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

use weights as defined in dataset.yaml #31

Draft
wants to merge 1 commit into
base: vinam/subtask-interval
Choose a base branch
from

Conversation

mazabou
Copy link
Member

@mazabou mazabou commented Nov 27, 2024

No description provided.

@mazabou mazabou requested a review from vinamarora8 November 27, 2024 23:13
Comment on lines +120 to +124
movement_periods.random_period: 1.0
movement_periods.hold_period: 0.1
movement_periods.reach_period: 5.0
movement_periods.return_period: 1.0
cursor_outlier_segments: 0.0
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does this work? I think movement_periods.reach_period is not present in the random-target sessions.

Comment on lines -351 to +334
"output_mask": pad8(output_mask),
"output_mask": track_mask8(output_session_index),
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I went with the pad8 route because that way we can & the evaluation mask with it (if doing eval) and don't have to pass a separate eval mask to the validator.

@vinamarora8
Copy link
Member

I like the change overall. Left some comments wrt this code being able to run like before.

@vinamarora8
Copy link
Member

Also note we're still missing invalid_periods here. Leaving it up to you to decide whether to have a separate PR for that.

@mazabou
Copy link
Member Author

mazabou commented Nov 28, 2024

Sounds good, I wanted to make a quick draft of the overall concept. Since you agree with the change overall, I will go ahead and implement it!

The invalid_periods is addressed through cursor_outlier_segments: 0.0.

@mazabou mazabou marked this pull request as draft November 28, 2024 16:27
@vinamarora8
Copy link
Member

The invalid_periods is addressed through cursor_outlier_segments: 0.0.

It was only outliers that had weight 0. Invalid trials had weight 0.1

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

Successfully merging this pull request may close these issues.

2 participants