EOF
  exit 1
fi
TFCI_BAZEL_COMMON_ARGS="$TFCI_BAZEL_COMMON_ARGS --config rbe_$TFCI_BAZEL_TARGET_SELECTING_CONFIG_PREFIX"

# These flags share the user's gcloud credentials with the container, so that bazel
# inside the container can authenticate. Note: TF's CI does not have any credential
# stored here.

  gcloud auth activate-service-account --key-file="${GOOGLE_APPLICATION_CREDENTIALS}"
fi
set -x

# When cross-compiling with RBE, we need to copy the macOS sysroot to be
# inside the TensorFlow root directory. We then define them as a filegroup
# target inside "tensorflow/tools/toolchains/cross_compile/cc" so that Bazel
# can register it as an input to compile/link actions and send it to the remote
# VMs when needed.

as a rough example of the steps below.

All the files referenced below are located in the same directory as this README,
unless indicated otherwise.

1) Add the new version to the `VERSIONS` variable inside
   `tensorflow/tools/toolchains/python/python_repo.bzl`. \
   While this isn't necessary for running the updater, it is required for
   actually using the new version with Tensorflow.

from tensorflow.python.tools import module_util as _module_util
from tensorflow.python.util.lazy_loader import KerasLazyLoader as _KerasLazyLoader

# Make sure code inside the TensorFlow codebase can use tf2.enabled() at import.
_os.environ["TF2_BEHAVIOR"] = "1"
from tensorflow.python import tf2 as _tf2
_tf2.enable()

# API IMPORTS PLACEHOLDER

# WRAPPER_PLACEHOLDER

  # $TF_VER_FULL will resolve to e.g. "2.15.0-rc2". Since $TF_VER_FULL comes
  # from get_versions.sh, which must be run *after* update_version.py, FINAL_URI
  # can't be set inside the rest of the _upload envs.
  FINAL_URI="$TFCI_ARTIFACT_FINAL_GCS_URI/$TF_VER_FULL"
  gsutil -m rsync -d -r "$DOWNLOADS" "$FINAL_URI"

  # Also mirror the latest-uploaded folder to the "latest" directory.

    *   `conversion_params` is now deprecated inside `TrtGraphConverterV2` in
        favor of direct arguments: `max_workspace_size_bytes`, `precision_mode`,
        `minimum_segment_size`, `maximum_cached_engines`, `use_calibration` and
        `allow_build_at_runtime`.
    *   Added a new parameter called `save_gpu_specific_engines` to the
        `.save()` function inside `TrtGraphConverterV2`. When `False`, the

  )

  if get_var(
      environ_cp, 'TF_OVERRIDE_EIGEN_STRONG_INLINE', 'Eigen strong inline',
      True, ('Would you like to override eigen strong inline for some C++ '
             'compilation to reduce the compilation time?'),
      'Eigen strong inline overridden.', 'Not overriding eigen strong inline, '
      'some compilations could take more than 20 mins.'):
    # Due to a known MSVC compiler issue

# See https://developer.nvidia.com/cuda-gpus#compute
# `compute_XY` enables PTX embedding in addition to SASS. PTX
# is forward compatible beyond the current compute capability major
# release while SASS is only forward compatible inside the current
# major release. Example: sm_80 kernels can run on sm_89 GPUs but
# not on sm_90 GPUs. compute_80 kernels though can also run on sm_90 GPUs.

)

config_setting(
    name = "using_rocm_hipcc",
    define_values = {"using_rocm_hipcc": "true"},
)

config_setting(
    name = "override_eigen_strong_inline",
    define_values = {"override_eigen_strong_inline": "true"},
    visibility = ["//visibility:public"],
)

# This flag specifies whether TensorFlow 2.0 API should be built instead

  };
  // A deque-backed stack, whose element references are not invalidated by
  // pushes and pops at the back.
  std::stack<AccumulatorCallState> call_state_;
};

// Template instantiations here

inline bool IsDtypeTrainable(DataType dtype) {
  switch (dtype) {
    case DT_HALF:
    case DT_BFLOAT16:
    case DT_FLOAT:
    case DT_DOUBLE:
    case DT_COMPLEX64:
    case DT_COMPLEX128: