keras-nightly 3.14.0.dev2026012204__py3-none-any.whl → 3.14.0.dev2026012404__py3-none-any.whl

keras/src/layers/preprocessing/image_preprocessing/random_erasing.py

@@ -2,6 +2,9 @@ from keras.src.api_export import keras_export
 from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
     BaseImagePreprocessingLayer,
 )
+from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
+    base_image_preprocessing_color_example,
+)
 from keras.src.random import SeedGenerator
 
 
@@ -41,6 +44,10 @@ class RandomErasing(BaseImagePreprocessingLayer):
             typically either `[0, 1]` or `[0, 255]` depending on how your
             preprocessing pipeline is set up.
         seed: Integer. Used to create a random seed.
+
+    Example:
+
+    {{base_image_preprocessing_color_example}}
     """
 
     _USE_BASE_FACTOR = False
@@ -326,3 +333,11 @@ class RandomErasing(BaseImagePreprocessingLayer):
         }
         base_config = super().get_config()
         return {**base_config, **config}
+
+
+RandomErasing.__doc__ = RandomErasing.__doc__.replace(
+    "{{base_image_preprocessing_color_example}}",
+    base_image_preprocessing_color_example.replace(
+        "{LayerName}", "RandomErasing"
+    ),
+)

keras/src/layers/preprocessing/image_preprocessing/random_flip.py

@@ -2,6 +2,9 @@ from keras.src.api_export import keras_export
 from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
     BaseImagePreprocessingLayer,
 )
+from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
+    base_image_preprocessing_transform_example,
+)
 from keras.src.layers.preprocessing.image_preprocessing.bounding_boxes.converters import (  # noqa: E501
     clip_to_image_size,
 )
@@ -46,6 +49,10 @@ class RandomFlip(BaseImagePreprocessingLayer):
         seed: Integer. Used to create a random seed.
         **kwargs: Base layer keyword arguments, such as
             `name` and `dtype`.
+
+    Example:
+
+    {{base_image_preprocessing_transform_example}}
     """
 
     _USE_BASE_FACTOR = False
@@ -234,3 +241,11 @@ class RandomFlip(BaseImagePreprocessingLayer):
             }
         )
         return config
+
+
+RandomFlip.__doc__ = RandomFlip.__doc__.replace(
+    "{{base_image_preprocessing_transform_example}}",
+    base_image_preprocessing_transform_example.replace(
+        "{LayerName}", "RandomFlip"
+    ),
+)

keras/src/layers/preprocessing/image_preprocessing/random_gaussian_blur.py

@@ -2,6 +2,9 @@ from keras.src.api_export import keras_export
 from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
     BaseImagePreprocessingLayer,
 )
+from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
+    base_image_preprocessing_color_example,
+)
 from keras.src.random import SeedGenerator
 
 
@@ -35,6 +38,10 @@ class RandomGaussianBlur(BaseImagePreprocessingLayer):
             typically either `[0, 1]` or `[0, 255]` depending on how your
             preprocessing pipeline is set up.
         seed: Integer. Used to create a random seed.
+
+    Example:
+
+    {{base_image_preprocessing_color_example}}
     """
 
     _USE_BASE_FACTOR = False
@@ -218,3 +225,11 @@ class RandomGaussianBlur(BaseImagePreprocessingLayer):
             }
         )
         return config
+
+
+RandomGaussianBlur.__doc__ = RandomGaussianBlur.__doc__.replace(
+    "{{base_image_preprocessing_color_example}}",
+    base_image_preprocessing_color_example.replace(
+        "{LayerName}", "RandomGaussianBlur"
+    ),
+)

keras/src/layers/preprocessing/image_preprocessing/random_grayscale.py

@@ -3,6 +3,9 @@ from keras.src.api_export import keras_export
 from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
     BaseImagePreprocessingLayer,
 )
+from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
+    base_image_preprocessing_color_example,
+)
 
 
 @keras_export("keras.layers.RandomGrayscale")
@@ -43,6 +46,10 @@ class RandomGrayscale(BaseImagePreprocessingLayer):
         Same as input shape. The output maintains the same number of channels
         as the input, even for grayscale-converted images where all channels
         will have the same value.
+
+    Example:
+
+    {{base_image_preprocessing_color_example}}
     """
 
     def __init__(self, factor=0.5, data_format=None, seed=None, **kwargs):
@@ -115,3 +122,11 @@ class RandomGrayscale(BaseImagePreprocessingLayer):
         config = super().get_config()
         config.update({"factor": self.factor})
         return config
+
+
+RandomGrayscale.__doc__ = RandomGrayscale.__doc__.replace(
+    "{{base_image_preprocessing_color_example}}",
+    base_image_preprocessing_color_example.replace(
+        "{LayerName}", "RandomGrayscale"
+    ),
+)

keras/src/layers/preprocessing/image_preprocessing/random_invert.py

@@ -2,6 +2,9 @@ from keras.src.api_export import keras_export
 from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
     BaseImagePreprocessingLayer,
 )
+from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
+    base_image_preprocessing_color_example,
+)
 
