Skip to content

[XPU] support 0-size tensor for MaskedSelectGrad #76905

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
cqulilujia wants to merge 1 commit into from
Closed

[XPU] support 0-size tensor for MaskedSelectGrad #76905

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 7 additions & 0 deletions paddle/phi/kernels/xpu/masked_select_grad_kernel.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,7 @@

#include "paddle/phi/backends/xpu/enforce_xpu.h"
#include "paddle/phi/core/kernel_registry.h"
#include "paddle/phi/kernels/full_kernel.h"

namespace phi {

Expand All @@ -25,6 +26,12 @@ void MaskedSelectGradKernel(const Context& dev_ctx,
const DenseTensor& mask,
const DenseTensor& out_grad,
DenseTensor* x_grad) {
if (out_grad.numel() == 0 && x_grad) {
// x = [1, 2], mask = [False, False], out = [], x_grad = [0, 0]
phi::Full<T, Context>(
dev_ctx, phi::IntArray(common::vectorize(x_grad->dims())), 0, x_grad);
return;
}
using XPUType = typename XPUTypeTrait<T>::Type;
auto* mask_data = mask.data<bool>();
auto* input_data = reinterpret_cast<const XPUType*>(out_grad.data<T>());
Expand Down
Loading