Merge branch 'main' into arm-backend-llm-export

Author: xingguo01

.ci/scripts/test_cortex_m_e2e.sh

@@ -1,5 +1,6 @@
 #!/usr/bin/env bash
 # Copyright (c) Meta Platforms, Inc. and affiliates.
+# Copyright 2026 Arm Limited and/or its affiliates.
 # All rights reserved.
 #
 # This source code is licensed under the BSD-style license found in the
@@ -18,7 +19,7 @@ mkdir -p "./cortex_m_e2e/${MODEL}"
 WORK_DIR=$(realpath "./cortex_m_e2e/${MODEL}")
 
 echo "=== Exporting ${MODEL} with cortex-m55+int8 ==="
-python -m examples.arm.aot_arm_compiler \
+python -m backends.arm.scripts.aot_arm_compiler \
     -m "${MODEL}" \
     --target=cortex-m55+int8 \
     --quantize \

.ci/scripts/test_qnn_static_llm.sh

@@ -47,11 +47,11 @@ if [[ "${TASK_NAME}" == "stories_110m" ]]; then
     $PYTHON_EXECUTABLE -m pytorch_tokenizers.tools.llama2c.convert -t tokenizer.model -o tokenizer.bin
 
     # Compile only as weight sharing is not applicable on x86.
-    $PYTHON_EXECUTABLE backends/qualcomm/tests/test_qnn_delegate.py -k TestExampleLLMScript.test_llama_stories_110m --model SM8650 --build_folder build-android/ --executorch_root . --artifact_dir ./stories_110m_pte_size --llama_artifacts . --compile_only
+    $PYTHON_EXECUTABLE backends/qualcomm/tests/test_qnn_delegate.py -k TestExampleLLMScript.test_llama_stories_110m --soc_model SM8650 --build_folder build-android/ --executorch_root . --artifact_dir ./stories_110m_pte_size --llama_artifacts . --compile_only
     exit_code1=$?
 
     # Checks accuracy with weight sharing disabled since x86 does not support weight sharing.
-    $PYTHON_EXECUTABLE backends/qualcomm/tests/test_qnn_delegate.py -k TestExampleLLMScript.test_llama_stories_110m --model SM8650 --build_folder build-x86/ --executorch_root . --artifact_dir ./stories_110m_accuracy --llama_artifacts . --enable_x86_64
+    $PYTHON_EXECUTABLE backends/qualcomm/tests/test_qnn_delegate.py -k TestExampleLLMScript.test_llama_stories_110m --soc_model SM8650 --build_folder build-x86/ --executorch_root . --artifact_dir ./stories_110m_accuracy --llama_artifacts . --enable_x86_64
     exit_code2=$?
 
     # Check the exit codes and print messages
@@ -84,7 +84,7 @@ elif [[ "${TASK_NAME}" == "smollm2_135m" ]]; then
     if [ -n "$2" ]; then
         EXTRA_FLAGS="$EXTRA_FLAGS --static_llm_eval_method $2"
     fi
-    $PYTHON_EXECUTABLE backends/qualcomm/tests/test_qnn_delegate.py -k TestExampleLLMScript.test_static_llm_model --model_name smollm2_135m --model SM8650 --build_folder build-x86/ --executorch_root . --artifact_dir ./static_smollm2 --enable_x86_64 $EXTRA_FLAGS
+    $PYTHON_EXECUTABLE backends/qualcomm/tests/test_qnn_delegate.py -k TestExampleLLMScript.test_static_llm_model --model_name smollm2_135m --soc_model SM8650 --build_folder build-x86/ --executorch_root . --artifact_dir ./static_smollm2 --enable_x86_64 $EXTRA_FLAGS
     exit_code1=$?
     if [ $exit_code1 -ne 0 ]; then
         exit 1

.ci/scripts/test_yolo12.sh .ci/scripts/test_yolo26.sh.ci/scripts/test_yolo12.sh renamed to .ci/scripts/test_yolo26.sh

@@ -6,6 +6,7 @@
 # LICENSE file in the root directory of this source tree.
 
 set -ex
+
 # shellcheck source=/dev/null
 source "$(dirname "${BASH_SOURCE[0]}")/utils.sh"
 
@@ -50,21 +51,21 @@ PT2E_QUANTIZE="${PT2E_QUANTIZE:-}"
 # Default CMake Build Type to release mode
 CMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release}
 
-if [[ $# -lt 5 ]]; then # Assuming 4 mandatory args
-    echo "Expecting atleast 5 positional arguments"
-    echo "Usage: [...]"
-fi
 if [[ -z "${MODEL_NAME:-}" ]]; then
   echo "Missing model name, exiting..."
   exit 1
 fi
 
-
 if [[ -z "${MODE:-}" ]]; then
   echo "Missing mode, choose openvino or xnnpack, exiting..."
   exit 1
 fi
 
