wehub-resource-sync
66 lines
2.1 KiB
C++
66 lines
2.1 KiB
C++
/* Copyright (c) 2022 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/kernels/unstack_grad_kernel.h"
|
|
|
|
#include "paddle/phi/backends/xpu/enforce_xpu.h"
|
|
#include "paddle/phi/core/kernel_registry.h"
|
|
|
|
namespace phi {
|
|
|
|
template <typename T, typename Context>
|
|
void UnStackGradKernel(const Context &dev_ctx,
|
|
const std::vector<const DenseTensor *> &x,
|
|
int axis,
|
|
DenseTensor *x_grad) {
|
|
using XPUType = typename XPUTypeTrait<T>::Type;
|
|
if (axis < 0) {
|
|
axis += x[0]->dims().size() + 1;
|
|
}
|
|
dev_ctx.template Alloc<T>(x_grad);
|
|
auto &dim = x[0]->dims();
|
|
std::vector<int64_t> xdims;
|
|
for (auto i = 0; i < dim.size(); ++i) {
|
|
xdims.push_back(dim[i]);
|
|
}
|
|
xdims.push_back(1);
|
|
std::vector<std::vector<int64_t>> xdims_list;
|
|
size_t n = static_cast<size_t>(x.size());
|
|
for (size_t i = 0; i < n; i++) {
|
|
xdims_list.push_back(xdims);
|
|
}
|
|
|
|
std::vector<const XPUType *> x_list;
|
|
for (size_t i = 0; i < n; i++) {
|
|
x_list.push_back(reinterpret_cast<const XPUType *>(x[i]->data<T>()));
|
|
}
|
|
|
|
int r = xpu::concat<XPUType>(dev_ctx.x_context(),
|
|
x_list,
|
|
reinterpret_cast<XPUType *>(x_grad->data<T>()),
|
|
xdims_list,
|
|
axis);
|
|
PADDLE_ENFORCE_XDNN_SUCCESS(r, "concat in unstack_grad op");
|
|
}
|
|
} // namespace phi
|
|
|
|
PD_REGISTER_KERNEL(unstack_grad,
|
|
XPU,
|
|
ALL_LAYOUT,
|
|
phi::UnStackGradKernel,
|
|
float,
|
|
phi::float16,
|
|
int,
|
|
int64_t) {}
|