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
181
182
183
184
185
186
187
188
189
190
191
192
|
/* Copyright 2017 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.
==============================================================================*/
#include "tensorflow/compiler/tf2xla/xla_helpers.h"
#include "tensorflow/compiler/tf2xla/xla_op_kernel.h"
#include "tensorflow/compiler/tf2xla/xla_op_registry.h"
#include "tensorflow/compiler/xla/client/xla_builder.h"
namespace tensorflow {
namespace {
void SpaceToBatch(XlaOpKernelContext* ctx, const xla::XlaOp& input,
DataType input_dtype, const TensorShape& input_tensor_shape,
absl::Span<const int64> block_shape,
const xla::Literal& paddings) {
const int input_rank = input_tensor_shape.dims();
const absl::InlinedVector<int64, 4> input_shape =
input_tensor_shape.dim_sizes();
const int block_rank = block_shape.size();
OP_REQUIRES(
ctx, input_rank >= 1 + block_rank,
errors::InvalidArgument("input rank should be >= ", 1 + block_rank,
" instead of ", input_rank));
absl::Span<const int64> remainder_shape(input_shape);
remainder_shape.remove_prefix(1 + block_rank);
OP_REQUIRES(
ctx,
xla::ShapeUtil::Rank(paddings.shape()) == 2 &&
block_rank == xla::ShapeUtil::GetDimension(paddings.shape(), 0) &&
2 == xla::ShapeUtil::GetDimension(paddings.shape(), 1),
errors::InvalidArgument("paddings should have shape [", block_rank,
", 2] instead of ",
xla::ShapeUtil::HumanString(paddings.shape())));
xla::XlaBuilder* b = ctx->builder();
// 1. Zero-pad the start and end of dimensions `[1, ..., M]` of the
// input according to `paddings` to produce `padded` of shape `padded_shape`.
xla::PaddingConfig padding_config;
std::vector<int64> padded_shape(input_shape.begin(), input_shape.end());
int64 block_num_elems = 1LL;
padding_config.add_dimensions(); // Don't pad the batch dimension.
for (int i = 0; i < block_rank; ++i) {
auto* dim = padding_config.add_dimensions();
int64 pad_start = paddings.Get<int64>({i, 0});
int64 pad_end = paddings.Get<int64>({i, 1});
OP_REQUIRES(ctx, pad_start >= 0 && pad_end >= 0,
errors::InvalidArgument("Paddings must be non-negative"));
dim->set_edge_padding_low(pad_start);
dim->set_edge_padding_high(pad_end);
padded_shape[1 + i] += pad_start + pad_end;
block_num_elems *= block_shape[i];
}
// Don't pad the remainder dimensions.
for (int i = 0; i < remainder_shape.size(); ++i) {
padding_config.add_dimensions();
}
OP_REQUIRES(ctx, block_num_elems > 0,
errors::InvalidArgument(
"The product of the block dimensions must be positive"));
xla::XlaOp padded =
xla::Pad(input, XlaHelpers::Zero(b, input_dtype), padding_config);
// 2. Reshape `padded` to `reshaped_padded` of shape:
//
// [batch] +
// [padded_shape[1] / block_shape[0],
// block_shape[0],
// ...,
// padded_shape[M] / block_shape[M-1],
// block_shape[M-1]] +
// remaining_shape
const int64 batch_size = input_shape[0];
std::vector<int64> reshaped_padded_shape(input_rank + block_rank);
reshaped_padded_shape[0] = batch_size;
for (int i = 0; i < block_rank; ++i) {
OP_REQUIRES(ctx, padded_shape[1 + i] % block_shape[i] == 0,
errors::InvalidArgument("padded_shape[", 1 + i,
