VulnTrain

VulnTrain offers a suite of commands to generate diverse AI datasets and train models using comprehensive vulnerability data from Vulnerability-Lookup. It harnesses over one million JSON records from all supported advisory sources (CVE, GitHub advisories, CSAF, PySecDB, CNVD) to build high-quality, domain-specific models.

Additionally, data from the vulnerability-lookup:meta container, including enrichment sources such as vulnrichment and Fraunhofer FKIE, is incorporated to enhance model quality.

Check out the datasets and models on Hugging Face:

For more information about the use of AI in Vulnerability-Lookup, please refer to the user manual.

Installation

pipx install VulnTrain

For development:

git clone https://github.com/vulnerability-lookup/VulnTrain.git
cd VulnTrain/
poetry install

Usage

Three types of commands are available:

• Dataset generation: Create and prepare datasets from vulnerability sources.
• Model training: Train models using the prepared datasets.
• Model validation: Assess the performance of trained models (validations, benchmarks, etc.).

CLI commands

| Command | Purpose | |---------|---------| | vulntrain-dataset-generation | Generate datasets from vulnerability sources | | vulntrain-train-severity-classification | Train severity classifier (RoBERTa/DistilBERT) | | vulntrain-train-severity-cnvd-classification | Train severity classifier for CNVD data | | vulntrain-train-description-generation | Train GPT-2 vulnerability description generator | | vulntrain-train-cwe-classification | Train CWE classifier from patches | | vulntrain-validate-severity-classification | Validate severity model | | vulntrain-validate-text-generation | Validate text generation model |

Models

• Severity classification:
• Description generation:

Distributed training on HPC clusters

VulnTrain supports distributed multi-GPU training via SLURM, making it suitable for EuroHPC-style GPU clusters. See the HPC documentation for Conda environment setup, single-node and multi-node SLURM job scripts, and NCCL configuration.

Documentation

Check out the full documentation for detailed usage instructions, dataset generation examples, and training recipes.

How to cite

Bonhomme, C., & Dulaunoy, A. (2025). VLAI: A RoBERTa-Based Model for Automated Vulnerability Severity Classification (Version 1.4.0) [Computer software]. https://doi.org/10.48550/arXiv.2507.03607

@misc{bonhomme2025vlai,
    title={VLAI: A RoBERTa-Based Model for Automated Vulnerability Severity Classification},
    author={Cédric Bonhomme and Alexandre Dulaunoy},
    year={2025},
    eprint={2507.03607},
    archivePrefix={arXiv},
    primaryClass={cs.CR}
}

License

VulnTrain is licensed under GNU General Public License version 3

Copyright (c) 2025-2026 Computer Incident Response Center Luxembourg (CIRCL)
Copyright (C) 2025-2026 Cédric Bonhomme - https://github.com/cedricbonhomme
Copyright (C) 2025 Léa Ulusan - https://github.com/3LS3-1F