Schedule Free #2
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode characters
| MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅ |
priyakasimbeg changed the base branch from main to rolling_leaderboard_submissions 
priyakasimbeg merged commit 76bf768 into mlcommons:rolling_leaderboard_submissions 1 of 2 checks passed
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
2 participants
Add this suggestion to a batch that can be applied as a single commit. This suggestion is invalid because no changes were made to the code. Suggestions cannot be applied while the pull request is closed. Suggestions cannot be applied while viewing a subset of changes. Only one suggestion per line can be applied in a batch. Add this suggestion to a batch that can be applied as a single commit. Applying suggestions on deleted lines is not supported. You must change the existing code in this line in order to create a valid suggestion. Outdated suggestions cannot be applied. This suggestion has been applied or marked resolved. Suggestions cannot be applied from pending reviews. Suggestions cannot be applied on multi-line comments. Suggestions cannot be applied while the pull request is queued to merge. Suggestion cannot be applied right now. Please check back later.
This pull request contains a version of Schedule-Free AdamW that makes use of the batch-norm fixes made after the competition deadline, and which follows the published version of Schedule-Free in a few small details for consistency.
I have also included a weighted dual-averaging version of Schedule-Free in the external tuning track, which shows improved convergence over the unweighted version on some workloads, particularly ViT.
This version is intended for the on-going public leaderboard that has been discussed in the workgroup meetings, and is designed for the original competition V100 hardware.
Note that the external-tuning track ResNet50 runs only just barely hit the baseline depending on the random seed, so there may be some non-determinism there.
@priyakasimbeg