auto collective_executor_handle = context->GetCollectiveExecutorHandle();
  tensorflow::Notification done;
  collective_executor_handle->get()->remote_access()->CheckPeerHealth(
      task, timeout_in_ms, [&done, status](const Status& s) {
        status->status = s;
        done.Notify();
      });
  done.WaitForNotification();
}

TF_ShapeAndTypeList* TF_NewShapeAndTypeList(int num_items) {

BASE="$1"
find "${BASE}" -type l | \
  while read l ; do
    if [[ "$(readlink "$l")" == \.\./* ]]; then
      CANONICAL="$(readlink "$l")";
      rm "$l";
      ln -s "${CANONICAL}" "$l"
    fi

  if [[ "$new_wheel" != "$wheel" ]]; then
    rm "$wheel"
    wheel="$new_wheel"
  fi

  TF_WHEEL="$wheel" bats /usertools/wheel_verification.bats --timing

    --enable_bzlmod=false \
    //:requirements_"$VERSION"_release.update
  sed -i '/^#/d' requirements_lock_"$VERSION".txt
  mv "requirements_lock_"$VERSION".txt" ../../../requirements_lock_"$VERSION".txt
done

for f in $(ls | grep python); do
  # set up symlink for devtoolset-10
  rm -f /dt10/usr/include/aarch64-linux-gnu/$f
  ln -s /usr/include/aarch64-linux-gnu/$f /dt10/usr/include/aarch64-linux-gnu/$f
done
popd

# Python 3.10 include headers fix:
# sysconfig.get_path('include') incorrectly points to /usr/local/include/python
# map /usr/include/python3.10 to /usr/local/include/python3.10

      # it, and print that list for debugging. Not really clear if this is
      # helpful since the only examples I've seen are enormous.
      bazel cquery "rdeps(kind(py_test, $(cat $BATS_TEST_TMPDIR/deps)), $dep, 1)"
    done < $BATS_TEST_TMPDIR/missing_deps
    exit 1
  fi
}

# The Python package is not allowed to depend on any CUDA packages.
@test "Pip package doesn't depend on CUDA" {
  bazel cquery \

    --enable_bzlmod=false \
    //:requirements_"$VERSION".update
  sed -i '/^#/d' requirements_lock_"$VERSION".txt
  mv requirements_lock_"$VERSION".txt ../../../requirements_lock_"$VERSION".txt
done

  }
  return absl::OkStatus();
}

// TODO(agarwal): use an automatic mechanism for handling None arguments to
// gradient functions.
//
// Some gradient functions can accept None arguments for gradients. The
// following maps the operation name to the indices at which the corresponding
// gradient function can accept None values. e.g. FusedBatchNorm outputs 5
// values and hence receives 5 gradient values during backprop. However the

}

TEST_P(CppGradients, TestMatMulGrad) {
  // TODO(vnvo2409): Figure out why `gradient_checker` does not work very
  // well with `MatMul` and remove `TestMatMul*` in
  // `mnist_gradients_test` when done.
  GTEST_SKIP();

  float A_vals[] = {1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f, 7.0f, 8.0f, 9.0f};
  int64_t A_dims[] = {3, 3};
  AbstractTensorHandlePtr A;
  {
    AbstractTensorHandle* A_raw;

PYTHON_VERSIONS=("python3.8" "python3.9" "python3.10" "python3.11")
for v in "${PYTHON_VERSIONS[@]}"; do
  ln -s "/usr/local/include/${v}" "${TARGET}/usr/include/aarch64-linux-gnu/${v}"