Tackling the tedious and often arduous task of uncovering and repairing software bugs can be a time-intensive process for software engineers. To ease this burden, Microsoft Research has developed BugLab, an innovative deep learning model that is capable of finding and rectifying errors without requiring labeled data. This new model uses a “hide and seek” approach to train itself by understanding both code structure as well as any natural language cues embedded in comments, variable names, or other places within the codebase. For example, consider the following snippet which resolves an issue in a GitHub open-source project: BugLab does not employ GPT-3 technology in its operation. By utilizing BugLab’s self-supervised bug detection and repair capabilities, developers are able to save valuable time while producing better quality software faster than ever before.
You May Also Like.
Product Information 01
