keras-team/keras · error · NotImplementedError

Unknown `fill_mode` {fill_mode}. Expected of one {self._SUPP

Error message

Unknown `fill_mode` {fill_mode}. Expected of one {self._SUPPORTED_FILL_MODE}.

What it means

Raised by RandomTranslation's constructor when `fill_mode` is not one of the supported modes ("constant", "reflect", "wrap", "nearest"). The mode controls how pixels emptied by the translation are filled.

Source

Thrown at keras/src/layers/preprocessing/image_preprocessing/random_translation.py:129

        fill_mode="reflect",
        interpolation="bilinear",
        seed=None,
        fill_value=0.0,
        data_format=None,
        **kwargs,
    ):
        super().__init__(data_format=data_format, **kwargs)
        self.height_factor = height_factor
        self.height_lower, self.height_upper = self._set_factor(
            height_factor, "height_factor"
        )
        self.width_factor = width_factor
        self.width_lower, self.width_upper = self._set_factor(
            width_factor, "width_factor"
        )

        if fill_mode not in self._SUPPORTED_FILL_MODE:
            raise NotImplementedError(
                f"Unknown `fill_mode` {fill_mode}. Expected of one "
                f"{self._SUPPORTED_FILL_MODE}."
            )
        if interpolation not in self._SUPPORTED_INTERPOLATION:
            raise NotImplementedError(
                f"Unknown `interpolation` {interpolation}. Expected of one "
                f"{self._SUPPORTED_INTERPOLATION}."
            )

        self.fill_mode = fill_mode
        self.fill_value = fill_value
        self.interpolation = interpolation
        self.seed = seed
        self.generator = SeedGenerator(seed)
        self.supports_jit = False

    def _set_factor(self, factor, factor_name):
        if isinstance(factor, (tuple, list)):

View on GitHub (pinned to 7a34a03db6)

Solutions

Use one of "constant", "reflect", "wrap", "nearest"
If porting from torchvision's "replicate", use "nearest" or "wrap" as the closest Keras equivalent
For a solid fill color, use fill_mode="constant" with fill_value

Example fix

// before
layer = RandomTranslation(0.1, 0.1, fill_mode="replicate")
// after
layer = RandomTranslation(0.1, 0.1, fill_mode="nearest")

Defensive patterns

Strategy: validation

Validate before calling

SUPPORTED_FILL = {"constant", "reflect", "wrap", "nearest"}
assert fill_mode in SUPPORTED_FILL, f"fill_mode must be one of {SUPPORTED_FILL}"

Type guard

def is_valid_fill_mode(m):
    return m in {"constant", "reflect", "wrap", "nearest"}

Try / catch

try:
    layer = RandomTranslation(0.1, 0.1, fill_mode=fill_mode)
except NotImplementedError as e:
    raise ValueError(f"Unsupported fill_mode {fill_mode!r}; valid: constant, reflect, wrap, nearest") from e

Prevention

Check layer docstrings for supported enum values when porting between torchvision/OpenCV and Keras
Centralize fill-mode mapping in one config-translation module

When it happens

Trigger: Calling RandomTranslation(height_factor=0.1, fill_mode="mirror") or fill_mode="replicate" — strings accepted by other frameworks but not Keras.

Common situations: Porting augmentation configs from torchvision ("reflect", "replicate") or OpenCV (cv2.BORDER_*); misspelling a supported mode.

Related errors

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/ff275c25e52b7a83. Report an issue: GitHub.