wehub/paddlepaddle--paddle
1
0
Fork 0
Code Issues Pull Requests Actions 2 Deployments Models Agent Notes Packages Projects Releases Wiki Activity
Files
ff97fb6869c2eeecb7b4c09af41ce69f43f2c574
paddlepaddle--paddle/paddle/phi/infermeta/binary.cc
T
wehub-resource-sync e25996e7db chore: import upstream snapshot with attribution
2026-07-13 12:40:42 +08:00

5148 lines
180 KiB
C++
Raw Blame History

/* Copyright (c) 2024 PaddlePaddle 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 "paddle/phi/infermeta/binary.h"
#include <algorithm>
#include <vector>
#include "glog/logging.h"
#include "paddle/common/ddim.h"
#include "paddle/common/flags.h"
#include "paddle/common/layout.h"
#include "paddle/phi/api/lib/data_type_set.h"
#include "paddle/phi/backends/onednn/onednn_helper.h"
#include "paddle/phi/common/data_type.h"
#include "paddle/phi/common/type_traits.h"
#include "paddle/phi/core/infermeta_utils.h"
#include "paddle/phi/core/utils/data_type.h"
#include "paddle/phi/infermeta/unary.h"
#include "paddle/phi/kernels/cpu/conv_util.h"
#include "paddle/phi/kernels/funcs/axis_utils.h"
#include "paddle/phi/kernels/funcs/common_infer_shape_functions.h"
#include "paddle/phi/kernels/funcs/common_shape.h"
#include "paddle/phi/kernels/funcs/correlation_funcs.h"
COMMON_DECLARE_bool(manually_trans_conv_filter);
namespace phi {
namespace detail {
static void BinarySameInputDimsCheck(const MetaTensor& x,
const MetaTensor& y,
MetaConfig config) {
auto input_dim = x.dims();
auto other_dim = y.dims();
PADDLE_ENFORCE_EQ(input_dim.size(),
other_dim.size(),
common::errors::PreconditionNotMet(
"Input(Input) and Input(Other) must have the same "
"dimension size."));
int n = input_dim.size();
bool is_runtime = config.is_runtime;
for (int i = 0; i < n; i++) {
if (is_runtime) {
PADDLE_ENFORCE_EQ(input_dim[i],
other_dim[i],
common::errors::PreconditionNotMet(
"The value at dim %d of Input(Input) is not "
"equal to the Input(Other): %ld != %ld.",
i,
input_dim[i],
other_dim[i]));
} else {
if (!(input_dim[i] < 0 || other_dim[i] < 0)) {
PADDLE_ENFORCE_EQ(input_dim[i],
other_dim[i],
common::errors::PreconditionNotMet(
"The value at dim %d of Input(Input) is not "
"equal to the Input(Other): %ld != %ld.",
i,
input_dim[i],
other_dim[i]));
}
}
}
}
// Used in MatrixRankTolInferMeta
static DDim CheckAndGetOutputDim(const DDim& dim_x) {
auto x_vec = vectorize(dim_x);
if (x_vec.size() == 2) {
return make_ddim({});
}
x_vec.erase(x_vec.end() - 2, x_vec.end());
return make_ddim(x_vec);
}
} // namespace detail
void AllValueCompareInferMeta(const MetaTensor& x,
const MetaTensor& y,
MetaTensor* out,
MetaConfig config) {
if (x.numel() != 0 && y.numel() != 0) {
detail::BinarySameInputDimsCheck(x, y, config);
}
out->set_dims(make_ddim({}));
out->set_dtype(DataType::BOOL);
}
void KLDivInferMeta(const MetaTensor& x,
const MetaTensor& label,
const std::string& reduction,
bool log_target,
MetaTensor* out,
MetaConfig config) {
auto dim_x = x.dims();
auto dim_target = label.dims();
PADDLE_ENFORCE_EQ(dim_x.size(),
dim_target.size(),
common::errors::InvalidArgument(
"Input(X) rank and Input(Target) rank should be "
"same, but received X rank(%d) != Target rank(%d)",
dim_x.size(),
dim_target.size()));
for (int i = 0; i < dim_x.size(); i++) {
if (config.is_runtime || (dim_x[i] > 0 && dim_target[i] > 0)) {
PADDLE_ENFORCE_EQ(
dim_x[i],
dim_target[i],
common::errors::InvalidArgument(
"Input(X) and Input(Target) should in same shape. but received "
"X dimension[%d](%d) != Target dimension[%d](%d)",
i,
dim_x[i],
i,
dim_target[i]));
}
}
auto reduction_valid = "mean" == reduction || "sum" == reduction ||
"batchmean" == reduction || "none" == reduction;
PADDLE_ENFORCE_EQ(
reduction_valid,
true,
common::errors::InvalidArgument(
"Attr(reduction) can only be 'none'|'batchmean'|'sum'|'mean'."));
if ("none" == reduction) {
out->set_dims(dim_x);
} else {
out->set_dims(make_ddim({}));
}
out->set_dtype(x.dtype());
}
void ArrayWriteInferMeta(const MetaTensor& array,
const MetaTensor& x,
MetaTensor* out,
MetaConfig config) {
DataType out_dtype = array.dtype();
if (x.dtype() != DataType::UNDEFINED) {
if (array.dtype() == DataType::UNDEFINED) {
out_dtype = x.dtype();
} else {
PADDLE_ENFORCE_EQ(array.dtype(),
x.dtype(),
common::errors::InvalidArgument(
"The dtype (%s) of input x shall be same as "
"dtype (%d) of array.",
x.dtype(),
array.dtype()));
}
}
out->set_dtype(out_dtype);
out->set_layout(array.layout());
}
void ArrayReadInferMeta(const MetaTensor& array,
const Scalar& i,
MetaTensor* out,
MetaConfig config) {
if (!config.is_runtime) {
auto dims = array.dims();
if (dims.size() > 1) {
for (int i = 0; i < dims.size(); ++i) {
dims[i] = -1;
}
out->set_dims(dims);
} else {
out->set_dims({-1});
}
} else {
double index = i.to<int64_t>();
out->set_dims(array.dims(index)); // NOLINT
out->share_lod(array, index); // NOLINT
}
out->set_dtype(array.dtype());
out->set_layout(array.layout());
}
void Atan2InferMeta(const MetaTensor& x, const MetaTensor& y, MetaTensor* out) {
ElementwiseInferMeta(x, y, out);
if (out->dtype() == DataType::INT32 || out->dtype() == DataType::INT64) {
out->set_dtype(DataType::FLOAT64);
}
}
void BCELossInferMeta(const MetaTensor& input,
const MetaTensor& label,
MetaTensor* out,
MetaConfig config) {
auto input_dims = input.dims();
auto label_dims = label.dims();
int rank = input_dims.size();
PADDLE_ENFORCE_EQ(rank,
label_dims.size(),
common::errors::InvalidArgument(
"Input(X) and Input(Label) shall have the same rank."
"But received: the rank of Input(X) is [%d], "
"the rank of Input(Label) is [%d].",
rank,
label_dims.size()));
bool check = true;
if ((!config.is_runtime) &&
(contain_unknown_dim(input_dims) || contain_unknown_dim(label_dims))) {
check = false;
}
if (check) {
PADDLE_ENFORCE_EQ(input_dims,
label_dims,
common::errors::InvalidArgument(
"Input(X) and Input(Label) shall have the same "
"shape. But received: the shape of Input(X) is "
"[%s], the shape of Input(Label) is [%s].",
input_dims,
label_dims));
}
out->set_dims(input_dims);
out->set_dtype(input.dtype());
out->share_lod(input);
}
void BeamSearchDecodeInferMeta(const MetaTensor& ids,
const MetaTensor& scores,
int beam_size,
int end_id,
MetaTensor* sentence_ids,
MetaTensor* sentence_scores,
MetaConfig config) {}
void BincountInferMeta(const MetaTensor& x,
const MetaTensor& weights,
const Scalar& minlength,
MetaTensor* out) {
auto input_dim = x.dims();
PADDLE_ENFORCE_EQ(input_dim.size(),
1,
common::errors::InvalidArgument(
"The 'shape' of Input(X) must be 1-D tensor."
"But the dimension of Input(X) is [%d]",
input_dim.size()));
VLOG(4) << "####### CHECK weights";
if (weights) {
auto weights_dim = weights.dims();
VLOG(4) << "##### weights_dim " << weights_dim;
PADDLE_ENFORCE_EQ(weights_dim.size(),
1,
common::errors::InvalidArgument(
"The 'shape' of Input(Weights) must be 1-D tensor."
"But the dimension of Input(Weights) is [%d]",
weights_dim.size()));
if (input_dim[0] != 0) {
PADDLE_ENFORCE_EQ(
weights_dim[0],
input_dim[0],
common::errors::InvalidArgument(
"The 'shape' of Input(Weights) must be equal to the 'shape' of "
"Input(X)."
"But received: the 'shape' of Input(Weights) is [%s],"
"the 'shape' of Input(X) is [%s]",
weights_dim,
input_dim));
}
}
out->set_dims(make_ddim({-1}));
if (weights) {
out->set_dtype(weights.dtype());
} else {
out->set_dtype(DataType::INT64);
}
out->share_lod(x);
}
void BinomialInferMeta(const MetaTensor& count,
const MetaTensor& prob,
MetaTensor* out,
MetaConfig config) {
auto count_dims = count.dims();
auto prob_dims = prob.dims();
bool check = true;
if ((!config.is_runtime) &&
(phi::product(count_dims) <= 0 || phi::product(prob_dims) <= 0)) {
check = false;
}
if (check) {
PADDLE_ENFORCE_EQ(count_dims,
prob_dims,
common::errors::InvalidArgument(
"Input(count) and Input(prob) shall have the same "
"shape. But received: the shape of Input(count) is "
"[%s], the shape of Input(prob) is [%s].",
count_dims,
prob_dims));
}
out->set_dims(count_dims);
out->set_dtype(DataType::INT64);
}
void BmmInferMeta(const MetaTensor& x, const MetaTensor& y, MetaTensor* out) {
std::vector<int64_t> x_dims = vectorize(x.dims());
std::vector<int64_t> y_dims = vectorize(y.dims());
std::size_t x_ndims = x_dims.size();
std::size_t y_ndims = y_dims.size();
PADDLE_ENFORCE_EQ(x_ndims,
3,
common::errors::InvalidArgument(
"Input(X) of BmmOp must be 3-dimensional in BmmOp, "
"but received X's shape: [%s].",
x_ndims));
PADDLE_ENFORCE_EQ(y_ndims,
3,
common::errors::InvalidArgument(
"Input(Y) of BmmOp must be 3-dimensional in BmmOp, "
"but received Y's shape: [%s].",
y_ndims));
std::vector<int64_t> dim_out;
auto cal_shape_fn = [](int64_t x, int64_t y, const std::string& error_str) {
if (x == -1) {
return y;
} else if (y == -1) {
return x;
}
PADDLE_ENFORCE_EQ(x, y, common::errors::InvalidArgument(error_str, x, y));
return x;
};
cal_shape_fn(x_dims[2],
y_dims[1],
"Input(X)'s width must be equal with Input(Y)'s height in BmmOp,"
"but receive X's width: [%s],"
"Y's height: [%s].");
dim_out.push_back(cal_shape_fn(
x_dims[0],
y_dims[0],
"Input(X) and Input(Y) must have the same batch size in BmmOp, "
"but received X's batch size: [%s],"
"Y's batch size [%s]"));
dim_out.push_back(x_dims[1]);
dim_out.push_back(y_dims[2]);
out->set_dims(make_ddim(dim_out));
out->share_lod(x);
out->set_dtype(x.dtype());
out->set_layout(x.layout());
}
void BoxClipInferMeta(const MetaTensor& input,
const MetaTensor& im_info,
MetaTensor* output,
MetaConfig config) {
const auto& input_box_dims = input.dims();
const auto& im_info_dims = im_info.dims();
if (config.is_runtime) {
auto input_box_size = input_box_dims.size();
PADDLE_ENFORCE_EQ(
input_box_dims[input_box_size - 1],
4,
common::errors::InvalidArgument(
"The last dimension of Input(Input) in BoxClipOp must be 4. "
"But received last dimension = %d",
input_box_dims[input_box_size - 1]));
PADDLE_ENFORCE_EQ(im_info_dims.size(),
2,
common::errors::InvalidArgument(
"The rank of Input(Input) in BoxClipOp must be 2."
" But received rank = %d",
im_info_dims.size()));
PADDLE_ENFORCE_EQ(
im_info_dims[1],
3,
common::errors::InvalidArgument(
"The last dimension of Input(ImInfo) of BoxClipOp must be 3. "
"But received last dimension = %d",
im_info_dims[1]));
}
output->set_dims(input.dims());
output->set_dtype(input.dtype());
output->share_lod(input);
}
void CholeskySolveInferMeta(const MetaTensor& x,
const MetaTensor& y,
bool upper,
MetaTensor* out) {
auto x_dims = x.dims();
auto y_dims = y.dims();
auto x_dims_n = x_dims.size();
auto y_dims_n = y_dims.size();
PADDLE_ENFORCE_GE(x_dims_n,
2,
common::errors::InvalidArgument(
"the rank of input Y must greater or equal to 2"));
PADDLE_ENFORCE_GE(y_dims_n,
2,
common::errors::InvalidArgument(
"the rank of input X must greater or equal to 2"));
PADDLE_ENFORCE_EQ(
y_dims[y_dims_n - 1],
y_dims[y_dims_n - 2],
common::errors::InvalidArgument("input Matrix Y should be square matrix,"
"But Got last shape of %ld x %ld",
y_dims[y_dims_n - 1],
y_dims[y_dims_n - 2]));
PADDLE_ENFORCE_EQ(x_dims[x_dims_n - 2],
y_dims[y_dims_n - 2],
common::errors::InvalidArgument(
"the first dim of Matrix X must be equal to "
"the first dim of Matrix Y,"
"But Got %ld and %ld",
x_dims[x_dims_n - 2],
y_dims[y_dims_n - 2]));
std::vector<int64_t> x_dims_vec = vectorize(x_dims);
std::vector<int64_t> y_dims_vec = vectorize(y_dims);
std::vector<int64_t> x_dims_vec_cut(x_dims_vec.begin(), x_dims_vec.end() - 2);
std::vector<int64_t> y_dims_vec_cut(y_dims_vec.begin(), y_dims_vec.end() - 2);
std::vector<int64_t> expand_batch_portion =
funcs::MatrixGetBroadcastBatchPortion(x_dims_vec_cut, y_dims_vec_cut);
std::vector<int64_t> x_broadcast_dims({expand_batch_portion});
x_broadcast_dims.insert(x_broadcast_dims.end(),
{x_dims_vec[x_dims_n - 2], x_dims_vec[x_dims_n - 1]});
// dim of 'out' is the same with 'X' after broadcast
out->set_dims(make_ddim(x_broadcast_dims));
out->set_dtype(x.dtype());
out->set_layout(x.layout());
out->share_lod(x);
}
void CompareRawInferMeta(const MetaTensor& x,
const MetaTensor& y,
int axis,
MetaTensor* out) {
auto dim_x = x.dims();
auto dim_y = y.dims();
if (dim_x == dim_y) {
out->share_meta(x);
} else {
int max_dim = std::max(dim_x.size(), dim_y.size());
int axis = std::abs(dim_x.size() - dim_y.size());
std::vector<int64_t> x_dims_array(max_dim);
std::vector<int64_t> y_dims_array(max_dim);
std::vector<int64_t> out_dims_array(max_dim);
funcs::GetBroadcastDimsArrays(dim_x,
dim_y,
x_dims_array.data(),
y_dims_array.data(),
out_dims_array.data(),
max_dim,
axis);
out->set_dims(make_ddim(out_dims_array));
out->share_lod(x);
}
if (!out->is_same_tensor(x)) {
out->set_dtype(DataType::BOOL);
}
}
void CompareInferMeta(const MetaTensor& x,
const MetaTensor& y,
MetaTensor* out) {
CompareRawInferMeta(x, y, -1, out);
}
void CompareAllInferMeta(const MetaTensor& x,
const MetaTensor& y,
MetaTensor* out) {
out->share_lod(x);
out->set_dims(make_ddim({}));
}
void ComplexInferMeta(const MetaTensor& x,
const MetaTensor& y,
MetaTensor* out) {
if (x.dims() == y.dims()) {
auto sizes = vectorize(x.dims());
out->set_dims(make_ddim(sizes));
out->set_dtype(dtype::ToComplex(x.dtype()));
// NOTE(chenfeiyu): lod & broadcasting is intrinsically contradictory
// so tensors with lod are not supported here
} else {
auto x_dims = x.dims();
auto y_dims = y.dims();
int max_dim = std::max(x_dims.size(), y_dims.size());
// start align axis
int axis = std::abs(x_dims.size() - y_dims.size());
std::vector<int64_t> x_dims_array(max_dim);
std::vector<int64_t> y_dims_array(max_dim);
std::vector<int64_t> out_dims_array(max_dim);
funcs::GetBroadcastDimsArrays(x_dims,
y_dims,
x_dims_array.data(),
y_dims_array.data(),
out_dims_array.data(),
max_dim,
axis);
out->set_dims(make_ddim(out_dims_array));
out->set_dtype(dtype::ToComplex(x.dtype()));
}
}
void ConvInferMeta(const MetaTensor& input,
const MetaTensor& filter,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const std::string& padding_algorithm,
const std::vector<int>& dilations,
int groups,
const std::string& data_format,
MetaTensor* out,
MetaConfig config) {
std::vector<int> paddings_ = paddings;
std::vector<int> dilations_ = dilations;
auto in_dims = input.dims();
auto filter_dims = filter.dims();
int dilation_size = static_cast<int>(dilations.size());
for (int i = 0; i < dilation_size; ++i) {
PADDLE_ENFORCE_GT(
dilations[i],
0,
common::errors::InvalidArgument(
"The dilation of Op(Conv) should be larger than 0, but received "
"dilation is %d.",
dilations[i]));
}
const bool channel_last = (config.is_run_onednn_kernel == false) &&
(data_format == "NHWC" || data_format == "NDHWC");
PADDLE_ENFORCE_EQ(
in_dims.size() == 4 || in_dims.size() == 5,
true,
common::errors::InvalidArgument(
"The input of Op(Conv) should be a 4-D or 5-D Tensor. But "
"received: input's dimension is %u, input's shape is [%s].",
in_dims.size(),
in_dims));
PADDLE_ENFORCE_EQ(
in_dims.size(),
filter_dims.size(),
common::errors::InvalidArgument(
"The input's dimension and filter's dimension of "
"Op(Conv) should be equal. But received: the input's shape is [%s], "
"the input's dimension is %d; the filter's shape is [%s], "
"the filter's dimension is %d.",
in_dims,
in_dims.size(),
filter_dims,
filter_dims.size()));
int stride_size = static_cast<int>(strides.size());
for (int i = 0; i < stride_size; ++i) {
PADDLE_ENFORCE_GT(
strides[i],
0,
common::errors::InvalidArgument(
"The stride of Op(Conv) should be larger than 0, but received "
"stride is %d.",
strides[i]));
}
int in_sub_stride_size = in_dims.size() - stride_size;
PADDLE_ENFORCE_EQ(
in_dims.size(),
strides.size() + 2U,
common::errors::InvalidArgument(
"The difference of input's dimension and Attr(strides)'s "
"length must be equal to 2 for Op(Conv). "
"But received: input's dimension is %d, input's shape is [%s]; "
"Attr(stride)'s length is %d, Attr(stride) is [%s]; "
"difference of input's dimension and Attr(strides)'s length = %u.",
in_dims.size(),
in_dims,
strides.size(),
make_ddim(strides),
in_sub_stride_size));
const auto input_channels =
channel_last ? in_dims[in_dims.size() - 1] : in_dims[1];
const auto filter_channels = channel_last && FLAGS_manually_trans_conv_filter
? filter_dims[filter_dims.size() - 1]
: filter_dims[1];
if (config.is_runtime) {
PADDLE_ENFORCE_EQ(
input_channels,
filter_channels * groups,
common::errors::InvalidArgument(
"The number of input's channels should be equal to filter's "
"channels "
"* groups for Op(Conv). But received: the input's channels is %d, "
"the input's shape is [%s]; the filter's channels is %d, the "
"filter's shape is [%s]; the groups is %d, the data_format is %s. "
"The error may come from wrong data_format setting.",
input_channels,
in_dims,
filter_channels,
filter_dims,
groups,
data_format));
PADDLE_ENFORCE_EQ(
filter_dims[0] % groups,
0,
common::errors::InvalidArgument(
"The number of output's channels (filter's first dimension) of "
"Op(Conv) should be divided by groups. But received: "
"the output channels is %d, the filter's shape is [%s], "
"the groups is %d.",
filter_dims[0],
filter_dims,
groups));
PADDLE_ENFORCE_GT(
filter_dims[0],
0,
common::errors::InvalidArgument(
"the size of filter at axis 0 should be greater than 0"));
for (int i = 2; i < filter_dims.size(); ++i) {
PADDLE_ENFORCE_GT(
filter_dims[i],
0,
common::errors::InvalidArgument(
"The kernel size of Op(Conv) should be greater than 0, but "
"received kernel size at dimension %d is %d. The filter's shape "
"is [%s].",
i,
filter_dims[i],
filter_dims));
}
}
DDim in_data_dims;
if (channel_last) {
in_data_dims = slice_ddim(in_dims, 1, in_dims.size() - 1);
} else {
in_data_dims = slice_ddim(in_dims, 2, in_dims.size());
}
DDim filter_data_dims;
if (channel_last && FLAGS_manually_trans_conv_filter) {
filter_data_dims = slice_ddim(filter_dims, 1, filter_dims.size() - 1);
} else {
filter_data_dims = slice_ddim(filter_dims, 2, filter_dims.size());
}
std::vector<int> ksize = vectorize<int>(filter_data_dims);
phi::UpdatePaddingAndDilation(
&paddings_, &dilations_, padding_algorithm, in_data_dims, strides, ksize);
std::vector<int64_t> output_shape({in_dims[0]});
if (!channel_last) {
if (filter_dims[1] == 0) {
output_shape.push_back(0);
} else {
output_shape.push_back(filter_dims[0]);
}
}
for (int i = 0; i < in_data_dims.size(); ++i) {
if ((!config.is_runtime) &&
(in_data_dims[i] < 0 || filter_dims[i + 2] < 0)) {
output_shape.push_back(-1);
} else {
const int64_t dkernel = dilations_[i] * (filter_data_dims[i] - 1) + 1;
int64_t output_size = (in_data_dims[i] + paddings_[2 * i] +
paddings_[2 * i + 1] - dkernel) /
strides[i] +
1;
output_shape.push_back(output_size);
}
}
if (channel_last) {
if (filter_dims[1] == 0) {
output_shape.push_back(0);
} else {
output_shape.push_back(filter_dims[0]);
}
}
out->set_dims(make_ddim(output_shape));
out->set_dtype(input.dtype());
}
void Conv3DInferMeta(const MetaTensor& input,
const MetaTensor& filter,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const std::string& padding_algorithm,
int groups,
const std::vector<int>& dilations,
const std::string& data_format,
MetaTensor* out,
MetaConfig config) {
ConvInferMeta(input,
filter,
strides,
paddings,
padding_algorithm,
dilations,
groups,
data_format,
out,
config);
}
void ConvTransposeInferMeta(const MetaTensor& x,
const MetaTensor& filter,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const std::vector<int>& output_padding,
const std::vector<int>& output_size,
const std::string& padding_algorithm,
int groups,
const std::vector<int>& dilations,
const std::string& data_format,
MetaTensor* out,
MetaConfig config) {
auto x_dims = x.dims();
auto filter_dims = filter.dims();
std::vector<int> paddings_ = paddings;
std::vector<int> dilations_ = dilations;
const DataLayout data_layout = config.is_run_onednn_kernel
? DataLayout::NCHW
: StringToDataLayout(data_format);
PADDLE_ENFORCE_EQ(
x_dims.size() == 4 || x_dims.size() == 5,
true,
errors::InvalidArgument("Input of Op(conv_transpose) should be 4-D or "
"5-D Tensor. But received: %u-D Tensor, "
"the shape of input is [%s]",
x_dims.size(),
x_dims));
PADDLE_ENFORCE_EQ(
x_dims.size(),
filter_dims.size(),
errors::InvalidArgument(
"The input's dimension size and filter's dimension size of "
"Op (conv_transpose) should be equal. But received: the shape of "
"input is [%s], the dimension size of input is [%d], the shape "
"of filter is [%s], the dimension size of filter is [%d]. ",
x_dims,
x_dims.size(),
filter_dims,
filter_dims.size()));
int stride_size = static_cast<int>(strides.size());
for (int i = 0; i < stride_size; ++i) {
PADDLE_ENFORCE_GT(
strides[i],
0,
errors::InvalidArgument(
"The stride of Op(Conv) should be larger than 0, but received "
"stride is %d.",
strides[i]));
}
int in_sub_stride_size = x_dims.size() - stride_size;
PADDLE_ENFORCE_EQ(
x_dims.size() - strides.size(),
2U,
errors::InvalidArgument(
"The input's dimension size minus Attr(stride)'s size must "
"be equal to 2 for Op(conv_transpose). But received: [%d], the "
"input's dimension size is [%d], the shape of input "
"is [%s], the Attr(stride)'s size is [%d].",
in_sub_stride_size,
x_dims.size(),
x_dims,
strides.size()));
if (!output_size.empty())
PADDLE_ENFORCE_EQ(
output_size.size(),
strides.size(),
errors::InvalidArgument(
"The Attr(output_size) and Attr(stride) of Op(conv_transpose) "
"should be the same."));
if (!output_padding.empty())
PADDLE_ENFORCE_EQ(
output_padding.size(),
strides.size(),
errors::InvalidArgument(
"The Attr(output_padding) and Attr(stride) of Op(conv_transpose) "
"should be the same."));
const int64_t C =
(data_layout != DataLayout::NHWC ? x_dims[1] : x_dims[x_dims.size() - 1]);
PADDLE_ENFORCE_EQ(
C,
filter_dims[0],
errors::InvalidArgument(
"The number of input channels should be equal to filter channels "
"for Op(conv_transpose). But received: the input's channels is "
"[%d], the shape of input is [%s], the filter's channels is [%d], "
"the shape of filter is [%s]. The data_format is %s."
