Main features & enhancements
- Projects: Add simple run/build/deploy wrappers + local pipeline, @yaronha
- Frameworks: Sklearn and XGboost auto-logging, @yaronha
- Frameworks: Added custom objects support and HP tuning, @guy1992l
- Model Monitoring: Automate the deployment process, @Michaelliv
- Model Monitoring: Add project and global functions for setting the model monitoring credentials, @eyalsol
- Feature Store: Add stream source to ingestion service, @dinal
- Feature Store: Support multiple keys for spark engine, @katyakats
- Feature Store: Change cache flush interval to flush after every event, @dinal
- Artifacts: Added Bokeh artifact, @guy1992l
- Azure DataStore: Handle storage options as secrets, @hayesgb
- Azure DataStore: Handle upload strings vs bytes and filepath formation when using adlfs, @hayesgb
- Pipelines: Enrich the pipeline with DAG on get_pipeline endpoint & Plot it (graphviz) in the client, @yaronha
- Datastore: Improve support for Dask DataFrames, @hayesgb
- Datastore: Extend Azure blob to support other auth methods, @hayesgb
- API+SDK: Fix propagation of errors from server to client, @Hedingber
- GPU Images: Bump CUDA version, @Hedingber
- Images: Bump Tensorflow to 2.4.1 and Horovod to 0.22.1, @Hedingber
- Docs: Add secrets documentation, @theSaarco
- Docs: Add documentation for model monitoring, @jasonnIguazio
- Docs: Add section on node affinity, @jasonnIguazio
- Runtimes: Support configuring auto mount type, @theSaarco
- Runtimes: Support pods priority, @AlonMaor14
- Runtimes: Support configuring default node selector, @Hedingber
- MPIJob: Change hyper-param loop to reuse the same CRD/containers, @yaronha
- Serving: Try to load json by default, @yaronha
- Serving: Add feature enrichment model/ensemble routers, @yaronha
- Sparkjob: Support deploying default image, @benbd86
- Nuclio: Support node selection, @AlonMaor14
- API: Add default images per kind and function image defaults in frontend spec, @Hedingber
- UI: Release notes