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 multiply the mask by the scaling? #2

Open
allensll opened this issue Jul 1, 2022 · 0 comments
Open

Why multiply the mask by the scaling? #2

allensll opened this issue Jul 1, 2022 · 0 comments

Comments

@allensll
Copy link

allensll commented Jul 1, 2022

Could you please give me some insight into why multiply the scaling? I found it slightly improved the training accuracy. Thanks.

    def compute_mask(self, temp, ticket):
        scaling = 1. / sigmoid(self.mask_initial_value)
        if ticket: mask = (self.mask_weight > 0).float()
        else: mask = F.sigmoid(temp * self.mask_weight)
        return scaling * mask
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
1 participant
@allensll