What causes "Invalid class_mode: {}; expected one of: {}" in keras-team/keras?

Use one of: 'categorical', 'binary', 'sparse', 'input', 'other', or None

keras-team/keras · error · ValueError

Invalid class_mode: {}; expected one of: {}

Error message

Invalid class_mode: {}; expected one of: {}

What it means

DirectoryIterator.__init__ validates class_mode against its allowed set (categorical, binary, sparse, input, other, None). class_mode decides the shape of returned labels, so an unrecognized string fails fast with this ValueError.

Source

Thrown at keras/src/legacy/preprocessing/image.py:445

            data_format = backend.image_data_format()
        if dtype is None:
            dtype = backend.floatx()
        super().set_processing_attrs(
            image_data_generator,
            target_size,
            color_mode,
            data_format,
            save_to_dir,
            save_prefix,
            save_format,
            subset,
            interpolation,
            keep_aspect_ratio,
        )
        self.directory = directory
        self.classes = classes
        if class_mode not in self.allowed_class_modes:
            raise ValueError(
                "Invalid class_mode: {}; expected one of: {}".format(
                    class_mode, self.allowed_class_modes
                )
            )
        self.class_mode = class_mode
        self.dtype = dtype
        # First, count the number of samples and classes.
        self.samples = 0

        if not classes:
            classes = []
            for subdir in sorted(os.listdir(directory)):
                if os.path.isdir(os.path.join(directory, subdir)):
                    classes.append(subdir)
        self.num_classes = len(classes)
        self.class_indices = dict(zip(classes, range(len(classes))))

        pool = multiprocessing.pool.ThreadPool()

View on GitHub (pinned to 7a34a03db6)

Solutions

Use one of: 'categorical', 'binary', 'sparse', 'input', 'other', or None
For no labels use class_mode=None (Python None), not the string 'none'
For integer class indices use 'sparse'; one-hot uses 'categorical'

Example fix

# before
it = gen.flow_from_directory(dir, class_mode='multiclass')

# after
it = gen.flow_from_directory(dir, class_mode='categorical')

Defensive patterns

Strategy: validation

Validate before calling

allowed = {'categorical', 'binary', 'sparse', 'input', 'other', None}
assert class_mode in allowed, f'bad class_mode: {class_mode}'

Type guard

def is_class_mode(v) -> bool: return v in {'categorical', 'binary', 'sparse', 'input', 'other'} or v is None

Try / catch

try:
    it = gen.flow_from_directory(d, class_mode=class_mode)
except ValueError as e:
    if 'Invalid class_mode' in str(e):
        class_mode = {'one_hot': 'categorical', 'multiclass': 'sparse'}.get(class_mode, class_mode)
    else:
        raise

Prevention

Use None (object) for 'no labels', never the string 'none'
sparse = integer labels, categorical = one-hot; pick deliberately per loss function

When it happens

Trigger: flow_from_directory(..., class_mode='multiclass'), 'one_hot', 'binary ' with a trailing space, wrong casing, or the string 'none' instead of the Python object None.

Common situations: Confusing sparse (integer labels) with categorical (one-hot), using class_mode='none' (string) where None is required, copy-pasting mode names from other APIs.

Understand the failure class

Background: Invalid enum value errors: "Unknown type", "Invalid scope", "must be one of" — when a string is not on the library's allowed list — this error's family across 23 libraries.

Related errors

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/e8c7c09bbd7302fb. Report an issue: GitHub.