Fix various links in Keras docstrings.

PiperOrigin-RevId: 306769967
Change-Id: I21491ccbe9679f34bbd4b28aa27c46c6ba01ac4e

tensorflow/python/keras/backend_config.py

@@ -86,8 +86,8 @@ def set_floatx(value):
   likely cause numeric stability issues. Instead, mixed precision, which is
   using a mix of float16 and float32, can be used by calling
   `tf.keras.mixed_precision.experimental.set_policy('mixed_float16')`. See the
-  [mixed precision
+  [mixed precision guide](
-  guide](https://www.tensorflow.org/guide/keras/mixed_precision) for details.
+    https://www.tensorflow.org/guide/keras/mixed_precision) for details.
 
   Arguments:
       value: String; `'float16'`, `'float32'`, or `'float64'`.

tensorflow/python/keras/callbacks_v1.py

@@ -83,15 +83,16 @@ class TensorBoard(callbacks.TensorBoard):
       embeddings_layer_names: a list of names of layers to keep eye on. If None
         or empty list all the embedding layer will be watched.
       embeddings_metadata: a dictionary which maps layer name to a file name in
-        which metadata for this embedding layer is saved. See the
+        which metadata for this embedding layer is saved.
-          [details](https://www.tensorflow.org/how_tos/embedding_viz/#metadata_optional)
+          [Here are details](
+            https://www.tensorflow.org/how_tos/embedding_viz/#metadata_optional)
             about metadata files format. In case if the same metadata file is
             used for all embedding layers, string can be passed.
       embeddings_data: data to be embedded at layers specified in
         `embeddings_layer_names`. Numpy array (if the model has a single input)
-        or list of Numpy arrays (if the model has multiple inputs). Learn [more
+        or list of Numpy arrays (if the model has multiple inputs). Learn more
-        about
+        about embeddings [in this guide](
-            embeddings](https://www.tensorflow.org/programmers_guide/embedding)
+          https://www.tensorflow.org/programmers_guide/embedding).
       update_freq: `'batch'` or `'epoch'` or integer. When using `'batch'`,
         writes the losses and metrics to TensorBoard after each batch. The same
         applies for `'epoch'`. If using an integer, let's say `1000`, the

tensorflow/python/keras/engine/input_layer.py

@@ -80,7 +80,7 @@ class InputLayer(base_layer.Layer):
       ragged: Boolean, whether the placeholder created is meant to be ragged.
           In this case, values of 'None' in the 'shape' argument represent
           ragged dimensions. For more information about RaggedTensors, see
-          https://www.tensorflow.org/guide/ragged_tensors.
+          [this guide](https://www.tensorflow.org/guide/ragged_tensors).
           Default to False.
       name: Optional name of the layer (string).
   """
@@ -231,7 +231,7 @@ def Input(  # pylint: disable=invalid-name
           ragged. Only one of 'ragged' and 'sparse' can be True. In this case,
           values of 'None' in the 'shape' argument represent ragged dimensions.
           For more information about RaggedTensors, see
-          https://www.tensorflow.org/guide/ragged_tensors.
+          [this guide](https://www.tensorflow.org/guide/ragged_tensors).
       **kwargs: deprecated arguments support. Supports `batch_shape` and
           `batch_input_shape`.
 

tensorflow/python/keras/engine/training.py

@@ -163,9 +163,6 @@ class Model(network.Network, version_utils.ModelVersionSelector):
   Once the model is created, you can config the model with losses and metrics
   with `model.compile()`, train the model with `model.fit()`, or use the model
   to do prediction with `model.predict()`.
-
-  Checkout [guide](https://www.tensorflow.org/guide/keras/overview) for
-  additional details.
   """
   _TF_MODULE_IGNORED_PROPERTIES = frozenset(
       itertools.chain(('_train_counter', '_test_counter', '_predict_counter',

tensorflow/python/keras/estimator/__init__.py

@@ -46,8 +46,8 @@ def model_to_estimator(
   model to an Estimator for use with downstream systems.
 
   For usage example, please see:
-  [Creating estimators from Keras
+  [Creating estimators from Keras Models](
-  Models](https://www.tensorflow.org/guide/estimators#creating_estimators_from_keras_models).
+    https://www.tensorflow.org/guide/estimators#creating_estimators_from_keras_models).
 
