[fx][split] Copy node metadata for placeholders (#107981) (#109297)

- Follow-up to #107248 which copies metadata for placeholder nodes in the top-level FX graph
- Currently, top-level placeholders do not have their metadata copied over, causing loss of `TensorMetadata` in some `torch.compile` backends

Fixes https://github.com/pytorch/TensorRT/issues/2258
Pull Request resolved: https://github.com/pytorch/pytorch/pull/107981
Approved by: https://github.com/angelayi

Co-authored-by: gs-olive <113141689+gs-olive@users.noreply.github.com>

.ci/docker/ci_commit_pins/triton-rocm.txt

@@ -1 +1 @@
-05d67b9418cacda0d356c2102d7c1a887948b013
+34f8189eae57a23cc15b4b4f032fe25757e0db8e

.ci/docker/common/install_inductor_benchmark_deps.sh

@@ -7,18 +7,14 @@ source "$(dirname "${BASH_SOURCE[0]}")/common_utils.sh"
 function install_huggingface() {
   local version
   version=$(get_pinned_commit huggingface)
-  pip_install pandas
-  pip_install scipy
-  pip_install z3-solver
+  pip_install pandas==2.0.3
   pip_install "transformers==${version}"
 }
 
 function install_timm() {
   local commit
   commit=$(get_pinned_commit timm)
-  pip_install pandas
-  pip_install scipy
-  pip_install z3-solver
+  pip_install pandas==2.0.3
   pip_install "git+https://github.com/rwightman/pytorch-image-models@${commit}"
 }
 

.ci/docker/common/install_onnx.sh

@@ -4,6 +4,10 @@ set -ex
 
 source "$(dirname "${BASH_SOURCE[0]}")/common_utils.sh"
 
+retry () {
+    "$@" || (sleep 10 && "$@") || (sleep 20 && "$@") || (sleep 40 && "$@")
+}
+
 # A bunch of custom pip dependencies for ONNX
 pip_install \
   beartype==0.10.4 \
@@ -18,22 +22,17 @@ pip_install \
 # onnx-weekly. Otherwise, onnx-weekly could be
 # overwritten by onnx.
 pip_install \
-  onnxruntime==1.15.1 \
   parameterized==0.8.1 \
   pytest-cov==4.0.0 \
   pytest-subtests==0.10.0 \
   tabulate==0.9.0 \
   transformers==4.31.0
 
-# Using 1.15dev branch for the following not yet released features and fixes.
-# - Segfault fix for shape inference.
-# - Inliner to workaround ORT segfault.
-pip_install onnx-weekly==1.15.0.dev20230717
+pip_install coloredlogs packaging
+retry pip_install -i https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/ORT-Nightly/pypi/simple/ --no-cache-dir --no-input ort-nightly==1.16.0.dev20230908001
 
-# TODO: change this when onnx-script is on testPypi
-# pip_install onnxscript-preview==0.1.0.dev20230809 --no-deps
-# NOTE: temp change for CI to run on unpublished onnxscript PR.
-pip_install "onnxscript@git+https://github.com/microsoft/onnxscript@f69be19ebd3f2e0d7efe64b0c7be3329cbab3822" --no-deps
+pip_install onnx==1.14.1
+pip_install onnxscript-preview==0.1.0.dev20230828 --no-deps
 
 # Cache the transformers model to be used later by ONNX tests. We need to run the transformers
 # package to download the model. By default, the model is cached at ~/.cache/huggingface/hub/

.ci/docker/requirements-ci.txt

@@ -271,7 +271,7 @@ pytest-cpp==2.3.0
 #Pinned versions: 2.3.0
 #test that import:
 
-z3-solver
+z3-solver==4.12.2.0
 #Description: The Z3 Theorem Prover Project
 #Pinned versions:
 #test that import:

.ci/pytorch/common_utils.sh

@@ -180,7 +180,7 @@ function install_numpy_pytorch_interop() {
 
 function clone_pytorch_xla() {
   if [[ ! -d ./xla ]]; then
-    git clone --recursive --quiet https://github.com/pytorch/xla.git
+    git clone --recursive -b r2.1 https://github.com/pytorch/xla.git
     pushd xla
     # pin the xla hash so that we don't get broken by changes to xla
     git checkout "$(cat ../.github/ci_commit_pins/xla.txt)"

.circleci/scripts/binary_checkout.sh

@@ -62,7 +62,7 @@ git --no-pager log --max-count 1
 popd
 
 # Clone the Builder main repo
-retry git clone -q https://github.com/pytorch/builder.git "$BUILDER_ROOT"
+retry git clone -q https://github.com/pytorch/builder.git -b release/2.1 "$BUILDER_ROOT"
 pushd "$BUILDER_ROOT"
 echo "Using builder from "
 git --no-pager log --max-count 1

.circleci/scripts/binary_linux_test.sh

@@ -90,7 +90,7 @@ if [[ "$PACKAGE_TYPE" == conda ]]; then
       if [[ "\${TORCH_CONDA_BUILD_FOLDER}" == "pytorch-nightly" ]]; then
               PYTORCH_CHANNEL="pytorch-nightly"
       fi
-      retry conda install \${EXTRA_CONDA_FLAGS} -yq -c nvidia -c "\${PYTORCH_CHANNEL}" "pytorch-cuda=\${cu_ver}"
+      retry conda install \${EXTRA_CONDA_FLAGS} -yq -c nvidia -c pytorch-test "pytorch-cuda=\${cu_ver}"
     fi
     conda install \${EXTRA_CONDA_FLAGS} -y "\$pkg" --offline
   )
@@ -98,9 +98,9 @@ elif [[ "$PACKAGE_TYPE" != libtorch ]]; then
   if [[ "$(uname -m)" == aarch64 ]]; then
     # Using "extra-index-url" until all needed aarch64 dependencies are
     # added to "https://download.pytorch.org/whl/nightly/"
-    pip install "\$pkg" --extra-index-url "https://download.pytorch.org/whl/nightly/${DESIRED_CUDA}"
+    pip install "\$pkg" --extra-index-url "https://download.pytorch.org/whl/test/${DESIRED_CUDA}"
   else
-    pip install "\$pkg" --index-url "https://download.pytorch.org/whl/nightly/${DESIRED_CUDA}"
+    pip install "\$pkg" --index-url "https://download.pytorch.org/whl/test/${DESIRED_CUDA}"
   fi
   retry pip install -q numpy protobuf typing-extensions
 fi

.circleci/scripts/binary_upload.sh

@@ -11,7 +11,7 @@ PKG_DIR=${PKG_DIR:-/tmp/workspace/final_pkgs}
 # currently set within `designate_upload_channel`
 UPLOAD_CHANNEL=${UPLOAD_CHANNEL:-nightly}
 # Designates what subfolder to put packages into
-UPLOAD_SUBFOLDER=${UPLOAD_SUBFOLDER:-cpu}
+UPLOAD_SUBFOLDER=${UPLOAD_SUBFOLDER:-}
 UPLOAD_BUCKET="s3://pytorch"
 BACKUP_BUCKET="s3://pytorch-backup"
 BUILD_NAME=${BUILD_NAME:-}
@@ -64,12 +64,17 @@ s3_upload() {
   local pkg_type
   extension="$1"
   pkg_type="$2"
-  s3_dir="${UPLOAD_BUCKET}/${pkg_type}/${UPLOAD_CHANNEL}/${UPLOAD_SUBFOLDER}/"
+  s3_root_dir="${UPLOAD_BUCKET}/${pkg_type}/${UPLOAD_CHANNEL}"
+  if [[ -z ${UPLOAD_SUBFOLDER:-} ]]; then
+    s3_upload_dir="${s3_root_dir}/"
+  else
+    s3_upload_dir="${s3_root_dir}/${UPLOAD_SUBFOLDER}/"
+  fi
   (
     for pkg in ${PKG_DIR}/*.${extension}; do
       (
         set -x
-        ${AWS_S3_CP} --no-progress --acl public-read "${pkg}" "${s3_dir}"
+        ${AWS_S3_CP} --no-progress --acl public-read "${pkg}" "${s3_upload_dir}"
       )
     done
   )
@@ -82,15 +87,17 @@ pip install -q awscli
 case "${PACKAGE_TYPE}" in
   conda)
     conda_upload
-    # Fetch  platform (eg. win-64, linux-64, etc.) from index file
-    # Because there's no actual conda command to read this
-    subdir=$(\
-      tar -xOf ${PKG_DIR}/*.bz2 info/index.json \
-        | grep subdir  \
-        | cut -d ':' -f2 \
-        | sed -e 's/[[:space:]]//' -e 's/"//g' -e 's/,//' \
-    )
-    BACKUP_DIR="conda/${subdir}"
+    for conda_archive in ${PKG_DIR}/*.tar.bz2; do
+      # Fetch  platform (eg. win-64, linux-64, etc.) from index file because
+      # there's no actual conda command to read this
+      subdir=$(\
+        tar -xOf "${conda_archive}" info/index.json \
+          | grep subdir  \
+          | cut -d ':' -f2 \
+          | sed -e 's/[[:space:]]//' -e 's/"//g' -e 's/,//' \
+      )
+      BACKUP_DIR="conda/${subdir}"
+    done
     ;;
   libtorch)
     s3_upload "zip" "libtorch"

.github/ci_commit_pins/xla.txt

@@ -1 +1 @@
-e1ee592d9806216d7ac0bb711cae6307b0c5b68a
+r2.1

.github/merge_rules.yaml

@@ -7,6 +7,7 @@
   - docs/source/onnx.rst
   - docs/source/onnx*
   - docs/source/scripts/onnx/**
+  - docs/source/_static/img/onnx/**
   - scripts/onnx/**
   - test/onnx/**
   - tools/onnx/**

.github/scripts/build_triton_wheel.py

@@ -60,12 +60,18 @@ def build_triton(
     build_conda: bool = False,
     build_rocm: bool = False,
     py_version: Optional[str] = None,
+    release: bool = False,
 ) -> Path:
     env = os.environ.copy()
     if "MAX_JOBS" not in env:
         max_jobs = os.cpu_count() or 1
         env["MAX_JOBS"] = str(max_jobs)
 
+    if not release:
+        # Nightly binaries include the triton commit hash, i.e. 2.1.0+e6216047b8
+        # while release build should only include the version, i.e. 2.1.0
+        version = f"{version}+{commit_hash[:10]}"
+
     with TemporaryDirectory() as tmpdir:
         triton_basedir = Path(tmpdir) / "triton"
         triton_pythondir = triton_basedir / "python"
@@ -80,7 +86,7 @@ def build_triton(
         if build_conda:
             with open(triton_basedir / "meta.yaml", "w") as meta:
                 print(
-                    f"package:\n  name: torchtriton\n  version: {version}+{commit_hash[:10]}\n",
+                    f"package:\n  name: torchtriton\n  version: {version}\n",
                     file=meta,
                 )
                 print("source:\n  path: .\n", file=meta)
@@ -103,7 +109,7 @@ def build_triton(
 
             patch_init_py(
                 triton_pythondir / "triton" / "__init__.py",
-                version=f"{version}+{commit_hash[:10]}",
+                version=f"{version}",
             )
             if py_version is None:
                 py_version = f"{sys.version_info.major}.{sys.version_info.minor}"
@@ -129,11 +135,11 @@ def build_triton(
         patch_setup_py(
             triton_pythondir / "setup.py",
             name=triton_pkg_name,
-            version=f"{version}+{commit_hash[:10]}",
+            version=f"{version}",
         )
         patch_init_py(
             triton_pythondir / "triton" / "__init__.py",
-            version=f"{version}+{commit_hash[:10]}",
+            version=f"{version}",
         )
 
         if build_rocm:
@@ -157,12 +163,14 @@ def main() -> None:
     from argparse import ArgumentParser
 
     parser = ArgumentParser("Build Triton binaries")
+    parser.add_argument("--release", action="store_true")
     parser.add_argument("--build-conda", action="store_true")
     parser.add_argument("--build-rocm", action="store_true")
     parser.add_argument("--py-version", type=str)
     parser.add_argument("--commit-hash", type=str)
     parser.add_argument("--triton-version", type=str, default=read_triton_version())
     args = parser.parse_args()
+
     build_triton(
         build_rocm=args.build_rocm,
         commit_hash=args.commit_hash
@@ -171,6 +179,7 @@ def main() -> None:
         version=args.triton_version,
         build_conda=args.build_conda,
         py_version=args.py_version,
+        release=args.release,
     )
 
 

.github/scripts/generate_binary_build_matrix.py

@@ -248,7 +248,8 @@ def generate_wheels_matrix(
                         "nvidia-cusolver-cu12==11.4.5.107; platform_system == 'Linux' and platform_machine == 'x86_64' | "
                         "nvidia-cusparse-cu12==12.1.0.106; platform_system == 'Linux' and platform_machine == 'x86_64' | "
                         "nvidia-nccl-cu12==2.18.1; platform_system == 'Linux' and platform_machine == 'x86_64' | "
-                        "nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64'",
+                        "nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+                        "triton==2.1.0; platform_system == 'Linux' and platform_machine == 'x86_64'",
                         "build_name": f"{package_type}-py{python_version}-{gpu_arch_type}{gpu_arch_version}-with-pypi-cudnn".replace(  # noqa: B950
                             ".", "_"
                         ),

.github/templates/common.yml.j2

@@ -8,7 +8,7 @@
 # NOTE: If testing pytorch/builder changes you can change this variable to change what pytorch/builder reference
 #       the binary builds will check out
 {%- set builder_repo = "pytorch/builder" -%}
-{%- set builder_branch = "main" -%}
+{%- set builder_branch = "release/2.1" -%}
 
 {%- macro concurrency(build_environment) -%}
 concurrency:

.github/templates/linux_binary_build_workflow.yml.j2

@@ -97,13 +97,13 @@ jobs:
         with:
           name: !{{ config["build_name"] }}
           path: "${{ runner.temp }}/artifacts/"
-      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
-      !{{ common.checkout(deep_clone=False, directory="builder", repository=common.builder_repo, branch=common.builder_branch) }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
+      !{{ common.checkout(deep_clone=False, directory="builder", repository=common.builder_repo, branch=common.builder_branch, checkout_pr_head=False) }}
       - name: ROCm set GPU_FLAG
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: !{{ config["container_image"] }}
       - name: Test Pytorch binary

.github/templates/macos_binary_build_workflow.yml.j2

@@ -74,8 +74,8 @@ jobs:
           /bin/bash "${RUNNER_TEMP}/conda.sh" -b -p "${RUNNER_TEMP}/anaconda"
           echo "${RUNNER_TEMP}/anaconda/bin" >> "${GITHUB_PATH}"
           echo "DEVELOPER_DIR=/Applications/Xcode_13.3.1.app/Contents/Developer" >> "${GITHUB_ENV}"
-      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
-      !{{ common.checkout(deep_clone=False, directory="builder", repository=common.builder_repo, branch=common.builder_branch) }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
+      !{{ common.checkout(deep_clone=False, directory="builder", repository=common.builder_repo, branch=common.builder_branch, checkout_pr_head=False) }}
       - name: Install sccache (only for non-forked PRs, and pushes to trunk)
         uses: nick-fields/retry@v2.8.2
         if: ${{ github.event_name == 'push' || github.event.pull_request.head.repo.full_name == github.repository }}

.github/templates/upload.yml.j2

@@ -67,6 +67,6 @@
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 {%- endmacro %}

.github/templates/windows_binary_build_workflow.yml.j2

@@ -62,8 +62,8 @@ jobs:
     steps:
       !{{ common.setup_ec2_windows() }}
       !{{ set_runner_specific_vars() }}
-      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
-      !{{ common.checkout(deep_clone=False, directory="builder", repository=common.builder_repo, branch=common.builder_branch) }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
+      !{{ common.checkout(deep_clone=False, directory="builder", repository=common.builder_repo, branch=common.builder_branch, checkout_pr_head=False) }}
       - name: Populate binary env
         shell: bash
         run: |
@@ -102,8 +102,8 @@ jobs:
         with:
           name: !{{ config["build_name"] }}
           path: "${{ env.PYTORCH_FINAL_PACKAGE_DIR }}"
-      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
-      !{{ common.checkout(deep_clone=False, directory="builder", repository=common.builder_repo, branch=common.builder_branch) }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
+      !{{ common.checkout(deep_clone=False, directory="builder", repository=common.builder_repo, branch=common.builder_branch, checkout_pr_head=False) }}
       - name: Populate binary env
         shell: bash
         run: |

.github/workflows/_android-build-test.yml

@@ -36,7 +36,7 @@ jobs:
       keep-going: ${{ steps.filter.outputs.keep-going }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           fetch-depth: 1
           submodules: false
@@ -58,25 +58,25 @@ jobs:
     runs-on: ${{ matrix.runner }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Setup Linux
         uses: ./.github/actions/setup-linux
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.1
         with:
           docker-image-name: ${{ inputs.docker-image-name }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -140,5 +140,5 @@ jobs:
         if: always()
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
         if: always()

.github/workflows/_android-full-build-test.yml

@@ -36,7 +36,7 @@ jobs:
       keep-going: ${{ steps.filter.outputs.keep-going }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           fetch-depth: 1
           submodules: false
@@ -58,25 +58,25 @@ jobs:
     runs-on: ${{ matrix.runner }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Setup Linux
         uses: ./.github/actions/setup-linux
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.1
         with:
           docker-image-name: ${{ inputs.docker-image-name }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -185,5 +185,5 @@ jobs:
         if: always()
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
         if: always()

.github/workflows/_bazel-build-test.yml

@@ -41,7 +41,7 @@ jobs:
       reenabled-issues: ${{ steps.filter.outputs.reenabled-issues }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           fetch-depth: 1
           submodules: false
@@ -63,30 +63,30 @@ jobs:
     runs-on: ${{ matrix.runner }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Setup Linux
         uses: ./.github/actions/setup-linux
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.1
         with:
           docker-image-name: ${{ inputs.docker-image-name }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
       - name: Install nvidia driver, nvidia-docker runtime, set GPU_FLAG
-        uses: pytorch/test-infra/.github/actions/setup-nvidia@main
+        uses: pytorch/test-infra/.github/actions/setup-nvidia@release/2.1
         if: ${{ inputs.cuda-version != 'cpu' }}
 
       - name: Output disk space left
@@ -197,5 +197,5 @@ jobs:
           file-suffix: bazel-${{ github.job }}_${{ steps.get-job-id.outputs.job-id }}
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
         if: always()

.github/workflows/_binary-build-linux.yml

@@ -139,12 +139,12 @@ jobs:
         run: env
 
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.github-token }}
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           no-sudo: ${{ inputs.build_environment == 'linux-aarch64-binary-manywheel' }}
 
@@ -159,10 +159,12 @@ jobs:
       - name: Clean workspace
         shell: bash
         run: |
+          set -eux
+
           rm -rf "${GITHUB_WORKSPACE}"
           mkdir "${GITHUB_WORKSPACE}"
 
-          if [[ inputs.build_environment == 'linux-aarch64-binary-manywheel' ]]; then
+          if [[ ${{ inputs.build_environment }} == 'linux-aarch64-binary-manywheel' ]]; then
             rm -rf "${RUNNER_TEMP}/artifacts"
             mkdir "${RUNNER_TEMP}/artifacts"
           fi
@@ -170,7 +172,6 @@ jobs:
       - name: Checkout PyTorch to pytorch dir
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -184,7 +185,7 @@ jobs:
       - name: Checkout pytorch/builder to builder dir
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -210,7 +211,7 @@ jobs:
 
       - name: Pull Docker image
         if: ${{ steps.filter.outputs.is-test-matrix-empty == 'False' }}
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ inputs.DOCKER_IMAGE }}
 
@@ -267,7 +268,7 @@ jobs:
 
       - name: Teardown Linux
         if: always()
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
 
       - name: Chown workspace
         if: always()

.github/workflows/_binary-test-linux.yml

@@ -127,13 +127,13 @@ jobs:
           } >> "${GITHUB_ENV} }}"
 
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.github-token }}
 
         # Setup the environment
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           no-sudo: ${{ inputs.build_environment == 'linux-aarch64-binary-manywheel' }}
 
@@ -154,7 +154,6 @@ jobs:
       - name: Checkout PyTorch to pytorch dir
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
 
@@ -167,7 +166,7 @@ jobs:
       - name: Checkout pytorch/builder to builder dir
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -198,12 +197,12 @@ jobs:
           path: "${{ runner.temp }}/artifacts/"
 
       - name: Install nvidia driver, nvidia-docker runtime, set GPU_FLAG
-        uses: pytorch/test-infra/.github/actions/setup-nvidia@main
+        uses: pytorch/test-infra/.github/actions/setup-nvidia@release/2.1
         if: ${{ inputs.GPU_ARCH_TYPE == 'cuda' && steps.filter.outputs.is-test-matrix-empty == 'False' }}
 
       - name: Pull Docker image
         if: ${{ steps.filter.outputs.is-test-matrix-empty == 'False' }}
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ inputs.DOCKER_IMAGE }}
 
@@ -213,7 +212,7 @@ jobs:
 
       - name: Teardown Linux
         if: always()
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
 
       - name: Chown workspace
         if: always()

.github/workflows/_binary-upload.yml

@@ -97,7 +97,7 @@ jobs:
       SHA1: ${{ github.event.pull_request.head.sha || github.sha }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           no-sudo: true
 
@@ -121,7 +121,7 @@ jobs:
         shell: bash -e -l {0}
         run: |
           # reference ends with an RC suffix
-          if [[ ${GITHUB_REF_NAME} = *-rc[0-9]* ]]; then
+          if [[ "${GITHUB_REF_NAME}" = *-rc[0-9]* ]]; then
             echo "UPLOAD_CHANNEL=test" >> "$GITHUB_ENV"
           fi
 

.github/workflows/_buck-build-test.yml

@@ -22,7 +22,7 @@ jobs:
       keep-going: ${{ steps.filter.outputs.keep-going }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           fetch-depth: 1
           submodules: false
@@ -43,7 +43,7 @@ jobs:
     runs-on: ${{ matrix.runner }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Set up JDK 8
         uses: actions/setup-java@v3
@@ -52,7 +52,7 @@ jobs:
           distribution: 'temurin'
 
       - name: Setup miniconda
-        uses: pytorch/test-infra/.github/actions/setup-miniconda@main
+        uses: pytorch/test-infra/.github/actions/setup-miniconda@release/2.1
         with:
           python-version: 3.8
           environment-file: .github/requirements/conda-env-${{ runner.os }}-${{ runner.arch }}

.github/workflows/_docs.yml

@@ -66,7 +66,7 @@ jobs:
     name: build-docs-${{ matrix.docs_type }}-${{ inputs.push }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
           instructions: |
@@ -77,19 +77,19 @@ jobs:
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Setup Linux
         uses: ./.github/actions/setup-linux
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.1
         with:
           docker-image-name: ${{ inputs.docker-image }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -187,5 +187,5 @@ jobs:
           s3-prefix: pytorch/pytorch/${{ github.event.pull_request.number }}/functorchdocs
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
         if: always()

.github/workflows/_ios-build-test.yml

@@ -7,14 +7,6 @@ on:
         required: true
         type: string
         description: Top-level label for what's being built/tested.
-      ios-platform:
-        required: true
-        type: string
-        description: Which iOS platform to build for.
-      ios-arch:
-        required: true
-        type: string
-        description: Which iOS arch to build for.
       sync-tag:
         required: false
         type: string
@@ -31,8 +23,6 @@ on:
 env:
   GIT_DEFAULT_BRANCH: ${{ github.event.repository.default_branch }}
   BUILD_ENVIRONMENT: ${{ inputs.build-environment }}
-  IOS_PLATFORM: ${{ inputs.ios-platform }}
-  IOS_ARCH: ${{ inputs.ios-arch }}
 
 jobs:
   filter:
@@ -43,7 +33,7 @@ jobs:
       keep-going: ${{ steps.filter.outputs.keep-going }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           fetch-depth: 1
           submodules: false
@@ -63,33 +53,30 @@ jobs:
       matrix: ${{ fromJSON(needs.filter.outputs.test-matrix) }}
       fail-fast: false
     runs-on: ${{ matrix.runner }}
+    env:
+      IOS_PLATFORM: ${{ matrix.ios_platform }}
+      IOS_ARCH: ${{ matrix.ios_arch }}
+      BUILD_LITE_INTERPRETER: ${{ matrix.use_lite_interpreter }}
+      USE_PYTORCH_METAL: ${{ matrix.use_metal }}
+      USE_COREML_DELEGATE: ${{ matrix.use_coreml }}
+      CUSTOM_OP_LIST: ${{ matrix.use_custom_op_list }}
+      # TODO: Bump it to 2.2.0 after cherry pick this or figure out a better way
+      # to get this version instead of hard coding it here
+      PYTORCH_VERSION: 2.1.0
     timeout-minutes: 240
     steps:
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Populate CI build options
+        shell: bash
         run: |
-          # Most builds use the lite interpreter, if certain builds shouldn't
-          # build the lite interpreter this env variable should get over-written
-          # in the following case statement
-          echo "BUILD_LITE_INTERPRETER=1" >> "${GITHUB_ENV}"
+          set -ex
 
-          case ${BUILD_ENVIRONMENT} in
-            *metal*)
-              echo "USE_PYTORCH_METAL=1" >> "${GITHUB_ENV}"
-              ;;
-            *full_jit*)
-              echo "BUILD_LITE_INTERPRETER=0" >> "${GITHUB_ENV}"
-              ;;
-            *custom*)
-              echo "SELECTED_OP_LIST=${GITHUB_WORKSPACE}/ios/TestApp/custom_build/mobilenetv2.yaml" >> "${GITHUB_ENV}"
-              ;;
-            *coreml*)
-              echo "USE_COREML_DELEGATE=1" >> "${GITHUB_ENV}"
-              ;;
-          esac
+          if [ -n "${CUSTOM_OP_LIST:-}" ]; then
+            echo "SELECTED_OP_LIST=${GITHUB_WORKSPACE}/ios/TestApp/custom_build/${CUSTOM_OP_LIST}" >> "${GITHUB_ENV}"
+          fi
 
       - name: Install brew dependencies
         uses: nick-fields/retry@v2.8.2
@@ -102,7 +89,7 @@ jobs:
             brew install libtool
 
       - name: Setup miniconda for iOS
-        uses: pytorch/test-infra/.github/actions/setup-miniconda@main
+        uses: pytorch/test-infra/.github/actions/setup-miniconda@release/2.1
         with:
           python-version: "3.9"
           environment-file: .github/requirements/conda-env-iOS
@@ -116,54 +103,67 @@ jobs:
           retry_wait_seconds: 90
           command: |
             set -x
-            cd ios/TestApp
-            # install fastlane
+
+            pushd ios/TestApp
+            # Install fastlane
             sudo gem install bundler && bundle install
             bundle update fastlane
+            popd
 
-      - name: Build PyTorch Mobile Runtime
+      - name: Build PyTorch mobile runtime
+        shell: bash
         run: |
+          set -eux
           # shellcheck disable=SC1091
           export TCLLIBPATH="/usr/local/lib"
-          python -VV
           ${CONDA_RUN} scripts/build_ios.sh
 
