To properly display this page you need a browser with JavaScript support.
The NER option on the ML Tools panel.
This option is only available in jobs with the architecture "ForTokenClassification" which is the one for data extraction.
In this example, a very basic one, we have a job where we want to fine-tune a model for extracting date on invoices.
This option is were we are going to create annotations for creating the data format used for the model/ architecture that we are using for training.
Lets have a quick view of the provided functions:
Reset annotations Button to remove all existing annotations on the current document.
E-Tags suggestions If we click this button the grid below will be populated with e-tags suggestions for helping us annotating the document from ChronoScan fed data.
ML Suggestions Similar to the E-tags suggestions, but in this case the suggestions are taken from previous pre-trained models that have been used to infer from.
No suggestions Click here if you don't want to list any of the previous explained suggestions.
Create new annotations Click here to create manual annotations.
Example of annotating from E-tags suggestions:
Example of annotating from ML Suggestions:
Example of annotating from manual annotations:
