tff.learning.BatchOutput

View source on GitHub

Class BatchOutput

A structure that holds the output of a tff.learning.Model.

Used in the tutorials:

• Federated Learning for Image Classification

NOTE: All fields are optional (may be None).

• loss: The scalar mean loss on the examples in the batch. If the model has multiple losses, it is the sum of all the individual losses.
• predictions: Tensor of predictions on the examples. The first dimension must be the same size (the size of the batch).
• num_examples: Number of examples seen in the batch.

__new__

__new__(
    _cls,
    loss,
    predictions,
    num_examples
)

Create new instance of BatchOutput(loss, predictions, num_examples)

Properties

loss

predictions

num_examples