 
 @keras_export("keras.layers.RandomInvert")
@@ -30,6 +33,10 @@ class RandomInvert(BaseImagePreprocessingLayer):
             represents the upper bound. Images passed to the layer should have
             values within `value_range`. Defaults to `(0, 255)`.
         seed: Integer. Used to create a random seed.
+
+    Example:
+
+    {{base_image_preprocessing_color_example}}
     """
 
     _USE_BASE_FACTOR = False
@@ -127,3 +134,11 @@ class RandomInvert(BaseImagePreprocessingLayer):
         }
         base_config = super().get_config()
         return {**base_config, **config}
+
+
+RandomInvert.__doc__ = RandomInvert.__doc__.replace(
+    "{{base_image_preprocessing_color_example}}",
+    base_image_preprocessing_color_example.replace(
+        "{LayerName}", "RandomInvert"
+    ),
+)

keras/src/layers/preprocessing/image_preprocessing/random_perspective.py

@@ -2,6 +2,9 @@ from keras.src.api_export import keras_export
 from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
     BaseImagePreprocessingLayer,
 )
+from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
+    base_image_preprocessing_transform_example,
+)
 from keras.src.layers.preprocessing.image_preprocessing.bounding_boxes.converters import (  # noqa: E501
     clip_to_image_size,
 )
@@ -43,6 +46,9 @@ class RandomPerspective(BaseImagePreprocessingLayer):
             boundaries when `fill_mode="constant"`.
         seed: Integer. Used to create a random seed.
 
+    Example:
+
+    {{base_image_preprocessing_transform_example}}
     """
 
     _USE_BASE_FACTOR = False
@@ -337,3 +343,11 @@ class RandomPerspective(BaseImagePreprocessingLayer):
             "seed": self.seed,
         }
         return {**base_config, **config}
+
+
+RandomPerspective.__doc__ = RandomPerspective.__doc__.replace(
+    "{{base_image_preprocessing_transform_example}}",
+    base_image_preprocessing_transform_example.replace(
+        "{LayerName}", "RandomPerspective"
+    ),
+)

keras/src/layers/preprocessing/image_preprocessing/random_posterization.py

@@ -2,6 +2,9 @@ from keras.src.api_export import keras_export
 from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
     BaseImagePreprocessingLayer,
 )
+from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
+    base_image_preprocessing_color_example,
+)
 
 
 @keras_export("keras.layers.RandomPosterization")
@@ -22,6 +25,10 @@ class RandomPosterization(BaseImagePreprocessingLayer):
             values within `value_range`. Defaults to `(0, 255)`.
         factor: integer, the number of bits to keep for each channel. Must be a
             value between 1-8.
+
+    Example:
+
+    {{base_image_preprocessing_color_example}}
     """
 
     _USE_BASE_FACTOR = False
@@ -152,3 +159,11 @@ class RandomPosterization(BaseImagePreprocessingLayer):
 
     def compute_output_shape(self, input_shape):
         return input_shape
+
+
+RandomPosterization.__doc__ = RandomPosterization.__doc__.replace(
+    "{{base_image_preprocessing_color_example}}",
+    base_image_preprocessing_color_example.replace(
+        "{LayerName}", "RandomPosterization"
+    ),
+)

keras/src/layers/preprocessing/image_preprocessing/random_rotation.py

@@ -2,6 +2,9 @@ from keras.src.api_export import keras_export
 from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
     BaseImagePreprocessingLayer,
 )
+from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
+    base_image_preprocessing_transform_example,
+)
 from keras.src.layers.preprocessing.image_preprocessing.bounding_boxes import (
     converters,
 )
@@ -75,6 +78,10 @@ class RandomRotation(BaseImagePreprocessingLayer):
             `image_data_format` value found in your Keras config file at
             `~/.keras/keras.json`. If you never set it, then it will be
             `"channels_last"`.
+
+    Example:
+
+    {{base_image_preprocessing_transform_example}}
     """
 
