Merge branch 'main' into update-compile-from-source-docs

Author: keyprocedure

.buckconfig

@@ -39,6 +39,7 @@
 
 [buck2]
 restarter=true
+file_watcher=notify
 
 [oss]
 folly_cxx_tests = False

.ci/docker/ci_commit_pins/buck2.txt

@@ -1 +1 @@
-2024-12-16
+2025-05-06

.ci/scripts/build-qnn-sdk.sh

@@ -33,13 +33,13 @@ set_up_aot() {
   cmake .. \
       -DCMAKE_INSTALL_PREFIX=$PWD \
       -DEXECUTORCH_BUILD_QNN=ON \
+      -DANDROID_NATIVE_API_LEVEL=30 \
       -DQNN_SDK_ROOT=${QNN_SDK_ROOT} \
       -DEXECUTORCH_BUILD_DEVTOOLS=ON \
       -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
       -DEXECUTORCH_BUILD_EXTENSION_TENSOR=ON \
       -DEXECUTORCH_ENABLE_EVENT_TRACER=ON \
-      -DPYTHON_EXECUTABLE=python3 \
-      -DEXECUTORCH_SEPARATE_FLATCC_HOST_PROJECT=OFF
+      -DPYTHON_EXECUTABLE=python3
   cmake --build $PWD --target "PyQnnManagerAdaptor" "PyQnnWrapperAdaptor" -j$(nproc)
   # install Python APIs to correct import path
   # The filename might vary depending on your Python and host version.

.ci/scripts/gather_benchmark_configs.py

@@ -21,6 +21,7 @@
     "apple_iphone_15": "arn:aws:devicefarm:us-west-2:308535385114:devicepool:02a2cf0f-6d9b-45ee-ba1a-a086587469e6/3b5acd2e-92e2-4778-b651-7726bafe129d",
     "apple_iphone_15+ios_18": "arn:aws:devicefarm:us-west-2:308535385114:devicepool:02a2cf0f-6d9b-45ee-ba1a-a086587469e6/12c8b15c-8d03-4e07-950d-0a627e7595b4",
     "samsung_galaxy_s22": "arn:aws:devicefarm:us-west-2:308535385114:devicepool:02a2cf0f-6d9b-45ee-ba1a-a086587469e6/e59f866a-30aa-4aa1-87b7-4510e5820dfa",
+    "samsung_galaxy_s22_private": "arn:aws:devicefarm:us-west-2:308535385114:devicepool:02a2cf0f-6d9b-45ee-ba1a-a086587469e6/ea6b049d-1508-4233-9a56-5d9eacbe1078",
     "samsung_galaxy_s24": "arn:aws:devicefarm:us-west-2:308535385114:devicepool:02a2cf0f-6d9b-45ee-ba1a-a086587469e6/98f8788c-2e25-4a3c-8bb2-0d1e8897c0db",
     "google_pixel_8_pro": "arn:aws:devicefarm:us-west-2:308535385114:devicepool:02a2cf0f-6d9b-45ee-ba1a-a086587469e6/d65096ab-900b-4521-be8b-a3619b69236a",
     "google_pixel_3_private_rooted": "arn:aws:devicefarm:us-west-2:308535385114:devicepool:02a2cf0f-6d9b-45ee-ba1a-a086587469e6/98d23ca8-ea9e-4fb7-b725-d402017b198d",

.ci/scripts/test_llava.sh

@@ -93,8 +93,7 @@ cmake_build_llava_runner_for_android() {
         -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK/build/cmake/android.toolchain.cmake \
         -DANDROID_ABI=arm64-v8a                                                 \
         ${LLAVA_COMMON_CMAKE_ARGS}                                              \
-        -DCMAKE_PREFIX_PATH="$python_lib"                  \
-        -DLLAVA_RUNNER_NO_TORCH_DUMMY_IMAGE=ON                                  \
+        -DCMAKE_PREFIX_PATH="$python_lib"                                       \
         -B${BUILD_DIR}/${dir}                                                   \
         ${dir}
 
@@ -107,11 +106,10 @@ export_llava() {
     $PYTHON_EXECUTABLE -m executorch.examples.models.llava.export_llava --pte-name llava.pte --with-artifacts
 }
 
-# Download a new image with different size, to test if the model can handle different image sizes
-prepare_image_tensor() {
+# Download a new image
+download_image() {
     echo "Downloading image"
     curl -o basketball.jpg https://upload.wikimedia.org/wikipedia/commons/7/73/Chicago_Bulls_and_New_Jersey_Nets%2C_March_28%2C_1991.jpg
-    $PYTHON_EXECUTABLE -m executorch.examples.models.llava.image_util --image-path basketball.jpg --output-path image.pt
 }
 
 run_and_verify() {
@@ -121,20 +119,18 @@ run_and_verify() {
         echo "Export failed. Abort"
         exit 1
     fi
-    if [[ ! -f "image.pt" ]]; then
-        echo "image.pt is missing."
+    if [[ ! -f "basketball.jpg" ]]; then
+        echo "basketball.jpg is missing."
         exit 1
     fi
     if [[ ! -f "tokenizer.bin" ]]; then
         echo "tokenizer.bin is missing."
         exit 1
     fi
 