       - name: Build TestApp
-        if: inputs.ios-platform == 'SIMULATOR'
+        if: matrix.ios_platform == 'SIMULATOR'
         timeout-minutes: 15
         run: |
-          # run the ruby build script
+          # Run the ruby build script
           if ! [ -x "$(command -v xcodebuild)" ]; then
             echo 'Error: xcodebuild is not installed.'
             exit 1
           fi
           ruby scripts/xcode_build.rb -i build_ios/install -x ios/TestApp/TestApp.xcodeproj -p "${IOS_PLATFORM}"
 
-      - name: Run Simulator Tests
-        if: inputs.ios-platform == 'SIMULATOR'
+      - name: Run simulator tests
+        if: matrix.ios_platform == 'SIMULATOR'
+        shell: bash
         run: |
+          set -eux
           # shellcheck disable=SC1091
-          # use the pytorch nightly build to generate models
-          ${CONDA_RUN} pip3 install --pre torch torchvision torchaudio -f https://download.pytorch.org/whl/nightly/cpu/torch_nightly.html
-          # generate models for differnet backends
-          cd "${GITHUB_WORKSPACE}/ios/TestApp/benchmark"
+          # Use the pytorch nightly build to generate models
+          ${CONDA_RUN} pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cpu
+
+          # Generate models for differnet backends
+          pushd "${GITHUB_WORKSPACE}/ios/TestApp/benchmark"
           mkdir -p ../models
+          # NB: Both of the following scripts only export models with lite interpreter
           if [ "${USE_COREML_DELEGATE}" == 1 ]; then
             ${CONDA_RUN} python coreml_backend.py
           else
-            cd "${GITHUB_WORKSPACE}"
+            pushd "${GITHUB_WORKSPACE}"
             ${CONDA_RUN} python test/mobile/model_test/gen_test_model.py ios-test
+            popd
           fi
-          cd "${GITHUB_WORKSPACE}/ios/TestApp/benchmark"
+
           if [ "${BUILD_LITE_INTERPRETER}" == 1 ]; then
             echo "Setting up the TestApp for LiteInterpreter"
             ruby setup.rb --lite 1
           else
+            # Generate some models for JIT without lite interpreter
+            ${CONDA_RUN} python trace_model.py
+
             echo "Setting up the TestApp for Full JIT"
             ruby setup.rb
           fi
-          cd "${GITHUB_WORKSPACE}/ios/TestApp"
-          # instruments -s -devices
+          popd
+
+          pushd "${GITHUB_WORKSPACE}/ios/TestApp"
+          # Instruments -s -devices
           if [ "${BUILD_LITE_INTERPRETER}" == 1 ]; then
             if [ "${USE_COREML_DELEGATE}" == 1 ]; then
               bundle exec fastlane scan --only_testing TestAppTests/TestAppTests/testCoreML
@@ -173,9 +173,282 @@ jobs:
           else
             bundle exec fastlane scan --only_testing TestAppTests/TestAppTests/testFullJIT
           fi
+          popd
 
-      - name: Dump Simulator Tests On a Failure
-        if: failure() && inputs.ios-platform == 'SIMULATOR'
+      - name: Dump simulator tests on failure
+        if: failure() && matrix.ios_platform == 'SIMULATOR'
         run: |
           echo "Simulator Tests Logs:"
           cat /Users/runner/Library/Logs/scan/*.log
+
+      - name: Prepare the build artifacts for upload
+        shell: bash
+        run: |
+          set -eux
+
+          # The structure of the folder is as follows:
+          #
+          # RUNNER_TEMP/
+          # └── IOS_ARCH/
+          #     ├── LICENSE
+          #     ├── install
+          #     │   ├── include
+          #     │   │   └── headers
+          #     │   └── lib
+          #     │       ├── libXNNPACK.a
+          #     │       ├── libc10.a
+          #     │       ├── libclog.a
+          #     │       ├── libcpuinfo.a
+          #     │       ├── libeigen_blas.a
+          #     │       ├── libpthreadpool.a
+          #     │       ├── libpytorch_qnnpack.a
+          #     │       ├── libtorch.a
+          #     │       └── libtorch_cpu.a
+          #     ├── src
+          #     │   └── LibTorch-Lite.h
+          #     └── version.txt
+          SETUP_DIR="${RUNNER_TEMP}/${IOS_ARCH}"
+          mkdir -p "${SETUP_DIR}/src"
+
+          cp -R "${GITHUB_WORKSPACE}/build_ios/install" "${SETUP_DIR}"
+          # Copy the umbrella header and license
+          if [ "${BUILD_LITE_INTERPRETER}" == 1 ]; then
+            cp "${GITHUB_WORKSPACE}/ios/LibTorch-Lite.h" "${SETUP_DIR}/src"
+          else
+            cp "${GITHUB_WORKSPACE}/ios/LibTorch.h" "${SETUP_DIR}/src"
+          fi
+
+          # Copy license and version
+          cp "${GITHUB_WORKSPACE}/LICENSE" "${SETUP_DIR}"
+          echo "${PYTORCH_VERSION}" > "${SETUP_DIR}"/version.txt
+
+          # Save the podspec for the upload job later
+          if [ "${BUILD_LITE_INTERPRETER}" == "1" ]; then
+            DATE=$(date -u +%Y%m%d)
+            cp "${GITHUB_WORKSPACE}"/ios/LibTorch-Lite-Nightly.podspec.template "${SETUP_DIR}"/LibTorch-Lite-Nightly.podspec
+            sed -i '' -e "s/IOS_NIGHTLY_BUILD_VERSION/${PYTORCH_VERSION}.${DATE}/g" "${SETUP_DIR}"/LibTorch-Lite-Nightly.podspec
+
+            cp "${GITHUB_WORKSPACE}"/ios/LibTorch-Lite.podspec.template "${SETUP_DIR}"/LibTorch-Lite.podspec
+            sed -i '' -e "s/IOS_BUILD_VERSION/${PYTORCH_VERSION}/g" "${SETUP_DIR}"/LibTorch-Lite.podspec
+          else
+            # NB: There is no nightly build without lite interpreter atm
+            cp "${GITHUB_WORKSPACE}"/ios/LibTorch.podspec.template "${SETUP_DIR}"/LibTorch.podspec
+            sed -i '' -e "s/IOS_BUILD_VERSION/${PYTORCH_VERSION}/g" "${SETUP_DIR}"/LibTorch.podspec
+          fi
+
+          pushd "${SETUP_DIR}"
+          # NB: It's important to zip all the files before uploading because the GHA will upload
+          # all files sequentially which is both slow and has too many requests. More info is at
+          # https://github.com/actions/upload-artifact#too-many-uploads-resulting-in-429-responses
+          zip -r "${IOS_ARCH}.zip" install src version.txt LICENSE ./*.podspec
+          popd
+
+      - uses: actions/upload-artifact@v3
+        with:
+          name: pytorch-ios-build-artifacts-${{ matrix.ios_arch }}
+          if-no-files-found: error
+          path: ${{ runner.temp }}/${{ matrix.ios_arch }}/${{ matrix.ios_arch }}.zip
+
+  upload-ios-artifacts:
+    # NB: this job run on GitHub MacOS ephemeral runner so that it can use lipo
+    # to create the fat iOS binaries for both x86_64 and arm64
+    runs-on: macos-12
+    needs: build
+    # NB: Only upload release build, if we need it, we could also turn on nightly here
+    environment: ${{ (github.event_name == 'push' && (github.event.ref == 'refs/heads/nightly' || startsWith(github.event.ref, 'refs/tags/v'))) && 'ios-upload' || '' }}
+    steps:
+      - uses: actions/checkout@v3
+
+      # For awscli S3 upload
+      - uses: actions/setup-python@v4
+        with:
+          python-version: '3.10'
+          cache: pip
+
+      # For cocoapods pod upload
+      - uses: ruby/setup-ruby@v1
+        with:
+          ruby-version: '3.2'
+          bundler-cache: true
+
+      - name: Download arm64 artifacts
+        uses: actions/download-artifact@v3
+        with:
+          name: pytorch-ios-build-artifacts-arm64
+
+      - name: Download x86_64 artifacts
+        uses: actions/download-artifact@v3
+        with:
+          name: pytorch-ios-build-artifacts-x86_64
+
+      - name: Unzip arm64 and x86_64 artifacts
+        shell: bash
+        run: |
+          set -eux
+
+          for ARCH in "arm64" "x86_64"; do
+            TMP_DIR="${RUNNER_TEMP}/${ARCH}"
+            mkdir -p "${TMP_DIR}"
+
+            cp "${ARCH}.zip" "${TMP_DIR}"
+
+            pushd "${TMP_DIR}"
+            unzip -o "${ARCH}.zip"
+            popd
+          done
+
+      - name: Prepare the artifact
+        env:
+          IS_NIGHTLY: ${{ github.event.ref == 'refs/heads/nightly' }}
+        shell: bash
+        working-directory: ${{ runner.temp }}/arm64
+        run: |
+          set -eux
+
+          DEST_DIR="${RUNNER_TEMP}"/ios
+          echo "DEST_DIR=${DEST_DIR}" >> "$GITHUB_ENV"
+
+          # Prepare all the sub directories
+          mkdir -p "${DEST_DIR}"/install/lib
+
+          # Copy header and share files, arm64 or x86_64 both work
+          cp -R install/include "${DEST_DIR}"/install
+          cp -R install/share "${DEST_DIR}"/install
+          # The last dash is important to copy only files under src
+          cp -R src "${DEST_DIR}"
+          cp LICENSE "${DEST_DIR}"
+
+          if [ "${IS_NIGHTLY}" == true ]; then
+            PYTORCH_VERSION=$(cat version.txt)
+            DATE=$(date -u +%Y%m%d)
+            echo "${PYTORCH_VERSION}.${DATE}" > "${DEST_DIR}"/version.txt
+          else
+            cp version.txt "${DEST_DIR}"
+          fi
+          PYTORCH_VERSION=$(cat "${DEST_DIR}"/version.txt)
+          echo "PYTORCH_VERSION=${PYTORCH_VERSION}" >> "$GITHUB_ENV"
+
+          pushd install/lib
+          # shellcheck disable=SC2207
+          LIBRARIES=($(ls ./*.a))
+          popd
+
+          for LIB in "${LIBRARIES[@]}"; do
+            FROM_LIBS=("${RUNNER_TEMP}"/arm64/install/lib/"${LIB}" "${RUNNER_TEMP}"/x86_64/install/lib/"${LIB}")
+            # Create a fat binary for both arm64 and x86_64
+            lipo -create "${FROM_LIBS[@]}" -o "${DEST_DIR}"/install/lib/"${LIB}"
+            # Print the info
+            lipo -i "${DEST_DIR}"/install/lib/"${LIB}"
+          done
+
+          BUILD_LITE_INTERPRETER=1
+          if [ -f "${RUNNER_TEMP}"/arm64/LibTorch.podspec ]; then
+            # If LibTorch.podspec is used instead of LibTorch-Lite.podspec, the artifact is built
+            # without lite interpreter
+            BUILD_LITE_INTERPRETER=0
+          fi
+          echo "BUILD_LITE_INTERPRETER=${BUILD_LITE_INTERPRETER}" >> "$GITHUB_ENV"
+
+      - name: Prepare the podspec
+        env:
+          IS_NIGHTLY: ${{ github.event.ref == 'refs/heads/nightly' }}
+        shell: bash
+        working-directory: ${{ env.DEST_DIR }}
+        run: |
+          set -eux
+
+          ARTIFACT_NAME=libtorch
+          SPEC_NAME=LibTorch
+
+          if [ "${BUILD_LITE_INTERPRETER}" == "1" ]; then
+            ARTIFACT_NAME="${ARTIFACT_NAME}_lite_ios"
+            SPEC_NAME="${SPEC_NAME}-Lite"
+          else
+            ARTIFACT_NAME="${ARTIFACT_NAME}_ios"
+          fi
+
+          if [ "${IS_NIGHTLY}" == true ]; then
+            ARTIFACT_NAME="${ARTIFACT_NAME}_nightly_${PYTORCH_VERSION}.zip"
+            SPEC_NAME="${SPEC_NAME}-Nightly"
+          else
+            ARTIFACT_NAME="${ARTIFACT_NAME}_${PYTORCH_VERSION}.zip"
+          fi
+
+          SPEC_NAME_WITH_VERSION="${SPEC_NAME}-${PYTORCH_VERSION}.podspec"
+          SPEC_NAME="${SPEC_NAME}.podspec"
+
+          # Also copy the spec file
+          cp "${RUNNER_TEMP}"/arm64/"${SPEC_NAME}" "${SPEC_NAME_WITH_VERSION}"
+
+          # NB: It's important to zip all the files before uploading because the GHA will upload
+          # all files sequentially which is both slow and has too many requests. More info is at
+          # https://github.com/actions/upload-artifact#too-many-uploads-resulting-in-429-responses
+          zip -r "${ARTIFACT_NAME}" install src version.txt LICENSE
+
+          {
+            echo "ARTIFACT_NAME=${ARTIFACT_NAME}"
+            echo "SPEC_NAME_WITH_VERSION=${SPEC_NAME_WITH_VERSION}"
+            echo "SPEC_NAME=${SPEC_NAME}"
+          } >> "$GITHUB_ENV"
+
+      - uses: actions/upload-artifact@v3
+        with:
+          name: pytorch-ios-artifacts
+          if-no-files-found: error
+          path: ${{ env.DEST_DIR }}/${{ env.ARTIFACT_NAME }}
+
+      - uses: actions/upload-artifact@v3
+        with:
+          name: pytorch-ios-podspec
+          if-no-files-found: error
+          path: ${{ env.DEST_DIR }}/${{ env.SPEC_NAME_WITH_VERSION }}
+
+      - name: Set DRY_RUN
+        if: ${{ github.event_name == 'push' && (github.event.ref == 'refs/heads/nightly' || (startsWith(github.event.ref, 'refs/tags/v'))) }}
+        shell: bash
+        run: |
+          echo "DRY_RUN=disabled" >> "$GITHUB_ENV"
+
+      - name: Upload the artifact to S3
+        env:
+          AWS_ACCESS_KEY_ID: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
+          AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
+          IS_NIGHTLY: ${{ github.event.ref == 'refs/heads/nightly' }}
+        shell: bash
+        working-directory: ${{ env.DEST_DIR }}
+        run: |
+          set -eux
+
+          pip install -q awscli==1.29.40
+
+          DRY_RUN=${DRY_RUN:-enabled}
+          AWS_S3_CP="aws s3 cp --dryrun"
+          if [ "${DRY_RUN}" == "disabled" ]; then
+            AWS_S3_CP="aws s3 cp"
+          fi
+
+          if [ "${IS_NIGHTLY}" == true ]; then
+            BUCKET_NAME="ossci-ios-build"
+          else
+            BUCKET_NAME="ossci-ios"
+          fi
+
+          ${AWS_S3_CP} "${ARTIFACT_NAME}" "s3://${BUCKET_NAME}/" --acl public-read
+          ${AWS_S3_CP} "${SPEC_NAME_WITH_VERSION}" "s3://${BUCKET_NAME}/" --acl public-read
+
+      - name: Upload the artifact to cocoapods (nightly only)
+        env:
+          # We need to set this secret to upload to cocoapods. However, we might want
+          # to NOT set this for PROD release so that we can upload the artifacts manually
+          COCOAPODS_TRUNK_TOKEN: ${{ secrets.COCOAPODS_TRUNK_TOKEN || '' }}
+        if: ${{ github.event_name == 'push' && github.event.ref == 'refs/heads/nightly' && env.COCOAPODS_TRUNK_TOKEN != '' }}
+        shell: bash
+        working-directory: ${{ runner.temp }}/arm64
+        run: |
+          set -eux
+
+          gem install cocoapods
+
+          pod trunk me
+          # Upload the spec to cocoapods
+          pod trunk push --verbose --allow-warnings --use-libraries --skip-import-validation "${SPEC_NAME}"

.github/workflows/_linux-build.yml

@@ -73,7 +73,7 @@ jobs:
       test-matrix: ${{ steps.filter.outputs.test-matrix }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
 
@@ -82,19 +82,19 @@ jobs:
       # checkout because when we run this action we don't *have* a local
       # checkout. In other cases you should prefer a local checkout.
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Setup Linux
         uses: ./.github/actions/setup-linux
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.1
         with:
           docker-image-name: ${{ inputs.docker-image-name }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -192,5 +192,5 @@ jobs:
           path: sccache-stats-*.json
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
         if: always()

.github/workflows/_linux-test.yml

@@ -57,7 +57,7 @@ jobs:
     timeout-minutes: ${{ inputs.timeout-minutes }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         if: ${{ !contains(matrix.runner, 'gcp.a100') }}
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -66,25 +66,25 @@ jobs:
               docker exec -it $(docker container ps --format '{{.ID}}') bash
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Setup Linux
         uses: ./.github/actions/setup-linux
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.1
         with:
           docker-image-name: ${{ inputs.docker-image }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
       - name: Install nvidia driver, nvidia-docker runtime, set GPU_FLAG
         id: install-nvidia-driver
-        uses: pytorch/test-infra/.github/actions/setup-nvidia@main
+        uses: pytorch/test-infra/.github/actions/setup-nvidia@release/2.1
         if: contains(inputs.build-environment, 'cuda') && !contains(matrix.config, 'nogpu')
 
       - name: Lock NVIDIA A100 40GB Frequency
@@ -292,7 +292,7 @@ jobs:
           path: ./**/core.[1-9]*
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
         if: always()
 
       # NB: We are currently having an intermittent GPU-related issue on G5 runners with

.github/workflows/_mac-build.yml

@@ -71,11 +71,11 @@ jobs:
       test-matrix: ${{ steps.filter.outputs.test-matrix }}
     steps:
       - name: Clean up disk space before running MacOS workflow
-        uses: pytorch/test-infra/.github/actions/check-disk-space@main
+        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.1
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Set xcode version
         env:
@@ -87,7 +87,7 @@ jobs:
 
       - name: Setup miniconda
         if: inputs.environment-file == ''
-        uses: pytorch/test-infra/.github/actions/setup-miniconda@main
+        uses: pytorch/test-infra/.github/actions/setup-miniconda@release/2.1
         with:
           python-version: ${{ inputs.python-version }}
           environment-file: .github/requirements/conda-env-${{ runner.os }}-${{ runner.arch }}
@@ -97,7 +97,7 @@ jobs:
       # environment even though the arch is x86-64
       - name: Setup miniconda using the provided environment file
         if: inputs.environment-file != ''
-        uses: pytorch/test-infra/.github/actions/setup-miniconda@main
+        uses: pytorch/test-infra/.github/actions/setup-miniconda@release/2.1
         with:
           python-version: ${{ inputs.python-version }}
           environment-file: ${{ inputs.environment-file }}
@@ -206,4 +206,4 @@ jobs:
       - name: Clean up disk space
         if: always()
         continue-on-error: true
-        uses: pytorch/test-infra/.github/actions/check-disk-space@main
+        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.1

.github/workflows/_mac-test-mps.yml

@@ -41,7 +41,7 @@ jobs:
       reenabled-issues: ${{ steps.filter.outputs.reenabled-issues }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           fetch-depth: 1
           submodules: false
@@ -85,7 +85,7 @@ jobs:
           use-gha: true
 
       - name: Setup miniconda
-        uses: pytorch/test-infra/.github/actions/setup-miniconda@main
+        uses: pytorch/test-infra/.github/actions/setup-miniconda@release/2.1
         with:
           python-version: ${{ inputs.python-version }}
           environment-file: .github/requirements/conda-env-${{ runner.os }}-${{ runner.arch }}

.github/workflows/_mac-test.yml

@@ -69,11 +69,11 @@ jobs:
           done
 
       - name: Clean up disk space before running MacOS workflow
-        uses: pytorch/test-infra/.github/actions/check-disk-space@main
+        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.1
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Download build artifacts
         uses: ./.github/actions/download-build-artifacts
@@ -82,7 +82,7 @@ jobs:
           use-gha: true
 
       - name: Setup miniconda
-        uses: pytorch/test-infra/.github/actions/setup-miniconda@main
+        uses: pytorch/test-infra/.github/actions/setup-miniconda@release/2.1
         with:
           python-version: ${{ inputs.python-version }}
           environment-file: .github/requirements/conda-env-${{ runner.os }}-${{ runner.arch }}
@@ -205,4 +205,4 @@ jobs:
       - name: Clean up disk space
         if: always()
         continue-on-error: true
-        uses: pytorch/test-infra/.github/actions/check-disk-space@main
+        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.1

.github/workflows/_rocm-test.yml

@@ -48,7 +48,7 @@ jobs:
     steps:
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           no-sudo: true
 
@@ -57,12 +57,12 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.1
         with:
           docker-image-name: ${{ inputs.docker-image }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 

.github/workflows/_run_android_tests.yml

@@ -22,7 +22,7 @@ jobs:
       keep-going: ${{ steps.filter.outputs.keep-going }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           fetch-depth: 1
           submodules: false
@@ -45,10 +45,10 @@ jobs:
     steps:
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Setup miniconda
-        uses: pytorch/test-infra/.github/actions/setup-miniconda@main
+        uses: pytorch/test-infra/.github/actions/setup-miniconda@release/2.1
         with:
           python-version: 3.8
           environment-file: .github/requirements/conda-env-${{ runner.os }}-${{ runner.arch }}

.github/workflows/_win-build.yml

@@ -60,10 +60,10 @@ jobs:
           git config --global core.fsmonitor false
 
       - name: Clean up leftover processes on non-ephemeral Windows runner
-        uses: pytorch/test-infra/.github/actions/cleanup-runner@main
+        uses: pytorch/test-infra/.github/actions/cleanup-runner@release/2.1
 
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
           instructions: |
@@ -78,7 +78,7 @@ jobs:
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           no-sudo: true
 

.github/workflows/_win-test.yml

@@ -48,10 +48,10 @@ jobs:
           git config --global core.fsmonitor false
 
       - name: Clean up leftover processes on non-ephemeral Windows runner
-        uses: pytorch/test-infra/.github/actions/cleanup-runner@main
+        uses: pytorch/test-infra/.github/actions/cleanup-runner@release/2.1
 
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
           instructions: |
@@ -67,7 +67,7 @@ jobs:
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           no-sudo: true
 

.github/workflows/build-ios-binaries.yml

@@ -0,0 +1,70 @@
+name: Build iOS binaries
+
+on:
+  push:
+    branches:
+      - nightly
+    tags:
+      # NOTE: Binary build pipelines should only get triggered on release candidate builds
+      # Release candidate tags look like: v1.11.0-rc1
+      - v[0-9]+.[0-9]+.[0-9]+-rc[0-9]+
+    paths:
+      - .github/workflows/build-ios-binaries.yml
+      - .github/workflows/_ios-build-test.yml
+  pull_request:
+    paths:
+      - .github/workflows/build-ios-binaries.yml
+      - .github/workflows/_ios-build-test.yml
+  # NB: We can use this workflow dispatch to test and build iOS binaries manually
+  workflow_dispatch:
+    inputs:
+      use_lite_interpreter:
+        description: "Use PyTorch lite interpreter?"
+        type: string
+        default: 1
+      use_coreml:
+        description: "Use Apple Core ML?"
+        type: string
+        default: 1
+      use_custom_op_list:
+        description: "Specify the custom ops list to include in the binaries"
+        type: string
+        default: ""
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
+  cancel-in-progress: true
+
+jobs:
+  # TODO: Figure out how to migrate this job to M1 runner
+  ios-build-test:
+    name: ios-build-test
+    uses: ./.github/workflows/_ios-build-test.yml
+    with:
+      build-environment: ios-build-test
+      sync-tag: ios-build-test
+      test-matrix: |
+        { include: [
+          { config: "default",
+            shard: 1,
+            num_shards: 1,
+            runner: "macos-12",
+            ios_platform: "SIMULATOR",
+            ios_arch: "x86_64",
+            use_lite_interpreter: ${{ inputs.use_lite_interpreter || 1 }},
+            use_metal: 0,
+            use_coreml: ${{ inputs.use_coreml || 1 }},
+            use_custom_op_list: ${{ inputs.use_custom_op_list || '' }}
+          },
+          { config: "default",
+            shard: 1,
+            num_shards: 1,
+            runner: "macos-12",
+            ios_platform: "OS",
+            ios_arch: "arm64",
+            use_lite_interpreter: ${{ inputs.use_lite_interpreter || 1 }},
+            use_metal: 1,
+            use_coreml: ${{ inputs.use_coreml || 1 }},
+            use_custom_op_list: ${{ inputs.use_custom_op_list || '' }}
+          }
+        ]}

.github/workflows/build-triton-wheel.yml

@@ -3,7 +3,11 @@ name: Build Triton wheels
 on:
   push:
     branches:
-      - main
+      - release/2.1
+    tags:
+      # NOTE: Binary build pipelines should only get triggered on release candidate builds
+      # Release candidate tags look like: v1.11.0-rc1
+      - v[0-9]+.[0-9]+.[0-9]+-rc[0-9]+
     paths:
       - .github/workflows/build-triton-wheel.yml
       - .github/scripts/build_triton_wheel.py
@@ -43,12 +47,12 @@ jobs:
       BUILD_DEVICE: ${{ matrix.device }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           submodules: false
 
@@ -56,11 +60,13 @@ jobs:
         uses: ./.github/actions/setup-linux
 
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ env.DOCKER_IMAGE }}
 
       - name: Build Triton wheel
+        env:
+          IS_RELEASE_TAG: ${{ startsWith(github.event.ref, 'refs/tags/v') }}
         run: |
           set -x
           mkdir -p "${RUNNER_TEMP}/artifacts/"
@@ -98,64 +104,75 @@ jobs:
             BUILD_ROCM="--build-rocm"
           fi
 
+          RELEASE=""
+          if [[ "${IS_RELEASE_TAG}" == true ]]; then
+            RELEASE="--release"
+          fi
+
           docker exec -t "${container_name}" yum install -y zlib-devel zip
           docker exec -t "${container_name}" "${PYTHON_EXECUTABLE}"  -m pip install -U setuptools==67.4.0
-          docker exec -t "${container_name}" "${PYTHON_EXECUTABLE}" /pytorch/.github/scripts/build_triton_wheel.py $BUILD_ROCM
+          docker exec -t "${container_name}" "${PYTHON_EXECUTABLE}" /pytorch/.github/scripts/build_triton_wheel.py $BUILD_ROCM $RELEASE
           docker exec -t "${container_name}" chown -R 1000.1000 /artifacts
 
