Merge branch 'keras-team:main' into fix_torch_bincount

examples/keras_io/tensorflow/nlp/end_to_end_mlm_with_bert.py

@@ -167,7 +167,7 @@ def get_vectorize_layer(texts, vocab_size, max_seq, special_tokens=["[MASK]"]):
       texts (list): List of string i.e input texts
       vocab_size (int): vocab size
       max_seq (int): Maximum sequence lenght.
-      special_tokens (list, optional): List of special tokens. Defaults to ['[MASK]'].
+      special_tokens (list, optional): List of special tokens. Defaults to `['[MASK]']`.
 
     Returns:
         layers.Layer: Return TextVectorization Keras Layer

examples/keras_io/tensorflow/vision/shiftvit.py

@@ -315,9 +315,9 @@ class ShiftViTBlock(layers.Layer):
     """A unit ShiftViT Block
 
     Args:
-        shift_pixel (int): The number of pixels to shift. Default to 1.
+        shift_pixel (int): The number of pixels to shift. Defaults to `1`.
         mlp_expand_ratio (int): The ratio with which MLP features are
-            expanded. Default to 2.
+            expanded. Defaults to `2`.
         mlp_dropout_rate (float): The dropout rate used in MLP.
         num_div (int): The number of divisions of the feature map's channel.
             Totally, 4/num_div of channels will be shifted. Defaults to 12.
@@ -493,10 +493,10 @@ class StackedShiftBlocks(layers.Layer):
         stochastic_depth_rate (float): The maximum drop path rate chosen.
         is_merge (boolean): A flag that determines the use of the Patch Merge
             layer after the shift vit blocks.
-        num_div (int): The division of channels of the feature map. Defaults to 12.
-        shift_pixel (int): The number of pixels to shift. Defaults to 1.
+        num_div (int): The division of channels of the feature map. Defaults to `12`.
+        shift_pixel (int): The number of pixels to shift. Defaults to `1`.
         mlp_expand_ratio (int): The ratio with which the initial dense layer of
-            the MLP is expanded Defaults to 2.
+            the MLP is expanded Defaults to `2`.
     """
 
     def __init__(
@@ -578,10 +578,10 @@ class ShiftViTModel(keras.Model):
         mlp_dropout_rate (float): The dropout rate used in the MLP block.
         stochastic_depth_rate (float): The maximum drop rate probability.
         num_div (int): The number of divisions of the channesl of the feature
-            map. Defaults to 12.
-        shift_pixel (int): The number of pixel to shift. Default to 1.
+            map. Defaults to `12`.
+        shift_pixel (int): The number of pixel to shift. Defaults to `1`.
         mlp_expand_ratio (int): The ratio with which the initial mlp dense layer
-            is expanded to. Defaults to 2.
+            is expanded to. Defaults to `2`.
     """
 
     def __init__(

keras_core/applications/efficientnet.py

@@ -192,7 +192,7 @@
     classifier_activation: A `str` or callable. The activation function to use
         on the "top" layer. Ignored unless `include_top=True`. Set
         `classifier_activation=None` to return the logits of the "top" layer.
-        Defaults to 'softmax'.
+        Defaults to `'softmax'`.
         When loading pretrained weights, `classifier_activation` can only
         be `None` or `"softmax"`.
 

keras_core/applications/imagenet_utils.py

@@ -62,7 +62,7 @@
       - torch: will scale pixels between 0 and 1 and then
           will normalize each channel with respect to the
           ImageNet dataset.
-      Defaults to "caffe".
+      Defaults to `"caffe"`.
   """
 
 PREPROCESS_INPUT_DEFAULT_ERROR_DOC = """
@@ -119,7 +119,7 @@ def decode_predictions(preds, top=5):
 
     Args:
         preds: NumPy array encoding a batch of predictions.
-        top: Integer, how many top-guesses to return. Defaults to 5.
+        top: Integer, how many top-guesses to return. Defaults to `5`.
 
