Quickstart TensorFlow

In this tutorial we will learn how to train a Convolutional Neural Network on CIFAR-10 using the Flower framework and TensorFlow. First of all, it is recommended to create a virtual environment and run everything within a virtualenv.

Let’s use flwr new to create a complete Flower+TensorFlow project. It will generate all the files needed to run, by default with the Flower Simulation Engine, a federation of 10 nodes using FedAvg. The dataset will be partitioned using Flower Dataset’s IidPartitioner.

Now that we have a rough idea of what this example is about, let’s get started. First, install Flower in your new environment:

# In a new Python environment
$ pip install flwr

Then, run the command below. You will be prompted to select one of the available templates (choose TensorFlow), give a name to your project, and type in your developer name:

$ flwr new

After running it you’ll notice a new directory with your project name has been created. It should have the following structure:

<your-project-name>
├── <your-project-name>
│   ├── __init__.py
│   ├── client_app.py   # Defines your ClientApp
│   ├── server_app.py   # Defines your ServerApp
│   └── task.py         # Defines your model, training and data loading
├── pyproject.toml      # Project metadata like dependencies and configs
└── README.md

If you haven’t yet installed the project and its dependencies, you can do so by:

# From the directory where your pyproject.toml is
$ pip install -e .

To run the project, do:

# Run with default arguments
$ flwr run .

With default arguments you will see an output like this one:

Loading project configuration...
Success
INFO :      Starting FedAvg strategy:
INFO :          ├── Number of rounds: 3
INFO :          ├── ArrayRecord (0.16 MB)
INFO :          ├── ConfigRecord (train): (empty!)
INFO :          ├── ConfigRecord (evaluate): (empty!)
INFO :          ├──> Sampling:
INFO :          │       ├──Fraction: train (0.50) | evaluate ( 1.00)
INFO :          │       ├──Minimum nodes: train (2) | evaluate (2)
INFO :          │       └──Minimum available nodes: 2
INFO :          └──> Keys in records:
INFO :                  ├── Weighted by: 'num-examples'
INFO :                  ├── ArrayRecord key: 'arrays'
INFO :                  └── ConfigRecord key: 'config'
INFO :
INFO :
INFO :      [ROUND 1/3]
INFO :      configure_train: Sampled 5 nodes (out of 10)
INFO :      aggregate_train: Received 5 results and 0 failures
INFO :          └──> Aggregated MetricRecord: {'train_loss': 2.0013, 'train_acc': 0.2624}
INFO :      configure_evaluate: Sampled 10 nodes (out of 10)
INFO :      aggregate_evaluate: Received 10 results and 0 failures
INFO :          └──> Aggregated MetricRecord: {'eval_acc': 0.1216, 'eval_loss': 2.2686}
INFO :
INFO :      [ROUND 2/3]
INFO :      configure_train: Sampled 5 nodes (out of 10)
INFO :      aggregate_train: Received 5 results and 0 failures
INFO :          └──> Aggregated MetricRecord: {'train_loss': 1.8099, 'train_acc': 0.3373}
INFO :      configure_evaluate: Sampled 10 nodes (out of 10)
INFO :      aggregate_evaluate: Received 10 results and 0 failures
INFO :          └──> Aggregated MetricRecord: {'eval_acc': 0.4273, 'eval_loss': 1.6684}
INFO :
INFO :      [ROUND 3/3]
INFO :      configure_train: Sampled 5 nodes (out of 10)
INFO :      aggregate_train: Received 5 results and 0 failures
INFO :          └──> Aggregated MetricRecord: {'train_loss': 1.6749, 'train_acc': 0.3965}
INFO :      configure_evaluate: Sampled 10 nodes (out of 10)
INFO :      aggregate_evaluate: Received 10 results and 0 failures
INFO :          └──> Aggregated MetricRecord: {'eval_acc': 0.4281, 'eval_loss': 1.5807}
INFO :
INFO :      Strategy execution finished in 16.60s
INFO :
INFO :      Final results:
INFO :
INFO :          Global Arrays:
INFO :                  ArrayRecord (0.163 MB)
INFO :
INFO :          Aggregated ClientApp-side Train Metrics:
INFO :          { 1: {'train_acc': '2.6240e-01', 'train_loss': '2.0014e+00'},
INFO :            2: {'train_acc': '3.3725e-01', 'train_loss': '1.8099e+00'},
INFO :            3: {'train_acc': '3.9655e-01', 'train_loss': '1.6750e+00'}}
INFO :
INFO :          Aggregated ClientApp-side Evaluate Metrics:
INFO :          { 1: {'eval_acc': '1.2160e-01', 'eval_loss': '2.2686e+00'},
INFO :            2: {'eval_acc': '4.2730e-01', 'eval_loss': '1.6684e+00'},
INFO :            3: {'eval_acc': '4.2810e-01', 'eval_loss': '1.5807e+00'}}
INFO :
INFO :          ServerApp-side Evaluate Metrics:
INFO :          {}
INFO :
Saving final model to disk as final_model.keras...

