WebML WG Teleconference – 24 September 2026 - 15:00-16:00 UTC

Latest version: https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md


WebML WG Teleconference – 24 September 2026 - 15:00-16:00 UTC
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md#webml-wg-teleconference--24-september-2026---1500-1600-utc>
See the timezone table ...
Logistics
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md#logistics>

  *   Chair: Anssi
  *   Scribe: Anssi
  *   IRC: irc://irc.w3.org:6667/#webmachinelearning

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

  *   Zoom joining instructions: internal-webmachinelearning<https://lists.w3.org/Archives/Member/internal-webmachinelearning/2023Jun/0000.html>
  *   Minutes: https://www.w3.org/2026/09/24-webmachinelearning-minutes.html


Web Neural Network API
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md#web-neural-network-api>

Tip

Specification<https://www.w3.org/TR/webnn/> | Explainer<https://github.com/webmachinelearning/webnn/blob/main/explainer.md> | Implementation Report<https://wpt.fyi/results/webnn> | Implementation Status<https://webmachinelearning.github.io/webnn-status/> | Awesome WebNN<https://github.com/webmachinelearning/awesome-webnn>

WebNN-WebGPU interop v2 features
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md#webnn-webgpu-interop-v2-features>

Discuss v2 feature proposals layered atop the v1 baseline.

  *   ⨀ zero-copy hints webmachinelearning/webnn#941<https://github.com/webmachinelearning/webnn/issues/941>
  *   ⨀ MLTensor export custom usages & flags webmachinelearning/webnn#954<https://github.com/webmachinelearning/webnn/issues/954>

Performance reporting API for client-cloud offload
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md#performance-reporting-api-for-client-cloud-offload>

Discuss new implementation experience and proposal for performance reporting API to inform client-cloud offload decisions.

  *   ⨀ webmachinelearning/webnn#934 (comment)<https://github.com/webmachinelearning/webnn/issues/934#issuecomment-5673970569>
  *   ≡ related work: Dynamic AI Offloading Protocol https://github.com/webmachinelearning/daop


Subgraphs and contemporary LLM requirements
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md#subgraphs-and-contemporary-llm-requirements>

Review GQA decomposition. Discuss contemporary LLM requirements.

  *   ⨀ webmachinelearning/webnn#907<https://github.com/webmachinelearning/webnn/issues/907>

[security] Formulaic mitigation for integer overflows
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md#security-formulaic-mitigation-for-integer-overflows>

Discuss and refine the proposal to specify relational limits as algorithmic steps, as opposed to hard-coded limits.

  *   ⨀ webmachinelearning/webnn#928<https://github.com/webmachinelearning/webnn/issues/928>
  *   ☑ Resolution<https://www.w3.org/2026/05/07-webmachinelearning-minutes.html#4d72>

Low-precision floating point data types
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md#low-precision-floating-point-data-types>

Review PRs for bfloat16 and float8e4m3 data type support.

  *   ⨀ webmachinelearning/webnn#930<https://github.com/webmachinelearning/webnn/issues/930>
  *   ⛙ explainer PR webmachinelearning/webnn#938<https://github.com/webmachinelearning/webnn/pull/938>
  *   ⛙ spec PR webmachinelearning/webnn#956<https://github.com/webmachinelearning/webnn/pull/956>

[security] Full-sequence output validation
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md#security-full-sequence-output-validation>

Review PR. Discuss only as appropriate.

  *   ⨀ webmachinelearning/webnn#949<https://github.com/webmachinelearning/webnn/issues/949>
  *   ⛙ webmachinelearning/webnn#955<https://github.com/webmachinelearning/webnn/pull/955>

Accuracy for reductions
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md#accuracy-for-reductions>

Review PR. Discuss only as appropriate.

  *   ⨀ webmachinelearning/webnn#947<https://github.com/webmachinelearning/webnn/issues/947>
  *   ⛙ webmachinelearning/webnn#957<https://github.com/webmachinelearning/webnn/pull/957>

Dynamic shapes
<https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-24-wg-agenda.md#dynamic-shapes>

Review PR. Discuss only as appropriate.

  *   ⨀ webmachinelearning/webnn#883<https://github.com/webmachinelearning/webnn/issues/883>
  *   ⛙ webmachinelearning/webnn#945<https://github.com/webmachinelearning/webnn/pull/945>

Received on Friday, 18 September 2026 10:02:04 UTC