arXiv 2408.04663
Dopamin: Transformer-based Comment Classifiers through Domain Post-Training and Multi-level Layer Aggregation
By Nam Le Hai and Nghi D. Q. Bui
Published 2024-08-06
Discussion
Read the public discussion and references gathered around this paper.
Code comments provide important information for understanding the source code. They can help developers understand the overall purpose of a function or class, as well as identify bugs and technical debt. However, an overabundance of comments is meaningless and counterproductive. As a result, it is critical to automatically filter out these comments for specific purposes. In this paper, we present Dopamin, a Transfor…