Peter Mattson
Peter Mattson
+Guenther Schmuelling +Cheng, Christine Thanks for raising this. I'd assumed we'd at least hit the 99% target. If not, we should correct. Guenther, would it be possible to extend your...
We should probably revise the results page with accuracy numbers if some of the results are for below 99. Would you mind extracting the column? On Sat, Nov 14, 2020...
Thanks Guenter. Will add soon. On Wed, Nov 18, 2020 at 4:09 PM Guenther Schmuelling < [email protected]> wrote: > I generated a new result sheet here and added a accuracy...
e.g. 2 of (GPUs, CPUs, TPUs, anything else that is FP32 and can run TF or pyT)
Closer! :-) IMO, the *only* change needed is to add this paragraph: (Un)padding or (un)packing are both allowed as offline or online preprocessing steps, including removal or addition of zero...
SWG: This is complicated. We should write clear rules, but it will take work. For now, review committee will decide if two systems are sufficiently identical for MLPerf purposes.
Backlogged :-) by proposer's request.
SWG: Approve week 3 Monday declaration of borrowing w/ specific hps. Make withdraw deadline more obvious.
Check status as part of v0.8 planning
Already passed for v0.7, but need to make a standard rule for v0.8