   Sample Weights:
   Estimators returned by `model_to_estimator` are configured so that they can
@@ -144,8 +144,8 @@ def model_to_estimator_v2(keras_model=None,
   model to an Estimator for use with downstream systems.
 
   For usage example, please see:
-  [Creating estimators from Keras
+  [Creating estimators from Keras Models](
-  Models](https://www.tensorflow.org/guide/estimators#creating_estimators_from_keras_models).
+    https://www.tensorflow.org/guide/estimators#creating_estimators_from_keras_models).
 
   Sample Weights:
   Estimators returned by `model_to_estimator` are configured so that they can

tensorflow/python/keras/layers/convolutional_recurrent.py

@@ -458,8 +458,8 @@ class ConvLSTM2DCell(DropoutRNNCellMixin, Layer):
     unit_forget_bias: Boolean.
       If True, add 1 to the bias of the forget gate at initialization.
       Use in combination with `bias_initializer="zeros"`.
-      This is recommended in [Jozefowicz et al.]
+      This is recommended in [Jozefowicz et al., 2015](
-      (http://www.jmlr.org/proceedings/papers/v37/jozefowicz15.pdf)
+        http://www.jmlr.org/proceedings/papers/v37/jozefowicz15.pdf)
     kernel_regularizer: Regularizer function applied to
       the `kernel` weights matrix.
     recurrent_regularizer: Regularizer function applied to
@@ -739,8 +739,8 @@ class ConvLSTM2D(ConvRNN2D):
     unit_forget_bias: Boolean.
       If True, add 1 to the bias of the forget gate at initialization.
       Use in combination with `bias_initializer="zeros"`.
-      This is recommended in [Jozefowicz et al.]
+      This is recommended in [Jozefowicz et al., 2015](
-      (http://www.jmlr.org/proceedings/papers/v37/jozefowicz15.pdf)
+        http://www.jmlr.org/proceedings/papers/v37/jozefowicz15.pdf)
     kernel_regularizer: Regularizer function applied to
       the `kernel` weights matrix.
     recurrent_regularizer: Regularizer function applied to
@@ -807,10 +807,9 @@ class ConvLSTM2D(ConvRNN2D):
     ValueError: in case of invalid constructor arguments.
 
   References:
-    - [Convolutional LSTM Network: A Machine Learning Approach for
+    - [Shi et al., 2015](http://arxiv.org/abs/1506.04214v1)
-    Precipitation Nowcasting](http://arxiv.org/abs/1506.04214v1)
+    (the current implementation does not include the feedback loop on the
-    The current implementation does not include the feedback loop on the
+    cells output).
-    cells output.
   """
 
