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

Reduce overheads in CustomCollateFn #1

Closed

khanhvu207 opened this issue May 4, 2022 · 1 comment

Labels

Milestone

Stable supervised...

Owner

khanhvu207 commented May 4, 2022

Apparently creating new tensors via torch.tensor() creates lots of overheads (as Kien pointed out). See here for details.

def pad(a, value, l):
        a.extend([value] * l)
        return a
    
batch_team_fleet_pos_x = [[pos_x for pos_x, pos_y in x["team_fleet_pos"]] for x in batch]
batch_team_fleet_pos_x = [pad(a, 21, max_team_fleet_len-len(a)) for a in batch_team_fleet_pos_x]
batch_team_fleet_pos_x = torch.tensor(batch_team_fleet_pos_x).long()

Possible solution.

khanhvu207 added the enhancement label

khanhvu207 added this to the Stable supervised learning pipeline milestone

khanhvu207 closed this as completed

Owner Author

khanhvu207 commented May 11, 2022

Not relevant anymore :D

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment