Merge remote-tracking branch 'origin/main' into mtk-android

Author: kirklandsign

.ci/docker/ci_commit_pins/pytorch.txt

@@ -1 +1 @@
-d1b87e26e5c4343f5b56bb1e6f89b479b389bfac
+bd5482c7c3e1197e10c46ff739027f917d9c1fcc

.github/scripts/propose_ghstack_orig_pr.py

@@ -26,9 +26,9 @@ def parse_args():
         required=True,
     )
     parser.add_argument(
-        "--pr",
-        type=int,
-        help="Number of the PR in the stack to check and create corresponding PR",
+        "--ref",
+        type=str,
+        help="Ref fo PR in the stack to check and create corresponding PR",
         required=True,
     )
     return parser.parse_args()
@@ -68,12 +68,18 @@ def extract_stack_from_body(pr_body: str) -> List[int]:
     return list(reversed(prs))
 
 
-def get_pr_stack_from_number(pr_number: int, repo: Repository) -> List[int]:
+def get_pr_stack_from_number(ref: str, repo: Repository) -> List[int]:
+    if ref.isnumeric():
+        pr_number = int(ref)
+    else:
+        branch_name = ref.replace("refs/heads/", "")
+        pr_number = repo.get_branch(branch_name).commit.get_pulls()[0].number
+
     pr_stack = extract_stack_from_body(repo.get_pull(pr_number).body)
 
     if not pr_stack:
         raise Exception(
-            f"Could not find PR stack in body of #{pr_number}. "
+            f"Could not find PR stack in body of ref. "
             + "Please make sure that the PR was created with ghstack."
         )
 
@@ -100,14 +106,15 @@ def create_prs_for_orig_branch(pr_stack: List[int], repo: Repository):
 ghstack PR base: https://github.com/pytorch/executorch/tree/{pr.base.ref}
 ghstack PR head: https://github.com/pytorch/executorch/tree/{pr.head.ref}
 Merge bot PR base: https://github.com/pytorch/executorch/tree/{orig_branch_merge_base}
-Merge bot PR head: https://github.com/pytorch/executorch/tree/{orig_branch_merge_head}"""
+Merge bot PR head: https://github.com/pytorch/executorch/tree/{orig_branch_merge_head}
+@diff-train-skip-merge"""
 
         existing_orig_pr = repo.get_pulls(
             head="pytorch:" + orig_branch_merge_head,
             base=orig_branch_merge_base,
-            state="open",
+            state="all",
         )
-        if existing_orig_pr.totalCount > 0:
+        if existing_orig_pr.totalCount > 0 and existing_orig_pr[0].title == pr.title:
             print(
                 f"PR for {orig_branch_merge_head} already exists {existing_orig_pr[0]}"
             )
@@ -128,7 +135,7 @@ def main():
 
     with Github(auth=Auth.Token(os.environ["GITHUB_TOKEN"])) as gh:
         repo = gh.get_repo(args.repo)
-        create_prs_for_orig_branch(get_pr_stack_from_number(args.pr, repo), repo)
+        create_prs_for_orig_branch(get_pr_stack_from_number(args.ref, repo), repo)
 
 
 if __name__ == "__main__":

.github/workflows/android.yml renamed to .github/workflows/_android.yml

@@ -1,28 +1,9 @@
 name: Android
 
 on:
-  push:
-    branches:
-      - main
-      - release/*
-    tags:
-      - ciflow/android/*
-  pull_request:
-    paths:
-      - .ci/docker/**
-      - .github/workflows/android.yml
-      - build/*android*.sh
-      - install_requirements.sh
-      - examples/demo-apps/android/**
-      - extension/android/**
-      - extension/benchmark/android/**
-      - extension/module/**
+  workflow_call:
   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:
   build-llm-demo:
     name: build-llm-demo

.github/workflows/apple-perf.yml

@@ -76,7 +76,7 @@ jobs:
           # on-demand and periodic benchmarking.
           CRON_DEFAULT_MODELS: "stories110M,mv3,mv2,ic4,ic3,resnet50,edsr,mobilebert,w2l"
           CRON_DEFAULT_DEVICES: "apple_iphone_15"
-          CRON_DEFAULT_DELEGATES: "nnpack,coreml,mps"
+          CRON_DEFAULT_DELEGATES: "xnnpack,coreml,mps"
         run: |
           set -ex
           MODELS="${{ inputs.models }}"

.github/workflows/ghstack_land.yml

@@ -11,10 +11,12 @@ on:
       - 'gh/kimishpatel/[0-9]+/base'
       - 'gh/kirklandsign/[0-9]+/base'
       - 'gh/larryliu0820/[0-9]+/base'
+      - 'gh/lucylq/[0-9]+/base'
       - 'gh/manuelcandales/[0-9]+/base'
       - 'gh/mcr229/[0-9]+/base'
       - 'gh/swolchok/[0-9]+/base'
       - 'gh/SS-JIA/[0-9]+/base'
+      - 'gh/trivedivivek/[0-9]+/base'
 
 jobs:
   ghstack_merge_to_main:
@@ -32,9 +34,7 @@ jobs:
       run: |
         pip install pygithub
 
