consistent with the original source files registering each operator. For example
since `REGISTER_OP("MatMul")` appears in ***core/math_ops.cc***, the "MatMul"
operator in the script should be in the "math" category, and it will be
generated in the output file `c/experimental/ops/math_ops.cc`.

Running this script should be a no-op, generating identical code other than

# ==============================================================================

# Need a newer version of patchelf as the installed version is buggy in 20.04
# so get patchelf source from 22.04 ie 'jammy' and build it to avoid dependency
# problems that would occur with a binary package

mkdir -p /patchelf
cd /patchelf
echo deb-src http://ports.ubuntu.com/ubuntu-ports/ jammy universe>>/etc/apt/sources.list
apt-get update

TensorFlow server (`tf.train.Server`). The TensorFlow server is intended for
internal communication only. It is not built for use in untrusted environments
or networks.

For performance reasons, the default TensorFlow server does not include any
authorization protocol and sends messages unencrypted. It accepts connections
from anywhere, and executes the graphs it is sent without performing any checks.

setup_file() {
    cd /tf/tensorflow
    bazel version  # Start the bazel server
}

# Do a bazel query specifically for the licenses checker. It searches for
# targets matching the provided query, which start with // or @ but not
# //tensorflow (so it looks for //third_party, //external, etc.), and then
# gathers the list of all packages (i.e. directories) which contain those
# targets.
license_query() {

for wheel in /tf/pkg/*.whl; do
  echo "Checking and renaming $wheel..."
  time python3 -m auditwheel repair --plat manylinux2014_aarch64 "$wheel" --wheel-dir /tf/pkg 2>&1 | tee check.txt

  # We don't need the original wheel if it was renamed
  new_wheel=$(grep --extended-regexp --only-matching '/tf/pkg/\S+.whl' check.txt)
  if [[ "$new_wheel" != "$wheel" ]]; then
    rm "$wheel"
    wheel="$new_wheel"
  fi

  // and making a new one in its place, analogous to unique_ptr
  // semantics.
  Cleanup& operator=(Cleanup&& src) {  // NOLINT
    if (!released_) f_();
    released_ = src.released_;
    f_ = src.release();
    return *this;
  }

  ~Cleanup() {
    if (!released_) f_();
  }

  // Releases the cleanup function instead of running it.
  // Hint: use c.release()() to run early.

# limitations under the License.
# ==============================================================================
#
# setup.python.sh: Install a specific Python version and packages for it.
# Usage: setup.python.sh <pyversion> <requirements.txt>

# Sets up custom apt sources for our TF images.

# Prevent apt install tzinfo from asking our location (assumes UTC)
export DEBIAN_FRONTEND=noninteractive

can view more detailed results that are probably easier to read than this log.
Try the links below:
EOF
  # Find any "Streaming build results to" line, then print the last word in it,
  # and don't print duplicates
  awk '/Streaming build results to/ {print $NF}' "$TFCI_OUTPUT_DIR/script.log" | uniq
}

# Print out any ResultStore URLs for Bazel invocations' results.

libcudnn8=8.9.6.50-1+cuda12.2
libcudnn8-dev=8.9.6.50-1+cuda12.2

# This can be removed once NVIDIA publishes a cuda-12.3.2 Docker image.
# For now it ensures that we install at least version 12.3.107 of PTXAS,
# since 12.3.103 has a bug.

```

### 3. Failure after conversion
If the conversion is successful, but the generated model is wrong, then state what is wrong:

- Model produces wrong results and/or has lesser accuracy.
- Model produces correct results, but it is slower than expected.

### 4. (optional) RNN conversion support
If converting TF RNN to TFLite fused RNN ops, please prefix [RNN] in the title.

### 5. (optional) Any other info / logs