Taxonomy data extraction

above shows validation station from data extraction from 2 separate invoices one pick up the product details one does not?
anyway to correct that? eventually want to make do without validation station

Hey Ray,

This is more related to the extraction method you’re using. Are you using the ML Invoices model or a forms extractor?

The out of the box machine learning model is the best extraction method if you want to extract invoices. It will be consistent across all invoices.

Let me know if you’re using an ML model or another extractor and we can find the issue from there.

Hi @sharon.palawandram

i am using the ML invoice model

Thanks for confirming. This would mean your table columns need more training data. You can iterate another train pipeline with more train data. To validate if your table columns are captured correctly, you can run an evaluation pipeline.