All pull requests

Pull requests

• Preserve ORT perf settings during WinMLCG GPU timing

  #1523·Qiong Wu (qiowu) (DingmaomaoBJTU) opened Nov 6, 2026Collaborator·

  ·

• Avoid quadratic matching in large ETW traces

  #1514·Qiong Wu (qiowu) (DingmaomaoBJTU) opened Nov 3, 2026Collaborator·

  ·

• Revert native DXCGC perf backend (#1473)

  #1512·Qiong Wu (qiowu) (DingmaomaoBJTU) opened Nov 3, 2026Collaborator·

  ·

• Honor NPZ calibration input in static quantization

  model-opt-by-skill
  #1507·ssss141414 opened Oct 30, 2026Contributor·

  ·

• build(deps): bump onnx from 1.18 to 1.22.0

  dependenciespython
  #1504·dependabot[bot] opened Oct 29, 2026Bot·

  ·

• Adopt Python 3.14 as the sole CLI runtime

  #1489·fangyangci opened Oct 23, 2026Contributor·

  ·
  1

  4

• Align Raw DXCGC Run timing with Runtime API perf

  #1478·Qiong Wu (qiowu) (DingmaomaoBJTU) opened Oct 18, 2026Collaborator·

  ·

• Fix native DXCGC constant initialization and bound parser input

  #1476·Qiong Wu (qiowu) (DingmaomaoBJTU) opened Oct 16, 2026Collaborator·

  ·

• Include cgc-onboarding.md in doc navigation

  #1466·Zhenchao Ni (zhenchaoni) opened Oct 10, 2026Member·

  ·

• Expose GenAI context and prefill lengths in winml build

  #1463·Qiong Wu (qiowu) (DingmaomaoBJTU) opened Oct 8, 2026Collaborator·

  ·

• fix(loader): resolve registered custom architectures and validate weights

  #1459·Qiong Wu (qiowu) (DingmaomaoBJTU) opened Sep 29, 2026Collaborator·

  ·
  2

• Fix perf loading ModelKit build output directories

  #1453·fangyangci opened Sep 23, 2026Contributor·

  ·
  1

• Add split Whisper encoder-decoder support

  model-scale-by-skill
  #1407·ssss141414 opened Sep 10, 2026Contributor·

  ·
  8

• fix(eval): respect image-to-text pipeline call contracts

  #1405·Yue Sun (KayMKM) opened Sep 9, 2026Contributor·

  ·

• recipe(swin2sr): add QNN NPU configs

  model-opt-by-skillmodel-scale-by-skill
  #1396·Qiong Wu (qiowu) (DingmaomaoBJTU) opened Sep 7, 2026Collaborator·

  ·
  1

• feat(eval): add translation evaluation and refresh opus-mt-en-ru recipes

  model-scale-by-skill
  #1394·ssss141414 opened Sep 3, 2026Contributor·

  ·
  1

• feat(eval): add audio classification evaluation and refresh Gustking deepfake recipes

  model-scale-by-skill
  #1390·ssss141414 opened Sep 3, 2026Contributor·

  ·
  1

• feat(eval): add audio classification evaluation and MMS language ID recipes

  model-scale-by-skill
  #1389·ssss141414 opened Sep 3, 2026Contributor·

  ·
  1

• feat(eval): add audio classification evaluation and music genre recipes

  model-scale-by-skill
  #1388·ssss141414 opened Sep 2, 2026Contributor·

  ·
  4

• fix(build): honor --no-optimize and add multilingual similarity recipes

  model-scale-by-skill
  #1370·ssss141414 opened Aug 28, 2026Contributor·

  ·
  2

• fix(optrace): include CPU fallback operators in QNN results

  #1368·xieofxie opened Aug 27, 2026Contributor·

  ·
  1

  2

• test(eval): cover functional VitPose keypoint path

  model-scale-by-skill
  #1360·Qiong Wu (qiowu) (DingmaomaoBJTU) opened Aug 25, 2026Collaborator·

  ·
  4

• recipe(lt-wikidata-comp-en): sentence-similarity recipe

  model-scale-by-skill
  #1359·ssss141414 opened Aug 25, 2026Contributor·

  ·
  2

• recipe(deberta-v2): add mxbai rerank xsmall recipes and fp16 exclusions

  model-scale-by-skill
  #1355·ssss141414 opened Aug 25, 2026Contributor·

  ·
  2

• Add mxbai reranking recipes and exact FP16 boundary exclusions

  model-scale-by-skill
  #1354·ssss141414 opened Aug 24, 2026Contributor·

  ·
  1

  2