jonathan-cohen-nvidia

Yes, sorry, you are correct. In the kernels above, the non-zero value is always 1 so it ends up being a copy. But you are correct these are highly specialized...

See: https://github.com/mlperf/policies/pull/32

@bitfort @petermattson If we are deferring to 0.8, then what did we decide to do for 0.7? I am unclear on the proposed rule or process.

Submitted draft PR: https://github.com/mlperf/training_policies/pull/317

What is the deadline for anyone who isn't Ali or Intel to request to make use of this exception? Seems like if we could agree on the rules for an...

See proposed language change: https://github.com/mlperf/policies/pull/22

We brought this up two weeks ago and Google specifically objected to evaluating more frequently. The 3M number was proposed by a Google engineer. I think any change at this...

I think there is a principle: once the submission deadline is passed, no one should be surprised by anyone's resubmissions. Resubmissions can happen for 2 reasons: 1. To address a...

Just want to add that today we identified a valid 3rd reason for resubmit: 3. In case the rules were slightly wrong/ambiguous, there are special cases where we will allow...

Suggest that we also ask (in a non-binding way) for which benchmarks each submitter expects to submit for. This matches the practice in Inference. Seems like it would be helpful...