Release v0.10.1
This release brings several new features around file-based input / output data handling and experimental support for logging metrics / artifacts to MLFlow or disk.
For more information on how to use the new logging capabilities please refer to the Tesseract Core documentation.
What’s Changed
Features
- Add --input-path to tesseract cli and TESSERACT_INPUT_PATH to runtime (#249)
- Introduce logging metrics, parameters and artifacts to file or MLflow (#229)
- Add top-level description field to OAS (#268)
Bug Fixes
- Ensure tracebacks are always propagated through Python client (#228)
- Ensure default workdir is writable (#263)
- Tesseract-runtime default io paths (#266)
- Add volume statements to dockerfile to ensure logs / metrics / data are always written to a volume (#270)
- Add tests for MLflow backend using MLflow’s capabillity to write to file (#271)
- Bug in volume error handling (#272)
Refactor
- Use RuntimeConfig for --input-path/–output-path (#264)
Documentation
- Fix docstring on metrics example (#269)
Full diff: Comparing v0.10.0...v0.10.1 · pasteurlabs/tesseract-core · GitHub