-
-
     RUNTIME_ARGS="--model_path=llava.pte    \
         --tokenizer_path=tokenizer.bin      \
-        --image_path=image.pt               \
+        --image_path=basketball.jpg         \
         --prompt=ASSISTANT:                 \
         --temperature=0                     \
         --seq_len=650"
@@ -149,13 +145,8 @@ run_and_verify() {
 
     # verify result.txt
     RESULT=$(cat result.txt)
-    # set the expected prefix to be the same as prompt because there's a bug in sdpa_with_kv_cache that causes <unk> tokens.
-    if [[ "$(uname)" == "Darwin" ]]; then
-        EXPECTED_PREFIX="ASSISTANT: image captures a basketball game in progress, with several players on the court. One of the players is dribbling the ball, while the others are in various"
-    else
-        # set the expected prefix to be the same as prompt because there's a bug in sdpa_with_kv_cache that causes <unk> tokens.
-        EXPECTED_PREFIX="ASSISTANT: image"
-    fi
+    EXPECTED_PREFIX="ASSISTANT: image captures a basketball game in progress, with several players on the court. "
+
     if [[ "${RESULT}" == *"${EXPECTED_PREFIX}"* ]]; then
         echo "Expected result prefix: ${EXPECTED_PREFIX}"
         echo "Actual result: ${RESULT}"
@@ -184,5 +175,5 @@ fi
 export_llava
 
 # Step3. Run
-prepare_image_tensor
+download_image
 run_and_verify

.ci/scripts/test_model.sh

@@ -87,10 +87,6 @@ test_model() {
     bash examples/models/llava/install_requirements.sh
     STRICT="--no-strict"
   fi
-  if [[ "$MODEL_NAME" == "llama3_2_vision_encoder" || "$MODEL_NAME" == "llama3_2_text_decoder" ]]; then
-    # Install requirements for llama vision.
-    bash examples/models/llama3_2_vision/install_requirements.sh
-  fi
   if [[ "${MODEL_NAME}" == "qwen2_5" ]]; then
       # Install requirements for export_llama
       bash examples/models/llama/install_requirements.sh
@@ -205,7 +201,7 @@ test_model_with_qnn() {
   # TODO(guangyang): Make QNN chipset matches the target device
   QNN_CHIPSET=SM8450
 
-  "${PYTHON_EXECUTABLE}" -m examples.qualcomm.scripts.${EXPORT_SCRIPT} -b ${CMAKE_OUTPUT_DIR} -m ${QNN_CHIPSET} --compile_only $EXTRA_FLAGS
+  "${PYTHON_EXECUTABLE}" -m examples.qualcomm.scripts.${EXPORT_SCRIPT} -b ${CMAKE_OUTPUT_DIR} -m ${QNN_CHIPSET} --ci --compile_only $EXTRA_FLAGS
   EXPORTED_MODEL=$(find "./${EXPORT_SCRIPT}" -type f -name "${MODEL_NAME}*.pte" -print -quit)
 }
 
@@ -222,7 +218,7 @@ test_model_with_coreml() {
 
   DTYPE=float16
 
-  "${PYTHON_EXECUTABLE}" -m examples.apple.coreml.scripts.export --model_name="${MODEL_NAME}" --compute_precision "${DTYPE}"
+  "${PYTHON_EXECUTABLE}" -m examples.apple.coreml.scripts.export --model_name="${MODEL_NAME}" --compute_precision "${DTYPE}" --use_partitioner
   EXPORTED_MODEL=$(find "." -type f -name "${MODEL_NAME}*.pte" -print -quit)
 
   if [ -n "$EXPORTED_MODEL" ]; then

.ci/scripts/unittest-linux.sh

@@ -24,9 +24,6 @@ if [[ "$BUILD_TOOL" == "cmake" ]]; then
     CMAKE_ARGS="-DEXECUTORCH_BUILD_PYBIND=ON -DEXECUTORCH_BUILD_XNNPACK=ON -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON" \
     .ci/scripts/setup-linux.sh "$@"
 
-    # Install llama3_2_vision dependencies.
-    PYTHON_EXECUTABLE=python ./examples/models/llama3_2_vision/install_requirements.sh
-
     .ci/scripts/unittest-linux-cmake.sh
 elif [[ "$BUILD_TOOL" == "buck2" ]]; then
     # Removing this breaks sccache in the Buck build, apparently