       - uses: actions/upload-artifact@v3
         with:
-          name: "pytorch-triton-wheel-${{ matrix.py_vers }}"
+          # NB: Use the same name here and all wheels can be downloaded by referring to the same artifact
+          name: pytorch-triton-wheel
           if-no-files-found: error
-          path:
-            ${{ runner.temp }}/artifacts/*
+          path: ${{ runner.temp }}/artifacts/*
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
         if: always()
+
   upload-wheel:
-    runs-on: linux.20_04.4x
+    runs-on: ubuntu-22.04
     needs: build-wheel
     container:
       image: continuumio/miniconda3:4.12.0
-    env:
-      GITHUB_TOKEN: ${{ secrets.github-token }}
+    environment: ${{ (github.event_name == 'push' && (github.event.ref == 'refs/heads/nightly' || startsWith(github.event.ref, 'refs/tags/v'))) && 'conda-aws-upload' || '' }}
     steps:
-      - name: Download Build Artifacts (3.8)
+      - uses: actions/checkout@v3
+
+      - name: Download Build Artifacts
         uses: actions/download-artifact@v3
         with:
-          name: "pytorch-triton-wheel-3.8"
-          path: "${{ runner.temp }}/artifacts/"
-      - name: Download Build Artifacts (3.9)
-        uses: actions/download-artifact@v3
-        with:
-          name: "pytorch-triton-wheel-3.9"
-          path: "${{ runner.temp }}/artifacts/"
-      - name: Download Build Artifacts (3.10)
-        uses: actions/download-artifact@v3
-        with:
-          name: "pytorch-triton-wheel-3.10"
-          path: "${{ runner.temp }}/artifacts/"
-      - name: Download Build Artifacts (3.11)
-        uses: actions/download-artifact@v3
-        with:
-          name: "pytorch-triton-wheel-3.11"
-          path: "${{ runner.temp }}/artifacts/"
-      - name: Upload binaries
-        if: ${{ github.event_name == 'push' && github.event.ref == 'refs/heads/main' }}
-        env:
-          PKG_DIR: "${{ runner.temp }}/artifacts"
-          # When running these on pull_request events these should be blank
-          AWS_ACCESS_KEY_ID: ${{ secrets.AWS_S3_UPDATE_ACCESS_KEY_ID }}
-          AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_S3_UPDATE_SECRET_ACCESS_KEY }}
-          UPLOAD_BUCKET: "s3://pytorch"
+          name: pytorch-triton-wheel
+          path: ${{ runner.temp }}/artifacts/
+
+      - name: Set DRY_RUN (only for tagged pushes)
+        if: ${{ github.event_name == 'push' && (github.event.ref == 'refs/heads/nightly' || (startsWith(github.event.ref, 'refs/tags/v'))) }}
+        shell: bash
         run: |
-            set -ex
-            pip install -q awscli
-            s3_dir="${UPLOAD_BUCKET}/whl/nightly/"
-            for pkg in "${PKG_DIR}/"*.whl; do
-              aws s3 cp --no-progress --acl public-read "${pkg}" "${s3_dir}"
-             done
+          echo "DRY_RUN=disabled" >> "$GITHUB_ENV"
+
+      - name: Set UPLOAD_CHANNEL (only for tagged pushes)
+        if: ${{ github.event_name == 'push' && startsWith(github.event.ref, 'refs/tags/v') }}
+        shell: bash
+        run: |
+          set -ex
+
+          # reference ends with an RC suffix
+          if [[ "${GITHUB_REF_NAME}" = *-rc[0-9]* ]]; then
+            echo "UPLOAD_CHANNEL=test" >> "$GITHUB_ENV"
+          fi
+
+      # NB: This step is gated by DRY_RUN, which is enabled everywhere except nightly and release branches
+      - name: Upload binaries
+        env:
+          PACKAGE_TYPE: wheel
+          # The UPLOAD_SUBFOLDER needs to be empty here so that triton wheels are uploaded
+          # to nightly or test
+          UPLOAD_SUBFOLDER: ""
+          PKG_DIR: ${{ runner.temp }}/artifacts
+          # When running these on pull_request events these should be blank
+          AWS_ACCESS_KEY_ID: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
+          AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
+        shell: bash
+        run: |
+          set -ex
+          bash .circleci/scripts/binary_upload.sh
+
   build-conda:
     name: "Build Triton Conda"
     runs-on: [self-hosted, linux.2xlarge]
@@ -164,19 +181,17 @@ jobs:
       matrix:
         py_vers: [ "3.8", "3.9", "3.10", "3.11" ]
     timeout-minutes: 40
-    environment: ${{ (github.event_name == 'push' && github.event.ref == 'refs/heads/main') && 'conda-aws-upload' || '' }}
     env:
       DOCKER_IMAGE: pytorch/conda-builder:cpu
       PY_VERS: ${{ matrix.py_vers }}
-      ANACONDA_API_TOKEN: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           submodules: false
 
@@ -184,11 +199,13 @@ jobs:
         uses: ./.github/actions/setup-linux
 
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ env.DOCKER_IMAGE }}
 
       - name: Build Triton conda package
+        env:
+          IS_RELEASE_TAG: ${{ startsWith(github.event.ref, 'refs/tags/v') }}
         run: |
           set -x
           mkdir -p "${RUNNER_TEMP}/artifacts/"
@@ -198,31 +215,76 @@ jobs:
             -v "${GITHUB_WORKSPACE}:/pytorch" \
             -v "${RUNNER_TEMP}/artifacts:/artifacts" \
             -w /artifacts/ \
-            -e ANACONDA_API_TOKEN \
             "${DOCKER_IMAGE}" \
           )
 
+          RELEASE=""
+          if [[ "${IS_RELEASE_TAG}" == true ]]; then
+            RELEASE="--release"
+          fi
+
           docker exec -t "${container_name}" yum install -y llvm11 llvm11-devel llvm11-static llvm11-libs zlib-devel
-          docker exec -t "${container_name}" python /pytorch/.github/scripts/build_triton_wheel.py --build-conda --py-version="${PY_VERS}"
-
-      - name: Upload artifacts to Anaconda
-        if: ${{ github.event_name == 'push' && github.event.ref == 'refs/heads/main' }}
-        run: |
-          container_name=$(docker container ps --format '{{.ID}}')
-          docker exec -t "${container_name}" sh -c "anaconda upload /artifacts/torch*.tar.bz2 -u pytorch-nightly --label main --no-progress --force"
-
-      - name: Chown artifacts
-        run: |
-          container_name=$(docker container ps --format '{{.ID}}')
+          docker exec -t "${container_name}" python /pytorch/.github/scripts/build_triton_wheel.py --build-conda --py-version="${PY_VERS}" $RELEASE
           docker exec -t "${container_name}" chown -R 1000.1000 /artifacts
 
       - uses: actions/upload-artifact@v3
         with:
-          name: "pytorch-triton-conda-${{ matrix.py_vers }}"
+          # NB: Use the same name here and all wheels can be downloaded by referring to the same artifact
+          name: pytorch-triton-conda
           if-no-files-found: error
-          path:
-            ${{ runner.temp }}/artifacts/*
+          path: ${{ runner.temp }}/artifacts/*
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
         if: always()
+
+  upload-conda:
+    runs-on: ubuntu-22.04
+    needs: build-conda
+    container:
+      image: continuumio/miniconda3:4.12.0
+    environment: ${{ (github.event_name == 'push' && (github.event.ref == 'refs/heads/nightly' || startsWith(github.event.ref, 'refs/tags/v'))) && 'conda-aws-upload' || '' }}
+    steps:
+      - uses: actions/checkout@v3
+
+      - name: Download Build Artifacts
+        uses: actions/download-artifact@v3
+        with:
+          name: pytorch-triton-conda
+          path: ${{ runner.temp }}/artifacts/
+
+      - name: Set DRY_RUN (only for tagged pushes)
+        if: ${{ github.event_name == 'push' && (github.event.ref == 'refs/heads/nightly' || (startsWith(github.event.ref, 'refs/tags/v'))) }}
+        shell: bash
+        run: |
+          echo "DRY_RUN=disabled" >> "$GITHUB_ENV"
+
+      - name: Set UPLOAD_CHANNEL (only for tagged pushes)
+        if: ${{ github.event_name == 'push' && startsWith(github.event.ref, 'refs/tags/v') }}
+        shell: bash
+        run: |
+          set -ex
+
+          # reference ends with an RC suffix
+          if [[ "${GITHUB_REF_NAME}" = *-rc[0-9]* ]]; then
+            echo "UPLOAD_CHANNEL=test" >> "$GITHUB_ENV"
+          fi
+
+      # NB: This step is gated by DRY_RUN, which is enabled everywhere except nightly and release branches
+      - name: Upload binaries to Anaconda
+        env:
+          PACKAGE_TYPE: conda
+          PKG_DIR: ${{ runner.temp }}/artifacts
+          # When running these on pull_request events these should be blank
+          CONDA_PYTORCHBOT_TOKEN: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+          CONDA_PYTORCHBOT_TOKEN_TEST: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
+        shell: bash
+        run: |
+          set -ex
+
+          if [[ "${UPLOAD_CHANNEL}" = "nightly" ]]; then
+            export ANACONDA_API_TOKEN="${CONDA_PYTORCHBOT_TOKEN}"
+          else
+            export ANACONDA_API_TOKEN="${CONDA_PYTORCHBOT_TOKEN_TEST}"
+          fi
+          bash .circleci/scripts/binary_upload.sh

.github/workflows/check-labels.yml

@@ -29,7 +29,7 @@ jobs:
     runs-on: linux.20_04.4x
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           submodules: false
           fetch-depth: 1

.github/workflows/close-nonexistent-disable-issues.yml

@@ -10,7 +10,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Run close_nonexistent_disable_issues.py
         env:

.github/workflows/docker-builds.yml

@@ -61,21 +61,21 @@ jobs:
       # [see note: pytorch repo ref]
       # deep clone (fetch-depth 0) required for git merge-base
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - name: Setup Linux
         uses: ./.github/actions/setup-linux
 
       - name: Build docker image
         id: build-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.1
         with:
           docker-image-name: ${{ matrix.docker-image-name }}
           always-rebuild: true
           push: true
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: ${{ steps.build-docker-image.outputs.docker-image }}
 
@@ -105,5 +105,5 @@ jobs:
         if: always()
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
         if: always()

.github/workflows/docker-release.yml

@@ -47,7 +47,7 @@ jobs:
       BUILD_PLATFORMS: ${{ matrix.platform }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.1
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
       # [see note: pytorch repo ref]
@@ -109,5 +109,5 @@ jobs:
                  ghcr.io/pytorch/pytorch-nightly:latest
           docker push ghcr.io/pytorch/pytorch-nightly:latest
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.1
         if: always()

.github/workflows/generated-linux-aarch64-binary-manywheel-nightly.yml

@@ -93,7 +93,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_9-cpu-aarch64-build:
@@ -153,7 +153,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_10-cpu-aarch64-build:
@@ -213,7 +213,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_11-cpu-aarch64-build:
@@ -273,5 +273,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-linux-binary-conda-nightly.yml

@@ -90,7 +90,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   conda-py3_8-cuda11_8-build:
@@ -150,7 +150,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   conda-py3_8-cuda12_1-build:
@@ -210,7 +210,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   conda-py3_9-cpu-build:
@@ -267,7 +267,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   conda-py3_9-cuda11_8-build:
@@ -327,7 +327,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   conda-py3_9-cuda12_1-build:
@@ -387,7 +387,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   conda-py3_10-cpu-build:
@@ -444,7 +444,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   conda-py3_10-cuda11_8-build:
@@ -504,7 +504,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   conda-py3_10-cuda12_1-build:
@@ -564,7 +564,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   conda-py3_11-cpu-build:
@@ -621,7 +621,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   conda-py3_11-cuda11_8-build:
@@ -681,7 +681,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   conda-py3_11-cuda12_1-build:
@@ -741,5 +741,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-linux-binary-libtorch-cxx11-abi-nightly.yml

@@ -93,7 +93,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cpu-shared-without-deps-cxx11-abi-build:
@@ -153,7 +153,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cpu-static-with-deps-cxx11-abi-build:
@@ -213,7 +213,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cpu-static-without-deps-cxx11-abi-build:
@@ -273,7 +273,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda11_8-shared-with-deps-cxx11-abi-build:
@@ -336,7 +336,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda11_8-shared-without-deps-cxx11-abi-build:
@@ -399,7 +399,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda11_8-static-with-deps-cxx11-abi-build:
@@ -462,7 +462,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda11_8-static-without-deps-cxx11-abi-build:
@@ -525,7 +525,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda12_1-shared-with-deps-cxx11-abi-build:
@@ -588,7 +588,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda12_1-shared-without-deps-cxx11-abi-build:
@@ -651,7 +651,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda12_1-static-with-deps-cxx11-abi-build:
@@ -714,7 +714,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda12_1-static-without-deps-cxx11-abi-build:
@@ -777,7 +777,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-rocm5_5-shared-with-deps-cxx11-abi-build:
@@ -828,7 +828,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -840,7 +839,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -854,7 +853,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/libtorch-cxx11-builder:rocm5.5
       - name: Test Pytorch binary
@@ -881,7 +880,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-rocm5_5-static-with-deps-cxx11-abi-build:
@@ -932,7 +931,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -944,7 +942,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -958,7 +956,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/libtorch-cxx11-builder:rocm5.5
       - name: Test Pytorch binary
@@ -985,7 +983,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-rocm5_6-shared-with-deps-cxx11-abi-build:
@@ -1036,7 +1034,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1048,7 +1045,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1062,7 +1059,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/libtorch-cxx11-builder:rocm5.6
       - name: Test Pytorch binary
@@ -1089,7 +1086,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-rocm5_6-static-with-deps-cxx11-abi-build:
@@ -1140,7 +1137,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1152,7 +1148,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1166,7 +1162,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/libtorch-cxx11-builder:rocm5.6
       - name: Test Pytorch binary
@@ -1193,5 +1189,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-linux-binary-libtorch-pre-cxx11-nightly.yml

@@ -93,7 +93,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cpu-shared-without-deps-pre-cxx11-build:
@@ -153,7 +153,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cpu-static-with-deps-pre-cxx11-build:
@@ -213,7 +213,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cpu-static-without-deps-pre-cxx11-build:
@@ -273,7 +273,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda11_8-shared-with-deps-pre-cxx11-build:
@@ -336,7 +336,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda11_8-shared-without-deps-pre-cxx11-build:
@@ -399,7 +399,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda11_8-static-with-deps-pre-cxx11-build:
@@ -462,7 +462,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda11_8-static-without-deps-pre-cxx11-build:
@@ -525,7 +525,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda12_1-shared-with-deps-pre-cxx11-build:
@@ -588,7 +588,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda12_1-shared-without-deps-pre-cxx11-build:
@@ -651,7 +651,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda12_1-static-with-deps-pre-cxx11-build:
@@ -714,7 +714,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-cuda12_1-static-without-deps-pre-cxx11-build:
@@ -777,7 +777,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-rocm5_5-shared-with-deps-pre-cxx11-build:
@@ -828,7 +828,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -840,7 +839,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -854,7 +853,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.5
       - name: Test Pytorch binary
@@ -881,7 +880,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-rocm5_5-static-with-deps-pre-cxx11-build:
@@ -932,7 +931,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -944,7 +942,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -958,7 +956,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.5
       - name: Test Pytorch binary
@@ -985,7 +983,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-rocm5_6-shared-with-deps-pre-cxx11-build:
@@ -1036,7 +1034,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1048,7 +1045,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1062,7 +1059,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.6
       - name: Test Pytorch binary
@@ -1089,7 +1086,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   libtorch-rocm5_6-static-with-deps-pre-cxx11-build:
@@ -1140,7 +1137,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1152,7 +1148,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1166,7 +1162,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.6
       - name: Test Pytorch binary
@@ -1193,5 +1189,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-linux-binary-manywheel-main.yml

@@ -86,7 +86,7 @@ jobs:
       DESIRED_PYTHON: "3.8"
       build_name: manywheel-py3_8-cuda12_1-with-pypi-cudnn
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==8.9.2.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.1.3.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.0.2.54; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.2.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.4.5.107; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.1.0.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.18.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==8.9.2.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.1.3.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.0.2.54; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.2.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.4.5.107; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.1.0.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.18.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | triton==2.1.0; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_8-cuda12_1-with-pypi-cudnn-test:  # Testing

.github/workflows/generated-linux-binary-manywheel-nightly.yml

@@ -90,7 +90,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_8-cpu-cxx11-abi-build:
@@ -150,7 +150,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_8-cuda11_8-build:
@@ -210,7 +210,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_8-cuda12_1-with-pypi-cudnn-build:
@@ -229,7 +229,7 @@ jobs:
       DESIRED_PYTHON: "3.8"
       build_name: manywheel-py3_8-cuda12_1-with-pypi-cudnn
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==8.9.2.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.1.3.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.0.2.54; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.2.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.4.5.107; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.1.0.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.18.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==8.9.2.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.1.3.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.0.2.54; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.2.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.4.5.107; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.1.0.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.18.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | triton==2.1.0; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_8-cuda12_1-with-pypi-cudnn-test:  # Testing
@@ -271,7 +271,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_8-cuda12_1-build:
@@ -331,7 +331,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_8-rocm5_5-build:
@@ -380,7 +380,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -392,7 +391,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -406,7 +405,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.5
       - name: Test Pytorch binary
@@ -432,7 +431,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_8-rocm5_6-build:
@@ -481,7 +480,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -493,7 +491,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -507,7 +505,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.6
       - name: Test Pytorch binary
@@ -533,7 +531,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_9-cpu-build:
@@ -590,7 +588,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_9-cpu-cxx11-abi-build:
@@ -650,7 +648,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_9-cuda11_8-build:
@@ -710,7 +708,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_9-cuda12_1-with-pypi-cudnn-build:
@@ -729,7 +727,7 @@ jobs:
       DESIRED_PYTHON: "3.9"
       build_name: manywheel-py3_9-cuda12_1-with-pypi-cudnn
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==8.9.2.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.1.3.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.0.2.54; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.2.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.4.5.107; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.1.0.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.18.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==8.9.2.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.1.3.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.0.2.54; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.2.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.4.5.107; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.1.0.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.18.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | triton==2.1.0; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_9-cuda12_1-with-pypi-cudnn-test:  # Testing
@@ -771,7 +769,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_9-cuda12_1-build:
@@ -831,7 +829,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_9-rocm5_5-build:
@@ -880,7 +878,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -892,7 +889,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -906,7 +903,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.5
       - name: Test Pytorch binary
@@ -932,7 +929,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_9-rocm5_6-build:
@@ -981,7 +978,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -993,7 +989,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1007,7 +1003,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.6
       - name: Test Pytorch binary
@@ -1033,7 +1029,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_10-cpu-build:
@@ -1090,7 +1086,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_10-cpu-cxx11-abi-build:
@@ -1150,7 +1146,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_10-cuda11_8-build:
@@ -1210,7 +1206,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_10-cuda12_1-with-pypi-cudnn-build:
@@ -1229,7 +1225,7 @@ jobs:
       DESIRED_PYTHON: "3.10"
       build_name: manywheel-py3_10-cuda12_1-with-pypi-cudnn
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==8.9.2.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.1.3.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.0.2.54; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.2.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.4.5.107; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.1.0.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.18.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==8.9.2.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.1.3.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.0.2.54; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.2.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.4.5.107; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.1.0.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.18.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | triton==2.1.0; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_10-cuda12_1-with-pypi-cudnn-test:  # Testing
@@ -1271,7 +1267,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_10-cuda12_1-build:
@@ -1331,7 +1327,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_10-rocm5_5-build:
@@ -1380,7 +1376,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1392,7 +1387,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1406,7 +1401,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.5
       - name: Test Pytorch binary
@@ -1432,7 +1427,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_10-rocm5_6-build:
@@ -1481,7 +1476,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1493,7 +1487,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1507,7 +1501,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.6
       - name: Test Pytorch binary
@@ -1533,7 +1527,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_11-cpu-build:
@@ -1590,7 +1584,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_11-cpu-cxx11-abi-build:
@@ -1650,7 +1644,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_11-cuda11_8-build:
@@ -1710,7 +1704,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_11-cuda12_1-with-pypi-cudnn-build:
@@ -1729,7 +1723,7 @@ jobs:
       DESIRED_PYTHON: "3.11"
       build_name: manywheel-py3_11-cuda12_1-with-pypi-cudnn
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==8.9.2.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.1.3.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.0.2.54; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.2.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.4.5.107; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.1.0.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.18.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==8.9.2.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.1.3.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.0.2.54; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.2.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.4.5.107; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.1.0.106; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.18.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.1.105; platform_system == 'Linux' and platform_machine == 'x86_64' | triton==2.1.0; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_11-cuda12_1-with-pypi-cudnn-test:  # Testing
@@ -1771,7 +1765,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_11-cuda12_1-build:
@@ -1831,7 +1825,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_11-rocm5_5-build:
@@ -1880,7 +1874,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1892,7 +1885,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1906,7 +1899,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.5
       - name: Test Pytorch binary
@@ -1932,7 +1925,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
 
   manywheel-py3_11-rocm5_6-build:
@@ -1981,7 +1974,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1993,7 +1985,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2007,7 +1999,7 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.1
         with:
           docker-image: pytorch/manylinux-builder:rocm5.6
       - name: Test Pytorch binary
@@ -2033,5 +2025,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-macos-arm64-binary-conda-nightly.yml

@@ -75,7 +75,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -87,7 +86,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -144,7 +143,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_9-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -187,7 +186,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -199,7 +197,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -256,7 +254,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_10-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -299,7 +297,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -311,7 +308,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -368,7 +365,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_11-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -411,7 +408,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -423,7 +419,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -480,5 +476,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-macos-arm64-binary-wheel-nightly.yml

@@ -75,7 +75,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -87,7 +86,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -144,7 +143,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_9-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -187,7 +186,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -199,7 +197,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -256,7 +254,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_10-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -299,7 +297,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -311,7 +308,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -368,7 +365,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_11-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -411,7 +408,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -423,7 +419,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -480,5 +476,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-macos-binary-conda-nightly.yml

@@ -73,7 +73,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -85,7 +84,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -142,7 +141,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_9-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -185,7 +184,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -197,7 +195,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -254,7 +252,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_10-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -297,7 +295,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -309,7 +306,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -366,7 +363,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_11-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -409,7 +406,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -421,7 +417,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -478,5 +474,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-macos-binary-libtorch-cxx11-abi-nightly.yml

@@ -77,7 +77,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -89,7 +88,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -147,7 +146,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cpu-shared-without-deps-cxx11-abi-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -194,7 +193,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -206,7 +204,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -264,7 +262,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cpu-static-with-deps-cxx11-abi-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -311,7 +309,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -323,7 +320,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -381,7 +378,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cpu-static-without-deps-cxx11-abi-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -428,7 +425,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -440,7 +436,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -498,5 +494,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-macos-binary-wheel-nightly.yml

@@ -73,7 +73,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -85,7 +84,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -142,7 +141,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_9-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -185,7 +184,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -197,7 +195,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -254,7 +252,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_10-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -297,7 +295,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -309,7 +306,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -366,7 +363,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_11-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -409,7 +406,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -421,7 +417,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -478,5 +474,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-windows-binary-conda-nightly.yml

@@ -92,7 +92,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -104,7 +103,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -208,7 +207,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -220,7 +218,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -268,7 +266,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_8-cuda11_8-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -331,7 +329,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -343,7 +340,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -448,7 +445,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -460,7 +456,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -509,7 +505,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_8-cuda12_1-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -572,7 +568,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -584,7 +579,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -689,7 +684,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -701,7 +695,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -750,7 +744,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_9-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -812,7 +806,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -824,7 +817,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -928,7 +921,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -940,7 +932,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -988,7 +980,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_9-cuda11_8-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1051,7 +1043,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1063,7 +1054,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1168,7 +1159,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1180,7 +1170,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1229,7 +1219,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_9-cuda12_1-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1292,7 +1282,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1304,7 +1293,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1409,7 +1398,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1421,7 +1409,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1470,7 +1458,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_10-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1532,7 +1520,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1544,7 +1531,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1648,7 +1635,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1660,7 +1646,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1708,7 +1694,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_10-cuda11_8-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1771,7 +1757,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1783,7 +1768,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1888,7 +1873,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1900,7 +1884,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1949,7 +1933,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_10-cuda12_1-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2012,7 +1996,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2024,7 +2007,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2129,7 +2112,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2141,7 +2123,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2190,7 +2172,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_11-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2252,7 +2234,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2264,7 +2245,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2368,7 +2349,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2380,7 +2360,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2428,7 +2408,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_11-cuda11_8-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2491,7 +2471,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2503,7 +2482,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2608,7 +2587,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2620,7 +2598,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2669,7 +2647,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   conda-py3_11-cuda12_1-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2732,7 +2710,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2744,7 +2721,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2849,7 +2826,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2861,7 +2837,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2910,5 +2886,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-windows-binary-libtorch-debug-main.yml

@@ -89,7 +89,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -101,7 +100,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -209,7 +208,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -221,7 +219,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder

.github/workflows/generated-windows-binary-libtorch-debug-nightly.yml

@@ -96,7 +96,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -108,7 +107,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -216,7 +215,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -228,7 +226,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -280,7 +278,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cpu-shared-without-deps-debug-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -346,7 +344,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -358,7 +355,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -466,7 +463,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -478,7 +474,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -530,7 +526,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cpu-static-with-deps-debug-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -596,7 +592,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -608,7 +603,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -716,7 +711,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -728,7 +722,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -780,7 +774,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cpu-static-without-deps-debug-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -846,7 +840,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -858,7 +851,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -966,7 +959,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -978,7 +970,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1030,7 +1022,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda11_8-shared-with-deps-debug-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1097,7 +1089,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1109,7 +1100,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1218,7 +1209,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1230,7 +1220,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1283,7 +1273,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda11_8-shared-without-deps-debug-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1350,7 +1340,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1362,7 +1351,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1471,7 +1460,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1483,7 +1471,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1536,7 +1524,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda11_8-static-with-deps-debug-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1603,7 +1591,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1615,7 +1602,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1724,7 +1711,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1736,7 +1722,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1789,7 +1775,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda11_8-static-without-deps-debug-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1856,7 +1842,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1868,7 +1853,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1977,7 +1962,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1989,7 +1973,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2042,7 +2026,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda12_1-shared-with-deps-debug-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2109,7 +2093,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2121,7 +2104,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2230,7 +2213,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2242,7 +2224,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2295,7 +2277,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda12_1-shared-without-deps-debug-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2362,7 +2344,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2374,7 +2355,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2483,7 +2464,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2495,7 +2475,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2548,7 +2528,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda12_1-static-with-deps-debug-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2615,7 +2595,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2627,7 +2606,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2736,7 +2715,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2748,7 +2726,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2801,7 +2779,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda12_1-static-without-deps-debug-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2868,7 +2846,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2880,7 +2857,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2989,7 +2966,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -3001,7 +2977,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -3054,5 +3030,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-windows-binary-libtorch-release-main.yml

@@ -89,7 +89,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -101,7 +100,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -209,7 +208,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -221,7 +219,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder

