TensorFlow 2.0 Beta is available Learn more

High-performance simulations with TFF

This tutorial will describe how to setup high-performance simulations with TFF in a variety of common scenarios.

NOTE: The mechanisms covered here are not included in the latest release, have not been tested yet, and the API may evolve.

TODO(b/134543154): Populate the content, some of the things to cover here: - tff.framework.set_default_executor() - tff.framework.create_local_executor() - single-machine setup with GPU - multi-machine setup on GCP/GKE with TPU - interfacing MapReduce - current limitations, and when/how they will be relaxed

Before we begin

We're not ready to begin quite yet.

Single-machine simulations

Coming soon.