You can also override the parameters defined in the [tool.flwr.app.config] section in pyproject.toml like this:

# Override some arguments
$ flwr run . --run-config "num-server-rounds=5 batch-size=16"

The Data

This tutorial uses Flower Datasets to easily download and partition the CIFAR-10 dataset. In this example you’ll make use of the IidPartitioner to generate num_partitions partitions. You can choose other partitioners available in Flower Datasets. Each ClientApp will call this function to create the NumPy arrays that correspond to their data partition.

partitioner = IidPartitioner(num_partitions=num_partitions)
fds = FederatedDataset(
    dataset="uoft-cs/cifar10",
    partitioners={"train": partitioner},
)
partition = fds.load_partition(partition_id, "train")
partition.set_format("numpy")

# Divide data on each node: 80% train, 20% test
partition = partition.train_test_split(test_size=0.2)
x_train, y_train = partition["train"]["img"] / 255.0, partition["train"]["label"]
x_test, y_test = partition["test"]["img"] / 255.0, partition["test"]["label"]

The Model

Next, we need a model. We defined a simple Convolutional Neural Network (CNN), but feel free to replace it with a more sophisticated model if you’d like:

def load_model(learning_rate: float = 0.001):
    # Define a simple CNN for CIFAR-10 and set Adam optimizer
    model = keras.Sequential(
        [
            keras.Input(shape=(32, 32, 3)),
            layers.Conv2D(32, kernel_size=(3, 3), activation="relu"),
            layers.MaxPooling2D(pool_size=(2, 2)),
            layers.Conv2D(64, kernel_size=(3, 3), activation="relu"),
            layers.MaxPooling2D(pool_size=(2, 2)),
            layers.Flatten(),
            layers.Dropout(0.5),
            layers.Dense(10, activation="softmax"),
        ]
    )
    model.compile(
        "adam",
        loss="sparse_categorical_crossentropy",
        metrics=["accuracy"],
    )
    return model

The ClientApp

The main changes we have to make to use Tensorflow with Flower have to do with converting the ArrayRecord received in the Message into numpy ndarrays for use with the built-in set_weights() function. After training, the get_weights() function can be used to extract then pack the updated numpy ndarrays into a Message from the ClientApp. We can make use of built-in methods in the ArrayRecord to make these conversions:

@app.train()
def train(msg: Message, context: Context):

    # Load the model
    model = load_model(context.run_config["learning-rate"])
    # Extract the ArrayRecord from Message and convert to numpy ndarrays
    model.set_weights(msg.content["arrays"].to_numpy_ndarrays())

    # Train the model
    ...

    # Pack the model weights into an ArrayRecord
    model_record = ArrayRecord(model.get_weights())

The rest of the functionality is directly inspired by the centralized case. The ClientApp comes with three core methods (train, evaluate, and query) that we can implement for different purposes. For example: train to train the received model using the local data; evaluate to assess its performance of the received model on a validation set; and query to retrieve information about the node executing the ClientApp. In this tutorial we will only make use of train and evaluate.