"The error may come from wrong data_format setting.",
C,
x_dims,
filter_dims[0],
filter_dims,
data_format));
DDim x_data_dims;
if (data_layout != DataLayout::NHWC) {
x_data_dims = slice_ddim(x_dims, 2, x_dims.size());
} else {
x_data_dims = slice_ddim(x_dims, 1, x_dims.size() - 1);
}
DDim filter_data_dims = slice_ddim(filter_dims, 2, filter_dims.size());
std::vector<int> ksize = vectorize<int>(filter_data_dims);
UpdatePaddingAndDilation(
&paddings_, &dilations_, padding_algorithm, x_data_dims, strides, ksize);
std::vector<int64_t> output_shape({x_dims[0]});
if (data_layout != DataLayout::NHWC) {
output_shape.push_back(filter_dims[1] * groups);
}
const int offset = (data_layout != DataLayout::NHWC ? 2 : 1);
for (int i = 0; i < static_cast<int>(strides.size()); ++i) {
auto filter_extent = dilations_[i] * (filter_dims[i + 2] - 1) + 1;
auto infer_shape = (config.is_runtime || x_dims[i + offset] > 0)
? (x_dims[i + offset] - 1) * strides[i] -
paddings_[2 * i] - paddings_[2 * i + 1] +
filter_extent
: -1;
if (!output_size.empty()) {
if (config.is_runtime) {
PADDLE_ENFORCE_GE(
output_size[i],
infer_shape,
errors::InvalidArgument(
"output_size of Op(ConvTransposeOp) should not be less than "
"the inferred output size. But received output_size = [%s], "
"whose dim %d is less than the inferred output size [%s]",
make_ddim(output_size).to_str(),
i,
infer_shape));
if (common::product(x_dims) != 0) {
PADDLE_ENFORCE_LT(
output_size[i],
infer_shape + strides[i],
errors::InvalidArgument(
"output_size of Op(ConvTransposeOp) should be less "
"than inferred size + stride. But received output_size = "
"[%s], "
"whose dim %d is not less than the inferred output size (%d) "
"+ "
"stride (%d) = %d",
make_ddim(output_size).to_str(),
i,
infer_shape,
strides[i],
infer_shape + strides[i]));
}
}
output_shape.push_back(output_size[i]);
} else if (!output_padding.empty()) {
if (config.is_runtime) {
PADDLE_ENFORCE_GE(
output_padding[i],
0,
errors::InvalidArgument(
"output_padding of Op(ConvTransposeOp) should not be "
"less than the 0. But received output_padding = "
"[%s], whose dim %d is less than 0",
make_ddim(output_padding).to_str(),
i));
PADDLE_ENFORCE_LT(
output_padding[i],
std::max(strides[i], dilations_[i]),
errors::InvalidArgument(
"output_padding of Op(ConvTransposeOp) should be less "
"than either stride or dilation. But received output_size = "
"[%s], "
"whose dim %d is not less than either stride (%d) or "
"dilation (%d)",
make_ddim(output_size).to_str(),
i,
strides[i],
dilations_[i]));
}
output_shape.push_back((infer_shape + output_padding[i]));
} else {
output_shape.push_back(infer_shape);
}
}
if (data_layout == DataLayout::NHWC) {
output_shape.push_back(filter_dims[1] * groups);
}
out->set_dims(make_ddim(output_shape));
out->set_dtype(x.dtype());
}
void Conv2dTransposeInferMeta(const MetaTensor& x,
const MetaTensor& filter,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const std::vector<int>& output_padding,
const IntArray& output_size,
const std::string& padding_algorithm,
int groups,
const std::vector<int>& dilations,
const std::string& data_format,
MetaTensor* out,
MetaConfig config) {
std::vector<int32_t> vec_output_size(output_size.GetData().begin(),
output_size.GetData().end());
ConvTransposeInferMeta(x,
filter,
strides,
paddings,
output_padding,
vec_output_size,
padding_algorithm,
groups,
dilations,
data_format,
out,
config);
}
void CorrelationInferMeta(const MetaTensor& input1,
const MetaTensor& input2,
int pad_size,
int kernel_size,
int max_displacement,
int stride1,
int stride2,
int corr_type_multiply,
MetaTensor* out) {
auto in_dims = input1.dims();
auto in2_dims = input2.dims();
PADDLE_ENFORCE_EQ(in_dims.size() == 4,
true,
common::errors::InvalidArgument(
"Input(X) of CorrelationOp must be 4 dims."
"But received dims is %d.",
in_dims.size()));
PADDLE_ENFORCE_EQ(in2_dims.size() == 4,
true,
common::errors::InvalidArgument(
"Input(Y) of CorrelationOp must be 4 dims."
"But received dims is %d.",
in2_dims.size()));
PADDLE_ENFORCE_GT(stride1,
0,
common::errors::InvalidArgument(
"stride1 of CorrelationOp must be greater than 0. "
"But received stride1 = %d.",
stride1));
PADDLE_ENFORCE_GT(stride2,
0,
common::errors::InvalidArgument(
"stride2 of CorrelationOp must be greater than 0. "
"But received stride2 = %d.",
stride2));
std::vector<int64_t> output_shape = CorrelationOutputSize(in_dims[0],
in_dims[2],
in_dims[3],
stride1,
stride2,
kernel_size,
pad_size,
max_displacement);
out->set_dims(make_ddim(output_shape));
out->set_dtype(input1.dtype());
}
void CrossInferMeta(const MetaTensor& x,
const MetaTensor& y,
int axis,
MetaTensor* out) {
auto x_dim = x.dims();
auto y_dim = y.dims();
auto dim = axis;
bool dims_match = funcs::CheckDims(x_dim, y_dim);
PADDLE_ENFORCE_EQ(dims_match,
true,
common::errors::InvalidArgument(
"The 'shape' of Input(X) should be equal to "
"the 'shape' of Input(Y). But received "
"Input(X).dimensions = [%s], "
"Input(Y).dimensions = [%s]",
x_dim,
y_dim));
if (dim != DDim::kMaxRank) {
PADDLE_ENFORCE_EQ(
dim < x_dim.size() && dim >= (0 - x_dim.size()),
true,
common::errors::OutOfRange(
"Attr(dim) is out of range, It's expected "
"to be in range of [-%d, %d]. But received Attr(dim) = %d.",
x_dim.size(),
x_dim.size() - 1,
dim));
if (dim < 0) {
dim += x_dim.size();
}
PADDLE_ENFORCE_EQ(x_dim[dim] == 3 && y_dim[dim] == 3,
true,
common::errors::InvalidArgument(
"Input(X/Y).dims()[dim] should be equal to 3."
"But received Input(X/Y).dims()[dim] = %d.",
x_dim[dim]));
}
out->set_dims(x_dim);
out->set_dtype(x.dtype());
out->set_layout(x.layout());
out->share_lod(x);
}
void CrossEntropyInferMeta(const MetaTensor& x,
const MetaTensor& label,
bool soft_label,
int ignore_index,
MetaTensor* out,
MetaConfig config) {
const auto& x_dims = x.dims();
const auto& label_dims = label.dims();
int rank = x_dims.size();
bool contain_unknown_dim = common::contain_unknown_dim(x_dims) ||
common::contain_unknown_dim(label_dims);
bool check = config.is_runtime || !contain_unknown_dim;
if (check) {
PADDLE_ENFORCE_EQ(
slice_ddim(x_dims, 0, rank - 1),
slice_ddim(label_dims, 0, rank - 1),
common::errors::InvalidArgument(
"Input(X) and Input(Label) shall have the same shape "
"except the last dimension. But received: the shape of Input(X) "
"is "
"[%s], the shape of Input(Label) is [%s].",
x_dims,
label_dims));
}
if (soft_label) {
PADDLE_ENFORCE_EQ(
rank,
label_dims.size(),
common::errors::InvalidArgument(
"If Attr(soft_label) == true, Input(X) and Input(Label) "
"shall have the same dimensions. But received: the dimensions of "
"Input(X) is [%d],"
"the shape of Input(X) is [%s], the dimensions of Input(Label) "
"is [%d], the shape of "
"Input(Label) is [%s]",
rank,
x_dims,
label_dims.size(),
label_dims));
if (check) {
PADDLE_ENFORCE_EQ(
x_dims[rank - 1],
label_dims[rank - 1],
common::errors::InvalidArgument(
"If Attr(soft_label) == true, the last dimension of "
"Input(X) and Input(Label) should be equal. But received: the "
"last dimension of Input(X) is [%d], the shape of Input(X) is "
"[%s], "
"the last dimension of Input(Label) is [%d], the shape of "
"Input(Label) "
"is [%s], the last dimension is [%d].",
x_dims[rank - 1],
x_dims,
label_dims[rank - 1],
label_dims,
rank - 1));
}
} else {
if (rank == label_dims.size()) {
PADDLE_ENFORCE_EQ(
label_dims[rank - 1],
1UL,
common::errors::InvalidArgument(
"the last dimension of Input(Label) should be 1."
"But received: the last dimension of Input(Label) is [%d],"
"the last dimension is [%d]",
label_dims[rank - 1],
rank - 1));
} else {
PADDLE_ENFORCE_EQ(
rank,
label_dims.size() + 1,
common::errors::InvalidArgument(
"ShapeError: The rank of Input(X) should be equal to "
"Input(Label) plus 1."
"But received: The dimension of Input(X) is [%d], "
"the shape of Input(X) is [%s],"
"the dimension of Input(Label) is [%d], the shape of "
"Input(Label) is [%s]",
rank,
x_dims,
label_dims.size(),
label_dims));
}
}
auto y_dims = label_dims;
if (rank == label_dims.size()) {
y_dims[rank - 1] = 1;
}
out->set_dims(y_dims);
out->set_dtype(x.dtype());
out->share_lod(x);
}
void CrossEntropy2InferMeta(const MetaTensor& x,
const MetaTensor& label,
int ignore_index,
MetaTensor* out,
MetaTensor* x_shape,
MetaTensor* match_x,
MetaConfig config) {
CrossEntropyInferMeta(x, label, false, ignore_index, out);
auto x_dims = x.dims();
auto x_dims_vec = vectorize(x_dims);
x_dims_vec.push_back(0);
x_shape->set_dims(make_ddim(x_dims_vec));
x_dims[x_dims.size() - 1] = 1;
match_x->set_dims(x_dims);
x_shape->set_dtype(x.dtype());
match_x->set_dtype(x.dtype());
x_shape->share_lod(x);
}
void CrossEntropyWithSoftmaxInferMeta(const MetaTensor& logits,
const MetaTensor& label,
bool soft_label,
bool use_softmax,
bool numeric_stable_mode,
int ignore_index,
int axis,
MetaTensor* softmax,
MetaTensor* loss,
MetaConfig config) {
auto logits_dims = logits.dims();
auto labels_dims = label.dims();
auto logits_rank = logits_dims.size();
PADDLE_ENFORCE_GE(axis,
-logits_rank,
common::errors::InvalidArgument(
"Attr(axis) value should be in range [-R, R-1], "
"R is the rank of Input(Logits)."));
PADDLE_ENFORCE_LT(axis,
logits_rank,
common::errors::InvalidArgument(
"Attr(axis) value should be in range [-R, R-1], "
"R is the rank of Input(Logits)."));
axis = funcs::CanonicalAxis(axis, logits_rank);
for (int i = 0; i < logits_rank; i++) {
if (i != axis) {
if (config.is_runtime || (logits_dims[i] > 0 && labels_dims[i] > 0)) {
PADDLE_ENFORCE_EQ(logits_dims[i],
labels_dims[i],
common::errors::InvalidArgument(
"Input(Logits) and Input(Label) should in "
"same shape in dimensions except axis."));
}
}
}
if (axis != logits_rank - 1) {
PADDLE_ENFORCE_EQ(
numeric_stable_mode,
true,
common::errors::InvalidArgument("Attr(axis) can only be -1 "
"when not in numeric_stable_mode."));
}
if (soft_label) {
if (config.is_runtime || (logits_dims[axis] > 0 && labels_dims[axis] > 0)) {
PADDLE_ENFORCE_EQ(logits_dims[axis],
labels_dims[axis],
common::errors::InvalidArgument(
"If Attr(soft_label) == true, "
"the axis dimension of "
"Input(X) and Input(Label) should be equal."));
}
} else {
if (config.is_runtime || labels_dims[axis] > 0) {
PADDLE_ENFORCE_EQ(
labels_dims[axis],
1UL,
common::errors::InvalidArgument("If Attr(soft_label) == false, "
"the axis dimension of "
"Input(Label) should be 1."));
}
}
softmax->set_dims(logits_dims);
if (softmax->dtype() == DataType::BFLOAT16) {
softmax->set_dtype(DataType::FLOAT32);
} else {
softmax->set_dtype(logits.dtype());
}
logits_dims[axis] = 1;
loss->set_dims(logits_dims);
if (logits.dtype() == DataType::BFLOAT16) {
loss->set_dtype(DataType::FLOAT32);
} else {
loss->set_dtype(logits.dtype());
}
softmax->share_lod(logits);
loss->share_lod(logits);
}
void CSoftmaxWithCrossEntropyInferMeta(const MetaTensor& logits,
const MetaTensor& label,
int64_t ignore_index,
int rank,
int nranks,
MetaTensor* softmax,
MetaTensor* loss,
MetaConfig config) {
auto logits_dims = logits.dims();
auto labels_dims = label.dims();
auto logits_rank = logits_dims.size();
auto axis = logits_rank - 1;
for (int i = 0; i < logits_rank; i++) {
if (i != axis) {
if (config.is_runtime || (logits_dims[i] > 0 && labels_dims[i] > 0)) {
PADDLE_ENFORCE_EQ(logits_dims[i],
labels_dims[i],
common::errors::InvalidArgument(
"Input(Logits) and Input(Label) should in "
"same shape in dimensions except axis."));
}
}
}
PADDLE_ENFORCE_EQ(
labels_dims[logits_rank - 1],
1UL,
common::errors::InvalidArgument(
"the last dimension of Input(Label) should be 1."