+if [[ -z "${VIDEO_PATH:-}" ]]; then
+  echo "Missing video path, exiting..."
+  exit 1
+fi
+
 if [[ -z "${PYTHON_EXECUTABLE:-}" ]]; then
   PYTHON_EXECUTABLE=python3
 fi
@@ -75,21 +76,13 @@ if [[ "${MODE}" =~ .*openvino.* ]]; then
   OPENVINO=ON
   TARGET_LIBS="$TARGET_LIBS openvino_backend "
 
-  git clone https://github.com/openvinotoolkit/openvino.git
-  cd openvino && git b16b776ac119dafda51f69a80f1e6b7376d02c3b
-  git submodule update --init --recursive
-  sudo ./install_build_dependencies.sh
-  mkdir build && cd build
-  cmake .. -DCMAKE_BUILD_TYPE=Release -DENABLE_PYTHON=ON
-  make -j$(nproc)
-
-  cd ..
-  cmake --install build --prefix dist
-
-  source dist/setupvars.sh
-  cd ../backends/openvino
-  pip install -r requirements.txt
-  cd ../../
+  # Install specific OpenVINO runtime from pip.
+  $PYTHON_EXECUTABLE -m pip install --pre openvino==2026.1.0.dev20260131 --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly
+  $PYTHON_EXECUTABLE -m pip install -r backends/openvino/requirements.txt
+
+  # Set OPENVINO_LIB_PATH so the C++ demo runner can also find libopenvino_c.so.
+  OPENVINO_LIB_PATH=$($PYTHON_EXECUTABLE -c "import openvino, os, glob; print(sorted(glob.glob(os.path.join(os.path.dirname(openvino.__file__), 'libs', 'libopenvino_c.so*')))[-1])")
+  export OPENVINO_LIB_PATH
 else
   OPENVINO=OFF
 fi
@@ -103,9 +96,10 @@ fi
 
 which "${PYTHON_EXECUTABLE}"
 
+TORCH_URL=https://download.pytorch.org/whl/cpu
 
-DIR="examples/models/yolo12"
-$PYTHON_EXECUTABLE -m pip install -r ${DIR}/requirements.txt
+DIR="examples/models/yolo26"
+$PYTHON_EXECUTABLE -m pip install --upgrade-strategy only-if-needed --extra-index-url "$TORCH_URL" -r ${DIR}/requirements.txt
 
 cmake_install_executorch_libraries() {
     rm -rf cmake-out
@@ -142,11 +136,11 @@ cmake_install_executorch_libraries() {
 
     echo $TARGET_LIBS
     export CMAKE_BUILD_ARGS="--target $TARGET_LIBS"
-    pip install . --no-build-isolation
+    $PYTHON_EXECUTABLE -m pip install . --no-build-isolation
 }
 
 cmake_build_demo() {
-    echo "Building yolo12 runner"
+    echo "Building yolo26 runner"
     retry cmake \
         -DCMAKE_BUILD_TYPE="$CMAKE_BUILD_TYPE" \
         -DUSE_OPENVINO_BACKEND="$OPENVINO" \
@@ -174,24 +168,29 @@ prepare_artifacts_upload() {
 
 
 # Export model.
-EXPORTED_MODEL_NAME="${MODEL_NAME}_fp32_${MODE}.pte"
-echo "Exporting ${EXPORTED_MODEL_NAME}"
 EXPORT_ARGS="--model_name=${MODEL_NAME} --backend=${MODE}"
+if [[ -n "${PT2E_QUANTIZE}" ]]; then
+  EXPORTED_MODEL_NAME="${MODEL_NAME}_int8_${MODE}.pte"
+  EXPORT_ARGS="${EXPORT_ARGS} --quantize --video_path=${VIDEO_PATH}"
+else
+  EXPORTED_MODEL_NAME="${MODEL_NAME}_fp32_${MODE}.pte"
+fi
+echo "Exporting ${EXPORTED_MODEL_NAME}"
 
 # Add dynamically linked library location
 cmake_install_executorch_libraries
 
-$PYTHON_EXECUTABLE -m examples.models.yolo12.export_and_validate ${EXPORT_ARGS}
+$PYTHON_EXECUTABLE -m examples.models.yolo26.export_and_validate ${EXPORT_ARGS}
 
 
 RUNTIME_ARGS="--model_path=${EXPORTED_MODEL_NAME} --input_path=${VIDEO_PATH}"
 # Check build tool.
 cmake_build_demo
-# Run yolo12 runner
+# Run yolo26 runner
 NOW=$(date +"%H:%M:%S")
-echo "Starting to run yolo12 runner at ${NOW}"
+echo "Starting to run yolo26 runner at ${NOW}"
 # shellcheck source=/dev/null
-cmake-out/examples/models/yolo12/Yolo12DetectionDemo ${RUNTIME_ARGS} > result.txt
+cmake-out/examples/models/yolo26/Yolo26DetectionDemo ${RUNTIME_ARGS} > result.txt
 NOW=$(date +"%H:%M:%S")
 echo "Finished at ${NOW}"
 

