weekly notes and new agenda for the MLCommons "education and reproducibility" conf-call
3 views
Skip to first unread message
Grigori Fursin
Oct 11, 2022, 1:13:04 PM10/11/22
to Arjun Suresh (OctoML)
Dear All,
We would like to thank everyone for their active participation and
feedback about our new MLCommons education and reproducibility
initiative - very exciting!
Please find the notes from our last conf-call and tentative agenda for
the next conf-call at
https://docs.google.com/document/d/1zMNK1m_LhWm6jimZK6YE05hu4VH9usdbKJ3nBy-ZPAw
. Feel free to comment!
Looking forward to further discussions,
Grigori Fursin and Arjun Suresh
-------------------
Grigori Fursin
* Vice President at OctoML.ai (MLOps)
* Author of the MLCommons benchmarking meta-framework (CK)
* A founding member of the ACM taskforce on reproducibility
We would like to thank everyone for their active participation and
feedback about our new MLCommons education and reproducibility
initiative - very exciting!
Please find the notes from our last conf-call and tentative agenda for
the next conf-call at
https://docs.google.com/document/d/1zMNK1m_LhWm6jimZK6YE05hu4VH9usdbKJ3nBy-ZPAw
. Feel free to comment!
Looking forward to further discussions,
Grigori Fursin and Arjun Suresh
-------------------
Grigori Fursin
* Vice President at OctoML.ai (MLOps)
* Author of the MLCommons benchmarking meta-framework (CK)
* A founding member of the ACM taskforce on reproducibility
Grigori Fursin
Oct 20, 2022, 9:00:39 AM10/20/22
to Arjun Suresh (OctoML)
Hi all,
Based on your requests, we are planning to make a demo of the CM automation workflows with modular Docker containers to make it easier to run and reproduce MLPerf inference benchmarks during our weekly conf-call today.
Please find the agenda and the notes for our conf-call here.
Looking forward to your comments and further suggestions,
Based on your requests, we are planning to make a demo of the CM automation workflows with modular Docker containers to make it easier to run and reproduce MLPerf inference benchmarks during our weekly conf-call today.
Please find the agenda and the notes for our conf-call here.
Looking forward to your comments and further suggestions,
Grigori Fursin
Nov 30, 2022, 10:18:57 AM11/30/22
to Grigori Fursin
Hi All,
Please find our agenda for the conf-call tomorrow (Dec 1, 2022): https://docs.google.com/document/d/1zMNK1m_LhWm6jimZK6YE05hu4VH9usdbKJ3nBy-ZPAw
After successfully validating our modular CK2 workflow to reproduce several reference implementations of the MLPerf inference benchmark at SuperComputing'22, we are ready to start modularizing and documenting other "non-reference" (optimized) implementations (Nvidia, Qualcomm, Neural Magic, Intel), adding power measurements, automating design space exploration experiments and visualizing results - we are very interested in your suggestions and feedback to prioritize our developments.
We also plan to use this public mailing list for further announcements - you will receive an invite shortly.
Thank you and looking forward to further discussions,
Grigori
-------------------
Grigori Fursin, PhD
* Vice President of MLOps at OctoML
* Author of the MLCommons CK automation meta-framework
Please find our agenda for the conf-call tomorrow (Dec 1, 2022): https://docs.google.com/document/d/1zMNK1m_LhWm6jimZK6YE05hu4VH9usdbKJ3nBy-ZPAw
After successfully validating our modular CK2 workflow to reproduce several reference implementations of the MLPerf inference benchmark at SuperComputing'22, we are ready to start modularizing and documenting other "non-reference" (optimized) implementations (Nvidia, Qualcomm, Neural Magic, Intel), adding power measurements, automating design space exploration experiments and visualizing results - we are very interested in your suggestions and feedback to prioritize our developments.
We also plan to use this public mailing list for further announcements - you will receive an invite shortly.
Thank you and looking forward to further discussions,
Grigori
-------------------
Grigori Fursin, PhD
* Vice President of MLOps at OctoML
* Author of the MLCommons CK automation meta-framework
Reply all
Reply to author
Forward
0 new messages