Skip to content

Patch for issue 391 #395

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
szmigacz merged 2 commits into from
Feb 7, 2020
Merged

Patch for issue 391 #395

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3ed70df
Merge pull request #5 from NVIDIA/master
swethmandava Feb 7, 2020
ddf6dfc
bug fix in beta decay
Feb 7, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 4 additions & 5 deletions TensorFlow/LanguageModeling/BERT/optimization.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -301,12 +301,12 @@ def __init__(self,
self.beta_2 = beta_2
self.epsilon = epsilon
self.exclude_from_weight_decay = exclude_from_weight_decay
self.steps = 0

def apply_gradients(self, grads_and_vars, global_step=None, name=None,
def apply_gradients(self, grads_and_vars, global_step, name=None,
manual_fp16=False):
"""See base class."""
assignments = []
steps = tf.cast(global_step, tf.float32)
for (grad, param) in grads_and_vars:
if grad is None or param is None:
continue
Expand Down Expand Up @@ -343,9 +343,8 @@ def apply_gradients(self, grads_and_vars, global_step=None, name=None,
tf.multiply(self.beta_2, v) + tf.multiply(1.0 - self.beta_2,
tf.square(grad)))

self.steps += 1
beta1_correction = (1 - self.beta_1 ** self.steps)
beta2_correction = (1 - self.beta_2 ** self.steps)
beta1_correction = (1 - self.beta_1 ** steps)
beta2_correction = (1 - self.beta_2 ** steps)

next_m_unbiased = next_m / beta1_correction
next_v_unbiased = next_v / beta2_correction
Expand Down