How to get better embeddings with code pre-trained models? An empirical study

TL;DR

This empirical study evaluates various code pre-trained models for software engineering tasks, revealing that token aggregation methods and model architecture significantly impact embedding quality and semantic richness.

Contribution

The paper systematically compares five code PTMs across different architectures and tasks, providing insights into effective embedding strategies for SE applications.

Findings

Token-based embeddings do not fully capture code semantics.

Combining code and text data as in pre-training yields poor embeddings.

Decoder-only PTMs can produce high-quality, semantically rich code embeddings.

Abstract

Pre-trained language models have demonstrated powerful capabilities in the field of natural language processing (NLP). Recently, code pre-trained model (PTM), which draw from the experiences of the NLP field, have also achieved state-of-the-art results in many software engineering (SE) downstream tasks. These code PTMs take into account the differences between programming languages and natural languages during pre-training and make adjustments to pre-training tasks and input data. However, researchers in the SE community still inherit habits from the NLP field when using these code PTMs to generate embeddings for SE downstream classification tasks, such as generating semantic embeddings for code snippets through special tokens and inputting code and text information in the same way as pre-training the PTMs. In this paper, we empirically study five different PTMs (i.e. CodeBERT, CodeT5, PLBART, CodeGPT and CodeGen) with three different architectures (i.e. encoder-only, decoder-only and encoder-decoder) on four SE downstream classification tasks (i.e. code vulnerability detection, code clone detection, just-in-time defect prediction and function docstring mismatch detection) with respect to the two aforementioned aspects. Our experimental results indicate that (1) regardless of the architecture of the code PTMs used, embeddings obtained through special tokens do not sufficiently aggregate the semantic information of the entire code snippet; (2) the quality of code embeddings obtained by combing code data and text data in the same way as pre-training the PTMs is poor and cannot guarantee richer semantic information; (3) using the method that aggregates the vector representations of all code tokens, the decoder-only PTMs can obtain code embeddings with semantics as rich as or even better quality than those obtained from the encoder-only and encoder-decoder PTMs.