keras-team/keras · error · ValueError

Layer {self.__class__.__name__} does not take a `factor` arg

Error message

Layer {self.__class__.__name__} does not take a `factor` argument. Received: factor={factor}

What it means

BaseImagePreprocessingLayer subclasses declare _USE_BASE_FACTOR=True only if augmentation strength is controlled by a factor (e.g. RandomContrast, Solarization). If a subclass that does not use factor (like AutoContrast) receives a non-None factor argument, the base __init__ raises this error to catch the misplaced configuration early.

Source

Thrown at keras/src/layers/preprocessing/image_preprocessing/base_image_preprocessing_layer.py:24

    densify_bounding_boxes,
)


class BaseImagePreprocessingLayer(DataLayer):
    _USE_BASE_FACTOR = True
    _FACTOR_BOUNDS = (-1, 1)

    def __init__(
        self, factor=None, bounding_box_format=None, data_format=None, **kwargs
    ):
        super().__init__(**kwargs)
        self.bounding_box_format = bounding_box_format
        self.data_format = backend_config.standardize_data_format(data_format)
        if self._USE_BASE_FACTOR:
            factor = factor or 0.0
            self._set_factor(factor)
        elif factor is not None:
            raise ValueError(
                f"Layer {self.__class__.__name__} does not take "
                f"a `factor` argument. Received: factor={factor}"
            )

    def _set_factor(self, factor):
        error_msg = (
            "The `factor` argument should be a number "
            "(or a list of two numbers) "
            "in the range "
            f"[{self._FACTOR_BOUNDS[0]}, {self._FACTOR_BOUNDS[1]}]. "
            f"Received: factor={factor}"
        )
        if isinstance(factor, (tuple, list)):
            if len(factor) != 2:
                raise ValueError(error_msg)
            if (
                factor[0] > self._FACTOR_BOUNDS[1]
                or factor[1] < self._FACTOR_BOUNDS[0]

View on GitHub (pinned to 7a34a03db6)

Solutions

1. Remove the factor argument from the constructor call of this layer.
2. If you intended factor-based augmentation, use a layer that supports it (e.g. RandomContrast).
3. Audit shared config dicts so layer-specific keys are not splatted via **kwargs into every layer.

Example fix

# before
layer = keras.layers.AutoContrast(value_range=(0,255), factor=0.5)
# after
layer = keras.layers.AutoContrast(value_range=(0,255))

import inspect
if "factor" not in inspect.signature(LayerClass.__init__).parameters:
    kwargs.pop("factor", None)

When it happens

Trigger: keras.layers.AutoContrast(factor=0.5) or any factor-free image layer constructed with a factor kwarg, often leaked from a copied config dict or **kwargs.

Common situations: Sharing a hyperparameter dict across multiple augmentation layers where only some accept factor; upgrading code where a layer signature changed and factor was removed.

Related errors

• {self._VALUE_RANGE_VALIDATION_ERROR}Received: value_range={v
• The `value_range` argument should be a list of two numbers.
• The `factor` argument should be a number (or a list of two n
• Invalid quantization mode. Expected one of {dtype_policies.Q
• Invalid value for argument `output_mode`. Expected one of {a

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/174101925076c1bc. Report an issue: GitHub.