arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.09054cs.SEcs.AIcs.PL
收藏

大语言模型在竞技编程中的性能评估:一项跨多年、跨年级的分析

Evaluating the Performance of Large Language Models in Competitive Programming: A Multi-Year, Multi-Grade Analysis

University of Bucharest(布加勒斯特大学);It Just Works Inc.(It Just Works公司)

Adrian Marius Dumitran, Adrian Catalin Badea, Stefan-Gabriel Muscalu

AI总结:

本研究基于2002-2023年罗马尼亚县级信息学奥赛304道题的数据集,评估多款大语言模型的竞技编程表现,发现其性能随年级、题型有显著差异,GPT-4表现突出可辅助初中编程教学,且不同模型代码质量与风格有别。

摘要:

本研究探讨了大语言模型(LLM)在求解罗马尼亚县级信息学奥林匹克竞赛的竞技编程题目时的表现。罗马尼亚作为计算机科学竞赛领域的领先国家,凭借其深厚的竞赛积淀与严格的竞赛标准,为评估LLM能力提供了理想环境。我们收集并分析了一个包含2002年至2023年间304道赛题的数据集,重点研究LLM针对这些题目分别用C++和Python编写的解题代码,核心目标是厘清LLM在不同任务上表现优劣的原因。我们采用含多次尝试与多轮反馈的标准化流程,评估了多款模型,包括GPT-4等闭源模型,以及CodeLlama、RoMistral等开源权重模型。分析显示,LLM的表现随年级和题目类型的不同存在显著差异。值得注意的是,GPT-4表现优异,显示出其可作为初中生教育工具的潜力。我们还观察到不同LLM生成的代码在质量与风格上存在差异。

英文摘要:

This study explores the performance of large language models (LLMs) in solving competitive programming problems from the Romanian Informatics Olympiad at the county level. Romania, a leading nation in computer science competitions, provides an ideal environment for evaluating LLM capabilities due to its rich history and stringent competition standards. We collected and analyzed a dataset comprising 304 challenges from 2002 to 2023, focusing on solutions written by LLMs in C++ and Python for these problems. Our primary goal is to understand why LLMs perform well or poorly on different tasks. We evaluated various models, including closed-source models like GPT-4 and open-weight models such as CodeLlama and RoMistral, using a standardized process involving multiple attempts and feedback rounds. The analysis revealed significant variations in LLM performance across different grades and problem types. Notably, GPT-4 showed strong performance, indicating its potential use as an educational tool for middle school students. We also observed differences in code quality and style across various LLMs

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