626_privacy/tensorflow_privacy
2
0
Fork 1
Code Issues Pull requests 1 Projects Releases Packages Wiki Activity Actions

Correct imports of keras loss utils

Browse source 
PiperOrigin-RevId: 486795765
This commit is contained in:
A. Unique TensorFlower 2022-11-07 16:33:28 -08:00
parent e334633466
commit ec747a8d75
3 changed files with 5 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

3
tensorflow_privacy/privacy/estimators/binary_class_head.py
View file

@ -14,7 +14,6 @@
"""Binary class head for Estimator that allow integration with TF Privacy.""" """Binary class head for Estimator that allow integration with TF Privacy."""
import tensorflow as tf import tensorflow as tf
from tensorflow.python.keras.utils import losses_utils # pylint: disable=g-direct-tensorflow-import
from tensorflow_estimator.python.estimator import model_fn from tensorflow_estimator.python.estimator import model_fn
from tensorflow_estimator.python.estimator.canned import prediction_keys from tensorflow_estimator.python.estimator.canned import prediction_keys
from tensorflow_estimator.python.estimator.export import export_output from tensorflow_estimator.python.estimator.export import export_output
@ -55,7 +54,7 @@ class DPBinaryClassHead(binary_class_head.BinaryClassHead):
labels = self._processed_labels(logits, labels) labels = self._processed_labels(logits, labels)
unweighted_loss, weights = self._unweighted_loss_and_weights( unweighted_loss, weights = self._unweighted_loss_and_weights(
logits, labels, features) logits, labels, features)
vector_training_loss = losses_utils.compute_weighted_loss( vector_training_loss = tf.keras.__internal__.losses.compute_weighted_loss(
unweighted_loss, unweighted_loss,
sample_weight=weights, sample_weight=weights,
reduction=tf.keras.losses.Reduction.NONE) reduction=tf.keras.losses.Reduction.NONE)

7
tensorflow_privacy/privacy/estimators/multi_class_head.py
View file

@ -14,7 +14,6 @@
"""Multiclass head for Estimator that allow integration with TF Privacy.""" """Multiclass head for Estimator that allow integration with TF Privacy."""
import tensorflow as tf import tensorflow as tf
from tensorflow.python.keras.utils import losses_utils # pylint: disable=g-direct-tensorflow-import
from tensorflow_estimator.python.estimator import model_fn from tensorflow_estimator.python.estimator import model_fn
from tensorflow_estimator.python.estimator.canned import prediction_keys from tensorflow_estimator.python.estimator.canned import prediction_keys
from tensorflow_estimator.python.estimator.export import export_output from tensorflow_estimator.python.estimator.export import export_output
@ -30,14 +29,14 @@ class DPMultiClassHead(multi_class_head.MultiClassHead):
n_classes, n_classes,
weight_column=None, weight_column=None,
label_vocabulary=None, label_vocabulary=None,
loss_reduction=losses_utils.ReductionV2.SUM_OVER_BATCH_SIZE, loss_reduction=tf.keras.losses.Reduction.SUM_OVER_BATCH_SIZE,
loss_fn=None, loss_fn=None,
name=None): name=None):
super().__init__( super().__init__(
n_classes=n_classes, n_classes=n_classes,
weight_column=weight_column, weight_column=weight_column,
label_vocabulary=label_vocabulary, label_vocabulary=label_vocabulary,
loss_reduction=losses_utils.ReductionV2.SUM_OVER_BATCH_SIZE, loss_reduction=tf.keras.losses.Reduction.SUM_OVER_BATCH_SIZE,
loss_fn=loss_fn, loss_fn=loss_fn,
name=name) name=name)
@ -55,7 +54,7 @@ class DPMultiClassHead(multi_class_head.MultiClassHead):
labels = self._processed_labels(logits, labels) labels = self._processed_labels(logits, labels)
unweighted_loss, weights = self._unweighted_loss_and_weights( unweighted_loss, weights = self._unweighted_loss_and_weights(
logits, labels, features) logits, labels, features)
vector_training_loss = losses_utils.compute_weighted_loss( vector_training_loss = tf.keras.__internal__.losses.compute_weighted_loss(
unweighted_loss, unweighted_loss,
sample_weight=weights, sample_weight=weights,
reduction=tf.keras.losses.Reduction.NONE) reduction=tf.keras.losses.Reduction.NONE)

3
tensorflow_privacy/privacy/estimators/multi_label_head.py
View file

@ -14,7 +14,6 @@
"""Multiclass head for Estimator that allow integration with TF Privacy.""" """Multiclass head for Estimator that allow integration with TF Privacy."""
import tensorflow as tf import tensorflow as tf
from tensorflow.python.keras.utils import losses_utils # pylint: disable=g-direct-tensorflow-import
from tensorflow_estimator.python.estimator import model_fn from tensorflow_estimator.python.estimator import model_fn
from tensorflow_estimator.python.estimator.canned import prediction_keys from tensorflow_estimator.python.estimator.canned import prediction_keys
from tensorflow_estimator.python.estimator.export import export_output from tensorflow_estimator.python.estimator.export import export_output
@ -61,7 +60,7 @@ class DPMultiLabelHead(multi_label_head.MultiLabelHead):
labels = self._processed_labels(logits, labels) labels = self._processed_labels(logits, labels)
unweighted_loss, weights = self._unweighted_loss_and_weights( unweighted_loss, weights = self._unweighted_loss_and_weights(
logits, labels, features) logits, labels, features)
vector_training_loss = losses_utils.compute_weighted_loss( vector_training_loss = tf.keras.__internal__.losses.compute_weighted_loss(
unweighted_loss, unweighted_loss,
sample_weight=weights, sample_weight=weights,
reduction=tf.keras.losses.Reduction.NONE) reduction=tf.keras.losses.Reduction.NONE)