Add fastai E2E test

.github/workflows/e2e.yml

@@ -253,6 +253,7 @@ jobs:
           cd e2e/scikit-learn
           python simulation.py
 
+
   opacus:
     runs-on: ubuntu-22.04
     timeout-minutes: 10
@@ -289,3 +290,41 @@ jobs:
         run: |
           cd e2e/opacus
           python simulation.py
+
+
+  fastai:
+    runs-on: ubuntu-22.04
+    timeout-minutes: 10
+    steps:
+      - uses: actions/checkout@v3
+      - name: Set up Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: 3.8.17
+      - name: Install
+        run: |
+          python -m pip install -U pip==22.3.1
+          python -m pip install -U setuptools==65.6.3
+          python -m pip install poetry==1.3.2
+          poetry config virtualenvs.create false
+      - name: Install dependencies
+        run: |
+          cd e2e/fastai
+          python -m poetry install
+      - name: Cache Datasets
+        uses: actions/cache@v2
+        with:
+          path: "~/.fastai"
+          key: fastai-datasets
+      - name: Download Datasets
+        run: |
+          cd e2e/fastai
+          python -c "from fastai.vision.all import *; untar_data(URLs.MNIST)"
+      - name: Run edge client test
+        run: |
+          cd e2e/fastai
+          ./test.sh
+      - name: Run virtual client test
+        run: |
+          cd e2e/fastai
+          python simulation.py

e2e/fastai/README.md

@@ -0,0 +1,5 @@
+# Flower with FastAI testing
+
+This directory is used for testing Flower with FastAI by using a simple MNIST recognition task.
+
+It uses the `FedAvg` strategy.

e2e/fastai/client.py

@@ -0,0 +1,57 @@
+import warnings
+from collections import OrderedDict
+
+import flwr as fl
+import numpy as np
+import torch
+from fastai.vision.all import *
+
+
+warnings.filterwarnings("ignore", category=UserWarning)
+
+# Download MNIST dataset
+path = untar_data(URLs.MNIST)
+
+# Load dataset
+dls = ImageDataLoaders.from_folder(
+    path, valid_pct=0.5, train="training", valid="testing", num_workers=0
+)
+
+subset_size = 100 # Or whatever
+selected_train = np.random.choice(dls.train_ds.items, subset_size, replace=False)
+selected_valid = np.random.choice(dls.valid_ds.items, subset_size, replace=False)
+# Swap in the subset for the whole thing (Note: this mutates dls, so re-initialize before full training!)
+dls.train = dls.test_dl(selected_train, with_labels=True)
+dls.valid = dls.test_dl(selected_valid, with_labels=True)
+
+# Define model
+learn = vision_learner(dls, squeezenet1_1, metrics=error_rate)
+
+
+# Define Flower client
+class FlowerClient(fl.client.NumPyClient):
+    def get_parameters(self, config):
+        return [val.cpu().numpy() for _, val in learn.model.state_dict().items()]
+
+    def set_parameters(self, parameters):
+        params_dict = zip(learn.model.state_dict().keys(), parameters)
+        state_dict = OrderedDict({k: torch.tensor(v) for k, v in params_dict})
+        learn.model.load_state_dict(state_dict, strict=True)
+
+    def fit(self, parameters, config):
+        self.set_parameters(parameters)
+        learn.fit(1)
+        return self.get_parameters(config={}), len(dls.train), {}
+
+    def evaluate(self, parameters, config):
+        self.set_parameters(parameters)
+        loss, error_rate = learn.validate()
+        return loss, len(dls.valid), {"accuracy": 1 - error_rate}
+
+
+if __name__ == "__main__":
+    # Start Flower client
+    fl.client.start_numpy_client(
+        server_address="127.0.0.1:8080",
+        client=FlowerClient(),
+    )

e2e/fastai/pyproject.toml

@@ -0,0 +1,14 @@
+[build-system]
+requires = ["poetry-core>=1.4.0"]
+build-backend = "poetry.core.masonry.api"
+
+[tool.poetry]
+name = "quickstart_fastai"
+version = "0.1.0"
+description = "Fastai Federated Learning E2E test with Flower"
+authors = ["The Flower Authors <hello@flower.dev>"]
+
+[tool.poetry.dependencies]
+python = ">=3.7,<3.10"
+flwr = { path = "../../", develop = true, extras = ["simulation"] }
+fastai = "^2.7.10"

e2e/fastai/server.py

@@ -0,0 +1,7 @@
+import flwr as fl
+
+hist = fl.server.start_server(
+    server_address="0.0.0.0:8080",
+    config=fl.server.ServerConfig(num_rounds=3),
+)
+assert (hist.losses_distributed[0][1] / hist.losses_distributed[-1][1]) >= 1

e2e/fastai/simulation.py

@@ -0,0 +1,14 @@
+import flwr as fl
+
+from client import FlowerClient
+
+def client_fn(cid):
+    _ = cid
+    return FlowerClient()
+
+hist = fl.simulation.start_simulation(
+    client_fn=client_fn,
+    num_clients=2,
+    config=fl.server.ServerConfig(num_rounds=3),
+)
+assert (hist.losses_distributed[0][1] / hist.losses_distributed[-1][1]) > 0.98

e2e/fastai/test.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+set -e
+cd "$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"/
+
+bash ../test.sh