Returns size of the affix table.

Named accessor/initializer:

Clear text.

Adds feature type and value pair to feature vector.

Returns the TopN to an empty state.

UTF-8-specific methods Return the data, length, and capacity of UTF-8-encoded version of the text. Length and capacity are measured in bytes.

Adds the token at given index as a new start token.

Returns the number of terms with positive frequency.

Returns feature value for an element in the feature vector.

Returns component name.

Removes all elements from the feature vector.

Create test input tensor. Args: batch_size: The number of images per batch or `None` if unknown. height: The height of each image or `None`

Create test input tensor. Args: batch_size: The number of images per batch or `None` if unknown. height: The height of each image or `None`

Returns the length of the affix.

Helper to create a Variable stored on CPU memory. Args: name: name of the variable shape: list of ints initializer: initializer for Var

Helper to create an initialized Variable with weight decay. Note that the Variable is initialized with a truncated normal distribution. A weight

Returns a RandomUniform Tensor between -param_init and param_init.

We are in a final state when we reached the end of the input and the stack is empty.

Name scope to create the clone. Args: clone_index: Int, representing the clone_index. Returns: A name_scope suitable for `tf.nam

Minimize using sanitized gradients. This gets a var_list which is the list of trainable variables. For each var in var_list, we defined a gra

Device used to create the clone and all the ops inside the clone. Args: clone_index: Int, representing the clone_index. Returns:

A shallow and thin ResNet v1 for faster tests.

A shallow and thin ResNet v2 for faster tests.

Builds a trainer to minimize the cross entropy cost function. Args: task_context: file path from which to read the task context batch

Tests that the output of the graph of the given spec has target_shape.

Convert a sentence to a list of ids, with special tokens added.

Device to use with the optimizer. Returns: A value suitable for `tf.device()`.

Override the default train() to freeze the BN parameters :return:

Returns the index of the k-th most recent start token.

Look up an existing TokenMorphology. If it is not present, return -1.

Returns the index associated with the given term. If the term does not exist, the unknown index is returned instead.

Tests that beam search generates the expected captions. Args: expected_captions: A sequence of pairs (sentence, probability), where

Batch normalization.

Relu, with optional leaky support.

Device to use to build the inputs. Returns: A value suitable for `tf.device()`.

Returns the device to use for variables created inside the clone. Returns: A value suitable for `tf.device()`.

Builds the forward network only without the training operation. Args: task_context: file path from which to read the task context. ba

Checks if the action is allowed in a given parser state.

Requests workspaces from the registry. Must be called after Init(), and before Preprocess(). Does not take ownership over registry. This should be the

Convolution.

Map a stride scalar to the stride array for tf.nn.conv2d.