"But received: the last dimension of Input(Label) is [%d],"
"the last dimension is [%d]",
labels_dims[logits_rank - 1],
logits_rank - 1));
softmax->set_dims(logits_dims);
logits_dims[axis] = 1;
loss->set_dims(logits_dims);
softmax->share_lod(logits);
loss->share_lod(logits);
}
void CtcAlignInferMeta(const MetaTensor& input,
const MetaTensor& input_length,
int blank,
bool merge_repeated,
int padding_value,
MetaTensor* output,
MetaTensor* output_length) {
auto input_dims = input.dims();
output->set_dims(input_dims);
if (input_length.initialized()) {
output_length->set_dims({input_dims[0], 1});
}
output->set_dtype(input.dtype());
}
void DepthwiseConvInferMeta(const MetaTensor& input,
const MetaTensor& filter,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const std::string& padding_algorithm,
int groups,
const std::vector<int>& dilations,
const std::string& data_format,
MetaTensor* out,
MetaConfig config) {
ConvInferMeta(input,
filter,
strides,
paddings,
padding_algorithm,
dilations,
groups,
data_format,
out,
config);
}
void DepthwiseConv2dBiasInferMeta(const MetaTensor& input,
const MetaTensor& filter,
const MetaTensor& bias,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const std::string& padding_algorithm,
int groups,
const std::vector<int>& dilations,
const std::string& data_format,
MetaTensor* out,
MetaConfig config) {
ConvInferMeta(input,
filter,
strides,
paddings,
padding_algorithm,
dilations,
groups,
data_format,
out,
config);
}
void DepthwiseConv3dBiasInferMeta(const MetaTensor& input,
const MetaTensor& filter,
const MetaTensor& bias,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const std::string& padding_algorithm,
int groups,
const std::vector<int>& dilations,
const std::string& data_format,
MetaTensor* out,
MetaConfig config) {
ConvInferMeta(input,
filter,
strides,
paddings,
padding_algorithm,
dilations,
groups,
data_format,
out,
config);
}
void CvmInferMeta(const MetaTensor& x,
const MetaTensor& cvm,
bool use_cvm,
MetaTensor* out) {
const auto& x_dims = x.dims();
PADDLE_ENFORCE_EQ(
x_dims.size(),
2UL,
common::errors::InvalidArgument("Input(X)'s rank should be 2, but got %d",
x_dims.size()));
if (use_cvm) {
out->set_dims({x_dims[0], x_dims[1]});
} else {
out->set_dims({x_dims[0], x_dims[1] - 2});
}
out->share_lod(x);
out->set_dtype(x.dtype());
}
void DequantizeAbsMaxInferMeta(const MetaTensor& x,
const MetaTensor& scale,
float max_range,
MetaTensor* out) {
out->set_dtype(x.dtype());
out->share_dims(x);
out->share_lod(x);
}
void DequantizeLogInferMeta(const MetaTensor& x,
const MetaTensor& dict,
MetaTensor* out) {
out->set_dtype(x.dtype());
out->share_dims(x);
out->share_lod(x);
}
void DistInferMeta(const MetaTensor& x,
const MetaTensor& y,
float p,
MetaTensor* out) {
out->set_dims(make_ddim({}));
out->set_dtype(x.dtype());
}
void DistributeLookupTableInferMeta(
const std::vector<const phi::MetaTensor*>& ids,
const MetaTensor& w,
int table_id,
bool is_distributed,
const std::string& lookup_table_version,
int64_t padding_idx,
DataType dtype,
bool is_test,
std::vector<MetaTensor*> outputs) {
auto table_dims = w.dims();
PADDLE_ENFORCE_EQ(w.dims().size(),
2,
errors::InvalidArgument(
"Only 2 dimensions of the 'Embedding' is supported."));
for (auto& id : ids) {
PADDLE_ENFORCE_EQ(id->dims().size(),
2,
errors::InvalidArgument(
"The dimension of the 'Ids' tensor must be 2."));
}
// for fluid.embedding
for (size_t i = 0; i < ids.size(); ++i) {
MetaTensor* output = outputs[i];
auto id_dims = ids[i]->dims();
if (lookup_table_version == "lookup_table") {
output->set_dims(make_ddim({id_dims[0], table_dims[1]}));
output->share_lod(*ids[i]);
output->set_dtype(w.dtype());
} else if (lookup_table_version == "lookup_table_v2") {
output->set_dims(make_ddim({static_cast<int64_t>(id_dims[0]),
static_cast<int64_t>(id_dims[1]),
static_cast<int64_t>(table_dims[1])}));
output->share_lod(*ids[i]);
output->set_dtype(w.dtype());
}
}
}
void DistributeFpnProposalsInferMeta(
const MetaTensor& fpn_rois,
const MetaTensor& rois_num,
int min_level,
int max_level,
int refer_level,
int refer_scale,
bool pixel_offset,
std::vector<MetaTensor*> multi_fpn_rois,
std::vector<MetaTensor*> multi_level_rois_num,
MetaTensor* restore_index,
MetaConfig config) {
PADDLE_ENFORCE_GE(
multi_fpn_rois.size(),
1UL,
errors::InvalidArgument("Outputs(MultiFpnRois) of "
"DistributeFpnProposalsOp should not be empty"));
PADDLE_ENFORCE_GE(
max_level,
min_level,
errors::InvalidArgument(
"max_level must not lower than "
"min_level. But received max_level = %d, min_level = %d",
max_level,
min_level));
// Set the output shape
for (auto& multi_fpn_roi : multi_fpn_rois) {
DDim out_dim = {-1, 4};
if (multi_fpn_roi == nullptr) {
continue;
}
multi_fpn_roi->set_dims(out_dim);
multi_fpn_roi->set_dtype(fpn_rois.dtype());
}
restore_index->set_dims({-1, 1});
restore_index->set_dtype(DataType::INT32);
for (auto& item : multi_level_rois_num) {
if (item == nullptr) {
continue;
}
item->set_dims({-1});
item->set_dtype(DataType::INT32);
}
if (!config.is_runtime) {
for (auto& multi_fpn_roi : multi_fpn_rois) {
multi_fpn_roi->share_lod(fpn_rois);
}
}
}
void DistributedFusedLambInitInferMeta(
const std::vector<const MetaTensor*>& param,
const std::vector<const MetaTensor*>& grad,
float beta1,
float beta2,
const std::vector<int>& apply_weight_decay,
int alignment,
int rank,
int nranks,
MetaTensor* fp32_fused_param,
MetaTensor* fp32_fused_grad,
MetaTensor* fp16_fused_param,
MetaTensor* fp16_fused_grad,
MetaTensor* moment1,
MetaTensor* moment2,
MetaTensor* beta1_pow,
MetaTensor* beta2_pow,
MetaTensor* fused_param_offsets,
MetaTensor* fp32_shard_fused_param_offsets,
MetaTensor* fp16_shard_fused_param_offsets,
MetaTensor* param_info,
MetaTensor* param_order,
std::vector<MetaTensor*> param_out,
std::vector<MetaTensor*> master_param_out,
std::vector<MetaTensor*> grad_out,
MetaTensor* global_scale,
MetaTensor* step) {
fp32_fused_param->set_dtype(DataType::FLOAT32);
fp32_fused_grad->set_dtype(DataType::FLOAT32);
fp16_fused_param->set_dtype(DataType::FLOAT16);
fp16_fused_grad->set_dtype(DataType::FLOAT16);
moment1->set_dtype(DataType::FLOAT32);
moment2->set_dtype(DataType::FLOAT32);
beta1_pow->set_dtype(DataType::FLOAT32);
beta2_pow->set_dtype(DataType::FLOAT32);
fused_param_offsets->set_dtype(DataType::INT32);
fp32_shard_fused_param_offsets->set_dtype(DataType::INT32);
fp16_shard_fused_param_offsets->set_dtype(DataType::INT32);
param_info->set_dtype(DataType::INT32);
param_order->set_dtype(DataType::INT32);
for (size_t i = 0; i < param.size(); ++i) {
param_out[i]->set_dtype(param[i]->dtype());
master_param_out[i]->set_dtype(DataType::FLOAT32);
}
for (size_t i = 0; i < grad.size(); ++i) {
grad_out[i]->set_dtype(grad[i]->dtype());
}
global_scale->set_dtype(DataType::FLOAT32);
step->set_dtype(DataType::INT64);
}
void DropoutInferMeta(const MetaTensor& x,
const MetaTensor& seed_tensor,
const Scalar& p,
bool is_test,
const std::string& mode,
int seed,
bool fix_seed,
MetaTensor* out,
MetaTensor* mask) {
auto x_dims = x.dims();
out->set_dims(x_dims);
out->share_lod(x);
out->set_dtype(x.dtype());
if (mask != nullptr) {
mask->set_dims(x_dims);
mask->set_dtype(DataType::UINT8);
}
}
void DropoutNdInferMeta(const MetaTensor& x,
const MetaTensor& seed_tensor,
const Scalar& p,
bool is_test,
const std::string& mode,
int seed,
bool fix_seed,
const std::vector<int>& axis,
MetaTensor* out,
MetaTensor* mask) {
auto x_dims = x.dims();
PADDLE_ENFORCE_LE(
axis.size(),
x_dims.size(),
common::errors::InvalidArgument(
"The length of axis is expected to be less than or equal to the "
"dimension size of x. But received the length of axis is %d, the "
"dimension size of x is %d, x's shape is {%s}.",
axis.size(),
x_dims.size(),
x_dims));
for (size_t i = 0; i < axis.size(); ++i) {
PADDLE_ENFORCE_EQ(
axis[i] >= 0 && axis[i] <= x_dims.size() - 1,
true,
common::errors::InvalidArgument(
"The %d-th value of axis is expected to be greater ot "
"equal to 0 and less than the dimensions of x. But "
"received axis is {%s}, the dimension size of x is %d.",
i,
make_ddim(axis),
x_dims.size()));
}
out->set_dims(x_dims);
out->share_lod(x);
out->set_dtype(x.dtype());
if (mask != nullptr) {
std::vector<int64_t> mask_dims(x.dims().size(), 1);
std::for_each(
axis.begin(), axis.end(), [&mask_dims, &x_dims](const int64_t& t) {
mask_dims[t] = x_dims[static_cast<int>(t)];
});
mask->set_dims(make_ddim(mask_dims));
mask->set_dtype(DataType::UINT8);
}
}
void DotInferMeta(const MetaTensor& x, const MetaTensor& y, MetaTensor* out) {
auto x_dims = x.dims();
int x_rank = static_cast<int>(x_dims.size());
PADDLE_ENFORCE_EQ(true,
1 == x_rank || 2 == x_rank,
common::errors::PreconditionNotMet(
"ShapeError: The dimensions of input tensor X (%s) "
"should be 1 or 2",
x_dims.to_str()));
auto y_dims = y.dims();
PADDLE_ENFORCE_EQ(
true,
x_rank == static_cast<int>(y_dims.size()),
common::errors::PreconditionNotMet(
"ShapeError: The shape of input tensor Y: %s should match with "
"input tensor X: %s",
y_dims.to_str(),
x_dims.to_str()));
bool shape_match = true;
for (int i = 0; i < x_rank; ++i) {
if (x_dims[i] == 0 || y_dims[i] == 0) {
continue;
}
if (x_dims[i] != y_dims[i]) {
shape_match = false;
break;
}
}
PADDLE_ENFORCE_EQ(true,
shape_match,
common::errors::PreconditionNotMet(
"ShapeError: The shape of input tensor X: %s should "
"be exactly the same "
"with input tensor Y: %s",
x_dims.to_str(),
y_dims.to_str()));
auto out_dims = x_dims;
// The output dims need to be modified.
if (x_rank == 2 && x_dims[0] != 0 && y_dims[0] == 0) {
out_dims[0] = 0;
}
std::vector<int64_t> out_dims_vec = vectorize(out_dims);
std::vector<int64_t> out_dims_vec_cut(out_dims_vec.begin(),
out_dims_vec.end() - 1);
out->set_dims(make_ddim(out_dims_vec_cut));
out->set_dtype(x.dtype());
out->set_layout(x.layout());
}
void ElementwiseInferMeta(const MetaTensor& x,
const MetaTensor& y,
MetaTensor* out) {
return ElementwiseRawInferMeta(x, y, -1, out);
}
void BitwiseShiftInferMeta(const MetaTensor& x,
const MetaTensor& y,
bool is_arithmetic,
MetaTensor* out) {
return ElementwiseRawInferMeta(x, y, -1, out);
}
void ElementwiseRawInferMeta(const MetaTensor& x,
const MetaTensor& y,
int axis,
MetaTensor* out,
MetaConfig config) {
if (x.dims() != y.dims()) {
auto x_dims = x.dims();
auto y_dims = y.dims();
int max_dim = std::max(x_dims.size(), y_dims.size());
if (x_dims.size() == y_dims.size()) {
PADDLE_ENFORCE_EQ((axis == -1) || (axis == 0),
true,
common::errors::InvalidArgument(
"axis should be -1 or 0 while the dimension of "
"tensor X (%s) is equal to the dimension of "
"tensor Y (%s), but received axis: %s",
x_dims.size(),
y_dims.size(),
axis));
}
PADDLE_ENFORCE_EQ((axis >= (-1 * max_dim)) && (axis < max_dim),
true,
common::errors::InvalidArgument(
"The axis range must be [%s, %s), but axis is %s. "
"Please set the axis again.",
-1 * max_dim,
max_dim,
axis));
axis = (axis < 0 ? (std::abs(x_dims.size() - y_dims.size()) + axis + 1)
: axis);
std::vector<int64_t> x_dims_array(max_dim);
std::vector<int64_t> y_dims_array(max_dim);
std::vector<int64_t> out_dims_array(max_dim);
#ifdef PADDLE_WITH_DNNL
bool should_rotate =
config.is_run_onednn_kernel &&
(phi::OneDNNContext::tls().get_cur_paddle_data_layout() ==
DataLayout::NHWC) &&
(x_dims.size() >= 3 || y_dims.size() >= 3);
if (should_rotate) {
// Pick bigger shape and rotate this one
bool x_over_y = (common::product(x_dims) > common::product(y_dims));
auto vdims =
x_over_y ? vectorize<int64_t>(x_dims) : vectorize<int64_t>(y_dims);
std::rotate(vdims.begin() + 1, vdims.begin() + 2, vdims.end());
if (x_over_y) {
x_dims = make_ddim(vdims);
} else {
y_dims = make_ddim(vdims);
}
}
#endif
funcs::GetBroadcastDimsArrays(x_dims,
y_dims,
x_dims_array.data(),
y_dims_array.data(),
out_dims_array.data(),
max_dim,
axis);
#ifdef PADDLE_WITH_DNNL
if (should_rotate) {
std::rotate(out_dims_array.begin() + 1,
out_dims_array.end() - 1,
out_dims_array.end());
}
#endif
auto out_dims = make_ddim(out_dims_array);
out->set_dims(out_dims);
} else {
out->set_dims(x.dims());
}
// dtype need promote when meet input dtype with more precision
paddle::experimental::DataTypeSet dtype_set{x.dtype()};
dtype_set = dtype_set | paddle::experimental::DataTypeSet(y.dtype());
DataType promote_result = PromoteTypes(dtype_set);
if (promote_result == DataType::UNDEFINED) {
promote_result = x.dtype();
}
out->set_dtype(promote_result);
// layout need change when meet input layout contain NHWC
auto layout = [&]() {
if (x.layout() == DataLayout::NHWC || y.layout() == DataLayout::NHWC)
return DataLayout::NHWC;
return x.layout();
}();
out->set_layout(layout);
out->share_lod(x);
}
void EmbeddingInferMeta(const MetaTensor& x,
const MetaTensor& weight,
int64_t padding_idx,
MetaTensor* out) {
const auto& table_dims = weight.dims();
const auto& ids_dims = x.dims();
int ids_rank = ids_dims.size();
VLOG(5) << "ids rank is " << ids_rank << std::endl;
PADDLE_ENFORCE_EQ(
table_dims.size(),
2,
common::errors::InvalidArgument(
"ShapeError: The dimensions of the 'lookup table' must be 2. "
"But received lookup table's dimensions = %d, "
"lookup table's shape = [%s].",
table_dims.size(),
table_dims));
auto output_dims = vectorize(ids_dims);
output_dims.push_back(table_dims[1]);
out->set_dims(make_ddim(output_dims));
out->set_dtype(weight.dtype());
out->share_lod(x);
}
void CEmbeddingInferMeta(const MetaTensor& weight,
const MetaTensor& x,
int64_t start_index,
MetaTensor* out) {
const auto& table_dims = weight.dims();
const auto& ids_dims = x.dims();
int ids_rank = ids_dims.size();
VLOG(5) << "ids rank is " << ids_rank << std::endl;
PADDLE_ENFORCE_EQ(
table_dims.size(),
2,
common::errors::InvalidArgument(
"ShapeError: The dimensions of the 'c_embedding' must be 2. "
"But received c_embedding's dimensions = %d, "
"c_embedding's shape = [%s].",
table_dims.size(),
table_dims));
auto output_dims = vectorize(ids_dims);
output_dims.push_back(table_dims[1]);
out->set_dims(make_ddim(output_dims));
out->set_dtype(weight.dtype());
out->share_lod(x);
const auto height = table_dims[0];
const auto width = table_dims[1];
PADDLE_ENFORCE_EQ(
(height > 0 && width > 0 && start_index >= 0),
true,
common::errors::InvalidArgument(
"height:%ld width:%ld start_index:%ld must not have negative values",
height,
width,
start_index));
}
void ExpandAsInferMeta(const MetaTensor& x,
const MetaTensor& y,
const std::vector<int64_t>& target_shape,
MetaTensor* out) {
#define MAX_RANK_SUPPORTED 8
auto x_dims = x.dims();
PADDLE_ENFORCE_GE(
target_shape.size(),
static_cast<size_t>(x_dims.size()),
common::errors::InvalidArgument(
"The rank of target_shape must be greater than or equal "
"to the rank of Input(X). But received Input(X): input "
"rank %u; received target_shape: rank %u.",
x_dims.size(),
target_shape.size()));
PADDLE_ENFORCE_LE(target_shape.size(),
MAX_RANK_SUPPORTED,
common::errors::InvalidArgument(
"The rank of target_shape must be less than or equal "
"to %d. But received: rank %u.",
MAX_RANK_SUPPORTED,
target_shape.size()));
out->set_dims(make_ddim(target_shape));
out->set_dtype(x.dtype());
#undef MAX_RANK_SUPPORTED
}
void FastRMSNormInfermeta(const MetaTensor& x,
const MetaTensor& scale,
float epsilon,
MetaTensor* y,
MetaTensor* invvar) {
auto x_dim = x.dims();
auto x_ndim = x_dim.size();
auto matrix_dim = flatten_to_2d(x_dim, x_ndim - 1);
int64_t right = matrix_dim[1];
if (scale) {
PADDLE_ENFORCE_EQ(scale.dims().size(),
1,
common::errors::InvalidArgument(
"The dimensions of Input(Scale) must be 1, but "
"received dimensions of "
"Input(Scale) is [%d]",
scale.dims().size()));
}
PADDLE_ENFORCE_EQ(
scale.dims()[0],
right,
common::errors::InvalidArgument(
"The first dimension value of Input(Scale) must equal to be the "
"second dimension value of the flattened 2D matrix of Input(X), "
"But received the first dimension value of Input(Scale) is "
"[%d], the second dimension value of the flattened 2D matrix of "
" Input(Scale) is [%d].",
scale.dims()[0],
right));
PADDLE_ENFORCE_EQ(epsilon >= 0.0f && epsilon <= 0.001f,
true,
common::errors::InvalidArgument(
"'epsilon' in Op(LayerNorm) should be between"
"0.0 and 0.001, But received [%s].",
epsilon));
DataType scale_dtype = scale.dtype();
y->set_dims(x_dim);
y->set_dtype(scale_dtype);
auto row_shape = slice_ddim(x_dim, 0, x_dim.size() - 1);
invvar->set_dims({row_shape});
invvar->set_dtype(DataType::FLOAT32);
}
void FakeDequantizeMaxAbsInferMeta(const MetaTensor& x,
const MetaTensor& scale,
float max_range,
MetaTensor* out) {
out->set_dtype(x.dtype());
out->share_dims(x);
out->share_lod(x);
}
void FillDiagonalTensorInferMeta(const MetaTensor& x,
const MetaTensor& y,
int64_t offset,
int dim1,
int dim2,
MetaTensor* out) {
PADDLE_ENFORCE_NOT_NULL(out,
common::errors::InvalidArgument(
"Output Tensor (out) should not be nullptr."));
out->set_dims(x.dims());
out->set_dtype(x.dtype());
}
void FusedDropoutAddInferMeta(const MetaTensor& x,
const MetaTensor& y,
MetaTensor* out,
MetaTensor* seed_offset) {
out->share_meta(x);
if (seed_offset) {
seed_offset->set_dims({2});
seed_offset->set_dtype(DataType::INT64);
}
}
// Used in FusedMatmulInferMeta
static std::vector<int64_t> GetInputShape(DDim dim,
std::vector<int> shape,
