All pull requests

Pull requests

• NXP backend: use channels-first for reference testing model

  CLA Signedmodule: nxprelease notes: nxp
  #23308·novak-vaclav opened Oct 1, 2026Collaborator·

  ·
  3

• Arm backend: Dynamic shape lowering for NSS & NFRU

  ciflow/trunkCLA Signedmodule: armpartner: armrelease notes: arm
  #23307·Michiel-Olieslagers opened Oct 1, 2026Collaborator·

  ·
  1

• Arm backend: Remove obsolete import ignores

  ciflow/trunkCLA Signedmodule: armpartner: armrelease notes: none
  #23306·perheld opened Oct 1, 2026Collaborator·

  ·
  1

• NXP backend: Enable model suite CI testing.

  CLA Signedmodule: nxprelease notes: nxp
  #23305·MartinPavella opened Oct 1, 2026Collaborator·

  ·
  2

  

• Fix CI breaks from the off-graph KV stack (ROCm build, muse-glimmer offgraph e2e)

  ciflow/rocmCLA Signedmodule: rocm
  #23304·Gasoonjia opened Oct 1, 2026Contributor·

  ·
  2

• Zephyr: document classifying your own image, and correct stale references

  CLA Signed
  #23300·psiddh opened Oct 1, 2026Contributor·

  ·
  11

• Share one FakeQuantize across cat under QAT (#23299)

  CLA Signedmeta-exportedmodule: qnnrelease notes: none
  #23299·billmguo opened Oct 1, 2026Contributor·

  ·
  3

• Make ConcatObserver deepcopy-safe (#23298)

  CLA Signedmeta-exportedmodule: qnnrelease notes: none
  #23298·billmguo opened Oct 1, 2026Contributor·

  ·
  3

• [LLM serving 6/x] Add a bounded multiplexed native serving worker

  CLA Signed
  #23291·metascroy opened Sep 30, 2026Contributor·

  ·
  1

• Fix FlatTensor serializer linkage and header installation

  #23290·quic-xiyushi opened Sep 30, 2026Contributor·

  ·
  3

• Automate stable release dependency finalization

  CLA Signed
  #23289·JacobSzwejbka opened Sep 30, 2026Contributor·

  ·
  1

• Arm backend: Upgrade model converter to 0.11.

  ciflow/trunkCLA Signedhelp wantedmodule: armpartner: armrelease notes: arm
  #23288·wwwind opened Sep 30, 2026Collaborator·

  ·
  3

• Automate release branch preparation and finalization

  CLA Signed
  #23287·JacobSzwejbka opened Sep 30, 2026Contributor·

  ·
  2

• [LLM] Reuse prompt prefixes only for new serving sessions

  CLA Signed
  #23286·metascroy opened Sep 30, 2026Contributor·

  ·
  1

• [LLM serving 4/x] Add batching-backed public text generation

  CLA Signed
  #23285·metascroy opened Sep 30, 2026Contributor·

  ·
  1

• [LLM serving 3/x] Bridge serving requests to the batching runner

  CLA Signed
  #23284·metascroy opened Sep 30, 2026Contributor·

  ·
  1

• Bump fast-uri from 3.0.3 to 3.1.8 in /examples/demo-apps/react-native/rnllama

  CLA Signeddependenciesjavascript
  #23282·dependabot[bot] opened Sep 30, 2026Bot·

  ·
  2

• Prevent unsafe view-copy replacement around mutations

  CLA Signed
  #23279·JacobSzwejbka opened Sep 30, 2026Contributor·

  ·
  3

• [CoreAI] Record ordered I/O names in the delegate manifest

  CLA Signed
  #23278·metascroy opened Sep 30, 2026Contributor·

  ·
  2

• [LLM serving 2/x] Add batching-backed serving session lifecycle

  CLA Signed
  #23277·metascroy opened Sep 30, 2026Contributor·

  ·
  2

• Fix Inspector compare_results crash on non-tensor outputs

  CLA Signedcommunity: contributionrelease notes: devtools
  #23276·u7k4rs6 opened Sep 30, 2026·

  ·
  4

• Add channels-last support to portable mean

  CLA Signedrelease notes: ops & kernels
  #23275·JakeStevens opened Sep 30, 2026Contributor·

  ·
  1

  1

• NXP backend: Enable MobileNetV2 tests

  CLA Signedmodule: nxprelease notes: nxp
  #23274·roman-janik-nxp opened Sep 30, 2026Collaborator·

  ·
  3

• [Vulkan] Enable native Kev inference with the Vulkan backend

  CLA Signedmodule: vulkan
  #23272·Annubyx opened Sep 30, 2026·

  ·
  1

  4

• [LLM serving 1/x] Define text-serving contracts and share prompt planning

  CLA Signed
  #23270·metascroy opened Sep 30, 2026Contributor·

  ·
  9