Matthew Frank

Yes, unfortunately this was left unfixed in the churn before v1.0 submission 1.5 years ago, so it is what it is, and it would not be productive to change the...

This PR seems to refer to 2 different benchmarks: the object_detection benchmark, and the now-retired ssd-v1 benchmark. I'm not sure whether we want to incorporate the changes to the object_detection...

I believe PR https://github.com/mlcommons/training/pull/590 (earlier https://github.com/mlcommons/training/pull/516) also addresses this issue.

Note that reinforcement/README.md has since been moved to retired_benchmarks/minigo/README.md, so this PR should be updated to reflect the new path.

PR https://github.com/mlcommons/training/pull/603 is my proposed fix for the small amount of this problem that remains.

We're in the process of cleaning up the repo. I fully agree with the sentiment that this repo should have tags corresponding to the repo states at each submission deadline....

I just closed https://github.com/mlcommons/training/issues/406 as a duplicate of this. Hopefully when we do the tagging above it will (sort of) address issue #406 . (Although as I said above: won't...

My memory is that the committee rejected this pytorch implementation of BERT as an alternate reference implementation. But I believe this is a useful artifact to save for posterity, so...