std::vector<int> axis) {
PADDLE_ENFORCE_GT(
dim.size(),
0,
common::errors::InvalidArgument(
"The input has not been initialized properly. The shape of input = "
"[%s].",
dim));
auto is_input_fused = (!shape.empty() && !axis.empty());
if (is_input_fused) {
dim = dim.reshape(shape).transpose(axis);
}
return vectorize(dim);
}
void FusedMatmulInferMeta(const MetaTensor& x,
const MetaTensor& y,
const MetaTensor& residual_data,
bool transpose_x,
bool transpose_y,
const float matmul_alpha,
const std::string& fuse_activation,
const float fuse_alpha,
const float fuse_beat,
const float fused_output_scale,
const std::vector<int>& fused_reshape_X,
const std::vector<int>& fused_transpose_X,
const std::vector<int>& fused_reshape_Y,
const std::vector<int>& fused_transpose_Y,
const std::vector<int>& fused_reshape_Out,
const std::vector<int>& fused_transpose_Out,
const std::string& onednn_data_type,
const float scale_x,
const float scale_y,
const float scale_scale_in_eltwise,
const float scale_out,
const bool force_fp32_output,
MetaTensor* out) {
std::vector<int64_t> dims_x =
GetInputShape(x.dims(), fused_reshape_X, fused_transpose_X);
std::vector<int64_t> dims_y =
GetInputShape(y.dims(), fused_reshape_Y, fused_transpose_Y);
auto ndims_x = dims_x.size();
auto ndims_y = dims_y.size();
PADDLE_ENFORCE_GT(ndims_x,
0,
common::errors::InvalidArgument(
"The Input(X) dims size must be greater than 0,"
" but received dims size is 0. "));
PADDLE_ENFORCE_GT(ndims_y,
0,
common::errors::InvalidArgument(
"The Input(Y) dims size must be greater than 0,"
" but received dims size is 0. "));
bool x_broadcasted = false;
bool y_broadcasted = false;
if (ndims_x == 1) {
dims_x.insert(dims_x.begin(), 1);
ndims_x = 2;
x_broadcasted = true;
}
if (ndims_y == 1) {
dims_y.push_back(1);
ndims_y = 2;
y_broadcasted = true;
}
size_t M = 0, N = 0;
if (transpose_x) {
M = dims_x[ndims_x - 1];
} else {
M = dims_x[ndims_x - 2];
}
if (transpose_y) {
N = dims_y[ndims_y - 2];
} else {
N = dims_y[ndims_y - 1];
}
std::vector<int64_t> new_dims;
if (ndims_x > ndims_y) {
new_dims.assign(dims_x.begin(), dims_x.end() - 2);
} else if (ndims_x < ndims_y) {
new_dims.assign(dims_y.begin(), dims_y.end() - 2);
} else {
new_dims.reserve(ndims_x);
for (size_t i = 0; i < ndims_x - 2; ++i) {
new_dims.push_back(std::max(dims_x[i], dims_y[i]));
}
}
if (!x_broadcasted) {
new_dims.push_back(M); // NOLINT
}
if (!y_broadcasted) {
new_dims.push_back(N); // NOLINT
}
if (x_broadcasted && y_broadcasted) {
new_dims.push_back(1);
}
auto ddim_out = make_ddim(new_dims);
std::vector<int> shape = fused_reshape_Out;
const std::vector<int>& axis = fused_transpose_Out;
auto is_output_fused = (!shape.empty() && !axis.empty());
if (is_output_fused) {
ddim_out = ddim_out.transpose(axis).reshape(shape);
}
out->set_dims(ddim_out);
bool is_int8 = (x.dtype() == DataType::UINT8 || x.dtype() == DataType::INT8);
bool is_bfloat16 = (x.dtype() == DataType::BFLOAT16);
bool fuse_relu = false;
if (fuse_activation == "relu" || fuse_activation == "relu6") {
fuse_relu = true;
}
if (force_fp32_output || ((!is_int8) && (!is_bfloat16))) {
out->set_dtype(DataType::FLOAT32);
} else if (is_bfloat16) {
out->set_dtype(DataType::BFLOAT16);
} else if (fuse_relu) {
out->set_dtype(DataType::UINT8);
} else {
out->set_dtype(DataType::INT8);
}
}
void GatherInferMeta(const MetaTensor& x,
const MetaTensor& index,
const Scalar& axis,
MetaTensor* out) {
auto index_dims = index.dims();
if (index_dims.size() == 2) {
if (index_dims[1] != 0) {
PADDLE_ENFORCE_EQ(
index_dims[1],
1,
common::errors::InvalidArgument("The last dim of index should be 0 "
"or 1 when it is 2D, but we get %d",
index_dims[1]));
}
} else {
PADDLE_ENFORCE_EQ(
index_dims.size() == 1 || index_dims.size() == 0,
true,
common::errors::InvalidArgument(
"The index should be 0D or 1D, when it is not 2D, but we get %d",
index_dims.size()));
}
auto input_dim = x.dims();
auto axis_v = axis.to<int>();
if (axis_v < 0) axis_v += static_cast<int>(input_dim.size());
PADDLE_ENFORCE_GE(
axis_v,
(0 - input_dim.size()),
common::errors::OutOfRange(
"Attr(axis) is out of range, It's expected "
"to be in range of [%d, %d]. But received Attr(axis) = %d.",
-input_dim.size(),
input_dim.size() - 1,
axis_v));
PADDLE_ENFORCE_LT(
axis_v,
input_dim.size(),
common::errors::OutOfRange(
"Attr(axis) is out of range, It's expected "
"to be in range of [%d, %d]. But received Attr(axis) = %d.",
-input_dim.size(),
input_dim.size() - 1,
axis_v));
if (index_dims.size() == 0) {
// 0D index will decrease the dimension
if (input_dim.size() == 1) {
// the index is a 0D tensor and the x is a 1D tensor
out->set_dims(DDim(phi::Dim<0>()));
out->set_dtype(x.dtype());
out->share_lod(x);
} else {
if (axis.FromTensor() || axis_v == 0) {
// decrease the output dimension
std::vector<int64_t> out_dim_vec;
for (int i = 1; i < input_dim.size(); ++i) {
out_dim_vec.emplace_back(input_dim[i]);
}
auto output_dims = make_ddim(out_dim_vec);
out->set_dims(output_dims);
out->set_dtype(x.dtype());
out->share_lod(x);
} else {
std::vector<int64_t> out_dim_vec;
for (int i = 0; i < axis_v; i++) {
out_dim_vec.push_back(input_dim[i]); // NOLINT
}
for (int i = axis_v + 1; i < input_dim.size(); i++) {
out_dim_vec.push_back(input_dim[i]); // NOLINT
}
auto output_dims = make_ddim(out_dim_vec);
out->set_dims(output_dims);
out->set_dtype(x.dtype());
out->share_lod(x);
}
}
} else {
if (axis.FromTensor() || axis_v == 0) {
// if axis.FromTensor(), we can not obtain correct shape of output
int64_t batch_size = static_cast<int64_t>(index_dims[0]);
if (index_dims.size() == 2 && index_dims[1] == 0) {
batch_size = 0;
}
DDim output_dims(input_dim);
output_dims[0] = batch_size;
out->set_dims(output_dims);
out->set_dtype(x.dtype());
out->share_lod(x);
} else {
int64_t index_size = static_cast<int64_t>(index_dims[0]);
if (index_dims.size() == 2 && index_dims[1] == 0) {
index_size = 0;
}
std::vector<int64_t> out_dim_vec;
for (int i = 0; i < axis_v; i++) {
out_dim_vec.push_back(input_dim[i]); // NOLINT
}
out_dim_vec.push_back(index_size);
for (int i = axis_v + 1; i < input_dim.size(); i++) {
out_dim_vec.push_back(input_dim[i]); // NOLINT
}
auto output_dims = make_ddim(out_dim_vec);
out->set_dims(output_dims);
out->set_dtype(x.dtype());
out->share_lod(x);
}
}
}
void GatherNdInferMeta(const MetaTensor& x,
const MetaTensor& index,
MetaTensor* out) {
auto x_dims = x.dims();
auto x_dims_size = x_dims.size();
auto index_dims = index.dims();
int index_dims_size = index_dims.size();
PADDLE_ENFORCE_LE(
index_dims[index_dims_size - 1],
x_dims_size,
common::errors::InvalidArgument(
"Input(Index).shape[-1] should be no greater than Input(X).rank"));
PADDLE_ENFORCE_GE(index_dims_size,
1UL,
common::errors::InvalidArgument(
"The rank of Input(Index) should be greater than 1"));
std::vector<int64_t> result_dims;
// The result dims is
// Index.shape[:-1] + X.shape[Index.shape[-1]:]
for (int i = 0; i < index_dims_size - 1; ++i) {
result_dims.emplace_back(index_dims[i]);
}
for (int64_t i = index_dims[index_dims_size - 1]; i < x_dims_size; ++i) {
result_dims.emplace_back(x_dims[i]);
}
out->set_dims(make_ddim(result_dims));
out->share_lod(x);
out->set_dtype(x.dtype());
}
void GatherTreeMeta(const MetaTensor& ids,
const MetaTensor& parents,
MetaTensor* out) {
auto ids_dims = ids.dims();
auto parents_dims = parents.dims();
if (common::product(ids_dims) != 0) {
PADDLE_ENFORCE_EQ(ids_dims == parents_dims,
true,
common::errors::InvalidArgument(
"The shape of Input(Parents) must be same with the "
"shape of Input(Ids)."));
}
out->set_dims(ids_dims);
out->set_dtype(ids.dtype());
}
void GridSampleBaseInferMeta(const MetaTensor& x,
const MetaTensor& grid,
MetaTensor* out,
MetaConfig config) {
auto x_dims = x.dims();
auto grid_dims = grid.dims();
PADDLE_ENFORCE_GE(x_dims.size(),
4,
common::errors::InvalidArgument(
"Input(X) of GridSampleOp should be 4-D Tensor, but "
"received X dimension size(%d)",
x_dims.size()));
PADDLE_ENFORCE_LE(x_dims.size(),
5,
common::errors::InvalidArgument(
"Input(X) of GridSampleOp should be 4-D Tensor, but "
"received X dimension size(%d)",
x_dims.size()));
PADDLE_ENFORCE_GE(grid_dims.size(),
4,
common::errors::InvalidArgument(
"Input(Grid) of GridSampleOp should be 4-D Tensor, "
"but received X dimension size(%d)",
grid_dims.size()));
PADDLE_ENFORCE_LE(grid_dims.size(),
5,
common::errors::InvalidArgument(
"Input(Grid) of GridSampleOp should be 4-D Tensor, "
"but received X dimension size(%d)",
grid_dims.size()));
if (grid_dims.size() == 4 && (config.is_runtime || grid_dims[3] > 0)) {
PADDLE_ENFORCE_EQ(
grid_dims[3],
2,
common::errors::InvalidArgument(
"Input(Grid) dimension[3] should be 2, but received %d",
grid_dims[3]));
}
if (grid_dims.size() == 5 && (config.is_runtime || grid_dims[4] > 0)) {
PADDLE_ENFORCE_EQ(
grid_dims[4],
3,
common::errors::InvalidArgument(
"Input(Grid) dimension[4] should be 3, but received %d",
grid_dims[4]));
}
if (config.is_runtime) {
PADDLE_ENFORCE_EQ(
grid_dims[0],
x_dims[0],
common::errors::InvalidArgument(
"Input(X) and Input(Grid) dimension[0] should be equal, but "
"received X dimension[0](%d) != Grid dimension[0](%d)",
x_dims[0],
grid_dims[0]));
}
if (grid_dims.size() == 4) {
out->set_dims({x_dims[0], x_dims[1], grid_dims[1], grid_dims[2]});
} else {
out->set_dims(
{x_dims[0], x_dims[1], grid_dims[1], grid_dims[2], grid_dims[3]});
}
out->set_dtype(x.dtype());
out->share_lod(x);
}
void HingeLossInferMeta(const MetaTensor& logits,
const MetaTensor& labels,
MetaTensor* loss) {
const auto& pred_dims = logits.dims();
const auto& label_dims = labels.dims();
PADDLE_ENFORCE_EQ(
pred_dims,
label_dims,
common::errors::InvalidArgument(
"The Input(input) and Input(label) should have the same "
"shape, but received input shape [%s] != label shape [%s]",
pred_dims,
label_dims));
PADDLE_ENFORCE_EQ(
pred_dims.size(),
2,
common::errors::InvalidArgument("Input(input) rank should be 2, "
"but received input rank(%d) != 2",
pred_dims.size()));
PADDLE_ENFORCE_EQ(pred_dims[1],
1,
common::errors::InvalidArgument(
"The second dimension of Input(input) should be 1, "
"as each row of input contains a real value, "
"but received second dimension of input (%d) != 1",
pred_dims[1]));
loss->set_dims({pred_dims[0], 1});
loss->share_lod(logits);
loss->set_dtype(logits.dtype());
}
void HistogramInferMeta(const MetaTensor& input,
const MetaTensor& weight,
int64_t bins,
float min,
float max,
bool density,
MetaTensor* out) {
PADDLE_ENFORCE_GE(bins,
1,
common::errors::InvalidArgument(
"The bins should be greater than or equal to 1."
"But received nbins is %d",
bins));
PADDLE_ENFORCE_GE(
max,
min,
common::errors::InvalidArgument("max must be larger or equal to min."
"But received max is %f, min is %f",
max,
min));
if (weight) {
auto weight_dims = weight.dims();
PADDLE_ENFORCE_EQ(
weight_dims,
input.dims(),
common::errors::InvalidArgument(
"The shape of weight should be equal to the shape of input."
"But received weight shape is [%s], input shape is [%s]",
weight_dims,
input.dims()));
}
out->set_dims({bins});
out->share_lod(input);
if (density || weight) {
out->set_dtype(DataType::FLOAT32);
} else {
out->set_dtype(DataType::INT64);
}
}
void HuberLossInferMeta(const MetaTensor& input,
const MetaTensor& label,
float delta,
MetaTensor* out,
MetaTensor* residual,
MetaConfig config) {
auto input_dims = input.dims();
auto label_dims = label.dims();
PADDLE_ENFORCE_EQ(input_dims.size(),
label_dims.size(),
common::errors::InvalidArgument(
"Input(input) rank and Input(label) rank should be "
"same, but received input rank(%d) != label rank(%d)",
input_dims.size(),
label_dims.size()));
bool contain_unknown_dim = common::contain_unknown_dim(input_dims) ||
common::contain_unknown_dim(label_dims);
if (config.is_runtime || !contain_unknown_dim) {
PADDLE_ENFORCE_EQ(
input_dims,
label_dims,
common::errors::InvalidArgument(
"The Input(input) and Input(label) should have the same "
"shape, but received input shape [%s] != label shape [%s]",
input_dims,
label_dims));
}
auto out_dims = label_dims;
residual->set_dims(out_dims);
out->set_dims(out_dims);
out->set_dtype(input.dtype());
residual->set_dtype(input.dtype());
out->share_lod(input);
}
void IdentityLossGradInferMeta(const MetaTensor& x,
const MetaTensor& out_grad,
const int reduction,
MetaTensor* x_grad) {
x_grad->set_dims(x.dims());
x_grad->share_lod(x);
x_grad->set_dtype(out_grad.dtype());
}
void IndexSampleInferMeta(const MetaTensor& x,
const MetaTensor& y,
MetaTensor* out,
MetaConfig config) {
auto input_dims = x.dims();
PADDLE_ENFORCE_EQ(input_dims.size(),
2,
errors::InvalidArgument(
"Inputs(X) shape of IndexSample op should be 2-D, but "
"got X's shape = [%s], please check X shape.",
input_dims));
auto index_dims = y.dims();
PADDLE_ENFORCE_EQ(
index_dims.size(),
2,
errors::InvalidArgument(
"Inputs(Index) shape of IndexSample op should be 2-D, but "
"got Index's shape [%s] , please check index shape.",
input_dims));
if (config.is_runtime && index_dims[0] != 0) { // 0-size not check
PADDLE_ENFORCE_EQ(input_dims[0],
index_dims[0],
errors::InvalidArgument(
"Inputs(X)'s value of dimension 0 must same with "
"Inputs(Index)'s value of dimension 0, but "
"got %d of Inputs(X), and got %d of Inputs(Index), "
"please check Inputs shape.",
input_dims[0],
index_dims[0]));
}
out->set_dtype(x.dtype());
out->set_dims(index_dims);
out->share_lod(y);
}
void Im2sequenceInferMeta(const MetaTensor& x,
const MetaTensor& y,
const std::vector<int>& kernels,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const std::vector<int>& out_stride,
MetaTensor* out,
MetaConfig config) {
PADDLE_ENFORCE_EQ(
x.initialized(),
true,
common::errors::NotFound("The input 'X' of Im2SequenceOp is not found."));
PADDLE_ENFORCE_EQ(out != nullptr,
true,
common::errors::NotFound(
"The output 'Out' of Im2SequenceOp is not found."));
const auto& in_dim = x.dims();
PADDLE_ENFORCE_EQ(in_dim.size(),
4,
common::errors::InvalidArgument(
"The dimensions size of input 'X' in Im2SequenceOp "
"should be 4. But "
"received dimensions size=[%d], dimensions=[%s].",
in_dim.size(),
in_dim));
auto img_channels = in_dim[1];
out->set_dims({in_dim[0], img_channels * kernels[0] * kernels[1]});
out->set_dtype(x.dtype());
}
void IndexSelectInferMeta(const MetaTensor& x,
const MetaTensor& index,
int dim,
MetaTensor* output) {
auto input_dim = x.dims();
auto index_dim = index.dims();
PADDLE_ENFORCE_EQ(
dim < input_dim.size() && dim >= (0 - input_dim.size()),
true,
common::errors::OutOfRange(
"Attr(dim) is out of range, It's expected "
"to be in range of [-%d, %d]. But received Attr(dim) = %d.",
input_dim.size(),
input_dim.size() - 1,
dim));
PADDLE_ENFORCE_EQ(
index_dim.size() == 1 || (index_dim.size() == 2 && index_dim[1] == 1),
true,
common::errors::InvalidArgument(
"The 'shape' of Input(Index) must be 1-D tensor. "
"But received: the 'shape' of Input(Index) is [%s], "
"the dimension of Input(Index) is [%d].",
index_dim,
index_dim.size()));
if (dim < 0) {
dim += input_dim.size();
}
auto output_dim = vectorize(input_dim);
output_dim[dim] = index_dim[0];
output->set_dims(make_ddim(output_dim));
output->set_dtype(x.dtype());
output->set_layout(x.layout());
output->share_lod(x);
}
void IndexSelectStridedInferMeta(const MetaTensor& x,
int64_t index,
int dim,
MetaTensor* output) {
auto input_dim = x.dims();
PADDLE_ENFORCE_EQ(
dim < input_dim.size() && dim >= (0 - input_dim.size()),
true,
common::errors::OutOfRange(
"Attr(dim) is out of range, It's expected "
"to be in range of [-%d, %d]. But received Attr(dim) = %d.",
input_dim.size(),
input_dim.size() - 1,
dim));
auto output_dim = vectorize(input_dim);
if (dim < 0) {
dim += input_dim.size();
}
output_dim.erase(output_dim.begin() + dim);
output->set_dims(make_ddim(output_dim));
output->set_dtype(x.dtype());
output->set_layout(x.layout());
output->share_lod(x);
}
void IndexAddInferMeta(const MetaTensor& x,
const MetaTensor& index,
const MetaTensor& add_value,
int axis,
MetaTensor* output) {
auto input_dim = x.dims();
if (common::product(input_dim) == 0) {
output->set_dims(input_dim);
output->set_dtype(x.dtype());
output->set_layout(x.layout());
return;
}
if (index.dims().size() == 1 && index.dims()[0] == 0) {
output->set_dims(input_dim);
output->set_dtype(x.dtype());
output->set_layout(x.layout());
output->share_lod(x);
return;
}
auto index_dim = index.dims();
auto add_value_dim = add_value.dims();
PADDLE_ENFORCE_EQ(
axis < input_dim.size() && axis >= (0 - input_dim.size()),
true,
common::errors::OutOfRange(
"Attr(dim) is out of range, It's expected "
"to be in range of [-%d, %d]. But received Attr(axis) = %d.",
input_dim.size(),
input_dim.size() - 1,
axis));
int real_axis = axis >= 0 ? axis : axis + input_dim.size();
PADDLE_ENFORCE_EQ(index_dim.size() == 1,
true,
common::errors::InvalidArgument(
"The 'shape' of Input(Index) must be 1-D tensor. "
"But received: the 'shape' of Input(Index) is [%s], "
"the dimension of Input(Index) is [%d].",
index_dim,
index_dim.size()));
if (common::product(add_value_dim) == 0) {
output->set_dims(input_dim);
output->set_dtype(x.dtype());
output->set_layout(x.layout());
output->share_lod(x);
return;
}
// Note, add_value does not support broadcast now.
