Open
Conversation
Added a new Retired Division section to outline rules for submitting benchmarks that are no longer active, including submission guidelines and limitations.
Clarified that retired division submissions cannot be compared in any public communications to ensure fairness.
Updated the submission timeline for retired benchmarks from two years to three years.
|
MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅ |
ShriyaRishab
commented
Feb 26, 2026
| * Since reviewing retired benchmarks is an extra overhead, all mlperf logging checkers from the <round-of-retirement>* should be used to verify the retired benchmark results. If the <round-of-retirement> logging checkers fail for a retired division submission, the submission will be rejected. | ||
| * No new RCPs can be added for retired benchmarks, no changes to the reference code and no modifications will be allowed to the logging checkers. | ||
| * Retired division submissions CANNOT be compared to any other official MLPerf Training results from the current round or any previous rounds in any public communications (including but not limited to press statements, analyst discussions, website, blogs) to ensure fairness. | ||
| * The retired division submissions will not be displayed in the final results table to avoid confusion around result interpretability. It will get a submitter ID and it should be submitted under the division retired_<round-of-retirement>*. For example, if submitting Stable Diffusion v2 (v5.0 was the last round when the benchmark was in the suite) in round v6.0, the division for the submission should be "retired_v5.0". |
Contributor
Author
There was a problem hiding this comment.
Maybe we put the result with invalid results on the website? So folks can find it if needed in a table?
There was a problem hiding this comment.
Retired results shouldn't be invalid because they pass a submission checker and invalid is used for other means. I agree they should be visible on the website for easy reference, and if classified as a separate division this should be reasonable to keep separate from closed and open.
There was a problem hiding this comment.
If not visible on the website then they can only be found by running an offline submission check to print the result. Is this the correct assumption?
Contributor
Author
There was a problem hiding this comment.
Yes
the logs will be available so score can be generated
pbaumstarck
approved these changes
Feb 27, 2026
|
|
||
| The retired division is governed by all existing closed division rules, with the sole modification being the retirement of the benchmark. These additional rules also apply | ||
|
|
||
| * Since reviewing retired benchmarks is an extra overhead, all mlperf logging checkers from the <round-of-retirement>* should be used to verify the retired benchmark results. If the <round-of-retirement> logging checkers fail for a retired division submission, the submission will be rejected. |
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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.
Addresses mlcommons/training#847