[async_tp] Base support ag-transpose-mm(mat_B) case

ghstack-source-id: edd51b9c46e46e8eca0c45e0ea53c1b26b375c01
Pull Request resolved: https://github.com/pytorch/pytorch/pull/163069

.ci/aarch64_linux/aarch64_ci_build.sh

@@ -5,9 +5,9 @@ GPU_ARCH_VERSION=${GPU_ARCH_VERSION:-}
 
 # Set CUDA architecture lists to match x86 build_cuda.sh
 if [[ "$GPU_ARCH_VERSION" == *"12.6"* ]]; then
-    export TORCH_CUDA_ARCH_LIST="8.0;9.0"
+    export TORCH_CUDA_ARCH_LIST="5.0;6.0;7.0;8.0;9.0"
 elif [[ "$GPU_ARCH_VERSION" == *"12.8"* ]]; then
-    export TORCH_CUDA_ARCH_LIST="8.0;9.0;10.0;12.0"
+    export TORCH_CUDA_ARCH_LIST="7.0;8.0;9.0;10.0;12.0"
 elif [[ "$GPU_ARCH_VERSION" == *"13.0"* ]]; then
     export TORCH_CUDA_ARCH_LIST="8.0;9.0;10.0;11.0;12.0+PTX"
 fi
@@ -15,8 +15,6 @@ fi
 # Compress the fatbin with -compress-mode=size for CUDA 13
 if [[ "$DESIRED_CUDA" == *"13"* ]]; then
     export TORCH_NVCC_FLAGS="-compress-mode=size"
-    # Bundle ptxas into the cu13 wheel, see https://github.com/pytorch/pytorch/issues/163801
-    export BUILD_BUNDLE_PTXAS=1
 fi
 
 SCRIPTPATH="$( cd -- "$(dirname "$0")" >/dev/null 2>&1 ; pwd -P )"
@@ -44,6 +42,9 @@ else
         echo "Bundling CUDA libraries with wheel for aarch64."
     else
         echo "Using nvidia libs from pypi for aarch64."
+        # Fix platform constraints in PYTORCH_EXTRA_INSTALL_REQUIREMENTS for aarch64
+        # Replace 'platform_machine == "x86_64"' with 'platform_machine == "aarch64"'
+        export PYTORCH_EXTRA_INSTALL_REQUIREMENTS="${PYTORCH_EXTRA_INSTALL_REQUIREMENTS//platform_machine == \'x86_64\'/platform_machine == \'aarch64\'}"
         echo "Updated PYTORCH_EXTRA_INSTALL_REQUIREMENTS for aarch64: $PYTORCH_EXTRA_INSTALL_REQUIREMENTS"
         export USE_NVIDIA_PYPI_LIBS=1
     fi

.ci/aarch64_linux/aarch64_wheel_ci_build.py

@@ -138,8 +138,6 @@ def package_cuda_wheel(wheel_path, desired_cuda) -> None:
     folder = os.path.dirname(wheel_path)
     os.mkdir(f"{folder}/tmp")
     os.system(f"unzip {wheel_path} -d {folder}/tmp")
-    # Delete original wheel since it will be repackaged
-    os.system(f"rm {wheel_path}")
 
     # Check if we should use PyPI NVIDIA libraries or bundle system libraries
     use_nvidia_pypi_libs = os.getenv("USE_NVIDIA_PYPI_LIBS", "0") == "1"
@@ -213,8 +211,7 @@ def package_cuda_wheel(wheel_path, desired_cuda) -> None:
         ]
 
         # CUDA version-specific libraries
-        if "13" in desired_cuda:
-            minor_version = desired_cuda[-1]
+        if "130" in desired_cuda:
             version_specific_libs = [
                 "/usr/local/cuda/extras/CUPTI/lib64/libcupti.so.13",
                 "/usr/local/cuda/lib64/libcublas.so.13",
@@ -224,7 +221,7 @@ def package_cuda_wheel(wheel_path, desired_cuda) -> None:
                 "/usr/local/cuda/lib64/libcusolver.so.12",
                 "/usr/local/cuda/lib64/libnvJitLink.so.13",
                 "/usr/local/cuda/lib64/libnvrtc.so.13",
-                f"/usr/local/cuda/lib64/libnvrtc-builtins.so.13.{minor_version}",
+                "/usr/local/cuda/lib64/libnvrtc-builtins.so.13.0",
             ]
         elif "12" in desired_cuda:
             # Get the last character for libnvrtc-builtins version (e.g., "129" -> "9")
@@ -240,8 +237,6 @@ def package_cuda_wheel(wheel_path, desired_cuda) -> None:
                 "/usr/local/cuda/lib64/libnvrtc.so.12",
                 f"/usr/local/cuda/lib64/libnvrtc-builtins.so.12.{minor_version}",
             ]
-        else:
-            raise ValueError(f"Unsupported CUDA version: {desired_cuda}.")
 
         # Combine all libraries
         libs_to_copy = common_libs + version_specific_libs
@@ -280,7 +275,14 @@ def complete_wheel(folder: str) -> str:
             f"/{folder}/dist/{repaired_wheel_name}",
         )
     else:
-        repaired_wheel_name = list_dir(f"/{folder}/dist")[0]
+        repaired_wheel_name = wheel_name.replace(
+            "linux_aarch64", "manylinux_2_28_aarch64"
+        )
+        print(f"Renaming {wheel_name} wheel to {repaired_wheel_name}")
+        os.rename(
+            f"/{folder}/dist/{wheel_name}",
+            f"/{folder}/dist/{repaired_wheel_name}",
+        )
 
     print(f"Copying {repaired_wheel_name} to artifacts")
     shutil.copy2(

.ci/docker/build.sh

@@ -214,7 +214,8 @@ case "$tag" in
     TRITON=yes
     ;;
   pytorch-linux-jammy-py3-gcc11-inductor-benchmarks)
-    ANACONDA_PYTHON_VERSION=3.10
+    # TODO (huydhn): Upgrade this to Python >= 3.10
+    ANACONDA_PYTHON_VERSION=3.9
     GCC_VERSION=11
     VISION=yes
     KATEX=yes
@@ -262,10 +263,13 @@ case "$tag" in
     TRITON_CPU=yes
     ;;
   pytorch-linux-jammy-linter)
-    PYTHON_VERSION=3.10
+    # TODO: Use 3.9 here because of this issue https://github.com/python/mypy/issues/13627.
+    # We will need to update mypy version eventually, but that's for another day. The task
+    # would be to upgrade mypy to 1.0.0 with Python 3.11
+    PYTHON_VERSION=3.9
     ;;
-  pytorch-linux-jammy-cuda12.8-cudnn9-py3.10-linter)
-    PYTHON_VERSION=3.10
+  pytorch-linux-jammy-cuda12.8-cudnn9-py3.9-linter)
+    PYTHON_VERSION=3.9
     CUDA_VERSION=12.8.1
     ;;
   pytorch-linux-jammy-aarch64-py3.10-gcc11)

.ci/docker/ci_commit_pins/triton.txt

@@ -1 +1 @@
-bbb06c0334a6772b92d24bde54956e675c8c6604
+fccfc522864cf8bc172abe0cd58ae5581e2d44b9

.ci/docker/common/patch_libstdc.sh

@@ -1,9 +0,0 @@
-#!/bin/bash
-set -xe
-# Script used in Linux x86 and aarch64 CD pipeline
-
-# Workaround for exposing statically linked libstdc++ CXX11 ABI symbols.
-# see: https://github.com/pytorch/pytorch/issues/133437
-LIBNONSHARED=$(gcc -print-file-name=libstdc++_nonshared.a)
-nm -g $LIBNONSHARED | grep " T " | grep recursive_directory_iterator | cut -c 20-  > weaken-symbols.txt
-objcopy --weaken-symbols weaken-symbols.txt $LIBNONSHARED $LIBNONSHARED

.ci/docker/manywheel/Dockerfile_2_28

@@ -130,8 +130,7 @@ ENV LD_LIBRARY_PATH=/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr/lib64:/op
 RUN for cpython_version in "cp312-cp312" "cp313-cp313" "cp313-cp313t"; do \
     /opt/python/${cpython_version}/bin/python -m pip install setuptools wheel; \
     done;
-ADD ./common/patch_libstdc.sh patch_libstdc.sh
-RUN bash ./patch_libstdc.sh && rm patch_libstdc.sh
+
 
 # cmake-3.18.4 from pip; force in case cmake3 already exists
 RUN yum install -y python3-pip && \

.ci/docker/manywheel/Dockerfile_2_28_aarch64

@@ -71,5 +71,3 @@ RUN rm -rf /opt/python/cp33-cp33m /opt/_internal/cpython-3.3.6
 RUN rm -rf /opt/python/cp34-cp34m /opt/_internal/cpython-3.4.6
 COPY --from=openblas     /opt/OpenBLAS/  /opt/OpenBLAS/
 ENV LD_LIBRARY_PATH=/opt/OpenBLAS/lib:$LD_LIBRARY_PATH
-ADD ./common/patch_libstdc.sh patch_libstdc.sh
-RUN bash ./patch_libstdc.sh && rm patch_libstdc.sh

.ci/docker/manywheel/Dockerfile_cuda_aarch64

@@ -95,5 +95,3 @@ COPY --from=nvpl /opt/nvpl/lib/  /usr/local/lib/
 COPY --from=nvpl /opt/nvpl/include/  /usr/local/include/
 RUN ln -sf /usr/local/cuda-${BASE_CUDA_VERSION} /usr/local/cuda
 ENV PATH=/usr/local/cuda/bin:$PATH
-ADD ./common/patch_libstdc.sh patch_libstdc.sh
-RUN bash ./patch_libstdc.sh && rm patch_libstdc.sh

.ci/docker/requirements-ci.txt

@@ -93,9 +93,8 @@ librosa==0.10.2 ; python_version == "3.12" and platform_machine != "s390x"
 #Pinned versions:
 #test that import:
 
-mypy==1.16.0 ; platform_system != "Windows"
+mypy==1.16.0
 # Pin MyPy version because new errors are likely to appear with each release
-# Skip on Windows as lots of type annotations are POSIX specific
 #Description: linter
 #Pinned versions: 1.16.0
 #test that import: test_typing.py, test_type_hints.py

