Build Aarch64 Linux CUDA Wheels

Workflow runs for Build Aarch64 Linux CUDA Wheels

• Load the torchci test-run reporter from the root conftest

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6056:Pull request#24992synchronizebypsiddh
  psiddh:export-D131320308

  Nov 13, 202631m 52s
  More actions for workflow run 6056

• Add quantized KV attention primitives

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6055:Pull request#25033openedbyxingguo01
  xingguo01:arm-quantized-kv-attention-primitives

  Nov 13, 202630m 16s
  More actions for workflow run 6055

• Enable Spark-X2.5 export and runner support for MLX

  requires action with the application:
  Build Aarch64 Linux CUDA Wheels#6054:Pull request#23523synchronizebydongjiang1989
  XHToken:add-spark2.5-mlx

  Nov 13, 2026Action required
  More actions for workflow run 6054

• Enable Spark-X2.5 export and runner support for MLX

  requires action with the application:
  Build Aarch64 Linux CUDA Wheels#6053:Pull request#23523synchronizebydongjiang1989
  XHToken:add-spark2.5-mlx

  Nov 13, 2026Action required
  More actions for workflow run 6053

• 2026-11-13 nightly release (1ab77be22c37536e133d7a7e9657f6276d7d7cf7)

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6052:Commita037f68pushedbypytorchbot
  nightly

  Nov 13, 202632m 6s
  More actions for workflow run 6052

• LFM3-VL-2B and LFM3-Omni-2B image, speech in and speech out (#25028)

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6051:Pull request#25028synchronizebypsiddh
  psiddh:export-D128739916

  Nov 13, 202629m 56s
  More actions for workflow run 6051

• Arm backend: Add example for collecting PMU counters.

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6050:Pull request#25029openedbyErik-Lundell
  Erik-Lundell:ethosu-etdump-example

  Nov 13, 202629m 21s
  More actions for workflow run 6050

• Phonon: run the Mimi transposed convolutions as convolutions on Vulkan, MLX and CUDA (#25020)

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6049:Pull request#25020synchronizebypsiddh
  psiddh:export-D131642733

  Nov 13, 202633m 13s
  More actions for workflow run 6049

• Document the version-1 hf_static I/O contract

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6048:Pull request#24573synchronizebypsiddh
  psiddh:stack2/pr08-docs

  Nov 13, 202632m 26s
  More actions for workflow run 6048

• LLM runner: validate and drive hf_static float-cache programs

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6047:Pull request#24572synchronizebypsiddh
  psiddh:stack2/pr07-hf-static-runtime

  Nov 13, 202632m 52s
  More actions for workflow run 6047

• Export hf_static float-cache programs and register their Python runner

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6046:Pull request#24571synchronizebypsiddh
  psiddh:stack2/pr06-hf-static-export

  Nov 13, 202632m 53s
  More actions for workflow run 6046

• QNN: bind delegate outputs in exported program order

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6045:Pull request#24570synchronizebypsiddh
  psiddh:stack2/pr05-qnn-output-order

  Nov 13, 202632m 40s
  More actions for workflow run 6045

• QNN: construct partitioners from options and allocate LLM I/O in rpcmem

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6044:Pull request#24569synchronizebypsiddh
  psiddh:stack2/pr04-qnn-options-allocator

  Nov 13, 202633m 15s
  More actions for workflow run 6044

• LLM runner: select I/O contracts and accept custom buffer allocators

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6043:Pull request#24568synchronizebypsiddh
  psiddh:stack2/pr03-runner-contracts

  Nov 13, 202632m 35s
  More actions for workflow run 6043

• Do not delegate quantized linears that have no Vulkan kernel (#24971)

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6042:Pull request#24971synchronizebypsiddh
  psiddh:export-D131277775

  Nov 13, 202631m 50s
  More actions for workflow run 6042

• Respect CPU quotas when sizing quantized GEMM compiler workers

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6041:Pull request#24889synchronizebyGasoonjia
  fix/muse-cgroup-compile-workers

  Nov 13, 202631m 19s
  More actions for workflow run 6041

• LFM3-VL-2B and LFM3-Omni-2B image, speech in and speech out (#25028)

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6040:Pull request#25028openedbypsiddh
  psiddh:export-D128739916

  Nov 13, 202637m 46s
  More actions for workflow run 6040

• LFM3 text decoder with conv-state support in LLM runners (#24687)

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6039:Pull request#24687synchronizebypsiddh
  psiddh:export-D128739913

  Nov 13, 202631m 9s
  More actions for workflow run 6039

• Respect CPU quotas when sizing quantized GEMM compiler workers

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6038:Pull request#24889synchronizebyGasoonjia
  fix/muse-cgroup-compile-workers

  Nov 13, 202629m 56s
  More actions for workflow run 6038

• Harden Android instrumentation CI and isolate performance checks (#24650)

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6037:Pull request#24650synchronizebypsiddh
  psiddh:export-D129513640

  Nov 13, 202633m 0s
  More actions for workflow run 6037

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

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6036:Pull request#24975synchronizebydigantdesai
  gh/digantdesai/204/head

  Nov 13, 202632m 18s
  More actions for workflow run 6036

• [executorch][native] Record embedding op samples

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6035:Pull request#24974synchronizebydigantdesai
  gh/digantdesai/203/head

  Nov 13, 202629m 44s
  More actions for workflow run 6035

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

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6034:Pull request#25024openedbydigantdesai
  gh/digantdesai/208/head

  Nov 13, 202631m 4s
  More actions for workflow run 6034

• [executorch][native] Record embedding op samples

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6033:Pull request#25023openedbydigantdesai
  gh/digantdesai/207/head

  Nov 13, 202633m 12s
  More actions for workflow run 6033

• Insert the .default overload of the affine quant ops in convert (#24993)

  completed successfully:
  Build Aarch64 Linux CUDA Wheels#6032:Pull request#24993synchronizebymeta-codesync[bot]Bot
  export-D130716026

  Nov 13, 202628m 12s
  More actions for workflow run 6032