Frequently Asked Questions
What is Cubyc?
Cubyc is a free and open-source experiment tracking platform for data science. It tracks and manages your experiment runs by capturing:
- Inputs: Data, parameters, and dependencies
- Code: Syntax and structure
- Outputs: Metrics and artifacts
How does Cubyc work?
Cubyc makes experiment tracking effortless:
- Use Cubyc's Python library to track your scripts and functions
- Cubyc saves the inputs, code, and outputs of your runs to your Git repository
- Query your Git-stored runs with SQL to filter, aggregate, and analyze them
Why should I track my experiments?
Experiment tracking is crucial for:
- Reproducibility: Accurately replicate results
- Debugging: Quickly identify and fix issues
- Benchmarking: Systematically compare models
- Collaboration: Seamlessly share experiment data with your team
- Compliance: Ensure adherence to regulatory and auditing standards
How is Cubyc unique?
Cubyc is lightweight, flexible, and secure:
- Data Science Focused: Cubyc is designed for all data science, not just AI
- Automatic Tracking: Say goodbye to manual configurations and boilerplate code
- Secure Storage: Keep your data safe and accessible in your preferred Git system
- SQL Querying: Analyze results with SQL and build custom reports
- Git Collaboration: Easily share and collaborate on experiments with your team
| Cubyc | Others* | |
|---|---|---|
| Applications | Data Science + AI | AI |
| Configurations | Automatic | Manual |
| Storage | Git | Server |
| Results | SQL | Web UI |
| Collaboration | Git | Limited |
*Other platforms include Weights & Biases, Comet.ml, Neptune.ai, and MLflow.