Add back-ticks around code in the docs of nlp modeling library, so that the...

Add back-ticks around code in the docs of nlp modeling library, so that the name will get auto-linked to that api page.

PiperOrigin-RevId: 357808494

official/nlp/modeling/layers/attention.py

@@ -27,7 +27,7 @@ MultiHeadAttention = tf.keras.layers.MultiHeadAttention
 class CachedAttention(tf.keras.layers.MultiHeadAttention):
   """Attention layer with cache used for auto-agressive decoding.
 
-  Arguments are the same as `MultiHeadAttention` layer.
+  Arguments are the same as `tf.keras.layers.MultiHeadAttention` layer.
   """
 
   def _update_cache(self, key, value, cache, decode_loop_step):

official/nlp/modeling/layers/dense_einsum.py

@@ -24,7 +24,7 @@ _CHR_IDX = ["a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m"]
 
 @tf.keras.utils.register_keras_serializable(package="Text")
 class DenseEinsum(tf.keras.layers.Layer):
-  """A densely connected layer that uses tf.einsum as the backing computation.
+  """A densely connected layer that uses `tf.einsum` as the backing computation.
 
   This layer can perform einsum calculations of arbitrary dimensionality.
 

official/nlp/modeling/layers/masked_softmax.py

@@ -22,7 +22,7 @@ def _large_compatible_negative(tensor_type):
   """Large negative number as Tensor.
 
   This function is necessary because the standard value for epsilon
-  in this module (-1e9) cannot be represented using tf.float16
+  in this module (-1e9) cannot be represented using `tf.float16`.
 
   Args:
     tensor_type: a dtype to determine the type.

official/nlp/modeling/layers/relative_attention.py

@@ -75,7 +75,7 @@ class MultiHeadRelativeAttention(tf.keras.layers.MultiHeadAttention):
   """A multi-head attention layer with relative attention + position encoding.
 
   This layer shares the same input/output projections as the common
-  MultiHeadAttention layer.
+  `tf.keras.layers.MultiHeadAttention` layer.
 
   When it calculates attention logits, position encoding is projected to form
   relative keys. The logits are composed by shifted relative logits and content
@@ -333,8 +333,9 @@ class TwoStreamRelativeAttention(MultiHeadRelativeAttention):
   The query stream only has access to contextual information and the position,
   but not the content.
 
-  This layer shares the same build signature as `MultiHeadRelativeAttention` but
-  has different input/output projections.
+  This layer shares the same build signature as
+  `tf.keras.layers.MultiHeadAttention` but has different input/output
+  projections.
 
   **Note: This layer is currently experimental.
 

official/nlp/modeling/layers/self_attention_mask.py

@@ -23,7 +23,7 @@ from official.nlp.keras_nlp import layers
 class SelfAttentionMask(layers.SelfAttentionMask):
   """Create 3D attention mask from a 2D tensor mask.
 
-    **Warning: Please use the keras_nlp.layers.SelfAttentionMask.**
+    **Warning: Please use the `keras_nlp.layers.SelfAttentionMask`.**
     inputs[0]: from_tensor: 2D or 3D Tensor of shape
       [batch_size, from_seq_length, ...].
     inputs[1]: to_mask: int32 Tensor of shape [batch_size, to_seq_length].

official/nlp/modeling/layers/talking_heads_attention.py

@@ -33,7 +33,7 @@ class TalkingHeadsAttention(tf.keras.layers.MultiHeadAttention):
   multi-head attention by including linearprojections across the attention-heads
   dimension, immediately before and after the softmax operation.
 
-  See the base class `MultiHeadAttention` for more details.
+  See the base class `tf.keras.layers.MultiHeadAttention` for more details.
 
   Args:
     num_heads: Number of attention heads.

official/nlp/modeling/layers/text_layers.py

@@ -97,8 +97,9 @@ class BertTokenizer(tf.keras.layers.Layer):
   """Wraps BertTokenizer with pre-defined vocab as a Keras Layer.
 
