Skip to content

Merge pixelshuffle function into module #375

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

Jump to bottom
Merged
soumith merged 1 commit into from
Dec 30, 2016
Merged

Merge pixelshuffle function into module #375

soumith merged 1 commit into from
Dec 30, 2016

Conversation

@colesbury
Copy link
Member

@colesbury colesbury commented Dec 30, 2016

The operations used by PixelShuffle are differentiable (via autograd), so we only have to implement the forward pass in the Module and get the backward automatically.

cc @alykhantejani

@soumith soumith merged commit 235d540 into pytorch:master Dec 30, 2016
@apaszke
Copy link
Contributor

apaszke commented Dec 30, 2016

We should put the implementation in functional, not in the module 😕

@alykhantejani
Copy link
Contributor

alykhantejani commented Dec 30, 2016

Ah cool, did not know that autograd would handle the permutations. As long as the following test passed:

        input = Variable(torch.Tensor(batch_size, channels, height, width).uniform_(), requires_grad=True)
        ps = nn.PixelShuffle(upscale_factor)
        output = ps(input)
        output.backward(output.data)
        self.assertEqual(input.data, input.grad)

i.e. the input == backwards(forwards(input)) then it LGTM.

This test is already in test_nn so that means it should work fine with autograd. Nice spot @colesbury

mrshenli pushed a commit to mrshenli/pytorch that referenced this pull request Apr 11, 2020
@yf225
Clean up build.sh (pytorch#375)
c64413c 
* Re-enable hybrid_frontend tutorials

* clean up
KsenijaS pushed a commit to KsenijaS/pytorch that referenced this pull request Dec 14, 2020
@pluradj-ibm @wenbingl
Removed tinyyolov2-1 (pytorch#375)
b148b55 
Model for opset 1 fails validation in two ways. Models for more recent
opsets are already available.

* Fixes pytorch#55 - all initializers not declared as inputs
* Fixes pytorch#13 - missing consumed_inputs for BatchNormalization nodes

Signed-off-by: Jason Plurad <[email protected]>

Co-authored-by: Wenbing Li <[email protected]>
KyleCZH pushed a commit to KyleCZH/pytorch that referenced this pull request Sep 20, 2021
@peterjc123 @soumith
Add version checks to smoke tests (pytorch#375)
afc3799
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

1 more reviewer

@soumith soumith soumith approved these changes

Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@colesbury @apaszke @alykhantejani @soumith