Skip to content

Don't end on inplace operators in einsum #22111

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
t-vi wants to merge 2 commits into from
Closed

Don't end on inplace operators in einsum #22111

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a73c1b0
Don't end on inplace updates in einsum
t-vi Jun 22, 2019
8d1ea8e
better bracing
t-vi Jun 23, 2019
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
18 changes: 13 additions & 5 deletions aten/src/ATen/native/Linear.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -119,9 +119,13 @@ static Tensor sumproduct_pair(const Tensor& left_, const Tensor& right_, IntArra

// finally squeeze summed dimensions if desired
if (! keepdim) {
for (int i = dim-1; i>=0; i--)
if (sum_dims[i])
result.squeeze_(i);
auto sizes = result.sizes().vec();
for (int i = dim-1; i>=0; i--) {
if (sum_dims[i]) {
sizes.erase(sizes.begin() + i);
Copy link
Contributor

@apaszke apaszke Jun 22, 2019

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: can you please add braces here? Braceless nested statements make me nervous 😬

Copy link
Collaborator Author

@t-vi t-vi Jun 23, 2019

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Me, too, even if I'm the one to blame for them in the first place.

}
}
result = result.view(sizes);
}
return result;
}
Expand Down Expand Up @@ -354,8 +358,12 @@ Tensor einsum(std::string eqn, TensorList tensors) {
result = at::native::sumproduct_pair(result, preprocessed_operands[i], sum_dims, true);
}
// finally, we squeeze out all non-result dimensions
for (int64_t dim = num_total_idxes-1; dim >= num_output_dims; dim--)
result.squeeze_(dim);
auto sizes = result.sizes().vec();
for (int64_t dim = num_total_idxes-1; dim >= num_output_dims; dim--) {
sizes.erase(sizes.begin() + dim);
}

result = result.view(sizes);
return result;
}

Expand Down