A question about the class 'MetaDAIN'

[Fairydetail]

        param_dict = dict()
        if params is not None:
            param_dict = extract_top_level_dict(current_dict=params)
            cur_output_rectified = self.rectifyNet(rectify_input, params=param_dict['rectifyNet']) + cur_output
        else:
            cur_output_rectified = self.rectifyNet(rectify_input) + cur_output

As the code above, the only difference in the forward function between class 'MetaDAIN' and 'DAIN' is the 'rectifyNet'. Does it mean that only params in 'rectifyNet' will be updated in the inner loop optimization? Hope for your reply, thanks a lot.