Merge branch 'main' into update_tokenizer_pin_2

Author: mergennachin

.ci/docker/ci_commit_pins/optimum-executorch.txt

@@ -1 +1 @@
-de4f3c4978b4d36cc0bb8f87c6877a4a040d7ae7
+732b11313b2006b4d8649500eaf5567ec6ac1e49

.ci/docker/ci_commit_pins/pytorch.txt

@@ -1 +1 @@
-7a064ed3eafa43f17412d434b395240c727b3000
+7a79b41e29a790ebb4b530eb98a89381e2d7de29

.ci/scripts/setup-samsung-linux-deps.sh

@@ -101,6 +101,9 @@ install_devicefarm_cli() {
 }
 
 reserve_if_needed() {
+  # Set default value
+  export DEVICE_RESERVED=0
+
   if ! command -v devicefarm-cli >/dev/null 2>&1; then
     echo "[WARN] devicefarm-cli is not installed." >&2
     return 1
@@ -144,7 +147,11 @@ reserve_if_needed() {
 
   if (( any_below_threshold )); then
     echo "[INFO] Reserving now."
-    devicefarm-cli -R
+    if ! devicefarm-cli -R; then
+      echo "::warning::Failed to reserve a device. No devices are currently available." >&2
+      echo "[WARN] Device reservation failed - continuing without device." >&2
+      return 0
+    fi
   else
     echo "[INFO] Don't need to be reserved."
   fi
@@ -174,7 +181,10 @@ reserve_if_needed() {
   if [[ -n "$reservation_id" ]]; then
     devicefarm-cli -C "$reservation_id"
     devicefarm-cli -E "ls /"
+    export DEVICE_RESERVED=1
+    echo "[INFO] Device successfully reserved and connected."
   else
+    echo "::warning::No available devices found." >&2
     echo "[WARN] There is no available devices."
   fi
 }

.ci/scripts/wheel/test_windows.py

@@ -30,9 +30,7 @@ def test_model_xnnpack(model: Model, quantize: bool) -> None:
 
     if quantize:
         quant_type = MODEL_NAME_TO_OPTIONS[str(model)].quantization
-        model_instance = torch.export.export_for_training(
-            model_instance, example_inputs
-        )
+        model_instance = torch.export.export(model_instance, example_inputs)
         model_instance = quantize_xnn(
             model_instance.module(), example_inputs, quant_type
         )

.github/workflows/add-unanswered-to-project.yml

@@ -2,14 +2,20 @@ name: Add Open External Contributor PRs and Issues to PyTorch Org Project 136
 