.ci/docker/requirements-docs.txt

@@ -1,7 +1,7 @@
 sphinx==5.3.0
 #Description: This is used to generate PyTorch docs
 #Pinned versions: 5.3.0
--e git+https://github.com/pytorch/pytorch_sphinx_theme.git@71e55749be14ceb56e7f8211a9fb649866b87ad4#egg=pytorch_sphinx_theme2
+-e git+https://github.com/pytorch/pytorch_sphinx_theme.git@1657ad2fc1acdc98aa719eebecbb0128a7c13ce4#egg=pytorch_sphinx_theme2
 
 # TODO: sphinxcontrib.katex 0.9.0 adds a local KaTeX server to speed up pre-rendering
 # but it doesn't seem to work and hangs around idly. The initial thought that it is probably

.ci/libtorch/build.sh

@@ -7,4 +7,4 @@ set -ex
 
 SCRIPTPATH="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
 
-USE_NVSHMEM=0 USE_CUSPARSELT=0 BUILD_PYTHONLESS=1 DESIRED_PYTHON="3.10" ${SCRIPTPATH}/../manywheel/build.sh
+USE_NVSHMEM=0 USE_CUSPARSELT=0 BUILD_PYTHONLESS=1 DESIRED_PYTHON="3.9" ${SCRIPTPATH}/../manywheel/build.sh

.ci/lumen_cli/cli/lib/core/vllm/lib.py

@@ -41,6 +41,7 @@ def sample_vllm_test_library():
                 "pytest -v -s basic_correctness/test_cumem.py",
                 "pytest -v -s basic_correctness/test_basic_correctness.py",
                 "pytest -v -s basic_correctness/test_cpu_offload.py",
+                "VLLM_TEST_ENABLE_ARTIFICIAL_PREEMPT=1 pytest -v -s basic_correctness/test_preemption.py",
             ],
         },
         "vllm_basic_models_test": {
@@ -67,12 +68,15 @@ def sample_vllm_test_library():
                         "-v",
                         "-s",
                         "entrypoints/llm",
+                        "--ignore=entrypoints/llm/test_lazy_outlines.py",
                         "--ignore=entrypoints/llm/test_generate.py",
+                        "--ignore=entrypoints/llm/test_generate_multiple_loras.py",
                         "--ignore=entrypoints/llm/test_collective_rpc.py",
                     ]
                 ),
-                "pytest -v -s entrypoints/llm/test_generate.py",
-                "pytest -v -s entrypoints/offline_mode",
+                "pytest -v -s entrypoints/llm/test_lazy_outlines.py",
+                "pytest -v -s entrypoints/llm/test_generate.py ",
+                "VLLM_USE_V1=0 pytest -v -s entrypoints/offline_mode",
             ],
         },
         "vllm_regression_test": {

.ci/pytorch/check_binary.sh

@@ -67,7 +67,7 @@ fi
 #       wheels with cxx11-abi
 
 echo "Checking that the gcc ABI is what we expect"
-if [[ "$(uname)" != 'Darwin' &&  "$(uname -m)" != "s390x" ]]; then
+if [[ "$(uname)" != 'Darwin' ]]; then
   # We also check that there are cxx11 symbols in libtorch
   #
   echo "Checking that symbols in libtorch.so have the right gcc abi"

.ci/pytorch/common_utils.sh

@@ -284,7 +284,7 @@ function install_torchrec_and_fbgemm() {
 
 function clone_pytorch_xla() {
   if [[ ! -d ./xla ]]; then
-    git clone --recursive -b r2.9 https://github.com/pytorch/xla.git
+    git clone --recursive --quiet 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)"

.ci/pytorch/cpp_doc_push_script.sh

@@ -58,7 +58,7 @@ time python tools/setup_helpers/generate_code.py \
 
 # Build the docs
 pushd docs/cpp
-time make VERBOSE=1 html
+time make VERBOSE=1 html -j
 
 popd
 popd

.ci/pytorch/macos-build.sh

@@ -35,11 +35,10 @@ fi
 
 print_cmake_info
 if [[ ${BUILD_ENVIRONMENT} == *"distributed"* ]]; then
-  # Needed for inductor benchmarks, as lots of HF networks make `torch.distribtued` calls
-  USE_DISTRIBUTED=1 USE_OPENMP=1 WERROR=1 python setup.py bdist_wheel
+  USE_OPENMP=1 WERROR=1 python setup.py bdist_wheel
 else
-  # Explicitly set USE_DISTRIBUTED=0 to align with the default build config on mac. This also serves as the sole CI config that tests
-  # that building with USE_DISTRIBUTED=0 works at all. See https://github.com/pytorch/pytorch/issues/86448
+  # NB: we always build with distributed; USE_DISTRIBUTED turns off all
+  # backends (specifically the gloo backend), so test that this case works too
   USE_DISTRIBUTED=0 USE_OPENMP=1 MACOSX_DEPLOYMENT_TARGET=11.0 WERROR=1 BUILD_TEST=OFF USE_PYTORCH_METAL=1 python setup.py bdist_wheel --plat-name macosx_11_0_arm64
 fi
 if which sccache > /dev/null; then

.ci/pytorch/macos-test.sh

@@ -13,9 +13,13 @@ if [[ ! $(python -c "import torch; print(int(torch.backends.openmp.is_available(
 fi
 popd
 
+python -mpip install -r requirements.txt
+
 # enable debug asserts in serialization
 export TORCH_SERIALIZATION_DEBUG=1
 
+python -mpip install --no-input -r requirements.txt
+
 setup_test_python() {
   # The CircleCI worker hostname doesn't resolve to an address.
   # This environment variable makes ProcessGroupGloo default to

.ci/pytorch/numba-cuda-13.patch

@@ -1,25 +0,0 @@
-From 6e08c9d08e9de59c7af28b720289debbbd384764 Mon Sep 17 00:00:00 2001
-From: Michael Wang <13521008+isVoid@users.noreply.github.com>
-Date: Tue, 1 Apr 2025 17:28:05 -0700
-Subject: [PATCH] Avoid bumping certain driver API to avoid future breakage
- (#185)
-
-Co-authored-by: isVoid <isVoid@users.noreply.github.com>
----
- numba_cuda/numba/cuda/cudadrv/driver.py | 3 +++
- 1 file changed, 3 insertions(+)
-
-diff --git a/numba_cuda/numba/cuda/cudadrv/driver.py b/numba_cuda/numba/cuda/cudadrv/driver.py
-index 1641bf77..233e9ed7 100644
---- a/numba_cuda/numba/cuda/cudadrv/driver.py
-+++ b/numba_cuda/numba/cuda/cudadrv/driver.py
-@@ -365,6 +365,9 @@ def _find_api(self, fname):
-         else:
-             variants = ('_v2', '')
- 
-+        if fname in ("cuCtxGetDevice", "cuCtxSynchronize"):
-+            return getattr(self.lib, fname)
-+
-         for variant in variants:
-             try:
-                 return getattr(self.lib, f'{fname}{variant}')

.ci/pytorch/smoke_test/check_binary_symbols.py

@@ -32,9 +32,6 @@ LIBTORCH_NAMESPACE_LIST = (
     "torch::",
 )
 
-# Patterns for detecting statically linked libstdc++ symbols
-STATICALLY_LINKED_CXX11_ABI = [re.compile(r".*recursive_directory_iterator.*")]
-
 
 def _apply_libtorch_symbols(symbols):
     return [
@@ -56,17 +53,12 @@ def get_symbols(lib: str) -> list[tuple[str, str, str]]:
     return [x.split(" ", 2) for x in lines.decode("latin1").split("\n")[:-1]]
 
 
-def grep_symbols(
-    lib: str, patterns: list[Any], symbol_type: str | None = None
-) -> list[str]:
+def grep_symbols(lib: str, patterns: list[Any]) -> list[str]:
     def _grep_symbols(
         symbols: list[tuple[str, str, str]], patterns: list[Any]
     ) -> list[str]:
         rc = []
         for _s_addr, _s_type, s_name in symbols:
-            # Filter by symbol type if specified
-            if symbol_type and _s_type != symbol_type:
-                continue
             for pattern in patterns:
                 if pattern.match(s_name):
                     rc.append(s_name)
@@ -88,18 +80,6 @@ def grep_symbols(
         return functools.reduce(list.__add__, (x.result() for x in tasks), [])
 
 
-def check_lib_statically_linked_libstdc_cxx_abi_symbols(lib: str) -> None:
-    cxx11_statically_linked_symbols = grep_symbols(
-        lib, STATICALLY_LINKED_CXX11_ABI, symbol_type="T"
-    )
-    num_statically_linked_symbols = len(cxx11_statically_linked_symbols)
-    print(f"num_statically_linked_symbols (T): {num_statically_linked_symbols}")
-    if num_statically_linked_symbols > 0:
-        raise RuntimeError(
-            f"Found statically linked libstdc++ symbols (recursive_directory_iterator): {cxx11_statically_linked_symbols[:100]}"
-        )
-
-
 def check_lib_symbols_for_abi_correctness(lib: str) -> None:
     print(f"lib: {lib}")
     cxx11_symbols = grep_symbols(lib, LIBTORCH_CXX11_PATTERNS)
@@ -127,7 +107,6 @@ def main() -> None:
 
     libtorch_cpu_path = str(install_root / "lib" / "libtorch_cpu.so")
     check_lib_symbols_for_abi_correctness(libtorch_cpu_path)
-    check_lib_statically_linked_libstdc_cxx_abi_symbols(libtorch_cpu_path)
 
 
 if __name__ == "__main__":

.ci/pytorch/test.sh

@@ -32,16 +32,6 @@ if [[ "$BUILD_ENVIRONMENT" != *rocm* && "$BUILD_ENVIRONMENT" != *s390x* && -d /v
   git config --global --add safe.directory /var/lib/jenkins/workspace
 fi
 
-
-# Patch numba to avoid CUDA-13 crash, see https://github.com/pytorch/pytorch/issues/162878
-NUMBA_CUDA_DIR=$(python -c "import os;import numba.cuda; print(os.path.dirname(numba.cuda.__file__))" 2>/dev/null || true)
-if [ -n "$NUMBA_CUDA_DIR" ]; then
-  NUMBA_PATCH="$(dirname "$(realpath "${BASH_SOURCE[0]}")")/numba-cuda-13.patch"
-  pushd "$NUMBA_CUDA_DIR"
-  patch -p4 <"$NUMBA_PATCH"
-  popd
-fi
-
 echo "Environment variables:"
 env
 
@@ -1624,25 +1614,6 @@ test_operator_benchmark() {
       --expected "expected_ci_operator_benchmark_eager_float32_cpu.csv"
 }
 
