We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
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
你好,请问论文中所说的step-wise patch embedding的实现具体体现在哪里呢?是通过不同stage设置不同patch size的patch embedding来体现的吗?
The text was updated successfully, but these errors were encountered:
你好,是这样的,step-wise patch embedding就是每一步做embedding时只做尺度为2x2大小的patch embedding,具体就是每次使用步长为2,卷积核大小为2的卷积,类似于CNN中的下采样层,同时通道数加倍。结果就是原来8x8的patch embedding被多个2x2的patch embedding替代。
Sorry, something went wrong.
多谢回复!
请问一般overlapping的cnn stem更好,还是(你们的)non-overlapping的更好呢?
No branches or pull requests
你好,请问论文中所说的step-wise patch embedding的实现具体体现在哪里呢?是通过不同stage设置不同patch size的patch embedding来体现的吗?
The text was updated successfully, but these errors were encountered: