Understanding Long Programming Languages with Structure-Aware Sparse Attention
This addresses a bottleneck for developers and researchers in code-related tasks like code search, where real-world code often exceeds length limits of existing models.
The paper tackles the problem of processing long code sequences in programming-based pre-trained language models by introducing SASA, a Structure-Aware Sparse Attention mechanism, which reduces computational complexity and improves performance on CodeXGLUE tasks compared to baselines.
Programming-based Pre-trained Language Models (PPLMs) such as CodeBERT have achieved great success in many downstream code-related tasks. Since the memory and computational complexity of self-attention in the Transformer grow quadratically with the sequence length, PPLMs typically limit the code length to 512. However, codes in real-world applications are generally long, such as code searches, which cannot be processed efficiently by existing PPLMs. To solve this problem, in this paper, we present SASA, a Structure-Aware Sparse Attention mechanism, which reduces the complexity and improves performance for long code understanding tasks. The key components in SASA are top-$k$ sparse attention and Abstract Syntax Tree (AST)-based structure-aware attention. With top-$k$ sparse attention, the most crucial attention relation can be obtained with a lower computational cost. As the code structure represents the logic of the code statements, which is a complement to the code sequence characteristics, we further introduce AST structures into attention. Extensive experiments on CodeXGLUE tasks show that SASA achieves better performance than the competing baselines.