-test_operator_microbenchmark() {
-  TEST_REPORTS_DIR=$(pwd)/test/test-reports
-  mkdir -p "$TEST_REPORTS_DIR"
-  TEST_DIR=$(pwd)
-
-  cd benchmarks/operator_benchmark/pt_extension
-  python -m pip install .
-
-  cd "${TEST_DIR}"/benchmarks/operator_benchmark
-
-  for OP_BENCHMARK_TESTS in matmul mm addmm bmm; do
-    $TASKSET python -m pt.${OP_BENCHMARK_TESTS}_test --tag-filter long \
-      --output-json-for-dashboard "${TEST_REPORTS_DIR}/operator_microbenchmark_${OP_BENCHMARK_TESTS}_compile.json" \
-      --benchmark-name "PyTorch operator microbenchmark" --use-compile
-    $TASKSET python -m pt.${OP_BENCHMARK_TESTS}_test --tag-filter long \
-      --output-json-for-dashboard "${TEST_REPORTS_DIR}/operator_microbenchmark_${OP_BENCHMARK_TESTS}.json" \
-      --benchmark-name "PyTorch operator microbenchmark"
-  done
-}
 
 if ! [[ "${BUILD_ENVIRONMENT}" == *libtorch* || "${BUILD_ENVIRONMENT}" == *-bazel-* ]]; then
   (cd test && python -c "import torch; print(torch.__config__.show())")
@@ -1697,8 +1668,6 @@ elif [[ "${TEST_CONFIG}" == *operator_benchmark* ]]; then
     test_operator_benchmark cpu ${TEST_MODE}
 
   fi
-elif [[ "${TEST_CONFIG}" == *operator_microbenchmark* ]]; then
-  test_operator_microbenchmark
 elif [[ "${TEST_CONFIG}" == *inductor_distributed* ]]; then
   test_inductor_distributed
 elif [[ "${TEST_CONFIG}" == *inductor-halide* ]]; then
@@ -1752,6 +1721,11 @@ elif [[ "${TEST_CONFIG}" == *inductor_cpp_wrapper* ]]; then
 elif [[ "${TEST_CONFIG}" == *inductor* ]]; then
   install_torchvision
   test_inductor_shard "${SHARD_NUMBER}"
+  if [[ "${SHARD_NUMBER}" == 1 ]]; then
+    if [[ "${BUILD_ENVIRONMENT}" != linux-jammy-py3.9-gcc11-build ]]; then
+      test_inductor_distributed
+    fi
+  fi
 elif [[ "${TEST_CONFIG}" == *einops* ]]; then
   test_einops
 elif [[ "${TEST_CONFIG}" == *dynamo_wrapped* ]]; then

.ci/pytorch/win-test-helpers/build_pytorch.bat

@@ -137,7 +137,7 @@ sccache --show-stats
 python -c "import os, glob; os.system('python -mpip install --no-index --no-deps ' + glob.glob('dist/*.whl')[0])"
 (
   if "%BUILD_ENVIRONMENT%"=="" (
-    echo NOTE: To run `import torch`, please make sure to activate the conda environment by running `call %CONDA_ROOT_DIR%\Scripts\activate.bat %CONDA_ROOT_DIR%\envs\py_tmp` in Command Prompt before running Git Bash.
+    echo NOTE: To run `import torch`, please make sure to activate the conda environment by running `call %CONDA_PARENT_DIR%\Miniconda3\Scripts\activate.bat %CONDA_PARENT_DIR%\Miniconda3` in Command Prompt before running Git Bash.
   ) else (
     copy /Y "dist\*.whl" "%PYTORCH_FINAL_PACKAGE_DIR%"
 

.ci/pytorch/win-test-helpers/installation-helpers/activate_miniconda3.bat

@@ -3,12 +3,12 @@ if "%BUILD_ENVIRONMENT%"=="" (
 ) else (
   set CONDA_PARENT_DIR=C:\Jenkins
 )
-set CONDA_ROOT_DIR=%CONDA_PARENT_DIR%\Miniconda3
+
 
 :: Be conservative here when rolling out the new AMI with conda. This will try
 :: to install conda as before if it couldn't find the conda installation. This
 :: can be removed eventually after we gain enough confidence in the AMI
-if not exist %CONDA_ROOT_DIR% (
+if not exist %CONDA_PARENT_DIR%\Miniconda3 (
   set INSTALL_FRESH_CONDA=1
 )
 
@@ -17,14 +17,10 @@ if "%INSTALL_FRESH_CONDA%"=="1" (
   if errorlevel 1 exit /b
   if not errorlevel 0 exit /b
 
-  %TMP_DIR_WIN%\Miniconda3-latest-Windows-x86_64.exe /InstallationType=JustMe /RegisterPython=0 /S /AddToPath=0 /D=%CONDA_ROOT_DIR%
+  %TMP_DIR_WIN%\Miniconda3-latest-Windows-x86_64.exe /InstallationType=JustMe /RegisterPython=0 /S /AddToPath=0 /D=%CONDA_PARENT_DIR%\Miniconda3
   if errorlevel 1 exit /b
   if not errorlevel 0 exit /b
 )
 
 :: Activate conda so that we can use its commands, i.e. conda, python, pip
-call %CONDA_ROOT_DIR%\Scripts\activate.bat %CONDA_ROOT_DIR%
-:: Activate conda so that we can use its commands, i.e. conda, python, pip
-call conda activate py_tmp
-
-call pip install -r .ci/docker/requirements-ci.txt
+call %CONDA_PARENT_DIR%\Miniconda3\Scripts\activate.bat %CONDA_PARENT_DIR%\Miniconda3

.ci/pytorch/win-test-helpers/setup_pytorch_env.bat

@@ -14,7 +14,7 @@ if not errorlevel 0 exit /b
 :: build\torch. Rather than changing all these references, making a copy of torch folder
 :: from conda to the current workspace is easier. The workspace will be cleaned up after
 :: the job anyway
-xcopy /s %CONDA_ROOT_DIR%\envs\py_tmp\Lib\site-packages\torch %TMP_DIR_WIN%\build\torch\
+xcopy /s %CONDA_PARENT_DIR%\Miniconda3\Lib\site-packages\torch %TMP_DIR_WIN%\build\torch\
 
 pushd .
 if "%VC_VERSION%" == "" (

.ci/pytorch/win-test.sh

@@ -38,14 +38,7 @@ if [[ "$BUILD_ENVIRONMENT" == *cuda* ]]; then
 fi
 
 # TODO: Move both of them to Windows AMI
-python -m pip install tensorboard==2.13.0 protobuf==5.29.4 pytest-subtests==0.13.1
-
-# Copied from https://github.com/pytorch/test-infra/blob/be01a40157c36cd5a48391fdf44a7bc3ebd4c7e3/aws/ami/windows/scripts/Installers/Install-Pip-Dependencies.ps1#L16 with some adjustments
-# pytest-rerunfailures==10.3 as 10.2 fails with INTERNALERROR> pluggy._manager.PluginValidationError: unknown hook 'pytest_configure_node'
-# scipy from 1.6.3 to 1.10
-# expecttest from 0.1.3 to 0.3.0
-# xdoctest from 1.0.2 to 1.3.0
-python -m pip install "future==0.18.2" "hypothesis==5.35.1" "expecttest==0.3.0" "librosa>=0.6.2" "scipy==1.10.1" "psutil==5.9.1" "pynvml==11.4.1" "pillow==9.2.0" "unittest-xml-reporting<=3.2.0,>=2.0.0" "pytest==7.1.3" "pytest-xdist==2.5.0" "pytest-flakefinder==1.1.0" "pytest-rerunfailures==10.3" "pytest-shard==0.1.2" "sympy==1.11.1" "xdoctest==1.3.0" "pygments==2.12.0" "opt-einsum>=3.3" "networkx==2.8.8" "mpmath==1.2.1" "pytest-cpp==2.3.0" "boto3==1.35.42"
+python -m pip install pytest-rerunfailures==10.3 pytest-cpp==2.3.0 tensorboard==2.13.0 protobuf==5.29.4 pytest-subtests==0.13.1
 
 # Install Z3 optional dependency for Windows builds.
 python -m pip install z3-solver==4.15.1.0
@@ -59,6 +52,9 @@ python -m pip install parameterized==0.8.1
 # Install pulp for testing ilps under torch\distributed\_tools
 python -m pip install pulp==2.9.0
 
+# Install expecttest to merge https://github.com/pytorch/pytorch/pull/155308
+python -m pip install expecttest==0.3.0
+
 run_tests() {
     # Run nvidia-smi if available
     for path in '/c/Program Files/NVIDIA Corporation/NVSMI/nvidia-smi.exe' /c/Windows/System32/nvidia-smi.exe; do

.ci/pytorch/windows/cuda128.bat

@@ -37,10 +37,10 @@ IF "%CUDA_PATH_V128%"=="" (
 )
 
 IF "%BUILD_VISION%" == "" (
-    set TORCH_CUDA_ARCH_LIST=7.0;7.5;8.0;8.6;9.0;10.0;12.0
+    set TORCH_CUDA_ARCH_LIST=6.1;7.0;7.5;8.0;8.6;9.0;10.0;12.0
     set TORCH_NVCC_FLAGS=-Xfatbin -compress-all
 ) ELSE (
-    set NVCC_FLAGS=-D__CUDA_NO_HALF_OPERATORS__ --expt-relaxed-constexpr -gencode=arch=compute_70,code=sm_70 -gencode=arch=compute_75,code=sm_75 -gencode=arch=compute_80,code=compute_80 -gencode=arch=compute_86,code=compute_86 -gencode=arch=compute_90,code=compute_90 -gencode=arch=compute_100,code=compute_100 -gencode=arch=compute_120,code=compute_120
+    set NVCC_FLAGS=-D__CUDA_NO_HALF_OPERATORS__ --expt-relaxed-constexpr -gencode=arch=compute_50,code=sm_50 -gencode=arch=compute_60,code=sm_60 -gencode=arch=compute_70,code=sm_70 -gencode=arch=compute_75,code=sm_75 -gencode=arch=compute_80,code=compute_80 -gencode=arch=compute_86,code=compute_86 -gencode=arch=compute_90,code=compute_90 -gencode=arch=compute_100,code=compute_100 -gencode=arch=compute_120,code=compute_120
 )
 
 set "CUDA_PATH=%CUDA_PATH_V128%"

