Exploring the Impact of the Output Format on the Evaluation of Large Language Models for Code Translation

Code translation between programming languages is a long-existing and critical task in software engineering, facilitating the modernization of legacy systems, ensuring cross-platform compatibility, and enhancing software performance. With the recent advances in large language models (LLMs) and their...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2024 IEEE/ACM First International Conference on AI Foundation Models and Software Engineering (Forge) Conference Acronym s. 57 - 68
Hlavní autoři: Macedo, Marcos, Tian, Yuan, Cogo, Filipe R., Adams, Bram
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: ACM 14.04.2024
Témata:
Accuracy
Benchmark testing
benchmarking
case study
code translation
Codes
Computational modeling
empirical study
evaluation
large language model
Large language models
LLM
output format
soft-ware engineering
Source coding
Terminology
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Code translation between programming languages is a long-existing and critical task in software engineering, facilitating the modernization of legacy systems, ensuring cross-platform compatibility, and enhancing software performance. With the recent advances in large language models (LLMs) and their applications to code translation, there is an increasing need for comprehensive evaluation of these models. In this study, we empirically analyze the generated outputs of eleven popular instruct-tuned LLMs with parameters ranging from 1B up to 46.7B on 3,820 translation pairs across five languages, including C, C++, Go, Java, and Python. Our analysis found that between 26.4% and 73.7% of code translations produced by our evaluated LLMs necessitate post-processing, as these translations often include a mix of code, quotes, and text rather than being purely source code. Overlooking the output format of these models can inadvertently lead to underestimation of their actual performance. This is particularly evident when evaluating them with execution-based metrics such as Computational Accuracy (CA). Our results demonstrate that a strategic combination of prompt engineering and regular expression can effectively extract the source code from the model generation output. In particular, our method can help eleven selected models achieve an average Code Extraction Success Rate (CSR) of 92.73%. Our findings shed light on and motivate future research to conduct more reliable benchmarks of LLMs for code translation.
DOI:10.1145/3650105.3652301