.github/workflows/generated-windows-binary-libtorch-release-nightly.yml

@@ -96,7 +96,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -108,7 +107,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -216,7 +215,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -228,7 +226,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -280,7 +278,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cpu-shared-without-deps-release-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -346,7 +344,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -358,7 +355,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -466,7 +463,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -478,7 +474,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -530,7 +526,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cpu-static-with-deps-release-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -596,7 +592,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -608,7 +603,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -716,7 +711,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -728,7 +722,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -780,7 +774,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cpu-static-without-deps-release-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -846,7 +840,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -858,7 +851,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -966,7 +959,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -978,7 +970,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1030,7 +1022,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda11_8-shared-with-deps-release-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1097,7 +1089,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1109,7 +1100,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1218,7 +1209,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1230,7 +1220,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1283,7 +1273,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda11_8-shared-without-deps-release-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1350,7 +1340,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1362,7 +1351,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1471,7 +1460,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1483,7 +1471,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1536,7 +1524,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda11_8-static-with-deps-release-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1603,7 +1591,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1615,7 +1602,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1724,7 +1711,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1736,7 +1722,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1789,7 +1775,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda11_8-static-without-deps-release-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1856,7 +1842,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1868,7 +1853,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1977,7 +1962,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1989,7 +1973,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2042,7 +2026,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda12_1-shared-with-deps-release-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2109,7 +2093,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2121,7 +2104,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2230,7 +2213,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2242,7 +2224,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2295,7 +2277,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda12_1-shared-without-deps-release-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2362,7 +2344,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2374,7 +2355,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2483,7 +2464,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2495,7 +2475,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2548,7 +2528,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda12_1-static-with-deps-release-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2615,7 +2595,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2627,7 +2606,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2736,7 +2715,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2748,7 +2726,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2801,7 +2779,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   libtorch-cuda12_1-static-without-deps-release-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2868,7 +2846,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2880,7 +2857,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2989,7 +2966,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -3001,7 +2977,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -3054,5 +3030,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/generated-windows-binary-wheel-nightly.yml

@@ -92,7 +92,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -104,7 +103,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -208,7 +207,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -220,7 +218,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -268,7 +266,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_8-cuda11_8-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -331,7 +329,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -343,7 +340,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -448,7 +445,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -460,7 +456,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -509,7 +505,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_8-cuda12_1-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -572,7 +568,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -584,7 +579,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -689,7 +684,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -701,7 +695,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -750,7 +744,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_9-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -812,7 +806,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -824,7 +817,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -928,7 +921,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -940,7 +932,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -988,7 +980,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_9-cuda11_8-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1051,7 +1043,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1063,7 +1054,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1168,7 +1159,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1180,7 +1170,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1229,7 +1219,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_9-cuda12_1-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1292,7 +1282,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1304,7 +1293,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1409,7 +1398,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1421,7 +1409,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1470,7 +1458,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_10-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1532,7 +1520,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1544,7 +1531,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1648,7 +1635,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1660,7 +1646,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1708,7 +1694,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_10-cuda11_8-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -1771,7 +1757,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1783,7 +1768,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1888,7 +1873,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -1900,7 +1884,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -1949,7 +1933,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_10-cuda12_1-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2012,7 +1996,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2024,7 +2007,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2129,7 +2112,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2141,7 +2123,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2190,7 +2172,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_11-cpu-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2252,7 +2234,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2264,7 +2245,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2368,7 +2349,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2380,7 +2360,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2428,7 +2408,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_11-cuda11_8-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2491,7 +2471,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2503,7 +2482,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2608,7 +2587,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2620,7 +2598,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2669,7 +2647,7 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml
   wheel-py3_11-cuda12_1-build:
     if: ${{ github.repository_owner == 'pytorch' }}
@@ -2732,7 +2710,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2744,7 +2721,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2849,7 +2826,6 @@ jobs:
       - name: Checkout PyTorch
         uses: malfet/checkout@silent-checkout
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           quiet-checkout: true
@@ -2861,7 +2837,7 @@ jobs:
       - name: Checkout pytorch/builder
         uses: malfet/checkout@silent-checkout
         with:
-          ref: main
+          ref: release/2.1
           submodules: recursive
           repository: pytorch/builder
           path: builder
@@ -2910,5 +2886,5 @@ jobs:
       github-token: ${{ secrets.GITHUB_TOKEN }}
       aws-pytorch-uploader-access-key-id: ${{ secrets.AWS_PYTORCH_UPLOADER_ACCESS_KEY_ID }}
       aws-pytorch-uploader-secret-access-key: ${{ secrets.AWS_PYTORCH_UPLOADER_SECRET_ACCESS_KEY }}
-      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN }}
+      conda-pytorchbot-token: ${{ secrets.CONDA_PYTORCHBOT_TOKEN_TEST }}
     uses: ./.github/workflows/_binary-upload.yml

.github/workflows/lint-bc.yml

@@ -26,7 +26,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Run BC Lint Action
-        uses: pytorch/test-infra/.github/actions/bc-lint@main
+        uses: pytorch/test-infra/.github/actions/bc-lint@release/2.1
         with:
           repo: ${{ github.event.pull_request.head.repo.full_name }}
           base_sha: ${{ github.event.pull_request.base.sha }}

.github/workflows/lint.yml

@@ -15,7 +15,7 @@ on:
 # When any other step fails, it's job will be retried once by retryBot.
 jobs:
   lintrunner:
-    uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job.yml@release/2.1
     with:
       runner: linux.2xlarge
       docker-image: pytorch-linux-focal-linter
@@ -62,7 +62,7 @@ jobs:
         exit $RC
 
   quick-checks:
-    uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job.yml@release/2.1
     with:
       runner: linux.2xlarge
       docker-image: pytorch-linux-focal-linter
@@ -103,7 +103,7 @@ jobs:
     if: github.event_name == 'pull_request' && !contains(github.event.pull_request.labels.*.name, 'skip-pr-sanity-checks')
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           submodules: false
           fetch-depth: -1
@@ -116,7 +116,7 @@ jobs:
           bash .github/scripts/pr-sanity-check.sh
 
   workflow-checks:
-    uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job.yml@release/2.1
     with:
       runner: linux.2xlarge
       docker-image: pytorch-linux-focal-linter
@@ -151,7 +151,7 @@ jobs:
         exit $RC
 
   toc:
-    uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job.yml@release/2.1
     with:
       runner: linux.2xlarge
       docker-image: pytorch-linux-focal-linter
@@ -189,7 +189,7 @@ jobs:
   test-tools:
     name: Test tools
     if: ${{ github.repository == 'pytorch/pytorch' }}
-    uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job.yml@release/2.1
     with:
       runner: linux.2xlarge
       docker-image: pytorch-linux-focal-linter
@@ -210,7 +210,7 @@ jobs:
     runs-on: linux.20_04.4x
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           submodules: false
           fetch-depth: 1
@@ -240,7 +240,7 @@ jobs:
       # [see note: pytorch repo ref]
       # deep clone (fetch-depth 0) required, to allow us to use git log
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           submodules: false
           fetch-depth: 1

.github/workflows/nightly-rockset-uploads.yml

@@ -21,7 +21,7 @@ jobs:
     environment: upload-stats
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           fetch-depth: 1
           submodules: false

.github/workflows/periodic.yml

@@ -112,30 +112,38 @@ jobs:
       cuda-version: "11.8"
       test-matrix: ${{ needs.win-vs2019-cuda11_8-py3-build.outputs.test-matrix }}
 
-  ios-12-5-1-x86-64-coreml:
-    name: ios-12-5-1-x86-64-coreml
+  # TODO: Figure out how to migrate this job to M1 runner
+  ios-build-test:
+    name: ios-build-test
     if: github.event_name != 'schedule' || github.event.schedule == '45 0,8,16 * * 1-5' || github.event.schedule == '45 4 * * 0,6'
     uses: ./.github/workflows/_ios-build-test.yml
     with:
-      build-environment: ios-12-5-1-x86-64-coreml
-      ios-platform: SIMULATOR
-      ios-arch: x86_64
+      build-environment: ios-build-test
+      sync-tag: ios-build-test
       test-matrix: |
         { include: [
-          { config: "default", shard: 1, num_shards: 1, runner: "macos-12" },
-        ]}
-
-  ios-12-5-1-arm64-custom-ops:
-    name: ios-12-5-1-arm64-custom-ops
-    if: github.event_name != 'schedule' || github.event.schedule == '45 0,8,16 * * 1-5' || github.event.schedule == '45 4 * * 0,6'
-    uses: ./.github/workflows/_ios-build-test.yml
-    with:
-      build-environment: ios-12-5-1-arm64-custom-ops
-      ios-platform: OS
-      ios-arch: arm64
-      test-matrix: |
-        { include: [
-          { config: "default", shard: 1, num_shards: 1, runner: "macos-12" },
+          { config: "default",
+            shard: 1,
+            num_shards: 1,
+            runner: "macos-12",
+            ios_platform: "SIMULATOR",
+            ios_arch: "x86_64",
+            use_lite_interpreter: 1,
+            use_metal: 0,
+            use_coreml: 1,
+            use_custom_op_list: ""
+          },
+          { config: "default",
+            shard: 1,
+            num_shards: 1,
+            runner: "macos-12",
+            ios_platform: "OS",
+            ios_arch: "arm64",
+            use_lite_interpreter: 1,
+            use_metal: 1,
+            use_coreml: 1,
+            use_custom_op_list: "mobilenetv2.yaml"
+          }
         ]}
 
   buck-build-test:

.github/workflows/update_pytorch_labels.yml

@@ -14,7 +14,7 @@ jobs:
     if: ${{ github.repository == 'pytorch/pytorch' }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           fetch-depth: 1
           submodules: false

.github/workflows/upload-alerts.yml

@@ -44,7 +44,7 @@ jobs:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
           AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
-        uses: pytorch/test-infra/.github/actions/upload-alerts@main
+        uses: pytorch/test-infra/.github/actions/upload-alerts@release/2.1
         with:
           alerts: '${{ steps.alert_creation_step.outputs.script-output }}'
           organization: "pytorch"

.github/workflows/upload-test-stats.yml

@@ -37,7 +37,7 @@ jobs:
         run: echo "${TRIGGERING_WORKFLOW}"
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
 
       - uses: actions/setup-python@v4
         with:

.github/workflows/upload-torch-dynamo-perf-stats.yml

@@ -29,7 +29,7 @@ jobs:
     name: Upload dynamo performance stats for ${{ github.event.workflow_run.id }}, attempt ${{ github.event.workflow_run.run_attempt }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.1
         with:
           submodules: false
           fetch-depth: 1

aten/src/ATen/CMakeLists.txt

@@ -125,6 +125,7 @@ file(GLOB native_ao_sparse_h
             "native/ao_sparse/quantized/cpu/*.h")
 file(GLOB native_quantized_h "native/quantized/*.h" "native/quantized/cpu/*.h" "native/quantized/cudnn/*.h")
 file(GLOB native_cpu_h "native/cpu/*.h")
+file(GLOB native_utils_h "native/utils/*.h")
 
 file(GLOB native_cuda_cu "native/cuda/*.cu")
 file(GLOB native_cuda_cpp "native/cuda/*.cpp")
@@ -540,7 +541,7 @@ install(FILES "${CMAKE_CURRENT_BINARY_DIR}/cmake-exports/ATenConfig.cmake"
 
 set(INSTALL_HEADERS ${base_h} ${ATen_CORE_HEADERS})
 if(NOT INTERN_BUILD_MOBILE)
-  list(APPEND INSTALL_HEADERS ${native_h} ${native_cpu_h} ${native_ao_sparse_h} ${native_quantized_h} ${cuda_h} ${native_cuda_h} ${native_hip_h} ${cudnn_h} ${hip_h} ${mps_h} ${native_mps_h} ${miopen_h})
+  list(APPEND INSTALL_HEADERS ${native_h} ${native_cpu_h} ${native_ao_sparse_h} ${native_quantized_h} ${cuda_h} ${native_cuda_h} ${native_hip_h} ${cudnn_h} ${hip_h} ${mps_h} ${native_mps_h} ${native_utils_h} ${miopen_h})
   # Metal
   if(USE_PYTORCH_METAL_EXPORT)
     # Add files needed from exporting metal models(optimized_for_mobile)

aten/src/ATen/native/ConvUtils.h

@@ -389,8 +389,9 @@ static inline bool mkldnn_conv_use_channels_last(const at::Tensor& input, const
       (input_memory_format  == at::MemoryFormat::ChannelsLast) ||
       (weight_memory_format == at::MemoryFormat::ChannelsLast);
 
-  // TODO: add channels last 3d support
-  bool can_use_mkldnn_channels_last_3d = false;
+  bool can_use_mkldnn_channels_last_3d =
+      (input_memory_format  == at::MemoryFormat::ChannelsLast3d) ||
+      (weight_memory_format == at::MemoryFormat::ChannelsLast3d);
 
   return can_use_mkldnn_channels_last_2d || can_use_mkldnn_channels_last_3d;
 }

aten/src/ATen/native/Convolution.cpp

@@ -508,9 +508,6 @@ struct ConvParams {
     if (transposed && is_output_padding_big()) {
       return false;
     }
-    if (transposed && groups > 1 && at::symint::size<T>(input, 1) == groups) {
-      return false;
-    }
     if (input.device().is_cpu() && input.scalar_type() == kBFloat16 && mkldnn_bf16_device_check()) {
       return true;
     }

aten/src/ATen/native/Copy.cpp

@@ -253,7 +253,9 @@ static Tensor & copy_impl(Tensor & self, const Tensor & src, bool non_blocking)
       self.storage_offset() == src.storage_offset() &&
       self.strides().equals(src.strides()) &&
       self.sizes().equals(src.sizes()) &&
-      self.scalar_type() == src.scalar_type()
+      self.scalar_type() == src.scalar_type() &&
+      self.is_conj() == src.is_conj() &&
+      self.is_neg() == src.is_neg()
     );
   if (is_same_data) {
     return self;

aten/src/ATen/native/mkldnn/Conv.cpp

@@ -727,7 +727,7 @@ Tensor _mkldnn_convolution_transpose(
 
   if (bias.defined()) {
     const ideep::tensor b = itensor_from_tensor(bias);
-    ideep::convolution_transpose_forward::compute(
+    ideep::convolution_transpose_forward::compute_v3(
         x,
         w,
         b,
@@ -738,9 +738,10 @@ Tensor _mkldnn_convolution_transpose(
         padding_r(padding_expanded, output_padding_expanded),
         dilation.vec(),
         groups,
+        use_channels_last,
         op_attr);
   } else {
-    ideep::convolution_transpose_forward::compute(
+    ideep::convolution_transpose_forward::compute_v3(
         x,
         w,
         output_sizes,
@@ -750,6 +751,7 @@ Tensor _mkldnn_convolution_transpose(
         padding_r(padding_expanded, output_padding_expanded),
         dilation.vec(),
         groups,
+        use_channels_last,
         op_attr);
   }
   if (input.is_mkldnn()) {
@@ -988,7 +990,7 @@ Tensor mkldnn_convolution_transpose_backward_input(
     grad_input.resize_(input_size, memory_format);
     grad_x = itensor_from_tensor(grad_input);
   }
-  ideep::convolution_transpose_backward_data::compute(
+  ideep::convolution_transpose_backward_data::compute_v3(
       grad_y,
       w,
       input_size.vec(),
@@ -997,7 +999,8 @@ Tensor mkldnn_convolution_transpose_backward_input(
       padding.vec(),
       padding_r(padding, output_padding),
       dilation.vec(),
-      groups);
+      groups,
+      is_channels_last);
 
   if (grad_output.is_mkldnn()) {
     return MKLDNNTensor(grad_x, grad_output.options());
@@ -1024,7 +1027,7 @@ std::tuple<Tensor,Tensor> mkldnn_convolution_transpose_backward_weights(
 
   ideep::tensor grad_w, grad_b;
   if (bias_defined) {
-    ideep::convolution_transpose_backward_weights::compute(
+    ideep::convolution_transpose_backward_weights::compute_v3(
         x,
         grad_y,
         weight_size.vec(),
@@ -1034,9 +1037,10 @@ std::tuple<Tensor,Tensor> mkldnn_convolution_transpose_backward_weights(
         padding.vec(),
         padding_r(padding, output_padding),
         dilation.vec(),
-        groups);
+        groups,
+        is_channels_last);
   } else {
-    ideep::convolution_transpose_backward_weights::compute(
+    ideep::convolution_transpose_backward_weights::compute_v3(
         x,
         grad_y,
         weight_size.vec(),
@@ -1045,7 +1049,8 @@ std::tuple<Tensor,Tensor> mkldnn_convolution_transpose_backward_weights(
         padding.vec(),
         padding_r(padding, output_padding),
         dilation.vec(),
-        groups);
+        groups,
+        is_channels_last);
   }
 
   if (!is_channels_last) {
@@ -1061,18 +1066,21 @@ std::tuple<Tensor,Tensor> mkldnn_convolution_transpose_backward_weights(
 }
 
 std::tuple<Tensor, Tensor, Tensor> mkldnn_convolution_transpose_backward(
-    const Tensor& input, const Tensor& grad_output_t, const Tensor& weight,
+    const Tensor& input_t, const Tensor& grad_output_t, const Tensor& weight_t,
     IntArrayRef padding, IntArrayRef output_padding, IntArrayRef stride, IntArrayRef dilation, int64_t groups,
     std::array<bool,3> output_mask)
 {
-  bool is_channels_last = mkldnn_conv_use_channels_last(input, weight);
-  auto memory_format = mkldnn_convolution_memory_format(input.ndimension(), is_channels_last);
+  bool is_channels_last = mkldnn_conv_use_channels_last(input_t, weight_t);
+  auto memory_format = mkldnn_convolution_memory_format(input_t.ndimension(), is_channels_last);
   Tensor grad_output = grad_output_t.is_mkldnn() ? grad_output_t : grad_output_t.contiguous(memory_format);
+  auto input = input_t.is_mkldnn() ? input_t : input_t.contiguous(memory_format);
+  auto weight = weight_t.is_mkldnn() ? weight_t : weight_t.contiguous(memory_format);
   int64_t dim = input.ndimension() - 2;
   const auto padding_expanded = expand_param_if_needed(padding, "padding", dim);
   const auto stride_expanded = expand_param_if_needed(stride, "stride", dim);
   const auto dilation_expanded = expand_param_if_needed(dilation, "dilation", dim);
   const auto output_padding_expanded = expand_param_if_needed(output_padding, "output_padding", dim);
+
   Tensor grad_input, grad_weight, grad_bias;
   if (output_mask[0]) {
     grad_input = mkldnn_convolution_transpose_backward_input(

aten/src/ATen/native/mps/operations/Copy.mm

@@ -293,7 +293,8 @@ at::Tensor& mps_copy_(at::Tensor& dst, const at::Tensor& src, bool non_blocking)
     dst.resize_as_(src);
   }
 
-  TORCH_CHECK(dst.dim() >= src.dim());
+  TORCH_CHECK(
+      dst.dim() >= src.dim(), "Destination ", dst.sym_sizes(), " doesn't match the broadcast shape ", src.sym_sizes());
   if (dst.dim() > src.dim()) {
     needs_broadcasting = true;
   } else {

aten/src/ATen/native/mps/operations/Scalar.mm

@@ -16,15 +16,14 @@ namespace at::native {
 Scalar _local_scalar_dense_mps(const Tensor& self) {
   Scalar r;
 
+  auto output = at::empty_like(self, TensorOptions(kCPU));
+  mps::mps_copy_(output, self, false);
   AT_DISPATCH_ALL_TYPES_AND_COMPLEX_AND3(at::ScalarType::Half,
                                          at::ScalarType::Bool,
                                          at::ScalarType::BFloat16,
                                          self.scalar_type(),
                                          "_local_scalar_dense_mps",
                                          [&] {
-                                           Tensor output = at::empty({1}, TensorOptions(at::CPU(self.scalar_type())));
-
-                                           mps::mps_copy_(output, self, false);
                                            scalar_t value = *output.data_ptr<scalar_t>();
                                            r = Scalar(value);
                                          });

cmake/iOS.cmake

@@ -53,9 +53,9 @@ set(CMAKE_RANLIB ranlib CACHE FILEPATH "" FORCE)
 set(PKG_CONFIG_EXECUTABLE pkg-config CACHE FILEPATH "" FORCE)
 
 # Setup iOS platform unless specified manually with IOS_PLATFORM
-if(NOT DEFINED IOS_PLATFORM)
+if(NOT IOS_PLATFORM)
     set(IOS_PLATFORM "OS")
-endif(NOT DEFINED IOS_PLATFORM)
+endif(NOT IOS_PLATFORM)
 set(IOS_PLATFORM ${IOS_PLATFORM} CACHE STRING "Type of iOS Platform")
 
 # Check the platform selection and setup for developer root
@@ -118,9 +118,9 @@ set(CMAKE_FIND_LIBRARY_SUFFIXES ".dylib" ".so" ".a")
 # (where install_name_tool was hardcoded) and where CMAKE_INSTALL_NAME_TOOL isn't in the cache
 # and still cmake didn't fail in CMakeFindBinUtils.cmake (because it isn't rerun)
 # hardcode CMAKE_INSTALL_NAME_TOOL here to install_name_tool, so it behaves as it did before, Alex
-if(NOT DEFINED CMAKE_INSTALL_NAME_TOOL)
+if(NOT CMAKE_INSTALL_NAME_TOOL)
     find_program(CMAKE_INSTALL_NAME_TOOL install_name_tool)
-endif(NOT DEFINED CMAKE_INSTALL_NAME_TOOL)
+endif(NOT CMAKE_INSTALL_NAME_TOOL)
 
 # Setup iOS deployment target
 set(IOS_DEPLOYMENT_TARGET ${IOS_DEPLOYMENT_TARGET} CACHE STRING "Minimum iOS version")
@@ -130,17 +130,17 @@ set(IOS_DEPLOYMENT_TARGET ${IOS_DEPLOYMENT_TARGET} CACHE STRING "Minimum iOS ver
 exec_program(/usr/bin/xcode-select ARGS -print-path OUTPUT_VARIABLE CMAKE_XCODE_DEVELOPER_DIR)
 set(XCODE_POST_43_ROOT "${CMAKE_XCODE_DEVELOPER_DIR}/Platforms/${IOS_PLATFORM_LOCATION}/Developer")
 set(XCODE_PRE_43_ROOT "/Developer/Platforms/${IOS_PLATFORM_LOCATION}/Developer")
-if(NOT DEFINED CMAKE_IOS_DEVELOPER_ROOT)
+if(NOT CMAKE_IOS_DEVELOPER_ROOT)
     if(EXISTS ${XCODE_POST_43_ROOT})
         set(CMAKE_IOS_DEVELOPER_ROOT ${XCODE_POST_43_ROOT})
     elseif(EXISTS ${XCODE_PRE_43_ROOT})
         set(CMAKE_IOS_DEVELOPER_ROOT ${XCODE_PRE_43_ROOT})
     endif(EXISTS ${XCODE_POST_43_ROOT})
-endif(NOT DEFINED CMAKE_IOS_DEVELOPER_ROOT)
+endif(NOT CMAKE_IOS_DEVELOPER_ROOT)
 set(CMAKE_IOS_DEVELOPER_ROOT ${CMAKE_IOS_DEVELOPER_ROOT} CACHE PATH "Location of iOS Platform")
 
 # Find and use the most recent iOS sdk unless specified manually with CMAKE_IOS_SDK_ROOT
-if(NOT DEFINED CMAKE_IOS_SDK_ROOT)
+if(NOT CMAKE_IOS_SDK_ROOT)
     file(GLOB _CMAKE_IOS_SDKS "${CMAKE_IOS_DEVELOPER_ROOT}/SDKs/*")
     if(_CMAKE_IOS_SDKS)
         list(SORT _CMAKE_IOS_SDKS)
@@ -150,7 +150,7 @@ if(NOT DEFINED CMAKE_IOS_SDK_ROOT)
         message(FATAL_ERROR "No iOS SDK's found in default search path ${CMAKE_IOS_DEVELOPER_ROOT}. Manually set CMAKE_IOS_SDK_ROOT or install the iOS SDK.")
     endif(_CMAKE_IOS_SDKS)
     message(STATUS "Toolchain using default iOS SDK: ${CMAKE_IOS_SDK_ROOT}")
-endif(NOT DEFINED CMAKE_IOS_SDK_ROOT)
+endif(NOT CMAKE_IOS_SDK_ROOT)
 set(CMAKE_IOS_SDK_ROOT ${CMAKE_IOS_SDK_ROOT} CACHE PATH "Location of the selected iOS SDK")
 
 # Set the sysroot default to the most recent SDK

docs/Makefile

@@ -18,8 +18,8 @@ figures:
 	@$(PYCMD) source/scripts/build_quantization_configs.py
 
 onnx:
-	@$(PYCMD) source/scripts/onnx/build_onnx_supported_aten_op_csv_table.py
-	@$(PYCMD) source/scripts/onnx/build_onnx_diagnostics_rules_md.py $(SOURCEDIR)/generated/onnx_diagnostics_rules
+	@$(PYCMD) source/scripts/onnx/build_onnx_torchscript_supported_aten_op_csv_table.py
+	@$(PYCMD) source/scripts/onnx/build_onnx_dynamo_diagnostics_rules_md.py $(SOURCEDIR)/generated/onnx_dynamo_diagnostics_rules
 
 opset:
 	@$(PYCMD) source/scripts/build_opsets.py

docs/source/autograd.rst

@@ -179,6 +179,7 @@ Tensor autograd functions
    torch.Tensor.detach
    torch.Tensor.detach_
    torch.Tensor.register_hook
+   torch.Tensor.register_post_accumulate_grad_hook
    torch.Tensor.retain_grad
 
