from absl import app
from absl import flags
from absl import logging
import tensorflow as tf
import tensorflow_federated as tff
from tensorflow_federated.python.research.differential_privacy import dp_utils
from tensorflow_federated.python.research.optimization.shared import optimizer_utils
from tensorflow_federated.python.research.utils import training_loop
from tensorflow_federated.python.research.utils import training_utils
from tensorflow_federated.python.research.utils import utils_impl
from tensorflow_federated.python.research.utils.datasets import emnist_dataset
from tensorflow_federated.python.research.utils.models import emnist_models
with utils_impl.record_hparam_flags():
# Experiment hyperparameters
flags.DEFINE_enum(
'model', 'cnn', ['cnn', '2nn'], 'Which model to use. This '
'can be a convolutional model (cnn) or a two hidden-layer '
'densely connected network (2nn).')
flags.DEFINE_integer('client_batch_size', 20,
'Batch size used on the client.')
flags.DEFINE_integer('clients_per_round', 10,
'How many clients to sample per round.')
flags.DEFINE_integer(
'client_epochs_per_round', 1,
'Number of client (inner optimizer) epochs per federated round.')
flags.DEFINE_boolean(
'uniform_weighting', False,
'Whether to weigh clients uniformly. If false, clients '
from tensorflow_federated.python.research.adaptive_lr_decay import callbacks
from tensorflow_federated.python.research.optimization.cifar100 import federated_cifar100
from tensorflow_federated.python.research.optimization.emnist import federated_emnist
from tensorflow_federated.python.research.optimization.emnist_ae import federated_emnist_ae
from tensorflow_federated.python.research.optimization.shakespeare import federated_shakespeare
from tensorflow_federated.python.research.optimization.shared import optimizer_utils
from tensorflow_federated.python.research.optimization.stackoverflow import federated_stackoverflow
from tensorflow_federated.python.research.optimization.stackoverflow_lr import federated_stackoverflow_lr
from tensorflow_federated.python.research.utils import utils_impl
_SUPPORTED_TASKS = [
'cifar100', 'emnist_cr', 'emnist_ae', 'shakespeare', 'stackoverflow_nwp',
'stackoverflow_lr'
]
with utils_impl.record_hparam_flags() as optimizer_flags:
optimizer_utils.define_optimizer_flags('client')
optimizer_utils.define_optimizer_flags('server')
with utils_impl.record_hparam_flags() as callback_flags:
flags.DEFINE_float(
'client_decay_factor', 0.1, 'Amount to decay the client learning rate '
'upon reaching a plateau.')
flags.DEFINE_float(
'server_decay_factor', 0.9, 'Amount to decay the server learning rate '
'upon reaching a plateau.')
flags.DEFINE_float(
'min_delta', 1e-4,
'Minimum delta for improvement in the learning rate '
'callbacks.')
flags.DEFINE_integer(
from tensorflow_federated.python.research.optimization.cifar100 import federated_cifar100
from tensorflow_federated.python.research.optimization.emnist import federated_emnist
from tensorflow_federated.python.research.optimization.emnist_ae import federated_emnist_ae
from tensorflow_federated.python.research.optimization.shakespeare import federated_shakespeare
from tensorflow_federated.python.research.optimization.shared import fed_avg_schedule
from tensorflow_federated.python.research.optimization.shared import optimizer_utils
from tensorflow_federated.python.research.optimization.stackoverflow import federated_stackoverflow
from tensorflow_federated.python.research.optimization.stackoverflow_lr import federated_stackoverflow_lr
from tensorflow_federated.python.research.utils import utils_impl
_SUPPORTED_TASKS = [
'cifar100', 'emnist_cr', 'emnist_ae', 'shakespeare', 'stackoverflow_nwp',
'stackoverflow_lr'
]
with utils_impl.record_hparam_flags() as optimizer_flags:
# Defining optimizer flags
optimizer_utils.define_optimizer_flags('client')
optimizer_utils.define_optimizer_flags('server')
optimizer_utils.define_lr_schedule_flags('client')
optimizer_utils.define_lr_schedule_flags('server')
with utils_impl.record_hparam_flags() as shared_flags:
# Federated training hyperparameters
flags.DEFINE_integer('client_epochs_per_round', 1,
'Number of epochs in the client to take per round.')
flags.DEFINE_integer('client_batch_size', 20, 'Batch size on the clients.')
flags.DEFINE_integer('clients_per_round', 10,
'How many clients to sample per round.')
flags.DEFINE_integer('client_datasets_random_seed', 1,
'Random seed for client sampling.')