-        PR_NUMBER=$(echo "$GITHUB_REF" | grep -oE '[0-9]+')
-
-        python .github/scripts/propose_ghstack_orig_pr.py --pr $PR_NUMBER --repo pytorch/executorch
+        python .github/scripts/propose_ghstack_orig_pr.py --ref $GITHUB_REF --repo pytorch/executorch
       env:
         GITHUB_TOKEN: ${{ secrets.GH_PYTORCHBOT_CHERRY_PICK_TOKEN }}
         GITHUB_REF: ${{ github.ref }}

.github/workflows/pull.yml

@@ -35,9 +35,6 @@ jobs:
     name: test-setup-linux-gcc
     uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
     strategy:
-      matrix:
-        include:
-          - build-tool: cmake
       fail-fast: false
     with:
       runner: linux.2xlarge
@@ -50,7 +47,7 @@ jobs:
         CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
         conda activate "${CONDA_ENV}"
 
-        BUILD_TOOL=${{ matrix.build-tool }}
+        BUILD_TOOL="cmake"
 
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "${BUILD_TOOL}"
         # Build and test ExecuTorch with the add model on portable backend.
@@ -89,20 +86,11 @@ jobs:
     strategy:
       matrix:
         dtype: [fp32]
-        build-tool: [buck2, cmake]
         mode: [portable, xnnpack+custom, xnnpack+custom+qe]
         include:
           - dtype: bf16
-            build-tool: cmake
-            mode: portable
-          - dtype: bf16
-            build-tool: buck2
             mode: portable
           - dtype: bf16
-            build-tool: cmake
-            mode: custom
-          - dtype: bf16
-            build-tool: buck2
             mode: custom
       fail-fast: false
     with:
@@ -117,11 +105,11 @@ jobs:
         conda activate "${CONDA_ENV}"
 
         DTYPE=${{ matrix.dtype }}
-        BUILD_TOOL=${{ matrix.build-tool }}
+        BUILD_TOOL="cmake"
         MODE=${{ matrix.mode }}
 
         # Setup executorch
-        PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh buck2
+        PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "${BUILD_TOOL}"
         # Install requirements for export_llama
         PYTHON_EXECUTABLE=python bash examples/models/llama/install_requirements.sh
         # Test llama2
@@ -131,9 +119,6 @@ jobs:
     name: test-llama-runner-linux-android
     uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
     strategy:
-      matrix:
-        include:
-          - build-tool: cmake
       fail-fast: false
     with:
       runner: linux.2xlarge
@@ -146,18 +131,14 @@ jobs:
         CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
         conda activate "${CONDA_ENV}"
 
-        BUILD_TOOL=${{ matrix.build-tool }}
+        BUILD_TOOL="cmake"
         PYTHON_EXECUTABLE=python \
         bash .ci/scripts/build_llama_android.sh  "${BUILD_TOOL}"
 
   test-custom-ops-linux:
     name: test-custom-ops-linux
     uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
     strategy:
-      matrix:
-        include:
-          - build-tool: buck2
-          - build-tool: cmake
       fail-fast: false
     with:
       runner: linux.2xlarge
@@ -170,7 +151,7 @@ jobs:
         CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
         conda activate "${CONDA_ENV}"
 
-        BUILD_TOOL=${{ matrix.build-tool }}
+        BUILD_TOOL="cmake"
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "${BUILD_TOOL}"
         # Test custom ops
         PYTHON_EXECUTABLE=python bash examples/portable/custom_ops/test_custom_ops.sh "${BUILD_TOOL}"
@@ -179,10 +160,6 @@ jobs:
     name: test-selective-build-linux
     uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
     strategy:
-      matrix:
-        include:
-          - build-tool: buck2
-          - build-tool: cmake
       fail-fast: false
     with:
       runner: linux.2xlarge
@@ -195,7 +172,7 @@ jobs:
         CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
         conda activate "${CONDA_ENV}"
 
-        BUILD_TOOL=${{ matrix.build-tool }}
+        BUILD_TOOL="cmake"
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "${BUILD_TOOL}"
         # Test selective build
         PYTHON_EXECUTABLE=python bash examples/selective_build/test_selective_build.sh "${BUILD_TOOL}"
@@ -235,9 +212,6 @@ jobs:
     name: test-quantized-aot-lib-linux
     uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
     strategy:
-      matrix:
-        include:
-          - build-tool: cmake
       fail-fast: false
     with:
       runner: linux.2xlarge
@@ -250,17 +224,14 @@ jobs:
         CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
         conda activate "${CONDA_ENV}"
 
