{ }
View source on GitHub |
Returns a tensor whose value represents the total loss.
tf.compat.v1.losses.get_total_loss(
add_regularization_losses=True, name='total_loss', scope=None
)
In particular, this adds any losses you have added with tf.add_loss()
to
any regularization losses that have been added by regularization parameters
on layers constructors e.g. tf.layers
. Be very sure to use this if you
are constructing a loss_op manually. Otherwise regularization arguments
on tf.layers
methods will not function.
Returns | |
---|---|
A Tensor whose value represents the total loss.
|
Raises | |
---|---|
ValueError
|
if losses is not iterable.
|