Get Involved

Join the forum

The MLPerf forum is open to everyone. To join, you will need to:

  1. Associate a Google account with your existing email address. To do so go here.
  2. Go here and click “apply for membership”.

Join working groups

MLPerf is run by a set of working groups. Before joining a working group, please first join the forum as described above. To join a working group please follow the group name hyperlink and click “apply for membership”.

Working Group Description Chair(s)
Training
Training Submitters Meeting-of-record for training rules. Submitters must attend. Victor Bittorf
Training Special Topics Explore training topics that require in depth discussion. Jonathan Cohen
Training Results Handle post-submission issues and presentation of Training results. Andy Hock
Inference
Inference Submitters Meeting-of-record for inference rules. Submitters must attend. Christine Cheng, David Kanter
Inference Special Topics Explore inference topics that require in depth discussion. Carole-Jean Wu, Vijay Janapa Reddi
Inference Results Handle post-submission issues and presentation of Inference results. Guenther Schmuelling
Power Determine how to measure power for MLPerf submissions. Bing Yu, Hanlin Tang, Jonah Alben
Benchmark Development
Time Series Benchmark Create a time series benchmark. Xinyuan Huang, Sergey Serebryakov, Ankur Ankur
Speech Benchmark Create a speech benchmark. Sam Davis, Peter Baldwin
HPC Explore ML benchmarking supercomputers. Steve Farrel, Jacob Balma, Abid Muslim
Silicon IP Explore ML benchmarking hardware in simulation. Colin Osborne, Debajyoti Pal
Pre-Silicon Explore ML benchmarking silicon IP. Jeffery Liao
ML Commons Scope Expansion
Governance Create host legal entity. Peter Mattson
Datasets Create large public datasets. Greg Diamos, Peter Mattson, Vijay Janapa Reddi
Best Practices Develop basic best practices for ML. Sarah Bird, Debo Dutta, Victor Bittorf
Research Promote ML in the research community. Cody Coleman, Gennady Pekhimenko

Attend community meetings

MLPerf organizes a community meeting every other month. The meetings are open to anyone who is a member of a working group. The dates and locations are shown below. More specific information is available on the discussion group.

Community Meeting Date Location Slides
Thursday, February 14, 2019 Xilinx, San Jose slides
Thursday, April 4, 2019 Synopsys, San Jose slides
Thursday, June 6, 2019 AMD, Santa Clara slides
Thursday, August 8, 2019 Cadence, San Jose slides
Thursday, October 10, 2019 Samsung, San Jose
Thursday, December 12, 2019 Facebook, Menlo Park

Become a supporting org

For the remainder of 2019 it is free to become a supporting organization and have your logo shown on the MLPerf home page. Please have a member of your marketing or legal teams send an email to info@mlperf.org stating the following exact permission "the MLPerf effort can use our company name and logo on the MLPerf.org website, documents, and press as a supporting organization” along with your preferred logo for a white background (jpeg or png).

Starting in 2020, MLPerf will be hosted by a non-profit called MLCommons. Membership in MLCommons will be required for display of logos, participation in most MLPerf activities, and submission of benchmark results. If your organization would like to become a founding member of MLCommons, please contact info@mlperf.org.

Questions or issues

If you have administrative questions, you can contact info@mlperf.org.

If you have technical questions or issues with the reference code or rules, please file issues against the appropriate Github repo.