Eland

-------------------

Fixed
^^^^^

* Fixed direct usage of TransformerModel (`619`_)

.. _619: https://github.com/elastic/eland/pull/619

-------------------

Added
^^^^^

* Published pre-built Docker images to docker.elastic.co/eland/eland (`613`_)
* Allowed importing private HuggingFace models (`608`_)
* Added Apple Silicon (arm64) support to Docker image (`615`_)
* Allowed importing some DPR models like ance-dpr-context-multi (`573`_)
* Allowed using the Pandas API without monitoring/main permissions (`581`_)

Changed
^^^^^^^

* Updated Docker image to Debian 12 Bookworm (`613`_)
* Reduced Docker image size by not installing unused PyTorch GPU support on amd64 (`615`_)
* Reduced model chunk size to 1MB (`605`_)

Fixed
^^^^^

* Fixed deprecations in preparation of Pandas 2.0 support (`593`_, `596`_, contributed by `bartbroere`_)

.. _bartbroere: https://github.com/bartbroere
.. _613: https://github.com/elastic/eland/pull/613
.. _608: https://github.com/elastic/eland/pull/608
.. _615: https://github.com/elastic/eland/pull/615
.. _573: https://github.com/elastic/eland/pull/573
.. _581: https://github.com/elastic/eland/pull/581
.. _605: https://github.com/elastic/eland/pull/605
.. _593: https://github.com/elastic/eland/pull/593
.. _596: https://github.com/elastic/eland/pull/596

------------------

Added
^^^^^

* Simplify embedding model support and loading (`569`_)
* Make eland_import_hub_model easier to find on Windows (`559`_)
* Update trained model inference endpoint (`556`_)
* Add BertJapaneseTokenizer support with bert_ja tokenization configuration (`534`_)
* Add ability to upload xlm-roberta tokenized models (`518`_)
* Tolerate different model output formats when measuring embedding size (`535`_)
* Generate valid NLP model id from file path (`541`_)
* Upgrade torch to 1.13.1 and check the cluster version before uploading a NLP model (`522`_)
* Set embedding_size config parameter for Text Embedding models (`532`_)
* Add support for the pass_through task (`526`_)

Fixed
^^^^^

* Fixed black to comply with the code style (`557`_)
* Fixed No module named 'torch' (`553`_)
* Fix autosummary directive by removing hack autosummaries (`548`_)
* Prevent TypeError with None check (`525`_)

.. _518: https://github.com/elastic/eland/pull/518
.. _522: https://github.com/elastic/eland/pull/522
.. _525: https://github.com/elastic/eland/pull/525
.. _526: https://github.com/elastic/eland/pull/526
.. _532: https://github.com/elastic/eland/pull/532
.. _534: https://github.com/elastic/eland/pull/534
.. _535: https://github.com/elastic/eland/pull/535
.. _541: https://github.com/elastic/eland/pull/541
.. _548: https://github.com/elastic/eland/pull/548
.. _553: https://github.com/elastic/eland/pull/553
.. _556: https://github.com/elastic/eland/pull/556
.. _557: https://github.com/elastic/eland/pull/557
.. _559: https://github.com/elastic/eland/pull/559
.. _569: https://github.com/elastic/eland/pull/569

------------------

Added
^^^^^

* Added a new NLP model task type "text_similarity" (`486`_)
* Added a new NLP model task type "text_expansion" (`520`_)
* Added support for exporting an Elastic ML model as a scikit-learn pipeline via ``MLModel.export_model()`` (`509`_)

Fixed
^^^^^

* Fixed an issue that occurred when LightGBM was installed but libomp wasn't installed on the system. (`499`_)

.. _486: https://github.com/elastic/eland/pull/486
.. _499: https://github.com/elastic/eland/pull/499
.. _509: https://github.com/elastic/eland/pull/509
.. _520: https://github.com/elastic/eland/pull/520

------------------

Added
^^^^^

* Added a new NLP model task type "auto" which infers the task type based on model configuration and architecture (`475`_)

Changed
^^^^^^^

* Changed required version of 'torch' package to `>=1.11.0,<1.12` to match required PyTorch version for Elasticsearch 8.3 (was `>=1.9.0,<2`) (`479`_)
* Changed the default value of the `--task-type` parameter for the `eland_import_hub_model` CLI to be "auto" (`475`_)

Fixed
^^^^^

* Fixed decision tree classifier serialization to account for probabilities (`465`_)
* Fixed PyTorch model quantization (`472`_)

.. _465: https://github.com/elastic/eland/pull/465
.. _472: https://github.com/elastic/eland/pull/472
.. _475: https://github.com/elastic/eland/pull/475
.. _479: https://github.com/elastic/eland/pull/479

------------------

Added
^^^^^

* Added support for passing Cloud ID via ``--cloud-id`` to ``eland_import_hub_model`` CLI tool (`462`_)
* Added support for authenticating via ``--es-username``, ``--es-password``, and ``--es-api-key`` to the ``eland_import_hub_model`` CLI tool (`461`_)
* Added support for XGBoost 1.6 (`458`_)
* Added support for ``question_answering`` NLP tasks (`457`_)

.. _457: https://github.com/elastic/eland/pull/457
.. _458: https://github.com/elastic/eland/pull/458
.. _461: https://github.com/elastic/eland/pull/461
.. _462: https://github.com/elastic/eland/pull/462