Merge branch 'main' into patch-2

.github/workflows/_docker-build.yml

@@ -1,4 +1,4 @@
-name: Reusable docker server image build workflow
+name: Reusable docker image build workflow
 
 on:
   workflow_call:
@@ -35,7 +35,7 @@ permissions:
 # based on https://docs.docker.com/build/ci/github-actions/multi-platform/#distribute-build-across-multiple-runners
 jobs:
   build:
-    name: Build server image
+    name: Build image
     runs-on: ubuntu-22.04
     timeout-minutes: 60
     outputs:
@@ -98,7 +98,7 @@ jobs:
           touch "/tmp/digests/${digest#sha256:}"
 
       - name: Upload digest
-        uses: actions/upload-artifact@c7d193f32edcb7bfad88892161225aeda64e9392 # v4.0.0
+        uses: actions/upload-artifact@1eb3cb2b3e0f29609092a73eb033bb759a334595 # v4.1.0
         with:
           name: digests-${{ steps.build-id.outputs.id }}-${{ matrix.platform.name }}
           path: /tmp/digests/*
@@ -114,7 +114,7 @@ jobs:
       metadata: ${{ steps.meta.outputs.json }}
     steps:
       - name: Download digests
-        uses: actions/download-artifact@f44cd7b40bfd40b6aa1cc1b9b5b7bf03d3c67110 # v4.1.0
+        uses: actions/download-artifact@6b208ae046db98c579e8a3aa621ab581ff575935 # v4.1.1
         with:
           pattern: digests-${{ needs.build.outputs.build-id }}-*
           path: /tmp/digests

.github/workflows/docker-base.yml

@@ -39,7 +39,7 @@ jobs:
           echo "ubuntu-version=${{ env.DEFAULT_UBUNTU }}" >> "$GITHUB_OUTPUT"
 
   build-base-images:
-    name: Build images
+    name: Build base images
     uses: ./.github/workflows/_docker-build.yml
     needs: parameters
     strategy:

.github/workflows/docker-client.yml

@@ -0,0 +1,36 @@
+name: Build docker client image
+
+on:
+  workflow_dispatch:
+    inputs:
+      flwr-version:
+        description: "Version of Flower e.g. (1.6.0)."
+        required: true
+        type: string
+
+permissions:
+  contents: read
+
+jobs:
+  build-client-images:
+    name: Build client images
+    uses: ./.github/workflows/_docker-build.yml
+    # run only on default branch when using it with workflow_dispatch
+    if: github.ref_name == github.event.repository.default_branch
+    strategy:
+      fail-fast: false
+      matrix:
+        python-version: ["3.8", "3.9", "3.10", "3.11"]
+    with:
+      namespace-repository: flwr/client
+      file-dir: src/docker/client
+      build-args: |
+        FLWR_VERSION=${{ github.event.inputs.flwr-version }}
+        BASE_IMAGE_TAG=py${{ matrix.python-version }}-ubuntu22.04
+      tags: |
+        ${{ github.event.inputs.flwr-version }}-py${{ matrix.python-version }}-ubuntu22.04
+        ${{ github.event.inputs.flwr-version }}
+        latest
+    secrets:
+      dockerhub-user: ${{ secrets.DOCKERHUB_USERNAME }}
+      dockerhub-token: ${{ secrets.DOCKERHUB_TOKEN }}

baselines/hfedxgboost/README.md

@@ -11,7 +11,7 @@ dataset: [a9a, cod-rna, ijcnn1, space_ga, cpusmall, YearPredictionMSD]
 
 **Paper:** [arxiv.org/abs/2304.07537](https://arxiv.org/abs/2304.07537)
 
-**Authors:** Chenyang Ma, Xinchi Qiu, Daniel J. Beutel, Nicholas D. Laneearly_stop_patience_rounds: 100
+**Authors:** Chenyang Ma, Xinchi Qiu, Daniel J. Beutel, Nicholas D. Lane
 
 **Abstract:** The privacy-sensitive nature of decentralized datasets and the robustness of eXtreme Gradient Boosting (XGBoost) on tabular data raise the need to train XGBoost in the context of federated learning (FL). Existing works on federated XGBoost in the horizontal setting rely on the sharing of gradients, which induce per-node level communication frequency and serious privacy concerns. To alleviate these problems, we develop an innovative framework for horizontal federated XGBoost which does not depend on the sharing of gradients and simultaneously boosts privacy and communication efficiency by making the learning rates of the aggregated tree ensembles are learnable. We conduct extensive evaluations on various classification and regression datasets, showing our approach achieve performance comparable to the state-of-the-art method and effectively improves communication efficiency by lowering both communication rounds and communication overhead by factors ranging from 25x to 700x.
 

