Commits

Branch selector

 main

User selector

All users

All time

Commit History

Commits on Apr 22, 2026

• NXP backend: Add Linear+BN fusion to conversion pipeline of unit tests (#18527)

  Show description for 941486e
  StrycekSimon

  authored
  941486e

  Copy full SHA for 941486e

• Arm backend: Add test transpose count suite (#19022)

  Show description for 656850a
  AdrianLundell

  authored
  656850a

  Copy full SHA for 656850a

• Metal backend: Materialize non-packed tensor views in reinterpret_tensor (#19033)

  Show description for 6be4fb5
  manuelcandales

  authored
  6be4fb5

  Copy full SHA for 6be4fb5

• Move permute optimization passes to shared transforms location (#19002)

  Show description for 273888f
  mcremon-meta

  authored
  273888f

  Copy full SHA for 273888f

• Precompute T1 offset for quantized conv2d NHWC in TIE kernel (#18960)

  Show description for 89600b3
  abeakkas

  authored
  89600b3

  Copy full SHA for 89600b3

Commits on Apr 21, 2026

• Android: LlmModule Closeable lifecycle and ModelRunner crash fix (#19012)

  Show description for 54b0148
  psiddh

  authored
  54b0148

  Copy full SHA for 54b0148

• Enable infinite generation with RoPE position remapping for attention sink (#19011)

  Show description for 239f7cc
  kirklandsign

  authored
  239f7cc

  Copy full SHA for 239f7cc

• Xnnpack flatbuffer: check non-null and num_dims matches array size in defineTensor (#18896)

  Show description for 2d53535
  

  lucylq

  and

  Github Executorch

  authored
  2d53535

  Copy full SHA for 2d53535

• Destroy unwrapped_vals instead of wrapped_vals (#18962)

  Show description for deaf73b
  

  lucylq

  and

  Github Executorch

  authored
  deaf73b

  Copy full SHA for deaf73b

• Add tribal knowledge base and /executorch-kb skill (#19003)

  Show description for 2739129
  abhinaykukkadapu

  authored
  2739129

  Copy full SHA for 2739129

• Enable debug logging in llm-debug preset (#19005)

  Show description for 1d37abd
  manuelcandales

  authored
  1d37abd

  Copy full SHA for 1d37abd

• Qwen 3.5 MoE Metal: Use max-sized prefill example for dynamic inputs (#18956)

  Show description for f13b783
  manuelcandales

  authored
  f13b783

  Copy full SHA for f13b783

• Qwen 3.5 MoE: Add Metal build target (#18881)

  Show description for 2c707e9
  manuelcandales

  authored
  2c707e9

  Copy full SHA for 2c707e9

• Qwen 3.5 MoE: Add --backend metal export path (#18880)

  Show description for 799bf5a
  manuelcandales

  authored
  799bf5a

  Copy full SHA for 799bf5a

• Qwen 3.5 MoE: Add Metal source transformations (#18879)

  Show description for 9600f63
  manuelcandales

  authored
  9600f63

  Copy full SHA for 9600f63

• Metal backend: Add gated delta rule kernel for linear attention (#18878)

  Show description for d408a10
  manuelcandales

  authored
  d408a10

  Copy full SHA for d408a10

• Relax SQNR check for mv2 (#19030)

  Show description for 2fce946
  JacobSzwejbka

  authored
  2fce946

  Copy full SHA for 2fce946

• Metal backend: Add gather_qmv kernel for MoE expert-indexed quantized matmul (#18877)

  Show description for 6875814
  manuelcandales

  authored
  6875814

  Copy full SHA for 6875814

• Add labeler for ARM path changes (#18606)

  Show description for ccaf17e
  Sebastian-Larsson

  authored
  ccaf17e

  Copy full SHA for ccaf17e

• Fix unchecked map access in xnnpack (#19008)

  Show description for e281726
  

  lucylq

  and

  Github Executorch

  authored
  e281726

  Copy full SHA for e281726

• Add instruction execution limit to prevent infinite loops (#18679)

  Show description for c4983ce
  

  lucylq

  and

  Github Executorch

  authored
  c4983ce

  Copy full SHA for c4983ce

• Fix float32 indices in test_batched_export_with_backprop

  Show description for 78a6689
  jeanschmidt

  authored
  78a6689

  Copy full SHA for 78a6689

• Enable Q/DQ opt with multiple inputs

  Show description for 8e5ec80
  boomitsnoom

  authored
  8e5ec80

  Copy full SHA for 8e5ec80

Commits on Apr 20, 2026

• Improve RemovePermutesAroundElementwiseOps robustness (#18980)

  Show description for c391738
  mcremon-meta

  authored
  c391738

  Copy full SHA for c391738

• Add PR-blocking CI for QNN model tests, operator/model testsuite, and pass unit tests (#18854)

  Show description for 6c82462
  abhinaykukkadapu

  authored
  6c82462

  Copy full SHA for 6c82462

• [Examples] Yolo26 Sample for OpenVINO/XNNPACK (#18583)

  Show description for 6a9d749
  

  daniil-lyakhov

  and

  suryasidd

  authored
  6a9d749

  Copy full SHA for 6a9d749

• Cap num_externs to 4096 (#18909)

  Show description for 0af4992
  lucylq

  authored
  0af4992

  Copy full SHA for 0af4992

• Qualcomm AI Engine Direct - Adding QNN backend support for reflection_pad1/2d core ATen ops (#18963)

  qti-horodnic

  authored
  8ed6e85

  Copy full SHA for 8ed6e85

• Metal backend: Add topk fallback kernel via MPSGraph (#18876)

  Show description for 66e4656
  manuelcandales

  authored
  66e4656

  Copy full SHA for 66e4656

• Add OSS Buck build support and CI for Qualcomm backend (#18666)

  Show description for 9dd342c
  abhinaykukkadapu

  authored
  9dd342c

  Copy full SHA for 9dd342c

• Android: unified error reporting — all sync errors throw, no silent f…

  Show description for eafc6e2
  psiddh

  authored
  eafc6e2

  Copy full SHA for eafc6e2

• Metal backend: Add SDPA head_dim=256 support (#18875)

  Show description for 4898af2
  manuelcandales

  authored
  4898af2

  Copy full SHA for 4898af2

• Metal backend: Add INT32 and BOOL dtype support (#18874)

  Show description for ad27a45
  manuelcandales

  authored
  ad27a45

  Copy full SHA for ad27a45

• Skip np.transpose for identity dim_order in process_node

  Show description for 401ea8e
  mcremon-meta

  authored
  401ea8e

  Copy full SHA for 401ea8e

• Fix mobilenet input download in ci (#19001)

  Show description for 0af4221
  JacobSzwejbka

  authored
  0af4221

  Copy full SHA for 0af4221