Lightning-AI/pytorch-lightning · error · MisconfigurationException

Cannot add arguments from: {lightning_class}. You should pro

Error message

Cannot add arguments from: {lightning_class}. You should provide either a callable or a subclass of: Trainer, LightningModule, LightningDataModule, or Callback.

What it means

LightningCLI.add_lightning_class_args (also used internally when building the parser) only accepts callables or subclasses of Trainer, LightningModule, LightningDataModule, or Callback. Passing any other class/instance (e.g. a plain nn.Module, a torch optim class directly, or an arbitrary object) raises this MisconfigurationException.

Source

Thrown at src/lightning/pytorch/cli.py:165

        """
        if callable(lightning_class) and not isinstance(lightning_class, type):
            lightning_class = class_from_function(lightning_class)

        if isinstance(lightning_class, type) and issubclass(
            lightning_class, (Trainer, LightningModule, LightningDataModule, Callback)
        ):
            if issubclass(lightning_class, Callback):
                self.callback_keys.append(nested_key)
            if subclass_mode:
                return self.add_subclass_arguments(lightning_class, nested_key, fail_untyped=False, required=required)
            return self.add_class_arguments(
                lightning_class,
                nested_key,
                fail_untyped=False,
                instantiate=not issubclass(lightning_class, Trainer),
                sub_configs=True,
            )
        raise MisconfigurationException(
            f"Cannot add arguments from: {lightning_class}. You should provide either a callable or a subclass of: "
            "Trainer, LightningModule, LightningDataModule, or Callback."
        )

    def add_optimizer_args(
        self,
        optimizer_class: Union[type[Optimizer], tuple[type[Optimizer], ...]] = (Optimizer,),
        nested_key: str = "optimizer",
        link_to: str = "AUTOMATIC",
    ) -> None:
        """Adds arguments from an optimizer class to a nested key of the parser.

        Args:
            optimizer_class: Any subclass of :class:`torch.optim.Optimizer`. Use tuple to allow subclasses.
            nested_key: Name of the nested namespace to store arguments.
            link_to: Dot notation of a parser key to set arguments or AUTOMATIC.

        """

View on GitHub (pinned to 9fed5c27d2)

Solutions

1. Use a supported base: subclass LightningModule/LightningDataModule/Callback/Trainer
2. Use the dedicated helpers add_optimizer_args / add_lr_scheduler_args for optimizers and schedulers
3. Pass the class (callable), not an instance

Example fix

# before
class MyWrapper: ...
parser.add_lightning_class_args(MyWrapper, "wrapper")  # MisconfigurationException
# after
class MyWrapper(Callback): ...
parser.add_lightning_class_args(MyWrapper, "wrapper")

from lightning.pytorch import Trainer, LightningModule, LightningDataModule, Callback

def addable(cls) -> bool:
    return callable(cls) or (isinstance(cls, type) and issubclass(cls, (Trainer, LightningModule, LightningDataModule, Callback)))
assert addable(MyClass)

from lightning.pytorch import Trainer, LightningModule, LightningDataModule, Callback

def is_addable_class(cls) -> bool:
    """True if cls can be passed to LightningCLI.add_lightning_class_args."""
    return callable(cls) or (isinstance(cls, type) and issubclass(cls, (Trainer, LightningModule, LightningDataModule, Callback)))

When it happens

Trigger: cli.add_lightning_class_args(torch.optim.Adam, 'optimizer') (not via add_optimizer_args), add_lightning_class_args(SomePlainClass), or add_core_arguments_to_parser encountering an unregistered type.

Common situations: Extending LightningCLI and trying to add arguments for arbitrary classes instead of the dedicated add_optimizer_args / add_lr_scheduler_args helpers; passing an instance instead of a class.

Understand the failure class

Background: Invalid argument type errors: "must be of type string", "expected X, got Y", and ERR_INVALID_ARG_TYPE explained — this error's family across 15 libraries.

Related errors

• Filter should be a dictionary, given {filter!r}
• Expected `fabric.save(filter=...)` for key {k!r} to be a cal
• Only PyTorch DataLoader are currently supported in `setup_da
• you tried to log {v} which is currently not supported. Try a
• Unsupported op {op!r} of type {type(op).__name__}

AI-assisted analysis of Lightning-AI/pytorch-lightning@9fed5c27d2 (2026-08-28). Data as JSON: /api/errors/c1cf56b1d543f9ac. Report an issue: GitHub.