datasets/e2e/tensorflow/pyproject.toml

@@ -9,7 +9,7 @@ description = "Flower Datasets with TensorFlow"
 authors = ["The Flower Authors <[email protected]>"]
 
 [tool.poetry.dependencies]
-python = "^3.8"
+python = ">=3.8,<3.11"
 flwr-datasets = { path = "./../../", extras = ["vision"] }
 tensorflow-cpu = "^2.9.1, !=2.11.1"
 parameterized = "==0.9.0"

doc/source/how-to-install-flower.rst

@@ -11,6 +11,9 @@ Flower requires at least `Python 3.8 <https://docs.python.org/3.8/>`_, but `Pyth
 Install stable release
 ----------------------
 
+Using pip
+~~~~~~~~~
+
 Stable releases are available on `PyPI <https://pypi.org/project/flwr/>`_::
 
   python -m pip install flwr
@@ -20,6 +23,25 @@ For simulations that use the Virtual Client Engine, ``flwr`` should be installed
   python -m pip install flwr[simulation]
 
 
+Using conda (or mamba)
+~~~~~~~~~~~~~~~~~~~~~~
+
+Flower can also be installed from the ``conda-forge`` channel.
+
+If you have not added ``conda-forge`` to your channels, you will first need to run the following::
+
+  conda config --add channels conda-forge
+  conda config --set channel_priority strict
+
+Once the ``conda-forge`` channel has been enabled, ``flwr`` can be installed with ``conda``::
+
+  conda install flwr
+
+or with ``mamba``::
+
+  mamba install flwr
+
+
 Verify installation
 -------------------
 

doc/source/tutorial-series-get-started-with-flower-pytorch.ipynb

@@ -484,7 +484,7 @@
     "    min_available_clients=10,  # Wait until all 10 clients are available\n",
     ")\n",
     "\n",
-    "# Specify the resources each of your clients need. By default, each \n",
+    "# Specify the resources each of your clients need. By default, each\n",
     "# client will be allocated 1x CPU and 0x CPUs\n",
     "client_resources = {\"num_cpus\": 1, \"num_gpus\": 0.0}\n",
     "if DEVICE.type == \"cuda\":\n",

examples/android/README.md

@@ -54,4 +54,4 @@ poetry run ./run.sh
 
 Download and install the `flwr_android_client.apk` on each Android device/emulator. The server currently expects a minimum of 4 Android clients, but it can be changed in the `server.py`.
 
-When the Android app runs, add the client ID (between 1-10), the IP and port of your server, and press `Load Dataset`. This will load the local CIFAR10 dataset in memory. Then press `Setup Connection Channel` which will establish connection with the server. Finally, press `Train Federated!` which will start the federated training.
+When the Android app runs, add the client ID (between 1-10), the IP and port of your server, and press `Start`. This will load the local CIFAR10 dataset in memory, establish connection with the server, and start the federated training. To abort the federated learning process, press `Stop`. You can clear and refresh the log messages by pressing `Clear` and `Refresh` buttons respectively.

examples/flower-simulation-step-by-step-pytorch/Part-I/main.py

@@ -24,7 +24,7 @@ def main(cfg: DictConfig):
     save_path = HydraConfig.get().runtime.output_dir
 
     ## 2. Prepare your dataset
-    # When simulating FL workloads we have a lot of freedom on how the FL clients behave,
+    # When simulating FL runs we have a lot of freedom on how the FL clients behave,
     # what data they have, how much data, etc. This is not possible in real FL settings.
     # In simulation you'd often encounter two types of dataset:
     #       * naturally partitioned, that come pre-partitioned by user id (e.g. FEMNIST,
@@ -91,7 +91,7 @@ def main(cfg: DictConfig):
             "num_gpus": 0.0,
         },  # (optional) controls the degree of parallelism of your simulation.
         # Lower resources per client allow for more clients to run concurrently
-        # (but need to be set taking into account the compute/memory footprint of your workload)
+        # (but need to be set taking into account the compute/memory footprint of your run)
         # `num_cpus` is an absolute number (integer) indicating the number of threads a client should be allocated
         # `num_gpus` is a ratio indicating the portion of gpu memory that a client needs.
     )

examples/mt-pytorch/driver.py

@@ -54,13 +54,13 @@ def weighted_average(metrics: List[Tuple[int, Metrics]]) -> Metrics:
 
 # -------------------------------------------------------------------------- Driver SDK
 driver.connect()
-create_workload_res: driver_pb2.CreateWorkloadResponse = driver.create_workload(
-    req=driver_pb2.CreateWorkloadRequest()
+create_run_res: driver_pb2.CreateRunResponse = driver.create_run(
+    req=driver_pb2.CreateRunRequest()
 )
 # -------------------------------------------------------------------------- Driver SDK
 
