- Document Understanding Release Notes
- ML Packages and Public Endpoints Release Notes
Document Understanding Release Notes
May 2024
What's new
- Deteriorating performance when asking more than 10 questions.
- Variations in responses to a particular question whenever a new question is added or an existing one is modified.
Improvements
For improved consistency, we updated the icon for the Upload button from the Build section.
Generative Extractor and Classifier in Discovery APIs
We are excited to announce that Generative Extractor and Generative Classifier are now returned in Discovery APIs. You can use this to process pages using Generative Classifier or Generative Extractor in APIs.
For more information, check the Use the Discovery APIs section from the Document Understanding APIs guide.
Boost extraction confidence when extracting Document Data by APIs
boostExtractionConfidence
, is now available. This
parameter is used to boost the extraction confidence when extracting Document Data
with APIs.
For more information, check the Use the Extraction Service section from the Document Understanding APIs user guide.
Bug fixes
- Fixed an issue causing a column field value to freeze in Document Manager sessions. This occasionally happened due to manual edits made to extracted field values.
- Fixed an issue in Action Center, ensuring that the Is Empty business rule is now displayed correctly.
What's new
- Deteriorating performance when asking more than 10 questions.
- Variations in responses to a particular question whenever a new question is added or an existing one is modified.
Bug fixes
We fixed a bug in Document Manager for classic projects which was causing differences between the score of the training pipeline and the performance obtained in a running automation. This issue affects in particular low diversity datasets (with under 10 layouts) in both Automation Cloud™ and Automation Suite. You'll notice the issue most clearly when a document processed in a robot workflow ends up with poorer results compared to those from the Excel evaluation file in the AI Center pipeline artifacts folder.
This fix will also be included in the June Cumulative Update for Automation Suite version 2023.10.
To benefit from this fix, re-export your dataset from the Document Manager session. Afterwards, simply proceed with the usual steps for training, testing, and deploying your model.
- About Pipelines
- Export datasets using Document Manager
- One Click Extraction to train documents
- Deploy ML Skills using AI Center
Erratum - added 23 May 2024: Added more information on specific bug behavior and links to related information.
Generative Extractor and Classifier in Discovery APIs
We are excited to announce that Generative Extractor and Generative Classifier are now returned in Discovery APIs. You can use this to process pages using Generative Classifier or Generative Extractor in APIs.
For more information, check the Use the Discovery APIs section from the Document Understanding APIs guide.
Boost extraction confidence when extracting Document Data by APIs
boostExtractionConfidence
, is now available. This
parameter is used to boost the extraction confidence when extracting Document Data
with APIs.
For more information, check the Use the Extraction Service section from the Document Understanding APIs user guide.