Refactor for students

Dockerfile

@@ -13,17 +13,20 @@ WORKDIR /opt/federation-lab
 
 # Update the Ubuntu software repository and fetch packages
 RUN apt-get update \
-    && apt-get install -y curl python3 python3-pip
+    && apt-get install -y python3.9
 
+# Setup pip3.9
+RUN apt install -y curl python3.9-distutils
+RUN curl https://bootstrap.pypa.io/get-pip.py -o get-pip.py
+RUN python3.9 get-pip.py
 # Add Pre-downloaded models (otherwise needs be run every-time)
 ADD data/ data/
 
 # Use cache for pip, otherwise we repeatedly pull from repository
-ADD requirements.txt ./
-RUN --mount=type=cache,target=/root/.cache/pip python3 -m pip install -r requirements.txt
+COPY requirements-cpu.txt ./requirements.txt
+RUN --mount=type=cache,target=/root/.cache/pip,mode=0777 python3.9 -m pip install -r requirements.txt
 
 # Add FLTK and configurations
 ADD fltk fltk
 ADD configs configs
 ADD experiments experiments
-ADD charts charts

fltk/__init__.py

@@ -1,2 +1 @@
-
 __version__ = '0.4.1'

fltk/core/comm/__init__.py

@@ -0,0 +1 @@
+from .rpc_util import *

fltk/core/rpc_util.py → fltk/core/comm/rpc_util.py

@@ -24,7 +24,14 @@ def _remote_method_async_by_info(method, worker_info, *args, **kwargs):
     args = [method, worker_info] + list(args)
     return rpc.rpc_async(worker_info, _call_method, args=args, kwargs=kwargs)
 
+
 def _remote_method_direct(method, other_node: str, *args, **kwargs):
+    """
+    Utility function for RPC communication between nodes.
+    :param method: A callable
+    :param other_node: reference to other node
+    :return: any
+    """
+
     args = [method, other_node] + list(args)
-    # return rpc.rpc_sync(other_node, _call_method, args=args, kwargs=kwargs)
     return rpc.rpc_sync(other_node, method, args=args, kwargs=kwargs)

fltk/core/distributed/client.py

@@ -29,7 +29,7 @@ def __init__(self, rank: int, task_id: str, world_size: int, config: Distributed
         @param task_id: String id representing the UID of the training task
         @type task_id: str
         @param config: Parsed configuration file representation to extract runtime information from.
-        @type config: BareConfig
+        @type config: DistributedConfig
         @param learning_params: Hyper-parameter configuration to be used during the training process by the learner.
         @type learning_params: LearningParameters
         """

fltk/core/node.py

@@ -1,3 +1,4 @@
+import abc
 import copy
 import os
 from typing import Callable, Any
@@ -12,18 +13,7 @@
 global_vars = {}
 
 
-def _remote_method_direct(method, other_node: str, *args, **kwargs):
-    """
-    Utility function for RPC communication between nodes
-    :param method: A callable
-    :param other_node: reference to other node
-    :return: any
-    """
-    args = [method, other_node] + list(args)
-    return rpc.rpc_sync(other_node, method, args=args, kwargs=kwargs)
-
-
-class Node:
+class Node(abc.ABC):
     """
     Implementation of any participating node.
     It handles communication and the basic functions for Deep Learning.

fltk/datasets/loader_util.py

@@ -1,5 +1,5 @@
 from fltk.datasets.distributed import DistMNISTDataset, DistFashionMNISTDataset, DistCIFAR100Dataset, DistCIFAR10Dataset
-from fltk.util.definitions import Dataset
+from fltk.util.config.definitions.dataset import Dataset
 
 
 def available_datasets():

