-
Notifications
You must be signed in to change notification settings - Fork 37
Adding TCN decoder. #9
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
base: alpha
Are you sure you want to change the base?
Conversation
| y = torch.normal( # pylint: disable=no-member | ||
| mean=batch["player_future"][..., :2], | ||
| mean=target_mean, | ||
| std=torch.ones_like(batch["player_future"][..., :2]) * noise_level, # pylint: disable=no-member |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
use target_mean instead of batch["player_future"][..., :2]
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@Filangel it looks like I'm already using target_mean. Can you clarify?
|
|
||
| # Calculates loss (NLL). | ||
| loss = -torch.mean(log_prob - logabsdet, dim=0) # pylint: disable=no-member | ||
| if use_tcn: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
the API for the decoder should be the same, regardless if it's an RNN or a TCN
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can you clarify what you mean by this? By API, do you want the TCN decoder to similarly have a .inverse method, as the Autoregressive decoder does?
No description provided.