Skip to content

Conversation

puhuk
Copy link
Contributor

@puhuk puhuk commented May 9, 2022

Add note for resnet architecture

Closes #5964

Add note for resnet architecture
while original implementation places the stride at the first 1x1 convolution(self.conv1)
according to "Deep residual learning for image recognition"https://arxiv.org/abs/1512.03385.
This variant is also known as ResNet V1.5 and improves accuracy according to
https://ngc.nvidia.com/catalog/model-scripts/nvidia:resnet_50_v1_5_for_pytorch.
Copy link
Contributor

Choose a reason for hiding this comment

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

This is added on the wrong place. ResNet18 and 34 use the basic BasicBlock not Bottleneck. It's the 50, 101 and 152 versions of the model that use the Bottleneck class.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Sorry, let me correct soon !

@NicolasHug NicolasHug changed the title To resolve issue #5964 Document ResNet architecture tweak May 10, 2022
@puhuk
Copy link
Contributor Author

puhuk commented May 12, 2022

@datumbox I'm having errors with unittest_onnx which I did not change. Could you give me some clue to resolve this error. Or is there any other reason that cause the error?

@datumbox
Copy link
Contributor

@puhuk that's not something you broke. The main branch is also failing. Feel free to ignore it.

@datumbox datumbox marked this pull request as draft May 12, 2022 13:56
@datumbox datumbox marked this pull request as ready for review May 20, 2022 10:00
@datumbox
Copy link
Contributor

The failures on the CI are due to the https://ghcr.io/v2/homebrew/core/gmp/manifests/6.2.1_1 being down temporarily. I checked the docs and they render fine.

@datumbox datumbox merged commit 37665a0 into pytorch:main May 20, 2022
facebook-github-bot pushed a commit that referenced this pull request Jun 1, 2022
Summary:
* To resolve issue #5964

Add note for resnet architecture

* Update resnet.py

* Update resnet.py

* Update resnet.rst

* Fix stylings

* Add the same notes on model builders

* Improve description

* Apply the change everywhere

* Remove trailing space

Reviewed By: NicolasHug

Differential Revision: D36760934

fbshipit-source-id: 044ff1d1f35f6354dbc7608a0d30951aa90190a2

Co-authored-by: Vasilis Vryniotis <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Document ResNet architecture tweak
3 participants