sufficient description, if applicable unit tests are added, if it is a
  reasonable contribution (meaning it is not a single liner cosmetic PR).

**2. Valid?**

-   If the PR passes all the quality checks then we go ahead and assign a
    reviewer.
-   If the PR didn't meet the validation criteria, we request for additional
    changes to be made to PR to pass quality checks and send it back or on a
    rare occasion we may reject it.

**3. Review**

  // be an empty string if no device was explicitly requested, but will
  // otherwise be the name of this custom device. Ops are placed onto a custom
  // device if any of their inputs are on that custom device, but custom devices
  // are free to set a bad status in order to require explicit placement.
  void (*execute)(const TFE_Op* op, int* num_outputs,

      vspace, target_tensor_ids, source_tensor_ids, sources_that_are_targets,
      output_gradients, result, /*build_default_zeros_grads*/ false));
  return absl::OkStatus();
}

// Helper functions which delegate to `AbstractOperation`, update
// the state of the ForwardOperation and call the tape as appropriate.
// These APIs are mainly to facilitate testing and are subject to change.
namespace internal {

  switch (dtype.code) {
    case DLDataTypeCode::kDLBool:
      if (dtype.bits != 8) {
        return tensorflow::errors::InvalidArgument(
            "Only DLPack bools of bitwidth 8 are supported, got: ", dtype.bits);
      }
      *tf_dtype = TF_DataType::TF_BOOL;
      return absl::OkStatus();

    case DLDataTypeCode::kDLUInt:
      switch (dtype.bits) {
        case 8:

                 absl::Span<AbstractTensorHandle* const> grad_outputs,
                 absl::Span<AbstractTensorHandle*> grad_inputs) override {
    /* Given upstream grad U and a matmul op A*B, the gradients are:
     *
     *    dA = U * B.T
     *    dB = A.T * U
     *
     *    where A.T means `transpose(A)`
     */
    AbstractTensorHandle* upstream_grad = grad_outputs[0];

    // Get transpose attrs

  }
  ~GraphOperation() override {}

 private:
  friend class GraphContext;  // For access to op_.
  TF_Graph* g_;
  std::unique_ptr<TF_OperationDescription> op_;
  // Hold `op_type` and `op_name` till both are available since we need both
  // to build a graph operation.
  string op_type_;
  const char* op_name_ = nullptr;
  // TODO(srbs): Use this.
  string device_name_;
};

// TODO(b/291142876) Simplify TFE_ContextSetServerDefWithTimeoutAndRetries and
// TFE_ContextUpdateServerDefWithTimeout to be simple wrappers around the same
// C++ function.
// Retries are used for CreateContext calls, which is used in
// ParameterServerStrategy initialization to be robust to worker preemption.
TF_CAPI_EXPORT extern void TFE_ContextSetServerDefWithTimeoutAndRetries(

# the dependencies of that target which include a license notice file are then
# included when generating that license. Necessary because the license targets
# in TensorFlow are manually enumerated rather than generated automatically.
do_external_licenses_check(){
  BUILD_TARGET="$1"
  LICENSES_TARGET="$2"

  # grep patterns for targets which are allowed to be missing from the licenses

};

// If `persistent_tape` is true, op_tape is not changed and none of the
// backwards functions are deleted.
// If `persistent_tape` is false, op_tape is cleared and backwards functions
// not needed for gradient computation are deleted. Backwards functions that
// are needed, are copied and returned in BackpropInitialState.
template <typename BackwardFunction, typename TapeTensor>

  auto* optimizer_options =
      config.mutable_graph_options()->mutable_optimizer_options();
  if (enable) {
    optimizer_options->set_global_jit_level(tensorflow::OptimizerOptions::ON_1);

    // These XLA flags are needed to trigger XLA properly from C (more generally
    // non-Python) clients. If this API is called again with `enable` set to
    // false, it is safe to keep these flag values as is.