fltk/launch.py

@@ -15,6 +15,7 @@
 from fltk.core.distributed import Orchestrator
 from fltk.core.distributed.extractor import download_datasets
 from fltk.core.federator import Federator
+from fltk.nets.util.reproducability import init_reproducibility
 from fltk.util.cluster.client import ClusterManager
 from fltk.util.config import DistributedConfig, Config
 from fltk.util.config.arguments import LearningParameters, extract_learning_parameters
@@ -41,7 +42,7 @@ def launch_distributed_client(task_id: str, conf: DistributedConfig = None,
     @param task_id: String representation (should be unique) corresponding to a client.
     @type task_id: str
     @param conf: Configuration for components, needed for spinning up components of the Orchestrator.
-    @type conf: BareConfig
+    @type conf: DistributedConfig
     @param learning_params: Parsed configuration of Hyper-Parameters for learning.
     @type: LearningParameters
     @return: None
@@ -71,7 +72,7 @@ def launch_orchestrator(args: Namespace = None, conf: DistributedConfig = None,
     @type args: Namespace
     @param config: Configuration for execution of Orchestrators components, needed for spinning up components of the
     Orchestrator.
-    @type config: BareConfig
+    @type config: Optional[DistributedConfig]
     @return: None
     @rtype: None
     """
@@ -86,7 +87,7 @@ def launch_orchestrator(args: Namespace = None, conf: DistributedConfig = None,
         logging.info("Pointing configuration to in cluster configuration.")
         conf.cluster_config.load_incluster_namespace()
         conf.cluster_config.load_incluster_image()
-    arrival_generator = DistributedExperimentGenerator() if simulate_arrivals else FederatedArrivalGenerator()
+    arrival_generator = (DistributedExperimentGenerator if simulate_arrivals else FederatedArrivalGenerator)(args.experiment)
     cluster_manager = ClusterManager()
 
     orchestrator = Orchestrator(cluster_manager, arrival_generator, conf)
@@ -114,7 +115,7 @@ def launch_extractor(base_path: Path, config_path: Path, args: Namespace = None,
     @param args: Arguments passed from CLI.
     @type args: Namespace
     @param conf: Parsed configuration file passed from the CLI.
-    @type conf: BareConfig
+    @type conf: Optional[DistributedConfig]
     @return: None
     @rtype: None
     """
@@ -295,5 +296,5 @@ def launch_cluster(arg_path, conf_path, args: Namespace = None, config: Distribu
                         datefmt='%m-%d %H:%M')
     # Set the seed for arrivals, torch seed is mostly ignored. Set the `arrival_seed` to a different value
     # for each repetition that you want to run an experiment with.
-    config.set_seed()
+    init_reproducibility(config.execution_config)
     launch_orchestrator(args=args, conf=config)

fltk/nets/__init__.py

@@ -2,7 +2,7 @@
 
 import torch
 
-from fltk.util.definitions import Nets
+from ..util.config.definitions.net import Nets
 from .cifar_100_resnet import Cifar100ResNet
 from .cifar_100_vgg import Cifar100VGG, vgg11_bn, vgg13_bn, vgg16_bn, vgg19_bn
 from .cifar_10_cnn import Cifar10CNN

fltk/nets/cifar_10_cnn.py

@@ -1,6 +1,6 @@
 # pylint: disable=missing-class-docstring,invalid-name
 import torch
-import torch.torch.nn.functional as F
+import torch.nn.functional as F
 
 class Cifar10CNN(torch.nn.Module):
 

fltk/nets/cifar_10_resnet.py

@@ -1,6 +1,6 @@
 # pylint: disable=missing-class-docstring,invalid-name
 import torch
-import torch.torch.nn.functional as F
+import torch.nn.functional as F
 
 
 class BasicBlock(torch.nn.Module):

fltk/nets/util/model.py

@@ -6,7 +6,7 @@
 import torch
 from torch.utils.tensorboard import SummaryWriter
 
-import fltk.util.config as config
+from fltk.util.config import DistributedConfig
 from fltk.util.results import EpochData
 
 
@@ -46,7 +46,7 @@ def recover_flattened(flat_params, model):
     return recovered_params
 
 
-def initialize_default_model(conf: config.DistributedConfig, model_class) -> torch.nn.Module:
+def initialize_default_model(conf: DistributedConfig, model_class) -> torch.nn.Module:
     """
     Load a default model dictionary into a torch model.
     @param model:

fltk/nets/util/reproducability.py

@@ -3,6 +3,8 @@
 import numpy as np
 import torch
 
+from fltk.util.config.distributed_config import ExecutionConfig
+
 
 def cuda_reproducible_backend(cuda: bool) -> None:
     """
@@ -21,25 +23,20 @@ def cuda_reproducible_backend(cuda: bool) -> None:
         torch.backends.cudnn.deterministic = False
 
