Abstract: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 pretrained 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.

A Comparison of Pretrained Models for Classifying Issue Reports

Automatic Issue Classifier: A Transfer Learning Framework for Classifying Issue Reports

Applying Large Language Models API to Issue Classification Problem

Automated classification for open-ended questions with BERT

How well do pre-trained contextual language representations recommend labels for GitHub issues?

Automated labelling of radiology reports using natural language processing: Comparison of traditional and newer methods

A Comparative Study of Transformer-based Neural Text Representation Techniques on Bug Triaging

Representation Learning for Stack Overflow Posts: How Far are We?

CatIss: An Intelligent Tool for Categorizing Issues Reports using Transformers

An Empirical Comparison of Pre-Trained Models of Source Code

RoBERTa: A Robustly Optimized BERT Pretraining Approach

BERT-Based Sentiment Analysis: A Software Engineering Perspective

Comparative analysis of real bugs in open-source Machine Learning projects -- A Registered Report

Comparison of Transfer Learning and Traditional Machine Learning Approach for Text Classification

On Calibration of Pre-trained Code Models

Evaluating text classification: A benchmark study

An Extensive Study on Pre-trained Models for Program Understanding and Generation

Emotion Classification In Software Engineering Texts: A Comparative Analysis of Pre-trained Transformers Language Models

Empirically Revisiting and Enhancing Automatic Classification of Bug and Non-Bug Issues

Comparing the Pretrained Models of Source Code by Re-pretraining under a Unified Setup.

Watch out for Version Mismtaching and Data Leakage! A Case Study of Their Influence in Bug Report Based Bug Localization Models