AI Models have undergone a major revamp and is now rechristened as AI Modeler that lets you build, train, and publish models to be used across your apps. To know more, check out our new help pages.
Prediction fields used in your applications work on an Artificial Intelligence (AI) based model that's capable of predicting field values for all future records. The AI Models page provides a consolidated view of the AI models used in all of your applications.
When a prediction field is deleted, its associated model will be deleted from the AI Models page.
View the AI models
Go to AI Models under Microservices.
The AI models of the prediction fields used in all your applications will be listed here. You can find the following information:
Application and Form name in which the prediction field is used.
Status of the model: Ready, Being trained, or Failed.
The date and time when the last time the model was trained.
Version: Whenever a model is retrained, a new version of the model is updated. You can choose to go back to the older version anytime.
Accuracy of the model's prediction.
Use the Filter dropdown to list only the models used in a particular application.
Click the record for a detailed view.
Re-train a model
You can retrain the prediction model to also consider the data added after the last time the model was trained. To do so:
Go to AI Models under Microservices.
Click on the horizontal ellipsis (three dots) icon adjacent to the required model, then click Re-train.
The model will begin to retrain and the status will be set to Being trained.
Once the training is complete, the status will be set to Ready and the prediction field will automatically start working with a newer updated version of the model. The version number and its accuracy will be reflected accordingly.
Note:
Alternatively, you can retrain your AI model from your app builder. Learn how.
When the model is being retrained, the previous working model will be used for prediction.
We recommend retraining the model when your data set has grown considerably beyond the last used train.
Change version
To make your prediction field work with a different version of the model:
Click AI Models under Microservices.
Click on the horizontal ellipsis (three dots) icon adjacent to the required model, then click Switch to version.
In the popup that appears, choose the version with which you want its corresponding prediction field to work.
The prediction field will immediately start working with the selected version and its accuracy will be reflected accordingly.