   Attributes:
-    tokenize_with_offsets: If true, calls BertTokenizer.tokenize_with_offsets()
-      instead of plain .tokenize() and outputs a triple of
+    tokenize_with_offsets: If true, calls
+      `text.BertTokenizer.tokenize_with_offsets()` instead of plain
+      `text.BertTokenizer.tokenize()` and outputs a triple of
       (tokens, start_offsets, limit_offsets).
     raw_table_access: An object with methods .lookup(keys) and .size()
       that operate on the raw lookup table of tokens. It can be used to
@@ -110,25 +111,26 @@ class BertTokenizer(tf.keras.layers.Layer):
                lower_case: bool,
                tokenize_with_offsets: bool = False,
                **kwargs):
-    """Initialize a BertTokenizer layer.
+    """Initialize a `BertTokenizer` layer.
 
     Args:
       vocab_file: A Python string with the path of the vocabulary file.
         This is a text file with newline-separated wordpiece tokens.
         This layer initializes a lookup table from it that gets used with
-        text.BertTokenizer.
-      lower_case: A Python boolean forwarded to text.BertTokenizer.
+        `text.BertTokenizer`.
+      lower_case: A Python boolean forwarded to `text.BertTokenizer`.
         If true, input text is converted to lower case (where applicable)
         before tokenization. This must be set to match the way in which
         the vocab_file was created.
       tokenize_with_offsets: A Python boolean. If true, this layer calls
-         BertTokenizer.tokenize_with_offsets() instead of plain .tokenize()
-         and outputs a triple of (tokens, start_offsets, limit_offsets)
+         `text.BertTokenizer.tokenize_with_offsets()` instead of plain
+         `text.BertTokenizer.tokenize()` and outputs a triple of
+         (tokens, start_offsets, limit_offsets)
          insead of just tokens.
       **kwargs: standard arguments to Layer().
 
     Raises:
-      ImportError: if importing tensorflow_text failed.
+      ImportError: if importing `tensorflow_text` failed.
     """
     _check_if_tf_text_installed()
 
@@ -162,18 +164,18 @@ class BertTokenizer(tf.keras.layers.Layer):
     return vocab_table, vocab_initializer
 
   def call(self, inputs: tf.Tensor):
-    """Calls text.BertTokenizer on inputs.
+    """Calls `text.BertTokenizer` on inputs.
 
     Args:
       inputs: A string Tensor of shape [batch_size].
 
     Returns:
-      One or three of RaggedTensors if tokenize_with_offsets is False or True,
-      respectively. These are
-      tokens: A RaggedTensor of shape [batch_size, (words), (pieces_per_word)]
+      One or three of `RaggedTensors` if `tokenize_with_offsets` is False or
+      True, respectively. These are
+      tokens: A `RaggedTensor` of shape [batch_size, (words), (pieces_per_word)]
         and type int32. tokens[i,j,k] contains the k-th wordpiece of the
         j-th word in the i-th input.
-      start_offsets, limit_offsets: If tokenize_with_offsets is True,
+      start_offsets, limit_offsets: If `tokenize_with_offsets` is True,
         RaggedTensors of type int64 with the same indices as tokens.
         Element [i,j,k] contains the byte offset at the start, or past the
         end, resp., for the k-th wordpiece of the j-th word in the i-th input.

official/nlp/modeling/models/bert_pretrainer.py

@@ -30,7 +30,7 @@ from official.nlp.modeling import networks
 class BertPretrainer(tf.keras.Model):
   """BERT pretraining model.
 
-  [Note] Please use the new BertPretrainerV2 for your projects.
+  [Note] Please use the new `BertPretrainerV2` for your projects.
 
   The BertPretrainer allows a user to pass in a transformer stack, and
   instantiates the masked language model and classification networks that are

official/nlp/modeling/models/electra_pretrainer.py

@@ -37,7 +37,7 @@ class ElectraPretrainer(tf.keras.Model):
   that are used to create the training objectives.
 
   *Note* that the model is constructed by Keras Subclass API, where layers are
-  defined inside __init__ and call() implements the computation.
+  defined inside `__init__` and `call()` implements the computation.
 
   Args:
     generator_network: A transformer network for generator, this network should