title: How to perform inference with LineVul on custom files?

[standard-librarian]

Description

I would like to use LineVul to test vulnerabilities in my own source code files. I need step-by-step instructions on how to set up and run inference with LineVul on custom files.

Current Understanding

• I am aware that LineVul is a deep learning-based vulnerability detection tool
• I want to use it for analyzing my own codebase rather than the pre-existing test sets

What I'm Looking For

Clear instructions on how to:

• Set up LineVul for inference mode
• Prepare my source code files in the correct format
• Run the inference process
• Interpret the results

Information about:

• Required file formats/structure
• Any preprocessing steps needed
• Configuration parameters that need to be set
• Expected output format

Questions

• What is the correct format for input files?
• Are there any size limitations for the files being analyzed?
• How should the model weights be loaded for inference?
• What is the recommended way to process multiple files?
• How can I interpret the confidence scores in the results?

Environment Details

My current environment:

• Operating System: Windows 10
• GPU: NVIDIA RTX 3050
• Python version: 3.11