All pull requests

Pull requests

• Ship the LLM custom ops in executorch_jni_full

  CLA Signedmeta-exported
  #25003·phaiting opened Nov 12, 2026Contributor·

  ·
  3

• feat: shortening test names POC

  CLA Signedmodule: nxprelease notes: nxp
  #25002·novak-vaclav opened Nov 12, 2026Collaborator·

  ·
  2

  

• Recompose decomposed hardswish in the XNNPACK provider

  CLA Signedmeta-exported
  #25001·JakeStevens opened Nov 12, 2026Contributor·

  ·
  3

• Add clamp and div to the XNNPACK provider (#25000)

  CLA Signedmeta-exported
  #25000·JakeStevens opened Nov 12, 2026Contributor·

  ·
  3

• [ET-VK][conv2d] Add a Winograd F(4x4, 3x3) route for 3x3 convs

  CLA Signedmeta-exported
  #24999·SS-JIA opened Nov 12, 2026Contributor·

  ·
  2

• Update tokenizers submodule for CLIP support

  CLA Signed
  #24998·kiymetakdemir opened Nov 12, 2026Contributor·

  ·
  2

• Fix int4 prepare storage fallback and tensors on mobile exceding maxImageDimension3D

  CLA Signedrelease notes: none
  #24997·BasGame1 opened Nov 12, 2026·

  ·
  6

• Fix Corstone-300 runner link regression from #24264

  ciflow/trunkCLA Signedmodule: arm
  #24996·luciusacarr opened Nov 12, 2026Contributor·

  ·
  2

• Drop the ESP-IDF dev version's date check and restructure the Espressif README

  CLA Signed
  #24994·rascani opened Nov 12, 2026Contributor·

  ·
  3

• Insert the .default overload of the affine quant ops in convert

  CLA Signedmeta-exported
  #24993·mcremon-meta opened Nov 12, 2026Contributor·

  ·
  2

• Load the torchci test-run reporter from the root conftest (#24992)

  CLA Signedmeta-exported
  #24992·psiddh opened Nov 12, 2026Contributor·

  ·
  8

  

• Move the backend-specific pre-quantize passes and _masked_softmax out of fused_quant

  CLA Signedmeta-exported
  #24988·mcremon-meta opened Nov 12, 2026Contributor·

  ·
  3

• Bump rnllama's react-server-dom-webpack to 19.0.9

  CLA Signed
  #24987·izaitsevfb opened Nov 12, 2026Contributor·

  ·
  3

• [Vulkan][CI] Install Windows glslc from a pinned shaderc build, not conda-forge

  CLA Signedmodule: vulkan
  #24986·atalman opened Nov 12, 2026Contributor·

  ·
  2

• Add shared elementwise LUT fusion

  ciflow/trunkCLA Signedmodule: armpartner: armrelease notes: arm
  #24985·usamahz opened Nov 12, 2026Collaborator·

  ·
  1

  

• Add explicit session checkpoints to a file for LLM serving

  CLA Signedrelease notes: llm
  #24984·mergennachin opened Nov 12, 2026Contributor·

  ·
  1

• [executorch][cuda] Add triton::convolution and the conv routing framework

  CLA Signed
  #24982·Gasoonjia opened Nov 12, 2026Contributor·

  ·
  2

• [executorch][cuda] Add planned replacement to the Triton replacement pass

  CLA Signed
  #24981·Gasoonjia opened Nov 12, 2026Contributor·

  ·
  2

• [executorch][cuda] Deprecate triton_kernel_mode=OFF

  CLA Signed
  #24980·Gasoonjia opened Nov 12, 2026Contributor·

  ·
  2

• Size Muse Glimmer sliding-window ring buffer from the prefill chunk

  CLA Signed
  #24979·Gasoonjia opened Nov 12, 2026Contributor·

  ·
  3

• Arm backend: Delegate int64 ops with narrower integer operands under EXT-INT64

  ciflow/trunkCLA Signedcommunity: contributionmodule: armpartner: armrelease notes: arm
  #24978·abm-77 opened Nov 12, 2026Contributor·

  ·
  14

• Bound Samsung device allocation waits and clean up queued requests

  community: contribution
  #24977·pk954061458 opened Nov 12, 2026·

  ·
  1

  4

• [executorch][vulkan] Fetch embedding_q4gsw indices of rank 2 and 3 from textures

  CLA Signedmeta-exportedmodule: vulkan
  #24976·digantdesai opened Nov 12, 2026Contributor·

  ·
  2

• [executorch][native] Keep float scales of int4 ggml embeddings exact

  CLA Signed
  #24975·digantdesai opened Nov 12, 2026Contributor·

  ·
  2

• [executorch][native] Record embedding op samples

  CLA Signedmeta-exported
  #24974·digantdesai opened Nov 12, 2026Contributor·

  ·
  2