 :hidden:`Function`

docs/source/export.rst

@@ -1,10 +1,551 @@
+.. _torch.export:
+
 torch.export
 =====================
 
-.. TODO: Add torch.export() tutorial here.
-
 .. warning::
-    This feature is a prototype and may have compatibility breaking changes in the future.
+    This feature is a prototype under active development and there WILL BE
+    BREAKING CHANGES in the future.
+
+
+Overview
+--------
+
+:func:`torch.export.export` takes an arbitrary Python callable (a
+:class:`torch.nn.Module`, a function or a method) and produces a traced graph
+representing only the Tensor computation of the function in an Ahead-of-Time
+(AOT) fashion, which can subsequently be executed with different outputs or
+serialized.
+
+::
+
+    import torch
+    from torch.export import export
+
+    def f(x: torch.Tensor, y: torch.Tensor) -> torch.Tensor:
+        a = torch.sin(x)
+        b = torch.cos(y)
+        return a + b
+
+    example_args = (torch.randn(10, 10), torch.randn(10, 10))
+
+    exported_program: torch.export.ExportedProgram = export(
+        f, args=example_args
+    )
+    print(exported_program)
+
+.. code-block::
+
+    ExportedProgram:
+        class GraphModule(torch.nn.Module):
+            def forward(self, arg0_1: f32[10, 10], arg1_1: f32[10, 10]):
+                # code: a = torch.sin(x)
+                sin: f32[10, 10] = torch.ops.aten.sin.default(arg0_1);
+
+                # code: b = torch.cos(y)
+                cos: f32[10, 10] = torch.ops.aten.cos.default(arg1_1);
+
+                # code: return a + b
+                add: f32[10, 10] = torch.ops.aten.add.Tensor(sin, cos);
+                return (add,)
+
+        Graph signature: ExportGraphSignature(
+            parameters=[],
+            buffers=[],
+            user_inputs=['arg0_1', 'arg1_1'],
+            user_outputs=['add'],
+            inputs_to_parameters={},
+            inputs_to_buffers={},
+            buffers_to_mutate={},
+            backward_signature=None,
+            assertion_dep_token=None,
+        )
+        Range constraints: {}
+        Equality constraints: []
+
+``torch.export`` produces a clean intermediate representation (IR) with the
+following invariants. More specifications about the IR can be found here (coming
+soon!).
+
+* **Soundness**: It is guaranteed to be a sound representation of the original
+  program, and maintains the same calling conventions of the original program.
+
+* **Normalized**: There are no Python semantics within the graph. Submodules
+  from the original programs are inlined to form one fully flattened
+  computational graph.
+
+* **Defined Operator Set**: The graph produced contains only a small defined
+  :ref:`Core ATen IR <torch.compiler_ir>` opset and registered custom
+  operators.
+
+* **Graph properties**: The graph is purely functional, meaning it does not
+  contain operations with side effects such as mutations or aliasing. It does
+  not mutate any intermediate values, parameters, or buffers.
+
+* **Metadata**: The graph contains metadata captured during tracing, such as a
+  stacktrace from user's code.
+
+Under the hood, ``torch.export`` leverages the following latest technologies:
+
+* **TorchDynamo (torch._dynamo)** is an internal API that uses a CPython feature
+  called the Frame Evaluation API to safely trace PyTorch graphs. This
+  provides a massively improved graph capturing experience, with much fewer
+  rewrites needed in order to fully trace the PyTorch code.
+
+* **AOT Autograd** provides a functionalized PyTorch graph and ensures the graph
+  is decomposed/lowered to the small defined Core ATen operator set.
+
+* **Torch FX (torch.fx)** is the underlying representation of the graph,
+  allowing flexible Python-based transformations.
+
+
+Existing frameworks
+^^^^^^^^^^^^^^^^^^^
+
+:func:`torch.compile` also utilizes the same PT2 stack as ``torch.export``, but
+is slightly different:
+
+* **JIT vs. AOT**: :func:`torch.compile` is a JIT compiler whereas
+  which is not intended to be used to produce compiled artifacts outside of
+  deployment.
+
+* **Partial vs. Full Graph Capture**: When :func:`torch.compile` runs into an
+  untraceable part of a model, it will "graph break" and fall back to running
+  the program in the eager Python runtime. In comparison, ``torch.export`` aims
+  to get a full graph representation of a PyTorch model, so it will error out
+  when something untraceable is reached. Since ``torch.export`` produces a full
+  graph disjoint from any Python features or runtime, this graph can then be
+  saved, loaded, and run in different environments and languages.
+
+* **Usability tradeoff**: Since :func:`torch.compile` is able to fallback to the
+  Python runtime whenever it reaches something untraceable, it is a lot more
+  flexible. ``torch.export`` will instead require users to provide more
+  information or rewrite their code to make it traceable.
+
+Compared to :func:`torch.fx.symbolic_trace`, ``torch.export`` traces using
+TorchDynamo which operates at the Python bytecode level, giving it the ability
+to trace arbitrary Python constructs not limited by what Python operator
+overloading supports. Additionally, ``torch.export`` keeps fine-grained track of
+tensor metadata, so that conditionals on things like tensor shapes do not
+fail tracing. In general, ``torch.export`` is expected to work on more user
+programs, and produce lower-level graphs (at the ``torch.ops.aten`` operator
+level). Note that users can still use :func:`torch.fx.symbolic_trace` as a
+preprocessing step before ``torch.export``.
+
+Compared to :func:`torch.jit.script`, ``torch.export`` does not capture Python
+control flow or data structures, but it supports more Python language features
+than TorchScript (as it is easier to have comprehensive coverage over Python
+bytecodes). The resulting graphs are simpler and only have straight line control
+flow (except for explicit control flow operators).
+
+Compared to :func:`torch.jit.trace`, ``torch.export`` is sound: it is able to
+trace code that performs integer computation on sizes and records all of the
+side-conditions necessary to show that a particular trace is valid for other
+inputs.
+
+
+Exporting a PyTorch Model
+-------------------------
+
+An Example
+^^^^^^^^^^
+
+The main entrypoint is through :func:`torch.export.export`, which takes a
+callable (:class:`torch.nn.Module`, function, or method) and sample inputs, and
+captures the computation graph into an :class:`torch.export.ExportedProgram`. An
+example:
+
+::
+
+    import torch
+    from torch.export import export
+
+    # Simple module for demonstration
+    class M(torch.nn.Module):
+        def __init__(self) -> None:
+            super().__init__()
+            self.conv = torch.nn.Conv2d(
+                in_channels=3, out_channels=16, kernel_size=3, padding=1
+            )
+            self.relu = torch.nn.ReLU()
+            self.maxpool = torch.nn.MaxPool2d(kernel_size=3)
+
+        def forward(self, x: torch.Tensor, *, constant=None) -> torch.Tensor:
+            a = self.conv(x)
+            a.add_(constant)
+            return self.maxpool(self.relu(a))
+
+    example_args = (torch.randn(1, 3, 256, 256),)
+    example_kwargs = {"constant": torch.ones(1, 16, 256, 256)}
+
+    exported_program: torch.export.ExportedProgram = export(
+        M(), args=example_args, kwargs=example_kwargs
+    )
+    print(exported_program)
+
+.. code-block::
+
+    ExportedProgram:
+        class GraphModule(torch.nn.Module):
+            def forward(self, arg0_1: f32[16, 3, 3, 3], arg1_1: f32[16], arg2_1: f32[1, 3, 256, 256], arg3_1: f32[1, 16, 256, 256]):
+
+                # code: a = self.conv(x)
+                convolution: f32[1, 16, 256, 256] = torch.ops.aten.convolution.default(
+                    arg2_1, arg0_1, arg1_1, [1, 1], [1, 1], [1, 1], False, [0, 0], 1
+                );
+
+                # code: a.add_(constant)
+                add: f32[1, 16, 256, 256] = torch.ops.aten.add.Tensor(convolution, arg3_1);
+
+                # code: return self.maxpool(self.relu(a))
+                relu: f32[1, 16, 256, 256] = torch.ops.aten.relu.default(add);
+                max_pool2d_with_indices = torch.ops.aten.max_pool2d_with_indices.default(
+                    relu, [3, 3], [3, 3]
+                );
+                getitem: f32[1, 16, 85, 85] = max_pool2d_with_indices[0];
+                return (getitem,)
+
+        Graph signature: ExportGraphSignature(
+            parameters=['L__self___conv.weight', 'L__self___conv.bias'],
+            buffers=[],
+            user_inputs=['arg2_1', 'arg3_1'],
+            user_outputs=['getitem'],
+            inputs_to_parameters={
+                'arg0_1': 'L__self___conv.weight',
+                'arg1_1': 'L__self___conv.bias',
+            },
+            inputs_to_buffers={},
+            buffers_to_mutate={},
+            backward_signature=None,
+            assertion_dep_token=None,
+        )
+        Range constraints: {}
+        Equality constraints: []
+
+Inspecting the ``ExportedProgram``, we can note the following:
+
+* The :class:`torch.fx.Graph` contains the computation graph of the original
+  program, along with records of the original code for easy debugging.
+
+* The graph contains only ``torch.ops.aten`` operators found in the
+  :ref:`Core ATen IR <torch.compiler_ir>` opset and custom operators, and is
+  fully functional, without any inplace operators such as ``torch.add_``.
+
+* The parameters (weight and bias to conv) are lifted as inputs to the graph,
+  resulting in no ``get_attr`` nodes in the graph, which previously existed in
+  the result of :func:`torch.fx.symbolic_trace`.
+
+* The :class:`torch.export.ExportGraphSignature` models the input and output
+  signature, along with specifying which inputs are parameters.
+
+* The resulting shape and dtype of tensors produced by each node in the graph is
+  noted. For example, the ``convolution`` node will result in a tensor of dtype
+  ``torch.float32`` and shape (1, 16, 256, 256).
+
+
+Expressing Dynamism
+^^^^^^^^^^^^^^^^^^^
+
+By default ``torch.export`` will trace the program assuming all input shapes are
+**static**, and specializing the exported program to those dimensions. However,
+some dimensions, such as a batch dimension, can be dynamic and vary from run to
+run. Such dimensions must be marked dynamic using the
+:func:`torch.export.dynamic_dim` API, and passed into
+:func:`torch.export.export` through the ``constraints`` argument. An example:
+
+::
+
+    import torch
+    from torch.export import export, dynamic_dim
+
+    class M(torch.nn.Module):
+        def __init__(self):
+            super().__init__()
+
+            self.branch1 = torch.nn.Sequential(
+                torch.nn.Linear(64, 32), torch.nn.ReLU()
+            )
+            self.branch2 = torch.nn.Sequential(
+                torch.nn.Linear(128, 64), torch.nn.ReLU()
+            )
+            self.buffer = torch.ones(32)
+
+        def forward(self, x1, x2):
+            out1 = self.branch1(x1)
+            out2 = self.branch2(x2)
+            return (out1 + self.buffer, out2)
+
+    example_args = (torch.randn(32, 64), torch.randn(32, 128))
+    constraints = [
+        # First dimension of each input is a dynamic batch size
+        dynamic_dim(example_args[0], 0),
+        dynamic_dim(example_args[1], 0),
+        # The dynamic batch size between the inputs are equal
+        dynamic_dim(example_args[0], 0) == dynamic_dim(example_args[1], 0),
+    ]
+
+    exported_program: torch.export.ExportedProgram = export(
+      M(), args=example_args, constraints=constraints
+    )
+    print(exported_program)
+
+.. code-block::
+
+    ExportedProgram:
+        class GraphModule(torch.nn.Module):
+            def forward(self, arg0_1: f32[32, 64], arg1_1: f32[32], arg2_1: f32[64, 128], arg3_1: f32[64], arg4_1: f32[32], arg5_1: f32[s0, 64], arg6_1: f32[s0, 128]):
+
+                # code: out1 = self.branch1(x1)
+                permute: f32[64, 32] = torch.ops.aten.permute.default(arg0_1, [1, 0]);
+                addmm: f32[s0, 32] = torch.ops.aten.addmm.default(arg1_1, arg5_1, permute);
+                relu: f32[s0, 32] = torch.ops.aten.relu.default(addmm);
+
+                # code: out2 = self.branch2(x2)
+                permute_1: f32[128, 64] = torch.ops.aten.permute.default(arg2_1, [1, 0]);
+                addmm_1: f32[s0, 64] = torch.ops.aten.addmm.default(arg3_1, arg6_1, permute_1);
+                relu_1: f32[s0, 64] = torch.ops.aten.relu.default(addmm_1);  addmm_1 = None
+
+                # code: return (out1 + self.buffer, out2)
+                add: f32[s0, 32] = torch.ops.aten.add.Tensor(relu, arg4_1);
+                return (add, relu_1)
+
+        Graph signature: ExportGraphSignature(
+            parameters=[
+                'branch1.0.weight',
+                'branch1.0.bias',
+                'branch2.0.weight',
+                'branch2.0.bias',
+            ],
+            buffers=['L__self___buffer'],
+            user_inputs=['arg5_1', 'arg6_1'],
+            user_outputs=['add', 'relu_1'],
+            inputs_to_parameters={
+                'arg0_1': 'branch1.0.weight',
+                'arg1_1': 'branch1.0.bias',
+                'arg2_1': 'branch2.0.weight',
+                'arg3_1': 'branch2.0.bias',
+            },
+            inputs_to_buffers={'arg4_1': 'L__self___buffer'},
+            buffers_to_mutate={},
+            backward_signature=None,
+            assertion_dep_token=None,
+        )
+        Range constraints: {s0: RangeConstraint(min_val=2, max_val=9223372036854775806)}
+        Equality constraints: [(InputDim(input_name='arg5_1', dim=0), InputDim(input_name='arg6_1', dim=0))]
+
+Some additional things to note:
+
+* Through the :func:`torch.export.dynamic_dim` API, we specified the first
+  dimension of each input to be dynamic. Looking at the inputs ``arg5_1`` and
+  ``arg6_1``, they have a symbolic shape of (s0, 64) and (s0, 128), instead of
+  the (32, 64) and (32, 128) shaped tensors that we passed in as example inputs.
+  ``s0`` is a symbol representing that this dimension can be a range
+  of values.
+
+* ``exported_program.range_constraints`` describes the ranges of each symbol
+  appearing in the graph. In this case, we see that ``s0`` has the range
+  [2, inf]. For technical reasons that are difficult to explain here, they are
+  assumed to be not 0 or 1. This is not a bug, and does not necessarily mean
+  that the exported program will not work for dimensions 0 or 1. See
+  `The 0/1 Specialization Problem <https://docs.google.com/document/d/16VPOa3d-Liikf48teAOmxLc92rgvJdfosIy-yoT38Io/edit?fbclid=IwAR3HNwmmexcitV0pbZm_x1a4ykdXZ9th_eJWK-3hBtVgKnrkmemz6Pm5jRQ#heading=h.ez923tomjvyk>`_
+  for an in-depth discussion of this topic.
+
+* ``exported_program.equality_constraints`` describes which dimensions are
+  required to be equal. Since we specified in the constraints that the first
+  dimension of each argument is equivalent,
+  (``dynamic_dim(example_args[0], 0) == dynamic_dim(example_args[1], 0)``),
+  we see in the equality constraints the tuple specifying that ``arg5_1``
+  dimension 0 and ``arg6_1`` dimension 0 are equal.
+
+
+Serialization
+^^^^^^^^^^^^^
+
+To save the ``ExportedProgram``, users can use the :func:`torch.export.save` and
+:func:`torch.export.load` APIs. A convention is to save the ``ExportedProgram``
+using a ``.pt2`` file extension.
+
+An example:
+
+::
+
+    import torch
+    import io
+
+    class MyModule(torch.nn.Module):
+        def forward(self, x):
+            return x + 10
+
+    exported_program = torch.export.export(MyModule(), torch.randn(5))
+
+    torch.export.save(exported_program, 'exported_program.pt2')
+    saved_exported_program = torch.export.load('exported_program.pt2')
+
+
+Specialization
+^^^^^^^^^^^^^^
+
+Input shapes
+~~~~~~~~~~~~
+
+As mentioned before, by default, ``torch.export`` will trace the program
+specializing on the input tensors' shapes, unless a dimension is specified as
+dynamic via the :func:`torch.export.dynamic_dim` API. This means that if there
+exists shape-dependent control flow, ``torch.export`` will specialize on the
+branch that is being taken with the given sample inputs. For example:
+
+::
+
+    import torch
+    from torch.export import export
+
+    def fn(x):
+        if x.shape[0] > 5:
+            return x + 1
+        else:
+            return x - 1
+
+    example_inputs = (torch.rand(10, 2),)
+    exported_program = export(fn, example_inputs)
+    print(exported_program)
+
+.. code-block::
+
+    ExportedProgram:
+        class GraphModule(torch.nn.Module):
+            def forward(self, arg0_1: f32[10, 2]):
+                add: f32[10, 2] = torch.ops.aten.add.Tensor(arg0_1, 1);
+                return (add,)
+
+The conditional of (``x.shape[0] > 5``) does not appear in the
+``ExportedProgram`` because the example inputs have the static
+shape of (10, 2). Since ``torch.export`` specializes on the inputs' static
+shapes, the else branch (``x - 1``) will never be reached. To preserve the dynamic
+branching behavior based on the shape of a tensor in the traced graph,
+:func:`torch.export.dynamic_dim` will need to be used to specify the dimension
+of the input tensor (``x.shape[0]``) to be dynamic, and the source code will
+need to be :ref:`rewritten <Data/Shape-Dependent Control Flow>`.
+
+Non-tensor inputs
+~~~~~~~~~~~~~~~~~
+
+``torch.export`` also specializes the traced graph based on the values of inputs
+that are not ``torch.Tensor``, such as ``int``, ``float``, ``bool``, and ``str``.
+However, we will likely change this in the near future to not specialize on
+inputs of primitive types.
+
+For example:
+
+::
+
+    import torch
+    from torch.export import export
+
+    def fn(x: torch.Tensor, const: int, times: int):
+        for i in range(times):
+            x = x + const
+        return x
+
+    example_inputs = (torch.rand(2, 2), 1, 3)
+    exported_program = export(fn, example_inputs)
+    print(exported_program)
+
+.. code-block::
+
+    ExportedProgram:
+        class GraphModule(torch.nn.Module):
+            def forward(self, arg0_1: f32[2, 2], arg1_1, arg2_1):
+                add: f32[2, 2] = torch.ops.aten.add.Tensor(arg0_1, 1);
+                add_1: f32[2, 2] = torch.ops.aten.add.Tensor(add, 1);
+                add_2: f32[2, 2] = torch.ops.aten.add.Tensor(add_1, 1);
+                return (add_2,)
+
+Because integers are specialized, the ``torch.ops.aten.add.Tensor`` operations
+are all computed with the inlined constant ``1``, rather than ``arg1_1``.
+Additionally, the ``times`` iterator used in the ``for`` loop is also "inlined"
+in the graph through the 3 repeated ``torch.ops.aten.add.Tensor`` calls, and the
+input ``arg2_1`` is never used.
+
+
+Limitations of torch.export
+---------------------------
+
+Graph Breaks
+^^^^^^^^^^^^
+
+As ``torch.export`` is a one-shot process for capturing a computation graph from
+a PyTorch program, it might ultimately run into untraceable parts of programs as
+it is nearly impossible to support tracing all PyTorch and Python features. In
+the case of ``torch.compile``, an unsupported operation will cause a "graph
+break" and the unsupported operation will be run with default Python evaluation.
+In contrast, ``torch.export`` will require users to provide additional
+information or rewrite parts of their code to make it traceable. As the
+tracing is based on TorchDynamo, which evaluates at the Python
+bytecode level, there will be significantly fewer rewrites required compared to
+previous tracing frameworks.
+
+When a graph break is encountered, :ref:`ExportDB <torch.export_db>` is a great
+resource for learning about the kinds of programs that are supported and
+unsupported, along with ways to rewrite programs to make them traceable.
+
+.. _Data/Shape-Dependent Control Flow:
+
+Data/Shape-Dependent Control Flow
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Graph breaks can also be encountered on data-dependent control flow (``if
+x.shape[0] > 2``) when shapes are not being specialized, as a tracing compiler cannot
+possibly deal with without generating code for a combinatorially exploding
+number of paths. In such cases, users will need to rewrite their code using
+special control flow operators (coming soon!).
+
+Data-Dependent Accesses
+^^^^^^^^^^^^^^^^^^^^^^^
+
+Data dependent behavior such as using the value inside of a tensor to construct
+another tensor, or using the value of a tensor to slice into another tensor, is
+also something the tracer cannot fully determine. Users will need to rewrite
+their code using the inline constraint APIs
+:func:`torch.export.constrain_as_size` and
+:func:`torch.export.constrain_as_value`.
+
+Missing Meta Kernels for Operators
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+When tracing, a META implementation (or "meta kernel") is required for all
+operators. This is used to reason about the input/output shapes for this
+operator.
+
+Note that the official API for registering custom meta kernels for custom ops is
+currently undergoing development. While the final API is being refined, you can
+refer to the documentation `here <https://docs.google.com/document/d/1GgvOe7C8_NVOMLOCwDaYV1mXXyHMXY7ExoewHqooxrs/edit#heading=h.64r4npvq0w0>`_.
+
+In the unfortunate case where your model uses an ATen operator that is does not
+have a meta kernel implementation yet, please file an issue.
+
+
+Read More
+---------
+
+.. toctree::
+   :caption: Additional Links for Export Users
+   :maxdepth: 1
+
+   torch.compiler_transformations
+   torch.compiler_ir
+   generated/exportdb/index
+
+.. toctree::
+   :caption: Deep Dive for PyTorch Developers
+   :maxdepth: 1
+
+   torch.compiler_deepdive
+   torch.compiler_dynamic_shapes
+   torch.compiler_fake_tensor
+
+
+API Reference
+-------------
 
 .. automodule:: torch.export
 .. autofunction:: export
@@ -24,10 +565,3 @@ torch.export
 .. autoclass:: ArgumentSpec
 .. autoclass:: ModuleCallSignature
 .. autoclass:: ModuleCallEntry
-
-
-.. toctree::
-   :glob:
-   :maxdepth: 1
-
-   generated/exportdb/index

docs/source/index.rst

@@ -94,7 +94,6 @@ Features described in this documentation are classified by release status:
    profiler
    nn.init
    onnx
-   onnx_diagnostics
    optim
    complex_numbers
    ddp_comm_hooks

docs/source/name_inference.rst

@@ -185,6 +185,7 @@ If you don't see an operation listed here, but it would help your use case, plea
    :meth:`Tensor.reciprocal_`,None
    :meth:`Tensor.refine_names`,See documentation
    :meth:`Tensor.register_hook`,None
+   :meth:`Tensor.register_post_accumulate_grad_hook`,None
    :meth:`Tensor.rename`,See documentation
    :meth:`Tensor.rename_`,See documentation
    :attr:`Tensor.requires_grad`,None

docs/source/onnx.rst

@@ -1,745 +1,64 @@
 torch.onnx
 ==========
 
-.. contents:: :local:
-
-.. automodule:: torch.onnx
+Overview
+--------
 
 `Open Neural Network eXchange (ONNX) <https://onnx.ai/>`_ is an open standard
-format for representing machine learning models. The torch.onnx module can export
-PyTorch models to ONNX. The model can then be consumed by any of the many
-`runtimes that support ONNX <https://onnx.ai/supported-tools.html#deployModel>`_.
+format for representing machine learning models. The ``torch.onnx`` module captures the computation graph from a
+native PyTorch :class:`torch.nn.Module` model and converts it into an
+`ONNX graph <https://github.com/onnx/onnx/blob/main/docs/IR.md>`_.
 
-Example: AlexNet from PyTorch to ONNX
--------------------------------------
+The exported model can be consumed by any of the many
+`runtimes that support ONNX <https://onnx.ai/supported-tools.html#deployModel>`_, including
+Microsoft's `ONNX Runtime <https://www.onnxruntime.ai>`_.
 
-Here is a simple script which exports a pretrained AlexNet to an ONNX file named ``alexnet.onnx``.
-The call to ``torch.onnx.export`` runs the model once to trace its execution and then exports the
-traced model to the specified file::
+**There are two flavors of ONNX exporter API that you can use, as listed below:**
 
-    import torch
-    import torchvision
+TorchDynamo-based ONNX Exporter
+-------------------------------
 
-    dummy_input = torch.randn(10, 3, 224, 224, device="cuda")
-    model = torchvision.models.alexnet(pretrained=True).cuda()
+*The TorchDynamo-based ONNX exporter is the newest (and Beta) exporter for PyTorch 2.0 and newer*
 
-    # Providing input and output names sets the display names for values
-    # within the model's graph. Setting these does not change the semantics
-    # of the graph; it is only for readability.
-    #
-    # The inputs to the network consist of the flat list of inputs (i.e.
-    # the values you would pass to the forward() method) followed by the
-    # flat list of parameters. You can partially specify names, i.e. provide
-    # a list here shorter than the number of inputs to the model, and we will
-    # only set that subset of names, starting from the beginning.
-    input_names = [ "actual_input_1" ] + [ "learned_%d" % i for i in range(16) ]
-    output_names = [ "output1" ]
+TorchDynamo engine is leveraged to hook into Python's frame evaluation API and dynamically rewrite its
+bytecode into an FX Graph. The resulting FX Graph is then polished before it is finally translated into an
+ONNX graph.
 
-    torch.onnx.export(model, dummy_input, "alexnet.onnx", verbose=True, input_names=input_names, output_names=output_names)
+The main advantage of this approach is that the `FX graph <https://pytorch.org/docs/stable/fx.html>`_ is captured using
+bytecode analysis that preserves the dynamic nature of the model instead of using traditional static tracing techniques.
 
-The resulting ``alexnet.onnx`` file contains a binary `protocol buffer <https://developers.google.com/protocol-buffers/>`_
-which contains both the network structure and parameters of the model you exported
-(in this case, AlexNet).  The argument ``verbose=True`` causes the
-exporter to print out a human-readable representation of the model::
+:doc:`Learn more about the TorchDynamo-based ONNX Exporter <onnx_dynamo>`
 
-    # These are the inputs and parameters to the network, which have taken on
-    # the names we specified earlier.
-    graph(%actual_input_1 : Float(10, 3, 224, 224)
-          %learned_0 : Float(64, 3, 11, 11)
-          %learned_1 : Float(64)
-          %learned_2 : Float(192, 64, 5, 5)
-          %learned_3 : Float(192)
-          # ---- omitted for brevity ----
-          %learned_14 : Float(1000, 4096)
-          %learned_15 : Float(1000)) {
-      # Every statement consists of some output tensors (and their types),
-      # the operator to be run (with its attributes, e.g., kernels, strides,
-      # etc.), its input tensors (%actual_input_1, %learned_0, %learned_1)
-      %17 : Float(10, 64, 55, 55) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[11, 11], pads=[2, 2, 2, 2], strides=[4, 4]](%actual_input_1, %learned_0, %learned_1), scope: AlexNet/Sequential[features]/Conv2d[0]
-      %18 : Float(10, 64, 55, 55) = onnx::Relu(%17), scope: AlexNet/Sequential[features]/ReLU[1]
-      %19 : Float(10, 64, 27, 27) = onnx::MaxPool[kernel_shape=[3, 3], pads=[0, 0, 0, 0], strides=[2, 2]](%18), scope: AlexNet/Sequential[features]/MaxPool2d[2]
-      # ---- omitted for brevity ----
-      %29 : Float(10, 256, 6, 6) = onnx::MaxPool[kernel_shape=[3, 3], pads=[0, 0, 0, 0], strides=[2, 2]](%28), scope: AlexNet/Sequential[features]/MaxPool2d[12]
-      # Dynamic means that the shape is not known. This may be because of a
-      # limitation of our implementation (which we would like to fix in a
-      # future release) or shapes which are truly dynamic.
-      %30 : Dynamic = onnx::Shape(%29), scope: AlexNet
-      %31 : Dynamic = onnx::Slice[axes=[0], ends=[1], starts=[0]](%30), scope: AlexNet
-      %32 : Long() = onnx::Squeeze[axes=[0]](%31), scope: AlexNet
-      %33 : Long() = onnx::Constant[value={9216}](), scope: AlexNet
-      # ---- omitted for brevity ----
-      %output1 : Float(10, 1000) = onnx::Gemm[alpha=1, beta=1, broadcast=1, transB=1](%45, %learned_14, %learned_15), scope: AlexNet/Sequential[classifier]/Linear[6]
-      return (%output1);
-    }
+TorchScript-based ONNX Exporter
+-------------------------------
 
-You can also verify the output using the `ONNX <https://github.com/onnx/onnx/>`_ library,
-which you can install using ``pip``::
+*The TorchScript-based ONNX exporter is available since PyTorch 1.2.0*
 
-    pip install onnx
+`TorchScript <https://pytorch.org/docs/stable/jit.html>`_ is leveraged to trace (through :func:`torch.jit.trace`)
+the model and capture a static computation graph.
 
