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

关于*local里面的base_size和train_szie #26

Open
xiaodaobuzhi opened this issue Dec 18, 2023 · 1 comment
Open

关于*local里面的base_size和train_szie #26

xiaodaobuzhi opened this issue Dec 18, 2023 · 1 comment

Comments

@xiaodaobuzhi
Copy link

你好,这个*local的类里面的base_size为什么设置成train size的1.5倍,还有这个base_size是什么意思呢,train_size是只网络训练的时候的数据尺寸吗?谢谢

@achusky
Copy link
Collaborator

achusky commented May 14, 2024

你好,抱歉现在才回复你

base_size可以理解为测试时“虚拟的数据裁切尺寸(local window)”。它并不会真的让网络在测试时裁切原始数据,但会影响“数据统计操作”的感受野。例如,SE中的avgpool的pool size。
它是一个超参数,可以根据需要进行设置。默认设置为训练数据尺寸(train_size)的1.5倍。该超参数是两方面的权衡:

  1. 更大的数值可以让“数据统计操作”有更大的感受野,从而获得更好的性能
  2. 更大的数值会增加训练和测试的不一致性,这会带来负面影响

因此默认值根据经验设置为1.5来平衡正面影响和负面影响。具体实验和解释见论文章节4.2

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