Let’s see how the train method can be implemented. It receives as input arguments a Message from the ServerApp. By default it carries:

• an ArrayRecord with the arrays of the model to federate. By default they can be retrieved with key "arrays" when accessing the message content.

• a ConfigRecord with the configuration sent from the ServerApp. By default it can be retrieved with key "config" when accessing the message content.

The train method also receives the Context, giving access to configs for your run and node. The run config hyperparameters are defined in the pyproject.toml of your Flower App. The node config can only be set when running Flower with the Deployment Runtime and is not directly configurable during simulations.

# Flower ClientApp
app = ClientApp()


@app.train()
def train(msg: Message, context: Context):
    """Train the model on local data."""

    # Reset local Tensorflow state
    keras.backend.clear_session()

    # Load the data
    partition_id = context.node_config["partition-id"]
    num_partitions = context.node_config["num-partitions"]
    x_train, y_train, _, _ = load_data(partition_id, num_partitions)

    # Load the model
    model = load_model(context.run_config["learning-rate"])
    model.set_weights(msg.content["arrays"].to_numpy_ndarrays())
    epochs = context.run_config["local-epochs"]
    batch_size = context.run_config["batch-size"]
    verbose = context.run_config.get("verbose")

    # Train the model
    history = model.fit(
        x_train,
        y_train,
        epochs=epochs,
        batch_size=batch_size,
        verbose=verbose,
    )

    # Get training metrics
    train_loss = history.history["loss"][-1] if "loss" in history.history else None
    train_acc = (
        history.history["accuracy"][-1] if "accuracy" in history.history else None
    )

    # Pack and send the model weights and metrics as a message
    model_record = ArrayRecord(model.get_weights())
    metrics = {"num-examples": len(x_train)}
    if train_loss is not None:
        metrics["train_loss"] = train_loss
    if train_acc is not None:
        metrics["train_acc"] = train_acc
    content = RecordDict({"arrays": model_record, "metrics": MetricRecord(metrics)})
    return Message(content=content, reply_to=msg)

The @app.evaluate() method would be near identical with two exceptions: (1) the model is not locally trained, instead it is used to evaluate its performance on the locally held-out validation set; (2) including the model in the reply Message is no longer needed because it is not locally modified.

The ServerApp

To construct a ServerApp we define its @app.main() method. This method receive as input arguments:

• a Grid object that will be used to interface with the nodes running the ClientApp to involve them in a round of train/evaluate/query or other.

• a Context object that provides access to the run configuration.

In this example we use the FedAvg and configure it with a specific value of fraction_train which is read from the run config. You can find the default value defined in the pyproject.toml. Then, the execution of the strategy is launched when invoking its start method. To it we pass:

• the Grid object.

• an ArrayRecord carrying a randomly initialized model that will serve as the global model to federated.

• the num_rounds parameter specifying how many rounds of FedAvg to perform.

# Create the ServerApp
app = ServerApp()


@app.main()
def main(grid: Grid, context: Context) -> None:
    """Main entry point for the ServerApp."""
    # Load config
    num_rounds = context.run_config["num-server-rounds"]
    fraction_train = context.run_config["fraction-train"]

    # Load initial model
    model = load_model()
    arrays = ArrayRecord(model.get_weights())

    # Define and start FedAvg strategy
    strategy = FedAvg(
        fraction_train=fraction_train,
    )

    result = strategy.start(
        grid=grid,
        initial_arrays=arrays,
        num_rounds=num_rounds,
    )

    # Save the final model
    ndarrays = result.arrays.to_numpy_ndarrays()
    final_model_name = "final_model.keras"
    print(f"Saving final model to disk as {final_model_name}...")
    model.set_weights(ndarrays)
    model.save(final_model_name)

Note the start method of the strategy returns a result object. This object contains all the relevant information about the FL process, including the final model weights as an ArrayRecord, and federated training and evaluation metrics as MetricRecords. You can easily log the metrics using Python’s pprint and save the final model weights using Tensorflow’s save() function.

Congratulations! You’ve successfully built and run your first federated learning system.

참고

Check the source code of the extended version of this tutorial in examples/quickstart-tensorflow in the Flower GitHub repository.