PADDLE_ENFORCE_EQ(input_dim.size() == add_value_dim.size(),
true,
common::errors::InvalidArgument(
"The add_value must be the same dimension as x."));
for (int i = 0; i < input_dim.size(); i++) {
if (i != real_axis) {
PADDLE_ENFORCE_EQ(input_dim[i] == add_value_dim[i],
true,
common::errors::InvalidArgument(
"The add_value parameter does not supported "
"broadcast, so input_dim[i](%d) must be equal to "
"add_value_dim[i](%d) when i(%d) != axis(%d).",
input_dim[i],
add_value_dim[i],
i,
real_axis));
}
}
const auto& index_type = index.dtype();
bool index_type_match =
index_type == DataType::INT64 || index_type == DataType::INT32;
PADDLE_ENFORCE_EQ(index_type_match,
true,
common::errors::InvalidArgument(
"Input(Index) holds the wrong type, it holds %s, but "
"desires to be %s or %s",
index_type,
DataType::INT32,
DataType::INT64));
output->set_dims(x.dims());
output->set_dtype(x.dtype());
output->set_layout(x.layout());
output->share_lod(x);
}
void IndexElementwisePutInferMeta(const MetaTensor& x,
const std::vector<const MetaTensor*>& index,
const Scalar& value,
const std::vector<int64_t>& input_dims,
const std::vector<int64_t>& input_strides,
const std::vector<int64_t>& index_dims,
const std::vector<int64_t>& index_strides,
const int64_t slice_offset,
MetaTensor* out) {
out->set_dims(x.dims());
out->set_dtype(x.dtype());
}
void IndexElementwisePutWithTensorInferMeta(
const MetaTensor& x,
const std::vector<const MetaTensor*>& index,
const MetaTensor& value,
const std::vector<int64_t>& input_dims,
const std::vector<int64_t>& input_strides,
const std::vector<int64_t>& index_dims,
const std::vector<int64_t>& index_strides,
const int64_t slice_offset,
MetaTensor* out) {
out->set_dims(x.dims());
out->set_dtype(x.dtype());
}
void IndexElementwiseGetInferMeta(const MetaTensor& x,
const std::vector<const MetaTensor*>& index,
const std::vector<int64_t>& input_dims,
const std::vector<int64_t>& input_strides,
const std::vector<int64_t>& index_dims,
const std::vector<int64_t>& index_stride,
const int64_t slice_offset,
const bool accumulate,
const bool is_combined,
MetaTensor* out) {
out->set_dims(make_ddim(input_dims));
out->set_dtype(x.dtype());
}
void KronInferMeta(const MetaTensor& x, const MetaTensor& y, MetaTensor* out) {
auto dim_x = x.dims();
auto dim_y = y.dims();
auto rank_x = dim_x.size();
auto rank_y = dim_y.size();
auto rank = (rank_x > rank_y) ? rank_x : rank_y;
std::vector<int64_t> dim_out;
dim_out.reserve(rank);
for (int i = 0; i < rank; i++) {
int64_t dim_xi = (i < rank - rank_x) ? 1 : dim_x.at(i - (rank - rank_x));
int64_t dim_yi = (i < rank - rank_y) ? 1 : dim_y.at(i - (rank - rank_y));
dim_out.push_back(dim_xi == -1 || dim_yi == -1 ? -1 : dim_xi * dim_yi);
}
out->set_dims(make_ddim(dim_out));
out->set_dtype(x.dtype());
}
void LegacyCropInferMeta(const MetaTensor& x,
const MetaTensor& y,
const IntArray& offsets,
const std::vector<int>& shape,
MetaTensor* out) {
const auto& x_dim = x.dims();
if (!y.initialized()) {
PADDLE_ENFORCE_EQ(
int64_t(shape.size()),
x_dim.size(),
common::errors::InvalidArgument(
"The number of elements (%d) of CropOp's "
"'shape' attribute should be equal to the number of dimensions "
"(%d) of the Input(X).",
shape.size(),
x_dim.size()));
std::vector<int64_t> tensor_shape(shape.size());
for (size_t i = 0; i < shape.size(); ++i) {
tensor_shape[i] = static_cast<int64_t>(shape[i]);
}
out->set_dims(make_ddim(tensor_shape));
out->set_dtype(x.dtype());
} else {
const auto& y_dim = y.dims();
PADDLE_ENFORCE_EQ(common::arity(x_dim),
common::arity(y_dim),
common::errors::InvalidArgument(
"The number of dimensions (%d) of CropOp's input(X)"
" must be equal to that (%d) of input(Y).",
common::arity(x_dim),
common::arity(y_dim)));
out->set_dims(y_dim);
out->set_dtype(y.dtype());
}
}
void LimitByCapacityInferMeta(const MetaTensor& expert_count,
const MetaTensor& capacity,
int n_worker,
MetaTensor* out) {
out->share_dims(expert_count);
out->share_lod(expert_count);
out->set_dtype(expert_count.dtype());
}
void LodResetInferMeta(const MetaTensor& x,
const MetaTensor& y,
const std::vector<int>& target_lod,
bool append,
MetaTensor* out,
MetaConfig config) {
if (y.initialized()) {
auto level0 = target_lod;
PADDLE_ENFORCE_GT(
static_cast<int64_t>(level0.size()),
0,
common::errors::InvalidArgument(
"If Input(Y) is not provided, the output's LoD should be "
"specified by attribute 'target_lod'. But the size of "
"'target_lod' is 0."));
} else if (config.is_runtime) {
out->share_lod(y);
}
if (append) {
out->share_lod(x);
}
out->set_dims(x.dims());
out->set_dtype(x.dtype());
}
void LogLossInferMeta(const MetaTensor& input,
const MetaTensor& label,
float epsilon,
MetaTensor* out,
MetaConfig config) {
auto pred_dims = input.dims();
auto label_dims = label.dims();
if (config.is_runtime ||
(common::product(pred_dims) > 0 && common::product(label_dims) > 0)) {
PADDLE_ENFORCE_EQ(
pred_dims,
label_dims,
common::errors::InvalidArgument(
"The dimensions of Input(Predicted) must be equal to the "
"dimensions of Input(Labels), but received dimensions of "
"Input(Predicted) "
"is [%s], received dimensions of Input(Labels) is [%s].",
pred_dims,
label_dims));
}
PADDLE_ENFORCE_EQ(pred_dims.size(),
2,
common::errors::InvalidArgument(
"The dimensions of Input(Predicted) must be 2,"
"But received dimensions of Input(Predicted)"
"is [%d]",
pred_dims.size()));
if (config.is_runtime) {
PADDLE_ENFORCE_EQ(pred_dims[1],
1,
common::errors::InvalidArgument(
"Each row of Input(Predicted) contains a real value, "
"so the 2nd dimension of Input(X) must be 1,"
"But got [%d]",
pred_dims[1]));
}
out->set_dims({pred_dims[0], 1});
out->set_dtype(input.dtype());
out->share_lod(input);
}
void LookupTableDequantInferMeta(const MetaTensor& w,
const MetaTensor& ids,
int64_t padding_idx,
MetaTensor* out) {
PADDLE_ENFORCE_EQ(
w.initialized(),
true,
common::errors::InvalidArgument(
"Input(W) of LookupTableDequantOp should not be null."));
PADDLE_ENFORCE_EQ(
ids.initialized(),
true,
common::errors::InvalidArgument(
"Input(Ids) of LookupTableDequantOp should not be null."));
PADDLE_ENFORCE_EQ(
out != nullptr,
true,
common::errors::InvalidArgument(
"Output(Out) of LookupTableDequantOp should not be null."));
const auto& table_dims = w.dims();
const auto& ids_dims = ids.dims();
int ids_rank = ids_dims.size();
VLOG(5) << "ids rank is " << ids_rank << std::endl;
PADDLE_ENFORCE_EQ(
table_dims.size(),
2,
common::errors::InvalidArgument(
"ShapeError: The dimensions of the 'lookup table' must be 2. "
"But received lookup table's dimensions = %d, "
"lookup table's shape = [%s].",
table_dims.size(),
table_dims));
PADDLE_ENFORCE_EQ(
ids_dims[ids_rank - 1],
1,
common::errors::InvalidArgument(
"ShapeError: The last dimensions of the 'Ids' tensor must be 1. "
"But received Ids's last dimensions = %d, Ids's shape = [%s].",
ids_dims[ids_rank - 1],
ids_dims));
auto output_dims = vectorize(slice_ddim(ids_dims, 0, ids_rank - 1));
PADDLE_ENFORCE_GE(table_dims[1],
2,
common::errors::InvalidArgument(
"the second dim of table_dims should be "
"greater or equal to 2, but the actual shape "
"is [%s]",
table_dims));
output_dims.push_back((table_dims[1] - 2) * 4);
out->set_dims(make_ddim(output_dims));
out->share_lod(ids);
out->set_dtype(w.dtype());
}
void LogicalBinaryInferMeta(const MetaTensor& x,
const MetaTensor& y,
MetaTensor* out) {
ElementwiseInferMeta(x, y, out);
if (!(out->is_same_tensor(x))) {
out->set_dtype(DataType::BOOL);
}
}
void LUUnpackInferMeta(const MetaTensor& x,
const MetaTensor& pivots,
bool unpack_ludata,
bool unpack_pivots,
MetaTensor* pmat,
MetaTensor* l,
MetaTensor* u) {
PADDLE_ENFORCE_NOT_NULL(
pmat,
common::errors::InvalidArgument("Output(Pmat) should not be nullptr."));
PADDLE_ENFORCE_NOT_NULL(
l, common::errors::InvalidArgument("Output(L) should not be nullptr."));
PADDLE_ENFORCE_NOT_NULL(
u, common::errors::InvalidArgument("Output(U) should not be nullptr."));
auto x_dims = x.dims();
int x_rank = x_dims.size();
PADDLE_ENFORCE_GE(x_rank,
2,
common::errors::InvalidArgument(
"The rank of input must greater than 2."));
int64_t m = x_dims[x_rank - 2];
int64_t n = x_dims[x_rank - 1];
int64_t min_mn = std::min(m, n);
if (unpack_ludata) {
auto ldims = x_dims;
auto udims = x_dims;
if (m >= n) {
udims[x_rank - 2] = min_mn;
} else {
ldims[x_rank - 1] = min_mn;
}
u->set_dims(udims);
u->set_dtype(x.dtype());
l->set_dims(ldims);
l->set_dtype(x.dtype());
}
if (unpack_pivots) {
auto pdims = x_dims;
pdims[x_rank - 1] = m;
pmat->set_dims(pdims);
pmat->set_dtype(x.dtype());
}
}
void LookupTableInferMeta(const MetaTensor& w,
const MetaTensor& ids,
MetaTensor* out) {
const auto& table_dims = w.dims();
const auto& ids_dims = ids.dims();
int ids_rank = ids_dims.size();
VLOG(5) << "ids rank is " << ids_rank << std::endl;
PADDLE_ENFORCE_EQ(
table_dims.size(),
2,
common::errors::InvalidArgument(
"ShapeError: The dimensions of the 'lookup table' must be 2. "
"But received lookup table's dimensions = %d, "
"lookup table's shape = [%s].",
table_dims.size(),
table_dims));
PADDLE_ENFORCE_EQ(
ids_dims[ids_rank - 1],
1,
common::errors::InvalidArgument(
"ShapeError: The last dimensions of the 'Ids' tensor must be 1. "
"But received Ids's last dimensions = %d, Ids's shape = [%s].",
ids_dims[ids_rank - 1],
ids_dims));
auto output_dims = vectorize(slice_ddim(ids_dims, 0, ids_rank - 1));
output_dims.push_back(table_dims[1]);
out->set_dims(make_ddim(output_dims));
out->set_dtype(w.dtype());
out->share_lod(ids);
}
void MarginCrossEntropyInferMeta(const MetaTensor& logits,
const MetaTensor& label,
bool return_softmax,
int ring_id,
int rank,
int nranks,
float margin1,
float margin2,
float margin3,
float scale,
MetaTensor* softmax,
MetaTensor* loss,
MetaConfig config) {
PADDLE_ENFORCE_NOT_NULL(
logits,
common::errors::InvalidArgument("Input of logits should not be null."));
PADDLE_ENFORCE_NOT_NULL(
label,
common::errors::InvalidArgument("Input of label should not be null."));
auto logits_dims = logits.dims();
auto labels_dims = label.dims();
auto logits_rank = logits_dims.size();
auto axis = logits_rank - 1;
for (int i = 0; i < logits_rank; i++) {
if (i != axis) {
if (config.is_runtime || (logits_dims[i] > 0 && labels_dims[i] > 0)) {
PADDLE_ENFORCE_EQ(logits_dims[i],
labels_dims[i],
common::errors::InvalidArgument(
"Input(Logits) and Input(Label) should in "
"same shape in dimensions except axis."));
}
}
}
if (labels_dims.size() > 1) {
PADDLE_ENFORCE_EQ(
labels_dims[logits_rank - 1],
1UL,
common::errors::InvalidArgument(
"the last dimension of Input(Label) should be 1."
"But received: the last dimension of Input(Label) is [%d],"
"the last dimension is [%d]",
labels_dims[logits_rank - 1],
logits_rank - 1));
}
softmax->set_dims(logits_dims);
softmax->set_dtype(logits.dtype());
logits_dims[axis] = 1;
loss->set_dims(logits_dims);
loss->set_dtype(logits.dtype());
softmax->share_lod(logits);
loss->share_lod(logits);
}
void MaskedSelectInferMeta(const MetaTensor& x,
const MetaTensor& mask,
MetaTensor* out) {
out->set_dims({-1}); // can not infer
out->set_dtype(x.dtype());
}
void MaskedFillInferMeta(const MetaTensor& x,
const MetaTensor& mask,
const MetaTensor& value,
MetaTensor* out) {
auto x_dims = x.dims();
auto mask_dims = mask.dims();
auto expanded_dims = funcs::BroadcastTwoDims(x_dims, mask_dims, -1);
out->set_dims(expanded_dims);
out->set_dtype(x.dtype());
}
void MatmulInferMeta(const MetaTensor& x,
const MetaTensor& y,
bool trans_x,
bool trans_y,
MetaTensor* out,
MetaConfig config) {
std::vector<int64_t> dims_x = vectorize(x.dims());
std::vector<int64_t> dims_y = vectorize(y.dims());
auto ndims_x = dims_x.size();
auto ndims_y = dims_y.size();
const int64_t lhs_reduce_dim = (ndims_x == 1) ? 0 : ndims_x - 1 - trans_x;
const int64_t rhs_reduce_dim = (ndims_y == 1) ? 0 : ndims_y - 2 + trans_y;
const int64_t K_lhs = dims_x[lhs_reduce_dim];
const int64_t K_rhs = dims_y[rhs_reduce_dim];
if (config.is_runtime || (K_rhs != -1 && K_lhs != -1)) {
PADDLE_ENFORCE_EQ(
K_lhs,
K_rhs,
common::errors::InvalidArgument(
"In operator matmul, the [%d] dimension of Input(X) must be equal "
"to "
"the [%d] dimension of Input(Y). But receiving the [%d]"
"dimension of Input(X) is [%d], and the [%d] dimension of "
"Input(Y) is [%d].",
lhs_reduce_dim,
rhs_reduce_dim,
lhs_reduce_dim,
K_lhs,
rhs_reduce_dim,
K_rhs));
}
PADDLE_ENFORCE_GT(ndims_x,
0UL,
common::errors::InvalidArgument(
"The Input(x) dims size must be greater than 0,"
" but received dims size is 0. "));
PADDLE_ENFORCE_GT(ndims_y,
0UL,
common::errors::InvalidArgument(
"The Input(y) dims size must be greater than 0,"
" but received dims size is 0. "));
bool x_broadcasted = false, y_broadcasted = false;
if (ndims_x == 1) {
dims_x.insert(dims_x.begin(), 1);
ndims_x = 2;
x_broadcasted = true;
}
if (ndims_y == 1) {
dims_y.push_back(1);
ndims_y = 2;
y_broadcasted = true;
}
size_t M = 0, N = 0;
if (trans_x) {
M = dims_x[ndims_x - 1];
} else {
M = dims_x[ndims_x - 2];
}
if (trans_y) {
N = dims_y[ndims_y - 2];
} else {
N = dims_y[ndims_y - 1];
}
std::vector<int64_t> new_dims;
if (ndims_x > ndims_y) {
new_dims.assign(dims_x.begin(), dims_x.end() - 2);
} else if (ndims_x < ndims_y) {
new_dims.assign(dims_y.begin(), dims_y.end() - 2);
} else {
new_dims.reserve(ndims_x);
for (size_t i = 0; i < ndims_x - 2; ++i) {
// If one of them is 0, choose 0.
if (dims_x[i] == 0 || dims_y[i] == 0) {
new_dims.push_back(0);
} else {
new_dims.push_back(std::max(dims_x[i], dims_y[i]));
}
}
}
if (!x_broadcasted) {
new_dims.push_back(M); // NOLINT
}
if (!y_broadcasted) {
new_dims.push_back(N); // NOLINT
}
auto ddim_out = make_ddim(new_dims);
out->set_dims(ddim_out);
if (x.dtype() == DataType::INT8) {
out->set_dtype(DataType::INT32);
} else if (x.dtype() == DataType::FLOAT8_E4M3FN ||
x.dtype() == DataType::FLOAT8_E5M2) {
out->set_dtype(DataType::FLOAT16);
} else {
out->set_dtype(x.dtype());
}
out->set_layout(x.layout());
out->share_lod(x);
}
void MmOutDtypeInferMeta(const MetaTensor& x,
const MetaTensor& y,
DataType out_dtype,
MetaTensor* out,
MetaConfig config) {
PADDLE_ENFORCE_EQ(
out_dtype,
DataType::FLOAT32,
common::errors::InvalidArgument(
"The out_dtype of paddle.mm currently only supports float32."));
PADDLE_ENFORCE_EQ(
x.dtype(),
DataType::BFLOAT16,
common::errors::InvalidArgument(
"The out_dtype of paddle.mm currently only supports bfloat16 "
"Input(X)."));
PADDLE_ENFORCE_EQ(
y.dtype(),
DataType::BFLOAT16,
common::errors::InvalidArgument(
"The out_dtype of paddle.mm currently only supports bfloat16 "
"Input(Y)."));
auto dims_x = vectorize(x.dims());
auto dims_y = vectorize(y.dims());
PADDLE_ENFORCE_EQ(
dims_x.size(),
2UL,
common::errors::InvalidArgument(
"The out_dtype of paddle.mm currently only supports 2-D Input(X)."));
PADDLE_ENFORCE_EQ(
dims_y.size(),
2UL,
common::errors::InvalidArgument(
"The out_dtype of paddle.mm currently only supports 2-D Input(Y)."));
const int64_t K_lhs = dims_x[1];
const int64_t K_rhs = dims_y[0];
if (config.is_runtime || (K_lhs >= 0 && K_rhs >= 0)) {
PADDLE_ENFORCE_EQ(
K_lhs,
K_rhs,
common::errors::InvalidArgument(
"Input(X)'s width must equal Input(Y)'s height, but received %d "
"and %d.",
K_lhs,
K_rhs));
}
out->set_dims(make_ddim({dims_x[0], dims_y[1]}));
out->set_dtype(DataType::FLOAT32);
out->set_layout(x.layout());
out->share_lod(x);
}
void MatmulWithFlattenInferMeta(const MetaTensor& x,
const MetaTensor& y,
int x_num_col_dims,
int y_num_col_dims,
MetaTensor* out) {
auto x_dims = x.dims();
auto y_dims = y.dims();
VLOG(3) << "mul operator x.shape=" << x_dims << " y.shape=" << y_dims
<< " x_num_col_dims=" << x_num_col_dims
<< " y_num_col_dims=" << y_num_col_dims;
PADDLE_ENFORCE_NE(common::product(y_dims),
0,
common::errors::PreconditionNotMet(
"The Input variable Y has not "
"been initialized. You may need to confirm "
"if you put exe.run(startup_program) "
"after optimizer.minimize function."));
PADDLE_ENFORCE_GT(
x_dims.size(),
x_num_col_dims,
common::errors::InvalidArgument(
"The input tensor X's dimensions of MulOp "
"should be larger than x_num_col_dims. But received X's "
"dimensions = %d, X's shape = [%s], x_num_col_dims = %d.",
x_dims.size(),
x_dims,
x_num_col_dims));
PADDLE_ENFORCE_GT(
y_dims.size(),
y_num_col_dims,
common::errors::InvalidArgument(
"The input tensor Y's dimensions of MulOp "
"should be larger than y_num_col_dims. But received Y's "
"dimensions = %d, Y's shape = [%s], y_num_col_dims = %d.",
y_dims.size(),
y_dims,
y_num_col_dims));
auto x_mat_dims = flatten_to_2d(x_dims, x_num_col_dims);
auto y_mat_dims = flatten_to_2d(y_dims, y_num_col_dims);
PADDLE_ENFORCE_EQ(
x_mat_dims[1],
y_mat_dims[0],
common::errors::InvalidArgument(
"After flatten the input tensor X and Y to 2-D dimensions matrix "
"X1 and Y1, the matrix X1's width must be equal with matrix Y1's "
"height. But received X's shape = [%s], X1's shape = [%s], X1's "
"width = %s; Y's shape = [%s], Y1's shape = [%s], Y1's height = "
"%s.",
x_dims,
x_mat_dims,
x_mat_dims[1],
y_dims,
y_mat_dims,
y_mat_dims[0]));
std::vector<int64_t> output_dims;
output_dims.reserve(
static_cast<size_t>(x_num_col_dims + y_dims.size() - y_num_col_dims));
for (int i = 0; i < x_num_col_dims; ++i) {
output_dims.push_back(x_dims[i]);
}
for (int i = y_num_col_dims; i < y_dims.size(); ++i) {
output_dims.push_back(y_dims[i]);
}
out->set_dims(make_ddim(output_dims));
if (x.dtype() == DataType::INT8) {
out->set_dtype(DataType::INT32);
} else {
out->set_dtype(x.dtype());
}
out->share_lod(x);
}
void MatrixNMSInferMeta(const MetaTensor& bboxes,
const MetaTensor& scores,
float score_threshold,
int nms_top_k,
int keep_top_k,
float post_threshold,
bool use_gaussian,
float gaussian_sigma,
int background_label,
bool normalized,
MetaTensor* out,
MetaTensor* index,
MetaTensor* roisnum,
MetaConfig config) {
auto box_dims = bboxes.dims();
auto score_dims = scores.dims();
auto score_size = score_dims.size();
if (config.is_runtime) {
PADDLE_ENFORCE_EQ(
score_size == 3,
true,
errors::InvalidArgument("The rank of Input(Scores) must be 3. "
"But received rank = %d.",
score_size));
PADDLE_ENFORCE_EQ(
box_dims.size(),
3,
errors::InvalidArgument("The rank of Input(BBoxes) must be 3."
