Hello friends, Good day
Wanted to know how the ML model trained in the back ground
for example i have two type of invoice
Want to extract po numbers from the invoice
in one invoice it is Your Reference:15246-1
and other is customer Po number : 23577-17
so whether the bot is going to find the number based on the text Po number or whether it is searching based on location.
How ml model is actually trained.
Hi
That’s a good question
Training a model happens in many ways
First would be proper labelling and tagging with Validation station if something couldn’t be identified especially the unlabelled data to identify patterns
Another would be leveraging pre-trained models and fine-tuning them for a specific task
So in short the training happens right from extraction, classification and finally with decision-making in validation station
I found this doc so interesting and informative
Hope this would add more insights
Cheers @aslam_ali1
Hi @aslam_ali1 ,
Model is Trained/ when trained with Different documents with Different formats and Templates, it is able to understand that the Field requires this particular value.
So a Combination of both based on Location and the Value is taken into consideration. It compares it with Different documents trained, Checks the value and also the location and assigns weightage to it accordingly. In this way, Model is learning/re-learning everytime the Model is trained.
Thank you @Palaniyappan it helped me a lot.
Thank you @supermanPunch .
This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.