Skip to content

Latest commit

 

History

History
14 lines (10 loc) · 760 Bytes

automl.md

File metadata and controls

14 lines (10 loc) · 760 Bytes
Raw

AutoML

The GCP AutoML service allows us to train custom models and then run predictions against them. The Node-RED AutoML node allows us to execute a prediction request. It encapsulates the predict API documented here. On execution the msg.payload is expected to have child nodes of:

  • msg.payload.payload
  • msg.payload.params (optional)

corresponding to the payload and params data requested by the underlying GCP API. The response is found in msg.payload.

The configuration of the node requires:

  • Project - The project containing the model.
  • Location - The region containing the model.
  • Model ID - The API key for the model (usually TCN…..).