   def __init__(self,

tensorflow/python/keras/layers/core.py

@@ -92,8 +92,8 @@ class Masking(Layer):
   # The time step 3 and 5 will be skipped from LSTM calculation.
   ```
 
-  See [the masking and padding
+  See [the masking and padding guide](
-  guide](https://www.tensorflow.org/guide/keras/masking_and_padding)
+    https://www.tensorflow.org/guide/keras/masking_and_padding)
   for more details.
   """
 
@@ -734,15 +734,15 @@ class Lambda(Layer):
   The `Lambda` layer exists so that arbitrary TensorFlow functions
   can be used when constructing `Sequential` and Functional API
   models. `Lambda` layers are best suited for simple operations or
-  quick experimentation. For more advanced usecases, follow 
+  quick experimentation. For more advanced usecases, follow
   [this guide](https://www.tensorflow.org/guide/keras/custom_layers_and_models)
-  for subclassing `tf.keras.layers.Layer`. 
+  for subclassing `tf.keras.layers.Layer`.
-  
+
-  The main reason to subclass `tf.keras.layers.Layer` instead of using a 
+  The main reason to subclass `tf.keras.layers.Layer` instead of using a
-  `Lambda` layer is saving and inspecting a Model. `Lambda` layers 
+  `Lambda` layer is saving and inspecting a Model. `Lambda` layers
-  are saved by serializing the Python bytecode, whereas subclassed 
+  are saved by serializing the Python bytecode, whereas subclassed
-  Layers can be saved via overriding their `get_config` method. Overriding 
+  Layers can be saved via overriding their `get_config` method. Overriding
-  `get_config` improves the portability of Models. Models that rely on 
+  `get_config` improves the portability of Models. Models that rely on
   subclassed Layers are also often easier to visualize and reason about.
 
   Examples:

tensorflow/python/keras/layers/normalization.py

@@ -88,8 +88,8 @@ class BatchNormalizationBase(Layer):
     gamma_regularizer: Optional regularizer for the gamma weight.
     beta_constraint: Optional constraint for the beta weight.
     gamma_constraint: Optional constraint for the gamma weight.
-    renorm: Whether to use Batch Renormalization
+    renorm: Whether to use [Batch Renormalization](
-      (https://arxiv.org/abs/1702.03275). This adds extra variables during
+      https://arxiv.org/abs/1702.03275). This adds extra variables during
       training. The inference is the same for either value of this parameter.
     renorm_clipping: A dictionary that may map keys 'rmax', 'rmin', 'dmax' to
       scalar `Tensors` used to clip the renorm correction. The correction
@@ -164,9 +164,9 @@ class BatchNormalizationBase(Layer):
 
     \\({y_i} = {\gamma * \hat{x_i} + \beta}\\)
 
-  References:
+  Reference:
-  - [Batch Normalization: Accelerating Deep Network Training by Reducing
+
-    Internal Covariate Shift](https://arxiv.org/abs/1502.03167)
+    - [Ioffe and Szegedy, 2015](https://arxiv.org/abs/1502.03167).
   """
 
   # By default, the base class uses V2 behavior. The BatchNormalization V1
@@ -998,8 +998,8 @@ class LayerNormalization(Layer):
   Output shape:
     Same shape as input.
 
-  References:
+  Reference:
-    - [Layer Normalization](https://arxiv.org/abs/1607.06450)
+    - [Lei Ba et al., 2016](https://arxiv.org/abs/1607.06450).
   """
 
   def __init__(self,

tensorflow/python/keras/layers/normalization_v2.py

@@ -76,8 +76,8 @@ class SyncBatchNormalization(normalization.BatchNormalizationBase):
     gamma_regularizer: Optional regularizer for the gamma weight.
     beta_constraint: Optional constraint for the beta weight.
     gamma_constraint: Optional constraint for the gamma weight.
-    renorm: Whether to use Batch Renormalization
+    renorm: Whether to use [Batch Renormalization](
-      (https://arxiv.org/abs/1702.03275). This adds extra variables during
+      https://arxiv.org/abs/1702.03275). This adds extra variables during
       training. The inference is the same for either value of this parameter.
     renorm_clipping: A dictionary that may map keys 'rmax', 'rmin', 'dmax' to
       scalar `Tensors` used to clip the renorm correction. The correction

tensorflow/python/keras/layers/recurrent.py

@@ -2233,8 +2233,8 @@ class LSTMCell(DropoutRNNCellMixin, Layer):
     unit_forget_bias: Boolean.
       If True, add 1 to the bias of the forget gate at initialization.
       Setting it to true will also force `bias_initializer="zeros"`.
-      This is recommended in [Jozefowicz et
+      This is recommended in [Jozefowicz et al., 2015](
-        al.](http://www.jmlr.org/proceedings/papers/v37/jozefowicz15.pdf)
+        http://www.jmlr.org/proceedings/papers/v37/jozefowicz15.pdf)
     kernel_regularizer: Regularizer function applied to
       the `kernel` weights matrix.
     recurrent_regularizer: Regularizer function applied to
@@ -2503,7 +2503,8 @@ class PeepholeLSTMCell(LSTMCell):
   well as the previous hidden state (which is what LSTMCell is limited to).