 
-def init_reproducibility(torch_seed: int = 42, cuda: bool = False, numpy_seed: int = 43, hash_seed: int = 44) -> None:
+def init_reproducibility(config: ExecutionConfig) -> None:
     """
     Function to pre-set all seeds for libraries used during training. Allows for re-producible network initialization,
     and non-deterministic number generation. Allows to prevent 'lucky' draws in network initialization.
-    @param torch_seed: Integer seed to use for the PyTorch PRNG and CUDA PRNG.
-    @type torch_seed: int
-    @param cuda: Flag to indicate whether the CUDA backend needs to be
-    @type cuda: bool
-    @param numpy_seed: Integer seed to use for NumPy's PRNG.
-    @type numpy_seed: int
-    @param hash_seed: Integer seed to use for Pythons Hash function PRNG, will set the
-    @type hash_seed: int
-
+    @param config: Execution configuration for the experiments to be run on the remote cluster.
+    @type config: ExecutionConfig
     @return: None
     @rtype: None
     """
+    random_seed = config.reproducibility.arrival_seed
+    torch_seed = config.reproducibility.torch_seed
     torch.manual_seed(torch_seed)
-    if cuda:
+    if config.cuda:
         torch.cuda.manual_seed_all(torch_seed)
         cuda_reproducible_backend(True)
-    np.random.seed(numpy_seed)
-    os.environ['PYTHONHASHSEED'] = str(hash_seed)
+    np.random.seed(random_seed)
+    os.environ['PYTHONHASHSEED'] = str(random_seed)

fltk/samplers/__init__.py

@@ -5,7 +5,7 @@
 from .dirichlet import DirichletSampler
 from .limit_labels import LimitLabelsSampler
 from .limit_labels_flex import LimitLabelsSamplerFlex
-from ..util.definitions import DataSampler
+from ..util.config.definitions import DataSampler
 from ..util.log import getLogger
 
 

fltk/strategy/__init__.py

@@ -1,4 +1,3 @@
 from .aggregation import *
 from .client_selection import *
 from .optimization import *
-from .offloading import OffloadingStrategy, parse_strategy

fltk/strategy/aggregation/__init__.py

@@ -1,10 +1,5 @@
-from typing import Callable
-
-import torch
-
-from fltk.util.definitions import Aggregations
-from .FedAvg import fed_avg
-from .aggregation import average_nn_parameters, average_nn_parameters_simple
+from fltk.util.config.definitions.aggregate import Aggregations
+from .fed_avg import fed_avg
 
 
 def get_aggregation(name: Aggregations):

fltk/strategy/optimization/__init__.py

@@ -2,20 +2,23 @@
 
 import torch
 
-from fltk.util.definitions import Optimizations
-from .fedprox import FedProx
-from .FedNova import FedNova
+from fltk.util.config.definitions.optim import Optimizations
+from .fed_prox import FedProx
+from .fed_nova import FedNova
 
 
 def get_optimizer(name: Optimizations) -> Type[torch.optim.Optimizer]:
     """
     Helper function to get specific Optimization class references.
     @param name: Optimizer class reference.
     @type name: Optimizations
-    @return: Class reference corresponding to the requested Optimizations definition.
+    @return: Class reference corresponding to the requested Optimizations definition. Requires instantiation with
+    pre-defined args and kwargs, depending on the Type of Optimizer.
     @rtype: Type[torch.optim.Optimizer]
     """
     optimizers = {
+            Optimizations.adam: torch.optim.Adam,
+            Optimizations.adam_w: torch.optim.AdamW,
             Optimizations.sgd: torch.optim.SGD,
             Optimizations.fedprox: FedProx,
             Optimizations.fednova: FedNova

fltk/util/__init__.py

@@ -1 +0,0 @@
-from fltk.util.reproducability import init_reproducibility