mirror of
https://github.com/zebrajr/pytorch.git
synced 2025-12-07 12:21:27 +01:00
7f33ec55b2
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/9350 Re-apply #9270 Breaking this out of #8338 This takes care of the Eigen failure we saw on Mac CUDA builds when BUILD_CAFFE2 and BUILD_ATEN were removed. Fix is to isolate Eigen from headers included by cu files and processed by nvcc. This was worked on with smessmer. Reviewed By: mingzhe09088 Differential Revision: D8794431 fbshipit-source-id: de656334af46c697802073f8e8d9a6aeb9ca65a7
75 lines
1.8 KiB
C++
75 lines
1.8 KiB
C++
#include "caffe2/operators/acos_op.h"
|
|
#include "caffe2/utils/eigen_utils.h"
|
|
|
|
#include <algorithm>
|
|
#include <functional>
|
|
|
|
namespace caffe2 {
|
|
|
|
template <>
|
|
template <typename T>
|
|
bool AcosGradientFunctor<CPUContext>::Forward(
|
|
const std::vector<int>& X_dims,
|
|
const std::vector<int>& /* dY_dims */,
|
|
const T* X,
|
|
const T* dY,
|
|
T* dX,
|
|
CPUContext* /* context */) const {
|
|
const int size = std::accumulate(
|
|
X_dims.cbegin(), X_dims.cend(), 1, std::multiplies<int>());
|
|
ConstEigenVectorArrayMap<T> dY_arr(dY, size);
|
|
ConstEigenVectorArrayMap<T> X_arr(X, size);
|
|
EigenVectorMap<T>(dX, size) = -dY_arr * (T(1) - X_arr.square()).rsqrt();
|
|
return true;
|
|
}
|
|
|
|
REGISTER_CPU_OPERATOR(
|
|
Acos,
|
|
UnaryElementwiseOp<
|
|
TensorTypes<float>,
|
|
CPUContext,
|
|
AcosFunctor<CPUContext>>);
|
|
REGISTER_CPU_OPERATOR(
|
|
AcosGradient,
|
|
BinaryElementwiseOp<
|
|
TensorTypes<float>,
|
|
CPUContext,
|
|
AcosGradientFunctor<CPUContext>>);
|
|
|
|
OPERATOR_SCHEMA(Acos)
|
|
.NumInputs(1)
|
|
.NumOutputs(1)
|
|
.IdenticalTypeAndShape()
|
|
.SetDoc(R"DOC(
|
|
Calculates the arccosine of the given input tensor, element-wise.
|
|
)DOC")
|
|
.Input(0, "input", "Input tensor")
|
|
.Output(
|
|
0,
|
|
"output",
|
|
"The arccosine of the input tensor computed element-wise");
|
|
|
|
OPERATOR_SCHEMA(AcosGradient)
|
|
.NumInputs(2)
|
|
.NumOutputs(1)
|
|
.IdenticalTypeAndShape();
|
|
|
|
namespace {
|
|
|
|
class GetAcosGradient : public GradientMakerBase {
|
|
using GradientMakerBase::GradientMakerBase;
|
|
std::vector<OperatorDef> GetGradientDefs() override {
|
|
return SingleGradientDef(
|
|
"AcosGradient",
|
|
"",
|
|
std::vector<std::string>{I(0), GO(0)},
|
|
std::vector<std::string>{GI(0)});
|
|
}
|
|
};
|
|
|
|
} // namespace
|
|
|
|
REGISTER_GRADIENT(Acos, GetAcosGradient);
|
|
|
|
} // namespace caffe2
|