论程序生成语言模型的可靠性和可解释性

IF 6.6 2区 计算机科学 Q1 COMPUTER SCIENCE, SOFTWARE ENGINEERING ACM Transactions on Software Engineering and Methodology Pub Date : 2024-01-18 DOI:10.1145/3641540
Yue Liu, Chakkrit Tantithamthavorn, Yonghui Liu, Li Li
{"title":"论程序生成语言模型的可靠性和可解释性","authors":"Yue Liu, Chakkrit Tantithamthavorn, Yonghui Liu, Li Li","doi":"10.1145/3641540","DOIUrl":null,"url":null,"abstract":"<p>Recent studies have adopted pre-trained language models, such as CodeT5 and CodeGPT, for automated program generation tasks like code generation, repair, and translation. Numerous language model-based approaches have been proposed and evaluated on various benchmark datasets, demonstrating promising performance. However, there is still uncertainty about the reliability of these models, particularly their realistic ability to consistently transform code sequences. This raises the question: are these techniques sufficiently trustworthy for automated program generation? Consequently, Further research is needed to understand model logic and assess reliability and explainability. To bridge these research gaps, we conduct a thorough empirical study of eight popular language models on five representative datasets to determine the capabilities and limitations of automated program generation approaches. We further employ advanced explainable AI approaches to highlight the tokens that significantly contribute to the code transformation. We discover that state-of-the-art approaches suffer from inappropriate performance evaluation stemming from severe data duplication, causing over-optimistic results. Our explainability analysis reveals that, in various experimental scenarios, language models can recognize code grammar and structural information, but they exhibit limited robustness to changes in input sequences. Overall, more rigorous evaluation approaches and benchmarks are critical to enhance the reliability and explainability of automated program generation moving forward. Our findings provide important guidelines for this goal.</p>","PeriodicalId":50933,"journal":{"name":"ACM Transactions on Software Engineering and Methodology","volume":"48 1","pages":""},"PeriodicalIF":6.6000,"publicationDate":"2024-01-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"On the Reliability and Explainability of Language Models for Program Generation\",\"authors\":\"Yue Liu, Chakkrit Tantithamthavorn, Yonghui Liu, Li Li\",\"doi\":\"10.1145/3641540\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p>Recent studies have adopted pre-trained language models, such as CodeT5 and CodeGPT, for automated program generation tasks like code generation, repair, and translation. Numerous language model-based approaches have been proposed and evaluated on various benchmark datasets, demonstrating promising performance. However, there is still uncertainty about the reliability of these models, particularly their realistic ability to consistently transform code sequences. This raises the question: are these techniques sufficiently trustworthy for automated program generation? Consequently, Further research is needed to understand model logic and assess reliability and explainability. To bridge these research gaps, we conduct a thorough empirical study of eight popular language models on five representative datasets to determine the capabilities and limitations of automated program generation approaches. We further employ advanced explainable AI approaches to highlight the tokens that significantly contribute to the code transformation. We discover that state-of-the-art approaches suffer from inappropriate performance evaluation stemming from severe data duplication, causing over-optimistic results. Our explainability analysis reveals that, in various experimental scenarios, language models can recognize code grammar and structural information, but they exhibit limited robustness to changes in input sequences. Overall, more rigorous evaluation approaches and benchmarks are critical to enhance the reliability and explainability of automated program generation moving forward. Our findings provide important guidelines for this goal.</p>\",\"PeriodicalId\":50933,\"journal\":{\"name\":\"ACM Transactions on Software Engineering and Methodology\",\"volume\":\"48 1\",\"pages\":\"\"},\"PeriodicalIF\":6.6000,\"publicationDate\":\"2024-01-18\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"ACM Transactions on Software Engineering and Methodology\",\"FirstCategoryId\":\"94\",\"ListUrlMain\":\"https://doi.org/10.1145/3641540\",\"RegionNum\":2,\"RegionCategory\":\"计算机科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"COMPUTER SCIENCE, SOFTWARE ENGINEERING\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"ACM Transactions on Software Engineering and Methodology","FirstCategoryId":"94","ListUrlMain":"https://doi.org/10.1145/3641540","RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, SOFTWARE ENGINEERING","Score":null,"Total":0}
引用次数: 0

