Help Center
Find answers, explore documentation, and get in touch with our team.
Frequently Asked Questions
Getting Started
How do I access the cyber dataset?
The dataset is available on Hugging Face Hub at spacetime-ml/expert-data. See our Getting Started guide for integration steps.
What formats does the dataset support?
All data is provided in JSON Lines format, making it easy to integrate into Python ML pipelines and evaluation frameworks.
How frequently is the dataset updated?
We maintain a 90-day rolling window of expert-curated examples, with new data added regularly as new security scenarios emerge.
Integration & Usage
Can I use the data for model fine-tuning?
Yes, the dataset is designed for both supervised fine-tuning and RLHF workflows. See our guides for best practices.
How do I cite SpacetimeML data?
Include proper attribution when publishing research using our datasets. Contact us for citation guidelines.
Is commercial use allowed?
Yes, all datasets are licensed for commercial use. Review the license details on Hugging Face.
Evaluation & Benchmarks
How are models evaluated on the dataset?
We provide comprehensive evaluation frameworks across 4 security domains: threat analysis, vulnerability assessment, incident response, and safety alignment.
What metrics should I use?
Common metrics include accuracy, precision, recall, F1-score for classification tasks, and expert ranking correlation for security decisions.
How can I contribute evaluations?
Contact our team to contribute expert evaluations and help expand the dataset with new security scenarios.
Get in Touch
Contact our team for support and inquiries.