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

batch_size #18

Open
Pumpkin123709 opened this issue Apr 5, 2024 · 1 comment
Open

batch_size #18

Pumpkin123709 opened this issue Apr 5, 2024 · 1 comment

Comments

@Pumpkin123709
Copy link

When I set batch_size to 8, the shape of the input_ids of the model's input is [9,n]. The first dimension of input_ids is always one more than bs.

@waxnkw
Copy link
Collaborator

waxnkw commented Apr 30, 2024

The reason is that: not every batch have location input or output data, which will lead to a running error caused by partially trained modules (e.g., box decoder can not be trained if there are no location out data). Therefore, we additionally include a mock data item to avoid the error. The weight of the data item for loss calculation is set to zero.

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