WebML WG Teleconference – 27 June 2024 - 14:00-15:00 UTC

Latest version: https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md


WebML WG Teleconference – 27 June 2024 - 14:00-15:00 UTC
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#webml-wg-teleconference--27-june-2024---1400-1500-utc>
See the timezone table ...
Logistics
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#logistics>

  *   Chair: Anssi
  *   Scribe: ? (see howto<https://github.com/webmachinelearning/meetings/blob/main/scribe-howto.md>, volunteers welcome!)
  *   IRC: irc://irc.w3.org:6667/#webmachinelearning

  *   IRC web client: https://irc.w3.org/?channels=#webmachinelearning

  *   Joining instructions: https://lists.w3.org/Archives/Member/internal-webmachinelearning/2023Jun/0000.html

  *   Minutes: https://www.w3.org/2024/06/27-webmachinelearning-minutes.html


Agenda
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#agenda>
ℹ️ Announcements
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#%E2%84%B9%EF%B8%8F-announcements>
🆕 WebML WG F2F at TPAC 2024
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#-webml-wg-f2f-at-tpac-2024>

W3C TPAC 2024 takes place in Anaheim, CA, USA at the Hilton Anaheim on 23–27 September 2024. The WebML WG will meet F2F on Monday, 23 September 2024, 09:00–18:00 PDT.

  *   📢 https://lists.w3.org/Archives/Public/public-webmachinelearning-wg/2024Jun/0003.html

  *   ⨀ Questions, comments: #23<https://github.com/webmachinelearning/meetings/issues/23>
  *   ☰ https://www.w3.org/2024/09/TPAC/


🆕 Summer break
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#-summer-break>

We'll have a break in July and will resume WG teleconferences 8 August 2024.

ℹ️ Transformers.js update
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#%E2%84%B9%EF%B8%8F-transformersjs-update>

A brief update on Transformers.js to inform the WG. Details and time TBC. (Joshua Lochner)

  *   https://huggingface.co/docs/transformers.js


ℹ️ Device selection
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#%E2%84%B9%EF%B8%8F-device-selection>

📌 We agreed to evolve MLContextOptions and other API controls for device selection informed by further implementation experience and new use cases from the wider web community. Kick off a new workstream to explore device selection mechanisms and assess early design proposals. Additional proposals are welcome. Discuss early proposals for a fallback device, multiple devices in a preferred order, an exclusion of a specific device. Assess error handling, ultimate fallback, and quantized operator aspects in this context.

Specification references:

  *   Device selection https://www.w3.org/TR/webnn/#programming-model-device-selection

  *   MLContextOptions https://www.w3.org/TR/webnn/#dictdef-mlcontextoptions


GitHub for comments:

  *   ⨀ webmachinelearning/webnn#623<https://github.com/webmachinelearning/webnn/issues/623>
  *   ⛙ webmachinelearning/webnn#696<https://github.com/webmachinelearning/webnn/pull/696> (merged, migrate applicable comments to issues)

Implementation experience:

  *   ⛭ Chromium implementation (NPU device type) https://chromium-review.googlesource.com/c/chromium/src/+/5330647


ℹ️ MLBuffer
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#%E2%84%B9%EF%B8%8F-mlbuffer>

📌 A placeholder for discussion on active MLBuffer topics as appropriate.

The group agreed to work out the details of the direct buffer sharing proposal and bring the proposal to the WebGPU group.

Discuss any design progress in the MLGraphBuilder and MLBuffer construction alignment.

  *   ⨀ webmachinelearning/webnn#688<https://github.com/webmachinelearning/webnn/issues/688>
  *   ⨀ webmachinelearning/webnn#697<https://github.com/webmachinelearning/webnn/issues/697>
  *   🏷️ All webgpu interop<https://github.com/webmachinelearning/webnn/labels/webgpu%20interop> issues

ℹ️ TAG feedback: resource contention considerations for GPU/CPU/accelerators (last call revised)
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#%E2%84%B9%EF%B8%8F-tag-feedback-resource-contention-considerations-for-gpucpuaccelerators-last-call-revised>

Last call for input to the revised TAG feedback response for resource contention considerations for GPU/CPU/accelerators. Proposal to add non-normative guidance on vectors and mitigations inspired by WebGPU DoS considerations.

  *   ⨀ w3ctag/design-reviews#933 (comment)<https://github.com/w3ctag/design-reviews/issues/933#issuecomment-2126805061>
  *   ⨀ Related: webmachinelearning/webnn#529<https://github.com/webmachinelearning/webnn/issues/529>
  *   ☰ Related: WebGPU DoS https://www.w3.org/TR/webgpu/#security-dos


ℹ️ Open issues and PRs
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#%E2%84%B9%EF%B8%8F-open-issues-and-prs>

We'll discuss open issues and review PRs based on your feedback:

  *   ⨀ All open issues<https://github.com/webmachinelearning/webnn/issues>
  *   ⛙ All open pull requests<https://github.com/webmachinelearning/webnn/pulls>

🏷️ Debrief on PRs merged recently
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#%EF%B8%8F-debrief-on-prs-merged-recently>

The editors and PR authors will debrief the group on substantive PRs that got merged in the last few weeks, answer question from the group.

  *   ⛙ Recently merged PRs<https://github.com/webmachinelearning/webnn/pulls?q=is%3Apr+is%3Amerged>

🏷️ Your proposals
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2024-06-27-wg-agenda.md#%EF%B8%8F-your-proposals>

🧪 This is a new experiment.

Please suggest issue(s) and PR(s) you'd like to see discussed on this teleconference in this issue:

  *   🆕 #24<https://github.com/webmachinelearning/meetings/issues/24>