 on:
   workflow_dispatch:
-  pull_request: 
-   paths: 
-     .github/workflows/add-unanswered-to-project.yml
+  schedule:
+    # GitHub Actions cron uses UTC. These run at:
+    #  - 14:00 UTC  -> 08:00 CST (UTC-6)
+    #  - 19:00 UTC  -> 13:00 CST (UTC-6)
+    - cron: "0 14 * * *"
+    - cron: "0 19 * * *"
+  pull_request:
+    paths:
+      - .github/workflows/add-unanswered-to-project.yml
 jobs:
   add_to_project:
     runs-on: ubuntu-latest
     steps:
-      - name: Add open issues and open, non-draft PRs to org project (excluding certain authors)
+      - name: Add open issues and open, non-draft PRs to org project (excluding certain authors and bots)
         uses: actions/github-script@v7
         with:
           github-token: ${{ secrets.ET_EXT_CONTRIB }}
@@ -41,13 +47,26 @@ jobs:
               "app/dependabot", "Erik-Lundell", "zingo", "AdrianLundell", "oscarandersson8218", "per", "Sebastian-Larsson", "SaoirseARM", 
               "robell", "mansnils", "martinlsm", "freddan80", "YufengShi-dudu", "tom-arm", "perheld", "Jerry-Ge", "gggekov", "fumchin", 
               "wwwind", "benkli01", "Tessil", "maddun01", "Michiel-Olieslagers", "armwaheed", "agrima1304", "emmakujala", "annietllnd", 
-              "MatthiasHertel80", "AlexTawseArm", "jmahbs", "morgolock", "Christoffer-JL", "ArmRyan", "xingguo01", "haowhsu-quic", 
-              "shewu-quic", "winskuo-quic", "chunit-quic", "DannyYuyang-quic", "chuntl", "thchenqti", "jethroqti", "chenweng-quic", 
-              "cymbalrush", "DenisVieriu97", "billmguo", "StrycekSimon", "jirioc", "robert-kalmar", "skywall", "MartinPavella", 
-              "roman-janik-nxp", "novak-vaclav ", "neuropilot-captain", "dijopaul", "cad-rlc", "cad-audio", "ynimmaga", "daniil-lyakhov", 
-              "emmanuel-ferdman", "cavusmustafa", "anzr299", "Jiseong-oh", "alexdean08"
+              "MatthiasHertel80", "AlexTawseArm", "jmahbs", "morgolock", "Christoffer-JL", "ArmRyan", "xingguo01", "tgonzalezorlandoarm", 
+              "haowhsu-quic", "shewu-quic", "winskuo-quic", "chunit-quic", "DannyYuyang-quic", "chuntl", "thchenqti", "jethroqti", 
+              "chenweng-quic", "cymbalrush", "DenisVieriu97", "billmguo", "StrycekSimon", "jirioc", "robert-kalmar", "skywall", 
+              "MartinPavella", "roman-janik-nxp", "novak-vaclav", "neuropilot-captain", "dijopaul", "cad-rlc", "cad-audio", "ynimmaga", 
+              "daniil-lyakhov", "emmanuel-ferdman", "cavusmustafa", "anzr299", "Jiseong-oh", "alexdean08",
+              // explicitly include the dependabot bot login seen in PRs
+              "dependabot[bot]"
             ]);
 
+            function isBotOrExcluded(user) {
+              if (!user) return false;
+              // GitHub sometimes marks bots with user.type === "Bot"
+              if (user.type && user.type.toLowerCase() === "bot") return true;
+              // Some bots use logins that end with [bot], e.g. dependabot[bot]
+              if (user.login && user.login.endsWith("[bot]")) return true;
+              // Explicit excluded list
+              if (excludedAuthors.has(user.login)) return true;
+              return false;
+            }
+
             async function addItem(contentId, type, number) {
               try {
                 await github.graphql(`
@@ -69,7 +88,7 @@ jobs:
             }
 
             try {
-              // Add open issues (not PRs) and exclude by author
+              // Add open issues (not PRs) and exclude by author/bots
               const issues = await github.paginate(
                 github.rest.issues.listForRepo,
                 {
@@ -80,12 +99,14 @@ jobs:
                 }
               );
               for (const issue of issues) {
-                if (!issue.pull_request && !excludedAuthors.has(issue.user.login)) {
+                if (!issue.pull_request && !isBotOrExcluded(issue.user)) {
                   await addItem(issue.node_id, 'issue', issue.number);
+                } else {
+                  console.log(`Skipping issue #${issue.number} by ${issue.user && issue.user.login}`);
                 }
               }
 
-              // Add open, non-draft PRs (regardless of review state), exclude by author
+              // Add open, non-draft PRs (regardless of review state), exclude by author/bots
               const prs = await github.paginate(
                 github.rest.pulls.list,
                 {
@@ -95,8 +116,10 @@ jobs:
                 }
               );
               for (const pr of prs) {
-                if (!pr.draft && !excludedAuthors.has(pr.user.login)) {
+                if (!pr.draft && !isBotOrExcluded(pr.user)) {
                   await addItem(pr.node_id, 'pr', pr.number);
+                } else {
+                  console.log(`Skipping PR #${pr.number} by ${pr.user && pr.user.login}`);
                 }
               }
             } catch (error) {

.github/workflows/cuda.yml

@@ -87,6 +87,36 @@ jobs:
         export LD_LIBRARY_PATH=/opt/conda/lib:$LD_LIBRARY_PATH
         PYTHON_EXECUTABLE=python source .ci/scripts/test_model.sh "${{ matrix.model }}" cmake cuda
 
