I ‘m trying to apply the document understanding product from UiPath on around 6 different suppliers’ invoices, using the below steps
- Created Data Labeling session and labeled all my train data invoices
- Used this dataset to retrain the out of the box model
- Then created ML Skill out of it to be used in the workflow
- then I made a validation station and validated, edited the extracted data
- I used the train extractor scope activity and added the Machine Learning Extractor Trainer
Then the workflow created folder with the below content
- Documents Folder
- Meta Folder
I uploaded them to a new dataset then used it for retraining the pipeline has failed with
alternatively I added them to the same dataset which is previously generated from Data Labeling session and retrain them and the process done successfully.
But when I tried to test it again using the invoices which was previously has extraction issues the issues still appeared …! which means the model hasn’t been enhanced using the validated data.
How Can I fix this…?