Module: tff

View source on GitHub

TensorFlow Federated Library.

Modules

backends module: This directory contains code for interfacing custom types of backends.

framework module: Interfaces for extensions, selectively lifted out of impl.

learning module: The public API for model developers using federated learning algorithms.

simulation module: The public API for experimenters running federated learning simulations.

utils module: Utility classes/functions built on top of TensorFlow Federated Core API.

Classes

class Computation: An abstract interface for all classes that represent computations.

class FederatedType: An implementation of tff.Type representing federated types in TFF.

class FunctionType: An implementation of tff.Type representing functional types in TFF.

class NamedTupleType: An implementation of tff.Type representing named tuple types in TFF.

class SequenceType: An implementation of tff.Type representing types of sequences in TFF.

class TensorType: An implementation of tff.Type representing types of tensors in TFF.

class Type: An abstract interface for all classes that represent TFF types.

class TypedObject: An abstract interface for things that possess TFF type signatures.

class Value: An abstract base class for all values in the bodies of TFF computations.

Functions

federated_aggregate(...): Aggregates value from tff.CLIENTS to tff.SERVER.

federated_apply(...): Applies a given function to a federated value on the tff.SERVER.

federated_broadcast(...): Broadcasts a federated value from the tff.SERVER to the tff.CLIENTS.

federated_collect(...): Returns a federated value from tff.CLIENTS as a tff.SERVER sequence.

federated_computation(...): Decorates/wraps Python functions as TFF federated/composite computations.

federated_map(...): Maps a federated value on tff.CLIENTS pointwise using a mapping function.

federated_mean(...): Computes a tff.SERVER mean of value placed on tff.CLIENTS.

federated_reduce(...): Reduces value from tff.CLIENTS to tff.SERVER using a reduction op.

federated_sum(...): Computes a sum at tff.SERVER of a value placed on the tff.CLIENTS.

federated_value(...): Returns a federated value at placement, with value as the constituent.

federated_zip(...): Converts an N-tuple of federated values into a federated N-tuple value.

sequence_map(...): Maps a TFF sequence value pointwise using a given function mapping_fn.

sequence_reduce(...): Reduces a TFF sequence value given a zero and reduction operator op.

sequence_sum(...): Computes a sum of elements in a sequence.

tf_computation(...): Decorates/wraps Python functions and defuns as TFF TensorFlow computations.

to_type(...): Converts the argument into an instance of tff.Type.

to_value(...): Converts the argument into an instance of the abstract class tff.Value.

Other Members

  • CLIENTS
  • SERVER
  • __version__ = '0.10.1'