[go: up one dir, main page]
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

Slightly more efficient fully-connected layer #235

Open
m-usama-z opened this issue Aug 7, 2024 · 0 comments
Open

Slightly more efficient fully-connected layer #235

m-usama-z opened this issue Aug 7, 2024 · 0 comments
Labels
enhancement New feature or request

Comments

@m-usama-z
Copy link
m-usama-z commented Aug 7, 2024 •
edited

Hi. I suggest changing the following piece of code in the reach_star_single_input function of the FullyConnectedLayer.m class

                V(1, 1, :, in_image.numPred + 1) = zeros(obj.OutputSize, 1, 'like', in_image.V);
                for i=1:n+1
                    I = in_image.V(:,:,:,i);
                    I = reshape(I,N,1); % flatten input
                    if i==1
                        V(1, 1,:,i) = obj.Weights*I + obj.Bias;
                    else
                        V(1, 1,:,i) = obj.Weights*I;
                    end
                end

to this:

                V(1, 1, :, :) = obj.Weights*reshape(in_image.V, N, n + 1);
                V(1, 1, :, 1) = reshape(V(1, 1, :, 1), obj.OutputSize, 1) + obj.Bias;

in order to improve efficiency for a large number of generators, for both cpu and gpu modes.
@mldiego mldiego added the enhancement New feature or request label Aug 20, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
enhancement New feature or request
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@mldiego @m-usama-z