-workload_id = create_workload_res.workload_id
-print(f"Created workload id {workload_id}")
+run_id = create_run_res.run_id
+print(f"Created run id {run_id}")
 
 history = History()
 for server_round in range(num_rounds):
@@ -93,7 +93,7 @@ def weighted_average(metrics: List[Tuple[int, Metrics]]) -> Metrics:
         # loop and wait until enough client nodes are available.
         while True:
             # Get a list of node ID's from the server
-            get_nodes_req = driver_pb2.GetNodesRequest(workload_id=workload_id)
+            get_nodes_req = driver_pb2.GetNodesRequest(run_id=run_id)
 
             # ---------------------------------------------------------------------- Driver SDK
             get_nodes_res: driver_pb2.GetNodesResponse = driver.get_nodes(
@@ -125,7 +125,7 @@ def weighted_average(metrics: List[Tuple[int, Metrics]]) -> Metrics:
         new_task_ins = task_pb2.TaskIns(
             task_id="",  # Do not set, will be created and set by the DriverAPI
             group_id="",
-            workload_id=workload_id,
+            run_id=run_id,
             task=task_pb2.Task(
                 producer=node_pb2.Node(
                     node_id=0,

examples/pytorch-from-centralized-to-federated/cifar.py

@@ -73,10 +73,10 @@ def apply_transforms(batch):
 
 
 def train(
-        net: Net,
-        trainloader: torch.utils.data.DataLoader,
-        epochs: int,
-        device: torch.device,  # pylint: disable=no-member
+    net: Net,
+    trainloader: torch.utils.data.DataLoader,
+    epochs: int,
+    device: torch.device,  # pylint: disable=no-member
 ) -> None:
     """Train the network."""
     # Define loss and optimizer
@@ -110,9 +110,9 @@ def train(
 
 
 def test(
-        net: Net,
-        testloader: torch.utils.data.DataLoader,
-        device: torch.device,  # pylint: disable=no-member
+    net: Net,
+    testloader: torch.utils.data.DataLoader,
+    device: torch.device,  # pylint: disable=no-member
 ) -> Tuple[float, float]:
     """Validate the network on the entire test set."""
     # Define loss and metrics

examples/pytorch-from-centralized-to-federated/client.py

@@ -24,10 +24,10 @@ class CifarClient(fl.client.NumPyClient):
     """Flower client implementing CIFAR-10 image classification using PyTorch."""
 
     def __init__(
-            self,
-            model: cifar.Net,
-            trainloader: DataLoader,
-            testloader: DataLoader,
+        self,
+        model: cifar.Net,
+        trainloader: DataLoader,
+        testloader: DataLoader,
     ) -> None:
         self.model = model
         self.trainloader = trainloader
@@ -61,15 +61,15 @@ def set_parameters(self, parameters: List[np.ndarray]) -> None:
             self.model.load_state_dict(state_dict, strict=True)
 
     def fit(
-            self, parameters: List[np.ndarray], config: Dict[str, str]
+        self, parameters: List[np.ndarray], config: Dict[str, str]
     ) -> Tuple[List[np.ndarray], int, Dict]:
         # Set model parameters, train model, return updated model parameters
         self.set_parameters(parameters)
         cifar.train(self.model, self.trainloader, epochs=1, device=DEVICE)
         return self.get_parameters(config={}), len(self.trainloader.dataset), {}
 
     def evaluate(
-            self, parameters: List[np.ndarray], config: Dict[str, str]
+        self, parameters: List[np.ndarray], config: Dict[str, str]
     ) -> Tuple[float, int, Dict]:
         # Set model parameters, evaluate model on local test dataset, return result
         self.set_parameters(parameters)

examples/quickstart-pytorch-lightning/client.py

@@ -10,6 +10,7 @@
 
 disable_progress_bar()
 
+
 class FlowerClient(fl.client.NumPyClient):
     def __init__(self, model, train_loader, val_loader, test_loader):
         self.model = model
@@ -55,7 +56,6 @@ def _set_parameters(model, parameters):
 
 
 def main() -> None:
-
     parser = argparse.ArgumentParser(description="Flower")
     parser.add_argument(
         "--node-id",

examples/quickstart-pytorch-lightning/mnist.py

@@ -86,16 +86,20 @@ def load_data(partition):
     # 60 % for the federated train and 20 % for the federated validation (both in fit)
     partition_train_valid = partition_full["train"].train_test_split(train_size=0.75)
     trainloader = DataLoader(
-        partition_train_valid["train"], batch_size=32,
-        shuffle=True, collate_fn=collate_fn, num_workers=1
+        partition_train_valid["train"],
+        batch_size=32,
+        shuffle=True,
+        collate_fn=collate_fn,
+        num_workers=1,
     )
     valloader = DataLoader(
-        partition_train_valid["test"], batch_size=32,
-        collate_fn=collate_fn, num_workers=1
+        partition_train_valid["test"],
+        batch_size=32,
+        collate_fn=collate_fn,
+        num_workers=1,
     )
     testloader = DataLoader(
-        partition_full["test"], batch_size=32,
-        collate_fn=collate_fn, num_workers=1
+        partition_full["test"], batch_size=32, collate_fn=collate_fn, num_workers=1
     )
     return trainloader, valloader, testloader
 

examples/quickstart-sklearn-tabular/client.py

@@ -68,4 +68,6 @@ def evaluate(self, parameters, config):  # type: ignore
             return loss, len(X_test), {"test_accuracy": accuracy}
 
