arrow
返回

Two Steps to Precision: Enhancing Reliable API Invocation in Code Generation

delete2026-01-01
delete0
PRE
AI
Y
Yanming Li
Z
Ziye Tang
W
Wang, Siyi
G
Guilin Qi *
DOI:10.1007/978-981-95-3346-6_33delete
delete原文链接
delete原文求助
delete分享
delete收藏
摘要

摘要

En 中文
自动化代码生成在现代软件开发中至关重要,但大型语言模型在代码版本控制和多API调用等实际挑战方面存在困难。现有方法包括直接生成和检索增强方法,往往无法确保精确的API使用。本文介绍了一种简单而有效的两步框架:先进行粗略代码生成或检索,再进行精细代码编辑。在VersiCode和BigCodeBench上的实验表明,该框架在版本特定代码补全和函数级编程方面取得了显著性能提升。这些结果表明该框架在增强基于LLM的代码生成系统方面具有实用性。
Keyword:
code generation
API invocation
code editing
large language models
version-specific completion

期刊

N
NATURAL LANGUAGE PROCESSING AND CHINESE COMPUTING, NLPCC 2025, PT II
IF:
0
论文数:
29
被引数:
0

机构

S
southeast university - china
学者数:
5.3W
论文数: 4.9W
被引数: 57
引用论文

引用论文

DevEval: A Manually-Annotated Code Generation Benchmark Aligned with Real-World Code Repositories
err2024-01-01
err0
PREAI
errJia Li; Ge Li; Yunfei Zhao; Yongmin Li; Huanyu Liu; Hao Zhu; Lecheng Wang; Kaibo Liu; Zheng Fang; Lanshen Wang; Jiazheng Ding; Xuanming Zhang; Yuqi Zhu; Yihong Dong; Zhi Jin; Binhua Li; Fei Huang; Yongbin Li; Bin Gu; Mengfei Yang
err分享
err收藏
Retrieval-Based Neural Code Generation
err2018-01-01
err0
errOAAI
errShirley Anugrah Hayati; Raphael Olivier; Pravalika Avvaru; Pengcheng Yin; Anthony Tomasic; Graham Neubig
err分享
err收藏