Details
-
Sub-task
-
Status: Resolved
-
P2
-
Resolution: Won't Fix
-
None
-
None
Description
The goal is to create a PTransform that integrates Google Cloud AI Platform Prediction functionality [1]. The transform should work both in batch and streaming mode.
One of the solutions could be using a tfx_bsl library and its `RunInference()` Beam transform[2] that can do inference by using a service endpoint.
[1] https://cloud.google.com/ai-platform/prediction/docs/
[2] https://github.com/tensorflow/tfx-bsl/blob/a9f5b6128309595570cc6212f8076e7a20063ac2/tfx_bsl/beam/run_inference.py#L78