-Then, you can run::
+As a consequence, the resulting graph has a couple limitations:
 
-    import onnx
+* It does not record any control-flow, like if-statements or loops;
+* Does not handle nuances between ``training`` and ``eval`` mode;
+* Does not truly handle dynamic inputs
 
-    # Load the ONNX model
-    model = onnx.load("alexnet.onnx")
+As an attempt to support the static tracing limitations, the exporter also supports TorchScript scripting
+(through :func:`torch.jit.script`), which adds support for data-dependent control-flow, for example. However, TorchScript
+itself is a subset of the Python language, so not all features in Python are supported, such as in-place operations.
 
-    # Check that the model is well formed
-    onnx.checker.check_model(model)
+:doc:`Learn more about the TorchScript-based ONNX Exporter <onnx_torchscript>`
 
-    # Print a human readable representation of the graph
-    print(onnx.helper.printable_graph(model.graph))
-
-You can also run the exported model with one of the many
-`runtimes that support ONNX <https://onnx.ai/supported-tools.html#deployModel>`_.
-For example after installing `ONNX Runtime <https://www.onnxruntime.ai>`_, you can
-load and run the model::
-
-    import onnxruntime as ort
-    import numpy as np
-
-    ort_session = ort.InferenceSession("alexnet.onnx")
-
-    outputs = ort_session.run(
-        None,
-        {"actual_input_1": np.random.randn(10, 3, 224, 224).astype(np.float32)},
-    )
-    print(outputs[0])
-
-Here is a more involved `tutorial on exporting a model and running it with ONNX Runtime <https://pytorch.org/tutorials/advanced/super_resolution_with_onnxruntime.html>`_.
-
-.. _tracing-vs-scripting:
-
-Tracing vs Scripting
---------------------
-
-Internally, :func:`torch.onnx.export()` requires a :class:`torch.jit.ScriptModule` rather than
-a :class:`torch.nn.Module`. If the passed-in model is not already a ``ScriptModule``,
-``export()`` will use *tracing* to convert it to one:
-
-.. TODO(justinchuby): Add a word on recommending tracing over scripting for most use cases.
-
-* **Tracing**: If ``torch.onnx.export()`` is called with a Module that is not already a
-  ``ScriptModule``, it first does the equivalent of :func:`torch.jit.trace`, which executes the model
-  once with the given ``args`` and records all operations that happen during that execution. This
-  means that if your model is dynamic, e.g., changes behavior depending on input data, the exported
-  model will *not* capture this dynamic behavior.
-  We recommend examining the exported model and making sure the operators look
-  reasonable. Tracing will unroll loops and if statements, exporting a static graph that is exactly
-  the same as the traced run. If you want to export your model with dynamic control flow, you will
-  need to use *scripting*.
-
-* **Scripting**: Compiling a model via scripting preserves dynamic control flow and is valid for inputs
-  of different sizes. To use scripting:
-
-  * Use :func:`torch.jit.script` to produce a ``ScriptModule``.
-  * Call ``torch.onnx.export()`` with the ``ScriptModule`` as the model. The ``args`` are still required,
-    but they will be used internally only to produce example outputs, so that the types and shapes of the
-    outputs can be captured. No tracing will be performed.
-
-See `Introduction to TorchScript <https://pytorch.org/tutorials/beginner/Intro_to_TorchScript_tutorial.html>`_
-and `TorchScript <jit.html>`_ for more details, including how to compose tracing and scripting to suit the
-particular requirements of different models.
-
-
-Avoiding Pitfalls
------------------
-
-Avoid NumPy and built-in Python types
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
-
-PyTorch models can be written using NumPy or Python types and functions, but
-during :ref:`tracing<tracing-vs-scripting>`, any variables of NumPy or Python
-types (rather than torch.Tensor) are converted to constants, which will produce
-the wrong result if those values should change depending on the inputs.
-
-For example, rather than using numpy functions on numpy.ndarrays: ::
-
-    # Bad! Will be replaced with constants during tracing.
-    x, y = np.random.rand(1, 2), np.random.rand(1, 2)
-    np.concatenate((x, y), axis=1)
-
-Use torch operators on torch.Tensors: ::
-
-    # Good! Tensor operations will be captured during tracing.
-    x, y = torch.randn(1, 2), torch.randn(1, 2)
-    torch.cat((x, y), dim=1)
-
-
-And rather than use :func:`torch.Tensor.item` (which converts a Tensor to a Python
-built-in number): ::
-
-    # Bad! y.item() will be replaced with a constant during tracing.
-    def forward(self, x, y):
-        return x.reshape(y.item(), -1)
-
-Use torch's support for implicit casting of single-element tensors: ::
-
-    # Good! y will be preserved as a variable during tracing.
-    def forward(self, x, y):
-        return x.reshape(y, -1)
-
-Avoid Tensor.data
-^^^^^^^^^^^^^^^^^
-
-Using the Tensor.data field can produce an incorrect trace and therefore an incorrect ONNX graph.
-Use :func:`torch.Tensor.detach` instead. (Work is ongoing to
-`remove Tensor.data entirely <https://github.com/pytorch/pytorch/issues/30987>`_).
-
-Avoid in-place operations when using tensor.shape in tracing mode
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
-
-In tracing mode, shapes obtained from ``tensor.shape`` are traced as tensors,
-and share the same memory. This might cause a mismatch the final output values.
-As a workaround, avoid the use of inplace operations in these scenarios.
-For example, in the model::
-
-    class Model(torch.nn.Module):
-      def forward(self, states):
-          batch_size, seq_length = states.shape[:2]
-          real_seq_length = seq_length
-          real_seq_length += 2
-          return real_seq_length + seq_length
-
-``real_seq_length`` and ``seq_length`` share the same memory in tracing mode.
-This could be avoided by rewriting the inplace operation::
-
-    real_seq_length = real_seq_length + 2
-
-Limitations
------------
-
-Types
-^^^^^
-
-* Only :class:`torch.Tensors`, numeric types that can be trivially converted to torch.Tensors (e.g. float, int),
-  and tuples and lists of those types are supported as model inputs or outputs. Dict and str inputs and
-  outputs are accepted in :ref:`tracing<tracing-vs-scripting>` mode, but:
-
-  * Any computation that depends on the value of a dict or a str input **will be replaced with the
-    constant value** seen during the one traced execution.
-  * Any output that is a dict will be silently replaced with a **flattened sequence of its values
-    (keys will be removed)**. E.g. ``{"foo": 1, "bar": 2}`` becomes ``(1, 2)``.
-  * Any output that is a str will be silently removed.
-
-* Certain operations involving tuples and lists are not supported in
-  :ref:`scripting<tracing-vs-scripting>` mode due to limited support in ONNX for nested sequences.
-  In particular appending a tuple to a list is not supported. In tracing mode, the nested sequences
-  will be flattened automatically during the tracing.
-
-Differences in Operator Implementations
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
-
-Due to differences in implementations of operators, running the exported model on different runtimes
-may produce different results from each other or from PyTorch. Normally these differences are
-numerically small, so this should only be a concern if your application is sensitive to these
-small differences.
-
-.. _tensor-indexing:
-
-Unsupported Tensor Indexing Patterns
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
-
-Tensor indexing patterns that cannot be exported are listed below.
-If you are experiencing issues exporting a model that does not include any of
-the unsupported patterns below, please double check that you are exporting with
-the latest ``opset_version``.
-
-Reads / Gets
-~~~~~~~~~~~~
-
-When indexing into a tensor for reading, the following patterns are not supported: ::
-
-  # Tensor indices that includes negative values.
-  data[torch.tensor([[1, 2], [2, -3]]), torch.tensor([-2, 3])]
-  # Workarounds: use positive index values.
-
-Writes / Sets
-~~~~~~~~~~~~~
-
-When indexing into a Tensor for writing, the following patterns are not supported: ::
-
-  # Multiple tensor indices if any has rank >= 2
-  data[torch.tensor([[1, 2], [2, 3]]), torch.tensor([2, 3])] = new_data
-  # Workarounds: use single tensor index with rank >= 2,
-  #              or multiple consecutive tensor indices with rank == 1.
-
-  # Multiple tensor indices that are not consecutive
-  data[torch.tensor([2, 3]), :, torch.tensor([1, 2])] = new_data
-  # Workarounds: transpose `data` such that tensor indices are consecutive.
-
-  # Tensor indices that includes negative values.
-  data[torch.tensor([1, -2]), torch.tensor([-2, 3])] = new_data
-  # Workarounds: use positive index values.
-
-  # Implicit broadcasting required for new_data.
-  data[torch.tensor([[0, 2], [1, 1]]), 1:3] = new_data
-  # Workarounds: expand new_data explicitly.
-  # Example:
-  #   data shape: [3, 4, 5]
-  #   new_data shape: [5]
-  #   expected new_data shape after broadcasting: [2, 2, 2, 5]
-
-Adding support for operators
-----------------------------
-
-When exporting a model that includes unsupported operators, you'll see an error message like:
-
-.. code-block:: text
-
-    RuntimeError: ONNX export failed: Couldn't export operator foo
-
-When that happens, there are a few things you can do:
-
-#. Change the model to not use that operator.
-#. Create a symbolic function to convert the operator and register it as a custom symbolic function.
-#. Contribute to PyTorch to add the same symbolic function to :mod:`torch.onnx` itself.
-
-If you decided to implement a symbolic function (we hope you will contribute it back to PyTorch!), here is how you can get started:
-
-ONNX exporter internals
-^^^^^^^^^^^^^^^^^^^^^^^
-
-A "symbolic function" is a function that decomposes a PyTorch operator into a
-composition of a series of ONNX operators.
-
-During export, each node (which contains a PyTorch operator) in the TorchScript
-graph is visited by the exporter in topological order.
-Upon visiting a node, the exporter looks for a registered symbolic functions for
-that operator. Symbolic functions are implemented in Python. A symbolic function for
-an op named ``foo`` would look something like::
-
-
-    def foo(
-      g,
-      input_0: torch._C.Value,
-      input_1: torch._C.Value) -> Union[None, torch._C.Value, List[torch._C.Value]]:
-      """
-      Adds the ONNX operations representing this PyTorch function by updating the
-      graph g with `g.op()` calls.
-
-      Args:
-        g (Graph): graph to write the ONNX representation into.
-        input_0 (Value): value representing the variables which contain
-            the first input for this operator.
-        input_1 (Value): value representing the variables which contain
-            the second input for this operator.
-
-      Returns:
-        A Value or List of Values specifying the ONNX nodes that compute something
-        equivalent to the original PyTorch operator with the given inputs.
-
-        None if it cannot be converted to ONNX.
-      """
-      ...
-
-The ``torch._C`` types are Python wrappers around the types defined in C++ in
-`ir.h <https://github.com/pytorch/pytorch/blob/main/torch/csrc/jit/ir/ir.h>`_.
-
-The process for adding a symbolic function depends on the type of operator.
-
-.. _adding-support-aten:
-
-ATen operators
-^^^^^^^^^^^^^^
-
-`ATen <https://pytorch.org/cppdocs/#aten>`_ is PyTorch's built-in tensor library.
-If the operator is an ATen operator (shows up in the TorchScript graph with the prefix
-``aten::``), make sure it is not supported already.
-
-List of supported operators
-~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-Visit the auto generated :doc:`list of supported TorchScript operators <../onnx_supported_aten_ops>`
-for details on which operator are supported in each ``opset_version``.
-
-Adding support for an aten or quantized operator
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-If the operator is not in the list above:
-
-* Define the symbolic function in ``torch/onnx/symbolic_opset<version>.py``, for example
-  `torch/onnx/symbolic_opset9.py <https://github.com/pytorch/pytorch/blob/main/torch/onnx/symbolic_opset9.py>`_.
-  Make sure the function has the same name as the ATen function, which may be declared in
-  ``torch/_C/_VariableFunctions.pyi`` or ``torch/nn/functional.pyi`` (these files are generated at
-  build time, so will not appear in your checkout until you build PyTorch).
-* By default, the first arg is the ONNX graph.
-  Other arg names must EXACTLY match the names in the ``.pyi`` file,
-  because dispatch is done with keyword arguments.
-* In the symbolic function, if the operator is in the
-  `ONNX standard operator set <https://github.com/onnx/onnx/blob/master/docs/Operators.md>`_,
-  we only need to create a node to represent the ONNX operator in the graph.
-  If not, we can compose several standard operators that have the
-  equivalent semantics to the ATen operator.
-
-Here is an example of handling missing symbolic function for the ``ELU`` operator.
-
-If we run the following code::
-
-    print(
-        torch.jit.trace(
-            torch.nn.ELU(), # module
-            torch.ones(1)   # example input
-        ).graph
-    )
-
-We see something like::
-
-  graph(%self : __torch__.torch.nn.modules.activation.___torch_mangle_0.ELU,
-        %input : Float(1, strides=[1], requires_grad=0, device=cpu)):
-    %4 : float = prim::Constant[value=1.]()
-    %5 : int = prim::Constant[value=1]()
-    %6 : int = prim::Constant[value=1]()
-    %7 : Float(1, strides=[1], requires_grad=0, device=cpu) = aten::elu(%input, %4, %5, %6)
-    return (%7)
-
-Since we see ``aten::elu`` in the graph, we know this is an ATen operator.
-
-We check the `ONNX operator list <https://github.com/onnx/onnx/blob/master/docs/Operators.md>`_,
-and confirm that ``Elu`` is standardized in ONNX.
-
-We find a signature for ``elu`` in ``torch/nn/functional.pyi``::
-
-    def elu(input: Tensor, alpha: float = ..., inplace: bool = ...) -> Tensor: ...
-
-We add the following lines to ``symbolic_opset9.py``::
-
-    def elu(g, input: torch.Value, alpha: torch.Value, inplace: bool = False):
-        return g.op("Elu", input, alpha_f=alpha)
-
-Now PyTorch is able to export models containing the ``aten::elu`` operator!
-
-See the ``torch/onnx/symbolic_opset*.py`` files for more examples.
-
-
-torch.autograd.Functions
-^^^^^^^^^^^^^^^^^^^^^^^^
-
-If the operator is a sub-class of :class:`torch.autograd.Function`, there are three ways
-to export it.
-
-Static Symbolic Method
-~~~~~~~~~~~~~~~~~~~~~~
-
-You can add a static method named ``symbolic`` to your function class. It should return
-ONNX operators that represent the function's behavior in ONNX. For example::
-
-    class MyRelu(torch.autograd.Function):
-        @staticmethod
-        def forward(ctx, input: torch.Tensor) -> torch.Tensor:
-            ctx.save_for_backward(input)
-            return input.clamp(min=0)
-
-        @staticmethod
-        def symbolic(g: torch.Graph, input: torch.Value) -> torch.Value:
-            return g.op("Clip", input, g.op("Constant", value_t=torch.tensor(0, dtype=torch.float)))
-
-.. FIXME(justinchuby): PythonOps are too complicated and the example below
-..  uses private methods we do not expose. We are looking to
-..  improve the experience. Since SymbolicContext is deprecated, we think
-..  defining a symbolic staticmethod is a better way to go for now.
-
-.. PythonOp Symbolic
-.. ~~~~~~~~~~~~~~~~~
-
-.. Alternatively, you can register a custom symbolic function.
-.. This gives the symbolic function access to more info through the
-.. ``torch.onnx.SymbolicContext`` object, which gets passed in as the first
-.. argument (before the ``Graph`` object).
-
-.. All autograd ``Function``\ s appear in the TorchScript graph as ``prim::PythonOp`` nodes.
-.. In order to differentiate between different ``Function`` subclasses, the
-.. symbolic function should use the ``name`` kwarg which gets set to the name of the class.
-
-.. Custom symbolic functions should add type and shape information by calling ``setType(...)``
-.. on Value objects before returning them (implemented in C++ by
-.. . ``torch::jit::Value::setType``). This is not required, but it can help the exporter's
-.. shape and type inference for down-stream nodes. For a non-trivial example of ``setType``, see
-.. ``test_aten_embedding_2`` in
-.. `test_operators.py <https://github.com/pytorch/pytorch/blob/main/test/onnx/test_operators.py>`_.
-
-.. The example below shows how you can access ``requires_grad`` via the ``Node`` object:
-
-..     class MyClip(torch.autograd.Function):
-..         @staticmethod
-..         def forward(ctx, input, min):
-..             ctx.save_for_backward(input)
-..             return input.clamp(min=min)
-
-..     class MyRelu(torch.autograd.Function):
-..         @staticmethod
-..         def forward(ctx, input):
-..             ctx.save_for_backward(input)
-..             return input.clamp(min=0)
-
-..     def symbolic_python_op(g: "GraphContext", *args, **kwargs):
-..         n = ctx.cur_node
-..         print("original node: ", n)
-..         for i, out in enumerate(n.outputs()):
-..             print("original output {}: {}, requires grad: {}".format(i, out, out.requiresGrad()))
-..         import torch.onnx.symbolic_helper as sym_helper
-..         for i, arg in enumerate(args):
-..             requires_grad = arg.requiresGrad() if sym_helper._is_value(arg) else False
-..             print("arg {}: {}, requires grad: {}".format(i, arg, requires_grad))
-
-..         name = kwargs["name"]
-..         ret = None
-..         if name == "MyClip":
-..             ret = g.op("Clip", args[0], args[1])
-..         elif name == "MyRelu":
-..             ret = g.op("Relu", args[0])
-..         else:
-..             # Logs a warning and returns None
-..             return _unimplemented("prim::PythonOp", "unknown node kind: " + name)
-..         # Copy type and shape from original node.
-..         ret.setType(n.type())
-..         return ret
-
-..     from torch.onnx import register_custom_op_symbolic
-.. .     register_custom_op_symbolic("prim::PythonOp", symbolic_python_op, 1)
-
-Inline Autograd Function
-~~~~~~~~~~~~~~~~~~~~~~~~
-
-In cases where a static symbolic method is not provided for its subsequent :class:`torch.autograd.Function` or
-where a function to register ``prim::PythonOp`` as custom symbolic functions is not provided,
-:func:`torch.onnx.export` tries to inline the graph that corresponds to that :class:`torch.autograd.Function` such that
-this function is broken down into individual operators that were used within the function.
-The export should be successful as long as these individual operators are supported. For example::
-
-    class MyLogExp(torch.autograd.Function):
-        @staticmethod
-        def forward(ctx, input: torch.Tensor) -> torch.Tensor:
-            ctx.save_for_backward(input)
-            h = input.exp()
-            return h.log().log()
-
-There is no static symbolic method present for this model, yet it is exported as follows::
-
-    graph(%input : Float(1, strides=[1], requires_grad=0, device=cpu)):
-        %1 : float = onnx::Exp[](%input)
-        %2 : float = onnx::Log[](%1)
-        %3 : float = onnx::Log[](%2)
-        return (%3)
-
-If you need to avoid inlining of :class:`torch.autograd.Function`, you should export models with
-``operator_export_type`` set to ``ONNX_FALLTHROUGH`` or ``ONNX_ATEN_FALLBACK``.
-
-Custom operators
-^^^^^^^^^^^^^^^^
-
-You can export your model with custom operators that includes a combination of many standard ONNX ops,
-or are driven by self-defined C++ backend.
-
-ONNX-script functions
-~~~~~~~~~~~~~~~~~~~~~
-
-If an operator is not a standard ONNX op, but can be composed of multiple existing ONNX ops, you can utilize
-`ONNX-script <https://github.com/microsoft/onnx-script>`_ to create an external ONNX function to support the operator.
-You can export it by following this example::
-
-    import onnxscript
-    # There are three opset version needed to be aligned
-    # This is (1) the opset version in ONNX function
-    from onnxscript.onnx_opset import opset15 as op
-    opset_version = 15
-
-    x = torch.randn(1, 2, 3, 4, requires_grad=True)
-    model = torch.nn.SELU()
-
-    custom_opset = onnxscript.values.Opset(domain="onnx-script", version=1)
-
-    @onnxscript.script(custom_opset)
-    def Selu(X):
-        alpha = 1.67326  # auto wrapped as Constants
-        gamma = 1.0507
-        alphaX = op.CastLike(alpha, X)
-        gammaX = op.CastLike(gamma, X)
-        neg = gammaX * (alphaX * op.Exp(X) - alphaX)
-        pos = gammaX * X
-        zero = op.CastLike(0, X)
-        return op.Where(X <= zero, neg, pos)
-
-    # setType API provides shape/type to ONNX shape/type inference
-    def custom_selu(g: jit_utils.GraphContext, X):
-        return g.onnxscript_op(Selu, X).setType(X.type())
-
-    # Register custom symbolic function
-    # There are three opset version needed to be aligned
-    # This is (2) the opset version in registry
-    torch.onnx.register_custom_op_symbolic(
-        symbolic_name="aten::selu",
-        symbolic_fn=custom_selu,
-        opset_version=opset_version,
-    )
-
-    # There are three opset version needed to be aligned
-    # This is (2) the opset version in exporter
-    torch.onnx.export(
-        model,
-        x,
-        "model.onnx",
-        opset_version=opset_version,
-        # only needed if you want to specify an opset version > 1.
-        custom_opsets={"onnx-script": 2}
-    )
-
-The example above exports it as a custom operator in the "onnx-script" opset.
-When exporting a custom operator, you can specify the custom domain version using the
-``custom_opsets`` dictionary at export. If not specified, the custom opset version defaults to 1.
-
-NOTE: Be careful to align the opset version mentioned in the above example, and make sure they are consumed in exporter step.
-The example usage of how to write a onnx-script function is a beta version in terms of the active development on onnx-script.
-Please follow the latest `ONNX-script <https://github.com/microsoft/onnx-script>`_
-
-C++ Operators
-~~~~~~~~~~~~~
-
-If a model uses a custom operator implemented in C++ as described in
-`Extending TorchScript with Custom C++ Operators <https://pytorch.org/tutorials/advanced/torch_script_custom_ops.html>`_,
-you can export it by following this example::
-
-    from torch.onnx import symbolic_helper
-
-
-    # Define custom symbolic function
-    @symbolic_helper.parse_args("v", "v", "f", "i")
-    def symbolic_foo_forward(g, input1, input2, attr1, attr2):
-        return g.op("custom_domain::Foo", input1, input2, attr1_f=attr1, attr2_i=attr2)
-
-
-    # Register custom symbolic function
-    torch.onnx.register_custom_op_symbolic("custom_ops::foo_forward", symbolic_foo_forward, 9)
-
-
-    class FooModel(torch.nn.Module):
-        def __init__(self, attr1, attr2):
-            super().__init__()
-            self.attr1 = attr1
-            self.attr2 = attr2
-
-        def forward(self, input1, input2):
-            # Calling custom op
-            return torch.ops.custom_ops.foo_forward(input1, input2, self.attr1, self.attr2)
-
-
-    model = FooModel(attr1, attr2)
-    torch.onnx.export(
-        model,
-        (example_input1, example_input1),
-        "model.onnx",
-        # only needed if you want to specify an opset version > 1.
-        custom_opsets={"custom_domain": 2}
-    )
-
-The example above exports it as a custom operator in the "custom_domain" opset.
-When exporting a custom operator, you can specify the custom domain version using the
-``custom_opsets`` dictionary at export. If not specified, the custom opset version defaults to 1.
-
-The runtime that consumes the model needs to support the custom op. See
-`Caffe2 custom ops <https://caffe2.ai/docs/custom-operators.html>`_,
-`ONNX Runtime custom ops <https://onnxruntime.ai/docs/reference/operators/add-custom-op.html>`_,
-or your runtime of choice's documentation.
-
-
-Discovering all unconvertible ATen ops at once
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
-
-When export fails due to an unconvertible ATen op, there may in fact be more
-than one such op but the error message only mentions the first. To discover
-all of the unconvertible ops in one go you can::
-
-    # prepare model, args, opset_version
-    ...
-
-    torch_script_graph, unconvertible_ops = torch.onnx.utils.unconvertible_ops(
-        model, args, opset_version=opset_version
-    )
-
-    print(set(unconvertible_ops))
-
-The set is approximated because some ops may be removed during the conversion
-process and don't need to be converted. Some other ops may have partial support
-that will fail conversion with particular inputs, but this should give you a
-general idea of what ops are not supported. Please feel free to open GitHub Issues
-for op support requests.
-
-Frequently Asked Questions
---------------------------
-Q: I have exported my LSTM model, but its input size seems to be fixed?
-
-  The tracer records the shapes of the example inputs. If the model should accept
-  inputs of dynamic shapes, set ``dynamic_axes`` when calling :func:`torch.onnx.export`.
-
-Q: How to export models containing loops?
-
-  See `Tracing vs Scripting`_.
-
-Q: How to export models with primitive type inputs (e.g. int, float)?
-
-  Support for primitive numeric type inputs was added in PyTorch 1.9.
-  However, the exporter does not support models with str inputs.
-
-Q: Does ONNX support implicit scalar datatype casting?
-
-  The ONNX standard does not, but the exporter will try to handle that part.
-  Scalars are exported as constant tensors.
-  The exporter will figure out the right data type for scalars. In rare cases when it is unable
-  to do so, you will need to manually specify the datatype with e.g. `dtype=torch.float32`.
-  If you see any errors, please [create a GitHub issue](https://github.com/pytorch/pytorch/issues).
-
-Q: Are lists of Tensors exportable to ONNX?
-
-  Yes, for ``opset_version`` >= 11, since ONNX introduced the Sequence type in opset 11.
-
-
-Contributing / developing
+Contributing / Developing
 -------------------------
-`Developer docs <https://github.com/pytorch/pytorch/wiki/PyTorch-ONNX-exporter>`_.
 
-Functions
----------
-.. autofunction:: export
-.. autofunction:: export_to_pretty_string
-.. autofunction:: register_custom_op_symbolic
-.. autofunction:: unregister_custom_op_symbolic
-.. autofunction:: select_model_mode_for_export
-.. autofunction:: is_in_onnx_export
-.. autofunction:: enable_log
-.. autofunction:: disable_log
-.. autofunction:: torch.onnx.verification.find_mismatch
+The ONNX exporter is a community project and we welcome contributions. We follow the
+`PyTorch guidelines for contributions <https://github.com/pytorch/pytorch/blob/main/CONTRIBUTING.md>`_, but you might
+also be interested in reading our `development wiki <https://github.com/pytorch/pytorch/wiki/PyTorch-ONNX-exporter>`_.
 