.claude/skills/qualcomm/SKILL.md

@@ -0,0 +1,98 @@
+---
+name: qualcomm
+description: Build, test, or develop the QNN (Qualcomm AI Engine Direct) backend. Use when working on backends/qualcomm/, building QNN (use backends/qualcomm/scripts/build.sh), adding new ops or passes, running QNN delegate
+  tests, or exporting models for Qualcomm HTP/GPU targets.
+---
+
+# QNN (Qualcomm AI Engine Direct) Backend
+
+## Advanced Topics
+
+When the user's request falls into one of these areas, read the corresponding file before proceeding:
+
+| Topic | File | When to read |
+|---|---|---|
+| Export / lowering / quantization options / pass pipelines | `lowering_export.md` | User asks about exporting, lowering, quantization config, QuantDtype, QuantRecipe, pass pipelines |
+| New op development | `new_op_development.md` | User asks to add/implement a new op or op builder |
+| Model enablement | `model_enablement.md` | User asks to enable a new model end-to-end |
+| Profiling & debugging | `profiling.md` | User asks about profiling, optrace, QHAS, QAIRT Visualizer *(file TBD)* |
+
+## Building
+
+Use `backends/qualcomm/scripts/build.sh`. Linux only (macOS not supported).
+
+**Environment variables:**
+- `QNN_SDK_ROOT` — path to QNN SDK (auto-downloaded if not set)
+- `ANDROID_NDK_ROOT` — path to Android NDK (auto-downloaded if not set)
+
+**Build targets:**
+
+| Target | Default | Build dir |
+|---|---|---|
+| x86_64 (Python interface + host tools) | enabled | `build-x86/` |
+| Android arm64-v8a (device runner) | enabled | `build-android/` |
+| Hexagon DSP (direct mode) | disabled | `build-hexagon/` |
+| OE Linux embedded | disabled | `build-oe-linux/` |
+
+**Common build commands:**
+
+```bash
+# Full build (x86_64 + Android)
+./backends/qualcomm/scripts/build.sh
+
+# x86_64 only (faster, for Python interface development)
+./backends/qualcomm/scripts/build.sh --skip_linux_android
+
+# Android only (skip x86_64)
+./backends/qualcomm/scripts/build.sh --skip_x86_64
+
+# Incremental build (skip clean)
+./backends/qualcomm/scripts/build.sh --no_clean
+
+# Enable Hexagon DSP direct mode (requires HEXAGON_SDK_ROOT, HEXAGON_TOOLS_ROOT, DSP_VERSION)
+./backends/qualcomm/scripts/build.sh --enable_hexagon
+
+# OE Linux embedded target (requires TOOLCHAIN_ROOT_HOST, TOOLCHAIN_ROOT_TARGET)
+./backends/qualcomm/scripts/build.sh --enable_linux_embedded
+
+# Release build
+./backends/qualcomm/scripts/build.sh --release
+
+# Control parallelism
+./backends/qualcomm/scripts/build.sh --job_number 8
+```
+
+**After x86_64 build**, the Python interface `.so` files are copied to `backends/qualcomm/python/` automatically.
+
+## Testing
+
+```bash
+QNN_SDK_ROOT=/path/to/qnn_sdk \
+ANDROID_NDK_ROOT=/path/to/android_ndk \
+LD_LIBRARY_PATH=/path/to/executorch/build-x86/lib:/path/to/qnn_sdk/lib/x86_64-linux-clang \
+PYTHONPATH=$(dirname $EXECUTORCH_ROOT) \
+python backends/qualcomm/tests/test_qnn_delegate.py \
+    TestQNNFloatingPointOperator.test_qnn_backend_abs \
+    -H $HOST -s $DEVICE_SERIAL -m SM8850 -b build-android -a /path/to/artifacts
+```
+
+> **Note (build from source):** Set `PYTHONPATH` to the parent directory of the executorch repo root. Required because `executorch.examples.qualcomm` lives in the source tree and is not installed into site-packages.
+
+Required flags: `-m` (SoC model), `-b` (Android build dir). Optional: `-s` (device serial), `-H` (host), `-a` (artifact dir), `-c` (compile only), `-x` (run on x86_64).
+
+**Test classes:**
+
+| Class | Description |
+|---|---|
+| `TestQNNFloatingPointOperator` | FP16 operator tests |
+| `TestQNNQuantizedOperator` | Quantized operator tests |
+| `TestQNNFloatingPointModel` | FP16 model-level tests |
+| `TestQNNQuantizedModel` | Quantized model-level tests |
+| `TestQNNFloatingPointUtils` | FP16 utility tests |
+| `TestQNNQuantizedUtils` | Quantized utility tests |
+| `TestExampleLLMScript` | LLM script tests |
+| `TestExampleMultimodalityScript` | Multimodality script tests |
+| `TestExampleOssScript` | OSS model script tests |
+| `TestExampleQaihubScript` | QAI Hub script tests |
+| `TestExampleScript` | General example script tests |
+| `TestUtilsScript` | Utility script tests |