghao Fan, Wen Heng, Bo Li, Sichen Liu, Yuxuan Song, Jing Su, Xiaoye Qu, Kai Shen, Wei Wei

机构 * Huazhong University of Science and Technology（华中科技大学）

专题命中其他AI编程：code model(abstract)；分类 cs.CL

AI总结 Stable-DiffCoder通过基于扩散的训练方法，在代码生成任务中超越了传统自回归模型，提升了代码建模的质量和效率。

2511.20654 2025-11-27 cs.HC cs.AI 57%

CodeVaani: A Multilingual, Voice-Based Code Learning Assistant

CodeVaani：一种多语言、基于语音的代码学习助手

Jayant Havare, Srikanth Tamilselvam, Ashish Mittal, Shalaka Thorat, Soham Jadia, Varsha Apte, Ganesh Ramakrishnan

机构 * Indian Institute of Technology Bombay（印度理工学院班加罗尔分校）； IBM Research（IBM研究院）

专题命中其他AI编程：code model(abstract)；分类 cs.AI

AI总结 CodeVaani通过多语言语音交互帮助非英语母语学生更高效地学习编程，实现75%的响应准确率和高用户满意度。

2511.03136 2025-11-06 cs.SE 57%

Automated Prompt Generation for Code Intelligence: An Empirical study and Experience in WeChat

Kexing Ji, Shiyun Fu, Cuiyun Gao, Yujia Chen, Zezhou Yang, Chaozheng Wang, Yuetang Deng

专题命中其他AI编程：code model(abstract)；分类 cs.SE

Comments Accepted by ASE 2025 Industry Track

2510.17158 2025-10-21 cs.DC cs.PF cs.SE 57%

Integrating Performance Tools in Model Reasoning for GPU Kernel Optimization

Daniel Nichols, Konstantinos Parasyris, Charles Jekel, Abhinav Bhatele, Harshitha Menon

专题命中其他AI编程：code model(abstract)；分类 cs.SE

2509.07867 2025-09-10 cs.AI 57%

CP-Model-Zoo: A Natural Language Query System for Constraint Programming Models

Augustin Crespin, Ioannis Kostis, Hélène Verhaeghe, Pierre Schaus

机构 * UCLouvain（乌尔斯坦大学）； ICTEAM（信息与计算技术团队）

专题命中其他AI编程：code model(abstract)；分类 cs.AI

Comments presented at"LLMs meet Constraint Solving" Workshop at CP2025 in Glasgow

2410.11203 2024-10-16 cs.LG 57%

Error Diffusion: Post Training Quantization with Block-Scaled Number Formats for Neural Networks

Alireza Khodamoradi, Kristof Denolf, Eric Dellinger

专题命中其他AI编程：code model(abstract)；分类 cs.LG

2405.16746 2024-10-01 cs.SE 57%

Ecosystem of Large Language Models for Code

Zhou Yang, Jieke Shi, Premkumar Devanbu, David Lo

专题命中其他AI编程：code model(abstract)；分类 cs.SE

2405.11466 2024-05-21 cs.SE 57%

Measuring Impacts of Poisoning on Model Parameters and Embeddings for Large Language Models of Code

Aftab Hussain, Md Rafiqul Islam Rabin, Mohammad Amin Alipour

专题命中其他AI编程：code model(abstract)；分类 cs.SE

Comments This work has been accepted at the 1st ACM International Conference on AI-powered Software (AIware), co-located with the ACM International Conference on the Foundations of Software Engineering (FSE) 2024, Porto de Galinhas, Brazil. arXiv admin note: substantial text overlap with arXiv:2402.12936

2201.00150 2024-03-13 cs.SE 57%

Cross-Domain Deep Code Search with Meta Learning

Yitian Chai, Hongyu Zhang, Beijun Shen, Xiaodong Gu

专题命中其他AI编程：code model(abstract)；分类 cs.SE

Comments Accepted by ICSE 2022 (The 44th International Conference on Software Engineering)

2402.12936 2024-03-06 cs.SE 57%

Measuring Impacts of Poisoning on Model Parameters and Neuron Activations: A Case Study of Poisoning CodeBERT

Aftab Hussain, Md Rafiqul Islam Rabin, Navid Ayoobi, Mohammad Amin Alipour

专题命中其他AI编程：code model(abstract)；分类 cs.SE

1. 其他AI编程 240 篇

On Trojan Signatures in Large Language Models of Code

Beyond Self-learned Attention: Mitigating Attention Bias in Transformer-based Models Using Attention Guidance

Data Portraits: Recording Foundation Model Training Data

Language Agnostic Code Embeddings

Natural Language to Code Translation with Execution

Scientific Machine Learning through Physics-Informed Neural Networks: Where we are and What's next

CLSRIL-23: Cross Lingual Speech Representations for Indic Languages

Precise Learning of Source Code Contextual Semantics via Hierarchical Dependence Structure and Graph Attention Networks

A Generative Approach for User-Centered, Collaborative, Domain-Specific Modeling Environments

Scalable Structure Learning for Probabilistic Soft Logic

Unsupervised Learning of Word-Sequence Representations from Scratch via Convolutional Tensor Decomposition

An NMF solution for the Flowgraphs case at the TTC 2013

Best-of-Evidence: Best-of-N Selection under Partial Verification

Towards Understanding What State Space Models Learn About Code

Synthesizing Best Abstract Transformers via Parallel Bit-Vector Optimization

Bayesian-Calibrated Detection of Hallucinated Package Imports in AI-Assisted Code

On Subquadratic Architectures: From Applications to Principles

Case Study: Fine-tuning Small Language Models for Accurate and Private CWE Detection in Python Code

LLM4CodeRE: Generative AI for Code Decompilation Analysis and Reverse Engineering

LLMs as Idiomatic Decompilers: Recovering High-Level Code from x86-64 Assembly for Dart

Stable-DiffCoder: Pushing the Frontier of Code Diffusion Large Language Model

CodeVaani: A Multilingual, Voice-Based Code Learning Assistant

Automated Prompt Generation for Code Intelligence: An Empirical study and Experience in WeChat

Integrating Performance Tools in Model Reasoning for GPU Kernel Optimization

CP-Model-Zoo: A Natural Language Query System for Constraint Programming Models

Error Diffusion: Post Training Quantization with Block-Scaled Number Formats for Neural Networks

Ecosystem of Large Language Models for Code

Measuring Impacts of Poisoning on Model Parameters and Embeddings for Large Language Models of Code

Cross-Domain Deep Code Search with Meta Learning

Measuring Impacts of Poisoning on Model Parameters and Neuron Activations: A Case Study of Poisoning CodeBERT