14 November 2022 | LTS Release

Released in AI Center | DocumentUnderstanding + DocumentClassifier + Data Extraction ML Packages

Stay up to date with all the latest news regarding the ML Packages by going through the next list of changes that have occurred since the last LTS release until now.

What's New

There are 18 new Preview ML packages available with a more advanced model architecture for our Document Understanding Machine Learning Packages in AI Center. You can easily identify them by the Preview attached to the end of the package name, eg.: InvoicesPreview, PurchaseOrderPreview, Acord125Preview, etc.
We've updated the public endpoints list with all the new Preview ML packages and can be consulted here.
Worth mentioning is the fact that these preview models don't consume DU/AI units from your licensing entitlement.

Ten new models have become generally available. Here is the complete list of all the available Out-of-the-box Pre-trained ML Packages.

The ML Classification endpoint is now available in -Preview.

The OCR for Chinese, Japanese, Korean public endpoint has become generally available.

Bug Fixes

• Fixed a bug on private skills usage and now the private skill can be used only with an API key that belongs to the same organization that is using the AI Center instance.
• The item splitting was stabilized by combining the eol classifier and line_detection methods into a single method.
• Fixed a bug that was causing the extracted fields to be shown on the wrong page in Validation Station.
• Fixed a bug that was causing the last line of text on some pages to not be digitized in Document Manager.
• Fixed a bug that was preventing displaying some F1 score items from the evaluation_F1_invoices.txt file in Full/Evaluation pipelines in AI Center.
• Fixed a bug that was causing the wrong overall F1 score to be calculated in evaluation_F1_invoices.txt file in Full/Evaluation pipelines in AI Center whenever a model had only column fields.
• Fixed a bug occurring when running an evaluation pipeline on a model trained with the special line_detection mode, causing predictions to be different than when called from the ML skill.

Known Issues

If you upgrade an offline Automation Suite environment from v2021.10 to v2022.10, an error results when trying to use the Document OCR skill deployed in the v2021.10. A temporary fix is to re-deploy the Document OCR skill. Before deploying make sure you are using the offline bundles for the v2022.10.

29 November 2022

What's New & Improvements

An upcoming deprecation is announced for the Invoices Australia pre-trained ML package. We recommend using instead the Invoices ML Package. Here you can find more details about it.

8 December 2022

Bug Fixes

• An Out-Of-the-Box pre-trained model that was running on a full pipeline was evaluating the training dataset instead of the dataset from the evaluation directory. The bug was fixed and the evaluation process now runs as expected.