Lightning-AI/pytorch-lightning · error · TypeError

Expected a method or a string, but got: {type(method).__name

Error message

Expected a method or a string, but got: {type(method).__name__}

What it means

FabricModule.mark_forward_method() only accepts a bound method or a method name string so it can register a user-defined method to be routed through the strategy wrapper (e.g. DDP sync). Passing any other type raises TypeError.

Source

Thrown at src/lightning/fabric/wrappers.py:168

    def state_dict(
        self, destination: Optional[T_destination] = None, prefix: str = "", keep_vars: bool = False
    ) -> Optional[dict[str, Any]]:
        return self._original_module.state_dict(
            destination=destination,  # type: ignore[type-var]
            prefix=prefix,
            keep_vars=keep_vars,
        )

    @override
    def load_state_dict(  # type: ignore[override]
        self, state_dict: Mapping[str, Any], strict: bool = True, **kwargs: Any
    ) -> _IncompatibleKeys:
        return self._original_module.load_state_dict(state_dict=state_dict, strict=strict, **kwargs)

    def mark_forward_method(self, method: Union[MethodType, str]) -> None:
        """Mark a method as a 'forward' method to prevent it bypassing the strategy wrapper (e.g., DDP)."""
        if not isinstance(method, (MethodType, str)):
            raise TypeError(f"Expected a method or a string, but got: {type(method).__name__}")
        name = method if isinstance(method, str) else method.__name__
        if name == "forward":
            raise ValueError("You cannot mark the forward method itself as a forward method.")
        if not isinstance(getattr(self._original_module, name, None), MethodType):
            raise AttributeError(
                f"You marked '{name}' as a forward method, but `{type(self._original_module).__name__}.{name}` does not"
                f" exist or is not a method."
            )
        self._forward_methods.add(name)

    def _redirection_through_forward(self, method_name: str) -> Callable:
        assert method_name != "forward"
        original_forward = self._original_module.forward

        def wrapped_forward(*args: Any, **kwargs: Any) -> Any:
            # Unpatch ourselves immediately before calling the method `method_name`
            # because itself may want to call the real `forward`
            self._original_module.forward = original_forward

View on GitHub (pinned to 9fed5c27d2)

Solutions

1. Pass the method name as a string: fabric_module.mark_forward_method('generate')
2. Or pass the bound method retrieved from the wrapped module instance itself, not from the class
3. Unwrap partials/callables and pass the underlying method name instead

Example fix

# before
fabric_module.mark_forward_method(MyModel.generate)
fabric_module.mark_forward_method(functools.partial(m.generate, temperature=0.7))

# after
fabric_module.mark_forward_method('generate')

name = method if isinstance(method, str) else getattr(method, '__name__', None)
assert name and callable(getattr(model, name, None)), 'pass a method name string instead'

from types import MethodType
from typing import Union

def as_method_name(m: Union[MethodType, str]) -> str:
    if isinstance(m, str):
        return m
    if isinstance(m, MethodType):
        return m.__name__
    raise TypeError(f'Expected MethodType or str, got {type(m).__name__}')

When it happens

Trigger: Calling fabric_module.mark_forward_method(...) with something that is not a MethodType or str, e.g. passing the unbound class attribute (Model.generate, a plain function), a functools.partial, a property, or the return value of calling the method (model.generate()).

Common situations: Passing Model.generate instead of the bound instance method; wrapping the method in partial() to bake in kwargs; passing a lambda or a callable object; getting the attribute from a compiled/OptimizedModule wrapper.

Understand the failure class

Background: "Wrong argument type", "must be a string", "expected Array or Prism::Scope": TypeError and ArgumentError when a library receives a value of the wrong type — this error's family across 28 libraries.

Related errors

• Expected a precision plugin, got {plugin}
• `devices` selected with `CPUAccelerator` should be an int >
• `setup_optimizers` requires at least one optimizer as input.
• An optimizer should be passed only once to the `setup_optimi
• The optimizer has references to the model's meta-device para

AI-assisted analysis of Lightning-AI/pytorch-lightning@9fed5c27d2 (2026-08-28). Data as JSON: /api/errors/d7be99e72ebd8669. Report an issue: GitHub.