Deployment of tensorflow serving

[kevin]

Hi,

Is there any way to deploy just the preprocessing part without a trained model?

The reason I ask is because I am trying to minimize the services that I use to put a model in production.

I currently have a service that gathers data using FastApi, but I want to deploy a “model” using tensorflow-serving that only gathers data and once it has enough data, i would use the end-to-end TFX solution.

I don’t know if tensorflow-serving has a way to keep the data it used for inference or preprocessing

Thank you,

[Aniket_Dubey]

Hi @kevin ,

@kevin

I don’t know if tensorflow-serving has a way to keep the data it used for inference or preprocessing

We can deploy the preprocessing step by including it within the model inference pipeline, but we cannot deploy the preprocessing step alone as a standalone component .

Thank You .