- Release Notes
April 2023
New Features and Improvements
You now have a Field Name and a Field ID for both column and regular fields created in a Document Understanding project. The Field Name represents the display name of the field, meaning that you can use uppercases, whitespace, and any type of characters. The Field ID is automatically generated based on the Field Name. The Field Name is meant to ease your work, while the Field ID is used for training with pretrained models.
A new option is now available under the Dataset Diagnostic menu, the Calculator Tab. Use the Calculator Tab to modify the information about the created document type. You can update any of the following fields: Out-of-the-box document type, Number of languages, or Number of layouts. Changes done on the Calculator Tab are influencing the size and accuracy of the entire Dataset, meaning that more labelled training data may be required.
The Search options available inside a Document Manager session have been redesigned, coming up with a fresh look and a cleaner way of searching/filtering the documents.
The ML Packages with v23.4, or higher, now have the option to train using Frozen Backbone. This new approach trains faster and gives better results for small or low diversity training sets below 400 pages. You can override this behavior by using the new Training Pipeline environment variables documented in the official documentation.
Released in Endpoints + DocumentClassifier ML Packages | v23.4.0
We've added new document types to the DocumentClassifier ML Package, made general improvements, and fixed some small bugs.
Released in Endpoints + DocumentUnderstanding + Data Extraction ML Packages | v23.4.0
A list of seven new Out-of-the-box pre-trained ML Packages is now available for general usage. Here's the list of the seven new models:
- Certificate of incorporation/Good Standing
- Certificate of Origin
- Children Product Certificate
- CMS1500
- EU Declaration of Conformity
- Invoices Shipping
- Pay slips