     Returns:
         A list of lists of top class prediction tuples

keras_core/applications/resnet.py

@@ -222,8 +222,8 @@ def residual_block_v1(
     Args:
         x: Input tensor.
         filters: No of filters in the bottleneck layer.
-        kernel_size: Kernel size of the bottleneck layer. Defaults to 3
-        stride: Stride of the first layer. Defaults to 1
+        kernel_size: Kernel size of the bottleneck layer. Defaults to `3`.
+        stride: Stride of the first layer. Defaults to `1`.
         conv_shortcut: Use convolution shortcut if `True`, otherwise
             use identity shortcut. Defaults to `True`
         name(optional): Name of the block
@@ -278,7 +278,7 @@ def stack_residual_blocks_v1(x, filters, blocks, stride1=2, name=None):
         x: Input tensor.
         filters: Number of filters in the bottleneck layer in a block.
         blocks: Number of blocks in the stacked blocks.
-        stride1: Stride of the first layer in the first block. Defaults to 2.
+        stride1: Stride of the first layer in the first block. Defaults to `2`.
         name: Stack label.
 
     Returns:
@@ -301,8 +301,8 @@ def residual_block_v2(
     Args:
         x: Input tensor.
         filters: No of filters in the bottleneck layer.
-        kernel_size: Kernel size of the bottleneck layer. Defaults to 3
-        stride: Stride of the first layer. Defaults to 1
+        kernel_size: Kernel size of the bottleneck layer. Defaults to `3`.
+        stride: Stride of the first layer. Defaults to `1`.
         conv_shortcut: Use convolution shortcut if `True`, otherwise
             use identity shortcut. Defaults to `True`
         name(optional): Name of the block
@@ -363,7 +363,7 @@ def stack_residual_blocks_v2(x, filters, blocks, stride1=2, name=None):
         x: Input tensor.
         filters: Number of filters in the bottleneck layer in a block.
         blocks: Number of blocks in the stacked blocks.
-        stride1: Stride of the first layer in the first block. Defaults to 2.
+        stride1: Stride of the first layer in the first block. Defaults to `2`.
         name: Stack label.
 
     Returns:

keras_core/backend/jax/distribution_lib.py

@@ -14,9 +14,9 @@ def list_devices(device_type=None):
     Note that this should return the global devices in a distributed setting.
 
     Args:
-        device_type: string of `"cpu"`, `"gpu"` or `"tpu"`. Default to `gpu` or
-            `tpu` if available when device_type is not provided. Otherwise
-            will return the `cpu` devices.
+        device_type: string of `"cpu"`, `"gpu"` or `"tpu"`. Defaults to `"gpu"`
+            or `"tpu"` if available when device_type is not provided. Otherwise
+            will return the `"cpu"` devices.
 
     Return:
         List of devices that are available for distribute computation.

keras_core/backend/jax/image.py

@@ -58,11 +58,11 @@ def resize(
     "bilinear": 1,
 }
 AFFINE_TRANSFORM_FILL_MODES = {
-    "constant": "grid-constant",
-    "nearest": "nearest",
-    "wrap": "grid-wrap",
-    "mirror": "mirror",
-    "reflect": "reflect",
+    "constant",
+    "nearest",
+    "wrap",
+    "mirror",
+    "reflect",
 }
 
 
@@ -80,11 +80,10 @@ def affine_transform(
             f"{set(AFFINE_TRANSFORM_INTERPOLATIONS.keys())}. Received: "
             f"interpolation={interpolation}"
         )
-    if fill_mode not in AFFINE_TRANSFORM_FILL_MODES.keys():
+    if fill_mode not in AFFINE_TRANSFORM_FILL_MODES:
         raise ValueError(
             "Invalid value for argument `fill_mode`. Expected of one "
-            f"{set(AFFINE_TRANSFORM_FILL_MODES.keys())}. "
-            f"Received: fill_mode={fill_mode}"
+            f"{AFFINE_TRANSFORM_FILL_MODES}. Received: fill_mode={fill_mode}"
         )
 
     transform = convert_to_tensor(transform)

keras_core/backend/numpy/image.py

@@ -57,11 +57,11 @@ def resize(
     "bilinear": 1,
 }
 AFFINE_TRANSFORM_FILL_MODES = {
-    "constant": "grid-constant",
-    "nearest": "nearest",
-    "wrap": "grid-wrap",
-    "mirror": "mirror",
-    "reflect": "reflect",
+    "constant",
+    "nearest",
+    "wrap",
+    "mirror",
+    "reflect",
 }
 
