3 ms·
We used TensorFlow Serving running in a docker container that is a rest endpoint for predictions. Then the backend can be agnostic to Serving. We personally use
by cercatrova 7y ago
We used TensorFlow Serving running in a docker container that is a rest endpoint for predictions. Then the backend can be agnostic to Serving. We personally used Node to query the container for whatever we needed.