WebML WG Teleconference – 7 March 2024 - meeting minutes

Hi All,

Thanks everyone for joining us today. If you were not able to make it, you can catch up via meeting minutes:

https://www.w3.org/2024/03/07-webmachinelearning-minutes.html


Summary of our discussions:

We acknowledged the Call for Consensus (CfC) to publish the Web Neural Network API as a new Candidate Recommendation Snapshot (CRS) passes. As the next step, we'll get the publication machinery moving. The group can continue work on the specification and merge PRs in parallel to this activity.

We discussed and received a presentation on Hybrid AI approaches that could flexibly shift ML workloads between server and client to support elasticity and avoid the problem of developers targeting only the weakest clients’ capabilities. Any feedback to inform the direction of the exploration is welcome via the proposals repo:

https://github.com/webmachinelearning/proposals/issues/5


Lastly, we discussed a few select issues: core operator set, new feature for improved error handling, and conventions updates. Please check out the newly triaged open issues and provide your comments:

https://github.com/webmachinelearning/webnn/issues


Thanks,

-Anssi (WebML WG chair)