std::vector<int64_t> size_as_vector;
};

std::vector<int64_t> IntTensorAsVector(const Tensor& t) {
  DCHECK(t.dtype() == DT_INT32 || t.dtype() == DT_INT64);
  std::vector<int64_t> result;
  result.reserve(t.NumElements());
  for (int i = 0; i < t.NumElements(); i++) {
    int64_t element = t.dtype() == DT_INT32
                          ? static_cast<int64_t>(t.flat<int32>()(i))
                          : t.flat<int64_t>()(i);

      shape_determination_fns_.layout_preference_fn(
          device_tensor->shape(), device_tensor->dtype(), std::nullopt);
  Status status = [&]() -> Status {
    TF_ASSIGN_OR_RETURN(xla::Shape shape,
                        shape_determination_fns_.shape_representation_fn(
                            device_tensor->shape(), device_tensor->dtype(),
                            /*fast_mem=*/false, layout_preference));

  FunctionDef make_ref_float = FunctionDefHelper::Define(
      "RefFloatFn", {}, {"r:float"}, {},
      {{{"var"},
        "VariableV2",
        {},
        {{"dtype", DT_FLOAT}, {"shape", TensorShape({})}}},
       {{"r"}, "Identity", {"var"}, {{"T", DT_FLOAT}}}});
  *fdef_lib->add_function() = make_ref_float;
}

void AddRegularFunctionFunctionDef(FunctionDefLibrary* fdef_lib) {

  const tensorflow::Tensor& arg_tensor = cc_ctx->input(index);
  absl::Status cc_status;
  if (arg_tensor.dtype() != tensorflow::DT_RESOURCE) {
    cc_status = absl::InvalidArgumentError(
        absl::StrCat("Trying to obtain resource handle from Input[", index,
                     "], which is not type DT_RESOURCE."));
    status->status = cc_status;
    return nullptr;
  }
  const tensorflow::ResourceHandle& handle =

  TF_Status* status = TF_NewStatus();
  // Create the variable handle.
  TFE_Op* op = TFE_NewOp(ctx, "VarHandleOp", status);
  if (TF_GetCode(status) != TF_OK) return nullptr;
  TFE_OpSetAttrType(op, "dtype", TF_FLOAT);
  TFE_OpSetAttrShape(op, "shape", {}, 0, status);
  TFE_OpSetAttrString(op, "container", "localhost", 0);
  TFE_OpSetAttrString(op, "shared_name", "", 0);
  if (!device_name.empty()) {

  tensorflow::TensorShape shape(variable.shape());
  tensorflow::DataType dtype = variable.dtype();
  std::vector<std::string> component_devices;

  for (const auto& component :
       variable.experimental_distributed_variable_components()) {
    component_devices.push_back(component.device());
  }

  TF_RETURN_IF_ERROR(Variable::CreateUninitialized(
      ctx, dtype, shape, name,

      tensorflow::OpInfo::TensorProperties in_props;
      Status s = tensorflow::BufferToMessage(in_props_buf[0], &in_props);
      TF_ASSERT_OK(s);

      EXPECT_EQ(DT_FLOAT, in_props.dtype());
      EXPECT_FALSE(in_props.shape().unknown_rank());
      EXPECT_EQ(2, in_props.shape().dim_size());
      EXPECT_EQ(10, in_props.shape().dim(0).size());
      EXPECT_EQ(1, in_props.shape().dim(1).size());

  }
  strings::StrAppend(&ret, "}");
  return ret;
}

string PrintTensor(const TensorProto& tensor_proto) {
  Tensor t(tensor_proto.dtype());
  CHECK(t.FromProto(tensor_proto));
  const int64_t num_elts = t.NumElements();
  switch (t.dtype()) {
    case DT_FLOAT:
      return PrintArray(num_elts, t.flat<float>().data());
    case DT_DOUBLE:
      return PrintArray(num_elts, t.flat<double>().data());

  }

  // Derive the output types. The result type is derived by using the
  // attributes attched to the result type of the signature. The attribute
  // value should be either in the attribute argument list or the derived
  // attribute from the input tensors. All the result type
  // are unranked, and shape inference should be applied afterwards.
  SmallVector<Type, 4> output_types;

  }

  tensorflow::AbstractTensorInterface* t =
      tensorflow::unwrap(ctx)->CreateTensor(
          static_cast<tensorflow::DataType>(dtype), dimvec);

  if (t == nullptr) {
    status->status =
        tensorflow::errors::InvalidArgument("Unsupported dtype: ", dtype);
    return nullptr;
  }

  return new TF_Tensor{t};
}

TFE_TensorHandle* TFE_NewTensorHandleFromTensor(TFE_Context* ctx, TF_Tensor* t,