"But received rank = %d.",
box_dims.size()));
PADDLE_ENFORCE_EQ(box_dims[2] == 4,
true,
errors::InvalidArgument(
"The last dimension of Input (BBoxes) must be 4, "
"represents the layout of coordinate "
"[xmin, ymin, xmax, ymax]."));
PADDLE_ENFORCE_EQ(
box_dims[1],
score_dims[2],
errors::InvalidArgument(
"The 2nd dimension of Input(BBoxes) must be equal to "
"last dimension of Input(Scores), which represents the "
"predicted bboxes."
"But received box_dims[1](%s) != score_dims[2](%s)",
box_dims[1],
score_dims[2]));
}
out->set_dims({-1, box_dims[2] + 2});
out->set_dtype(bboxes.dtype());
index->set_dims({-1, 1});
index->set_dtype(DataType::INT32);
if (roisnum != nullptr) {
roisnum->set_dims({score_dims[0]});
roisnum->set_dtype(DataType::INT32);
}
}
void MatrixRankStaticInferMeta(const MetaTensor& x,
const MetaTensor& atol_tensor,
bool use_default_tol,
bool hermitian,
MetaTensor* out) {
if (atol_tensor) {
MatrixRankTolInferMeta(x, atol_tensor, use_default_tol, hermitian, out);
} else {
MatrixRankInferMeta(x, use_default_tol, hermitian, out);
}
}
void MatrixRankTolInferMeta(const MetaTensor& x,
const MetaTensor& atol_tensor,
bool use_default_tol,
bool hermitian,
MetaTensor* out) {
auto dim_x = x.dims();
PADDLE_ENFORCE_GE(dim_x.size(),
2,
common::errors::InvalidArgument(
"The dims of input must be greater than 2"));
if (hermitian && x.numel() != 0) {
int64_t rows = static_cast<int64_t>(dim_x[dim_x.size() - 2]);
int64_t cols = static_cast<int64_t>(dim_x[dim_x.size() - 1]);
// if x is 0 size tensor,ignore rows == cols check.
PADDLE_ENFORCE_EQ(rows,
cols,
common::errors::InvalidArgument(
"if hermitian == true, matrix should be n*n"));
}
DDim dim_x_batch = detail::CheckAndGetOutputDim(dim_x);
auto dim_tol = atol_tensor.dims();
if (x.numel() == 0) {
if (dim_x.size() == 2) {
out->set_dims(make_ddim({}));
} else {
out->set_dims(dim_x_batch);
}
} else if (dim_x_batch == dim_tol) {
out->set_dims(dim_x_batch);
} else {
int max_dim = std::max(dim_x_batch.size(), dim_tol.size());
int axis = std::abs(dim_x_batch.size() - dim_tol.size());
std::vector<int64_t> x_batch_dims_array(max_dim);
std::vector<int64_t> tol_dims_array(max_dim);
std::vector<int64_t> out_dims_array(max_dim);
funcs::GetBroadcastDimsArrays(dim_x_batch,
dim_tol,
x_batch_dims_array.data(),
tol_dims_array.data(),
out_dims_array.data(),
max_dim,
axis);
out->set_dims(make_ddim(out_dims_array));
}
out->share_lod(x);
}
void MulticlassNmsv1InferMeta(const MetaTensor& bboxes,
const MetaTensor& scores,
float score_threshold,
int nms_top_k,
int keep_top_k,
float nms_threshold,
float nms_eta,
bool normalized,
int background_label,
MetaTensor* out,
MetaConfig config) {
const auto& box_dims = bboxes.dims();
const auto& score_dims = scores.dims();
int score_size = static_cast<int>(score_dims.size());
if (config.is_runtime) {
PADDLE_ENFORCE_EQ(score_size == 2 || score_size == 3,
true,
common::errors::InvalidArgument(
"The rank of Input(Scores) must be 2 or 3"
". But received rank = %d",
score_size));
PADDLE_ENFORCE_EQ(
box_dims.size(),
3,
common::errors::InvalidArgument("The rank of Input(BBoxes) must be 3"
". But received rank = %d",
box_dims.size()));
if (score_size == 3) {
PADDLE_ENFORCE_EQ(box_dims[2] == 4 || box_dims[2] == 8 ||
box_dims[2] == 16 || box_dims[2] == 24 ||
box_dims[2] == 32,
true,
common::errors::InvalidArgument(
"The last dimension of Input"
"(BBoxes) must be 4 or 8, "
"represents the layout of coordinate "
"[xmin, ymin, xmax, ymax] or "
"4 points: [x1, y1, x2, y2, x3, y3, x4, y4] or "
"8 points: [xi, yi] i= 1,2,...,8 or "
"12 points: [xi, yi] i= 1,2,...,12 or "
"16 points: [xi, yi] i= 1,2,...,16"));
PADDLE_ENFORCE_EQ(
box_dims[1],
score_dims[2],
common::errors::InvalidArgument(
"The 2nd dimension of Input(BBoxes) must be equal to "
"last dimension of Input(Scores), which represents the "
"predicted bboxes."
"But received box_dims[1](%s) != score_dims[2](%s)",
box_dims[1],
score_dims[2]));
} else {
PADDLE_ENFORCE_EQ(box_dims[2],
4,
common::errors::InvalidArgument(
"The last dimension of Input"
"(BBoxes) must be 4. But received dimension = %d",
box_dims[2]));
PADDLE_ENFORCE_EQ(
box_dims[1],
score_dims[1],
common::errors::InvalidArgument(
"The 2nd dimension of Input"
"(BBoxes) must be equal to the 2nd dimension of Input(Scores). "
"But received box dimension = %d, score dimension = %d",
box_dims[1],
score_dims[1]));
}
}
// Here the box_dims[0] is not the real dimension of output.
// It will be rewritten in the computing kernel.
out->set_dims({-1, box_dims[2] + 2});
out->set_dtype(scores.dtype());
}
void MvInferMeta(const MetaTensor& x, const MetaTensor& vec, MetaTensor* out) {
auto dim_x = x.dims();
auto dim_vec = vec.dims();
PADDLE_ENFORCE_EQ(
dim_x.size(),
2,
common::errors::InvalidArgument(
"The rank of input X should be 2, but is %d", dim_x.size()));
PADDLE_ENFORCE_EQ(
dim_vec.size(),
1,
common::errors::InvalidArgument(
"The rank of input Vec should be 1, but is %d", dim_vec.size()));
PADDLE_ENFORCE_EQ(dim_x[1],
dim_vec[0],
common::errors::InvalidArgument(
"X's second dimension is expected to be equal to "
"Vec's first dimension, "
"but received X'shape = [%s], Vec's shape = [%s]",
dim_x,
dim_vec));
auto dim_out = make_ddim({dim_x[0]});
out->set_dims(dim_out);
out->set_dtype(x.dtype());
out->set_layout(x.layout());
out->share_lod(x);
}
void PReluInferMeta(const MetaTensor& x,
const MetaTensor& alpha,
const std::string& data_format,
const std::string& mode,
MetaTensor* out,
MetaConfig config) {
auto x_dim = x.dims();
if (mode == "all") {
PADDLE_ENFORCE_EQ(common::product(alpha.dims()),
1,
common::errors::InvalidArgument(
"For mode 'all', size of weight Alpha must be one. "
"But received alpha's size: %d.",
product(alpha.dims())));
} else if (mode == "channel") {
auto x_rank = x_dim.size();
PADDLE_ENFORCE_GE(x_rank,
2,
common::errors::InvalidArgument(
"For mode 'channel', rank of input X must be "
"equal or larger than 2. But received X's "
"rank: %d",
x_rank));
PADDLE_ENFORCE_EQ(data_format == "NCHW" || data_format == "NHWC",
true,
common::errors::InvalidArgument(
"For mode 'channel', data_format must be one of "
"NCHW and NHWC. But received data_format: %s",
data_format));
if (data_format == "NCHW" || config.is_run_onednn_kernel) {
PADDLE_ENFORCE_EQ(product(alpha.dims()) == x_dim[1],
true,
common::errors::InvalidArgument(
"For mode 'channel', size of weight Alpha must be "
"equal to the number of channels of input(x). But "
"received alpha's size: %d, x_dim[1]: %d",
product(alpha.dims()),
x_dim[1]));
} else {
PADDLE_ENFORCE_EQ(product(alpha.dims()) == x_dim[x_rank - 1],
true,
common::errors::InvalidArgument(
"For mode 'channel', size of weight Alpha must be "
"equal to the number of channels of input(x). But "
"received alpha's size: %d, x_dim[%d]: %d",
product(alpha.dims()),
x_rank - 1,
x_dim[x_rank - 1]));
}
} else if (mode == "element") {
auto alpha_dim = alpha.dims();
auto alpha_rank = alpha_dim.size();
int x_rank = x_dim.size();
PADDLE_ENFORCE_GE(x_rank,
1,
common::errors::InvalidArgument(
"For mode 'element', rank of input X must be "
"equal or larger than 1. But received X's "
"rank: %d",
x_rank));
PADDLE_ENFORCE_EQ(
alpha_rank,
x_rank,
common::errors::InvalidArgument(
"For mode 'element', rank of weight Alpha must be equal to the "
"rank of input(x). But received alpha's rank: %d, x's rank: %d.",
alpha_rank,
x_rank));
size_t x_product = 1;
size_t alpha_product = 1;
for (int i = x_rank - 1; i > 0; i--) {
x_product *= x_dim[i];
alpha_product *= alpha_dim[i];
}
PADDLE_ENFORCE_EQ(
alpha_product,
x_product,
common::errors::InvalidArgument(
"For mode 'element', the size of weight Alpha must be "
"equal to the size of input(x). But received alpha's size: %d, "
"x's size: %d.",
alpha_product,
x_product));
} else {
PADDLE_THROW(common::errors::InvalidArgument(
"Attr(mode) of prelu must be one of 'all', 'channel', or 'element'. "
"But received "
"mode: '%s'.",
mode));
}
out->set_dims(x_dim);
out->set_dtype(x.dtype());
out->set_layout(x.layout());
out->share_lod(x);
}
void PullGpupsSparseInferMeta(const MetaTensor& w,
const std::vector<const MetaTensor*>& ids,
const std::vector<int>& size,
bool is_sparse,
bool is_distributed,
std::vector<MetaTensor*> out) {
PADDLE_ENFORCE_GE(
ids.size(),
1UL,
common::errors::InvalidArgument(
"Inputs(Ids) of PullGpuPSSparseOp should not be empty."));
PADDLE_ENFORCE_GE(
out.size(),
1UL,
common::errors::InvalidArgument(
"Outputs(Out) of PullGpuPSSparseOp should not be empty."));
PADDLE_ENFORCE_EQ(
ids.size(),
size.size(),
common::errors::InvalidArgument("The ids size: %lu must be equal to "
"the length of embedding size: %lu.",
ids.size(),
size.size()));
const size_t n_ids = ids.size();
std::vector<DDim> outs_dims;
outs_dims.resize(n_ids);
for (size_t i = 0; i < n_ids; ++i) {
int64_t embedding_size = size[i];
const auto ids_dims = ids[i]->dims();
int ids_rank = ids_dims.size();
PADDLE_ENFORCE_EQ(ids_dims[ids_rank - 1],
1,
common::errors::InvalidArgument(
"Shape error in %lu id, the last dimension of the "
"'Ids' tensor must be 1.",
i));
auto out_dim = vectorize(slice_ddim(ids_dims, 0, ids_rank - 1));
out_dim.push_back(embedding_size);
outs_dims[i] = make_ddim(out_dim);
}
for (size_t i = 0; i < n_ids; ++i) {
out[i]->set_dims(outs_dims[i]);
out[i]->share_lod(*ids[i], i);
out[i]->set_dtype(w.dtype());
}
}
void PullSparseV2InferMeta(const std::vector<const MetaTensor*>& ids,
const std::vector<const MetaTensor*>& w,
int embedding_dim,
int table_id,
const std::string& accessor_class,
const std::string& ctrlabel_name,
int padding_id,
bool scale_sparse_grad,
const std::vector<std::string>& input_names,
bool is_distributed,
std::vector<MetaTensor*> out) {
PADDLE_ENFORCE_GE(ids.size(),
1UL,
common::errors::InvalidArgument(
"Input(Ids) of PullSparseV2Op can not be null"));
PADDLE_ENFORCE_GE(out.size(),
1UL,
common::errors::InvalidArgument(
"Output(Out) of PullSparseV2Op can not be null"));
auto hidden_size = embedding_dim;
const size_t n_ids = ids.size();
std::vector<DDim> outs_dims;
outs_dims.resize(n_ids);
for (size_t i = 0; i < n_ids; ++i) {
const auto ids_dims = ids[i]->dims();
auto out_dim = vectorize(ids_dims);
out_dim.push_back(hidden_size);
outs_dims[i] = make_ddim(out_dim);
}
for (size_t i = 0; i < n_ids; ++i) {
out[i]->set_dims(outs_dims[i]);
out[i]->share_lod(*ids[i], i);
out[i]->set_dtype(w[i]->dtype());
}
}
void ApplyPerChannelScaleInferMeta(const MetaTensor& x,
const MetaTensor& scales,
MetaTensor* out) {
auto x_dim = x.dims();
auto scales_dim = scales.dims();
PADDLE_ENFORCE_EQ(
x_dim.size(),
2,
common::errors::InvalidArgument(
"The rank of Input(x) must be 2, but received %d.", x_dim.size()));
PADDLE_ENFORCE_EQ(scales_dim.size(),
1,
common::errors::InvalidArgument(
"The rank of Input(scales) must be 1, but received %d.",
scales_dim.size()));
PADDLE_ENFORCE_EQ(
x_dim[1],
scales_dim[0],
common::errors::InvalidArgument(
"The second dim of Input(x) must be equal to the first dim of scales,"
"but received %d and %d.",
x_dim[2],
scales_dim[1]));
out->set_dtype(x.dtype());
out->set_dims(x_dim);
out->set_layout(x.layout());
}
inline void ExpandAspectRatios(const std::vector<float>& input_aspect_ratio,
bool flip,
std::vector<float>* output_aspect_ratio) {
constexpr float epsilon = 1e-6;
output_aspect_ratio->clear();
output_aspect_ratio->push_back(1.0f);
for (auto ar : input_aspect_ratio) {
bool already_exist = false;
for (auto item : *output_aspect_ratio) {
if (fabs(ar - item) < epsilon) {
already_exist = true;
break;
}
}
if (!already_exist) {
output_aspect_ratio->push_back(ar);
if (flip) {
output_aspect_ratio->push_back(1.0f / ar);
}
}
}
}
void PriorBoxInferMeta(const MetaTensor& input,
const MetaTensor& image,
const std::vector<float>& min_sizes,
const std::vector<float>& max_sizes,
const std::vector<float>& aspect_ratios,
const std::vector<float>& variances,
bool flip,
bool clip,
float step_w,
float step_h,
float offset,
bool min_max_aspect_ratios_order,
MetaTensor* out,
MetaTensor* var) {
auto image_dims = image.dims();
auto input_dims = input.dims();
PADDLE_ENFORCE_EQ(
image_dims.size(),
4,
common::errors::InvalidArgument(
"The Input(Image) of Op(PriorBoxOp) should be a 4-D Tensor "
"and data format is NCHW. But received Image's dimensions = %d, "
"shape = [%s].",
image_dims.size(),
image_dims));
PADDLE_ENFORCE_EQ(
input_dims.size(),
4,
common::errors::InvalidArgument(
"The Input(Input) of Op(PriorBoxOp) should be a 4-D Tensor "
"and data format is NCHW. But received Input's dimensions = %d, "
"shape = [%s].",
input_dims.size(),
input_dims));
std::vector<float> aspect_ratios_vec;
ExpandAspectRatios(aspect_ratios, flip, &aspect_ratios_vec);
size_t num_priors = aspect_ratios_vec.size() * min_sizes.size();
if (!max_sizes.empty()) {
PADDLE_ENFORCE_EQ(
max_sizes.size(),
min_sizes.size(),
common::errors::InvalidArgument(
"The length of min_size and "
"max_size must be equal. But received: min_size's length is %d, "
"max_size's length is %d.",
min_sizes.size(),
max_sizes.size()));
num_priors += max_sizes.size();
for (size_t i = 0; i < max_sizes.size(); ++i) {
PADDLE_ENFORCE_GT(
max_sizes[i],
min_sizes[i],
common::errors::InvalidArgument(
"max_size[%d] must be greater "
"than min_size[%d]. But received: max_size[%d] is %f, "
"min_size[%d] is %f.",
i,
i,
i,
max_sizes[i],
i,
min_sizes[i]));
}
}
std::vector<int64_t> dim_vec(4);
dim_vec[0] = input_dims[2];
dim_vec[1] = input_dims[3];
dim_vec[2] = static_cast<int64_t>(num_priors);
dim_vec[3] = 4;
out->set_dtype(input.dtype());
var->set_dtype(input.dtype());
out->set_dims(make_ddim(dim_vec));
var->set_dims(make_ddim(dim_vec));
}
void PruneGateByCapacityInferMeta(const MetaTensor& gate_idx,
const MetaTensor& expert_count,
int64_t n_expert,
int64_t n_worker,
MetaTensor* new_gate_idx) {
auto expert_count_dims = expert_count.dims();
int64_t expert_count_num_ele = 1;
for (int i = 0; i < static_cast<int>(expert_count_dims.size()); i++) {
expert_count_num_ele *= expert_count_dims[i];
}
PADDLE_ENFORCE_EQ(
expert_count_num_ele,
n_expert * n_worker,
common::errors::Unavailable(
"The number of elements for expert_count is ( %ld ) incorrect. "
"Because the number of expert_count must equal the "
"product of n_worker ( %ld ) and n_expert ( %ld ). "
"Please input appropriate expert_count again!",
expert_count_num_ele,
n_worker,
n_expert));
auto gate_idx_in_dims = gate_idx.dims();
new_gate_idx->set_dims(gate_idx_in_dims);
new_gate_idx->set_dtype(gate_idx.dtype());
}
void PullBoxSparseInferMeta(const MetaTensor& w,
const std::vector<const MetaTensor*>& ids,
bool is_sparse,
bool is_distributed,
int size,
std::vector<MetaTensor*> out) {
auto hidden_size = static_cast<int64_t>(size);
const size_t n_ids = ids.size();
for (size_t i = 0; i < n_ids; ++i) {
MetaTensor* output = out[i];
auto ids_dims = ids[i]->dims();
int ids_rank = ids_dims.size();
PADDLE_ENFORCE_EQ(ids_dims[ids_rank - 1],
1UL,
common::errors::InvalidArgument(
"Shape error in %lu id, the last dimension of the "
"'Ids' tensor must be 1.",
i));
auto out_dim = vectorize(slice_ddim(ids_dims, 0, ids_rank - 1));
out_dim.push_back(hidden_size);
output->set_dims(make_ddim(out_dim));
output->share_lod(*ids[i]);
output->set_dtype(w.dtype());
}
}
void RepeatInterleaveWithTensorIndexInferMeta(const MetaTensor& x,
const MetaTensor& repeats,
int dim,
int64_t output_size,
MetaTensor* out) {
const auto& input_dim = x.dims();
auto output_dim = vectorize(input_dim);
PADDLE_ENFORCE_EQ(
dim < input_dim.size() && dim >= (0 - input_dim.size()),
true,
common::errors::OutOfRange(
"Attr(dim) is out of range, It's expected "
"to be in range of [-%d, %d]. But received Attr(dim) = %d.",
input_dim.size(),
input_dim.size() - 1,
dim));
auto repeats_dim = repeats.dims();
PADDLE_ENFORCE_EQ(
repeats_dim.size() == 1 ||
(repeats_dim.size() == 2 && repeats_dim[1] == 1),
true,
common::errors::InvalidArgument(
"The 'shape' of Input(RepeatsTensor) must be 1-D tensor. "
"But received: the 'shape' of Input(Index) is [%s], "
"the dimension of Input(Index) is [%d].",
repeats_dim,
repeats_dim.size()));
if (input_dim.size() == 1 && input_dim[0] == 0) {
output_dim[0] = 0;
} else {
PADDLE_ENFORCE_EQ(repeats_dim[0] != 0,
true,
common::errors::InvalidArgument(
"The length of Input(RepeatsTensor) can't be 0."));
PADDLE_ENFORCE_NE(
out,
nullptr,
common::errors::InvalidArgument(
"repeat_interleave's output tensor can't be nullptr"));
if (dim < 0) {
dim += input_dim.size();
}
if (output_size > 0) {
// Use provided output_size to avoid stream synchronization
output_dim[dim] = output_size;
} else {
output_dim[dim] = -1;