     _SUPPORTED_FILL_MODE = ("reflect", "wrap", "constant", "nearest")
@@ -247,3 +254,11 @@ class RandomRotation(BaseImagePreprocessingLayer):
         }
         base_config = super().get_config()
         return {**base_config, **config}
+
+
+RandomRotation.__doc__ = RandomRotation.__doc__.replace(
+    "{{base_image_preprocessing_transform_example}}",
+    base_image_preprocessing_transform_example.replace(
+        "{LayerName}", "RandomRotation"
+    ),
+)

keras/src/layers/preprocessing/image_preprocessing/random_sharpness.py

@@ -2,6 +2,9 @@ from keras.src.api_export import keras_export
 from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
     BaseImagePreprocessingLayer,
 )
+from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
+    base_image_preprocessing_color_example,
+)
 from keras.src.random import SeedGenerator
 
 
@@ -32,6 +35,10 @@ class RandomSharpness(BaseImagePreprocessingLayer):
             typically either `[0, 1]` or `[0, 255]` depending on how your
             preprocessing pipeline is set up.
         seed: Integer. Used to create a random seed.
+
+    Example:
+
+    {{base_image_preprocessing_color_example}}
     """
 
     _USE_BASE_FACTOR = False
@@ -169,3 +176,11 @@ class RandomSharpness(BaseImagePreprocessingLayer):
 
     def compute_output_shape(self, input_shape):
         return input_shape
+
+
+RandomSharpness.__doc__ = RandomSharpness.__doc__.replace(
+    "{{base_image_preprocessing_color_example}}",
+    base_image_preprocessing_color_example.replace(
+        "{LayerName}", "RandomSharpness"
+    ),
+)

keras/src/layers/preprocessing/image_preprocessing/random_shear.py

@@ -2,6 +2,9 @@ from keras.src.api_export import keras_export
 from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
     BaseImagePreprocessingLayer,
 )
+from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
+    base_image_preprocessing_transform_example,
+)
 from keras.src.layers.preprocessing.image_preprocessing.bounding_boxes.converters import (  # noqa: E501
     clip_to_image_size,
 )
@@ -61,6 +64,10 @@ class RandomShear(BaseImagePreprocessingLayer):
         fill_value: A float representing the value to be filled outside the
             boundaries when `fill_mode="constant"`.
         seed: Integer. Used to create a random seed.
+
+    Example:
+
+    {{base_image_preprocessing_transform_example}}
     """
 
     _USE_BASE_FACTOR = False
@@ -402,3 +409,11 @@ class RandomShear(BaseImagePreprocessingLayer):
 
     def compute_output_shape(self, input_shape):
         return input_shape
+
+
+RandomShear.__doc__ = RandomShear.__doc__.replace(
+    "{{base_image_preprocessing_transform_example}}",
+    base_image_preprocessing_transform_example.replace(
+        "{LayerName}", "RandomShear"
+    ),
+)

keras/src/layers/preprocessing/image_preprocessing/random_translation.py

@@ -2,6 +2,9 @@ from keras.src.api_export import keras_export
 from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
     BaseImagePreprocessingLayer,
 )
+from keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer import (  # noqa: E501
+    base_image_preprocessing_transform_example,
+)
 from keras.src.layers.preprocessing.image_preprocessing.bounding_boxes.converters import (  # noqa: E501
     clip_to_image_size,
 )
@@ -87,6 +90,10 @@ class RandomTranslation(BaseImagePreprocessingLayer):
             `~/.keras/keras.json`. If you never set it, then it will be
             `"channels_last"`.
         **kwargs: Base layer keyword arguments, such as `name` and `dtype`.
+
+    Example:
+
+    {{base_image_preprocessing_transform_example}}
     """
 
     _USE_BASE_FACTOR = False
@@ -382,3 +389,11 @@ class RandomTranslation(BaseImagePreprocessingLayer):
             "data_format": self.data_format,
         }
         return {**base_config, **config}
+
+
+RandomTranslation.__doc__ = RandomTranslation.__doc__.replace(
+    "{{base_image_preprocessing_transform_example}}",
+    base_image_preprocessing_transform_example.replace(
+        "{LayerName}", "RandomTranslation"
+    ),
+)

keras/src/layers/rnn/time_distributed.py

@@ -71,37 +71,43 @@ class TimeDistributed(Wrapper):
         super().build(child_input_shape)
 
