Accelerating ML innovation at MLCommons

Origins

ML Commons emerged from the collaborative efforts of industry leaders and academics, initially rooted in the MLPerf benchmarking initiative. David Kanter highlights the diverse founding members, including representatives from major tech companies like Google and Nvidia, as well as academic researchers 1. This collective aimed to establish a trusted community with fair and useful benchmarks for machine learning. Kanter describes ML Commons as a global engineering consortium, emphasizing its open-source principles and the importance of consensus-driven collaboration 2.

ML Commons is that container, but the other thing is, we knew that this was one leg of the tripod and that we had other projects that we wanted to get done.

--- David Kanter

The organization operates with a startup-like DNA, focusing on collective engineering efforts distinct from marketing or policy-driven groups 3.

   

Innovation

ML Commons drives innovation through three pillars: performance benchmarks, datasets, and best practices. David Kanter explains that these elements are crucial for advancing machine learning, akin to the precision required in the industrial revolution 4. The organization aims to make AI tools more accessible and reproducible, reducing the time needed to implement models across various platforms 5.

Our goal is advancing innovation in machine learning to drive the whole industry forward.

--- David Kanter

By standardizing components and practices, ML Commons seeks to extend AI capabilities beyond digitally native companies to traditional industries like Goodyear and Macy's 6.

   

Future Vision

David Kanter envisions a future where AI and machine learning significantly impact various sectors, including medicine and language translation. He emphasizes the potential of AI to empower innovations that benefit everyone, such as expanding speech-to-text capabilities across multiple languages 7. The organization aims to fill gaps in the industry, supporting pioneering innovations like BERT while enhancing efficiency through comprehensive datasets.

The name of the game is how do we build a better world and a bigger role for AI?

--- David Kanter

Kanter foresees a future where AI becomes as ubiquitous as Excel, with best practices evolving to remove frictions and enhance interoperability 6.