-Classes
--------
+.. toctree::
+    :hidden:
 
-.. autosummary::
-    :toctree: generated
-    :nosignatures:
-    :template: classtemplate.rst
-
-    JitScalarType
-    torch.onnx.verification.GraphInfo
-    torch.onnx.verification.VerificationOptions
-
-Preview: torch.onnx TorchDynamo Exporter
-----------------------------------------
-
-.. warning::
-  The ONNX exporter for TorchDynamo is under active development and is
-  subject to rapid change.
-
-.. autofunction:: torch.onnx.dynamo_export
-.. autofunction:: torch.onnx.enable_fake_mode
-.. autofunction:: torch.onnx.is_onnxrt_backend_supported
-
-.. autosummary::
-    :toctree: generated
-    :nosignatures:
-    :template: classtemplate.rst
-
-    torch.onnx.DiagnosticOptions
-    torch.onnx.ExportOptions
-    torch.onnx.ExportOutput
-    torch.onnx.ExportOutputSerializer
-    torch.onnx.OnnxExporterError
-    torch.onnx.OnnxRegistry
+    onnx_dynamo
+    onnx_dynamo_onnxruntime_backend
+    onnx_torchscript

docs/source/onnx_diagnostics.rst

@@ -1,26 +0,0 @@
-torch.onnx diagnostics
-======================
-
-.. contents:: :local:
-.. automodule:: torch.onnx._internal.diagnostics
-.. currentmodule:: torch.onnx._internal.diagnostics
-
-Overview
---------
-
-NOTE: This feature is underdevelopment and is subject to change.
-
-The goal is to improve the diagnostics to help users debug and improve their model export to ONNX.
-
-- The diagnostics are emitted in machine parsable `Static Analysis Results Interchange Format (SARIF) <https://docs.oasis-open.org/sarif/sarif/v2.1.0/sarif-v2.1.0.html>`__.
-- A new clearer, structured way to add new and keep track of diagnostic rules.
-- Serve as foundation for more future improvements consuming the diagnostics.
-
-
-Diagnostic Rules
-----------------
-
-.. toctree::
-    :glob:
-
-    generated/onnx_diagnostics_rules/*

docs/source/onnx_dynamo.rst

@@ -0,0 +1,156 @@
+TorchDynamo-based ONNX Exporter
+===============================
+
+.. automodule:: torch.onnx
+  :noindex:
+
+.. contents:: :local:
+    :depth: 3
+
+.. warning::
+  The ONNX exporter for TorchDynamo is a rapidly evolving beta technology.
+
+Overview
+--------
+
+The ONNX exporter leverages TorchDynamo engine to hook into Python's frame evaluation API
+and dynamically rewrite its bytecode into an FX Graph.
+The resulting FX Graph is then polished before it is finally translated into an ONNX graph.
+
+The main advantage of this approach is that the `FX graph <https://pytorch.org/docs/stable/fx.html>`_ is captured using
+bytecode analysis that preserves the dynamic nature of the model instead of using traditional static tracing techniques.
+
+The exporter is designed to be modular and extensible. It is composed of the following components:
+
+  - **ONNX Exporter**: :class:`Exporter` main class that orchestrates the export process.
+  - **ONNX Export Options**: :class:`ExportOptions` has a set of options that control the export process.
+  - **ONNX Registry**: :class:`OnnxRegistry` is the registry of ONNX operators and functions.
+  - **FX Graph Extractor**: :class:`FXGraphExtractor` extracts the FX graph from the PyTorch model.
+  - **Fake Mode**: :class:`ONNXFakeContext` is a context manager that enables fake mode for large scale models.
+  - **ONNX Export Output**: :class:`ExportOutput` is the output of the exporter that contains the exported ONNX graph and diagnostics.
+  - **ONNX Export Output Serializer**: :class:`ExportOutputSerializer` serializes the exported model to a file.
+  - **ONNX Diagnostic Options**: :class:`DiagnosticOptions` has a set of options that control the diagnostics emitted by the exporter.
+
+Dependencies
+------------
+
+The ONNX exporter depends on extra Python packages:
+
+  - `ONNX <https://onnx.ai>`_
+  - `ONNX Script <https://onnxscript.ai>`_
+
+They can be installed through `pip <https://pypi.org/project/pip/>`_:
+
+.. code-block:: bash
+
+  pip install --upgrade onnx onnxscript
+
+A simple example
+----------------
+
+See below a demonstration of exporter API in action with a simple Multilayer Perceptron (MLP) as example:
+
+.. code-block:: python
+
+  import torch
+
+  class MLPModel(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.fc0 = nn.Linear(8, 8, bias=True)
+        self.fc1 = nn.Linear(8, 4, bias=True)
+        self.fc2 = nn.Linear(4, 2, bias=True)
+        self.fc3 = nn.Linear(2, 2, bias=True)
+
+    def forward(self, tensor_x: torch.Tensor):
+        tensor_x = self.fc0(tensor_x)
+        tensor_x = torch.sigmoid(tensor_x)
+        tensor_x = self.fc1(tensor_x)
+        tensor_x = torch.sigmoid(tensor_x)
+        tensor_x = self.fc2(tensor_x)
+        tensor_x = torch.sigmoid(tensor_x)
+        output = self.fc3(tensor_x)
+        return output
+
+  model = MLPModel()
+  tensor_x = torch.rand((97, 8), dtype=torch.float32)
+  export_output = torch.onnx.dynamo_export(model, tensor_x)
+
+As the code above shows, all you need is to provide :func:`torch.onnx.dynamo_export` with an instance of the model and its input.
+The exporter will then return an instance of :class:`torch.onnx.ExportOutput` that contains the exported ONNX graph along with extra information.
+
+The in-memory model available through ``export_output.model_proto`` is an ``onnx.ModelProto`` object in compliance with the `ONNX IR spec <https://github.com/onnx/onnx/blob/main/docs/IR.md>`_.
+The ONNX model may then be serialized into a `Protobuf file <https://protobuf.dev/>`_ using the :meth:`torch.onnx.ExportOutput.save` API.
+
+.. code-block:: python
+
+  export_output.save("mlp.onnx")
+
+Inspecting the ONNX model using GUI
+-----------------------------------
+
+You can view the exported model using `Netron <https://netron.app/>`__.
+
+.. image:: _static/img/onnx/onnx_dynamo_mlp_model.png
+    :width: 40%
+    :alt: MLP model as viewed using Netron
+
+Note that each layer is represented in a rectangular box with a *f* icon in the top right corner.
+
+.. image:: _static/img/onnx/onnx_dynamo_mlp_model_function_highlight.png
+    :width: 40%
+    :alt: ONNX function highlighted on MLP model
+
+By expanding it, the function body is shown.
+
+.. image:: _static/img/onnx/onnx_dynamo_mlp_model_function_body.png
+    :width: 50%
+    :alt: ONNX function body
+
+The function body is a sequence of ONNX operators or other functions.
+
+Diagnosing issues with SARIF
+----------------------------
+
+ONNX diagnostics goes beyond regular logs through the adoption of
+`Static Analysis Results Interchange Format (aka SARIF) <https://docs.oasis-open.org/sarif/sarif/v2.1.0/sarif-v2.1.0.html>`__
+to help users debug and improve their model using a GUI, such as
+Visual Studio Code's `SARIF Viewer <https://marketplace.visualstudio.com/items?itemName=MS-SarifVSCode.sarif-viewer>`_.
+
+The main advantages are:
+
+  - The diagnostics are emitted in machine parseable `Static Analysis Results Interchange Format (SARIF) <https://docs.oasis-open.org/sarif/sarif/v2.1.0/sarif-v2.1.0.html>`__.
+  - A new clearer, structured way to add new and keep track of diagnostic rules.
+  - Serve as foundation for more future improvements consuming the diagnostics.
+
+.. toctree::
+   :maxdepth: 1
+   :caption: ONNX Diagnostic SARIF Rules
+   :glob:
+
+   generated/onnx_dynamo_diagnostics_rules/*
+
+API Reference
+-------------
+
+.. autofunction:: torch.onnx.dynamo_export
+
+.. autoclass:: torch.onnx.ExportOptions
+    :members:
+
+.. autofunction:: torch.onnx.enable_fake_mode
+
+.. autoclass:: torch.onnx.ExportOutput
+    :members:
+
+.. autoclass:: torch.onnx.ExportOutputSerializer
+    :members:
+
+.. autoclass:: torch.onnx.OnnxExporterError
+    :members:
+
+.. autoclass:: torch.onnx.OnnxRegistry
+    :members:
+
+.. autoclass:: torch.onnx.DiagnosticOptions
+    :members:

docs/source/onnx_dynamo_onnxruntime_backend.rst

@@ -0,0 +1,9 @@
+ONNX Backend for TorchDynamo
+============================
+
+For a quick overview of ``torch.compiler``, see :ref:`torch.compiler_overview`.
+
+.. warning::
+  The ONNX backend for torch.compile is a rapidly evolving beta technology.
+
+.. autofunction:: torch.onnx.is_onnxrt_backend_supported

docs/source/onnx_torchscript.rst

@@ -0,0 +1,719 @@
+TorchScript-based ONNX Exporter
+===============================
+
+.. note::
+    To export an ONNX model using TorchDynamo instead of TorchScript, see :func:`torch.onnx.dynamo_export`.
+
+.. contents:: :local:
+
+Example: AlexNet from PyTorch to ONNX
+-------------------------------------
+
+Here is a simple script which exports a pretrained AlexNet to an ONNX file named ``alexnet.onnx``.
+The call to ``torch.onnx.export`` runs the model once to trace its execution and then exports the
+traced model to the specified file::
+
+    import torch
+    import torchvision
+
+    dummy_input = torch.randn(10, 3, 224, 224, device="cuda")
+    model = torchvision.models.alexnet(pretrained=True).cuda()
+
+    # Providing input and output names sets the display names for values
+    # within the model's graph. Setting these does not change the semantics
+    # of the graph; it is only for readability.
+    #
+    # The inputs to the network consist of the flat list of inputs (i.e.
+    # the values you would pass to the forward() method) followed by the
+    # flat list of parameters. You can partially specify names, i.e. provide
+    # a list here shorter than the number of inputs to the model, and we will
+    # only set that subset of names, starting from the beginning.
+    input_names = [ "actual_input_1" ] + [ "learned_%d" % i for i in range(16) ]
+    output_names = [ "output1" ]
+
+    torch.onnx.export(model, dummy_input, "alexnet.onnx", verbose=True, input_names=input_names, output_names=output_names)
+
+The resulting ``alexnet.onnx`` file contains a binary `protocol buffer <https://developers.google.com/protocol-buffers/>`_
+which contains both the network structure and parameters of the model you exported
+(in this case, AlexNet).  The argument ``verbose=True`` causes the
+exporter to print out a human-readable representation of the model::
+
+    # These are the inputs and parameters to the network, which have taken on
+    # the names we specified earlier.
+    graph(%actual_input_1 : Float(10, 3, 224, 224)
+          %learned_0 : Float(64, 3, 11, 11)
+          %learned_1 : Float(64)
+          %learned_2 : Float(192, 64, 5, 5)
+          %learned_3 : Float(192)
+          # ---- omitted for brevity ----
+          %learned_14 : Float(1000, 4096)
+          %learned_15 : Float(1000)) {
+      # Every statement consists of some output tensors (and their types),
+      # the operator to be run (with its attributes, e.g., kernels, strides,
+      # etc.), its input tensors (%actual_input_1, %learned_0, %learned_1)
+      %17 : Float(10, 64, 55, 55) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[11, 11], pads=[2, 2, 2, 2], strides=[4, 4]](%actual_input_1, %learned_0, %learned_1), scope: AlexNet/Sequential[features]/Conv2d[0]
+      %18 : Float(10, 64, 55, 55) = onnx::Relu(%17), scope: AlexNet/Sequential[features]/ReLU[1]
+      %19 : Float(10, 64, 27, 27) = onnx::MaxPool[kernel_shape=[3, 3], pads=[0, 0, 0, 0], strides=[2, 2]](%18), scope: AlexNet/Sequential[features]/MaxPool2d[2]
+      # ---- omitted for brevity ----
+      %29 : Float(10, 256, 6, 6) = onnx::MaxPool[kernel_shape=[3, 3], pads=[0, 0, 0, 0], strides=[2, 2]](%28), scope: AlexNet/Sequential[features]/MaxPool2d[12]
+      # Dynamic means that the shape is not known. This may be because of a
+      # limitation of our implementation (which we would like to fix in a
+      # future release) or shapes which are truly dynamic.
+      %30 : Dynamic = onnx::Shape(%29), scope: AlexNet
+      %31 : Dynamic = onnx::Slice[axes=[0], ends=[1], starts=[0]](%30), scope: AlexNet
+      %32 : Long() = onnx::Squeeze[axes=[0]](%31), scope: AlexNet
+      %33 : Long() = onnx::Constant[value={9216}](), scope: AlexNet
+      # ---- omitted for brevity ----
+      %output1 : Float(10, 1000) = onnx::Gemm[alpha=1, beta=1, broadcast=1, transB=1](%45, %learned_14, %learned_15), scope: AlexNet/Sequential[classifier]/Linear[6]
+      return (%output1);
+    }
+
+You can also verify the output using the `ONNX <https://github.com/onnx/onnx/>`_ library,
+which you can install using ``pip``::
+
+    pip install onnx
+
+Then, you can run::
+
+    import onnx
+
+    # Load the ONNX model
+    model = onnx.load("alexnet.onnx")
+
+    # Check that the model is well formed
+    onnx.checker.check_model(model)
+
+    # Print a human readable representation of the graph
+    print(onnx.helper.printable_graph(model.graph))
+
+You can also run the exported model with one of the many
+`runtimes that support ONNX <https://onnx.ai/supported-tools.html#deployModel>`_.
+For example after installing `ONNX Runtime <https://www.onnxruntime.ai>`_, you can
+load and run the model::
+
+    import onnxruntime as ort
+    import numpy as np
+
+    ort_session = ort.InferenceSession("alexnet.onnx")
+
+    outputs = ort_session.run(
+        None,
+        {"actual_input_1": np.random.randn(10, 3, 224, 224).astype(np.float32)},
+    )
+    print(outputs[0])
+
+Here is a more involved `tutorial on exporting a model and running it with ONNX Runtime <https://pytorch.org/tutorials/advanced/super_resolution_with_onnxruntime.html>`_.
+
+.. _tracing-vs-scripting:
+
+Tracing vs Scripting
+--------------------
+
+Internally, :func:`torch.onnx.export()` requires a :class:`torch.jit.ScriptModule` rather than
+a :class:`torch.nn.Module`. If the passed-in model is not already a ``ScriptModule``,
+``export()`` will use *tracing* to convert it to one:
+
+.. TODO(justinchuby): Add a word on recommending tracing over scripting for most use cases.
+
+* **Tracing**: If ``torch.onnx.export()`` is called with a Module that is not already a
+  ``ScriptModule``, it first does the equivalent of :func:`torch.jit.trace`, which executes the model
+  once with the given ``args`` and records all operations that happen during that execution. This
+  means that if your model is dynamic, e.g., changes behavior depending on input data, the exported
+  model will *not* capture this dynamic behavior.
+  We recommend examining the exported model and making sure the operators look
+  reasonable. Tracing will unroll loops and if statements, exporting a static graph that is exactly
+  the same as the traced run. If you want to export your model with dynamic control flow, you will
+  need to use *scripting*.
+
+* **Scripting**: Compiling a model via scripting preserves dynamic control flow and is valid for inputs
+  of different sizes. To use scripting:
+
+  * Use :func:`torch.jit.script` to produce a ``ScriptModule``.
+  * Call ``torch.onnx.export()`` with the ``ScriptModule`` as the model. The ``args`` are still required,
+    but they will be used internally only to produce example outputs, so that the types and shapes of the
+    outputs can be captured. No tracing will be performed.
+
+See `Introduction to TorchScript <https://pytorch.org/tutorials/beginner/Intro_to_TorchScript_tutorial.html>`_
+and `TorchScript <jit.html>`_ for more details, including how to compose tracing and scripting to suit the
+particular requirements of different models.
+
+
+Avoiding Pitfalls
+-----------------
+
+Avoid NumPy and built-in Python types
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+PyTorch models can be written using NumPy or Python types and functions, but
+during :ref:`tracing<tracing-vs-scripting>`, any variables of NumPy or Python
+types (rather than torch.Tensor) are converted to constants, which will produce
+the wrong result if those values should change depending on the inputs.
+
+For example, rather than using numpy functions on numpy.ndarrays: ::
+
+    # Bad! Will be replaced with constants during tracing.
+    x, y = np.random.rand(1, 2), np.random.rand(1, 2)
+    np.concatenate((x, y), axis=1)
+
+Use torch operators on torch.Tensors: ::
+
+    # Good! Tensor operations will be captured during tracing.
+    x, y = torch.randn(1, 2), torch.randn(1, 2)
+    torch.cat((x, y), dim=1)
+
+
+And rather than use :func:`torch.Tensor.item` (which converts a Tensor to a Python
+built-in number): ::
+
+    # Bad! y.item() will be replaced with a constant during tracing.
+    def forward(self, x, y):
+        return x.reshape(y.item(), -1)
+
+Use torch's support for implicit casting of single-element tensors: ::
+
+    # Good! y will be preserved as a variable during tracing.
+    def forward(self, x, y):
+        return x.reshape(y, -1)
+
+Avoid Tensor.data
+^^^^^^^^^^^^^^^^^
+
+Using the Tensor.data field can produce an incorrect trace and therefore an incorrect ONNX graph.
+Use :func:`torch.Tensor.detach` instead. (Work is ongoing to
+`remove Tensor.data entirely <https://github.com/pytorch/pytorch/issues/30987>`_).
+
+Avoid in-place operations when using tensor.shape in tracing mode
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+In tracing mode, shapes obtained from ``tensor.shape`` are traced as tensors,
+and share the same memory. This might cause a mismatch the final output values.
+As a workaround, avoid the use of inplace operations in these scenarios.
+For example, in the model::
+
+    class Model(torch.nn.Module):
+      def forward(self, states):
+          batch_size, seq_length = states.shape[:2]
+          real_seq_length = seq_length
+          real_seq_length += 2
+          return real_seq_length + seq_length
+
+``real_seq_length`` and ``seq_length`` share the same memory in tracing mode.
+This could be avoided by rewriting the inplace operation::
+
+    real_seq_length = real_seq_length + 2
+
+Limitations
+-----------
+
+Types
+^^^^^
+
+* Only :class:`torch.Tensors`, numeric types that can be trivially converted to torch.Tensors (e.g. float, int),
+  and tuples and lists of those types are supported as model inputs or outputs. Dict and str inputs and
+  outputs are accepted in :ref:`tracing<tracing-vs-scripting>` mode, but:
+
+  * Any computation that depends on the value of a dict or a str input **will be replaced with the
+    constant value** seen during the one traced execution.
+  * Any output that is a dict will be silently replaced with a **flattened sequence of its values
+    (keys will be removed)**. E.g. ``{"foo": 1, "bar": 2}`` becomes ``(1, 2)``.
+  * Any output that is a str will be silently removed.
+
+* Certain operations involving tuples and lists are not supported in
+  :ref:`scripting<tracing-vs-scripting>` mode due to limited support in ONNX for nested sequences.
+  In particular appending a tuple to a list is not supported. In tracing mode, the nested sequences
+  will be flattened automatically during the tracing.
+
+Differences in Operator Implementations
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Due to differences in implementations of operators, running the exported model on different runtimes
+may produce different results from each other or from PyTorch. Normally these differences are
+numerically small, so this should only be a concern if your application is sensitive to these
+small differences.
+
+.. _tensor-indexing:
+
+Unsupported Tensor Indexing Patterns
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Tensor indexing patterns that cannot be exported are listed below.
+If you are experiencing issues exporting a model that does not include any of
+the unsupported patterns below, please double check that you are exporting with
+the latest ``opset_version``.
+
+Reads / Gets
+~~~~~~~~~~~~
+
+When indexing into a tensor for reading, the following patterns are not supported: ::
+
+  # Tensor indices that includes negative values.
+  data[torch.tensor([[1, 2], [2, -3]]), torch.tensor([-2, 3])]
+  # Workarounds: use positive index values.
+
+Writes / Sets
+~~~~~~~~~~~~~
+
+When indexing into a Tensor for writing, the following patterns are not supported: ::
+
+  # Multiple tensor indices if any has rank >= 2
+  data[torch.tensor([[1, 2], [2, 3]]), torch.tensor([2, 3])] = new_data
+  # Workarounds: use single tensor index with rank >= 2,
+  #              or multiple consecutive tensor indices with rank == 1.
+
+  # Multiple tensor indices that are not consecutive
+  data[torch.tensor([2, 3]), :, torch.tensor([1, 2])] = new_data
+  # Workarounds: transpose `data` such that tensor indices are consecutive.
+
+  # Tensor indices that includes negative values.
+  data[torch.tensor([1, -2]), torch.tensor([-2, 3])] = new_data
+  # Workarounds: use positive index values.
+
+  # Implicit broadcasting required for new_data.
+  data[torch.tensor([[0, 2], [1, 1]]), 1:3] = new_data
+  # Workarounds: expand new_data explicitly.
+  # Example:
+  #   data shape: [3, 4, 5]
+  #   new_data shape: [5]
+  #   expected new_data shape after broadcasting: [2, 2, 2, 5]
+
+Adding support for operators
+----------------------------
+
+When exporting a model that includes unsupported operators, you'll see an error message like:
+
+.. code-block:: text
+
+    RuntimeError: ONNX export failed: Couldn't export operator foo
+
+When that happens, there are a few things you can do:
+
+#. Change the model to not use that operator.
+#. Create a symbolic function to convert the operator and register it as a custom symbolic function.
+#. Contribute to PyTorch to add the same symbolic function to :mod:`torch.onnx` itself.
+
+If you decided to implement a symbolic function (we hope you will contribute it back to PyTorch!), here is how you can get started:
+
+ONNX exporter internals
+^^^^^^^^^^^^^^^^^^^^^^^
+
+A "symbolic function" is a function that decomposes a PyTorch operator into a
+composition of a series of ONNX operators.
+
+During export, each node (which contains a PyTorch operator) in the TorchScript
+graph is visited by the exporter in topological order.
+Upon visiting a node, the exporter looks for a registered symbolic functions for
+that operator. Symbolic functions are implemented in Python. A symbolic function for
+an op named ``foo`` would look something like::
+
+
+    def foo(
+      g,
+      input_0: torch._C.Value,
+      input_1: torch._C.Value) -> Union[None, torch._C.Value, List[torch._C.Value]]:
+      """
+      Adds the ONNX operations representing this PyTorch function by updating the
+      graph g with `g.op()` calls.
+
+      Args:
+        g (Graph): graph to write the ONNX representation into.
+        input_0 (Value): value representing the variables which contain
+            the first input for this operator.
+        input_1 (Value): value representing the variables which contain
+            the second input for this operator.
+
+      Returns:
+        A Value or List of Values specifying the ONNX nodes that compute something
+        equivalent to the original PyTorch operator with the given inputs.
+
+        None if it cannot be converted to ONNX.
+      """
+      ...
+
+The ``torch._C`` types are Python wrappers around the types defined in C++ in
+`ir.h <https://github.com/pytorch/pytorch/blob/main/torch/csrc/jit/ir/ir.h>`_.
+
+The process for adding a symbolic function depends on the type of operator.
+
+.. _adding-support-aten:
+
+ATen operators
+^^^^^^^^^^^^^^
+
+`ATen <https://pytorch.org/cppdocs/#aten>`_ is PyTorch's built-in tensor library.
+If the operator is an ATen operator (shows up in the TorchScript graph with the prefix
+``aten::``), make sure it is not supported already.
+
+List of supported operators
+~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Visit the auto generated :doc:`list of supported TorchScript operators <../onnx_torchscript_supported_aten_ops>`
+for details on which operator are supported in each ``opset_version``.
+
+Adding support for an aten or quantized operator
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+If the operator is not in the list above:
+
+* Define the symbolic function in ``torch/onnx/symbolic_opset<version>.py``, for example
+  `torch/onnx/symbolic_opset9.py <https://github.com/pytorch/pytorch/blob/main/torch/onnx/symbolic_opset9.py>`_.
+  Make sure the function has the same name as the ATen function, which may be declared in
+  ``torch/_C/_VariableFunctions.pyi`` or ``torch/nn/functional.pyi`` (these files are generated at
+  build time, so will not appear in your checkout until you build PyTorch).
+* By default, the first arg is the ONNX graph.
+  Other arg names must EXACTLY match the names in the ``.pyi`` file,
+  because dispatch is done with keyword arguments.
+* In the symbolic function, if the operator is in the
+  `ONNX standard operator set <https://github.com/onnx/onnx/blob/master/docs/Operators.md>`_,
+  we only need to create a node to represent the ONNX operator in the graph.
+  If not, we can compose several standard operators that have the
+  equivalent semantics to the ATen operator.
+
+Here is an example of handling missing symbolic function for the ``ELU`` operator.
+
+If we run the following code::
+
+    print(
+        torch.jit.trace(
+            torch.nn.ELU(), # module
+            torch.ones(1)   # example input
+        ).graph
+    )
+
+We see something like::
+
+  graph(%self : __torch__.torch.nn.modules.activation.___torch_mangle_0.ELU,
+        %input : Float(1, strides=[1], requires_grad=0, device=cpu)):
+    %4 : float = prim::Constant[value=1.]()
+    %5 : int = prim::Constant[value=1]()
+    %6 : int = prim::Constant[value=1]()
+    %7 : Float(1, strides=[1], requires_grad=0, device=cpu) = aten::elu(%input, %4, %5, %6)
+    return (%7)
+
+Since we see ``aten::elu`` in the graph, we know this is an ATen operator.
+
+We check the `ONNX operator list <https://github.com/onnx/onnx/blob/master/docs/Operators.md>`_,
+and confirm that ``Elu`` is standardized in ONNX.
+
+We find a signature for ``elu`` in ``torch/nn/functional.pyi``::
+
+    def elu(input: Tensor, alpha: float = ..., inplace: bool = ...) -> Tensor: ...
+
+We add the following lines to ``symbolic_opset9.py``::
+
+    def elu(g, input: torch.Value, alpha: torch.Value, inplace: bool = False):
+        return g.op("Elu", input, alpha_f=alpha)
+
+Now PyTorch is able to export models containing the ``aten::elu`` operator!
+
+See the ``torch/onnx/symbolic_opset*.py`` files for more examples.
+
+
+torch.autograd.Functions
+^^^^^^^^^^^^^^^^^^^^^^^^
+
+If the operator is a sub-class of :class:`torch.autograd.Function`, there are three ways
+to export it.
+
+Static Symbolic Method
+~~~~~~~~~~~~~~~~~~~~~~
+
+You can add a static method named ``symbolic`` to your function class. It should return
+ONNX operators that represent the function's behavior in ONNX. For example::
+
+    class MyRelu(torch.autograd.Function):
+        @staticmethod
+        def forward(ctx, input: torch.Tensor) -> torch.Tensor:
+            ctx.save_for_backward(input)
+            return input.clamp(min=0)
+
+        @staticmethod
+        def symbolic(g: torch.Graph, input: torch.Value) -> torch.Value:
+            return g.op("Clip", input, g.op("Constant", value_t=torch.tensor(0, dtype=torch.float)))
+
+.. FIXME(justinchuby): PythonOps are too complicated and the example below
+..  uses private methods we do not expose. We are looking to
+..  improve the experience. Since SymbolicContext is deprecated, we think
+..  defining a symbolic staticmethod is a better way to go for now.
+
+.. PythonOp Symbolic
+.. ~~~~~~~~~~~~~~~~~
+
+.. Alternatively, you can register a custom symbolic function.
+.. This gives the symbolic function access to more info through the
+.. ``torch.onnx.SymbolicContext`` object, which gets passed in as the first
+.. argument (before the ``Graph`` object).
+
+.. All autograd ``Function``\ s appear in the TorchScript graph as ``prim::PythonOp`` nodes.
+.. In order to differentiate between different ``Function`` subclasses, the
+.. symbolic function should use the ``name`` kwarg which gets set to the name of the class.
+
+.. Custom symbolic functions should add type and shape information by calling ``setType(...)``
+.. on Value objects before returning them (implemented in C++ by
+.. . ``torch::jit::Value::setType``). This is not required, but it can help the exporter's
+.. shape and type inference for down-stream nodes. For a non-trivial example of ``setType``, see
+.. ``test_aten_embedding_2`` in
+.. `test_operators.py <https://github.com/pytorch/pytorch/blob/main/test/onnx/test_operators.py>`_.
+
+.. The example below shows how you can access ``requires_grad`` via the ``Node`` object:
+
+..     class MyClip(torch.autograd.Function):
+..         @staticmethod
+..         def forward(ctx, input, min):
+..             ctx.save_for_backward(input)
+..             return input.clamp(min=min)
+
+..     class MyRelu(torch.autograd.Function):
+..         @staticmethod
+..         def forward(ctx, input):
+..             ctx.save_for_backward(input)
+..             return input.clamp(min=0)
+
+..     def symbolic_python_op(g: "GraphContext", *args, **kwargs):
+..         n = ctx.cur_node
+..         print("original node: ", n)
+..         for i, out in enumerate(n.outputs()):
+..             print("original output {}: {}, requires grad: {}".format(i, out, out.requiresGrad()))
+..         import torch.onnx.symbolic_helper as sym_helper
+..         for i, arg in enumerate(args):
+..             requires_grad = arg.requiresGrad() if sym_helper._is_value(arg) else False
+..             print("arg {}: {}, requires grad: {}".format(i, arg, requires_grad))
+
+..         name = kwargs["name"]
+..         ret = None
+..         if name == "MyClip":
+..             ret = g.op("Clip", args[0], args[1])
+..         elif name == "MyRelu":
+..             ret = g.op("Relu", args[0])
+..         else:
+..             # Logs a warning and returns None
+..             return _unimplemented("prim::PythonOp", "unknown node kind: " + name)
+..         # Copy type and shape from original node.
+..         ret.setType(n.type())
+..         return ret
+
+..     from torch.onnx import register_custom_op_symbolic
+.. .     register_custom_op_symbolic("prim::PythonOp", symbolic_python_op, 1)
+
+Inline Autograd Function
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+In cases where a static symbolic method is not provided for its subsequent :class:`torch.autograd.Function` or
+where a function to register ``prim::PythonOp`` as custom symbolic functions is not provided,
+:func:`torch.onnx.export` tries to inline the graph that corresponds to that :class:`torch.autograd.Function` such that
+this function is broken down into individual operators that were used within the function.
+The export should be successful as long as these individual operators are supported. For example::
+
+    class MyLogExp(torch.autograd.Function):
+        @staticmethod
+        def forward(ctx, input: torch.Tensor) -> torch.Tensor:
+            ctx.save_for_backward(input)
+            h = input.exp()
+            return h.log().log()
+
+There is no static symbolic method present for this model, yet it is exported as follows::
+
+    graph(%input : Float(1, strides=[1], requires_grad=0, device=cpu)):
+        %1 : float = onnx::Exp[](%input)
+        %2 : float = onnx::Log[](%1)
+        %3 : float = onnx::Log[](%2)
+        return (%3)
+
+If you need to avoid inlining of :class:`torch.autograd.Function`, you should export models with
+``operator_export_type`` set to ``ONNX_FALLTHROUGH`` or ``ONNX_ATEN_FALLBACK``.
+
+Custom operators
+^^^^^^^^^^^^^^^^
+
+You can export your model with custom operators that includes a combination of many standard ONNX ops,
+or are driven by self-defined C++ backend.
+
+ONNX-script functions
+~~~~~~~~~~~~~~~~~~~~~
+
+If an operator is not a standard ONNX op, but can be composed of multiple existing ONNX ops, you can utilize
+`ONNX-script <https://github.com/microsoft/onnx-script>`_ to create an external ONNX function to support the operator.
+You can export it by following this example::
+
+    import onnxscript
+    # There are three opset version needed to be aligned
+    # This is (1) the opset version in ONNX function
+    from onnxscript.onnx_opset import opset15 as op
+    opset_version = 15
+
+    x = torch.randn(1, 2, 3, 4, requires_grad=True)
+    model = torch.nn.SELU()
+
+    custom_opset = onnxscript.values.Opset(domain="onnx-script", version=1)
+
+    @onnxscript.script(custom_opset)
+    def Selu(X):
+        alpha = 1.67326  # auto wrapped as Constants
+        gamma = 1.0507
+        alphaX = op.CastLike(alpha, X)
+        gammaX = op.CastLike(gamma, X)
+        neg = gammaX * (alphaX * op.Exp(X) - alphaX)
+        pos = gammaX * X
+        zero = op.CastLike(0, X)
+        return op.Where(X <= zero, neg, pos)
+
+    # setType API provides shape/type to ONNX shape/type inference
+    def custom_selu(g: jit_utils.GraphContext, X):
+        return g.onnxscript_op(Selu, X).setType(X.type())
+
+    # Register custom symbolic function
+    # There are three opset version needed to be aligned
+    # This is (2) the opset version in registry
+    torch.onnx.register_custom_op_symbolic(
+        symbolic_name="aten::selu",
+        symbolic_fn=custom_selu,
+        opset_version=opset_version,
+    )
+
+    # There are three opset version needed to be aligned
+    # This is (2) the opset version in exporter
+    torch.onnx.export(
+        model,
+        x,
+        "model.onnx",
+        opset_version=opset_version,
+        # only needed if you want to specify an opset version > 1.
+        custom_opsets={"onnx-script": 2}
+    )
+
+The example above exports it as a custom operator in the "onnx-script" opset.
+When exporting a custom operator, you can specify the custom domain version using the
+``custom_opsets`` dictionary at export. If not specified, the custom opset version defaults to 1.
+
+NOTE: Be careful to align the opset version mentioned in the above example, and make sure they are consumed in exporter step.
+The example usage of how to write a onnx-script function is a beta version in terms of the active development on onnx-script.
+Please follow the latest `ONNX-script <https://github.com/microsoft/onnx-script>`_
+
+C++ Operators
+~~~~~~~~~~~~~
+
+If a model uses a custom operator implemented in C++ as described in
+`Extending TorchScript with Custom C++ Operators <https://pytorch.org/tutorials/advanced/torch_script_custom_ops.html>`_,
+you can export it by following this example::
+
+    from torch.onnx import symbolic_helper
+
+
+    # Define custom symbolic function
+    @symbolic_helper.parse_args("v", "v", "f", "i")
+    def symbolic_foo_forward(g, input1, input2, attr1, attr2):
+        return g.op("custom_domain::Foo", input1, input2, attr1_f=attr1, attr2_i=attr2)
+
+
+    # Register custom symbolic function
+    torch.onnx.register_custom_op_symbolic("custom_ops::foo_forward", symbolic_foo_forward, 9)
+
+
+    class FooModel(torch.nn.Module):
+        def __init__(self, attr1, attr2):
+            super().__init__()
+            self.attr1 = attr1
+            self.attr2 = attr2
+
+        def forward(self, input1, input2):
+            # Calling custom op
+            return torch.ops.custom_ops.foo_forward(input1, input2, self.attr1, self.attr2)
+
+
+    model = FooModel(attr1, attr2)
+    torch.onnx.export(
+        model,
+        (example_input1, example_input1),
+        "model.onnx",
+        # only needed if you want to specify an opset version > 1.
+        custom_opsets={"custom_domain": 2}
+    )
+
+The example above exports it as a custom operator in the "custom_domain" opset.
+When exporting a custom operator, you can specify the custom domain version using the
+``custom_opsets`` dictionary at export. If not specified, the custom opset version defaults to 1.
+
+The runtime that consumes the model needs to support the custom op. See
+`Caffe2 custom ops <https://caffe2.ai/docs/custom-operators.html>`_,
+`ONNX Runtime custom ops <https://onnxruntime.ai/docs/reference/operators/add-custom-op.html>`_,
+or your runtime of choice's documentation.
+
+
+Discovering all unconvertible ATen ops at once
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+When export fails due to an unconvertible ATen op, there may in fact be more
+than one such op but the error message only mentions the first. To discover
+all of the unconvertible ops in one go you can::
+
+    # prepare model, args, opset_version
+    ...
+
+    torch_script_graph, unconvertible_ops = torch.onnx.utils.unconvertible_ops(
+        model, args, opset_version=opset_version
+    )
+
+    print(set(unconvertible_ops))
+
+The set is approximated because some ops may be removed during the conversion
+process and don't need to be converted. Some other ops may have partial support
+that will fail conversion with particular inputs, but this should give you a
+general idea of what ops are not supported. Please feel free to open GitHub Issues
+for op support requests.
+
+Frequently Asked Questions
+--------------------------
+Q: I have exported my LSTM model, but its input size seems to be fixed?
+
+  The tracer records the shapes of the example inputs. If the model should accept
+  inputs of dynamic shapes, set ``dynamic_axes`` when calling :func:`torch.onnx.export`.
+
+Q: How to export models containing loops?
+
+  See `Tracing vs Scripting`_.
+
+Q: How to export models with primitive type inputs (e.g. int, float)?
+
+  Support for primitive numeric type inputs was added in PyTorch 1.9.
+  However, the exporter does not support models with str inputs.
+
+Q: Does ONNX support implicit scalar datatype casting?
+
+  The ONNX standard does not, but the exporter will try to handle that part.
+  Scalars are exported as constant tensors.
+  The exporter will figure out the right data type for scalars. In rare cases when it is unable
+  to do so, you will need to manually specify the datatype with e.g. `dtype=torch.float32`.
+  If you see any errors, please [create a GitHub issue](https://github.com/pytorch/pytorch/issues).
+
+Q: Are lists of Tensors exportable to ONNX?
+
+  Yes, for ``opset_version`` >= 11, since ONNX introduced the Sequence type in opset 11.
+
+Python API
+----------
+
+.. automodule:: torch.onnx
+
+Functions
+^^^^^^^^^
+
+.. autofunction:: export
+.. autofunction:: export_to_pretty_string
+.. autofunction:: register_custom_op_symbolic
+.. autofunction:: unregister_custom_op_symbolic
+.. autofunction:: select_model_mode_for_export
+.. autofunction:: is_in_onnx_export
+.. autofunction:: enable_log
+.. autofunction:: disable_log
+.. autofunction:: torch.onnx.verification.find_mismatch
+
+Classes
+^^^^^^^
+
+.. autosummary::
+    :toctree: generated
+    :nosignatures:
+    :template: classtemplate.rst
+
+    JitScalarType
+    torch.onnx.verification.GraphInfo
+    torch.onnx.verification.VerificationOptions

