Why naming `call` as `_call_impl` in module.py?

marsggbo · April 1, 2021, 6:58am

What’s the purpose to name __call__ as _call_impl in module.py?

pytorch/pytorch/blob/6145ac07b5431a95e7d29f84ab979b332883b4ea/torch/nn/modules/module.py#L1065


        grad_fn = var.grad_fn
        if grad_fn is not None:
            for hook in non_full_backward_hooks:
                wrapper = functools.partial(hook, self)
                functools.update_wrapper(wrapper, hook)
                grad_fn.register_hook(wrapper)
            self._maybe_warn_non_full_backward_hook(input, result, grad_fn)

    return result

__call__ : Callable[..., Any] = _call_impl

def __setstate__(self, state):
    self.__dict__.update(state)
    # Support loading old checkpoints that don't have the following attrs:
    if '_forward_pre_hooks' not in self.__dict__:
        self._forward_pre_hooks = OrderedDict()
    if '_state_dict_hooks' not in self.__dict__:
        self._state_dict_hooks = OrderedDict()
    if '_load_state_dict_pre_hooks' not in self.__dict__:
        self._load_state_dict_pre_hooks = OrderedDict()

Why naming `__call__` as `_call_impl` in module.py?

Why naming `call` as `_call_impl` in module.py?