.ci/pytorch/windows/internal/driver_update.bat

@@ -1,9 +1,9 @@
-set WIN_DRIVER_VN=580.88
-set "DRIVER_DOWNLOAD_LINK=https://ossci-windows.s3.amazonaws.com/%WIN_DRIVER_VN%-data-center-tesla-desktop-win10-win11-64bit-dch-international.exe" & REM @lint-ignore
-curl --retry 3 -kL %DRIVER_DOWNLOAD_LINK% --output %WIN_DRIVER_VN%-data-center-tesla-desktop-win10-win11-64bit-dch-international.exe
+set WIN_DRIVER_VN=528.89
+set "DRIVER_DOWNLOAD_LINK=https://ossci-windows.s3.amazonaws.com/%WIN_DRIVER_VN%-data-center-tesla-desktop-winserver-2016-2019-2022-dch-international.exe" & REM @lint-ignore
+curl --retry 3 -kL %DRIVER_DOWNLOAD_LINK% --output %WIN_DRIVER_VN%-data-center-tesla-desktop-winserver-2016-2019-2022-dch-international.exe
 if errorlevel 1 exit /b 1
 
-start /wait %WIN_DRIVER_VN%-data-center-tesla-desktop-win10-win11-64bit-dch-international.exe -s -noreboot
+start /wait %WIN_DRIVER_VN%-data-center-tesla-desktop-winserver-2016-2019-2022-dch-international.exe -s -noreboot
 if errorlevel 1 exit /b 1
 
-del %WIN_DRIVER_VN%-data-center-tesla-desktop-win10-win11-64bit-dch-international.exe || ver > NUL
+del %WIN_DRIVER_VN%-data-center-tesla-desktop-winserver-2016-2019-2022-dch-international.exe || ver > NUL

.ci/wheel/build_wheel.sh

@@ -189,7 +189,8 @@ pip install requests ninja typing-extensions
 retry pip install -r "${pytorch_rootdir}/requirements.txt" || true
 retry brew install libomp
 
-# For USE_DISTRIBUTED=1 on macOS, need libuv, which is build as part of tensorpipe submodule
+# For USE_DISTRIBUTED=1 on macOS, this enables gloo, which needs libuv, which
+# is build as part of tensorpipe submodule
 export USE_DISTRIBUTED=1
 
 export USE_MKLDNN=OFF

.github/actions/setup-win/action.yml

@@ -6,12 +6,6 @@ inputs:
   cuda-version:
     description: which cuda version to install, 'cpu' for none
     required: true
-  python-version:
-    required: false
-    type: string
-    default: "3.10"
-    description: |
-      The python version to be used. Will be 3.10 by default
 
 runs:
   using: composite
@@ -44,24 +38,18 @@ runs:
         CONDA="C:\Jenkins\Miniconda3\condabin\conda.bat"
 
         {
-          echo "CONDA=${CONDA}";
           echo "CONDA_RUN=${CONDA} run --no-capture-output";
           echo "CONDA_BUILD=${CONDA} run conda-build";
           echo "CONDA_INSTALL=${CONDA} install";
         } >> "${GITHUB_ENV}"
 
     - name: Setup Python3
-      env:
-          PYTHON_VERSION: ${{ inputs.python-version }}
       shell: bash
       run: |
         set +e
         set -x
 
-        # Create new py_tmp env with python-version
-        ${CONDA} create -y -n py_tmp python=${PYTHON_VERSION} intel-openmp libuv
-
-        PYTHON3=$(${CONDA_RUN} -n py_tmp which python3)
+        PYTHON3=$(${CONDA_RUN} which python3)
         EXIT_CODE=$?
 
         if [[ "${EXIT_CODE}" == "0" ]]; then
@@ -74,7 +62,7 @@ runs:
           # installation, which is Python 3 based. Its Python is default to Python 3. Further, there
           # is also the Miniconda installation that is Python 2 based, and both can be installed if
           # needed. In both cases, Python binary is just called python
-          PYTHON=$(${CONDA_RUN} -n py_tmp which python)
+          PYTHON=$(${CONDA_RUN} which python)
           EXIT_CODE=$?
 
           if [[ "${EXIT_CODE}" == "0" ]]; then

.github/ci_commit_pins/vllm.txt

@@ -1 +1 @@
-78a47f87ce259a48f0391fa9ae15add05ea7432b
+e10fef08838612b4560e9c72e5cb1414a5edfa13

.github/ci_commit_pins/xla.txt

@@ -1 +1 @@
-r2.9
+6c5478ff7c3d50dd1e3047d72ec5909bea474073

.github/scripts/filter_test_configs.py

@@ -41,9 +41,9 @@ SUPPORTED_PERIODICAL_MODES: dict[str, Callable[[Optional[str]], bool]] = {
 }
 
 # The link to the published list of disabled jobs
-DISABLED_JOBS_URL = "https://ossci-metrics.s3.amazonaws.com/disabled-jobs.json?versionId=hjktHz2WOejHpxKpkqpDknTt5rMTM9KK"
+DISABLED_JOBS_URL = "https://ossci-metrics.s3.amazonaws.com/disabled-jobs.json"
 # and unstable jobs
-UNSTABLE_JOBS_URL = "https://ossci-metrics.s3.amazonaws.com/unstable-jobs.json?versionId=wrjdvvQTJxgvMO.rGw5MEuMsj6XbjuV7"
+UNSTABLE_JOBS_URL = "https://ossci-metrics.s3.amazonaws.com/unstable-jobs.json"
 
 # Some constants used to handle disabled and unstable jobs
 JOB_NAME_SEP = "/"

.github/scripts/generate_binary_build_matrix.py

@@ -43,55 +43,55 @@ CUDA_AARCH64_ARCHES = ["12.6-aarch64", "12.8-aarch64", "13.0-aarch64"]
 
 PYTORCH_EXTRA_INSTALL_REQUIREMENTS = {
     "12.6": (
-        "nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | "
-        "nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | "
-        "nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | "
-        "nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | "
-        "nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | "
-        "nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | "
-        "nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | "
-        "nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | "
-        "nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | "
-        "nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | "
-        "nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | "
-        "nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | "
-        "nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | "
-        "nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | "
-        "nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'"
+        "nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'"
     ),
     "12.8": (
-        "nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | "
-        "nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | "
-        "nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | "
-        "nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | "
-        "nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | "
-        "nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | "
-        "nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | "
-        "nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | "
-        "nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | "
-        "nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | "
-        "nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | "
-        "nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | "
-        "nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | "
-        "nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | "
-        "nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'"
+        "nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'"
     ),
     "13.0": (
-        "nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | "
-        "nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | "
-        "nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | "
-        "nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | "
-        "nvidia-cublas==13.0.0.19; platform_system == 'Linux' | "
-        "nvidia-cufft==12.0.0.15; platform_system == 'Linux' | "
-        "nvidia-curand==10.4.0.35; platform_system == 'Linux' | "
-        "nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | "
-        "nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | "
-        "nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | "
-        "nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | "
-        "nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | "
-        "nvidia-nvtx==13.0.39; platform_system == 'Linux' | "
-        "nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | "
-        "nvidia-cufile==1.15.0.42; platform_system == 'Linux'"
+        "nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'"
     ),
     "xpu": (
         "intel-cmplr-lib-rt==2025.2.1 | "

.github/scripts/generate_ci_workflows.py

@@ -135,7 +135,7 @@ ROCM_SMOKE_WORKFLOWS = [
         build_configs=generate_binary_build_matrix.generate_wheels_matrix(
             OperatingSystem.LINUX,
             arches=["6.4"],
-            python_versions=["3.10"],
+            python_versions=["3.9"],
         ),
         ciflow_config=CIFlowConfig(
             labels={

.github/templates/common.yml.j2

@@ -32,7 +32,7 @@ concurrency:
 {%- macro setup_ec2_windows() -%}
       !{{ display_ec2_information() }}
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}

.github/templates/linux_binary_build_workflow.yml.j2

@@ -56,7 +56,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -135,7 +135,7 @@ jobs:
       contents: read
     steps:
       - name: Setup XPU
-        uses: pytorch/pytorch/.github/actions/setup-xpu@release/2.9
+        uses: pytorch/pytorch/.github/actions/setup-xpu@main
       - name: configure aws credentials
         id: aws_creds
         uses: aws-actions/configure-aws-credentials@v4
@@ -150,10 +150,10 @@ jobs:
         with:
           name: !{{ config["build_name"] }}
           path: "${{ runner.temp }}/artifacts/"
-      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: !{{ config["container_image"] }}
@@ -161,7 +161,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -182,7 +182,7 @@ jobs:
         with:
           name: !{{ config["build_name"] }}
           path: "${{ runner.temp }}/artifacts/"
-      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
       - 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}"
@@ -196,7 +196,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: !{{ config["container_image"] }}
@@ -204,7 +204,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary

.github/templates/macos_binary_build_workflow.yml.j2

@@ -68,7 +68,7 @@ jobs:
           chmod +x "${RUNNER_TEMP}/conda.sh"
           /bin/bash "${RUNNER_TEMP}/conda.sh" -b -p "${RUNNER_TEMP}/anaconda"
           echo "${RUNNER_TEMP}/anaconda/bin" >> "${GITHUB_PATH}"
-      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
       - name: Populate binary env
         run: |
           # shellcheck disable=SC1091

.github/templates/windows_binary_build_workflow.yml.j2

@@ -64,7 +64,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -135,7 +135,7 @@ jobs:
 {%- else %}
       !{{ set_runner_specific_vars() }}
       !{{ common.setup_ec2_windows() }}
-      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
 {%- endif %}
       - name: Populate binary env
         shell: bash
@@ -211,7 +211,7 @@ jobs:
           "pytorch/.ci/pytorch/windows/arm64/bootstrap_rust.bat"
 {%- else %}
       !{{ common.setup_ec2_windows() }}
-      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
       !{{ set_runner_specific_vars() }}
 {%- endif %}
       - uses: !{{ common.download_artifact_action }}

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

@@ -47,7 +47,7 @@ jobs:
       reenabled-issues: ${{ steps.filter.outputs.reenabled-issues }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           fetch-depth: 1
           submodules: false
@@ -69,25 +69,25 @@ jobs:
     runs-on: ${{ matrix.runner }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
 
       - 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@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-image-name: ${{ inputs.docker-image-name }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -97,7 +97,7 @@ jobs:
         run: echo "IN_CONTAINER_RUNNER=$(if [ -f /.inarc ] || [ -f /.incontainer ]; then echo true ; else echo false; fi)" >> "$GITHUB_OUTPUT"
 
       - name: Install nvidia driver, nvidia-docker runtime, set GPU_FLAG
-        uses: pytorch/test-infra/.github/actions/setup-nvidia@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-nvidia@main
         if: ${{ inputs.cuda-version != 'cpu' && steps.check_container_runner.outputs.IN_CONTAINER_RUNNER == 'false' }}
 
