Model Efficiency

Hi,

We have created an project for invoices and deployed it to production. We download the data tagged by users in Action Center and retrain the model using Training Pipelines every week.
Is there a was or report we can use to check if the model efficiency is increasing after training the model with the tagged data from users.

I need this report to show to business whether the model is able to predict better over time and training.

@nanmishra

Training will give you output files which would contain details about the model as well…please check the same

Cheers

Thanks a lot@Anil_G . You mean the evaluation_invoices.xlsx which is generated after the pipeline is complete?
One more thing is that every time I run the pipeline the score is different. Sometimes it is 0.9561, next time I suppose it should be higher but it shows 0.92.

Is there any documentation available around this?

@nanmishra

It need not be higher it depends on how it assumes its accuracy…

Sometimes when you train with new different files accuracy might go down as well a little and as you train more similar templates it would increase

Cheers

1 Like

Thanks for the information.

1 Like

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.