docs/source/onnx_torchscript_supported_aten_ops.rst

@@ -5,7 +5,7 @@ ONNX supported TorchScript operators
 
 .. This file is automatically generated during the documentation build
 .. by cross referencing ONNX operator symbolics with TorchScript operators via
-.. ``docs/source/scripts/build_onnx_supported_aten_op_csv_table.py``.
+.. ``docs/source/scripts/build_onnx_torchscript_supported_aten_op_csv_table.py``.
 .. Do not modify directly and instead `rebuild the docs <https://github.com/pytorch/pytorch#building-the-documentation>`_.
 
 This page lists the TorchScript operators that are supported/unsupported by ONNX export.

docs/source/scripts/exportdb/generate_example_rst.py

@@ -119,7 +119,9 @@ def generate_index_rst(example_cases, tag_to_modules, support_level_to_modules):
         blurb = file.read()
 
     # Generate contents of the .rst file
-    doc_contents = f"""ExportDB
+    doc_contents = f""".. _torch.export_db:
+
+ExportDB
 ========
 
 {blurb}

docs/source/tensors.rst

@@ -575,6 +575,7 @@ Tensor class reference
     Tensor.reciprocal_
     Tensor.record_stream
     Tensor.register_hook
+    Tensor.register_post_accumulate_grad_hook
     Tensor.remainder
     Tensor.remainder_
     Tensor.renorm

docs/source/torch.compiler.rst

@@ -54,6 +54,10 @@ Some of the most commonly used backends include:
      - Uses the TorchInductor backend. `Read more <https://dev-discuss.pytorch.org/t/torchinductor-a-pytorch-native-compiler-with-define-by-run-ir-and-symbolic-shapes/747>`__
    * - ``torch.compile(m, backend="cudagraphs")``
      - CUDA graphs with AOT Autograd. `Read more <https://github.com/pytorch/torchdynamo/pull/757>`__
+   * - ``torch.compile(m, backend="ipex")``
+     - Uses IPEX on CPU. `Read more <https://github.com/intel/intel-extension-for-pytorch>`__
+   * - ``torch.compile(m, backend="onnxrt")``
+     - Uses ONNX Runtime for training on CPU/GPU. :doc:`Read more <onnx_dynamo_onnxruntime_backend>`
 
 **Inference-only backends**
 
@@ -63,10 +67,8 @@ Some of the most commonly used backends include:
 
    * - Backend
      - Description
-   * - ``torch.compile(m, backend="onnxrt")``
-     - Uses ONNXRT for inference on CPU/GPU. `Read more <https://onnxruntime.ai/>`__
    * - ``torch.compile(m, backend="tensorrt")``
-     - Uses ONNXRT to run TensorRT for inference optimizations. `Read more <https://github.com/onnx/onnx-tensorrt>`__
+     - Uses ONNX Runtime to run TensorRT for inference optimizations. `Read more <https://github.com/onnx/onnx-tensorrt>`__
    * - ``torch.compile(m, backend="ipex")``
      - Uses IPEX for inference on CPU. `Read more <https://github.com/intel/intel-extension-for-pytorch>`__
    * - ``torch.compile(m, backend="tvm")``

docs/source/torch.compiler_faq.rst

@@ -317,47 +317,12 @@ them by default: ``env TORCHDYNAMO_DYNAMIC_SHAPES=0 python model.py`` 2.
 CUDA graphs with Triton are enabled by default in inductor but removing
 them may alleviate some OOM issues: ``torch._inductor.config.triton.cudagraphs = False``.
 
-``torch.func`` does not work with ``torch.compile``
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+``torch.func`` works with ``torch.compile`` (for `grad` and `vmap` transforms)
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 Applying a ``torch.func`` transform to a function that uses ``torch.compile``
 does not work:
 
-.. code-block:: python
-
-    import torch
-
-    @torch.compile
-    def f(x):
-        return torch.sin(x)
-
-    def g(x):
-        return torch.grad(f)(x)
-
-    x = torch.randn(2, 3)
-    g(x)
-
-As a workaround, use ``torch.compile`` outside of the ``torch.func`` function:
-
-.. code-block:: python
-
-    import torch
-
-    def f(x):
-        return torch.sin(x)
-
-    @torch.compile
-    def g(x):
-        return torch.vmap(f)(x)
-
-    x = torch.randn(2, 3)
-    g(x)
-
-Applying a ``torch.func`` transform to a function handled with ``torch.compile``
---------------------------------------------------------------------------------
-
-For example, you have the following code:
-
 .. code-block:: python
 
     import torch
@@ -374,12 +339,18 @@ For example, you have the following code:
 
 This code will not work. There is an `issue <https://github.com/pytorch/pytorch/issues/100320>`__
 that you can track for this.
-As a workaround, please put the ``torch.compile`` outside of ``torch.func`` transform:
+
+As a workaround, use ``torch.compile`` outside of the ``torch.func`` function:
+
+.. note::
+    This is an experimental feature and can be used by setting `torch._dynamo.config.capture_func_transforms=True`
 
 .. code-block:: python
 
     import torch
 
+    torch._dynamo.config.capture_func_transforms=True
+
     def f(x):
         return torch.sin(x)
 
@@ -393,18 +364,137 @@ As a workaround, please put the ``torch.compile`` outside of ``torch.func`` tran
 Calling ``torch.func`` transform inside of a function handled with ``torch.compile``
 ------------------------------------------------------------------------------------
 
+
+Compiling ``torch.func.grad`` with ``torch.compile``
+----------------------------------------------------
+
 .. code-block:: python
 
     import torch
 
-    @torch.compile
-    def f(x):
-        return torch.vmap(torch.sum)(x)
+    torch._dynamo.config.capture_func_transforms=True
 
-    x = torch.randn(2, 3)
-    f(x)
+    def wrapper_fn(x):
+        return torch.func.grad(lambda x: x.sin().sum())(x)
 
-This doesn't work yet. As a workaround, use ``torch._dynamo.allow_in_graph``
+    x = torch.randn(3, 3, 3)
+    grad_x = torch.compile(wrapper_fn)(x)
+
+Compiling ``torch.vmap`` with ``torch.compile``
+-----------------------------------------------
+
+.. code-block:: python
+
+    import torch
+
+    torch._dynamo.config.capture_func_transforms=True
+
+    def my_fn(x):
+        return torch.vmap(lambda x: x.sum(1))(x)
+
+    x = torch.randn(3, 3, 3)
+    output = torch.compile(my_fn)(x)
+
+Limitations
+-----------
+
+There are currently a few cases which are not supported and lead to graph breaks
+(that is, torch.compile falls back to eager-mode PyTorch on these). We are working
+on improving the situation for the next release (PyTorch 2.2)
+
+1. The inputs and outputs of the function being transformed over must be tensors.
+We do not yet support things like tuple of Tensors.
+
+.. code-block:: python
+
+    import torch
+
+    torch._dynamo.config.capture_func_transforms=True
+
+    def fn(x):
+        x1, x2 = x
+        return x1 + x2
+
+    def my_fn(x):
+        return torch.func.vmap(fn)(x)
+
+    x1 = torch.randn(3, 3, 3)
+    x2 = torch.randn(3, 3, 3)
+    # Unsupported, falls back to eager-mode PyTorch
+    output = torch.compile(my_fn)((x1, x2))
+
+2. Keyword arguments are not supported.
+
+.. code-block:: python
+
+    import torch
+
+    torch._dynamo.config.capture_func_transforms=True
+
+    def fn(x, y):
+        return (x + y).sum()
+
+    def my_fn(x, y):
+        return torch.func.grad(fn)(x, y=y)
+
+    x = torch.randn(3, 3)
+    y = torch.randn(3, 3)
+    # Unsupported, falls back to eager-mode PyTorch
+    output = torch.compile(my_fn)(x, y)
+
+3. Functions with observable side effects. For example, it is OK to mutate a list created in the function,
+but not OK to mutate a list created outside of the function.
+
+.. code-block:: python
+
+    import torch
+
+    torch._dynamo.config.capture_func_transforms=True
+
+    some_list = []
+
+    def f(x, y):
+        some_list.append(1)
+        return x + y
+
+    def my_fn(x, y):
+        return torch.func.vmap(f)(x, y)
+
+    x = torch.ones(2, 3)
+    y = torch.randn(2, 3)
+    # Unsupported, falls back to eager-mode PyTorch
+    output = torch.compile(my_fn)(x, y)
+
+4. ``torch.vmap`` over a function that calls one or more operators in the following list.
+
+.. note::
+    'stride', 'requires_grad', 'storage_offset', 'layout', 'data', 'is_coalesced', 'is_complex',
+    'is_conj', 'is_contiguous', 'is_cpu', 'is_cuda', 'is_distributed', 'is_floating_point',
+    'is_inference', 'is_ipu', 'is_leaf', 'is_meta', 'is_mkldnn', 'is_mps', 'is_neg', 'is_nested',
+    'is_nonzero', 'is_ort', 'is_pinned', 'is_quantized', 'is_same_size', 'is_set_to', 'is_shared',
+    'is_signed', 'is_sparse', 'is_sparse_csr', 'is_vulkan', 'is_xla', 'is_xpu'
+
+.. code-block:: python
+
+    import torch
+
+    torch._dynamo.config.capture_func_transforms=True
+
+    def bad_fn(x):
+        x.stride()
+        return x
+
+    def my_fn(x):
+        return torch.func.vmap(bad_fn)(x)
+
+    x = torch.randn(3, 3, 3)
+    # Unsupported, falls back to eager-mode PyTorch
+    output = torch.compile(my_fn)(x)
+
+Compiling functions besides the ones which are supported (escape hatch)
+-----------------------------------------------------------------------
+
+For other transforms, as a workaround, use ``torch._dynamo.allow_in_graph``
 
 ``allow_in_graph`` is an escape hatch. If your code does not work with
 ``torch.compile``, which introspects Python bytecode, but you believe it

docs/source/torch.compiler_ir.rst

@@ -1,3 +1,5 @@
+.. _torch.compiler_ir:
+
 IRs
 ===============
 

ios/LibTorch-Lite.podspec.template

@@ -1,6 +1,6 @@
 Pod::Spec.new do |s|
     s.name             = 'LibTorch-Lite'
-    s.version          = '1.13.0'
+    s.version          = 'IOS_BUILD_VERSION'
     s.authors          = 'PyTorch Team'
     s.license          = { :type => 'BSD' }
     s.homepage         = 'https://github.com/pytorch/pytorch'
@@ -33,5 +33,5 @@ Pod::Spec.new do |s|
         'VALID_ARCHS' => 'x86_64 arm64'
     }
     s.library = ['c++', 'stdc++']
-    s.frameworks = 'Accelerate'
+    s.frameworks = 'Accelerate', 'MetalPerformanceShaders', 'CoreML'
 end

ios/LibTorch.podspec.template

@@ -1,6 +1,6 @@
 Pod::Spec.new do |s|
     s.name             = 'LibTorch'
-    s.version          = '1.13.0'
+    s.version          = 'IOS_BUILD_VERSION'
     s.authors          = 'PyTorch Team'
     s.license          = { :type => 'BSD' }
     s.homepage         = 'https://github.com/pytorch/pytorch'
@@ -33,5 +33,5 @@ Pod::Spec.new do |s|
         'VALID_ARCHS' => 'x86_64 arm64'
     }
     s.library = ['c++', 'stdc++']
-    s.frameworks = 'Accelerate'
+    s.frameworks = 'Accelerate', 'MetalPerformanceShaders', 'CoreML'
 end

ios/TestApp/benchmark/coreml_backend.py

@@ -32,6 +32,7 @@ def main():
     mlmodel = torch._C._jit_to_backend("coreml", model, compile_spec)
     print(mlmodel._c._get_method("forward").graph)
     mlmodel._save_for_lite_interpreter("../models/model_coreml.ptl")
+    torch.jit.save(mlmodel, "../models/model_coreml.pt")
 
 
 if __name__ == "__main__":

ios/TestApp/benchmark/trace_model.py

@@ -9,5 +9,5 @@ traced_script_module = torch.jit.trace(model, example)
 optimized_scripted_module = optimize_for_mobile(traced_script_module)
 torch.jit.save(optimized_scripted_module, "../models/model.pt")
 exported_optimized_scripted_module = (
-    optimized_scripted_module._save_for_lite_interpreter("../models/model_lite.ptl")
+    optimized_scripted_module._save_for_lite_interpreter("../models/model.ptl")
 )

setup.py

@@ -1069,6 +1069,50 @@ def configure_extension_build():
     return extensions, cmdclass, packages, entry_points, extra_install_requires
 
 
+def add_triton(install_requires, extras_require) -> None:
+    """
+    Add triton package as a dependency when it's needed
+    """
+    # NB: If the installation requirments list already includes triton dependency,
+    # there is no need to add it one more time as an extra dependency. In nightly
+    # or when release PyTorch, that is done by setting PYTORCH_EXTRA_INSTALL_REQUIREMENTS
+    # environment variable on pytorch/builder
+    has_triton = any("triton" in pkg for pkg in install_requires)
+    if has_triton:
+        return
+
+    cmake_cache_vars = get_cmake_cache_vars()
+    use_rocm = cmake_cache_vars["USE_ROCM"]
+    use_cuda = cmake_cache_vars["USE_CUDA"]
+
+    # Triton is only needed for CUDA or ROCm
+    if not use_rocm and not use_cuda:
+        return
+
+    if use_rocm:
+        triton_text_file = "triton-rocm.txt"
+        triton_package_name = "pytorch-triton-rocm"
+    else:
+        triton_text_file = "triton.txt"
+        triton_package_name = "pytorch-triton"
+    triton_pin_file = os.path.join(
+        cwd, ".ci", "docker", "ci_commit_pins", triton_text_file
+    )
+    triton_version_file = os.path.join(cwd, ".ci", "docker", "triton_version.txt")
+
+    if os.path.exists(triton_pin_file) and os.path.exists(triton_version_file):
+        with open(triton_pin_file) as f:
+            triton_pin = f.read().strip()
+        with open(triton_version_file) as f:
+            triton_version = f.read().strip()
+
+        if "dynamo" not in extras_require:
+            extras_require["dynamo"] = []
+        extras_require["dynamo"].append(
+            triton_package_name + "==" + triton_version + "+" + triton_pin[:10]
+        )
+
+
 # post run, warnings, printed at the end to make them more visible
 build_update_message = """
     It is no longer necessary to use the 'build' or 'rebuild' targets
@@ -1102,29 +1146,6 @@ def main():
         "fsspec",
     ]
 
-    extras_require = {"opt-einsum": ["opt-einsum>=3.3"]}
-    if platform.system() == "Linux":
-        cmake_cache_vars = get_cmake_cache_vars()
-        if cmake_cache_vars["USE_ROCM"]:
-            triton_text_file = "triton-rocm.txt"
-            triton_package_name = "pytorch-triton-rocm"
-        else:
-            triton_text_file = "triton.txt"
-            triton_package_name = "pytorch-triton"
-        triton_pin_file = os.path.join(
-            cwd, ".ci", "docker", "ci_commit_pins", triton_text_file
-        )
-        triton_version_file = os.path.join(cwd, ".ci", "docker", "triton_version.txt")
-        if os.path.exists(triton_pin_file) and os.path.exists(triton_version_file):
-            with open(triton_pin_file) as f:
-                triton_pin = f.read().strip()
-            with open(triton_version_file) as f:
-                triton_version = f.read().strip()
-            extras_require["dynamo"] = [
-                triton_package_name + "==" + triton_version + "+" + triton_pin[:10],
-                "jinja2",
-            ]
-
     # Parse the command line and check the arguments before we proceed with
     # building deps and setup. We need to set values so `--help` works.
     dist = Distribution()
@@ -1150,6 +1171,14 @@ def main():
 
     install_requires += extra_install_requires
 
+    extras_require = {
+        "opt-einsum": ["opt-einsum>=3.3"],
+    }
+    # Triton is only available on Linux atm
+    if platform.system() == "Linux":
+        extras_require["dynamo"] = ["jinja2"]
+        add_triton(install_requires=install_requires, extras_require=extras_require)
+
     # Read in README.md for our long_description
     with open(os.path.join(cwd, "README.md"), encoding="utf-8") as f:
         long_description = f.read()
@@ -1209,6 +1238,7 @@ def main():
         "include/ATen/native/mps/*.h",
         "include/ATen/native/quantized/*.h",
         "include/ATen/native/quantized/cpu/*.h",
+        "include/ATen/native/utils/*.h",
         "include/ATen/quantized/*.h",
         "include/caffe2/serialize/*.h",
         "include/c10/*.h",