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.

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

Resnet34 test on CDF #18

Open
YU-SHAO-XU opened this issue Sep 29, 2022 · 8 comments
Open

Resnet34 test on CDF #18

YU-SHAO-XU opened this issue Sep 29, 2022 · 8 comments

Comments

@YU-SHAO-XU
Copy link

YU-SHAO-XU commented Sep 29, 2022

Hi
I train from scratch by using ResNet-34 with the Adam optimizer, unfortunately test on CDF auc only 59%. is there provide pretrained ResNet-34 model and weight!?
ResNet-34

@YU-SHAO-XU
Copy link
Author

@mapooon

@YU-SHAO-XU
Copy link
Author

is efficientnet-b4 also pretrained on SBI ? or only pretrained on imagenet??
so resnet34 need pretrained on imagenet too ?

@mapooon
Copy link
Owner

mapooon commented Oct 5, 2022

Hi,
We provide efnb4 weights trained on SBIs.
If you train other models(e.g., ResNet34) on SBIs, you need to

  • start with imagenet pretrained weights or other pretrained ones because we observe training may fail when start with randomly initialized weights.
  • resize input images to the proper size, e.g., 224*224 for ResNet34

@YU-SHAO-XU
Copy link
Author

what is the adam parameter if i use resnet 34 ? or still same ?

@YU-SHAO-XU
Copy link
Author

would u mind provide file of model.py (resnet) to me ? my mail : gyes503770@gmail.com

@YU-SHAO-XU
Copy link
Author

resize input images to 224*224 for ResNet34 is including base.json and sbi.py ? or just base.json !?

@YU-SHAO-XU
Copy link
Author

i use you provided efnb4 weights test on CDF is 93.18, but if i train from scratch efnb4 only 0.8995 !

@YU-SHAO-XU
Copy link
Author

would u mind provide me the version of packages, i wanna doble check !! beacuse my result not as good as paper

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants