All pull requests

Pull requests

• [WebGPU] Restore ORT 1.24.4 compatibility with serial legacy execution

  #33082·Xiaofei Han (xiaofeihan1) opened Oct 3, 2026Contributor·

  ·
  3

• [js/react_native] Fix Android install() crash under Bridgeless/New Architecture

  #33081·Alex (exzos28) opened Oct 3, 2026·2 of 3 tasks·

  ·
  3

• CI: run MLAS SVE tests under QEMU on arm64 (fixes #33069)

  #33080·tanveer-arch opened Oct 3, 2026·

  ·
  1

  6

• Fix LayerNorm fusion with zero normalized dimensions

  #33079·Hulk (hulkbig) opened Oct 3, 2026·

  ·
  1

  2

• [CUDA] Load fpA_intB tactic cache on session create and save on close

  #33078·Tianlei Wu (tianleiwu) opened Oct 3, 2026Contributor·

  ·

• Add external Dawn API package CI coverage for the WebGPU plugin

  #33077·Mustapha Jaber (mustjab) opened Oct 3, 2026Contributor·

  ·
  1

  1

• [WebGPU] Tile large block-FP8 QMoE expert matmuls

  #33075·Sayan Shaw (sayanshaw24) opened Oct 2, 2026Contributor·2/2·

  ·
  1

• [WebGPU] Support block-quantized FP8 dense MatMul

  #33074·Sayan Shaw (sayanshaw24) opened Oct 2, 2026Contributor·

  ·
  5

• Fix unbounded npm package reference

  #33072·eserscor opened Oct 2, 2026Contributor·

  ·
  6

• [WebGPU] Add FP16 GEMV specialization for single-row decode projections

  #33070·aciddelgado opened Oct 2, 2026Contributor·

  ·
  13

• Raise ir_version to 10 when MatMulNBitsQuantizer emits INT4/UINT4 tensors

  #33068·Andreas Fehlner (andife) opened Oct 2, 2026Contributor·

  ·
  2

• Align C API tensor element type values with ONNX TensorProto

  #33066·tianleiwu with CopilotAI opened Oct 2, 2026·

  ·
  8

  

• Bind external data validation to opened files

  #33063·Akshay Sonawane (apsonawane) opened Oct 1, 2026Contributor·

  ·
  30

• Retain external-data file handles and original model directories

  #33062·Gopalakrishnan Nallasamy (GopalakrishnanN) opened Oct 1, 2026Contributor·

  ·
  1

• [WebGPU] Reduce QMoE single-token gate sorting to stable TopK

  #33058·Sayan Shaw (sayanshaw24) opened Oct 1, 2026Contributor·

  ·
  6

• [WebGPU EP] fixes fp16 accumulation bug in gemm kernel

  #33057·Prathik Rao (prathikr) opened Oct 1, 2026Contributor·

  ·
  19

• [WebGPU] Accelerated model loading using Win32 and D3D12

  #33055·Sushanth Rajasankar (sushraja-msft) opened Oct 1, 2026Contributor·

  ·
  92

• Reuse device streams across threads when graph capture is disabled

  #33053·tianleiwu with CopilotAI opened Oct 1, 2026·

  ·
  1

  1

  

• Fix provider bridge initialization failures without exceptions

  #33051·Mustapha Jaber (mustjab) opened Oct 1, 2026Contributor·

  ·
  8

• [MLAS] Cover non-finite and lane-position inputs in the fp16 exp test

  #33050·Vijay Kapse (vijay-kapse) opened Oct 1, 2026Contributor·

  ·
  5

• [MLAS] Build the SVE asm generator's input TUs as C++20

  #33047·Vijay Kapse (vijay-kapse) opened Oct 1, 2026Contributor·

  ·
  2

• Fix LayerNorm fusion exponent and epsilon validation

  #33046·Raunak (raun6k) opened Oct 1, 2026·

  ·
  1

  2

• [CUDA] Plumb the attention dispatch policy session option into the L1 estimator (reader-only)

  #33042·Chi Lo (chilo-ms) opened Oct 1, 2026Contributor·

  ·

• Enable static CPU offload for FP16 CUDA MoE

  #33039·Xavier Dupré (xadupre) opened Oct 1, 2026Member·

  ·
  73

• fix(qnn/cpu): enable stft on cpu backend of QNN EP

  #33035·Nicolas Gry (grybouilli) opened Oct 1, 2026Contributor·

  ·
  22