Finetuning Large Language Models for Vulnerability Detection

TL;DR

This paper demonstrates that finetuning large language models like WizardCoder can significantly improve vulnerability detection in source code, especially when addressing class imbalance and optimizing training regimes.

Contribution

It introduces a method for finetuning WizardCoder for vulnerability detection, enhancing training efficiency and handling class imbalance, with improved detection performance.

Findings

Improved ROC AUC and F1 scores over baseline models

Enhanced training speed without performance loss

Effective handling of imbalanced datasets

Abstract

This paper presents the results of finetuning large language models (LLMs) for the task of detecting vulnerabilities in source code. We leverage WizardCoder, a recent improvement of the state-of-the-art LLM StarCoder, and adapt it for vulnerability detection through further finetuning. To accelerate training, we modify WizardCoder's training procedure, also we investigate optimal training regimes. For the imbalanced dataset with many more negative examples than positive, we also explore different techniques to improve classification performance. The finetuned WizardCoder model achieves improvement in ROC AUC and F1 measures on balanced and imbalanced vulnerability datasets over CodeBERT-like model, demonstrating the effectiveness of adapting pretrained LLMs for vulnerability detection in source code. The key contributions are finetuning the state-of-the-art code LLM, WizardCoder, increasing its training speed without the performance harm, optimizing the training procedure and regimes, handling class imbalance, and improving performance on difficult vulnerability detection datasets. This demonstrates the potential for transfer learning by finetuning large pretrained language models for specialized source code analysis tasks.