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

PreResNet-110 on Cifar100, Top1 error rate is 26.47 rather than 23.65. #37

Open
WeixiangXu opened this issue Sep 11, 2019 · 3 comments
Open

Comments

@WeixiangXu
Copy link

I trained PreResNet-110 on Cifar100 as training recipes. And I get top1 error rate: 26.47%.

@hukkai
Copy link

hukkai commented Nov 6, 2019

Same here, cannot reach 23.65. But I can only get 28. Could you give me some hints about the dropped accuracy?

@BeierZhu
Copy link

Same here, I get top1 error rate ~ 28%

@hukkai
Copy link

hukkai commented Dec 30, 2019

@BeierZhu Now I can confirm it should be 28%.
Please see this issue #36 (comment).

The reason is that some people use the easier version which is indeed PreResNet 164 with Bottleneck (higher accuracy) while the current version is PreResNet110 with BasicBlock.

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

3 participants