Automated rollback of change 153358570
Change: 153718592
This commit is contained in:
parent
f12c967893
commit
900652aed4
@ -107,13 +107,8 @@ def optimize_loss(loss,
|
||||
gradient_noise_scale: float or None, adds 0-mean normal noise scaled by this
|
||||
value.
|
||||
gradient_multipliers: dict of variables or variable names to floats.
|
||||
Most often it should be an instance of
|
||||
collections.defaultdict(lambda: 1.0), so that
|
||||
unspecified variable would still be updated with
|
||||
normal rate. If a common dict is provided, only
|
||||
gradients for specified variables will be
|
||||
multiplied by given constant and unspecified
|
||||
gradients will be dropped.
|
||||
If present, gradients for specified
|
||||
variables will be multiplied by given constant.
|
||||
clip_gradients: float, callable or `None`. If float, is provided, a global
|
||||
clipping is applied to prevent the norm of the gradient to exceed this
|
||||
value. Alternatively, a callable can be provided e.g.: adaptive_clipping.
|
||||
|
Loading…
Reference in New Issue
Block a user