search

mlcommons / training_policies

Issues related to MLPerf™ training policies, including rules and suggested changes
https://mlcommons.org/en/groups/training
Apache License 2.0
92 stars 66 forks source link

Define "acceptable" convergence (w/ error bars) (re: backbones) #392

Open bitfort opened 4 years ago

johntran-nv commented 3 years ago

Here's the saw SSD data: https://docs.google.com/spreadsheets/d/1OEKCmkZcRHCa7Bwdj51GJ1HaLXc426E1Dc42Rj2UuFE/edit#gid=882662286.

johntran-nv commented 3 years ago

Putting a link to the bounded convergence doc here: https://docs.google.com/document/d/15DBV5mM8KHYMjGRsJiztQaz-uxKaekOr2pnwmQl_RT0/edit?usp=sharing. We still need to make this doc "official" somehow.

johntran-nv commented 3 years ago

From this week's meeting, next AI is to @petermattson to propose a formal rules update, working with @bitfort .