     def call(self, inputs, training=None, mask=None):
-        input_shape = ops.shape(inputs)
-        mask_shape = None if mask is None else ops.shape(mask)
-        batch_size = input_shape[0]
-        timesteps = input_shape[1]
-
-        # For TF backend with graph mode and `partial_batch_size`, skip
-        # evaluation of `batch_size` as it can be a `strided_slice` and
-        # not a constant.
-        if backend.backend() == "tensorflow":
-            from keras.src.utils.module_utils import tensorflow as tf
-
-            if (
-                not tf.executing_eagerly
-                and mask_shape is not None
-                and mask_shape[1:2] != (timesteps,)
-            ):
+        # Validate mask shape using static shape info when available
+        if mask is not None:
+            mask_shape = mask.shape
+            input_shape = inputs.shape
+
+            # Check if mask has at least 2 dimensions (batch and timesteps)
+            if len(mask_shape) < 2:
                 raise ValueError(
-                    "`TimeDistributed` Layer should be passed a `mask` of "
-                    f"shape ({batch_size}, {timesteps}, ...), "
-                    f"received: mask.shape={mask_shape}"
+                    "The `mask` passed to the `TimeDistributed` layer must be "
+                    "at least 2D (e.g., `(batch_size, timesteps)`), but it has "
+                    f"{len(mask_shape)} dimension(s) with shape {mask_shape}."
                 )
-        elif mask_shape is not None and mask_shape[:2] != (
-            batch_size,
-            timesteps,
-        ):
-            raise ValueError(
-                "`TimeDistributed` Layer should be passed a `mask` of "
-                f"shape ({batch_size}, {timesteps}, ...), "
-                f"received: mask.shape={mask_shape}"
+
+            # Check batch size and timesteps dimensions match
+            batch_mismatch = (
+                input_shape[0] is not None
+                and mask_shape[0] is not None
+                and input_shape[0] != mask_shape[0]
+            )
+            time_mismatch = (
+                input_shape[1] is not None
+                and mask_shape[1] is not None
+                and input_shape[1] != mask_shape[1]
             )
 
+            if batch_mismatch or time_mismatch:
+                raise ValueError(
+                    "The `mask` passed to the `TimeDistributed` layer has a "
+                    f"shape {mask_shape} that is incompatible with the input "
+                    f"shape {input_shape}. The first two dimensions of the "
+                    "mask (batch size and timesteps) must match the input's "
+                    "first two dimensions. Expected mask shape prefix: "
+                    f"({input_shape[0]}, {input_shape[1]})."
+                )
+
+        input_shape = ops.shape(inputs)
+
         def time_distributed_transpose(data):
             """Swaps the timestep and batch dimensions of a tensor."""
             axes = [1, 0, *range(2, len(data.shape))]
@@ -129,5 +135,7 @@ class TimeDistributed(Wrapper):
 
         # Implementation #2: use backend.vectorized_map.
 
-        outputs = backend.vectorized_map(step_function, ops.arange(timesteps))
+        outputs = backend.vectorized_map(
+            step_function, ops.arange(input_shape[0])
+        )
         return time_distributed_transpose(outputs)

keras/src/ops/core.py

@@ -277,7 +277,7 @@ def associative_scan(f, elems, reverse=False, axis=0):
     [0, 1, 3, 6, 10]
 
     >>> sum_fn = lambda x, y: [x[0] + y[0], x[1] + y[1], x[2] + y[2]]
-    >>> xs = [keras.ops.array([[1, 2]]) for _ in range(3)]
+    >>> xs = [keras.ops.array([1, 2]) for _ in range(3)]
     >>> ys = keras.ops.associative_scan(sum_fn, xs, axis=0)
     >>> ys
     [[1, 3], [1, 3], [1, 3]]

keras/src/ops/numpy.py

@@ -5064,6 +5064,119 @@ def moveaxis(x, source, destination):
     return backend.numpy.moveaxis(x, source=source, destination=destination)
 
