Roadmap
Next up
Features
- Improve large scale experience
- Result saving function (Samir)
- Priorband default sampling / pass evaluated configs to neps.run (Samir)
- Document large scale
- Evaluate and maybe improve ease-of-use of NePS for DDP (Gopalji)
- Optimize dependencies (Anton)
Documentation
- Remove templates (Danny)
- Add New Lightning example (Gopalji)
- Add DDP examples (Gopalji)
- Add some larger examples (Gopalji, Anton)
- Improve new lightning example by adding Tensorboard (Tarek)
- Add optimizer pages (Anton, Neeratyoy)
Before 1.0.0 version
Features
- Utility neps.clean to manage existing run results
- Generate pdf plot after each evaluation
- Finegrained control over user prior
- Print search space upon run
- Utility to generate code for best architecture
- Core algorithmic feature set (research)
Documentation
- Add example for spawning cloud instances via run pipeline
- Add NAS documentation
- Keep a changelog, add to it before each release
Tests
- Regression tests to run on cluster on each version release