Skip to content

Supports DP optimizers #251

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

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Supports DP optimizers #251

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
17 changes: 14 additions & 3 deletions tensorflow_privacy/privacy/optimizers/dp_optimizer_keras.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -239,7 +239,11 @@ def _compute_gradients(self, loss, var_list, grad_loss=None, tape=None):
else:
num_microbatches = self._num_microbatches
microbatch_losses = tf.reduce_mean(
tf.reshape(loss, [num_microbatches, -1]), axis=1)
tf.reshape(
loss,
[num_microbatches,
tf.shape(loss)[0] / num_microbatches]),
axis=1)

if callable(var_list):
var_list = var_list()
Expand All @@ -250,7 +254,11 @@ def _compute_gradients(self, loss, var_list, grad_loss=None, tape=None):
else:
num_microbatches = self._num_microbatches
microbatch_losses = tf.reduce_mean(
tf.reshape(loss, [num_microbatches, -1]), axis=1)
tf.reshape(
loss,
[num_microbatches,
tf.shape(loss)[0] / num_microbatches]),
axis=1)

var_list = tf.nest.flatten(var_list)

Expand Down Expand Up @@ -294,7 +302,10 @@ def get_gradients(self, loss, params):
# This code mostly follows the logic in the original DPOptimizerClass
# in dp_optimizer.py, except that this returns only the gradients,
# not the gradients and variables.
microbatch_losses = tf.reshape(loss, [self._num_microbatches, -1])
microbatch_losses = tf.reshape(
loss,
[self._num_microbatches,
tf.shape(loss)[0] / self._num_microbatches])
sample_params = (
self._dp_sum_query.derive_sample_params(self._global_state))

Expand Down