Merge branch 'main' of https://github.com/pytorch/executorch

training in oss

Author: JacobSzwejbka

.github/workflows/trunk.yml

@@ -223,7 +223,6 @@ jobs:
     strategy:
       matrix:
         dtype: [fp32]
-        build-tool: [buck2, cmake]
         mode: [portable, xnnpack+kv+custom, mps, coreml]
       fail-fast: false
     with:
@@ -235,25 +234,12 @@ jobs:
       script: |
 
         DTYPE=${{ matrix.dtype }}
-        BUILD_TOOL=${{ matrix.build-tool }}
         MODE=${{ matrix.mode }}
 
-        if [[ "${BUILD_TOOL}" == "buck2" ]]; then
-          # TODO: Will add more modes that don't support buck2
-          if [[ "${MODE}" == "mps" ]]; then
-            echo "mps doesn't support buck2."
-            exit 0
-          fi
-          if [[ "${MODE}" == "coreml" ]]; then
-            echo "coreml doesn't support buck2."
-            exit 0
-          fi
-        fi
-
         bash .ci/scripts/setup-conda.sh
 
         # Setup executorch
-        PYTHON_EXECUTABLE=python ${CONDA_RUN} bash .ci/scripts/setup-macos.sh "${BUILD_TOOL}"
+        PYTHON_EXECUTABLE=python ${CONDA_RUN} bash .ci/scripts/setup-macos.sh cmake
 
         if [[ "${MODE}" == "mps" ]]; then
           # Install mps delegate
@@ -268,7 +254,7 @@ jobs:
         # Install requirements for export_llama
         PYTHON_EXECUTABLE=python ${CONDA_RUN} bash examples/models/llama2/install_requirements.sh
         # Test llama2
-        PYTHON_EXECUTABLE=python ${CONDA_RUN} bash .ci/scripts/test_llama.sh stories110M "${BUILD_TOOL}" "${DTYPE}" "${MODE}"
+        PYTHON_EXECUTABLE=python ${CONDA_RUN} bash .ci/scripts/test_llama.sh stories110M cmake "${DTYPE}" "${MODE}"
 
   # # TODO(jackzhxng): Runner consistently runs out of memory before test finishes. Try to find a more powerful runner.
   # test-llava-runner-macos:

backends/apple/coreml/scripts/install_requirements.sh

@@ -24,7 +24,7 @@ rm -rf "$COREML_DIR_PATH/third-party"
 mkdir "$COREML_DIR_PATH/third-party"
 
 echo "${green}ExecuTorch: Cloning coremltools."
-git clone --depth 1 --branch 8.0b2 "https://github.com/apple/coremltools.git" $COREMLTOOLS_DIR_PATH
+git clone --depth 1 --branch 8.0 "https://github.com/apple/coremltools.git" $COREMLTOOLS_DIR_PATH
 cd $COREMLTOOLS_DIR_PATH
 
 STATUS=$?

backends/cadence/aot/utils.py

@@ -104,11 +104,11 @@ def get_ops_count(graph_module: torch.fx.GraphModule) -> Dict[str, int]:
             ):
                 continue
             # If the op is already present, increment the count
-            if get_edge_overload_packet(node.target).__name__ in freq:
-                freq[get_edge_overload_packet(node.target).__name__] += 1
+            if node.target._name in freq:
+                freq[node.target._name] += 1
             # else, add a new entry
             else:
-                freq[get_edge_overload_packet(node.target).__name__] = 1
+                freq[node.target._name] = 1
     return freq
 
 

backends/qualcomm/tests/utils.py

@@ -350,7 +350,6 @@ def lower_module_and_test_output(
                 # Therefore, won't want to pre-allocate
                 # by memory manager in runtime.
                 memory_planning_pass=MemoryPlanningPass(
-                    memory_planning_algo="greedy",
                     alloc_graph_input=not self.shared_buffer,
                     alloc_graph_output=not self.shared_buffer,
                 ),