     # Start Flower client
-    fl.client.start_client(server_address="0.0.0.0:8080", client=IrisClient().to_client())
+    fl.client.start_client(
+        server_address="0.0.0.0:8080", client=IrisClient().to_client()
+    )

examples/secaggplus-mt/driver.py

@@ -23,7 +23,8 @@ def merge(_task: task_pb2.Task, _merge_task: task_pb2.Task) -> task_pb2.Task:
         task_pb2.TaskIns(
             task_id="",  # Do not set, will be created and set by the DriverAPI
             group_id="",
-            workload_id=workload_id,
+            run_id=run_id,
+            run_id=run_id,
             task=merge(
                 task,
                 task_pb2.Task(
@@ -84,13 +85,13 @@ def weighted_average(metrics: List[Tuple[int, Metrics]]) -> Metrics:
 
 # -------------------------------------------------------------------------- Driver SDK
 driver.connect()
-create_workload_res: driver_pb2.CreateWorkloadResponse = driver.create_workload(
-    req=driver_pb2.CreateWorkloadRequest()
+create_run_res: driver_pb2.CreateRunResponse = driver.create_run(
+    req=driver_pb2.CreateRunRequest()
 )
 # -------------------------------------------------------------------------- Driver SDK
 
-workload_id = create_workload_res.workload_id
-print(f"Created workload id {workload_id}")
+run_id = create_run_res.run_id
+print(f"Created run id {run_id}")
 
 history = History()
 for server_round in range(num_rounds):
@@ -119,7 +120,7 @@ def weighted_average(metrics: List[Tuple[int, Metrics]]) -> Metrics:
         # loop and wait until enough client nodes are available.
         while True:
             # Get a list of node ID's from the server
-            get_nodes_req = driver_pb2.GetNodesRequest(workload_id=workload_id)
+            get_nodes_req = driver_pb2.GetNodesRequest(run_id=run_id)
 
             # ---------------------------------------------------------------------- Driver SDK
             get_nodes_res: driver_pb2.GetNodesResponse = driver.get_nodes(

pyproject.toml

@@ -82,7 +82,7 @@ rest = ["requests", "starlette", "uvicorn"]
 types-dataclasses = "==0.6.6"
 types-protobuf = "==3.19.18"
 types-requests = "==2.31.0.10"
-types-setuptools = "==68.2.0.0"
+types-setuptools = "==69.0.0.20240115"
 clang-format = "==17.0.4"
 isort = "==5.12.0"
 black = { version = "==23.10.1", extras = ["jupyter"] }

src/docker/client/Dockerfile

@@ -0,0 +1,8 @@
+# Copyright 2023 Flower Labs GmbH. All Rights Reserved.
+
+ARG BASE_REPOSITORY=flwr/base
+ARG BASE_IMAGE_TAG
+FROM $BASE_REPOSITORY:$BASE_IMAGE_TAG
+
+ARG FLWR_VERSION
+RUN python -m pip install -U --no-cache-dir flwr[rest]==${FLWR_VERSION}

src/proto/flwr/proto/driver.proto

@@ -21,8 +21,8 @@ import "flwr/proto/node.proto";
 import "flwr/proto/task.proto";
 
 service Driver {
-  // Request workload_id
-  rpc CreateWorkload(CreateWorkloadRequest) returns (CreateWorkloadResponse) {}
+  // Request run_id
+  rpc CreateRun(CreateRunRequest) returns (CreateRunResponse) {}
 
   // Return a set of nodes
   rpc GetNodes(GetNodesRequest) returns (GetNodesResponse) {}
@@ -34,12 +34,12 @@ service Driver {
   rpc PullTaskRes(PullTaskResRequest) returns (PullTaskResResponse) {}
 }
 
-// CreateWorkload
-message CreateWorkloadRequest {}
-message CreateWorkloadResponse { sint64 workload_id = 1; }
+// CreateRun
+message CreateRunRequest {}
+message CreateRunResponse { sint64 run_id = 1; }
 
 // GetNodes messages
-message GetNodesRequest { sint64 workload_id = 1; }
+message GetNodesRequest { sint64 run_id = 1; }
 message GetNodesResponse { repeated Node nodes = 1; }
 
 // PushTaskIns messages