- Support Population Based Training #1382
- Support ASHA
- Support Auto Model Compression #460
- Support Auto Feature Engineering #475
- Various CRDs for HP, NAS and other AutoML techniques.
- Migrate to the new Katib UI Project 1
- Hyperparameter importances visualization with fANOVA algorithm
- Finish AWS CI/CD migration
- Support various parameter distribution #1207
- Finish validation for Algorithms #1126
- Refactor Hyperband #1389
- Support multiple CRD version with conversion webhook
- MLMD integration with Katib Experiments
- Support Early Stopping #692
- Support Advanced NAS Algorithms like DARTs, ProxylessNAS #461
- Delete Suggestion deployment after Experiment is finished #1061
- Save Suggestion state after deployment is deleted #1062
- Reconsider the design of Trial Template #906
- Design an extensible model for integrating with custom resources.
- Add validation for algorithms (a.k.a suggestions) #1126
- Katib UI fixes and enhancements
- Investigate Kubeflow Metadata integration
- Investigate the alignment with concept and implementation of "experiments" and "jobs/runs" in KFP #4955