.ci/scripts/unittest-macos.sh

@@ -29,7 +29,6 @@ if [[ "$BUILD_TOOL" == "cmake" ]]; then
     # Install llama3_2_vision dependencies.
     PYTHON_EXECUTABLE=python \
     ${CONDA_RUN} --no-capture-output \
-    ./examples/models/llama3_2_vision/install_requirements.sh
 
     .ci/scripts/unittest-macos-cmake.sh
 elif [[ "$BUILD_TOOL" == "buck2" ]]; then

.github/scripts/extract_benchmark_results.py

@@ -349,7 +349,10 @@ def transform(
     # Overwrite the device name here with the job name as it has more information about
     # the device, i.e. Samsung Galaxy S22 5G instead of just Samsung
     for r in benchmark_results:
-        r["deviceInfo"]["device"] = job_name
+        is_private_device = job_report.get("is_private_instance", False)
+        r["deviceInfo"]["device"] = (
+            f"{job_name} (private)" if is_private_device else job_name
+        )
 
     # From https://github.com/pytorch/pytorch/wiki/How-to-integrate-with-PyTorch-OSS-benchmark-database
     return [
@@ -363,6 +366,7 @@ def transform(
                     "benchmark_config": json.dumps(benchmark_config),
                     "job_conclusion": "SUCCESS",
                     "job_arn": job_report.get("arn", ""),
+                    "instance_arn": job_report.get("instance_arn", ""),
                 },
             },
             "model": {

.github/scripts/label_utils.py

@@ -22,12 +22,12 @@
 
 LABEL_ERR_MSG_TITLE = "This PR needs a `release notes:` label"
 LABEL_ERR_MSG = f"""# {LABEL_ERR_MSG_TITLE}
-If your changes are user facing and intended to be a part of release notes, please use a label starting with `release notes:`.
+If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with `release notes:`.
 
-If not, please add the `topic: not user facing` label.
+If not, please add the `release notes: none` label.
 
 To add a label, you can comment to pytorchbot, for example
-`@pytorchbot label "topic: not user facing"`
+`@pytorchbot label "release notes: none"`
 
 For more information, see
 https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.
@@ -115,7 +115,7 @@ def has_required_labels(pr: "GitHubPR") -> bool:
     pr_labels = pr.get_labels()
     # Check if PR is not user facing
     is_not_user_facing_pr = any(
-        label.strip() == "topic: not user facing" for label in pr_labels
+        label.strip() == "release notes: none" for label in pr_labels
     )
     return is_not_user_facing_pr or any(
         label.strip() in get_release_notes_labels(pr.org, pr.project)

.github/workflows/_link_check.yml

@@ -0,0 +1,57 @@
+on:
+  workflow_call:
+    inputs:
+      ref:
+        type: string
+        required: true
+
+jobs:
+  lint-urls:
+    if: ${{ github.event_name != 'pull_request' || !contains(github.event.pull_request.labels.*.name, 'skip-url-lint') }}
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    with:
+      runner: linux.2xlarge
+      docker-image: executorch-ubuntu-22.04-linter
+      submodules: false
+      fetch-depth: 0
+      ref: ${{ inputs.ref }}
+      timeout: 120
+      script: |
+        ./scripts/lint_urls.sh $(
+          { [ "${{ github.event_name }}" = "pull_request" ] \
+              && git diff --name-only "${{ github.event.pull_request.base.sha }}...${{ github.event.pull_request.head.sha }}"; } \
+          || \
+          { [ "${{ github.event_name }}" = "push" ] \
+              && git diff --name-only "${{ github.event.before }}...${{ github.sha }}"; }
+        ) || {
+          echo
+          echo "URL lint failed."
+          echo "If this is a transient outage, you can bypass it by adding the \`skip-url-lint\` label to your PR."
+          echo "Or add \`@lint-ignore\` somewhere on the same line as the URL you want to skip checking."
+          exit 1
+        }
+
+  lint-xrefs:
+    if: ${{ github.event_name != 'pull_request' || !contains(github.event.pull_request.labels.*.name, 'skip-xref-lint') }}
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    with:
+      runner: linux.2xlarge
+      docker-image: executorch-ubuntu-22.04-linter
+      submodules: false
+      fetch-depth: 0
+      ref: ${{ inputs.ref }}
+      timeout: 60
+      script: |
+        ./scripts/lint_xrefs.sh $(
+          { [ "${{ github.event_name }}" = "pull_request" ] \
+              && git diff --name-only "${{ github.event.pull_request.base.sha }}...${{ github.event.pull_request.head.sha }}"; } \
+          || \
+          { [ "${{ github.event_name }}" = "push" ] \
+              && git diff --name-only "${{ github.event.before }}...${{ github.sha }}"; }
+        ) || {
+          echo
+          echo "Xref lint failed."
+          echo "If this is a transient outage, you can bypass it by adding the \`skip-xref-lint\` label to your PR."
+          echo "Or add \`@lint-ignore\` somewhere on the same line as the reference you want to skip checking."
+          exit 1
+        }