       - name: Output disk space left
@@ -209,5 +209,5 @@ jobs:
           file-suffix: bazel-${{ github.job }}_${{ steps.get-job-id.outputs.job-id }}
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.9
+        uses: pytorch/test-infra/.github/actions/teardown-linux@main
         if: always()

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

@@ -142,13 +142,13 @@ jobs:
 
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
         if: inputs.build_environment != 'linux-s390x-binary-manywheel'
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.github-token }}
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           no-sudo: ${{ inputs.build_environment == 'linux-aarch64-binary-manywheel' || inputs.build_environment == 'linux-s390x-binary-manywheel' }}
 
@@ -178,6 +178,7 @@ jobs:
       - name: Checkout PyTorch to pytorch dir
         uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -212,9 +213,9 @@ jobs:
       - name: Calculate docker image
         id: calculate-docker-image
         if: ${{ steps.filter.outputs.is-test-matrix-empty == 'False' && inputs.build_environment != 'linux-s390x-binary-manywheel' }}
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
-          # If doing this in release/2.9 or release branch, use docker.io. Otherwise
+          # If doing this in main or release branch, use docker.io. Otherwise
           # use ECR
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: ${{ inputs.DOCKER_IMAGE }}
@@ -226,7 +227,7 @@ jobs:
 
       - name: Pull Docker image
         if: ${{ steps.filter.outputs.is-test-matrix-empty == 'False' && inputs.build_environment != 'linux-s390x-binary-manywheel' }}
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -282,7 +283,7 @@ jobs:
 
       - name: Teardown Linux
         if: always() && inputs.build_environment != 'linux-s390x-binary-manywheel'
-        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.9
+        uses: pytorch/test-infra/.github/actions/teardown-linux@main
 
       - name: Chown workspace
         if: always() && inputs.build_environment != 'linux-s390x-binary-manywheel'

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

@@ -125,14 +125,14 @@ jobs:
 
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
         if: inputs.build_environment != 'linux-s390x-binary-manywheel'
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.github-token }}
 
         # Setup the environment
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           no-sudo: ${{ inputs.build_environment == 'linux-aarch64-binary-manywheel' || inputs.build_environment == 'linux-s390x-binary-manywheel' }}
 
@@ -155,6 +155,7 @@ jobs:
       - name: Checkout PyTorch to pytorch dir
         uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           show-progress: false
           path: pytorch
@@ -185,7 +186,9 @@ jobs:
           path: "${{ runner.temp }}/artifacts/"
 
       - name: Install nvidia driver, nvidia-docker runtime, set GPU_FLAG
-        uses: pytorch/test-infra/.github/actions/setup-nvidia@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-nvidia@main
+        with:
+          driver-version: ${{ startsWith(inputs.GPU_ARCH_VERSION, '13') && '580.65.06' || '570.133.07' }}
         if: ${{ inputs.GPU_ARCH_TYPE == 'cuda' && steps.filter.outputs.is-test-matrix-empty == 'False' }}
 
       - name: configure aws credentials
@@ -200,7 +203,7 @@ jobs:
       - name: Calculate docker image
         id: calculate-docker-image
         if: ${{ steps.filter.outputs.is-test-matrix-empty == 'False' && inputs.build_environment != 'linux-s390x-binary-manywheel' }}
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: ${{ inputs.DOCKER_IMAGE }}
@@ -210,7 +213,7 @@ jobs:
 
       - name: Pull Docker image
         if: ${{ steps.filter.outputs.is-test-matrix-empty == 'False' && inputs.build_environment != 'linux-s390x-binary-manywheel' }}
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -222,7 +225,7 @@ jobs:
 
       - name: Teardown Linux
         if: always() && inputs.build_environment != 'linux-s390x-binary-manywheel'
-        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.9
+        uses: pytorch/test-infra/.github/actions/teardown-linux@main
 
       - name: Chown workspace
         if: always() && inputs.build_environment != 'linux-s390x-binary-manywheel'

.github/workflows/_binary-upload.yml

@@ -81,7 +81,7 @@ jobs:
       SHA1: ${{ github.event.pull_request.head.sha || github.sha }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           no-sudo: true
 

.github/workflows/_docs.yml

@@ -67,7 +67,7 @@ jobs:
             # an OOM issue when running the job, so this upgrades the runner from 4xlarge
             # to the next available tier of 12xlarge. So much memory just to generate cpp
             # doc
-            runner: ${{ inputs.runner_prefix }}linux.12xlarge.memory
+            runner: ${{ inputs.runner_prefix }}linux.12xlarge
             # TODO: Nightly cpp docs take longer and longer to finish (more than 3h now)
             # Let's try to figure out how this can be improved
             timeout-minutes: 360
@@ -84,7 +84,7 @@ jobs:
     name: build-docs-${{ matrix.docs_type }}-${{ inputs.push }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
           instructions: |
@@ -95,7 +95,7 @@ jobs:
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
 
       - name: Setup Linux
         uses: ./.github/actions/setup-linux
@@ -110,12 +110,12 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-image-name: ${{ inputs.docker-image }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -222,5 +222,5 @@ jobs:
           s3-prefix: pytorch/pytorch/${{ github.event.pull_request.number }}/functorchdocs
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.9
+        uses: pytorch/test-infra/.github/actions/teardown-linux@main
         if: always()

.github/workflows/_link_check.yml

@@ -11,7 +11,7 @@ on:
 jobs:
   lint-urls:
     if: ${{ github.event_name != 'pull_request' || !contains(github.event.pull_request.labels.*.name, 'skip-url-lint') }}
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@release/2.9
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
     with:
       job-name: lint-urls
       timeout: 120
@@ -37,7 +37,7 @@ jobs:
 
   lint-xrefs:
     if: ${{ github.event_name != 'pull_request' || !contains(github.event.pull_request.labels.*.name, 'skip-xref-lint') }}
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@release/2.9
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
     with:
       job-name: lint-xrefs
       timeout: 60

.github/workflows/_linux-build.yml

@@ -134,7 +134,7 @@ 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@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         if: inputs.build-environment != 'linux-s390x-binary-manywheel'
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -147,7 +147,7 @@ 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@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           no-sudo: true
 
@@ -183,7 +183,7 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         if: inputs.build-environment != 'linux-s390x-binary-manywheel'
         with:
           docker-image-name: ${{ inputs.docker-image-name }}
@@ -199,7 +199,7 @@ jobs:
           echo "docker pull ghcr.io/pytorch/ci-image:${tag/:/-}"
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         if: inputs.build-environment != 'linux-s390x-binary-manywheel' && steps.use-old-whl.outputs.reuse != 'true'
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
@@ -457,7 +457,7 @@ jobs:
           artifact_prefix: usage_log_build_${{ steps.get-job-id.outputs.job-id }}
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.9
+        uses: pytorch/test-infra/.github/actions/teardown-linux@main
         if: always() && inputs.build-environment != 'linux-s390x-binary-manywheel'
 
       - name: Cleanup docker

.github/workflows/_linux-test.yml

@@ -99,7 +99,7 @@ jobs:
       contents: read
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         if: ${{ !contains(matrix.runner, 'b200') && inputs.build-environment != 'linux-s390x-binary-manywheel' }}
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -108,7 +108,7 @@ jobs:
               docker exec -it $(docker container ps --format '{{.ID}}') bash
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           no-sudo: true
 
@@ -139,7 +139,7 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         if: inputs.build-environment != 'linux-s390x-binary-manywheel'
         with:
           docker-image-name: ${{ inputs.docker-image }}
@@ -155,7 +155,7 @@ jobs:
           echo "docker pull ghcr.io/pytorch/ci-image:${tag/:/-}"
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         if: inputs.build-environment != 'linux-s390x-binary-manywheel'
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
@@ -167,9 +167,9 @@ jobs:
 
       - name: Install nvidia driver, nvidia-docker runtime, set GPU_FLAG
         id: install-nvidia-driver
-        uses: pytorch/test-infra/.github/actions/setup-nvidia@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-nvidia@main
         with:
-          driver-version: ${{ matrix.config == 'legacy_nvidia_driver' && '525.105.17' || '580.82.07' }}
+          driver-version: ${{ matrix.config == 'legacy_nvidia_driver' && '525.105.17' || '570.133.07' }}
         if: ${{ contains(inputs.build-environment, 'cuda') && !contains(matrix.config, 'nogpu') && steps.check_container_runner.outputs.IN_CONTAINER_RUNNER == 'false' && !contains(matrix.runner, 'b200') }}
 
       - name: Setup GPU_FLAG for docker run
@@ -273,8 +273,6 @@ jobs:
           TEST_CONFIG: ${{ matrix.config }}
           SHARD_NUMBER: ${{ matrix.shard }}
           NUM_TEST_SHARDS: ${{ matrix.num_shards }}
-          EXTRA_FLAGS: ${{ matrix.extra_flags || '' }}
-          OP_BENCHMARK_TESTS: ${{ matrix.op_benchmark_tests }}
           REENABLED_ISSUES: ${{ steps.keep-going.outputs.reenabled-issues }}
           CONTINUE_THROUGH_ERROR: ${{ steps.keep-going.outputs.keep-going }}
           VERBOSE_TEST_LOGS: ${{ steps.keep-going.outputs.ci-verbose-test-logs }}
@@ -420,7 +418,7 @@ jobs:
           aws-region: us-east-1
 
       - name: Upload the benchmark results
-        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@release/2.9
+        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@main
         if: inputs.build-environment != 'linux-s390x-binary-manywheel'
         with:
           benchmark-results-dir: test/test-reports
@@ -478,7 +476,7 @@ jobs:
           workflow_attempt: ${{github.run_attempt}}
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.9
+        uses: pytorch/test-infra/.github/actions/teardown-linux@main
         if: always() && steps.check_container_runner.outputs.IN_CONTAINER_RUNNER == 'false'
 
       # NB: We are currently having an intermittent GPU-related issue on G5 runners with

.github/workflows/_mac-build.yml

@@ -67,11 +67,11 @@ 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@release/2.9
+        uses: pytorch/test-infra/.github/actions/check-disk-space@main
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
 
       - name: Set xcode version
         env:
@@ -82,7 +82,7 @@ jobs:
           fi
 
       - name: Setup Python
-        uses: pytorch/test-infra/.github/actions/setup-python@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-python@main
         with:
           python-version: ${{ inputs.python-version }}
           pip-requirements-file: .github/requirements/pip-requirements-macOS.txt
@@ -188,4 +188,4 @@ jobs:
       - name: Clean up disk space
         if: always()
         continue-on-error: true
-        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.9
+        uses: pytorch/test-infra/.github/actions/check-disk-space@main

