forked from tensorflow/tensorflow
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathattention_ops.py
More file actions
34 lines (29 loc) · 1.11 KB
/
attention_ops.py
File metadata and controls
34 lines (29 loc) · 1.11 KB
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
"""Operations for implementing attention.
"""
import tensorflow.python.platform
from tensorflow.python.framework import ops
from tensorflow.python.framework import tensor_shape
from tensorflow.python.framework import tensor_util
from tensorflow.python.ops import gen_attention_ops
# pylint: disable=wildcard-import
from tensorflow.python.ops.gen_attention_ops import *
# TODO(bsteiner): Implement the gradient function for extract_glimpse
ops.NoGradient("ExtractGlimpse")
@ops.RegisterShape("ExtractGlimpse")
def _ExtractGlimpseShape(op):
"""Shape function for ExtractGlimpse op."""
input_shape = op.inputs[0].get_shape().with_rank(4)
unused_size_shape = op.inputs[1].get_shape().merge_with(
tensor_shape.vector(2))
offsets_shape = op.inputs[2].get_shape().merge_with(
input_shape[:1].concatenate([2]))
offsets_shape = offsets_shape
size_value = tensor_util.ConstantValue(op.inputs[1])
if size_value is not None:
height = size_value[0]
width = size_value[1]
else:
height = None
width = None
return [tensor_shape.TensorShape(
[input_shape[0], height, width, input_shape[3]])]