An extensive study on pre-trained models for program understanding and generation

Zeng，Zhengran1,6; Tan，Hanzhuo2; Zhang，Haotian3; Li，Jing4; Zhang，Yuqun1; Zhang，Lingming5

10.1145/3533767.3534390

2022-07-18

ISSTA 2022 - Proceedings of the 31st ACM SIGSOFT International Symposium on Software Testing and Analysis

39-51

Automatic program understanding and generation techniques could significantly advance the productivity of programmers and have been widely studied by academia and industry. Recently, the advent of pre-trained paradigm enlightens researchers to develop general-purpose pre-trained models which can be applied for a broad range of program understanding and generation tasks. Such pre-trained models, derived by self-supervised objectives on large unlabelled corpora, can be fine-tuned in downstream tasks (such as code search and code generation) with minimal adaptations. Although these pre-trained models claim superiority over the prior techniques, they seldom follow equivalent evaluation protocols, e.g., they are hardly evaluated on the identical benchmarks, tasks, or settings. Consequently, there is a pressing need for a comprehensive study of the pre-trained models on their effectiveness, versatility as well as the limitations to provide implications and guidance for the future development in this area. To this end, we first perform an extensive study of eight open-access pre-trained models over a large benchmark on seven representative code tasks to assess their reproducibility. We further compare the pre-trained models and domain-specific state-of-the-art techniques for validating pre-trained effectiveness. At last, we investigate the robustness of the pre-trained models by inspecting their performance variations under adversarial attacks. Through the study, we find that while we can in general replicate the original performance of the pre-trained models on their evaluated tasks and adopted benchmarks, subtle performance fluctuations can refute the findings in their original papers. Moreover, none of the existing pre-trained models can dominate over all other models. We also find that the pre-trained models can significantly outperform non-pre-trained state-of-the-art techniques in program understanding tasks. Furthermore, we perform the first study for natural language-programming language pre-trained model robustness via adversarial attacks and find that a simple random attack approach can easily fool the state-of-the-art pre-trained models and thus incur security issues. At last, we also provide multiple practical guidelines for advancing future research on pre-trained models for program understanding and generation.

Adversarial Attack Code Representation Deep Learning Pre-Trained Language Models

第一 ; 通讯

英语

EI

20223512667335

Deep learning ; Natural language processing systems

Ergonomics and Human Factors Engineering:461.4 ; Data Processing and Image Processing:723.2

2-s2.0-85134872006

Scopus

1.Southern University of Science and Technology,China
2.Southern University of Science and Technology,Hong Kong Polytechnic University,China
3.Kwai,China
4.Hong Kong Polytechnic University,China
5.University of Illinois at Urbana-Champaign,United States
6.Research Institute of Trustworthy Autonomous Systems,Shenzhen,China

Zeng，Zhengran,Tan，Hanzhuo,Zhang，Haotian,et al. An extensive study on pre-trained models for program understanding and generation[C],2022:39-51.