Happy New Year and a questionnaire to select a better time slot for our weekly conf-call
0 views
Skip to first unread message
Grigori Fursin
Jan 10, 2023, 11:00:05 AM1/10/23
to Collective Knowledge
Dear colleagues,
We wish you Happy New Year with many exciting achievements!
We are searching for a better time slot for our weekly conf-calls to discuss modular and collaborative benchmarking, design space exploration and deployment of Pareto-efficient ML Systems. Please fill in the questionnaire with your preferred time slot at https://forms.gle/Nuy3YiPyWGRikvQy5 .
Our MLCommons taskforce on education and reproducibility has just finished modularizing the reference implementation of the MLPerf inference benchmark to plug in different models, data sets, frameworks, compilers and run-times using the new CK2 (CM) framework. Our ultimate goal is to help you reduce development, benchmarking, optimization and deployment time for your ML and AI systems!
Please get in touch if you want to work with us to automate your MLPerf inference benchmark submissions across different models, frameworks and platforms for the next v3.0 round!
You can reach the community via our public Discord channel at https://discord.gg/7SmbHwuC and/or get in touch with us directly if you have any questions and suggestions!
You can also talk to us in person at the HiPEAC'23 in Toulouse next week.
Finally, we will be helping with artifact evaluation at ACM/IEEE MICRO'23 and Student Cluster Competition'23 - please stay tuned if you are interested to help with these important community efforts to reproduce experimental results from research papers and make them portable across rapidly evolving software, hardware and data!
Looking forward to collaborating with you in 2023!
========================================================================
Grigori Fursin, PhD
* Vice President at OctoML
* Author of the Collective Knowledge framework
* Co-founder of the taskforce on education and reproducibility at MLCommons
We wish you Happy New Year with many exciting achievements!
We are searching for a better time slot for our weekly conf-calls to discuss modular and collaborative benchmarking, design space exploration and deployment of Pareto-efficient ML Systems. Please fill in the questionnaire with your preferred time slot at https://forms.gle/Nuy3YiPyWGRikvQy5 .
Our MLCommons taskforce on education and reproducibility has just finished modularizing the reference implementation of the MLPerf inference benchmark to plug in different models, data sets, frameworks, compilers and run-times using the new CK2 (CM) framework. Our ultimate goal is to help you reduce development, benchmarking, optimization and deployment time for your ML and AI systems!
Please get in touch if you want to work with us to automate your MLPerf inference benchmark submissions across different models, frameworks and platforms for the next v3.0 round!
You can reach the community via our public Discord channel at https://discord.gg/7SmbHwuC and/or get in touch with us directly if you have any questions and suggestions!
You can also talk to us in person at the HiPEAC'23 in Toulouse next week.
Finally, we will be helping with artifact evaluation at ACM/IEEE MICRO'23 and Student Cluster Competition'23 - please stay tuned if you are interested to help with these important community efforts to reproduce experimental results from research papers and make them portable across rapidly evolving software, hardware and data!
Looking forward to collaborating with you in 2023!
========================================================================
Grigori Fursin, PhD
* Vice President at OctoML
* Author of the Collective Knowledge framework
* Co-founder of the taskforce on education and reproducibility at MLCommons
Reply all
Reply to author
Forward
0 new messages