mlsanityData to 7 Oct 2026

Can we learn from developer mistakes? Learning to localize and repair real bugs from real bug fixes

arXiv 2207.00301 ↗First posted 1 Jul 2022cs.SE cs.LGcited by 4

Not on the map. Why?

Abstract (arXiv)

Real bug fixes found in open source repositories seem to be the perfect source for learning to localize and repair real bugs. However, the absence of large scale bug fix collections has made it difficult to effectively exploit real bug fixes in the training of larger neural models in the past. In contrast, artificial bugs -- produced by mutating existing source code -- can be easily obtained at a sufficient scale and are therefore often preferred in the training of existing approaches. Still, localization and repair models that are trained on artificial bugs usually underperform when faced with real bugs. This raises the question whether bug localization and repair models trained on real bug fixes are more effective in localizing and repairing real bugs. We address this question by introducing RealiT, a pre-train-and-fine-tune approach for effectively learning to localize and repair real bugs from real bug fixes. RealiT is first pre-trained on a large number of artificial bugs produced by traditional mutation operators and then fine-tuned on a smaller set of real bug fixes. Fine-tuning does not require any modifications of the learning algorithm and hence can be easily adopted in various training scenarios for bug localization or repair (even when real training data is scarce). In addition, we found that training on real bug fixes with RealiT is empirically powerful by nearly doubling the localization performance of an existing model on real bugs while maintaining or even improving the repair performance.

Key facts. Cited by 4 papers in the map (data through 7 Oct 2026). Builds on 25 of its 33 references that are in the map. First posted on arXiv 1 Jul 2022.

Builds on

25 of 33 references.

References · 25

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
  3. 2016Pointer Sentinel Mixture Modelscited by 2,992
  4. 2015Neural Machine Translation of Rare Words with Subword Unitscited by 2,898
  5. 2020CodeBERT: A Pre-Trained Model for Programming and Natural Languagescited by 1,286
  6. 2018Self-Attention with Relative Position Representationscited by 1,112
  7. 2019Big Transfer (BiT): General Visual Representation Learningcited by 766
  8. 2017Learning to Represent Programs with Graphscited by 219
  9. 2016Incorporating Copying Mechanism in Sequence-to-Sequence Learningcited by 217
  10. 2018SequenceR: Sequence-to-Sequence Learning for End-to-End Program Repaircited by 144
  11. 2018The Adverse Effects of Code Duplication in Machine Learning Models of Codecited by 123
  12. 2018An Empirical Study on Learning Bug-Fixing Patches in the Wild via Neural Machine Translationcited by 106
13 more
  1. 2020Graph-based, Self-Supervised Program Repair from Diagnostic Feedbackcited by 93
  2. 2021Break-It-Fix-It: Unsupervised Learning for Program Repaircited by 71
  3. 2018DeepBugs: A Learning Approach to Name-based Bug Detectioncited by 65
  4. 2021Self-Supervised Bug Detection and Repaircited by 59
  5. 2019Getafix: Learning to Fix Bugs Automaticallycited by 40
  6. 2019Neural Program Repair by Jointly Learning to Localize and Repaircited by 39
  7. 2019How Often Do Single-Statement Bugs Occur? The ManySStuBs4J Datasetcited by 35
  8. 2019Learning and Evaluating Contextual Embedding of Source Codecited by 33
  9. 2022On Distribution Shift in Learning-based Bug Detectorscited by 15
  10. 2022TSSB-3M: Mining single statement bugs at massive scalecited by 14
  11. 2018Learning How to Mutate Source Code from Bug-Fixescited by 13
  12. 2013Recurrent Convolutional Neural Networks for Discourse Compositionalitycited by 7
  13. 2021An Exploratory Study of Debugging Episodescited by 3

Built on by

Cited by year

Citing papers per year: 2023: 3; 2024: 1.

4 papers.

  1. 2023Large Language Models of Code Fail at Completing Code with Potential Bugscited by 28
  2. 2024Performance-Aligned LLMs for Generating Fast Codecited by 20
  3. 2023HPC-Coder: Modeling Parallel Programs using Large Language Modelscited by 11
  4. 2023RunBugRun -- An Executable Dataset for Automated Program Repaircited by 10