.github/workflows/_mac-test.yml

@@ -105,11 +105,11 @@ jobs:
           done
 
       - name: Clean up disk space before running MacOS workflow
-        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.9
+        uses: pytorch/test-infra/.github/actions/check-disk-space@main
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
 
       - name: Get workflow job id
         id: get-job-id
@@ -119,7 +119,7 @@ jobs:
           github-token: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Setup Python
-        uses: pytorch/test-infra/.github/actions/setup-python@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-python@main
         with:
           python-version: ${{ inputs.python-version }}
           pip-requirements-file: .github/requirements/pip-requirements-macOS.txt
@@ -257,7 +257,7 @@ jobs:
           file-suffix: ${{ github.job }}-${{ matrix.config }}-${{ matrix.shard }}-${{ matrix.num_shards }}-${{ matrix.runner }}_${{ steps.get-job-id.outputs.job-id }}
 
       - name: Upload the benchmark results
-        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@release/2.9
+        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@main
         with:
           benchmark-results-dir: test/test-reports
           dry-run: false
@@ -287,4 +287,4 @@ jobs:
       - name: Clean up disk space
         if: always()
         continue-on-error: true
-        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.9
+        uses: pytorch/test-infra/.github/actions/check-disk-space@main

.github/workflows/_rocm-test.yml

@@ -81,7 +81,7 @@ jobs:
     steps:
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           no-sudo: true
 
@@ -113,12 +113,12 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-image-name: ${{ inputs.docker-image }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -330,7 +330,7 @@ jobs:
           aws-region: us-east-1
 
       - name: Upload the benchmark results
-        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@release/2.9
+        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@main
         with:
           benchmark-results-dir: test/test-reports
           dry-run: false

.github/workflows/_runner-determinator.yml

@@ -59,7 +59,7 @@ jobs:
       PR_NUMBER: ${{ github.event.pull_request.number }}
     steps:
       # - name: Checkout PyTorch
-      #   uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+      #   uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
       #   with:
       #     fetch-depth: 1
       #     submodules: true

.github/workflows/_win-build.yml

@@ -85,10 +85,10 @@ 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@release/2.9
+        uses: pytorch/test-infra/.github/actions/cleanup-runner@main
 
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
           instructions: |
@@ -103,7 +103,7 @@ jobs:
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           no-sudo: true
 
@@ -151,7 +151,7 @@ jobs:
           BUILD_WHEEL: 1
           MAX_JOBS: 8
           CUDA_VERSION: ${{ inputs.cuda-version }}
-          PYTHON_VERSION: "3.10"
+          PYTHON_VERSION: "3.9"
           SCCACHE_BUCKET: "ossci-compiler-cache"
           SCCACHE_S3_KEY_PREFIX: ${{ github.workflow }}
           SCCACHE_REGION: us-east-1

.github/workflows/_win-test.yml

@@ -78,10 +78,10 @@ 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@release/2.9
+        uses: pytorch/test-infra/.github/actions/cleanup-runner@main
 
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
           instructions: |
@@ -97,7 +97,7 @@ jobs:
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           no-sudo: true
 
@@ -184,7 +184,7 @@ jobs:
         env:
           USE_CUDA: ${{ inputs.cuda-version != 'cpu' && '1' || '0' }}
           INSTALL_WINDOWS_SDK: 1
-          PYTHON_VERSION: "3.10"
+          PYTHON_VERSION: 3.9
           CONTINUE_THROUGH_ERROR: ${{ steps.keep-going.outputs.keep-going }}
           VERBOSE_TEST_LOGS: ${{ steps.keep-going.outputs.ci-verbose-test-logs }}
           TEST_SHOWLOCALS: ${{ steps.keep-going.outputs.ci-test-showlocals }}

.github/workflows/_xpu-test.yml

@@ -77,7 +77,7 @@ jobs:
     steps:
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
 
       - name: Setup XPU
         uses: ./.github/actions/setup-xpu
@@ -95,7 +95,7 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-image-name: ${{ inputs.docker-image }}
 
@@ -109,7 +109,7 @@ jobs:
           echo "docker pull ghcr.io/pytorch/ci-image:${tag/:/-}"
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 

.github/workflows/build-almalinux-images.yml

@@ -39,7 +39,7 @@ jobs:
         tag: ["cuda12.6", "cuda12.8", "cuda12.9", "cuda13.0", "rocm6.3", "rocm6.4", "cpu"]
     steps:
       - name: Build docker image
-        uses: pytorch/pytorch/.github/actions/binary-docker-build@release/2.9
+        uses: pytorch/pytorch/.github/actions/binary-docker-build@main
         with:
           docker-image-name: almalinux-builder
           custom-tag-prefix: ${{matrix.tag}}

.github/workflows/build-libtorch-images.yml

@@ -32,7 +32,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -58,7 +58,7 @@ jobs:
         ]
     steps:
       - name: Build docker image
-        uses: pytorch/pytorch/.github/actions/binary-docker-build@release/2.9
+        uses: pytorch/pytorch/.github/actions/binary-docker-build@main
         with:
           docker-image-name: libtorch-cxx11-builder
           custom-tag-prefix: ${{ matrix.tag }}

.github/workflows/build-manywheel-images-s390x.yml

@@ -25,7 +25,7 @@ jobs:
     runs-on: linux.s390x
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           submodules: false
           no-sudo: true

.github/workflows/build-manywheel-images.yml

@@ -32,7 +32,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -63,7 +63,7 @@ jobs:
     name: ${{ matrix.name }}:${{ matrix.tag }}
     steps:
       - name: Build docker image
-        uses: pytorch/pytorch/.github/actions/binary-docker-build@release/2.9
+        uses: pytorch/pytorch/.github/actions/binary-docker-build@main
         with:
           docker-image-name: ${{ matrix.name }}
           custom-tag-prefix: ${{ matrix.tag }}

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

@@ -3,7 +3,7 @@ name: Build Triton wheels
 on:
   push:
     branches:
-      - release/2.9
+      - main
     tags:
       # NOTE: Binary build pipelines should only get triggered on release candidate builds
       # Release candidate tags look like: v1.11.0-rc1
@@ -36,7 +36,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -74,12 +74,12 @@ jobs:
       PLATFORM: 'manylinux_2_28_x86_64'
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           submodules: false
 
@@ -87,7 +87,7 @@ jobs:
         uses: ./.github/actions/setup-linux
 
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ env.DOCKER_IMAGE }}
 
@@ -184,7 +184,7 @@ jobs:
           path: ${{ runner.temp }}/artifacts/wheelhouse/*
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.9
+        uses: pytorch/test-infra/.github/actions/teardown-linux@main
         if: always()
 
   build-wheel-win:
@@ -217,7 +217,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}

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

@@ -42,12 +42,12 @@ jobs:
       BUILD_DEVICE: ${{ matrix.device }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           submodules: false
 
@@ -170,7 +170,7 @@ jobs:
           path: ${{ runner.temp }}/artifacts/externals/vllm/wheels/*.whl
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.9
+        uses: pytorch/test-infra/.github/actions/teardown-linux@main
         if: always()
 
   # Copied from build-triton-wheel workflow (mostly)

.github/workflows/check-labels.yml

@@ -38,7 +38,7 @@ jobs:
     runs-on: linux.24_04.4x
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           submodules: false
           fetch-depth: 1

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

@@ -13,7 +13,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           submodules: false
           fetch-depth: 1

.github/workflows/create_release.yml

@@ -19,7 +19,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

.github/workflows/docker-builds.yml

@@ -33,7 +33,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -70,7 +70,7 @@ jobs:
           pytorch-linux-jammy-py3-clang18-asan,
           pytorch-linux-jammy-py3-clang12-onnx,
           pytorch-linux-jammy-linter,
-          pytorch-linux-jammy-cuda12.8-cudnn9-py3.10-linter,
+          pytorch-linux-jammy-cuda12.8-cudnn9-py3.9-linter,
           # Executorch pin needs update
           # pytorch-linux-jammy-py3-clang12-executorch,
           pytorch-linux-jammy-py3.12-triton-cpu,
@@ -96,21 +96,21 @@ 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@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
 
       - 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@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-image-name: ci-image:${{ matrix.docker-image-name }}
           always-rebuild: true
           push: true
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.build-docker-image.outputs.docker-image }}
 
@@ -141,5 +141,5 @@ jobs:
         if: always()
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.9
+        uses: pytorch/test-infra/.github/actions/teardown-linux@main
         if: always()

.github/workflows/docker-cache-mi300.yml

@@ -20,7 +20,7 @@ jobs:
     runs-on: rocm-docker
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           no-sudo: true
 
@@ -39,13 +39,13 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-image-name: ci-image:pytorch-linux-jammy-rocm-n-py3
           push: false
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 

.github/workflows/docker-release.yml

@@ -37,7 +37,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -52,7 +52,7 @@ jobs:
       matrix: ${{ steps.generate-matrix.outputs.matrix }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           fetch-depth: 1
           submodules: true
@@ -82,7 +82,7 @@ jobs:
       CUDNN_VERSION: ${{ matrix.cudnn_version }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
       # [see note: pytorch repo ref]
@@ -164,12 +164,12 @@ jobs:
           fi
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.9
+        uses: pytorch/test-infra/.github/actions/teardown-linux@main
         if: always()
 
   validate:
     needs: build
-    uses: pytorch/test-infra/.github/workflows/validate-docker-images.yml@release/2.9
+    uses: pytorch/test-infra/.github/workflows/validate-docker-images.yml@main
     with:
-      channel: test
+      channel: nightly
       ref: main

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

@@ -41,7 +41,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -132,7 +132,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_10-cuda-aarch64-12_6
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -178,7 +178,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_10-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -224,7 +224,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_10-cuda-aarch64-13_0
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -335,7 +335,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_11-cuda-aarch64-12_6
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -381,7 +381,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_11-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -427,7 +427,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_11-cuda-aarch64-13_0
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -538,7 +538,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_12-cuda-aarch64-12_6
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -584,7 +584,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_12-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -630,7 +630,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_12-cuda-aarch64-13_0
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -741,7 +741,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_13-cuda-aarch64-12_6
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -787,7 +787,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_13-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -833,7 +833,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_13-cuda-aarch64-13_0
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -944,7 +944,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_13t-cuda-aarch64-12_6
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -990,7 +990,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_13t-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -1036,7 +1036,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_13t-cuda-aarch64-13_0
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -1147,7 +1147,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_14-cuda-aarch64-12_6
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -1193,7 +1193,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_14-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -1239,7 +1239,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_14-cuda-aarch64-13_0
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -1350,7 +1350,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_14t-cuda-aarch64-12_6
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -1396,7 +1396,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_14t-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -1442,7 +1442,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_14t-cuda-aarch64-13_0
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}

