Merge branch 'main' into dev/refactor_unittests

Author: AdrianLundell

.ci/scripts/utils.sh

@@ -17,17 +17,17 @@ retry () {
 }
 
 clean_executorch_install_folders() {
-  ./install_requirements.sh --clean
+  ./install_executorch.sh --clean
 }
 
 install_executorch() {
   which pip
   # Install executorch, this assumes that Executorch is checked out in the
   # current directory.
   if [[ "${1:-}" == "use-pt-pinned-commit" ]]; then
-    ./install_requirements.sh --pybind xnnpack --use-pt-pinned-commit
+    ./install_executorch.sh --pybind xnnpack --use-pt-pinned-commit
   else
-    ./install_requirements.sh --pybind xnnpack
+    ./install_executorch.sh --pybind xnnpack
   fi
   # Just print out the list of packages for debugging
   pip list

.github/workflows/apple.yml

@@ -9,7 +9,7 @@ on:
     paths:
       - .ci/scripts/setup-ios.sh
       - .github/workflows/apple.yml
-      - install_requirements.sh
+      - install_executorch.sh
       - backends/apple/**
       - build/build_apple_frameworks.sh
       - build/build_apple_llm_demo.sh

.github/workflows/pull.yml

@@ -200,7 +200,7 @@ jobs:
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "cmake"
 
         # install pybind
-        bash install_requirements.sh --pybind xnnpack
+        bash install_executorch.sh --pybind xnnpack
 
         # install Llava requirements
         bash examples/models/llama/install_requirements.sh
@@ -333,6 +333,9 @@ jobs:
 
   unittest-arm:
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    permissions:
+      id-token: write
+      contents: read
     with:
       runner: linux.2xlarge
       docker-image: executorch-ubuntu-22.04-arm-sdk
@@ -433,7 +436,7 @@ jobs:
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "cmake"
 
         # install pybind
-        bash install_requirements.sh --pybind xnnpack
+        bash install_executorch.sh --pybind xnnpack
 
         # install phi-3-mini requirements
         bash examples/models/phi-3-mini/install_requirements.sh
@@ -460,7 +463,7 @@ jobs:
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "cmake"
 
         # install pybind
-        bash install_requirements.sh --pybind xnnpack
+        bash install_executorch.sh --pybind xnnpack
 
         # install llama requirements
         bash examples/models/llama/install_requirements.sh
@@ -487,7 +490,7 @@ jobs:
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "cmake"
 
         # install pybind
-        bash install_requirements.sh --pybind xnnpack
+        bash install_executorch.sh --pybind xnnpack
 
         # install llama requirements
         bash examples/models/llama/install_requirements.sh
@@ -514,7 +517,7 @@ jobs:
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "cmake"
 
         # install pybind
-        bash install_requirements.sh --pybind xnnpack
+        bash install_executorch.sh --pybind xnnpack
 
         # install llama requirements
         bash examples/models/llama/install_requirements.sh

.github/workflows/trunk.yml

@@ -132,6 +132,9 @@ jobs:
   test-arm-backend-delegation:
     name: test-arm-backend-delegation
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    permissions:
+      id-token: write
+      contents: read
     with:
       runner: linux.2xlarge
       docker-image: executorch-ubuntu-22.04-arm-sdk
@@ -159,6 +162,9 @@ jobs:
   test-arm-reference-delegation:
     name: test-arm-reference-delegation
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    permissions:
+      id-token: write
+      contents: read
     with:
       runner: linux.2xlarge
       docker-image: executorch-ubuntu-22.04-arm-sdk

backends/apple/mps/setup.md

@@ -97,7 +97,7 @@ I 00:00:00.122615 executorch:mps_executor_runner.mm:501] Model verified successf
 ### [Optional] Run the generated model directly using pybind
 1. Make sure `pybind` MPS support was installed:
 ```bash
-./install_requirements.sh --pybind mps
+./install_executorch.sh --pybind mps
 ```
 2. Run the `mps_example` script to trace the model and run it directly from python:
 ```bash

backends/cadence/build_cadence_fusionG3.sh

@@ -12,7 +12,7 @@ unset XTENSA_CORE
 export XTENSA_CORE=FCV_FG3GP
 git submodule sync
 git submodule update --init
-./install_requirements.sh
+./install_executorch.sh
 
 rm -rf cmake-out
 

backends/cadence/build_cadence_hifi4.sh

@@ -12,7 +12,7 @@ unset XTENSA_CORE
 export XTENSA_CORE=nxp_rt600_RI23_11_newlib
 git submodule sync
 git submodule update --init
-./install_requirements.sh
+./install_executorch.sh
 
 rm -rf cmake-out
 

backends/cadence/fusion_g3/operators/op_dequantize.cpp

@@ -67,8 +67,8 @@ void check_dequantize_per_tensor_args(
 
   ET_CHECK_MSG(
       input.scalar_type() == dtype,
-      "input.scalar_type() %" PRId8 " is not matching dtype argumenta:",
-      static_cast<int8_t>(input.scalar_type()));
+      "input.scalar_type() %s is not matching dtype arguments:",
+      ::executorch::runtime::toString(input.scalar_type()));
 
   if (out_dtype.has_value()) {
     ET_CHECK_MSG(
@@ -561,11 +561,12 @@ Tensor& dequantize_per_tensor_out(
     const Tensor& input,
     double scale,
     int64_t zero_point,
-    int64_t quant_min,
-    int64_t quant_max,
+    __ET_UNUSED int64_t quant_min,
+    __ET_UNUSED int64_t quant_max,
     ScalarType dtype,
-    ::executorch::aten::optional<ScalarType> out_dtype,
     Tensor& out) {
+  constexpr ScalarType out_dtype = ScalarType::Float;
+
 #ifdef OP_ARG_CHECK
   torch::executor::Error err = resize_tensor(out, input.sizes());
   ET_CHECK_MSG(

backends/cadence/reference/operators/quantized_conv_out.cpp

@@ -119,7 +119,7 @@ __attribute__((noinline)) void conv2d_nchw_core_generic(
                     if (((_h + d0 * _wh - p0) >= 0) &&
                         ((_h + d0 * _wh - p0) < h) &&
                         ((_w + d1 * _ww - p1) >= 0) &&
-                        ((_w + d1 * _ww - p1 < w))) {
+                        ((_w + d1 * _ww - p1) < w)) {
                       int ioff =
                           (_h + d0 * _wh - p0) * w + (_w + d1 * _ww - p1);
                       int woff = _wh * ww + _ww;

backends/qualcomm/_passes/insert_requantize.py

@@ -89,15 +89,9 @@ def _single_output_annotation(
         requantize_dict = n.meta.pop(QCOM_REQUANTIZE)
         # {quant_attr: user_node_name_list}
         group_quant_attr_dict = self._invert_dict(requantize_dict)
-        # TODO: If users of the node contain output node,
-        # we replace the node with to_copy op. However, it would
-        # be problem when the node has multiple to_copy ops
-        add_output = len(group_quant_attr_dict) == 1
 
         for hashable_quant_attr, user_nodes in group_quant_attr_dict.items():
             user_nodes_copy = user_nodes.copy()
-            if add_output:
-                user_nodes_copy.append("output")
             self._insert_to_copy(gm, n, dict(hashable_quant_attr), user_nodes_copy)
 
     def _insert(self, graph_module: torch.fx.GraphModule) -> torch.fx.GraphModule: