forked from google-research/google-research
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathkeras_utils.py
48 lines (38 loc) · 1.38 KB
/
keras_utils.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
# coding=utf-8
# Copyright 2020 The Google Research Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Utils for tensorflow/keras."""
import tensorflow.compat.v2 as tf
def my_reset_states(metric):
"""Resets metric states.
Args:
metric: A keras metric to reset states for.
"""
for var in metric.variables:
var.assign(0)
def orthogonal_regularization(model, reg_coef=1e-4):
"""Orthogonal regularization v2.
See equation (3) in https://arxiv.org/abs/1809.11096.
Args:
model: A keras model to apply regualization for.
reg_coef: Orthogonal regularization coefficient.
Returns:
A regularization loss term.
"""
reg = 0
for layer in model.layers:
if isinstance(layer, tf.keras.layers.Dense):
prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)
reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))
return reg * reg_coef