tf_agents.utils.common.aggregate_losses
Aggregates and scales per example loss and regularization losses.
tf_agents.utils.common.aggregate_losses(
per_example_loss=None,
sample_weight=None,
global_batch_size=None,
regularization_loss=None
)
If global_batch_size
is given it would be used for scaling, otherwise it
would use the batch_dim of per_example_loss and number of replicas.
Args |
per_example_loss
|
Per-example loss [B] or [B, T, ...].
|
sample_weight
|
Optional weighting for each example, Tensor shaped [B] or [B,
T, ...], or a scalar float.
|
global_batch_size
|
Optional global batch size value. Defaults to (size of
first dimension of losses ) * (number of replicas).
|
regularization_loss
|
Regularization loss.
|
Returns |
An AggregatedLosses named tuple with scalar losses to optimize.
|
Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. For details, see the Google Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.
Last updated 2024-04-26 UTC.
[null,null,["Last updated 2024-04-26 UTC."],[],[]]