Skip to content
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

Add new weights for ResNet101 and ResNet151 #4811

Merged
merged 2 commits into from
Nov 1, 2021
Merged

Add new weights for ResNet101 and ResNet151 #4811

merged 2 commits into from
Nov 1, 2021

Conversation

datumbox
Copy link
Contributor

@datumbox datumbox commented Nov 1, 2021 •
edited
Loading

Fixes partially #3995

torchrun --nproc_per_node=1 train.py --test-only --weights ImageNet1K_RefV2 --model resnet101 -b 1
Acc@1 81.728 Acc@5 95.670

torchrun --nproc_per_node=1 train.py --test-only --weights ImageNet1K_RefV2 --model resnet152 -b 1
Acc@1 82.042 Acc@5 95.926

cc @datumbox

@facebook-github-bot
Copy link

facebook-github-bot commented Nov 1, 2021 •
edited
Loading

💊 CI failures summary and remediations

As of commit bdc7c67 (more details on the Dr. CI page):

💚 💚 Looks good so far! There are no failures yet. 💚 💚

This comment was automatically generated by Dr. CI (expand for details).

Please report bugs/suggestions to the (internal) Dr. CI Users group.

Click here to manually regenerate this comment.

jdsgomes
jdsgomes approved these changes Nov 1, 2021
View reviewed changes
torchvision/prototype/models/resnet.py
@@ -127,6 +137,16 @@ class ResNet152Weights(Weights):
"acc@5": 94.046,
},
)
ImageNet1K_RefV2 = WeightEntry(
url="https://download.pytorch.org/models/resnet152-089c0848.pth",
transforms=partial(ImageNetEval, crop_size=224, resize_size=232),
Copy link
Contributor

Choose a reason for hiding this comment

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

the resize_size was not used in the original training?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Nope, that's a new optimization only available on the new weights. :-)

@datumbox datumbox merged commit 6139997 into pytorch:main Nov 1, 2021
@datumbox datumbox deleted the models/new_resnet_weights branch November 1, 2021 14:30
@datumbox datumbox linked an issue Nov 1, 2021 that may be closed by this pull request
Improve the accuracy of Classification models by using SOTA recipes and primitives #3995
Closed
facebook-github-bot pushed a commit that referenced this pull request Nov 1, 2021
@fmassa @facebook-github-bot
[fbsync] Add new weights for ResNet101 and ResNet151. (#4811)
d17cd25 
Reviewed By: datumbox

Differential Revision: D32064684

fbshipit-source-id: f42aec92bb919a7264023406b85df48656c673d5
cyyever pushed a commit to cyyever/vision that referenced this pull request Nov 16, 2021
@datumbox @cyyever
Add new weights for ResNet101 and ResNet151. (pytorch#4811)
d3bfba7
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@jdsgomes jdsgomes jdsgomes approved these changes

Assignees
No one assigned
Labels
ciflow/default cla signed enhancement module: models
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

Improve the accuracy of Classification models by using SOTA recipes and primitives
3 participants
@datumbox @facebook-github-bot @jdsgomes