sfu-rsl/rust-mizan-leaderboard

RustMizan Leaderboard

A Gradio leaderboard for RustMizan, a benchmark that evaluates LLMs on real-world Rust memory safety vulnerabilities from CVEs and security advisories.

Models are scored on the full vulnerability analysis pipeline: detection, CWE classification, and function- and line-level localization. Results are reported across dataset variants (vanilla plus semantic-preserving mutations) to probe contamination and robustness.

Tabs

• —Leaderboard: Aggregate metrics per model and dataset variant, with filters for granularity, vulnerability type, and columns.
• —Sample-wise Comparison: Per-CVE, per-granularity correctness across models.
• —Metrics / Variants / About: Definitions of the metrics, the mutation variants, and the task.

Run Locally

pip install -r requirements.txt
python app.py

The app reads pre-processed results from data/experiments/. To add your own, see CONTRIBUTING.md.

License

Licensed under the Apache License, Version 2.0. See LICENSE.