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

[timeseries] Use different random seed for each model #4099

Merged
merged 3 commits into from
Apr 15, 2024

Conversation

shchur
Copy link
Collaborator

@shchur shchur commented Apr 15, 2024

Description of changes:

  • Increment the random seed used to train each consecutive model by 1. This fixes a weird regression for GluonTS models: when we use the same seed for all GluonTS models, their performance becomes correlated and sometimes leads to an overall performance regression. This likely happens because, when the seed is fixed, the GluonTS dataloaders returns batches in the same order.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

@yinweisu
Copy link
Collaborator

Previous CI Run Current CI Run

Copy link
Contributor

@canerturkmen canerturkmen left a comment

Choose a reason for hiding this comment

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

LGTM! Thanks 🦸

@shchur shchur merged commit 949a781 into autogluon:master Apr 15, 2024
27 checks passed
@shchur shchur deleted the increment-seed branch April 15, 2024 13:31
Copy link

Job PR-4099-52add0e is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-4099/52add0e/index.html

LennartPurucker pushed a commit to LennartPurucker/autogluon that referenced this pull request Jun 1, 2024
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.

None yet

3 participants