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

[DLRMv2] Include dlrmv2 training rules #515

Closed janekl closed 1 year ago

janekl commented 1 year ago

Author: Jan Lasek, Nvidia (jlasek_at_nvidia_dot_com)

I'm updating training rules to include the new DLRMv2 recommender benchmark.

Summary:

  1. Introducing the model
  2. Listing optimizer parameters with links to the reference code
  3. Specifying evaluation frequency
  4. Remark on training dataset shuffling
  5. Remark on incomplete training batch
github-actions[bot] commented 1 year ago

MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