Tensorflow Implementation with More than 2 clients

This question was migrated from Github Discussions.

Original Question:
Hello,
I followed the Tensorflow Introduction at Federated Learning in less than 20 lines of code.
Is it possible to have more than 2 clients sampled? When I tried running more than 2 instances of client.py, in my server.py terminal I still see only 2 clients (out of 3). Do I need to do anything additional to get the server to sample 3 out of 3 clients?

Answer:
Hi, there are two things you need to do to train on more clients at once:

  1. Start more instances of client.py
  2. Configure the strategy to select more clients (this will happen by default since Flower 1.0)

The Flower tutorial covers this in more detail: Starting the training