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

@@ -41,7 +41,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -368,6 +368,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -389,7 +390,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: libtorch-cxx11-builder
@@ -397,7 +398,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -481,6 +482,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -502,7 +504,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: libtorch-cxx11-builder
@@ -510,7 +512,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary

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

@@ -36,7 +36,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

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

@@ -36,7 +36,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -60,7 +60,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_12-cuda12_8
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_12-cuda12_8-test:  # Testing

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

@@ -41,7 +41,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -127,7 +127,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_10-cuda12_6
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_10-cuda12_6-test:  # Testing
@@ -193,7 +193,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_10-cuda12_8
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_10-cuda12_8-test:  # Testing
@@ -259,7 +259,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_10-cuda13_0
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_10-cuda13_0-test:  # Testing
@@ -357,6 +357,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -378,7 +379,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -386,7 +387,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -467,6 +468,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -488,7 +490,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -496,7 +498,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -570,7 +572,7 @@ jobs:
       contents: read
     steps:
       - name: Setup XPU
-        uses: pytorch/pytorch/.github/actions/setup-xpu@release/2.9
+        uses: pytorch/pytorch/.github/actions/setup-xpu@main
       - name: configure aws credentials
         id: aws_creds
         uses: aws-actions/configure-aws-credentials@v4
@@ -588,6 +590,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -598,7 +601,7 @@ jobs:
         working-directory: pytorch
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -606,7 +609,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -716,7 +719,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_11-cuda12_6
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_11-cuda12_6-test:  # Testing
@@ -782,7 +785,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_11-cuda12_8
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_11-cuda12_8-test:  # Testing
@@ -848,7 +851,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_11-cuda13_0
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_11-cuda13_0-test:  # Testing
@@ -946,6 +949,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -967,7 +971,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -975,7 +979,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -1056,6 +1060,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -1077,7 +1082,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -1085,7 +1090,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -1159,7 +1164,7 @@ jobs:
       contents: read
     steps:
       - name: Setup XPU
-        uses: pytorch/pytorch/.github/actions/setup-xpu@release/2.9
+        uses: pytorch/pytorch/.github/actions/setup-xpu@main
       - name: configure aws credentials
         id: aws_creds
         uses: aws-actions/configure-aws-credentials@v4
@@ -1177,6 +1182,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -1187,7 +1193,7 @@ jobs:
         working-directory: pytorch
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -1195,7 +1201,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -1305,7 +1311,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_12-cuda12_6
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_12-cuda12_6-test:  # Testing
@@ -1371,7 +1377,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_12-cuda12_8
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_12-cuda12_8-test:  # Testing
@@ -1437,7 +1443,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_12-cuda13_0
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_12-cuda13_0-test:  # Testing
@@ -1535,6 +1541,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -1556,7 +1563,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -1564,7 +1571,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -1645,6 +1652,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -1666,7 +1674,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -1674,7 +1682,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -1748,7 +1756,7 @@ jobs:
       contents: read
     steps:
       - name: Setup XPU
-        uses: pytorch/pytorch/.github/actions/setup-xpu@release/2.9
+        uses: pytorch/pytorch/.github/actions/setup-xpu@main
       - name: configure aws credentials
         id: aws_creds
         uses: aws-actions/configure-aws-credentials@v4
@@ -1766,6 +1774,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -1776,7 +1785,7 @@ jobs:
         working-directory: pytorch
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -1784,7 +1793,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -1894,7 +1903,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_13-cuda12_6
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_13-cuda12_6-test:  # Testing
@@ -1960,7 +1969,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_13-cuda12_8
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_13-cuda12_8-test:  # Testing
@@ -2026,7 +2035,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_13-cuda13_0
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_13-cuda13_0-test:  # Testing
@@ -2124,6 +2133,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -2145,7 +2155,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -2153,7 +2163,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -2234,6 +2244,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -2255,7 +2266,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -2263,7 +2274,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -2337,7 +2348,7 @@ jobs:
       contents: read
     steps:
       - name: Setup XPU
-        uses: pytorch/pytorch/.github/actions/setup-xpu@release/2.9
+        uses: pytorch/pytorch/.github/actions/setup-xpu@main
       - name: configure aws credentials
         id: aws_creds
         uses: aws-actions/configure-aws-credentials@v4
@@ -2355,6 +2366,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -2365,7 +2377,7 @@ jobs:
         working-directory: pytorch
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -2373,7 +2385,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -2483,7 +2495,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_13t-cuda12_6
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_13t-cuda12_6-test:  # Testing
@@ -2549,7 +2561,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_13t-cuda12_8
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_13t-cuda12_8-test:  # Testing
@@ -2615,7 +2627,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_13t-cuda13_0
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_13t-cuda13_0-test:  # Testing
@@ -2713,6 +2725,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -2734,7 +2747,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -2742,7 +2755,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -2823,6 +2836,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -2844,7 +2858,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -2852,7 +2866,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -2926,7 +2940,7 @@ jobs:
       contents: read
     steps:
       - name: Setup XPU
-        uses: pytorch/pytorch/.github/actions/setup-xpu@release/2.9
+        uses: pytorch/pytorch/.github/actions/setup-xpu@main
       - name: configure aws credentials
         id: aws_creds
         uses: aws-actions/configure-aws-credentials@v4
@@ -2944,6 +2958,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -2954,7 +2969,7 @@ jobs:
         working-directory: pytorch
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -2962,7 +2977,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -3072,7 +3087,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_14-cuda12_6
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_14-cuda12_6-test:  # Testing
@@ -3138,7 +3153,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_14-cuda12_8
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_14-cuda12_8-test:  # Testing
@@ -3204,7 +3219,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_14-cuda13_0
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_14-cuda13_0-test:  # Testing
@@ -3302,6 +3317,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -3323,7 +3339,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -3331,7 +3347,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -3412,6 +3428,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -3433,7 +3450,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -3441,7 +3458,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -3515,7 +3532,7 @@ jobs:
       contents: read
     steps:
       - name: Setup XPU
-        uses: pytorch/pytorch/.github/actions/setup-xpu@release/2.9
+        uses: pytorch/pytorch/.github/actions/setup-xpu@main
       - name: configure aws credentials
         id: aws_creds
         uses: aws-actions/configure-aws-credentials@v4
@@ -3533,6 +3550,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -3543,7 +3561,7 @@ jobs:
         working-directory: pytorch
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -3551,7 +3569,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -3661,7 +3679,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_14t-cuda12_6
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_14t-cuda12_6-test:  # Testing
@@ -3727,7 +3745,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_14t-cuda12_8
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.10.2.21; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.83; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.3.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.7.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.27.5; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu12==3.3.20; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.90; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.93; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.1.3; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_14t-cuda12_8-test:  # Testing
@@ -3793,7 +3811,7 @@ jobs:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       build_name: manywheel-py3_14t-cuda13_0
       build_environment: linux-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' | nvidia-curand==10.4.0.35; platform_system == 'Linux' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' | nvidia-nvtx==13.0.39; platform_system == 'Linux' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' | nvidia-cufile==1.15.0.42; platform_system == 'Linux'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti==13.0.48; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu13==9.13.0.50; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas==13.0.0.19; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft==12.0.0.15; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand==10.4.0.35; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver==12.0.3.29; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse==12.6.2.49; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu13==0.8.0; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu13==2.27.7; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvshmem-cu13==3.3.24; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink==13.0.39; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile==1.15.0.42; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_14t-cuda13_0-test:  # Testing
@@ -3891,6 +3909,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -3912,7 +3931,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -3920,7 +3939,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -4001,6 +4020,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -4022,7 +4042,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -4030,7 +4050,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary
@@ -4104,7 +4124,7 @@ jobs:
       contents: read
     steps:
       - name: Setup XPU
-        uses: pytorch/pytorch/.github/actions/setup-xpu@release/2.9
+        uses: pytorch/pytorch/.github/actions/setup-xpu@main
       - name: configure aws credentials
         id: aws_creds
         uses: aws-actions/configure-aws-credentials@v4
@@ -4122,6 +4142,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -4132,7 +4153,7 @@ jobs:
         working-directory: pytorch
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -4140,7 +4161,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary

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

@@ -38,13 +38,13 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
       curr_branch: ${{ github.head_ref || github.ref_name }}
       curr_ref_type: ${{ github.ref_type }}
-  manywheel-py3_10-rocm6_4-build:
+  manywheel-py3_9-rocm6_4-build:
     if: ${{ github.repository_owner == 'pytorch' }}
     uses: ./.github/workflows/_binary-build-linux.yml
     needs: get-label-type
@@ -58,16 +58,16 @@ jobs:
       GPU_ARCH_TYPE: rocm
       DOCKER_IMAGE: manylinux2_28-builder
       DOCKER_IMAGE_TAG_PREFIX: rocm6.4
-      DESIRED_PYTHON: "3.10"
+      DESIRED_PYTHON: "3.9"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
-      build_name: manywheel-py3_10-rocm6_4
+      build_name: manywheel-py3_9-rocm6_4
       build_environment: linux-binary-manywheel-rocm
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  manywheel-py3_10-rocm6_4-test:  # Testing
+  manywheel-py3_9-rocm6_4-test:  # Testing
     if: ${{ github.repository_owner == 'pytorch' }}
     needs:
-      - manywheel-py3_10-rocm6_4-build
+      - manywheel-py3_9-rocm6_4-build
       - get-label-type
     runs-on: linux.rocm.gpu.mi250
     timeout-minutes: 240
@@ -82,18 +82,19 @@ jobs:
       SKIP_ALL_TESTS: 1
       DOCKER_IMAGE: manylinux2_28-builder
       DOCKER_IMAGE_TAG_PREFIX: rocm6.4
-      DESIRED_PYTHON: "3.10"
+      DESIRED_PYTHON: "3.9"
     steps:
       - name: Setup ROCm
         uses: ./.github/actions/setup-rocm
       - uses: actions/download-artifact@v4.1.7
         name: Download Build Artifacts
         with:
-          name: manywheel-py3_10-rocm6_4
+          name: manywheel-py3_9-rocm6_4
           path: "${{ runner.temp }}/artifacts/"
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -115,7 +116,7 @@ jobs:
           role-duration-seconds: 18000
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
           docker-image-name: manylinux2_28-builder
@@ -123,7 +124,7 @@ jobs:
           docker-build-dir: .ci/docker
           working-directory: pytorch
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.9
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
       - name: Test Pytorch binary

