Update Status of this document for CR Snapshot #532
Conversation
|
This PR pre-empts the group addresses #531 as discussed in https://www.w3.org/2024/01/25-webmachinelearning-minutes.html#t05 |
|
@wchao1115 @huningxin also PTAL This is crafted based on our discussion in https://www.w3.org/2024/01/25-webmachinelearning-minutes.html#t06 with consideration for "what's new" since initial CR March 2023 and what people outside this group should know given an opportunity to do an elevator pitch to them. This means we need to abstract the details. |
index.bs
Outdated
| integration. A proposal to | ||
| <a href="https://github.com/webmachinelearning/webnn/pull/322">simplify | ||
| MLContext creation</a> is being discussed. This document is maintained and | ||
| Since the <a href="https://www.w3.org/TR/2023/CR-webnn-20230330/">initial Candidate Recommendation Snapshot</a> the Working Group has gathered further <a href="https://webmachinelearning.github.io/webnn-status/">implementation experience</a> and added new operations and data types needed for well-known <a href="https://github.com/webmachinelearning/webnn/issues/375">transformers to support generative AI use cases</a>. In addition, the group has removed select 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 <a href="https://github.com/webmachinelearning/webnn/issues/482">backend-agnostic storage type</a>, to improve performance and interoperability between the WebNN, WebGPU APIs and purpose-built hardware for ML. |
There was a problem hiding this comment.
Looks good to me. Should we expand a bit more on what 'further implementation experience' is? i.e. should we be more specific about the ongoing work in the Chromium code base across Google, Microsoft, and Intel, etc.?
There was a problem hiding this comment.
Thanks @wchao1115 for your review. It may be hard to succinctly expand 'further implementation experience' in the limited space we have in this status section.
For this reason, we direct interested readers to https://webmachinelearning.github.io/webnn-status/ that also provides further pointers to Web Platform Tests. WPT is a resource W3C stakeholders are interested in in particular when they evaluate implementation experience. I think we are good in this respect with the materials we have.
|
@dontcallmedom PTAL, PR updated to note |
index.bs
Outdated
| integration. A proposal to | ||
| <a href="https://github.com/webmachinelearning/webnn/pull/322">simplify | ||
| MLContext creation</a> is being discussed. This document is maintained and | ||
| Since the <a href="https://www.w3.org/TR/2023/CR-webnn-20230330/">initial Candidate Recommendation Snapshot</a> the Working Group has gathered further <a href="https://webmachinelearning.github.io/webnn-status/">implementation experience</a> and added new operations and data types needed for well-known <a href="https://github.com/webmachinelearning/webnn/issues/375">transformers to support generative AI use cases</a>. In addition, informed by this implementation experience, the group removed <code>MLCommandEncoder</code></a>, support for synchronous execution, and higher-level operations that can be expressed in terms of lower-level primitives in a performant manner. 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 <a href="https://github.com/webmachinelearning/webnn/issues/482">backend-agnostic storage type</a>, to improve performance and interoperability between the WebNN, WebGPU APIs and purpose-built hardware for ML. |
There was a problem hiding this comment.
Thinking some more about the backend-agnostic storage new feature - I think we should indicate whether we expect to republish a CR Snapshot if/when that feature lands (vs keeping it for a potential post-Recommendation iteration); I'm happy to propose some suggested language if you confirm the former is our plan.
There was a problem hiding this comment.
I believe we don’t want to rush to post-CR stages, especially considering more implementers just joined the WG. I’d expect us to do another CRS when this new feature has matured adequately.
Contributions welcome.
anssiko
force-pushed
the
cr2-sotd
branch
2 times, most recently
from
680c17a to
15168d3
Compare
This status text is a high-level summary for busy people. - Implementation experience update - Significant new & removed features, use cases, conventions update - Significant new features in development, MLBuffer Co-authored-by: Dominique Hazael-Massieux <[email protected]>
|
@dontcallmedom I'll merge this SOTD update now even if we're not going to publish immediately. I feel it is a good practice to have this section reflect reality, and especially new participants might glimpse here for the latest high-level status. We should pick a habit to maintain this on an ongoing basis, not just for major milestones. |
SHA: d48e394 Reason: push, by anssiko Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
This status text is a high-level summary for busy people.
Preview | Diff