}
}
out->set_dims(make_ddim(output_dim));
out->share_lod(x);
out->set_dtype(x.dtype());
}
void RmsNormInferMeta(const MetaTensor& x,
const MetaTensor& scale,
const std::vector<int64_t>& normalized_shape,
double epsilon,
MetaTensor* y,
MetaTensor* invvar) {
auto x_dim = x.dims();
// std::vector<int64_t> normalized_shape_data = normalized_shape.GetData();
int normalized_shape_size = normalized_shape.size();
int x_dims_size = x_dim.size();
int begin_norm_axis = x_dims_size - normalized_shape_size;
PADDLE_ENFORCE_GT(begin_norm_axis,
0,
common::errors::InvalidArgument(
"'begin_norm_axis' in Op(LayerNorm) should be "
"greater than zero. But received [%d].",
begin_norm_axis));
PADDLE_ENFORCE_LT(
begin_norm_axis,
x_dims_size,
common::errors::InvalidArgument(
"'begin_norm_axis' must be less than the dimensions of X,"
"But received 'begin_norm_axis' is [%d],"
"received the dimensions of X is [%d].",
begin_norm_axis,
x_dims_size));
for (int i = 0; i < normalized_shape_size; i++) {
PADDLE_ENFORCE_EQ(x_dim[x_dims_size - i - 1],
normalized_shape[normalized_shape_size - i - 1],
common::errors::InvalidArgument(
"The %d-th dimension of X is not equal to the %d-th "
"dimension of NormalizedShape.",
x_dims_size - i - 1,
normalized_shape_size - i - 1));
}
if (scale) {
auto scale_dim = scale.dims();
PADDLE_ENFORCE_EQ(scale_dim.size(),
normalized_shape_size,
common::errors::InvalidArgument(
"The dimensions of Input(Scale) must be equal to the "
"dimensions of NormalizedShape. "
"But received: the dimensions of Input(Scale) is "
"[%d], the dimensions of NormalizedShape is [%d].",
scale_dim.size(),
normalized_shape_size));
for (int i = 0; i < normalized_shape_size; i++) {
PADDLE_ENFORCE_EQ(scale_dim[i],
normalized_shape[i],
common::errors::InvalidArgument(
"The %d-th dimension of Input(Scale) is not equal "
"to the %d-th dimension of NormalizedShape.",
i,
i));
}
}
auto before_norm_dims = slice_ddim(x_dim, 0, begin_norm_axis);
PADDLE_ENFORCE_EQ(epsilon >= 0.0f && epsilon <= 0.001f,
true,
common::errors::InvalidArgument(
"'epsilon' in Op(LayerNorm) should be between"
"0.0 and 0.001, But received [%s].",
epsilon));
DataType x_dtype = x.dtype();
y->set_dims(x_dim);
y->set_dtype(x_dtype);
DataType param_type =
(x_dtype == DataType::BFLOAT16 || x_dtype == DataType::FLOAT16)
? DataType::FLOAT32
: x_dtype;
invvar->set_dims({before_norm_dims});
invvar->set_dtype(param_type);
}
void RowConvInferMeta(const MetaTensor& x,
const MetaTensor& filter,
MetaTensor* out) {
auto filter_dims = filter.dims();
PADDLE_ENFORCE_EQ(filter_dims.size(),
2,
common::errors::InvalidArgument(
"Input(Filter)'s dimensions should be 2. Received: "
"Input(Filter)'s shape: [%s].",
filter_dims));
out->set_dims(x.dims());
out->share_lod(x);
out->set_dtype(x.dtype());
}
void SearchsortedInferMeta(const MetaTensor& sorted_sequence,
const MetaTensor& value,
bool out_int32,
bool right,
MetaTensor* out) {
auto sequences_dims = sorted_sequence.dims();
auto values_dims = value.dims();
PADDLE_ENFORCE_GE(
sequences_dims.size(),
1,
common::errors::InvalidArgument(
"Input sequences's dimension(%d) must be greater or equal than 1",
sequences_dims.size()));
bool flag = true;
if (sequences_dims.size() != values_dims.size()) {
flag = false;
}
const int& sequences_dims_size = sequences_dims.size();
for (int dim = 0; dim < sequences_dims_size - 1; ++dim) {
if (sequences_dims[dim] != values_dims[dim]) {
flag = false;
break;
}
}
if (sequences_dims.size() != 1) {
PADDLE_ENFORCE_EQ(
flag,
true,
common::errors::Unavailable(
"The dimensions of sorted_sequence tensor ( %s ) and values "
"tensor ( %s ) can not match. Because the input sorted_sequence "
"tensor must be 1 dimension or the first N-1 dimensions of "
"sorted_sequence tensor and input values tensor must match. "
"Please input appropriate sorted_sequence and values again! ",
sequences_dims,
values_dims));
}
if (out_int32) {
PADDLE_ENFORCE_LT(
sequences_dims[sequences_dims.size() - 1],
std::numeric_limits<int>::max(),
common::errors::Unavailable(
"The size of sorted_sequence %d exceed the maximum limit %d. "
"Because the size of sorted_sequence should be less than the "
"output maximum value for int32 bit. Please set appropriate "
"sorted_sequence to meet this requirement! ",
sequences_dims[sequences_dims.size() - 1],
std::numeric_limits<int>::max()));
}
out->set_dims(values_dims);
if (out_int32) {
out->set_dtype(DataType::INT32);
} else {
out->set_dtype(DataType::INT64);
}
}
void SequenceExpandInferMeta(const MetaTensor& x,
const MetaTensor& y,
int ref_level,
MetaTensor* out,
MetaConfig config) {
const auto& x_dims = x.dims();
auto out_dims = x_dims;
PADDLE_ENFORCE_GE(
x_dims.size(),
2,
common::errors::InvalidArgument(
"Dimension number of Input(X) should be at least 2. But "
"received: input rank %u, input shape [%s].",
x_dims.size(),
x_dims));
if (config.is_runtime) {
} else {
out_dims[0] = -1;
}
out->set_dims(out_dims);
out->share_lod(x);
out->set_dtype(x.dtype());
}
void ShapeBroadcastInferMeta(const MetaTensor& x,
const MetaTensor& y,
MetaTensor* out) {
const auto& x_dims = x.dims();
const auto& y_dims = y.dims();
PADDLE_ENFORCE_EQ(
x_dims.size(),
1,
common::errors::InvalidArgument(
"The rank of x must be 1. But received: %d", x_dims.size()));
PADDLE_ENFORCE_EQ(
y_dims.size(),
1,
common::errors::InvalidArgument(
"The rank of y must be 1. But received: %d", y_dims.size()));
if (x_dims[0] <= y_dims[0]) {
out->set_dims(y_dims);
} else {
out->set_dims(x_dims);
}
out->set_dtype(x.dtype());
}
void ShuffleBatchInferMeta(const MetaTensor& x,
const MetaTensor& seed,
int startup_seed,
MetaTensor* out,
MetaTensor* shuffle_idx,
MetaTensor* seed_out
) {
out->share_dims(x);
out->share_lod(x);
out->set_dtype(x.dtype());
seed_out->share_dims(seed);
seed_out->share_lod(seed);
seed_out->set_dtype(seed.dtype());
shuffle_idx->set_dims(make_ddim({-1}));
}
void SlowConvDilatedInferMeta(const MetaTensor& input,
const MetaTensor& filter,
const MetaTensor& bias,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const std::string& padding_algorithm,
const std::vector<int>& dilations,
int groups,
const std::string& data_format,
MetaTensor* out,
MetaConfig config) {
ConvInferMeta(input,
filter,
strides,
paddings,
padding_algorithm,
dilations,
groups,
data_format,
out,
config);
}
void SlowConv3DDilatedInferMeta(const MetaTensor& input,
const MetaTensor& filter,
const MetaTensor& bias,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const std::string& padding_algorithm,
int groups,
const std::vector<int>& dilations,
const std::string& data_format,
MetaTensor* out,
MetaConfig config) {
ConvInferMeta(input,
filter,
strides,
paddings,
padding_algorithm,
dilations,
groups,
data_format,
out,
config);
}
void SequenceMaskInferMeta(const MetaTensor& x,
const MetaTensor& max_len_tensor,
int maxlen,
DataType out_dtype,
MetaTensor* y) {
auto dim = vectorize<int>(x.dims());
if (max_len_tensor) {
dim.push_back(-1);
} else {
dim.push_back(maxlen > 0 ? maxlen : -1);
}
y->set_dims(make_ddim(dim));
y->set_dtype(out_dtype);
}
void ReduceAsInferMeta(const MetaTensor& x,
const MetaTensor& target,
MetaTensor* out) {
DataType out_dtype;
if (x.dtype() == DataType::BOOL || x.dtype() == DataType::INT32) {
out_dtype = DataType::INT64;
} else {
out_dtype = x.dtype();
}
out->set_dtype(out_dtype);
out->set_dims(target.dims());
out->set_layout(x.layout());
}
void SoftmaxMaskFuseInferMeta(const MetaTensor& x,
const MetaTensor& mask,
MetaTensor* out) {
auto x_dims = x.dims();
auto mask_dims = mask.dims();
PADDLE_ENFORCE_EQ(
x_dims.size(),
4,
common::errors::InvalidArgument("Input x must be in 4D dimension but "
"received the dimension of X is %d",
x_dims.size()));
PADDLE_ENFORCE_EQ(
mask_dims.size(),
4,
common::errors::InvalidArgument("Input mask must be in 4D dimension but "
"received the dimension of mask is %d",
mask_dims.size()));
out->share_meta(x);
}
void SegmentPoolInferMeta(const MetaTensor& x,
const MetaTensor& segment_ids,
const std::string& pooltype,
MetaTensor* out,
MetaTensor* summed_ids,
MetaConfig config) {
auto x_dims = x.dims();
auto seg_dims = segment_ids.dims();
auto dims = x_dims;
dims[0] = -1;
out->set_dims(dims);
out->set_dtype(x.dtype());
out->set_layout(x.layout());
if (pooltype == "MEAN") {
summed_ids->set_dims({-1, 1});
summed_ids->set_dtype(x.dtype());
summed_ids->set_layout(x.layout());
}
// Dimension validation: check only at runtime or when dimensions are known
// Runtime: config.is_runtime = true (dynamic graph/PIR)
// Compile time: config.is_runtime = false (static graph building)
bool contain_unknown_dim = common::contain_unknown_dim(x_dims) ||
common::contain_unknown_dim(seg_dims);
bool check = config.is_runtime || !contain_unknown_dim;
if (check) {
PADDLE_ENFORCE_EQ(
seg_dims[0],
x_dims[0],
common::errors::InvalidArgument(
"Segment_ids should be the same size as dimension 0 of input X."));
PADDLE_ENFORCE_EQ(seg_dims.size(),
1UL,
common::errors::InvalidArgument(
"Segment_ids should be 1-D tensor, or it's other "
"dimension size is 1. Segment_ids's shape is: [%s].",
seg_dims));
}
}
void StftInferMeta(const MetaTensor& x,
const MetaTensor& window,
int n_fft,
int hop_length,
bool normalized,
bool onesided,
MetaTensor* out) {
const auto& x_dims = x.dims();
const int x_rank = x_dims.size();
const auto& window_dims = window.dims();
const int64_t window_size = window_dims[0];
PADDLE_ENFORCE_EQ(
x_rank,
2,
common::errors::InvalidArgument(
"Input(X) of StftOp should be a tensor with shape [N, T], "
"but got rank %s.",
x_rank));
PADDLE_ENFORCE_GT(
hop_length,
0,
common::errors::InvalidArgument(
"Attribute(hop_length) should be greater than 0, but got %s.",
hop_length));
PADDLE_ENFORCE_EQ(
window_size,
n_fft,
common::errors::InvalidArgument(
"Input(Window) of StftOp should be equal with n_fft %s, "
"but got %s.",
n_fft,
window_size));
int64_t seq_length = x_dims[x_rank - 1];
int64_t n_frames = 1 + (seq_length - n_fft) / hop_length;
PADDLE_ENFORCE_LE(n_fft,
seq_length,
common::errors::InvalidArgument(
"Attribute(frame_length) should be less equal than "
"sequence length, but got (%s) > (%s).",
n_fft,
seq_length));
std::vector<int64_t> output_shape;
output_shape.push_back(x_dims[0]);
if (onesided) {
output_shape.push_back(n_fft / 2 + 1);
} else {
output_shape.push_back(n_fft);
}
output_shape.push_back(n_frames);
out->set_dims(make_ddim(output_shape));
out->set_dtype(phi::dtype::ToComplex(x.dtype()));
}
void TakeAlongAxisInferMeta(const MetaTensor& x,
const MetaTensor& index,
int axis,
MetaTensor* out) {
auto input_dim = x.dims();
auto index_dim = index.dims();
PADDLE_ENFORCE_GT(
input_dim.size(),
0,
common::errors::InvalidArgument(
"Dimension of the input(Input) of TakeAlongAxisOp should be greater "
"than 0, but received %d.",
input_dim.size()));
PADDLE_ENFORCE_GT(
index_dim.size(),
0,
common::errors::InvalidArgument(
"Dimension of the input(Index) of TakeAlongAxisOp should be greater "
"than 0, but received %d.",
index_dim.size()));
out->set_dims(index_dim);
out->set_dtype(x.dtype());
}
void TdmChildInferMeta(const MetaTensor& x,
const MetaTensor& tree_info,
int child_nums,
DataType dtype,
MetaTensor* child,
MetaTensor* leaf_mask) {
PADDLE_ENFORCE_GT(
child_nums,
0,
common::errors::InvalidArgument(
"ValueError: The value of the 'child_nums' must greater than 0. "
"But received child_nums value = %d, ",
child_nums));
const auto& info_dims = tree_info.dims();
const auto& input_dims = x.dims();
PADDLE_ENFORCE_EQ(
info_dims.size(),
2,
common::errors::InvalidArgument(
"ShapeError: The dimensions of the 'tree info' must be 2. "
"But received tree info's dimensions = %d, "
"tree info's shape = [%s].",
info_dims.size(),
info_dims));
auto output_dims = vectorize(input_dims);
output_dims.push_back(child_nums);
if (child != nullptr) {
child->set_dims(make_ddim(output_dims));
leaf_mask->set_dims(make_ddim(output_dims));
child->share_lod(x);
leaf_mask->share_lod(x);
child->set_dtype(x.dtype());
leaf_mask->set_dtype(x.dtype());
}
}
void TriangularSolveInferMeta(const MetaTensor& x,
const MetaTensor& y,
bool upper,
bool transpose,
bool unitriangular,
MetaTensor* out) {
auto x_dims = x.dims();
auto y_dims = y.dims();
auto x_dims_n = x_dims.size();
auto y_dims_n = y_dims.size();
PADDLE_ENFORCE_GE(x_dims_n,
2,
common::errors::InvalidArgument(
"The input tensor X's dimensions of TriangularSolveOp "
"should be >= 2. But received X's "
"dimensions = %d, X's shape = [%s]",
x_dims.size(),
x_dims));
PADDLE_ENFORCE_GE(y_dims_n,
2,
common::errors::InvalidArgument(
"The input tensor Y's dimensions of TriangularSolveOp "
"should be >=2. But received Y's "
"dimensions = %d, Y's shape = [%s]",
y_dims.size(),
y_dims));
PADDLE_ENFORCE_EQ(x_dims[x_dims_n - 2],
x_dims[x_dims_n - 1],
common::errors::InvalidArgument(
"The inner-most 2 dimensions of Input(X) all should "
"be square matrices "
"But received X's shape[-2] = %d and shape[-1] = %d.",
x_dims[x_dims_n - 2],
x_dims[x_dims_n - 1]));
std::vector<int64_t> x_dims_vec = vectorize(x_dims);
std::vector<int64_t> y_dims_vec = vectorize(y_dims);
std::vector<int64_t> x_dims_vec_cut(x_dims_vec.begin(), x_dims_vec.end() - 2);
std::vector<int64_t> y_dims_vec_cut(y_dims_vec.begin(), y_dims_vec.end() - 2);
std::vector<int64_t> expand_batch_portion =
funcs::MatrixGetBroadcastBatchPortion(x_dims_vec_cut, y_dims_vec_cut);
std::vector<int64_t> y_broadcast_dims({expand_batch_portion});
y_broadcast_dims.insert(y_broadcast_dims.end(),
{y_dims_vec[y_dims_n - 2], y_dims_vec[y_dims_n - 1]});
// dim of 'out' is the same with 'Y' after broadcast
out->set_dims(make_ddim(y_broadcast_dims));
out->set_dtype(y.dtype());
out->set_layout(y.layout());
out->share_lod(y);
}
void LstsqInferMeta(const MetaTensor& x,
const MetaTensor& y,
const Scalar& rcond,
const std::string& driver,
MetaTensor* solution,
MetaTensor* residuals,
MetaTensor* rank,
MetaTensor* singular_values) {
auto x_dims = x.dims();
auto y_dims = y.dims();
int x_rank = x_dims.size();
int y_rank = y_dims.size();
int64_t m = x_dims[x_rank - 2];
int64_t n = x_dims[x_rank - 1];
int64_t nrhs = y_dims[x_rank - 1];
PADDLE_ENFORCE_GE(x_rank,
2,
common::errors::InvalidArgument(
"Expects input tensor x to be not less than "
"2 dimensions, but got dimension %d",
x_rank));
PADDLE_ENFORCE_GE(y_rank,
2,
common::errors::InvalidArgument(
"Expects input tensor y to be not less than "
"2 dimensions, but got dimension %d",
y_rank));
PADDLE_ENFORCE_EQ(
x_rank,
y_rank,
common::errors::InvalidArgument(
"Expects input tensor x and y to have the same dimension "
"but got x's dimension [%d] and y's dimension [%d]",
x_rank,
y_rank));
std::vector<int> batch_dims_vec{};
for (int i = 0; i < x_rank - 2; ++i) {
PADDLE_ENFORCE_EQ(x_dims[i],
y_dims[i],
common::errors::InvalidArgument(
"Expects input tensor x and y to have the same batch "
"dimension, but got x's batch dimension [%d] and "
"y's batch dimension [%d] in %d-th dim",
x_dims[i],
y_dims[i],
i));
batch_dims_vec.emplace_back(x_dims[i]);
}
PADDLE_ENFORCE_EQ(
m,
y_dims[y_rank - 2],
common::errors::InvalidArgument(
"Expects input tensor x and y to have the same row dimension "
"of the inner-most 2-dims matrix, "
"but got x's row dimension [%d] and y's row dimension [%d]",
m,
y_dims[y_rank - 2]));
if (x.numel() == 0 || y.numel() == 0) {
rank->set_dims(make_ddim({0}));
} else {
rank->set_dims(make_ddim(batch_dims_vec));
}
if (m > n && driver != "gelsy") {
if (driver == "gelss" || driver == "gelsd") {
residuals->set_dims(make_ddim({-1}));
} else {
batch_dims_vec.emplace_back(nrhs);
residuals->set_dims(make_ddim(batch_dims_vec));
batch_dims_vec.pop_back();
}
} else {
residuals->set_dims(make_ddim({0}));
}
residuals->set_dtype(y.dtype());
batch_dims_vec.emplace_back(std::min(m, n));
if (x.numel() == 0 || y.numel() == 0) {
singular_values->set_dims(make_ddim({0}));
} else {
singular_values->set_dims(make_ddim(batch_dims_vec));
}
singular_values->set_dtype(y.dtype());
batch_dims_vec[x_rank - 2] = n;
batch_dims_vec.emplace_back(nrhs);
solution->set_dims(make_ddim(batch_dims_vec));
solution->set_dtype(y.dtype());
}
void YoloBoxInferMeta(const MetaTensor& x,
const MetaTensor& img_size,
const std::vector<int>& anchors,
int class_num,
float conf_thresh,
int downsample_ratio,
bool clip_bbox,
float scale_x_y,
bool iou_aware,
float iou_aware_factor,
MetaTensor* boxes,
MetaTensor* scores,
MetaConfig config) {
auto dim_x = x.dims();
auto dim_imgsize = img_size.dims();
int anchor_num = static_cast<int>(anchors.size() / 2);
PADDLE_ENFORCE_EQ(
dim_x.size(),
4,
common::errors::InvalidArgument("Input(X) should be a 4-D tensor."
