regenerate pkgdown site

Author: t-kalinowski

docs/dev/articles/examples/timeseries/timeseries_anomaly_detection.html

@@ -465,7 +465,7 @@ documentation.
   : Randomly performs the color degeneration operation on given images.
 
 - [`layer_random_color_jitter()`](https://keras3.posit.co/dev/reference/layer_random_color_jitter.md)
-  : Randomly apply brightness, contrast, saturation
+  : Randomly apply brightness, contrast, saturation, and hue.
 
 - [`layer_random_contrast()`](https://keras3.posit.co/dev/reference/layer_random_contrast.md)
   : A preprocessing layer which randomly adjusts contrast during
@@ -513,7 +513,8 @@ documentation.
   : Randomly performs the sharpness operation on given images.
 
 - [`layer_random_shear()`](https://keras3.posit.co/dev/reference/layer_random_shear.md)
-  : A preprocessing layer that randomly applies shear transformations
+  : A preprocessing layer that randomly applies shear transformations to
+  images.
 
 - [`layer_random_translation()`](https://keras3.posit.co/dev/reference/layer_random_translation.md)
   : A preprocessing layer which randomly translates images during
@@ -921,7 +922,8 @@ Functions that are safe to call with both symbolic and eager tensor.
   Return the ceiling of the input, element-wise.
 
 - [`op_cholesky_inverse()`](https://keras3.posit.co/dev/reference/op_cholesky_inverse.md)
-  : Compute a matrix inverse from its Cholesky factor.
+  : Compute the inverse of a symmetric positive-definite matrix from its
+  Cholesky factor.
 
 - [`op_clip()`](https://keras3.posit.co/dev/reference/op_clip.md) : Clip
   (limit) the values in a tensor.
@@ -1783,7 +1785,7 @@ Functions that are safe to call with both symbolic and eager tensor.
   Crop `images` to a specified `height` and `width`.
 
 - [`op_image_extract_patches()`](https://keras3.posit.co/dev/reference/op_image_extract_patches.md)
-  : Extracts patches from the image(s).
+  : Extracts patches from image(s) or volume(s).
 
 - [`op_image_extract_patches_3d()`](https://keras3.posit.co/dev/reference/op_image_extract_patches_3d.md)
   : Extract patches from 3D volumes.
@@ -2244,7 +2246,7 @@ transform your input dataset before training.
   : Applies elastic deformation to the image(s).
 
 - [`op_image_extract_patches()`](https://keras3.posit.co/dev/reference/op_image_extract_patches.md)
-  : Extracts patches from the image(s).
+  : Extracts patches from image(s) or volume(s).
 
 - [`op_image_extract_patches_3d()`](https://keras3.posit.co/dev/reference/op_image_extract_patches_3d.md)
   : Extract patches from 3D volumes.

docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html

@@ -28,7 +28,7 @@ articles:
   understanding_masking_and_padding: understanding_masking_and_padding.html
   writing_a_custom_training_loop_in_tensorflow: writing_a_custom_training_loop_in_tensorflow.html
   writing_your_own_callbacks: writing_your_own_callbacks.html
-last_built: 2026-07-31T18:34Z
+last_built: 2026-08-04T02:57Z
 urls:
   reference: https://keras3.posit.co/reference
   article: https://keras3.posit.co/articles

docs/dev/articles/examples/vision/oxford_pets_image_segmentation.html

@@ -16,8 +16,10 @@ adapt(object, data, ..., batch_size = NULL, steps = NULL)
 
 - data:
 
-  The data to train on. It can be passed either as a `tf.data.Dataset`
-  or as an R array.
+  The data to train on. It can be an R array, a backend-native eager
+  tensor, a batched `tf.data.Dataset`, a Grain dataset, a
+  `keras.utils.PyDataset`, or an iterable of batches such as a list of
+  arrays or a generator. Dataset and iterable inputs must be batched.
 
 - ...:
 

docs/dev/llms.txt

@@ -37,7 +37,9 @@ application_mobilenet_v3_large(
 
   controls the width of the network. This is known as the depth
   multiplier in the MobileNetV3 paper, but the name is kept for
-  consistency with MobileNetV1 in Keras.
+  consistency with MobileNetV1 in Keras. When `weights = "imagenet"`,
+  `alpha` can be `0.75` or `1` for non-minimalistic models, and must be
+  `1` for minimalistic models.
 
   - If `alpha < 1.0`, proportionally decreases the number of filters in
     each layer.