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
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
|
/* Copyright 2015 The TensorFlow Authors. All Rights Reserved.
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.
==============================================================================*/
// See docs in ../ops/nn_ops.cc.
#ifndef TENSORFLOW_KERNELS_RELU_OP_H_
#define TENSORFLOW_KERNELS_RELU_OP_H_
#define EIGEN_USE_THREADS
#include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"
#include "tensorflow/core/framework/numeric_op.h"
#include "tensorflow/core/framework/op_kernel.h"
#include "tensorflow/core/framework/register_types.h"
#include "tensorflow/core/framework/tensor.h"
#include "tensorflow/core/kernels/relu_op_functor.h"
#include "tensorflow/core/lib/core/errors.h"
namespace tensorflow {
template <typename Device, typename T>
class ReluOp : public UnaryElementWiseOp<T, ReluOp<Device, T>> {
public:
using UnaryElementWiseOp<T, ReluOp<Device, T>>::UnaryElementWiseOp;
void Operate(OpKernelContext* context, const Tensor& input, Tensor* output) {
functor::Relu<Device, T> functor;
functor(context->eigen_device<Device>(), input.flat<T>(),
output->flat<T>());
}
};
// Out of line check to save code space (we have this code once, rather
// than once for every NDIMS * NumTypes * Num_different_relu_variants
// functions.
struct ReluHelpers {
static void ValidateSameSizeHelper(OpKernelContext* context, const Tensor& g,
const Tensor& a) {
OP_REQUIRES(context, a.IsSameSize(g),
errors::InvalidArgument("g and a must be the same size"));
}
static bool ValidateSameSize(OpKernelContext* context, const Tensor& g,
const Tensor& a) {
ValidateSameSizeHelper(context, g, a);
return context->status().ok();
}
};
template <typename Device, typename T>
class ReluGradOp : public BinaryElementWiseOp<T, ReluGradOp<Device, T>> {
public:
using BinaryElementWiseOp<T, ReluGradOp<Device, T>>::BinaryElementWiseOp;
void OperateNoTemplate(OpKernelContext* context, const Tensor& g,
const Tensor& a, Tensor* output);
// INPUTS:
// g (gradients): backpropagated gradients
// a (inputs): either the inputs that were passed to ReluOp(), or its
// outputs (using either one yields the same result here).
// OUTPUT:
// gradients to backprop
template <int NDIMS>
void Operate(OpKernelContext* context, const Tensor& g, const Tensor& a,
Tensor* output) {
OperateNoTemplate(context, g, a, output);
}
};
template <typename Device, typename T>
void ReluGradOp<Device, T>::OperateNoTemplate(OpKernelContext* context,
const Tensor& g, const Tensor& a,
Tensor* output) {
if (!ReluHelpers::ValidateSameSize(context, g, a)) return;
functor::ReluGrad<Device, T> functor;
functor(context->eigen_device<Device>(), g.flat<T>(), a.flat<T>(),
output->flat<T>());
}
template <typename Device, typename T>
class Relu6Op : public UnaryElementWiseOp<T, Relu6Op<Device, T>> {
public:
using UnaryElementWiseOp<T, Relu6Op<Device, T>>::UnaryElementWiseOp;
void Operate(OpKernelContext* context, const Tensor& input, Tensor* output) {
functor::Relu6<Device, T> functor;
functor(context->eigen_device<Device>(), input.flat<T>(),
output->flat<T>());
}
};
template <typename Device, typename T>
class Relu6GradOp : public BinaryElementWiseOp<T, Relu6GradOp<Device, T>> {
public:
using BinaryElementWiseOp<T, Relu6GradOp<Device, T>>::BinaryElementWiseOp;
void OperateNoTemplate(OpKernelContext* context, const Tensor& g,
const Tensor& a, Tensor* output);
// INPUTS:
// g (gradients): backpropagated gradients
// a (inputs): inputs that were passed to Relu6Op()
// OUTPUT:
// gradients to backprop
template <int NDIMS>
void Operate(OpKernelContext* context, const Tensor& g, const Tensor& a,
Tensor* output) {
OperateNoTemplate(context, g, a, output);
}
};
template <typename Device, typename T>
void Relu6GradOp<Device, T>::OperateNoTemplate(OpKernelContext* context,
const Tensor& g, const Tensor& a,
Tensor* output) {
if (!ReluHelpers::ValidateSameSize(context, g, a)) return;
functor::Relu6Grad<Device, T> functor;
functor(context->eigen_device<Device>(), g.flat<T>(), a.flat<T>(),
output->flat<T>());
}
template <typename Device, typename T>
class EluOp : public UnaryElementWiseOp<T, EluOp<Device, T>> {
public:
using UnaryElementWiseOp<T, EluOp<Device, T>>::UnaryElementWiseOp;
void Operate(OpKernelContext* context, const Tensor& input, Tensor* output) {
functor::Elu<Device, T> functor;
functor(context->eigen_device<Device>(), input.flat<T>(),
output->flat<T>());
}
};
template <typename Device, typename T>
class EluGradOp : public BinaryElementWiseOp<T, EluGradOp<Device, T>> {
public:
using BinaryElementWiseOp<T, EluGradOp<Device, T>>::BinaryElementWiseOp;
void OperateNoTemplate(OpKernelContext* context, const Tensor& g,
const Tensor& a, Tensor* output);
// INPUTS:
// g (gradients): backpropagated gradients
// a (outputs): outputs of the EluOp()
// OUTPUT:
// gradients to backprop
template <int NDIMS>
void Operate(OpKernelContext* context, const Tensor& g, const Tensor& a,
Tensor* output) {
OperateNoTemplate(context, g, a, output);
}
};
template <typename Device, typename T>
void EluGradOp<Device, T>::OperateNoTemplate(OpKernelContext* context,
const Tensor& g, const Tensor& a,
Tensor* output) {
if (!ReluHelpers::ValidateSameSize(context, g, a)) return;
functor::EluGrad<Device, T> functor;
functor(context->eigen_device<Device>(), g.flat<T>(), a.flat<T>(),
output->flat<T>());
}
} // namespace tensorflow
#undef EIGEN_USE_THREADS
#endif // TENSORFLOW_KERNELS_RELU_OP_H_
|