"]=", padded_shape[1 + i],
" is not divisible by block_shape[", i,
"]=", block_shape[i]));
reshaped_padded_shape[1 + i * 2] = padded_shape[1 + i] / block_shape[i];
reshaped_padded_shape[1 + i * 2 + 1] = block_shape[i];
}
std::copy(remainder_shape.begin(), remainder_shape.end(),
reshaped_padded_shape.begin() + 1 + 2 * block_rank);
xla::XlaOp reshaped_padded = xla::Reshape(padded, reshaped_padded_shape);
// 3. Permute dimensions of `reshaped_padded` to produce
// `permuted_reshaped_padded` of shape:
//
// block_shape +
// [batch] +
// [padded_shape[1] / block_shape[0],
// ...,
// padded_shape[M] / block_shape[M-1]] +
// remaining_shape
std::vector<int64> permutation(reshaped_padded_shape.size());
for (int i = 0; i < block_rank; ++i) {
permutation[i] = 1 + 2 * i + 1;
permutation[block_rank + 1 + i] = 1 + 2 * i;
}
permutation[block_rank] = 0;
std::iota(permutation.begin() + 1 + block_rank * 2, permutation.end(),
1 + block_rank * 2);
xla::XlaOp permuted_reshaped_padded =
xla::Transpose(reshaped_padded, permutation);
// 4. Reshape `permuted_reshaped_padded` to flatten `block_shape` into the
// batch dimension, producing an output tensor of shape:
//
// [batch * prod(block_shape)] +
// [padded_shape[1] / block_shape[0],
// ...,
// padded_shape[M] / block_shape[M-1]] +
// remaining_shape
// Determine the length of the prefix of block dims that can be combined
// into the batch dimension due to having no padding and block_shape=1.
std::vector<int64> output_shape(input_rank);
output_shape[0] = batch_size * block_num_elems;
for (int i = 0; i < block_rank; ++i) {
output_shape[1 + i] = padded_shape[1 + i] / block_shape[i];
}
std::copy(remainder_shape.begin(), remainder_shape.end(),
output_shape.begin() + 1 + block_rank);
xla::XlaOp output = xla::Reshape(permuted_reshaped_padded, output_shape);
ctx->SetOutput(0, output);
}
class SpaceToBatchNDOp : public XlaOpKernel {
public:
explicit SpaceToBatchNDOp(OpKernelConstruction* ctx) : XlaOpKernel(ctx) {}
void Compile(XlaOpKernelContext* ctx) override {
std::vector<int64> block_shape;
OP_REQUIRES_OK(ctx, ctx->ConstantInputAsIntVector(1, &block_shape));
xla::Literal paddings;
OP_REQUIRES_OK(ctx, ctx->ConstantInputAsInt64Literal(2, &paddings));
SpaceToBatch(ctx, ctx->Input(0), input_type(0), ctx->InputShape(0),
block_shape, paddings);
}
};
REGISTER_XLA_OP(Name("SpaceToBatchND")
.CompileTimeConstInput("paddings")
.CompileTimeConstInput("block_shape"),
SpaceToBatchNDOp);
class SpaceToBatchOp : public XlaOpKernel {
public:
explicit SpaceToBatchOp(OpKernelConstruction* ctx) : XlaOpKernel(ctx) {
OP_REQUIRES_OK(ctx, ctx->GetAttr("block_size", &block_size_));
OP_REQUIRES(
ctx, block_size_ > 1,
errors::InvalidArgument("Block size should be > 1: ", block_size_));
}
void Compile(XlaOpKernelContext* ctx) override {
xla::Literal paddings;
OP_REQUIRES_OK(ctx, ctx->ConstantInputAsInt64Literal(1, &paddings));
SpaceToBatch(ctx, ctx->Input(0), input_type(0), ctx->InputShape(0),
{block_size_, block_size_}, paddings);
}
private:
int block_size_;
};
REGISTER_XLA_OP(Name("SpaceToBatch").CompileTimeConstInput("paddings"),
SpaceToBatchOp);
} // namespace
} // namespace tensorflow
|
