PolarDB:Procedure to build a custom model

When a built-in model doesn't meet your requirements — for example, when you need to optimize an algorithm or adjust a model's structure — PolarDB for AI lets you build and deploy a custom model. Custom models run in confidential containers, keeping your model and data secure.

Building a custom model involves three sequential steps:

1. Create a model

2. Evaluate a model

3. Perform model inference