When is a Workflow's custom AI model sufficiently trained?

Accuracy score indicators in the Training module help you determine whether a document category within a Workflow requires additional training.

There are two types of accuracy scores:

• Categorization accuracy score: tells you how confident your Workflow's custom AI model is in its auto-categorization ability in its current training state.
• Labelling accuracy score: tells you how confident your Workflow's custom AI model is in the correctness of each AI-generated label on a document within a specific document category. 

We recommend you conclude your training after:

1. A Workflow's assisted categorization accuracy score is high.
2. You have labelled all training documents within that Workflow. 
3. All AI-generated labels within that Workflow's document categories are correct and hold a high accuracy score indicating a green color coating.
   
   

Need to revisit a Workflow's Training module? How to improve your AI model