- March 2022: Grigori Fursin began developing CM (Collective Mind), also referred to as CK2, as a successor to CK at OctoML.
- April 2022: Arjun Suresh joined OctoML and collaborated with Grigori on developing CM Automation tools.
- May 2022: The CM CLI and Python interface were successfully implemented and stabilized by Grigori.
- Arjun completed the development of the MLPerf Inference Script within CM.
- OctoML achieved first MLPerf Inference submission (v2.1) using CM Automation (progress here).
- End-to-end MLPerf inference automations using CM was successfully completed in CM.
- Additional benchmarks and Power Measurement support were integrated into CM.
- cTuning achieved a successful MLPerf Inference v3.0 submission using CM Automation.
- Arjun and Grigori departed OctoML and resumed CM development under funding from cKnowledge.org and cTuning.
- MLPerf inference automations were extended to support NVIDIA implementations.
- cTuning achieved the largest-ever MLPerf Inference submission (v3.1) using CM Automation.
- MLCommons began funding CM development to enhance support for NVIDIA MLPerf inference and introduce support for Intel and Qualcomm MLPerf inference implementations.
- MLPerf inference automations were expanded to support NVIDIA, Intel, and Qualcomm implementations.
- cTuning completed the MLPerf Inference v4.0 submission using CM Automation.
- MLCommons contracted Arjun Suresh via GATEOverflow to improve MLPerf inference documentation and enhance CM Automation on various platforms.
- Anandhu Sooraj joined MLCommons to collaborate with Arjun Suresh on CM development.
- Dedicated documentation site launched for MLPerf inference.
- CM scripts were developed for MLPerf Automotive.
- CM Docker support was stabilized.
- GitHub Actions workflows were added for MLPerf inference reference implementations and NVIDIA integrations (see updates).