debug_info_txt_raw.ptr(), enable_mlir_converter,
            quantization_py_function_library));
      },
      py::arg("model_flags_proto_txt_raw"), py::arg("toco_flags_proto_txt_raw"),
      py::arg("input_contents_txt_raw"), py::arg("extended_return") = false,
      py::arg("debug_info_txt_raw") = py::none(),
      py::arg("enable_mlir_converter") = false,

    model_flags_str,
    toco_flags_str,
    input_data_str,
    debug_info_str,
    enable_mlir_converter,
):
  """Wraps TocoConvert with lazy loader."""
  return _pywrap_converter_api.Convert(
      model_flags_str,
      toco_flags_str,
      input_data_str,
      False,  # extended_return
      debug_info_str,
      enable_mlir_converter,
      py_function_lib.PyFunctionLibrary(),
  )

// flag is set to true returns a dictionary that contains string representation
// of the converted model and some statistics like arithmetic ops count.
// `debug_info_str` contains the `GraphDebugInfo` proto. When
// `enable_mlir_converter` is True, use MLIR-based conversion instead of
// TOCO conversion.
PyObject* Convert(PyObject* model_flags_proto_txt_raw,
                  PyObject* toco_flags_proto_txt_raw,

def Convert(model_flags_proto_txt_raw: object, toco_flags_proto_txt_raw: object, input_contents_txt_raw: object, extended_return: bool = ..., debug_info_txt_raw: object = ..., enable_mlir_converter: bool = ..., quantization_py_function_library = ...) -> object: ...

                  PyObject* toco_flags_proto_txt_raw,
                  PyObject* input_contents_txt_raw, bool extended_return,
                  PyObject* debug_info_txt_raw, bool enable_mlir_converter,
                  const tensorflow::quantization::PyFunctionLibrary*
                      quantization_py_function_library) {
  // Use Python C API to validate and convert arguments. In py3 (bytes),