.github/workflows/generated-linux-s390x-binary-manywheel-nightly.yml

@@ -41,7 +41,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

.github/workflows/generated-macos-arm64-binary-libtorch-release-nightly.yml

@@ -70,6 +70,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false

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

@@ -66,6 +66,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -205,6 +206,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -344,6 +346,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -483,6 +486,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -622,6 +626,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -761,6 +766,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -900,6 +906,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false

.github/workflows/generated-windows-arm64-binary-libtorch-debug-nightly.yml

@@ -41,7 +41,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

.github/workflows/generated-windows-arm64-binary-libtorch-release-nightly.yml

@@ -41,7 +41,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

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

@@ -41,7 +41,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

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

@@ -28,7 +28,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -77,7 +77,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -109,6 +109,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -182,7 +183,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -214,6 +215,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false

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

@@ -35,7 +35,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -84,7 +84,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -116,6 +116,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -189,7 +190,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -221,6 +222,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -330,7 +332,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -362,6 +364,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -436,7 +439,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -468,6 +471,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -578,7 +582,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -610,6 +614,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -684,7 +689,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -716,6 +721,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -826,7 +832,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -858,6 +864,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -932,7 +939,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -964,6 +971,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false

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

@@ -28,7 +28,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -77,7 +77,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -109,6 +109,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -182,7 +183,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -214,6 +215,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false

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

@@ -35,7 +35,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -84,7 +84,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -116,6 +116,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -189,7 +190,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -221,6 +222,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -330,7 +332,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -362,6 +364,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -436,7 +439,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -468,6 +471,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -578,7 +582,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -610,6 +614,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -684,7 +689,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -716,6 +721,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -826,7 +832,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -858,6 +864,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -932,7 +939,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.9
+        uses: pytorch/test-infra/.github/actions/setup-ssh@main
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -964,6 +971,7 @@ jobs:
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
+          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false

.github/workflows/h100-cutlass-backend.yml

@@ -27,7 +27,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

.github/workflows/h100-distributed.yml

@@ -24,7 +24,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

.github/workflows/h100-symm-mem.yml

@@ -24,7 +24,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

.github/workflows/inductor-micro-benchmark.yml

@@ -20,7 +20,7 @@ permissions:
 jobs:
   get-default-label-prefix:
     name: get-default-label-prefix
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}

.github/workflows/inductor-nightly.yml

@@ -23,7 +23,7 @@ permissions:
 jobs:
   get-default-label-prefix:
     name: get-default-label-prefix
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}
@@ -37,7 +37,7 @@ jobs:
     uses: ./.github/workflows/_linux-build.yml
     needs: get-default-label-prefix
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       docker-image-name: ci-image:pytorch-linux-jammy-py3-gcc11-inductor-benchmarks
       runner_prefix: "${{ needs.get-default-label-prefix.outputs.label-type }}"
       test-matrix: |
@@ -56,7 +56,7 @@ jobs:
     uses: ./.github/workflows/_linux-test.yml
     needs: nightly-dynamo-benchmarks-build
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       docker-image: ${{ needs.nightly-dynamo-benchmarks-build.outputs.docker-image }}
       test-matrix: ${{ needs.nightly-dynamo-benchmarks-build.outputs.test-matrix }}
       timeout-minutes: 720

.github/workflows/inductor-perf-compare.yml

@@ -18,7 +18,7 @@ jobs:
   get-default-label-prefix:
     if: github.repository_owner == 'pytorch'
     name: get-default-label-prefix
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

.github/workflows/inductor-perf-test-b200.yml

@@ -70,7 +70,7 @@ permissions:
 jobs:
   get-label-type:
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}

.github/workflows/inductor-perf-test-nightly-aarch64.yml

@@ -55,7 +55,7 @@ permissions:
 jobs:
   get-label-type:
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}

.github/workflows/inductor-perf-test-nightly-h100.yml

@@ -75,7 +75,7 @@ permissions:
 jobs:
   get-label-type:
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}

.github/workflows/inductor-perf-test-nightly-rocm.yml

@@ -70,7 +70,7 @@ permissions: read-all
 jobs:
   get-label-type:
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}

.github/workflows/inductor-perf-test-nightly-x86-zen.yml

@@ -60,7 +60,7 @@ permissions:
 jobs:
   get-label-type:
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}
@@ -75,7 +75,7 @@ jobs:
     needs: get-label-type
     with:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       docker-image-name: ci-image:pytorch-linux-jammy-py3-gcc11-inductor-benchmarks
       test-matrix: |
         { include: [
@@ -101,7 +101,7 @@ jobs:
     needs: inductor-build
     if: github.event.schedule == '0 7 * * *'
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       dashboard-tag: training-false-inference-true-default-true-dynamic-true-cppwrapper-true-aotinductor-true
       docker-image: ${{ needs.inductor-build.outputs.docker-image }}
       test-matrix: ${{ needs.inductor-build.outputs.test-matrix }}
@@ -118,7 +118,7 @@ jobs:
     needs: inductor-build
     if: github.event_name == 'workflow_dispatch'
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       dashboard-tag: training-${{ inputs.training }}-inference-${{ inputs.inference }}-default-${{ inputs.default }}-dynamic-${{ inputs.dynamic }}-cppwrapper-${{ inputs.cppwrapper }}-aotinductor-${{ inputs.aotinductor }}
       docker-image: ${{ needs.inductor-build.outputs.docker-image }}
       test-matrix: ${{ needs.inductor-build.outputs.test-matrix }}

.github/workflows/inductor-perf-test-nightly-x86.yml

@@ -65,7 +65,7 @@ permissions:
 jobs:
   get-label-type:
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}
@@ -80,7 +80,7 @@ jobs:
     needs: get-label-type
     with:
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       docker-image-name: ci-image:pytorch-linux-jammy-py3-gcc11-inductor-benchmarks
       test-matrix: |
         { include: [
@@ -107,7 +107,7 @@ jobs:
     needs: inductor-build
     if: github.event.schedule == '0 7 * * *'
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       dashboard-tag: training-false-inference-true-default-true-dynamic-true-cppwrapper-true-aotinductor-true-freezing-true
       docker-image: ${{ needs.inductor-build.outputs.docker-image }}
       test-matrix: ${{ needs.inductor-build.outputs.test-matrix }}
@@ -124,7 +124,7 @@ jobs:
     needs: inductor-build
     if: github.event_name == 'workflow_dispatch'
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       dashboard-tag: training-${{ inputs.training }}-inference-${{ inputs.inference }}-default-${{ inputs.default }}-dynamic-${{ inputs.dynamic }}-cppwrapper-${{ inputs.cppwrapper }}-aotinductor-${{ inputs.aotinductor }}-freezing-${{ inputs.freezing }}
       docker-image: ${{ needs.inductor-build.outputs.docker-image }}
       test-matrix: ${{ needs.inductor-build.outputs.test-matrix }}

.github/workflows/inductor-perf-test-nightly.yml

@@ -70,7 +70,7 @@ permissions:
 jobs:
   get-label-type:
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}

.github/workflows/inductor-periodic.yml

@@ -22,7 +22,7 @@ permissions:
 jobs:
   get-default-label-prefix:
     name: get-default-label-prefix
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}
@@ -154,7 +154,7 @@ jobs:
     uses: ./.github/workflows/_linux-build.yml
     needs: get-default-label-prefix
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       docker-image-name: ci-image:pytorch-linux-jammy-py3-gcc11-inductor-benchmarks
       runner_prefix: "${{ needs.get-default-label-prefix.outputs.label-type }}"
       test-matrix: |
@@ -200,7 +200,7 @@ jobs:
     uses: ./.github/workflows/_linux-test.yml
     needs: periodic-dynamo-benchmarks-cpu-build
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       docker-image: ${{ needs.periodic-dynamo-benchmarks-cpu-build.outputs.docker-image }}
       test-matrix: ${{ needs.periodic-dynamo-benchmarks-cpu-build.outputs.test-matrix }}
     secrets: inherit

.github/workflows/inductor-rocm-mi300.yml

@@ -28,7 +28,7 @@ jobs:
 
   get-label-type:
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}

.github/workflows/inductor-rocm.yml

@@ -20,7 +20,7 @@ permissions:
 jobs:
   get-label-type:
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}

.github/workflows/inductor-unittest.yml

@@ -19,7 +19,7 @@ permissions:
 jobs:
   get-label-type:
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}
@@ -110,7 +110,7 @@ jobs:
     uses: ./.github/workflows/_linux-build.yml
     needs: get-label-type
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       docker-image-name: ci-image:pytorch-linux-jammy-py3-gcc11-inductor-benchmarks
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       test-matrix: |
@@ -127,7 +127,7 @@ jobs:
     uses: ./.github/workflows/_linux-test.yml
     needs: inductor-cpu-build
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       docker-image: ${{ needs.inductor-cpu-build.outputs.docker-image }}
       test-matrix: ${{ needs.inductor-cpu-build.outputs.test-matrix }}
     secrets: inherit

.github/workflows/inductor.yml

@@ -35,7 +35,7 @@ jobs:
 
   get-label-type:
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.9
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
     if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
     with:
       triggering_actor: ${{ github.triggering_actor }}
@@ -79,7 +79,7 @@ jobs:
     uses: ./.github/workflows/_linux-build.yml
     needs: get-label-type
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       docker-image-name: ci-image:pytorch-linux-jammy-py3-gcc11-inductor-benchmarks
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       test-matrix: |
@@ -101,7 +101,7 @@ jobs:
     uses: ./.github/workflows/_linux-test.yml
     needs: inductor-cpu-build
     with:
-      build-environment: linux-jammy-py3.10-gcc11-build
+      build-environment: linux-jammy-py3.9-gcc11-build
       docker-image: ${{ needs.inductor-cpu-build.outputs.docker-image }}
       test-matrix: ${{ needs.inductor-cpu-build.outputs.test-matrix }}
     secrets: inherit

.github/workflows/lint-autoformat.yml

@@ -13,7 +13,7 @@ jobs:
     if: ${{ github.repository_owner == 'pytorch' && contains(github.event.pull_request.labels.*.name, 'autoformat') }}
     steps:
       - name: Checkout pytorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.9
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
         with:
           submodules: true
           fetch-depth: 0