facebookresearch/detectron2 · error · TypeError

Unsupported type for argument `serailzie`: {serialize}

Error message

Unsupported type for argument `serailzie`: {serialize}

What it means

Raised by SerializeList's __init__ when the `serialize` argument is neither a bool nor a callable. The class uses `serialize` to decide whether to serialize list elements (and optionally accepts a custom serializer callable); any other type (e.g. a string like 'w' or an int) is rejected.

Source

Thrown at detectron2/data/common.py:229

        self,
        lst: list,
        copy: bool = True,
        serialize: Union[bool, Callable] = True,
    ):
        """
        Args:
            lst (list): a list which contains elements to produce.
            copy (bool): whether to deepcopy the element when producing it,
                so that the result can be modified in place without affecting the
                source in the list.
            serialize (bool or callable): whether to serialize the stroage to other
                backend. If `True`, the default serialize method will be used, if given
                a callable, the callable will be used as serialize method.
        """
        self._lst = lst
        self._copy = copy
        if not isinstance(serialize, (bool, Callable)):
            raise TypeError(f"Unsupported type for argument `serailzie`: {serialize}")
        self._serialize = serialize is not False

        if self._serialize:
            serialize_method = (
                serialize
                if isinstance(serialize, Callable)
                else _DEFAULT_DATASET_FROM_LIST_SERIALIZE_METHOD
            )
            logger.info(f"Serializing the dataset using: {serialize_method}")
            self._lst = serialize_method(self._lst)

    def __len__(self):
        return len(self._lst)

    def __getitem__(self, idx):
        if self._copy and not self._serialize:
            return copy.deepcopy(self._lst[idx])
        else:

View on GitHub (pinned to a2f4a8771a)

Solutions

1. Pass a bool: SerializeList(lst, serialize=True) or serialize=False
2. Or pass a callable used as the serialize method, e.g. SerializeList(lst, serialize=pickle.dumps)
3. If forwarding a config value, validate it before constructing: assert isinstance(cfg.serialize, (bool, Callable))

Example fix

// before
srl = SerializeList(lst, serialize='w')
// after
srl = SerializeList(lst, serialize=True)

from typing import Callable
assert serialize is None or isinstance(serialize, (bool, Callable)), \
    f"serialize must be bool or Callable, got {type(serialize)}"

def is_valid_serialize(x) -> bool:
    return isinstance(x, bool) or callable(x)

When it happens

Trigger: Constructing SerializeList(lst, serialize=<non-bool-non-callable>) e.g. SerializeList(data, serialize='w') or serialize=1; typically hit when passing an open-mode-like string or file object where a bool/callable was expected, or when a variable of the wrong type is forwarded from user config.

Common situations: Copy-pasting a torch.utils.data style API where an argument like a file mode string is accepted; passing a serialization function object that is actually None or a string; version changes where the constructor signature changed.

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

• Unknown segmentation type {type(segmentation)}!
• Cannot match one checkpoint key to multiple keys in the mode
• Class with @configurable must have a 'from_config' classmeth
• {name} must take 'cfg' as the first argument!
• target of LazyCall must be a callable or defines a callable!

AI-assisted analysis of facebookresearch/detectron2@a2f4a8771a (2026-08-27). Data as JSON: /api/errors/150629661a872b91. Report an issue: GitHub.