Merge pull request #9575 from SamuelMarks:args-for-google-style-docstrings-official

PiperOrigin-RevId: 348853056

official/nlp/modeling/networks/xlnet_base.py

@@ -52,7 +52,7 @@ def _create_causal_attention_mask(
   We then flip the matrix values in order to match the representation where
   real values are 1s.
 
-  Arguments:
+  Args:
     seq_length: int, The length of each sequence.
     memory_length: int, The length of memory blocks.
     dtype: dtype of the mask.
@@ -392,7 +392,7 @@ class RelativePositionEncoding(tf.keras.layers.Layer):
   def call(self, pos_seq, batch_size=None):
     """Implements call() for the layer.
 
-    Arguments:
+    Args:
       pos_seq: A 1-D `Tensor`
       batch_size: The optionally provided batch size that tiles the relative
         positional encoding.

official/nlp/projects/bigbird/encoder.py

@@ -30,7 +30,7 @@ class BigBirdEncoder(tf.keras.Model):
   *Note* that the network is constructed by
   [Keras Functional API](https://keras.io/guides/functional_api/).
 
-  Arguments:
+  Args:
     vocab_size: The size of the token vocabulary.
     hidden_size: The size of the transformer hidden layers.
     num_layers: The number of transformer layers.

official/nlp/xlnet/xlnet_modeling.py

@@ -114,7 +114,7 @@ class RelativePositionEncoding(tf.keras.layers.Layer):
   def call(self, pos_seq, batch_size=None):
     """Implements call() for the layer.
 
-    Arguments:
+    Args:
       pos_seq: A 1-D `Tensor`
       batch_size: The optionally provided batch size that tiles the relative
         positional encoding.

official/staging/training/grad_utils.py

@@ -57,7 +57,7 @@ def _filter_and_allreduce_gradients(grads_and_vars,
   The allreduced gradients are then passed to optimizer.apply_gradients(
   experimental_aggregate_gradients=False).
 
-  Arguments:
+  Args:
       grads_and_vars: gradients and variables pairs.
       allreduce_precision: Whether to allreduce gradients in float32 or float16.
       bytes_per_pack: A non-negative integer. Breaks collective operations into
@@ -101,7 +101,7 @@ def minimize_using_explicit_allreduce(tape,
   For TPU and GPU training using FP32, explicit allreduce will aggregate
   gradients in FP32 format.
 
-  Arguments:
+  Args:
       tape: An instance of `tf.GradientTape`.
       optimizer: An instance of `tf.keras.optimizers.Optimizer`.
       loss: the loss tensor.

official/vision/beta/losses/retinanet_losses.py

@@ -68,7 +68,7 @@ class FocalLoss(tf.keras.losses.Loss):
                name=None):
     """Initializes `FocalLoss`.
 
-    Arguments:
+    Args:
       alpha: The `alpha` weight factor for binary class imbalance.
       gamma: The `gamma` focusing parameter to re-weight loss.
       num_classes: Number of foreground classes.
@@ -91,7 +91,7 @@ class FocalLoss(tf.keras.losses.Loss):
   def call(self, y_true, y_pred):
     """Invokes the `FocalLoss`.
 
-    Arguments:
+    Args:
       y_true: Ordered Dict with level to [batch, height, width, num_anchors].
         for example,
         {3: tf.Tensor(shape=[32, 512, 512, 9], dtype=tf.float32),
@@ -143,7 +143,7 @@ class RetinanetBoxLoss(tf.keras.losses.Loss):
                name=None):
     """Initializes `RetinanetBoxLoss`.
 
-    Arguments:
+    Args:
       delta: A float, the point where the Huber loss function changes from a
         quadratic to linear.
       reduction: (Optional) Type of `tf.keras.losses.Reduction` to apply to
@@ -167,7 +167,7 @@ class RetinanetBoxLoss(tf.keras.losses.Loss):
 
     Computes total detection loss including box and class loss from all levels.
 
-    Arguments:
+    Args:
       y_true: Ordered Dict with level to [batch, height, width,
         num_anchors * 4] for example,
         {3: tf.Tensor(shape=[32, 512, 512, 9 * 4], dtype=tf.float32),

official/vision/detection/executor/distributed_executor.py

@@ -108,7 +108,7 @@ class SummaryWriter(object):
   def __init__(self, model_dir: Text, name: Text):
     """Inits SummaryWriter with paths.
 
-    Arguments:
+    Args:
       model_dir: the model folder path.
       name: the summary subfolder name.
     """

official/vision/keras_cv/layers/deeplab.py

@@ -42,7 +42,7 @@ class SpatialPyramidPooling(tf.keras.layers.Layer):
       **kwargs):
     """Initializes `SpatialPyramidPooling`.
 
-    Arguments:
+    Args:
       output_channels: Number of channels produced by SpatialPyramidPooling.
       dilation_rates: A list of integers for parallel dilated conv.
       pool_kernel_size: A list of integers or None. If None, global average

official/vision/keras_cv/losses/focal_loss.py

@@ -31,7 +31,7 @@ class FocalLoss(tf.keras.losses.Loss):
                name=None):
     """Initializes `FocalLoss`.
 
-    Arguments:
+    Args:
       alpha: The `alpha` weight factor for binary class imbalance.
       gamma: The `gamma` focusing parameter to re-weight loss.
       reduction: (Optional) Type of `tf.keras.losses.Reduction` to apply to
@@ -52,7 +52,7 @@ class FocalLoss(tf.keras.losses.Loss):
   def call(self, y_true, y_pred):
     """Invokes the `FocalLoss`.
 
-    Arguments:
+    Args:
       y_true: A tensor of size [batch, num_anchors, num_classes]
       y_pred: A tensor of size [batch, num_anchors, num_classes]
 

official/vision/keras_cv/losses/loss_utils.py

@@ -20,7 +20,7 @@ import tensorflow as tf
 def multi_level_flatten(multi_level_inputs, last_dim=None):
   """Flattens a multi-level input.
 
-  Arguments:
+  Args:
     multi_level_inputs: Ordered Dict with level to [batch, d1, ..., dm].
     last_dim: Whether the output should be [batch_size, None], or [batch_size,
       None, last_dim]. Defaults to `None`.

official/vision/keras_cv/metrics/iou.py

@@ -48,7 +48,7 @@ class PerClassIoU(tf.keras.metrics.Metric):
   def __init__(self, num_classes, name=None, dtype=None):
     """Initializes `PerClassIoU`.
 
-    Arguments:
+    Args:
       num_classes: The possible number of labels the prediction task can have.
         This value must be provided, since a confusion matrix of dimension =
         [num_classes, num_classes] will be allocated.