摘要

最近的研究采用了预先训练好的语言模型,如 CodeT5 和 CodeGPT,用于自动程序生成任务,如代码生成、修复和翻译。许多基于语言模型的方法已被提出,并在各种基准数据集上进行了评估,显示出良好的性能。然而,这些模型的可靠性仍存在不确定性,特别是它们持续转换代码序列的实际能力。这就提出了一个问题:这些技术对于自动程序生成是否足够可靠?因此,我们需要开展进一步研究,以了解模型逻辑并评估可靠性和可解释性。为了弥补这些研究空白,我们在五个代表性数据集上对八个流行语言模型进行了全面的实证研究,以确定自动程序生成方法的能力和局限性。我们进一步采用了先进的可解释人工智能方法,以突出对代码转换有重大贡献的标记。我们发现,最先进的方法因数据严重重复而导致性能评估不当,造成结果过于乐观。我们的可解释性分析表明,在各种实验场景中,语言模型可以识别代码语法和结构信息,但它们对输入序列变化的鲁棒性有限。总之,更严格的评估方法和基准对于提高自动程序生成的可靠性和可解释性至关重要。我们的研究结果为实现这一目标提供了重要指导。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
On the Reliability and Explainability of Language Models for Program Generation

Recent studies have adopted pre-trained language models, such as CodeT5 and CodeGPT, for automated program generation tasks like code generation, repair, and translation. Numerous language model-based approaches have been proposed and evaluated on various benchmark datasets, demonstrating promising performance. However, there is still uncertainty about the reliability of these models, particularly their realistic ability to consistently transform code sequences. This raises the question: are these techniques sufficiently trustworthy for automated program generation? Consequently, Further research is needed to understand model logic and assess reliability and explainability. To bridge these research gaps, we conduct a thorough empirical study of eight popular language models on five representative datasets to determine the capabilities and limitations of automated program generation approaches. We further employ advanced explainable AI approaches to highlight the tokens that significantly contribute to the code transformation. We discover that state-of-the-art approaches suffer from inappropriate performance evaluation stemming from severe data duplication, causing over-optimistic results. Our explainability analysis reveals that, in various experimental scenarios, language models can recognize code grammar and structural information, but they exhibit limited robustness to changes in input sequences. Overall, more rigorous evaluation approaches and benchmarks are critical to enhance the reliability and explainability of automated program generation moving forward. Our findings provide important guidelines for this goal.

求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
ACM Transactions on Software Engineering and Methodology
ACM Transactions on Software Engineering and Methodology 工程技术-计算机:软件工程
CiteScore
6.30
自引率
4.50%
发文量
164
审稿时长
>12 weeks
期刊介绍: Designing and building a large, complex software system is a tremendous challenge. ACM Transactions on Software Engineering and Methodology (TOSEM) publishes papers on all aspects of that challenge: specification, design, development and maintenance. It covers tools and methodologies, languages, data structures, and algorithms. TOSEM also reports on successful efforts, noting practical lessons that can be scaled and transferred to other projects, and often looks at applications of innovative technologies. The tone is scholarly but readable; the content is worthy of study; the presentation is effective.
期刊最新文献
Effective, Platform-Independent GUI Testing via Image Embedding and Reinforcement Learning Bitmap-Based Security Monitoring for Deeply Embedded Systems Harmonising Contributions: Exploring Diversity in Software Engineering through CQA Mining on Stack Overflow An Empirical Study on the Characteristics of Database Access Bugs in Java Applications Self-planning Code Generation with Large Language Models
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1