Authors: David de-Fitero-Dominguez, Eva Garcia-Lopez, Antonio Garcia-Cabot, Jose-Javier Martinez-Herraiz

Published on: January 08, 2024

Impact Score: 8.22

Arxiv code: Arxiv:2401.03741

Summary

• What is new: The introduction of a novel format for code modification representation using advanced Large Language Models like Code Llama and Mistral.
• Why this is important: The challenge of automating the repair of code vulnerabilities to enhance digital security.
• What the research proposes: Fine-tuning advanced Large Language Models on datasets with C code vulnerabilities to improve the automated code repair process.
• Results: Models showed an enhanced repair accuracy over previous methods like VulRepair, and highlighted the importance of using test datasets not included in training for better effectiveness.

Technical Details

Technological frameworks used: Large Language Models (LLMs)

Models used: Code Llama, Mistral

Data used: Datasets featuring C code vulnerabilities

Potential Impact

Cybersecurity providers, digital security services, AI-based coding tools companies, software development firms

Want to implement this idea in a business?

We have generated a startup concept here: SecureCodeAI.