Skip to content
This repository has been archived by the owner on Jul 2, 2023. It is now read-only.

Gradient explosion occurs after modifying the weight initialization method #167

Open
eadstry opened this issue Oct 22, 2022 · 0 comments
Open

Gradient explosion occurs after modifying the weight initialization method #167

eadstry opened this issue Oct 22, 2022 · 0 comments

Comments

@eadstry
Copy link

eadstry commented Oct 22, 2022

I wanted to change the initialization method of the convolution from 0 to a positive-terrestrial distribution, but this led to a gradient explosion after a few iters.Why does this happen?
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@eadstry