Skip to content

fix a typo for natural gradient #3034

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

Closed
ahban wants to merge 1 commit into from
Closed

fix a typo for natural gradient #3034

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion src/nnet3/natural-gradient-online.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -375,7 +375,7 @@ namespace nnet3 {
* Initialization *

Now, a note on what we do on time t = 0, i.e. for the first minibatch. We
initialize X_0 to the top R eigenvectors of 1/N X_0 X_0^T, where N is the
initialize R_0 to the top R eigenvectors of 1/N X_0 X_0^T, where N is the
Copy link
Contributor

@danpovey danpovey Feb 15, 2019

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think there might be another problem here. I believe X is (N by D) where N is the minibatch size. (Although elsewhere I seem to have used M for the minibatch size so that might be a better letter). We want the eigenvectors of a D x D matrix, so it should be X_0^T X_0. That appears in the lines immediately below, too.

minibatch size (num-rows of R0). If L is the corresponding RxR diagonal
matrix of eigenvalues, then we will set D_0 = L - \rho_0 I. We set \rho_0
to ensure that
Expand Down