 
@@ -79,11 +79,10 @@ def affine_transform(
             f"{set(AFFINE_TRANSFORM_INTERPOLATIONS.keys())}. Received: "
             f"interpolation={interpolation}"
         )
-    if fill_mode not in AFFINE_TRANSFORM_FILL_MODES.keys():
+    if fill_mode not in AFFINE_TRANSFORM_FILL_MODES:
         raise ValueError(
             "Invalid value for argument `fill_mode`. Expected of one "
-            f"{set(AFFINE_TRANSFORM_FILL_MODES.keys())}. "
-            f"Received: fill_mode={fill_mode}"
+            f"{AFFINE_TRANSFORM_FILL_MODES}. Received: fill_mode={fill_mode}"
         )
 
     transform = convert_to_tensor(transform)
@@ -153,13 +152,12 @@ def affine_transform(
     # apply affine transformation
     affined = np.stack(
         [
-            scipy.ndimage.map_coordinates(
+            map_coordinates(
                 image[i],
                 coordinates[i],
                 order=AFFINE_TRANSFORM_INTERPOLATIONS[interpolation],
-                mode=AFFINE_TRANSFORM_FILL_MODES[fill_mode],
-                cval=fill_value,
-                prefilter=False,
+                fill_mode=fill_mode,
+                fill_value=fill_value,
             )
             for i in range(batch_size)
         ],

keras_core/distribution/distribution_lib.py

@@ -29,7 +29,7 @@ def list_devices(device_type=None):
 
     Args:
         device_type: string, one of `"cpu"`, `"gpu"` or `"tpu"`.
-            Default to `"gpu"` or `"tpu"` if available when
+            Defaults to `"gpu"` or `"tpu"` if available when
             `device_type` is not provided. Otherwise
             will return the `"cpu"` devices.
 
@@ -60,7 +60,7 @@ class DeviceMesh:
             the rank of the `shape`. The `axis_names` will be used to
             match/create the `TensorLayout` when distribute the data and
             variables.
-        devices: Optional list of devices. Default to all the available
+        devices: Optional list of devices. Defaults to all the available
             devices locally from `keras_core.distribution.list_devices()`.
     """
 

keras_core/layers/activations/elu.py

@@ -15,7 +15,7 @@ class ELU(Layer):
     ```
 
     Args:
-        alpha: float, slope of negative section. Defaults to 1.0.
+        alpha: float, slope of negative section. Defaults to `1.0`.
         **kwargs: Base layer keyword arguments, such as `name` and `dtype`.
     """
 

keras_core/layers/activations/leaky_relu.py

@@ -29,7 +29,7 @@ class LeakyReLU(Layer):
 
     Args:
         negative_slope: Float >= 0.0. Negative slope coefficient.
-          Defaults to 0.3.
+          Defaults to `0.3`.
         **kwargs: Base layer keyword arguments, such as
             `name` and `dtype`.
 

keras_core/layers/activations/relu.py

@@ -30,9 +30,10 @@ class ReLU(Layer):
     Args:
         max_value: Float >= 0. Maximum activation value. None means unlimited.
             Defaults to `None`.
-        negative_slope: Float >= 0. Negative slope coefficient. Defaults to 0.0.
+        negative_slope: Float >= 0. Negative slope coefficient.
+            Defaults to `0.0`.
         threshold: Float >= 0. Threshold value for thresholded activation.
-            Defaults to 0.0.
+            Defaults to `0.0`.
         **kwargs: Base layer keyword arguments, such as `name` and `dtype`.
     """
 

keras_core/layers/attention/additive_attention.py

@@ -26,7 +26,7 @@ class AdditiveAttention(Attention):
         use_scale: If `True`, will create a scalar variable to scale the
             attention scores.
         dropout: Float between 0 and 1. Fraction of the units to drop for the
-            attention scores. Defaults to 0.0.
+            attention scores. Defaults to `0.0`.
 