-        BUILD_TOOL=${{ matrix.build-tool }}
+        BUILD_TOOL="cmake"
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "${BUILD_TOOL}"
         PYTHON_EXECUTABLE=python bash examples/xnnpack/quantization/test_quantize.sh "${BUILD_TOOL}" mv2
 
   test-pybind-build-linux:
     name: test-pybind-build-linux
     uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
     strategy:
-      matrix:
-        include:
-          - build-tool: cmake
       fail-fast: false
     with:
       runner: linux.2xlarge
@@ -274,7 +245,7 @@ jobs:
         conda activate "${CONDA_ENV}"
 
         # build module for executorch.extension.pybindings.portable_lib
-        BUILD_TOOL=${{ matrix.build-tool }}
+        BUILD_TOOL="cmake"
         PYTHON_EXECUTABLE=python \
         EXECUTORCH_BUILD_XNNPACK=ON \
         EXECUTORCH_BUILD_PYBIND=ON \
@@ -347,17 +318,16 @@ jobs:
           exit 1
         fi
 
+  android:
+    uses: ./.github/workflows/_android.yml
+
   unittest:
     uses: ./.github/workflows/_unittest.yml
     with:
       docker-image: executorch-ubuntu-22.04-clang12
 
   unittest-arm:
     uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
-    strategy:
-      matrix:
-        include:
-          - build-tool: buck2
     with:
       runner: linux.2xlarge
       docker-image: executorch-ubuntu-22.04-arm-sdk
@@ -371,7 +341,7 @@ jobs:
         CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
         conda activate "${CONDA_ENV}"
 
-        BUILD_TOOL=${{ matrix.build-tool }}
+        BUILD_TOOL="cmake"
 
         # Setup MacOS dependencies as there is no Docker support on MacOS atm
         PYTHON_EXECUTABLE=python \
@@ -393,7 +363,6 @@ jobs:
     strategy:
       matrix:
         dtype: [fp32]
-        build-tool: [cmake]
         mode: [qnn]
       fail-fast: false
     with:
@@ -408,14 +377,14 @@ jobs:
         conda activate "${CONDA_ENV}"
 
         DTYPE=${{ matrix.dtype }}
-        BUILD_TOOL=${{ matrix.build-tool }}
+        BUILD_TOOL="cmake"
         MODE=${{ matrix.mode }}
 
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-qnn-deps.sh
         PYTHON_EXECUTABLE=python bash .ci/scripts/build-qnn-sdk.sh
 
         # Setup executorch
-        PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh buck2
+        PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "${BUILD_TOOL}"
         # Install requirements for export_llama
         PYTHON_EXECUTABLE=python bash examples/models/llama/install_requirements.sh
         # Test llama2

.github/workflows/update-viablestrict.yml

@@ -20,7 +20,7 @@ jobs:
         with:
           repository: pytorch/executorch
           stable-branch: viable/strict
-          requires: '[\"pull\", \"lint\", \"trunk\", \"Build documentation\", \"^Android$\", \"^Apple$\"]'
+          requires: '[\"pull\", \"lint\", \"trunk\", \"Build documentation\", \"^Apple$\"]'
           secret-bot-token: ${{ secrets.UPDATEBOT_TOKEN }}
           clickhouse-url: ${{ secrets.CLICKHOUSE_URL }}
           clickhouse-username: ${{ secrets.CLICKHOUSE_VIABLESTRICT_USERNAME }}

README.md

@@ -25,7 +25,7 @@ Check out the [Getting Started](https://pytorch.org/executorch/stable/getting-st
 Check out the examples of [Llama](./examples/models/llama/README.md), [Llava](./examples/models/llava/README.md) and [other models](./examples/README.md) running on edge devices using ExecuTorch.
 
 
-**[UPDATE - 09/25]** We have added support for running [Llama 3.2 1B/3B](./examples/models/llama/README.md) models via ExecuTorch.
+**[UPDATE - 10/24]** We have added support for running [Llama 3.2 Quantized 1B/3B](./examples/models/llama/README.md) models via ExecuTorch.
 
 ## Feedback
 
@@ -92,7 +92,7 @@ tools.
 ├── runtime                         #  Core C++ runtime.
 |   ├── backend                     #  Backend delegate runtime APIs.
 |   ├── core                        #  Core structures used across all levels of the runtime.
-|   ├── executor                    #  Model loading, initalization, and execution.
+|   ├── executor                    #  Model loading, initialization, and execution.
 |   ├── kernel                      #  Kernel registration and management.
 |   ├── platform                    #  Layer between architecture specific code and portable C++.
 ├── schema                          #  ExecuTorch PTE file format flatbuffer