Returns feature type for an element in the feature vector.

Returns the next gold action for a given state according to the underlying annotated sentence.

Returns true if 'c' is in the range [0, 0xD800) or [0xE000, 0x10FFFF] (i.e., is not a surrogate codepoint). See also IsValidCodepoint(const char* src)

Returns a new transition state that can be used to put additional information in a parser state. By specifying if we are in training_mode (true) or no

Makes a shift by pushing the next input token on the stack and moving to the next position.

Runs ComputeValue for each token in the sentence.

Generates a name for the op, using a user-supplied name if possible. Args: op_str: String representing the parsed op. index:

Wrapper for inserting bytes features into Example proto.

Converts `int_or_tuple` to height, width. Several of the functions that follow accept arguments as either a tuple of 2 integers or a single integ

Creates all ops for training and evaluation.

Returns the device to use for caching variables. Variables are cached on the worker CPU when using replicas. Returns: A device string

This implementation of clear() deallocates the buffer if we're an owner. That's not strictly necessary; we could just set size_ to 0.

Returns a python list of all (sharded) data subset files. Returns: python list of all (sharded) data set files. Raises: ValueErro

Helper function to create a weight variable initialized with a normal distribution Parameters ---------- shape : list Size of

Returns the lengths of the batch of elements in the given dimension. WARNING: The returned sizes may not exactly match TF's calculation. Args

Returns the term index or the unknown value. Used inside GetTokenIndex() specializations for convenience.

Returns the number of tokens in the sentence.

Add a model parameter w.r.t. we expect to compute gradients. _AddParam creates both regular parameters (usually for training) and averaged no

Builds a feed-forward part of the net given features as input. The network topology is already defined in the constructor, so multiple calls

Collapses CTC output to regular output. Args: ctc_labels: List of class labels including null characters to remove. merge_dups: If Tr

Wrapper for inserting bytes features into Example proto.

Wrapper for inserting int64 features into Example proto.

Helper function to create a bias variable initialized with a constant value. Parameters ---------- shape : list Size of weight

Convert a list of ids to a sentence, with space inserted.

Subtracts mean of image and divides by adjusted standard variance (for stability). Operations are per image but performed for the entire array.

Download a set of files in temporary local folder :param directory: the directory where to download :return: a tuple of filepaths correspondin

Print a message out and log it to file.

Reserves space in the underlying feature vector.

Run a step of the network.

Returns the transition system-specific state.

Returns a string representation of a parser action.

Adds ops to save and restore model parameters. Args: slim_model: whether only averaged variables are saved. Returns: the saver o

Returns a sanitized percent error rate from the raw counts. Prevents div by 0 and clips return to 100%. Args: error_count: Number of errors.

Extracts features from an object with a focus. This invokes all the top-level feature functions in the feature extractor. Only feature functions belon

Returns true if this byte is a trailing UTF-8 byte (10xx xxxx)

Returns the number of start tokens that have already been identified. In other words, number of start tokens between the first token of the sentence a

Number of unique values.

Outputs the given batch of sentences as a tensor and deletes them.

Runs a step for eval in the session. Args: sess: Session in which to run the model. Returns: output tensor result, lab

Runs a training step in the session. Args: sess: Session in which to train the model. Returns: loss, global_step.

Adds the given rate to the summary with the given tag. Args: tag: Name for this value. rate: Value to add to the summary. Perhaps an err

Load the model from GraphDef and Checkpoint. Args: gd_file: GraphDef proto text file. ckpt_file: TensorFlow Checkpoint file. Returns:

Verifies that the model produces expected outputs. Args: expected_shapes: A dict mapping Tensor or Tensor name to expected output s

Wrapper for inserting float features into Example proto.

Wrapper for inserting int64 features into Example proto.

Round robin the gpu device. (Reserve last gpu for expensive op).

Verifies the number of parameters in the inception model.

Returns the textual representation of the affix.

Returns unique id of affix.

Gets/sets the flag which says that the state was obtained though gold transitions only.

Returns the maximum affix length.

Run multiple tests at lower batch size to save memory.

Returns the next component in the component list.

Generator for v1 ResNet models. This function generates a family of ResNet v1 models. See the resnet_v1_*() methods for specific model instantiat