Reading Feature Data for Inference
Feature data can be read for inference using the methods in the following table.
|Using the HTTP API||This is the most common method used to read online feature data for inference. The HTTP API provides reads at low latency for single vector lookups.||Docs |
|Using the Java client||Use to read online feature data for inference. The client is a wrapper around the HTTP API and implements best practices for response deserialization and concurrent request handling.||Docs |
|Using Feature Output Streams||Feature View Output Streams enable your application to subscribe to the outputs of streaming feature pipelines. Feature View Output Streams are designed to be used for asynchronous predictions, where model inference is triggered by newly arriving feature data.||Docs|
|Using the Python SDK (Method 1)||Use to read batch feature data for offline inference.||Docs |
|Using the Python SDK (Method 2)||Use to read online feature data to test inferences. This method is not suitable for production use.||Docs |