.github/workflows/android-perf-private-device-experiment.yml

@@ -23,7 +23,7 @@ on:
         description: Target devices to run benchmark
         required: false
         type: string
-        default: google_pixel_3_private_rooted
+        default: samsung_galaxy_s22_private
       benchmark_configs:
         description: The list of configs used the benchmark
         required: false
@@ -39,7 +39,7 @@ on:
         description: Target devices to run benchmark
         required: false
         type: string
-        default: google_pixel_3_private_rooted
+        default: samsung_galaxy_s22_private
       benchmark_configs:
         description: The list of configs used the benchmark
         required: false
@@ -58,5 +58,5 @@ jobs:
       contents: read
     with:
       models: ${{ inputs.models || 'mv3,meta-llama/Llama-3.2-1B-Instruct-SpinQuant_INT4_EO8,meta-llama/Llama-3.2-1B-Instruct-QLORA_INT4_EO8' }}
-      devices: google_pixel_3_private_rooted
+      devices: samsung_galaxy_s22_private
       benchmark_configs: ${{ inputs.benchmark_configs }}

.github/workflows/apple-perf-private-device-experiment.yml

@@ -1,18 +1,16 @@
 name: apple-perf (private devices)
 
 on:
-  # TODO (huydhn): Disable the schedule run until we land the change to add device pool and device name
-  # to separate between public and private iOS devices
-  # schedule:
-  # - cron: 0 0,4,8,12,16,20 * * *
+  schedule:
+   - cron: 0 0,4,8,12,16,20 * * *
   pull_request:
     paths:
       - .github/workflows/apple-perf-private-device-experiment.yml
-  # push:
-  #   branches:
-  #     - main
-  #   paths:
-  #     - .github/workflows/apple-perf-private-device-experiment.yml
+  push:
+    branches:
+      - main
+    paths:
+      - .github/workflows/apple-perf-private-device-experiment.yml
   # Note: GitHub has an upper limit of 10 inputs
   workflow_dispatch:
     inputs:

.github/workflows/apple.yml

@@ -5,6 +5,8 @@ on:
     branches:
       - main
       - release/*
+    tags:
+      - ciflow/trunk/*
   pull_request:
     paths:
       - .ci/scripts/setup-ios.sh

.github/workflows/build-presets.yml

@@ -0,0 +1,30 @@
+name: Build Presets
+
+on:
+  pull_request:
+  push:
+    branches:
+      - main
+      - release/*
+  workflow_dispatch:
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.sha }}-${{ github.event_name == 'workflow_dispatch' }}-${{ github.event_name == 'schedule' }}
+  cancel-in-progress: true
+
+jobs:
+  apple:
+    uses: pytorch/test-infra/.github/workflows/macos_job.yml@main
+    strategy:
+      matrix:
+        preset: [macos-arm64]
+    with:
+      job-name: build
+      runner: macos-latest-xlarge
+      python-version: 3.12
+      submodules: recursive
+      script: |
+        set -eux
+        ${CONDA_RUN} ./install_requirements.sh > /dev/null
+        ${CONDA_RUN} cmake --preset ${{ matrix.preset }}
+        ${CONDA_RUN} cmake --build cmake-out --parallel

.github/workflows/doc-build.yml

@@ -14,20 +14,6 @@ on:
     - cron: '0 0 * * *'
 
 jobs:
-  check-urls:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v3
-      - name: Check URLs
-        run: bash ./scripts/check_urls.sh
-
-  check-xrefs:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v3
-      - name: Check Links
-        run: bash ./scripts/check_xrefs.sh
-
   build:
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
     permissions:

.github/workflows/lint.yml

@@ -64,6 +64,11 @@ jobs:
 
         exit $RC
 
+  link-check:
+    uses: ./.github/workflows/_link_check.yml
+    with:
+      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+
   android-java-format:
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
     permissions:

.github/workflows/nightly.yml

@@ -30,3 +30,9 @@ jobs:
           test-infra-ref: main
           updatebot-token: ${{ secrets.UPDATEBOT_TOKEN }}
           pytorchbot-token: ${{ secrets.GH_PYTORCHBOT_TOKEN }}
+
+  link-check:
+    needs: update-pytorch-commit-hash
+    uses: ./.github/workflows/_link_check.yml
+    with:
+      ref: ${{ github.sha }}