Empirical Study of Transformers for Source Code
This work addresses the need for systematic comparisons of Transformer modifications for source code processing, which is incremental as it consolidates existing approaches rather than introducing new methods.
The study investigated how well Transformers can use syntactic information in source code across three tasks (code completion, function naming, bug fixing) by re-implementing modifications in a unified framework, showing they make meaningful predictions and identifying best practices for performance improvement.
Initially developed for natural language processing (NLP), Transformers are now widely used for source code processing, due to the format similarity between source code and text. In contrast to natural language, source code is strictly structured, i.e., it follows the syntax of the programming language. Several recent works develop Transformer modifications for capturing syntactic information in source code. The drawback of these works is that they do not compare to each other and consider different tasks. In this work, we conduct a thorough empirical study of the capabilities of Transformers to utilize syntactic information in different tasks. We consider three tasks (code completion, function naming and bug fixing) and re-implement different syntax-capturing modifications in a unified framework. We show that Transformers are able to make meaningful predictions based purely on syntactic information and underline the best practices of taking the syntactic information into account for improving the performance of the model.