   This allows PeepholeLSTMCell to better learn precise timings over LSTMCell.
 
-  From [Gers et al.](http://www.jmlr.org/papers/volume3/gers02a/gers02a.pdf):
+  From [Gers et al., 2002](
+    http://www.jmlr.org/papers/volume3/gers02a/gers02a.pdf):
 
   "We find that LSTM augmented by 'peephole connections' from its internal
   cells to its multiplicative gates can learn the fine distinction between
@@ -2512,9 +2513,7 @@ class PeepholeLSTMCell(LSTMCell):
 
   The peephole implementation is based on:
 
-  [Long short-term memory recurrent neural network architectures for
+  [Sak et al., 2014](https://research.google.com/pubs/archive/43905.pdf)
-   large scale acoustic modeling.
-  ](https://research.google.com/pubs/archive/43905.pdf)
 
   Example:
 
@@ -2601,8 +2600,8 @@ class LSTM(RNN):
     unit_forget_bias: Boolean.
       If True, add 1 to the bias of the forget gate at initialization.
       Setting it to true will also force `bias_initializer="zeros"`.
-      This is recommended in [Jozefowicz et
+      This is recommended in [Jozefowicz et al., 2015](
-        al.](http://www.jmlr.org/proceedings/papers/v37/jozefowicz15.pdf).
+        http://www.jmlr.org/proceedings/papers/v37/jozefowicz15.pdf).
     kernel_regularizer: Regularizer function applied to
       the `kernel` weights matrix.
     recurrent_regularizer: Regularizer function applied to

tensorflow/python/keras/losses.py

@@ -63,8 +63,8 @@ class Loss(object):
   types, and reduce losses explicitly in your training loop. Using 'AUTO' or
   'SUM_OVER_BATCH_SIZE' will raise an error.
 
-  Please see this custom training [tutorial]
+  Please see this custom training [tutorial](
-  (https://www.tensorflow.org/tutorials/distribute/custom_training) for more
+    https://www.tensorflow.org/tutorials/distribute/custom_training) for more
   details on this.
 
   You can implement 'SUM_OVER_BATCH_SIZE' using global batch size like:
@@ -88,8 +88,8 @@ class Loss(object):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op.
     """
@@ -222,8 +222,8 @@ class LossFunctionWrapper(Loss):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: (Optional) name for the loss.
       **kwargs: The keyword arguments that are passed on to `fn`.
@@ -305,8 +305,8 @@ class MeanSquaredError(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to 'mean_squared_error'.
     """
@@ -364,8 +364,8 @@ class MeanAbsoluteError(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to 'mean_absolute_error'.
     """
@@ -424,8 +424,8 @@ class MeanAbsolutePercentageError(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to
         'mean_absolute_percentage_error'.
@@ -485,8 +485,8 @@ class MeanSquaredLogarithmicError(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to
         'mean_squared_logarithmic_error'.
@@ -561,8 +561,8 @@ class BinaryCrossentropy(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: (Optional) Name for the op. Defaults to 'binary_crossentropy'.
     """
@@ -641,8 +641,8 @@ class CategoricalCrossentropy(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to 'categorical_crossentropy'.
     """
@@ -718,8 +718,8 @@ class SparseCategoricalCrossentropy(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to
         'sparse_categorical_crossentropy'.
@@ -782,8 +782,8 @@ class Hinge(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to 'hinge'.
     """
@@ -843,8 +843,8 @@ class SquaredHinge(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to 'squared_hinge'.
     """
@@ -903,8 +903,8 @@ class CategoricalHinge(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to 'categorical_hinge'.
     """
@@ -960,8 +960,8 @@ class Poisson(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to 'poisson'.
     """
@@ -1017,8 +1017,8 @@ class LogCosh(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to 'log_cosh'.
     """
@@ -1077,8 +1077,8 @@ class KLDivergence(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to 'kl_divergence'.
     """
@@ -1145,8 +1145,8 @@ class Huber(LossFunctionWrapper):
         this defaults to `SUM_OVER_BATCH_SIZE`. When used with
         `tf.distribute.Strategy`, outside of built-in training loops such as
         `tf.keras` `compile` and `fit`, using `AUTO` or `SUM_OVER_BATCH_SIZE`
-        will raise an error. Please see this custom training [tutorial]
+        will raise an error. Please see this custom training [tutorial](
-        (https://www.tensorflow.org/tutorials/distribute/custom_training)
+          https://www.tensorflow.org/tutorials/distribute/custom_training)
         for more details.
       name: Optional name for the op. Defaults to 'huber_loss'.
     """

tensorflow/python/keras/metrics.py

@@ -1416,8 +1416,8 @@ class Recall(Metric):
 class SensitivitySpecificityBase(Metric):
   """Abstract base class for computing sensitivity and specificity.
 
