Release v0.39.0
What's Changed
New Features 🎉
- feat: update observation publisher to use newer sdk by @khorshuheng in #509
- feat: Support model observability to pyfunc model by @tiopramayudi in #502
- Add run_pyfunc_model() to build and run pyfunc model locally by @ariefrahmansyah in #514
- feat: Model schemas API and migrate client to OpenAPI 3 by @tiopramayudi in #518
- feat(api): Introduce CPU and memory limit request factors to API server by @deadlycoconuts in #519
- Propagate image building and model deployment errors by @ariefrahmansyah in #524
- feat: Add support for multiple sinks by @khorshuheng in #513
Other Changes
- Miscellaneous Doc updates by @krithika369 in #510
- fix: Set grpc as the container name when raw deployment mode is used by @krithika369 in #511
- Update User and Maintainer Docs by @krithika369 in #512
- Adding back SUMMARY.md file by @eric-lidong in #520
- fix: move git initialization to function body by @khorshuheng in #521
- fix: make sure transformer property on version endpoint is not undefined by @khorshuheng in #523
- fix: Incorrect request body for create and patch version by @tiopramayudi in #522
- fix: Reduce memory usage when publishing prediction log to kafka by @tiopramayudi in #525
- fix(api): Fix failing recovery of deployment attempts interrupted by API server pod termination by @deadlycoconuts in #517
- fix: Retraining model for PyFunc E2E test by @tiopramayudi in #527
- Add troubleshooting deployment errors page by @ariefrahmansyah in #530
- [sdk] Set experiment in model_version run by @leonlnj in #532
- preload pyfunc v2 model to avoid pydantic schema generation error by @khorshuheng in #535
- fix(cicd): Reduce memory footprint of e2e tests by @deadlycoconuts in #534
Full Changelog: v0.38.1...v0.39.0