Missed TensorFlow Dev Summit? Check out the video playlist. Watch recordings

tff.learning.framework.ServerState

View source on GitHub

Represents the state of the server carried between rounds.

tff.learning.framework.ServerState(
    model, optimizer_state, delta_aggregate_state, model_broadcast_state
)

Attributes:

  • model: A ModelWeights structure, containing Tensors or Variables.
  • optimizer_state: A list of Tensors or Variables, in the order returned by optimizer.variables()
  • delta_aggregate_state: State (possibly empty) of the delta_aggregate_fn.
  • model_broadcast_state: State (possibly empty) of the model_broadcast_fn.