"But received X dimension(%s)",
dim_x.size()));
if (iou_aware) {
PADDLE_ENFORCE_EQ(
dim_x[1],
anchor_num * (6 + class_num),
common::errors::InvalidArgument(
"Input(X) dim[1] should be equal to (anchor_mask_number * (6 "
"+ class_num)) while iou_aware is true."
"But received dim[1](%s) != (anchor_mask_number * "
"(6+class_num)(%s).",
dim_x[1],
anchor_num * (6 + class_num)));
PADDLE_ENFORCE_GE(
iou_aware_factor,
0,
common::errors::InvalidArgument(
"Attr(iou_aware_factor) should greater than or equal to 0."
"But received iou_aware_factor (%s)",
iou_aware_factor));
PADDLE_ENFORCE_LE(
iou_aware_factor,
1,
common::errors::InvalidArgument(
"Attr(iou_aware_factor) should less than or equal to 1."
"But received iou_aware_factor (%s)",
iou_aware_factor));
} else {
PADDLE_ENFORCE_EQ(
dim_x[1],
anchor_num * (5 + class_num),
common::errors::InvalidArgument(
"Input(X) dim[1] should be equal to (anchor_mask_number * (5 "
"+ class_num))."
"But received dim[1](%s) != (anchor_mask_number * "
"(5+class_num)(%s).",
dim_x[1],
anchor_num * (5 + class_num)));
}
PADDLE_ENFORCE_EQ(
dim_imgsize.size(),
2,
common::errors::InvalidArgument("Input(ImgSize) should be a 2-D tensor."
"But received Imgsize size(%s)",
dim_imgsize.size()));
if ((dim_imgsize[0] > 0 && dim_x[0] > 0) || config.is_runtime) {
PADDLE_ENFORCE_EQ(
dim_imgsize[0],
dim_x[0],
common::errors::InvalidArgument(
"Input(ImgSize) dim[0] and Input(X) dim[0] should be same."));
}
PADDLE_ENFORCE_EQ(
dim_imgsize[1],
2,
common::errors::InvalidArgument("Input(ImgSize) dim[1] should be 2."
"But received imgsize dim[1](%s).",
dim_imgsize[1]));
PADDLE_ENFORCE_GT(anchors.size(),
0,
common::errors::InvalidArgument(
"Attr(anchors) length should be greater than 0."
"But received anchors length(%s).",
anchors.size()));
PADDLE_ENFORCE_EQ(anchors.size() % 2,
0,
common::errors::InvalidArgument(
"Attr(anchors) length should be even integer."
"But received anchors length (%s)",
anchors.size()));
PADDLE_ENFORCE_GT(class_num,
0,
common::errors::InvalidArgument(
"Attr(class_num) should be an integer greater than 0."
"But received class_num (%s)",
class_num));
int64_t box_num = 0;
if ((dim_x[2] > 0 && dim_x[3] > 0) || config.is_runtime) {
box_num = dim_x[2] * dim_x[3] * anchor_num;
} else {
box_num = -1;
}
std::vector<int64_t> dim_boxes({dim_x[0], box_num, 4});
boxes->set_dims(make_ddim(dim_boxes));
boxes->set_dtype(x.dtype());
std::vector<int64_t> dim_scores({dim_x[0], box_num, class_num});
scores->set_dims(make_ddim(dim_scores));
}
void YoloBoxHeadInferMeta(const MetaTensor& x,
const std::vector<int>& anchors UNUSED,
int class_num UNUSED,
MetaTensor* out,
MetaConfig config) {
out->set_dims(x.dims());
out->set_dtype(x.dtype());
}
void ValueCompareInferMeta(const MetaTensor& x,
const MetaTensor& y,
MetaTensor* out,
MetaConfig config) {
detail::BinarySameInputDimsCheck(x, y, config);
out->set_dims(x.dims());
out->set_dtype(DataType::BOOL);
}
void SolveInferMeta(const MetaTensor& x, const MetaTensor& y, MetaTensor* out) {
auto x_dims = x.dims();
auto y_dims = y.dims();
std::vector<int64_t> x_dims_vec = vectorize(x.dims());
std::vector<int64_t> y_dims_vec = vectorize(y.dims());
auto x_dims_n = x_dims_vec.size();
auto y_dims_n = y_dims_vec.size();
PADDLE_ENFORCE_GT(x_dims_n,
1,
common::errors::InvalidArgument(
"The input tensor X's dimensions of SolveOp "
"should be larger than 1. But received X's "
"dimensions = %d, X's shape = [%s]",
x_dims_n,
x_dims));
PADDLE_ENFORCE_GE(y_dims_n,
1,
common::errors::InvalidArgument(
"The input tensor Y's dimensions of SolveOp "
"should be larger than or equal 1. But received Y's "
"dimensions = %d, Y's shape = [%s]",
y_dims_n,
y_dims));
PADDLE_ENFORCE_EQ(x_dims[x_dims_n - 2],
x_dims[x_dims_n - 1],
common::errors::InvalidArgument(
"The inner-most 2 dimensions of Input(X) all should "
"be square matrices "
"But received X's shape[-2] = %d and shape[-1] = %d.",
x_dims[x_dims_n - 2],
x_dims[x_dims_n - 1]));
bool x_broadcasted = false, y_broadcasted = false;
bool trans_x = false, trans_y = false;
if (x_dims_n == 1) {
x_dims_vec.insert(x_dims_vec.begin(), 1);
x_dims_n = 2;
x_broadcasted = true;
}
if (y_dims_n == 1) {
y_dims_vec.push_back(1);
y_dims_n = 2;
y_broadcasted = true;
}
size_t M = 0, N = 0;
if (trans_x) {
M = x_dims_vec[x_dims_n - 1];
} else {
M = x_dims_vec[x_dims_n - 2];
}
if (trans_y) {
N = y_dims_vec[y_dims_n - 2];
} else {
N = y_dims_vec[y_dims_n - 1];
}
std::vector<int64_t> new_dims;
if (x_dims_n >= y_dims_n) {
new_dims.assign(x_dims_vec.begin(), x_dims_vec.end() - 2);
} else {
new_dims.assign(y_dims_vec.begin(), y_dims_vec.end() - 2);
}
if (!x_broadcasted) {
new_dims.push_back(M); // NOLINT
}
if (!y_broadcasted) {
new_dims.push_back(N); // NOLINT
}
if (x_broadcasted && y_broadcasted) {
new_dims.push_back(1);
}
auto out_dims = make_ddim(new_dims);
out->set_dims(out_dims);
out->set_dtype(x.dtype());
out->set_layout(x.layout());
out->share_lod(x);
}
void SwiGLUInferMeta(const MetaTensor& x,
const MetaTensor& y,
MetaTensor* out) {
if (y) {
auto x_numel = common::product(x.dims());
auto y_numel = common::product(y.dims());
// skip 0-size
if (x_numel != 0 && y_numel != 0) {
PADDLE_ENFORCE_EQ(
x.dims(),
y.dims(),
common::errors::InvalidArgument("The shape of Input(X) should be "
"equal of the shape of Input(Y)."));
}
out->share_meta(x);
// If y is 0-size, out is 0-size
if (x_numel != 0 && y_numel == 0) {
out->set_dims(y.dims());
}
} else {
auto dims = x.dims();
PADDLE_ENFORCE_EQ(
dims[dims.size() - 1] % 2,
0,
common::errors::InvalidArgument(
"The last dim of Input(X) should be exactly divided by 2."));
dims[dims.size() - 1] /= 2;
out->set_dims(dims);
out->set_dtype(x.dtype());
out->set_layout(x.layout());
out->share_lod(x);
}
}
void UnpoolInferMeta(const MetaTensor& x,
const MetaTensor& indices,
const std::vector<int>& ksize,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const IntArray& output_size,
const std::string& data_format,
MetaTensor* out,
MetaConfig config) {
auto in_x_dims = x.dims();
auto in_y_dims = indices.dims();
PADDLE_ENFORCE_EQ(in_x_dims.size() == 4,
true,
common::errors::InvalidArgument(
"Unpool Input(X) must be of 4-dimensional, but "
"received Input(X)'s dimensions is %d.",
in_x_dims.size()));
PADDLE_ENFORCE_EQ(in_x_dims,
in_y_dims,
common::errors::InvalidArgument(
"The dimensions of Input(X) must equal to be "
"the dimensions of Input(Indices), but received "
"dimensions of Input(X) is [%d], received dimensions "
"of Input(Indices) is [%d]",
in_x_dims,
in_y_dims));
std::vector<int64_t> output_shape({in_x_dims[0], in_x_dims[1]});
std::vector<int64_t> output_size_val(output_size.size(), -1);
if (config.is_runtime || !output_size.FromTensor()) {
output_size_val = output_size.GetData();
}
for (int i = 0; i < static_cast<int>(ksize.size()); ++i) {
if (!config.is_runtime && in_x_dims[i + 2] <= 0) {
output_shape.push_back(-1);
} else {
output_shape.push_back(output_size_val[i]);
}
}
if (out != nullptr) {
out->set_dims(make_ddim(output_shape));
out->set_dtype(x.dtype());
}
}
void Unpool3dInferMeta(const MetaTensor& x,
const MetaTensor& indices,
const std::vector<int>& ksize,
const std::vector<int>& strides,
const std::vector<int>& paddings,
const std::vector<int>& output_size,
const std::string& data_format,
MetaTensor* out,
MetaConfig config) {
auto in_x_dims = x.dims();
auto in_y_dims = indices.dims();
PADDLE_ENFORCE_EQ(in_x_dims.size() == 5,
true,
common::errors::InvalidArgument(
"Unpool Input(X) must be of 5-dimensional, but "
"received Input(X)'s dimensions is %d.",
in_x_dims.size()));
PADDLE_ENFORCE_EQ(in_x_dims,
in_y_dims,
common::errors::InvalidArgument(
"The dimensions of Input(X) must equal to be "
"the dimensions of Input(Indices), but received "
"dimensions of Input(X) is [%d], received dimensions "
"of Input(Indices) is [%d]",
in_x_dims,
in_y_dims));
std::vector<int64_t> output_shape({in_x_dims[0], in_x_dims[1]});
for (int i = 0; i < static_cast<int>(ksize.size()); ++i) {
if (!config.is_runtime && in_x_dims[i + 2] <= 0) {
output_shape.push_back(-1);
} else {
output_shape.push_back(output_size[i]);
}
}
if (out != nullptr) {
out->set_dims(make_ddim(output_shape));
out->set_dtype(x.dtype());
}
}
void WeightDequantizeInferMeta(const MetaTensor& x,
const MetaTensor& scale,
const std::string& algo,
const int32_t group_size,
MetaTensor* out) {
PADDLE_ENFORCE_EQ(x.dims().size(),
2UL,
common::errors::InvalidArgument(
"The x tensor of dequantize op must be 2D, but got[%d]",
x.dims().size()));
PADDLE_ENFORCE_EQ(
(group_size == -1 || group_size == 64 || group_size == 128),
true,
common::errors::InvalidArgument("group_size must be -1, 64 or 128."));
auto dim_scale = scale.dims();
int64_t real_channel_shape = -1;
if (algo == "weight_only_int8") {
real_channel_shape = x.dims()[0];
} else if (algo == "weight_only_int4") {
real_channel_shape = x.dims()[0] * 2;
} else {
PADDLE_THROW(common::errors::InvalidArgument(
"Currently, we only support weight_only_int8"
" and weight_only_int4 algo."));
}
// per-channel dequantization
if (group_size == -1) {
PADDLE_ENFORCE_EQ(dim_scale.size(),
1UL,
common::errors::InvalidArgument(
"The scale tensor of dequantize op must "
"be 1D in per-channel mode, but got[%d]",
scale.dims().size()));
PADDLE_ENFORCE_EQ(dim_scale[0],
real_channel_shape,
common::errors::InvalidArgument(
"The scale tensor's shape must be equal to the x "
"tensor's shape, but got [%d] not equal to [%d]",
scale.dims()[0],
x.dims()[0]));
} else /* groupwise dequantization */ {
PADDLE_ENFORCE_EQ(dim_scale.size(),
2UL,
common::errors::InvalidArgument(
"The scale tensor of dequantize op must "
"be 2D in group-wise mode, but got[%d]",
scale.dims().size()));
PADDLE_ENFORCE_EQ(
dim_scale[0],
(x.dims()[1] + (group_size - 1)) / group_size,
errors::InvalidArgument("The input(weight_scale) dim[0] must be equal "
"to (Input(weight).dim[1] + (group_size -1))"
" / group_size. "
"But receive %d and %d",
dim_scale[0],
(x.dims()[1] + (group_size - 1)) / group_size));
PADDLE_ENFORCE_EQ(dim_scale[1],
real_channel_shape,
common::errors::InvalidArgument(
"The scale tensor's shape must be equal to the real "
"channel size, but got [%d] not equal to [%d]",
scale.dims()[0],
real_channel_shape));
}
int64_t n = x.dims()[1];
int64_t k = real_channel_shape;
out->set_dims(make_ddim({n, k}));
out->set_dtype(scale.dtype());
}
void FusedRMSNormInferMeta(const MetaTensor& x,
const MetaTensor& scale,
float epsilon,
MetaTensor* y,
MetaTensor* invvar) {
auto x_shape = x.dims();
auto scale_shape = scale.dims();
PADDLE_ENFORCE_EQ(
scale_shape.size(),
1,
common::errors::InvalidArgument(
"The scale tensor must be 1D, but got[%d]", scale_shape.size()));
PADDLE_ENFORCE_EQ(scale_shape[0],
x_shape[x_shape.size() - 1],
common::errors::InvalidArgument(
"The scale tensor's shape must be equal to the last "
"dimension of x tensor, but got [%d] not equal to [%d]",
scale_shape[0],
x_shape[x_shape.size() - 1]));
PADDLE_ENFORCE_EQ(
x.dtype() == DataType::FLOAT32 || x.dtype() == DataType::FLOAT16 ||
x.dtype() == DataType::BFLOAT16,
true,
common::errors::InvalidArgument(
"The dtype of x must be FLOAT32, FLOAT16 or BFLOAT16, but got [%s]",
x.dtype()));
PADDLE_ENFORCE_EQ(
scale.dtype() == DataType::FLOAT32 ||
scale.dtype() == DataType::FLOAT16 ||
scale.dtype() == DataType::BFLOAT16,
true,
common::errors::InvalidArgument("The dtype of scale must be FLOAT32, "
"FLOAT16 or BFLOAT16, but got [%s]",
scale.dtype()));
y->set_dims(x.dims());
y->set_dtype(scale.dtype());
invvar->set_dims({-1});
invvar->set_dtype(DataType::FLOAT32);
}
void BatchedGemmInferMeta(const MetaTensor& lhs,
const MetaTensor& rhs,
const std::vector<int64_t>& batch_sizes,
const bool trans_lhs,
const bool trans_rhs,
MetaTensor* output) {
const bool is_layout_invalid = (trans_lhs == true) && (trans_rhs == true);
const auto lhs_shape = lhs.dims();
const auto rhs_shape = rhs.dims();
const int64_t total_tokens = lhs_shape[0];
const int64_t num_experts = batch_sizes.size();
PADDLE_ENFORCE_EQ(
is_layout_invalid,
false,
common::errors::InvalidArgument(
"We don't support both lhs and rhs are transposed at the same time"));
PADDLE_ENFORCE_EQ(
(lhs.dtype() == DataType::BFLOAT16 || lhs.dtype() == DataType::FLOAT32) &&
(rhs.dtype() == DataType::BFLOAT16 ||
rhs.dtype() == DataType::FLOAT32) &&
lhs.dtype() == rhs.dtype(),
true,
common::errors::InvalidArgument(
"The dtype of lhs and rhs must both be BFLOAT16 or both be FLOAT32, "
"but got [%s] and [%s]",
lhs.dtype(),
rhs.dtype()));
PADDLE_ENFORCE_EQ(
lhs_shape.size(),
2,
common::errors::InvalidArgument(
"The lhs's dimension must be 2, but got[%d]", lhs_shape.size()));
PADDLE_ENFORCE_EQ(
rhs_shape.size(),
trans_lhs ? 2 : 3,
common::errors::InvalidArgument(
"The rhs's dimension must be 3, but got[%d]", rhs_shape.size()));
// We expect layout below:
// 1. trans_lhs = false && trans_rhs = false (group forward) :
// [M_total, input_hidden_size] x [num_experts, input_hidden_size,
// output_hidden_size] output: [M_total, output_hidden_size]
//
// 2. trans_lhs = false && trans_rhs = true (backward for lhs_grad, or
// specialized forward):
// [M_total, output_hidden_size] x [num_experts, input_hidden_size,
// output_hidden_size]' output: [M_total, input_hidden_size]
//
// 3. trans_lhs = true && trans_rhs = false (backward for rhs_grad) :
// [M_total, input_hidden_size]' x [M_total, output_hidden_size]
// output: [num_experts, input_hidden_size, output_hidden_size]
if (!trans_lhs) {
// =============================================================================
// Case 1 and 2: group forward or lhs_grad (input_grad)
// Note that this case implements grouped gemm, mapping hidden_lhs to
// hidden_out For each expert i, This case views lhs as [Mi x K] and rhs as
// [E x K x N] or [E x N x K], so the output is [Mtotal x N], N could be
// input_hidden_size or output_hidden_size.
const int64_t hidden_out = trans_rhs ? rhs_shape[1] : rhs_shape[2];
output->set_dims(make_ddim({total_tokens, hidden_out}));
} else {
// =============================================================================
// Case 3: group backward for rhs_grad (weight_grad)
// Note that this case implements k-grouped gemm
// For each expert i, this case views lhs as [K x Mi] and rhs as [Mi x N],
// so the output is [E x K x N].
PADDLE_ENFORCE_EQ(lhs_shape[0],
rhs_shape[0],
common::errors::InvalidArgument(
"The lhs's first dim must be equal to the rhs's "
"proposal, but got[%d] instead of [%d]",
lhs_shape[0],
rhs_shape[0]));
const int64_t hidden_in = lhs_shape[1];
const int64_t hidden_out = rhs_shape[1];
output->set_dims(make_ddim({num_experts, hidden_in, hidden_out}));
}
output->set_dtype(DataType::BFLOAT16);
}
} // namespace phi
PD_REGISTER_INFER_META_FN(add_raw, phi::ElementwiseRawInferMeta);
Reference in New Issue View Git Blame Copy Permalink