dontcallmedom commented 8 months ago

Document URLs

https://webmachinelearning.github.io/webnn/

Link to group's decision to request transition

https://lists.w3.org/Archives/Public/public-webmachinelearning-wg/2024Mar/0000.html

Link to previous Candidate Recommendation transition or update request

https://github.com/w3c/transitions/issues/491

Substantive changes

Since the initial Candidate Recommendation Snapshot, the Working Group has gathered further implementation experience and added new operations and data types needed for well-known transformers https://github.com/webmachinelearning/webnn/issues/375. In addition, the group has removed selected features informed by this implementation experience: higher-level operations that can be expressed in terms of lower-level primitives in a performant manner, and support for synchronous execution. The group has also updated the specification to use modern authoring conventions to improve interoperability and precision of normative definitions and is developing a new feature, a https://github.com/webmachinelearning/webnn/issues/482, to improve performance and interoperability between the WebNN, WebGPU APIs and purpose-built hardware for ML.

Any changes in requirements?

None

Wide Review of substantive changes

https://github.com/webmachinelearning/webnn/issues/239 (updated TAG review requested a month ago, but still untriaged)

Issues status

There are 49 substantive issues, mostly related to integration with WebGPU, and improvements to the definitions of existing operations.

Formal Objections

None

Any changes in implementation information?

Active implementation work in Chromium, shipping behind a flag in Chrome and Edge https://webmachinelearning.github.io/webnn-status/

Deadline for further comments

publication date + 6 months

Any changes in patent disclosures?

None

See Requirements for updating a Candidate Recommendation Snapshot for further information.

cc @anssiko