Skip to content

Foundation Model

The foundation model is a self-supervised representation layer trained on your raw event and attribute data. It learns behavioral patterns across all available tables and time periods, producing embeddings — behavioral profiles of your entities — that power every downstream scenario.

In this section

Guide Description
Basic Configuration Set up the minimum viable foundation model definition
Managing Data Expand on data sources, selection, and transformations
Advanced Options Fine-tune hyperparameters, scaling, and performance
Training Execute training and verify outputs