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

The ckpt has a social decoder, but the code does not. #20

Open

xushilin1 opened this issue Jul 17, 2022 · 1 comment

xushilin1 commented Jul 17, 2022

Thanks for your code. I encountered several questions when reproducing this paper.
I hope you can help me solve these questions.

Code did not have a social decoder, but the ckpt you provided has. I'm interested in how you trained this ckpt.
What is the classification loss you choose for six trajectory proposals? CrossEntropy or KL and what is the target ?
What is the loss weight for reg loss and cls loss?
Thanks for your project again and I really hope to get your reply.

DinghaoXi commented Jul 17, 2022 via email

这是来自QQ邮箱的假期自动回复邮件。您好，我已经收到邮件，会尽快给您回复。by小溪

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