+  unittest-cuda:
+    name: unittest-cuda
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    permissions:
+      id-token: write
+      contents: read
+    with:
+      timeout: 90
+      runner: linux.g5.4xlarge.nvidia.gpu
+      gpu-arch-type: cuda
+      gpu-arch-version: 12.6
+      use-custom-docker-registry: false
+      submodules: recursive
+      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+      script: |
+        set -eux
+        # Install executorch in editable mode so custom op libs land in-tree
+        bash ./install_executorch.sh
+
+        # Build ExecuTorch with CUDA support
+        cmake --workflow --preset llm-release-cuda
+
+        # Build and run CUDA shim tests (C++)
+        pushd backends/cuda/runtime/shims/tests
+        cmake --workflow --preset default
+        popd
+
+        # Run CUDA backend Python tests, overrides addopts so that we don't run all tests in pytest.ini
+        python -m pytest backends/cuda/tests backends/cuda/passes/tests -v -o "addopts="
+
   export-model-cuda-artifact:
     name: export-model-cuda-artifact
     # Skip this job if the pull request is from a fork (HuggingFace secrets are not available)

.github/workflows/pull.yml

@@ -490,17 +490,19 @@ jobs:
       build-tool: buck2
       docker-image: ci-image:executorch-ubuntu-22.04-clang12
 
-  unittest-arm-backend-with-no-fvp:
-    name: unittest-arm-backend-with-no-fvp
+  unittest-arm-backend-with-no-deps:
+    name: unittest-arm-backend-with-no-deps
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
     permissions:
       id-token: write
       contents: read
     strategy:
       matrix:
         include:
-          - test_arm_baremetal: test_pytest_ops
-          - test_arm_baremetal: test_pytest_models
+          - test_arm_baremetal: test_pytest_ops_no_target
+          - test_arm_baremetal: test_pytest_ops_tosa
+          - test_arm_baremetal: test_pytest_models_tosa
+          - test_arm_baremetal: test_run_tosa
       fail-fast: false
     with:
       runner: linux.2xlarge
@@ -516,7 +518,7 @@ jobs:
         source .ci/scripts/utils.sh
         install_executorch "--use-pt-pinned-commit"
 
-        .ci/scripts/setup-arm-baremetal-tools.sh
+        .ci/scripts/setup-arm-baremetal-tools.sh --disable-ethos-u-deps
 
         ARM_TEST=${{ matrix.test_arm_baremetal }}
 
@@ -945,6 +947,12 @@ jobs:
         export SAMSUNG_AI_LITECORE_KEY=$SECRET_SAMSUNG_AI_LITECORE_KEY
         source .ci/scripts/setup-samsung-linux-deps.sh
 
+        # Check if device was reserved
+        if [[ "${DEVICE_RESERVED:-0}" != "1" ]]; then
+          echo "::warning::Skipping tests - no Samsung device available"
+          exit 0
+        fi
+
         # Test quant models
         model_scripts="deeplab_v3 edsr inception_v3 inception_v4 mobilenet_v2 mobilenet_v3 resnet18 resnet50 vit wav2letter"
         for m_script in $model_scripts; do
@@ -981,6 +989,12 @@ jobs:
         export SAMSUNG_AI_LITECORE_KEY=$SECRET_SAMSUNG_AI_LITECORE_KEY
         source .ci/scripts/setup-samsung-linux-deps.sh
 
+        # Check if device was reserved
+        if [[ "${DEVICE_RESERVED:-0}" != "1" ]]; then
+          echo "::warning::Skipping tests - no Samsung device available"
+          exit 0
+        fi
+
         # Test models
         python -m unittest discover -s backends/samsung/test/models -p "test_*.py"
 

.github/workflows/trunk.yml

@@ -273,21 +273,21 @@ jobs:
         # Test selective build
         PYTHON_EXECUTABLE=python bash examples/portable/scripts/test_demo_backend_delegation.sh "${BUILD_TOOL}"
 
