driftflow Documentation Index
This directory contains the maintained documentation for driftflow, a typed Python library for DVS-driven adaptive sampling in graph-diffusion workflows.
Quick Navigation
| Document | What you will find |
|---|---|
| Mathematical background | Mathematical preliminaries, equations, notation, and derivations |
| Architecture | System architecture, data flow, component boundaries |
| API reference | Complete API reference for all public modules, classes, and functions |
| Usage guide | Step-by-step usage examples and integration patterns |
| Developer guide | How to run tests, lint, extend the codebase, and contribute |
| Engineering principles | The design and reliability rules behind the library |
| Deployment | Production deployment, Docker, monitoring, and troubleshooting |
| Production inference | Checkpoint loading, integrity, rollout gates, and operations |
| Extension guide | Optional enhancements and pluggable components |
| v1 to v2 migration | Upgrade notes for the current API |
For First-Time Users
- Start with Usage guide for a quick working example.
- Read Architecture to understand how the pieces fit together.
- Refer to API reference when writing your own integration code.
For Researchers Reproducing the Paper
- Read Mathematical background for the equation-by-equation mapping.
- Check Architecture for the algorithm-to-code mapping.
- Check API reference for the implementation surface.
For Contributors
- See Developer guide for the development workflow.
- Review CONTRIBUTING.md for style guidelines.
- Read Engineering principles before extending the core.
- Consult Extension guide for ideas on modular enhancements.