std::unordered_map<const Node*, std::vector<int>>* input_nodes)
    TF_EXCLUSIVE_LOCKS_REQUIRED(fn_body->mu) {
  input_tensors->reserve(ninputs);
  for (int i = 0; i < ninputs; ++i) {
    Node* node = inputs[i].oper ? &inputs[i].oper->node : nullptr;
    int idx = inputs[i].index;

    TF_RETURN_WITH_CONTEXT_IF_ERROR(
        fn_body->graph.IsValidOutputTensor(node, idx),

    // below. Allocate enough space so that no reallocation happens, which will
    // make the pointers invalid.
    all_input_tensors.reserve(num_inputs);
    for (int i = 0; i < num_inputs; ++i) {
      if (input_tensors[i] == nullptr) continue;
      all_input_tensors.emplace_back();
      Tensor& input_tensor = all_input_tensors.back();
      status->status = TF_TensorToTensor(input_tensors[i], &input_tensor);

/**
 * A future whose value is derived from a collection of input futures.
 *
 * @param <InputT> the type of the individual inputs
 * @param <OutputT> the type of the output (i.e. this) future
 */
@GwtCompatible
@ElementTypesAreNonnullByDefault
abstract class AggregateFuture<InputT extends @Nullable Object, OutputT extends @Nullable Object>
    extends AggregateFutureState<OutputT> {

      : forward_inputs_(f_inputs) {
    for (auto input : forward_inputs_) {
      if (input) {
        input->Ref();
      }
    }
  }

  Status Compute(AbstractContext* ctx,
                 absl::Span<AbstractTensorHandle* const> grad_outputs,
                 absl::Span<AbstractTensorHandle*> grad_inputs) override {
    // TODO(vnvo2409): Add control dependency

// to the passed operator
func checkPolicyCond(op string, input1, input2 string) bool {
	switch op {
	case policyCondEqual:
		return input1 == input2
	case policyCondStartsWith:
		return strings.HasPrefix(input1, input2)
	}
	return false
}

// checkPostPolicy - apply policy conditions and validate input values.
// (http://docs.aws.amazon.com/AmazonS3/latest/API/sigv4-HTTPPOSTConstructPolicy.html)

    std::vector<TFE_TensorHandle*> device_inputs;
    device_inputs.reserve(inputs.size());
    for (int input_index = 0; input_index < inputs.size(); ++input_index) {
      // Parallel tensors are divided between operations by device.
      device_inputs.push_back(inputs[input_index]->tensor(device_index));
    }
    device_thread->StartExecute(
        context, operation_name, std::move(device_inputs), attributes,

            final VaErrorHook hook) {
        return getEntity(form).map(entity -> {
            final String[] newInputs = splitLine(form.inputs);
            validateMappingString(action, newInputs, "inputs", hook);
            entity.setNewInputs(newInputs);
            final String newOutput = form.output;
            entity.setNewOutput(newOutput);
            return entity;
        });
    }

void Tape::RecordOperation(absl::Span<AbstractTensorHandle* const> inputs,
                           absl::Span<AbstractTensorHandle* const> outputs,
                           GradientFunction* gradient_function,
                           const string& op_name) {
  std::vector<int64_t> input_ids(inputs.size());
  std::vector<tensorflow::DataType> input_dtypes(inputs.size());
  for (int i = 0; i < inputs.size(); i++) {
    input_ids[i] = ToId(inputs[i]);

      Iterable<? extends T>... inputs) {
    return concatNoDefensiveCopy(Arrays.copyOf(inputs, inputs.length));
  }

  /**
   * Returns a fluent iterable that combines several iterables. The returned iterable has an
   * iterator that traverses the elements of each iterable in {@code inputs}. The input iterators
   * are not polled until necessary.
   *

      Iterable<? extends T>... inputs) {
    return concatNoDefensiveCopy(Arrays.copyOf(inputs, inputs.length));
  }

  /**
   * Returns a fluent iterable that combines several iterables. The returned iterable has an
   * iterator that traverses the elements of each iterable in {@code inputs}. The input iterators
   * are not polled until necessary.
   *