johntran-nv
johntran-nv
Shuffling rules about DLRM were not clear enough in the v0.7 round and they left a lot of room for interpretation. This update makes a clear rule that is easy...
We agreed to this doc, but never converted it to official policy. https://docs.google.com/document/d/15SJJqVBuLLC6k8Tks1TiyPWFwEGCzFmBhgJPU8ph_7s/edit?usp=sharing
In v1.0, we discovered after code freeze that the reference does clipnorm-before-reduce. Mathematically, that is not as good as clipnom-after-reduce, but since it was after code freeze, we decided to...
Copying updates from https://github.com/mlcommons/training/pull/516. @ascendhuawei please double check
Proposing formal text for how to invalidate preview submissions. This has been discussed in the Inference and Training WG's at some length already, and folks recommended putting something in writing...
It looks like we were not clear on this, as I don't see it in the rules doc, but I believe the goal in the past was that software must...