-  For additional information about specificity and sensitivity, see the
+  For additional information about specificity and sensitivity, see
-  following: https://en.wikipedia.org/wiki/Sensitivity_and_specificity
+  [the following](https://en.wikipedia.org/wiki/Sensitivity_and_specificity).
   """
 
   def __init__(self, value, num_thresholds=200, name=None, dtype=None):
@@ -1523,8 +1523,8 @@ class SensitivityAtSpecificity(SensitivitySpecificityBase):
   If `sample_weight` is `None`, weights default to 1.
   Use `sample_weight` of 0 to mask values.
 
-  For additional information about specificity and sensitivity, see the
+  For additional information about specificity and sensitivity, see
-  following: https://en.wikipedia.org/wiki/Sensitivity_and_specificity
+  [the following](https://en.wikipedia.org/wiki/Sensitivity_and_specificity).
 
   Args:
     specificity: A scalar value in range `[0, 1]`.
@@ -1598,8 +1598,8 @@ class SpecificityAtSensitivity(SensitivitySpecificityBase):
   If `sample_weight` is `None`, weights default to 1.
   Use `sample_weight` of 0 to mask values.
 
-  For additional information about specificity and sensitivity, see the
+  For additional information about specificity and sensitivity, see
-  following: https://en.wikipedia.org/wiki/Sensitivity_and_specificity
+  [the following](https://en.wikipedia.org/wiki/Sensitivity_and_specificity).
 
   Args:
     sensitivity: A scalar value in range `[0, 1]`.
@@ -1828,13 +1828,14 @@ class AUC(Metric):
       use when discretizing the roc curve. Values must be > 1.
     curve: (Optional) Specifies the name of the curve to be computed, 'ROC'
       [default] or 'PR' for the Precision-Recall-curve.
-    summation_method: (Optional) Specifies the Riemann summation method used
+    summation_method: (Optional) Specifies the [Riemann summation method](
-      (https://en.wikipedia.org/wiki/Riemann_sum): 'interpolation' [default],
+        https://en.wikipedia.org/wiki/Riemann_sum) used.
-        applies mid-point summation scheme for `ROC`. For PR-AUC, interpolates
+        'interpolation' (default) applies mid-point summation scheme for `ROC`.
-        (true/false) positives but not the ratio that is precision (see Davis
+        For PR-AUC, interpolates (true/false) positives but not the ratio that
-        & Goadrich 2006 for details); 'minoring' that applies left summation
+        is precision (see Davis & Goadrich 2006 for details);
+        'minoring' applies left summation
         for increasing intervals and right summation for decreasing intervals;
-        'majoring' that does the opposite.
+        'majoring' does the opposite.
     name: (Optional) string name of the metric instance.
     dtype: (Optional) data type of the metric result.
     thresholds: (Optional) A list of floating point values to use as the
@@ -2226,8 +2227,9 @@ class AUC(Metric):
 class CosineSimilarity(MeanMetricWrapper):
   """Computes the cosine similarity between the labels and predictions.
 
-  cosine similarity = (a . b) / ||a|| ||b||
+  `cosine similarity = (a . b) / ||a|| ||b||`
-  [Cosine Similarity](https://en.wikipedia.org/wiki/Cosine_similarity)
+
+  See: [Cosine Similarity](https://en.wikipedia.org/wiki/Cosine_similarity).
 
   This metric keeps the average cosine similarity between `predictions` and
   `labels` over a stream of data.

tensorflow/python/keras/mixed_precision/__init__.py

@@ -15,8 +15,8 @@
 
 """Keras mixed precision API.
 
-See [the mixed precision
+See [the mixed precision guide](
-guide](https://www.tensorflow.org/guide/keras/mixed_precision) to learn how to
+  https://www.tensorflow.org/guide/keras/mixed_precision) to learn how to
 use the API.
 """
 from __future__ import absolute_import

tensorflow/python/keras/mixed_precision/experimental/policy.py

@@ -57,8 +57,8 @@ class Policy(object):
   not have a single dtype. When the variable dtype does not match the compute
   dtype, variables will be automatically casted to the compute dtype to avoid
   type errors. In this case, `tf.keras.layers.Layer.dtype` refers to the
-  variable dtype, not the compute dtype. See [the mixed precision
+  variable dtype, not the compute dtype. See [the mixed precision guide](
-  guide](https://www.tensorflow.org/guide/keras/mixed_precision) for more
+    https://www.tensorflow.org/guide/keras/mixed_precision) for more
   information on how to use mixed precision.
 
   Certain policies also have a `tf.mixed_precision.experimental.LossScale`
@@ -119,8 +119,8 @@ class Policy(object):
   `'mixed_bfloat16'`, no loss scaling is done and loss scaling never needs to be
   manually applied.
 
-  See [the mixed precision
+  See [the mixed precision guide](
-  guide](https://www.tensorflow.org/guide/keras/mixed_precision) for more
+    https://www.tensorflow.org/guide/keras/mixed_precision) for more
   information on using mixed precision
 
   ### How to use float64 in a Keras model