[go: nahoru, domu]
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

Why do NetworkCIFAR and NetworkImageNet have different stems? #65

Open
e-271 opened this issue Jan 22, 2019 · 1 comment
Open

Why do NetworkCIFAR and NetworkImageNet have different stems? #65

e-271 opened this issue Jan 22, 2019 · 1 comment

Comments

@e-271
Copy link
e-271 commented Jan 22, 2019

Hi, Thanks for this great project. I noticed that in model.py you defined different initial stem layers for CIFAR and ImageNet, with ImageNet being more complex. Is there an intuitive reason for doing it this way, like due to the larger image sizes in ImageNet? Or is it just empirically better performance when you do it this way?
@Catosine
Copy link

Because Cifar and ImageNet have different size! For the former, each picture has a size of 32x32, while the latter has 224x224. Since the ImageNet uses cells searched on Cifar10/100, therefore it is reasonable to keep the input have the same size.

Best,

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@e-271 @Catosine