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.

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

question abour temporal feature aggregation #1

Open
eliliu2233 opened this issue May 1, 2024 · 1 comment
Open

question abour temporal feature aggregation #1

eliliu2233 opened this issue May 1, 2024 · 1 comment

Comments

@eliliu2233
Copy link

Thanks for your exciting work! I have two questions about the temporal feature aggregation:

  1. Does the AFFM use deformable attention, or only use original attention?
  2. Does the AFFM shared parameters? Suppose I have temporal BEV features at {T, T-1, T-2}, and I need 4 AFFM to finish the temporal aggregation as introduced in section3.2. Does these 4 AFFM shared parameters? Or I need to initiate 4 attention modules with different parameters?
    Looking forward to your reply.
@xiazhongyv
Copy link
Collaborator

  1. We use the original cross attention.
  2. All AFFMs of the forward process share parameters, and all AFFMs of the reverse process share parameters, so only two AFFM modules need to be instantiated.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants