-
Notifications
You must be signed in to change notification settings - Fork 54
System Configuration yaml and rules + bonus rule #163
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
Conversation
Added clear rule on consecutive runs having less time between runs than there is time in a run
|
MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅ |
|
@wvaske this is awesome - thanks! Do we want to indicate that Rack Units are mandatory only for certain types of systems? |
Submission_guidelines.md
Outdated
| * The metric for throughput is the minimum across all GPUs. | ||
|
|
||
| Each benchmark setup must be executed five times, and logs from all five runs must be submitted. The final metrics are the average across the five runs. | ||
| Each benchmark setup must be executed five times, and logs from all five runs must be submitted. The final metrics are the average across the five runs. Runs must be consecutive with not failed runs between the selected runs. Runs can not be cherry picked from a range of runs excepting that all five runs are consecutive within the large sequence of runs. |
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.
Should this read "Runs must be consecutive with no failed runs ..."?
# Conflicts: # Submission_guidelines.md
…e time at the top of the Benchmark Overview
|
I merged main into the branch. It looks good to me and I think I addressed the issues here. |
Added clear rule on consecutive runs having less time between runs than there is time in a run
Updated system-configuration.yaml example file
Updated rule that system-configuration.pdf needs topology