-  test-arm-backend:
-    name: test-arm-backend
+  test-arm-backend-ethos-u:
+    name: test-arm-backend-ethos-u
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
     permissions:
       id-token: write
       contents: read
     strategy:
       matrix:
         include:
-          - test_arm_baremetal: test_pytest_ops_ethosu_fvp
-          - test_arm_baremetal: test_pytest_models_ethosu_fvp
-          - test_arm_baremetal: test_run_ethosu_fvp
-          - test_arm_baremetal: test_models_tosa
-          - test_arm_baremetal: test_models_ethos-u55
-          - test_arm_baremetal: test_models_ethos-u85
+          - test_arm_baremetal: test_pytest_ops_ethos_u55
+          - test_arm_baremetal: test_pytest_models_ethos_u55
+          - test_arm_baremetal: test_run_ethos_u55
+          - test_arm_baremetal: test_pytest_ops_ethos_u85
+          - test_arm_baremetal: test_pytest_models_ethos_u85
+          - test_arm_baremetal: test_run_ethos_u85
           - test_arm_baremetal: test_smaller_stories_llama
           - test_arm_baremetal: test_memory_allocation
           - test_arm_baremetal: test_model_smollm2-135M

CODEOWNERS

@@ -18,7 +18,7 @@
 
 /devtools @Gasoonjia
 
-/docs @mergennachin
+/docs @mergennachin @AlannaBurke
 
 /examples/apple @shoumikhin
 /examples/apple/coreml @cccclai @metascroy @cymbalrush @YifanShenSZ

README.md

@@ -192,7 +192,7 @@ See [examples/models/llama](examples/models/llama/README.md) for complete workfl
 | macOS            | XNNPACK, MPS, Metal *(experimental)*                     |
 | Embedded / MCU   | XNNPACK, ARM Ethos-U, NXP, Cadence DSP                   |
 
-See [Backend Documentation](https://docs.pytorch.org/executorch/main/backends-overview.html) for detailed hardware requirements and optimization guides.
+See [Backend Documentation](https://docs.pytorch.org/executorch/main/backends-overview.html) for detailed hardware requirements and optimization guides. For desktop/laptop GPU inference with CUDA and Metal, see the [Desktop Guide](desktop/README.md). For Zephyr RTOS integration, see the [Zephyr Guide](zephyr/README.md).
 
 ## Production Deployments
 
@@ -204,9 +204,9 @@ ExecuTorch powers on-device AI at scale across Meta's family of apps, VR/AR devi
 
 **Multimodal:** [Llava](examples/models/llava/README.md) (vision-language), [Voxtral](examples/models/voxtral/README.md) (audio-language), [Gemma](examples/models/gemma3) (vision-language)
 
-**Vision/Speech:** [MobileNetV2](https://github.com/meta-pytorch/executorch-examples/tree/main/mv2), [DeepLabV3](https://github.com/meta-pytorch/executorch-examples/tree/main/dl3), [Whisper](https://github.com/meta-pytorch/executorch-examples/tree/main/whisper/android/WhisperApp)
+**Vision/Speech:** [MobileNetV2](https://github.com/meta-pytorch/executorch-examples/tree/main/mv2), [DeepLabV3](https://github.com/meta-pytorch/executorch-examples/tree/main/dl3), [Whisper](examples/models/whisper/README.md) <!-- @lint-ignore -->
 
-**Resources:** [`examples/`](examples/) directory • [executorch-examples](https://github.com/meta-pytorch/executorch-examples) out-of-tree demos • [Optimum-ExecuTorch](https://github.com/huggingface/optimum-executorch) for HuggingFace models
+**Resources:** [`examples/`](examples/) directory • [executorch-examples](https://github.com/meta-pytorch/executorch-examples) out-of-tree demos • [Optimum-ExecuTorch](https://github.com/huggingface/optimum-executorch) for HuggingFace models • [Unsloth](https://docs.unsloth.ai/new/deploy-llms-phone) for fine-tuned LLM deployment <!-- @lint-ignore -->
 
 ## Key Features