Add ClearML SDK 1.10 release notes (#509)

This commit is contained in:
pollfly 2023-03-26 12:00:36 +03:00 committed by GitHub
parent 640bca6662
commit 547e25ce19
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -2,6 +2,51 @@
title: Version 1.10
---
### ClearML SDK 1.10.0
**Breaking Changes**
* `StorageManager.download_file()` and `StorageManager.download_folder()` from MinIO and Azure no longer include bucket name
in target local path [ClearML GitHub issue #709](https://github.com/allegroai/clearml/issues/709)
**New Features and Improvements**
* Add support for attaching plots and scalars to models
* Add `Task.get_debug_sample()` to retrieve task's debug samples [ClearML GitHub issue #761](https://github.com/allegroai/clearml/issues/761)
* Add custom artifact serialization for pipelines [ClearML GitHub issue #689](https://github.com/allegroai/clearml/issues/689)
* Add table layout configuration to `Logger.report_table()` through `extra_data` parameter [#796](https://github.com/allegroai/clearml/issues/796)
* Add `PipelineController.connect_configuration()` to add configuration dictionaries to a pipeline
* Add `Dataset.delete()` configurable hosted contents deletion support:
* Delete ClearML file server hosted files
* Delete external storage hosted files
* Add `force_download` argument to `Model.get_local_copy()` to force download a new copy of a model even if already in cache
* Add `PipelineController.is_successful()` criteria customization
* Add support for API v2.23
* Add `Getting Started` Notebooks [ClearML PR #948](https://github.com/allegroai/clearml/pull/948)
* Improve `hyper_parameter` example [ClearML PR #927](https://github.com/allegroai/clearml/pull/927)
* Correct `Task.connect()` docstring [ClearML Docs issue #473](https://github.com/allegroai/clearml-DOCS/issues/473)
* Improve the `Task.mark_completed()` and `Task.close()` docstrings [ClearML Docs issue #486](https://github.com/allegroai/clearml-docs/issues/486)
**Bug Fixes**
* Fix logging dependencies that use the subdirectory argument when pip installing from a git repo [ClearML GitHub issue #946](https://github.com/allegroai/clearml/issues/946)
* Fix `Task.import_offline_session()` does not import offline models [ClearML GitHub issue #653](https://github.com/allegroai/clearml/issues/653)
* Fix `clearml-init` incorrectly sets Web and API server ports [Clearml Server GitHub issue #181](https://github.com/allegroai/clearml-server/issues/181) & [ClearML GitHub issue #910](https://github.com/allegroai/clearml/issues/910)
* Fix multiple models trained by the same framework are not all automatically logged [ClearML GitHub issue #767](https://github.com/allegroai/clearml/issues/767)
* Fix parallel coordinates plot displays categorical variables unclearly [ClearML GitHub issue #907](https://github.com/allegroai/clearml/issues/907)
* Fix runtime toggling task offline mode in the context of an open task
* Fix information not being captured from Jupyter Notebook in Sagemaker studio
* Fix `python-fire` support
* Fix mishandling nested dictionaries in `Task.connect()` in `PipelineDecorator`
* Fix pipeline version number isn't updated when `auto_version_bump` is set to `True`
* Fix `get_or_create_project` crashes when run in parallel
* Fix task auto-publishing is not backwards compatible
* Fix `task.get_user_properties(value_only=True)` raises exception
* Fix `CLEARML_VCS_DIFF` set to empty string does not cause task to log empty uncommitted changes
* Fix `silence_errors` parameter in `StorageManager.get_file_size_bytes()` not working
* Fix setting `Dataset` metadata name to a string prefixed with `data_` breaks `Dataset.get()`
* Fix Pipeline fails creating tasks from multi-argument functions
### ClearML Server 1.10.0
**New Features and Improvements**