     Call Args:
         inputs: List of the following tensors:

keras_core/layers/attention/attention.py

@@ -26,7 +26,7 @@ class Attention(Layer):
         use_scale: If `True`, will create a scalar variable to scale the
             attention scores.
         dropout: Float between 0 and 1. Fraction of the units to drop for the
-            attention scores. Defaults to 0.0.
+            attention scores. Defaults to `0.0`.
         score_mode: Function to use to compute attention scores, one of
             `{"dot", "concat"}`. `"dot"` refers to the dot product between the
             query and key vectors. `"concat"` refers to the hyperbolic tangent

keras_core/layers/layer.py

@@ -80,7 +80,7 @@ class Layer(BackendLayer, Operation):
         dtype: The dtype of the layer's computations and weights. Can also be a
             `keras_core.mixed_precision.DTypePolicy`,
             which allows the computation and
-            weight dtype to differ. Default of `None` means to use
+            weight dtype to differ. Defaults to `None`. `None` means to use
             `keras_core.mixed_precision.dtype_policy()`,
             which is a `float32` policy unless set to different value
             (via `keras_core.mixed_precision.set_dtype_policy()`).

keras_core/layers/normalization/group_normalization.py

@@ -34,7 +34,7 @@ class GroupNormalization(Layer):
         axis: Integer or List/Tuple. The axis or axes to normalize across.
             Typically, this is the features axis/axes. The left-out axes are
             typically the batch axis/axes. -1 is the last dimension in the
-            input. Defaults to -1.
+            input. Defaults to `-1`.
         epsilon: Small float added to variance to avoid dividing by zero.
             Defaults to 1e-3.
         center: If `True`, add offset of `beta` to normalized tensor.

keras_core/layers/preprocessing/discretization_test.py

@@ -1,6 +1,7 @@
 import os
 
 import numpy as np
+import pytest
 from tensorflow import data as tf_data
 
 from keras_core import backend
@@ -35,6 +36,9 @@ def test_adapt_flow(self):
         output = layer(np.array([[0.0, 0.1, 0.3]]))
         self.assertTrue(output.dtype, "int32")
 
+    @pytest.mark.skipif(
+        backend.backend() in ("torch", "numpy"), reason="TODO: fix me"
+    )
     def test_correctness(self):
         # int mode
         layer = layers.Discretization(

keras_core/layers/preprocessing/integer_lookup.py

@@ -61,15 +61,15 @@ class IntegerLookup(IndexLookup):
             If this value is more than 1, OOV inputs are modulated to
             determine their OOV value.
             If this value is 0, OOV inputs will cause an error when calling
-            the layer. Defaults to 1.
+            the layer. Defaults to `1`.
         mask_token: An integer token that represents masked inputs. When
             `output_mode` is `"int"`, the token is included in vocabulary
             and mapped to index 0. In other output modes,
             the token will not appear in the vocabulary and instances
             of the mask token in the input will be dropped.
             If set to None, no mask term will be added. Defaults to `None`.
         oov_token: Only used when `invert` is `True`. The token to return
-            for OOV indices. Defaults to -1.
+            for OOV indices. Defaults to `-1`.
         vocabulary: Optional. Either an array of integers or a string path to a
             text file. If passing an array, can pass a tuple, list,
             1D NumPy array, or 1D tensor containing the integer vocbulary terms.

keras_core/layers/preprocessing/string_lookup.py

@@ -60,7 +60,7 @@ class StringLookup(IndexLookup):
             If this value is more than 1, OOV inputs are modulated to
             determine their OOV value.
             If this value is 0, OOV inputs will cause an error when calling
-            the layer. Defaults to 1.
+            the layer. Defaults to `1`.
         mask_token: A token that represents masked inputs. When `output_mode` is
             `"int"`, the token is included in vocabulary and mapped to index 0.
             In other output modes, the token will not appear

keras_core/layers/rnn/conv_lstm.py

@@ -34,7 +34,7 @@ class ConvLSTMCell(Layer, DropoutRNNCell):
             `channels_first`. When unspecified, uses
             `image_data_format` value found in your Keras config file at
             `~/.keras/keras.json` (if exists) else 'channels_last'.
-            Defaults to 'channels_last'.
+            Defaults to `'channels_last'`.
         dilation_rate: An integer or tuple/list of n integers, specifying the
             dilation rate to use for dilated convolution.
             Currently, specifying any `dilation_rate` value != 1 is
@@ -405,7 +405,7 @@ class ConvLSTM(RNN):
             When unspecified, uses
             `image_data_format` value found in your Keras config file at
             `~/.keras/keras.json` (if exists) else 'channels_last'.
-            Defaults to 'channels_last'.
+            Defaults to `'channels_last'`.
         dilation_rate: An integer or tuple/list of n integers, specifying
             the dilation rate to use for dilated convolution.
             Currently, specifying any `dilation_rate` value != 1 is