 
+class Nanmax(Operation):
+    def __init__(self, axis=None, keepdims=False, *, name=None):
+        super().__init__(name=name)
+        self.axis = axis
+        self.keepdims = keepdims
+
+    def call(self, x):
+        return backend.numpy.nanmax(x, axis=self.axis, keepdims=self.keepdims)
+
+    def compute_output_spec(self, x):
+        dtype = dtypes.result_type(getattr(x, "dtype", backend.floatx()))
+
+        if backend.backend() == "torch" and dtype == "uint32":
+            dtype = "int32"
+
+        return KerasTensor(
+            reduce_shape(x.shape, axis=self.axis, keepdims=self.keepdims),
+            dtype=dtype,
+        )
+
+
+@keras_export(["keras.ops.nanmax", "keras.ops.numpy.nanmax"])
+def nanmax(x, axis=None, keepdims=False):
+    """Maximum of a tensor over the given axes, ignoring NaNs.
+
+    Args:
+        x: Input tensor.
+        axis: Axis or axes along which the maximum is computed.
+            The default is to compute the maximum of the flattened tensor.
+        keepdims: If this is set to `True`, the axes which are reduced are left
+            in the result as dimensions with size one. Defaults
+            to `False`.
+
+    Returns:
+        Output tensor containing the maximum, with NaN values ignored. If all
+        values along a reduced axis are NaN, the result is NaN.
+
+    Examples:
+    >>> import numpy as np
+    >>> from keras import ops
+    >>> x = np.array([[1.0, np.nan, 3.0],
+    ...               [np.nan, 2.0, 1.0]])
+    >>> ops.nanmax(x)
+    3.0
+
+    >>> ops.nanmax(x, axis=1)
+    array([3., 2.])
+
+    >>> ops.nanmax(x, axis=1, keepdims=True)
+    array([[3.],
+           [2.]])
+    """
+    if any_symbolic_tensors((x,)):
+        return Nanmax(axis=axis, keepdims=keepdims).symbolic_call(x)
+    return backend.numpy.nanmax(x, axis=axis, keepdims=keepdims)
+
+
+class Nanmin(Operation):
+    def __init__(self, axis=None, keepdims=False, *, name=None):
+        super().__init__(name=name)
+        self.axis = axis
+        self.keepdims = keepdims
+
+    def call(self, x):
+        return backend.numpy.nanmin(x, axis=self.axis, keepdims=self.keepdims)
+
+    def compute_output_spec(self, x):
+        dtype = dtypes.result_type(getattr(x, "dtype", backend.floatx()))
+
+        if backend.backend() == "torch" and dtype == "uint32":
+            dtype = "int32"
+
+        return KerasTensor(
+            reduce_shape(x.shape, axis=self.axis, keepdims=self.keepdims),
+            dtype=dtype,
+        )
+
+
+@keras_export(["keras.ops.nanmin", "keras.ops.numpy.nanmin"])
+def nanmin(x, axis=None, keepdims=False):
+    """Minimum of a tensor over the given axes, ignoring NaNs.
+
+    Args:
+        x: Input tensor.
+        axis: Axis or axes along which the minimum is computed.
+            The default is to compute the minimum of the flattened tensor.
+        keepdims: If this is set to `True`, the axes which are reduced are left
+            in the result as dimensions with size one.
+
+    Returns:
+        Output tensor containing the minimum, with NaN values ignored. If all
+        values along a reduced axis are NaN, the result is NaN.
+
+    Examples:
+    >>> import numpy as np
+    >>> from keras import ops
+    >>> x = np.array([[1.0, np.nan, 3.0],
+    ...               [np.nan, 2.0, 1.0]])
+    >>> ops.nanmin(x)
+    1.0
+
+    >>> ops.nanmin(x, axis=1)
+    array([1., 1.])
+
+    >>> ops.nanmin(x, axis=1, keepdims=True)
+    array([[1.],
+           [1.]])
+    """
+    if any_symbolic_tensors((x,)):
+        return Nanmin(axis=axis, keepdims=keepdims).symbolic_call(x)
+    return backend.numpy.nanmin(x, axis=axis, keepdims=keepdims)
+
+
 class Nansum(Operation):
     def __init__(self, axis=None, keepdims=False, *, name=None):
         super().__init__(name=name)

keras/src/regularizers/regularizers.py

@@ -45,8 +45,8 @@ class Regularizer:
     >>> out = layer(tensor)
 
     >>> # The kernel regularization term is 0.25
-    >>> # The activity regularization term (after dividing by the batch size)
-    >>> # is 5
+    >>> # The activity regularization term (after dividing by batch size of 5)
+    >>> # is 5.0
     >>> ops.sum(layer.losses)
     5.25
 

keras/src/version.py

@@ -1,7 +1,7 @@
 from keras.src.api_export import keras_export
 
 # Unique source of truth for the version number.
-__version__ = "3.14.0.dev2026012204"
+__version__ = "3.14.0.dev2026012404"
 
 
 @keras_export("keras.version")

{keras_nightly-3.14.0.dev2026012204.dist-info → keras_nightly-3.14.0.dev2026012404.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: keras-nightly
-Version: 3.14.0.dev2026012204
+Version: 3.14.0.dev2026012404
 Summary: Multi-backend Keras
 Author-email: Keras team <keras-users@googlegroups.com>
 License: Apache License 2.0