Our first run: MNIST

We present how to run an example script for APPFL. We first make sure that the dependencies are installed and change the directory to examples directory.

$ git clone https://github.com/APPFL/APPFL.git
$ cd APPFL
$ pip install -e ".[examples]"
$ cd examples

The first example can be simply run by this:

$ python mnist.py

If we want to run it in parallel using MPI (assuming that all clients are trained in the same cluster), we can run the same example as

$ mpiexec -np 5 python mnist.py

We can also simulate PPFL with gRPC.

$ mpiexec -np 5 python grpc_mnist.py

Note

mpiexec may need to specify additional argument to use CUDA: --mca opal_cuda_support 1

Our first run is training MNIST in a federated learning setting with the default configuration. Learn more about How to set configuration.