Helper function: computes argmax along last dimension of array to obtain labels (max prob or max logit value) :param probs: numpy array where p

Computes the logmgf value given q and privacy eps. The bound used is the min of three terms. The first term is from https://arxiv.org/pdf/1605.02

Returns the nearest neighbors to the query (a word or vector).

Returns the number of examples in the data subset.

Decodes and processes an image string. Args: encoded_image: A scalar string Tensor; the encoded image. thread_id: Preprocessing threa

Random data pair for duplication task. Total length should be <= l.

Random data pair for key-value sort. Total length should be <= l.

Random data pair for a task. Total length should be <= l.

Random data pair for reverse2 task. Total length should be <= l.

Random data pair for search task. Total length should be <= l.

Reads text file with one number per line to an array.

Return a reader for a single entry from the data set. See io_ops.py for details of Reader class. Returns: Reader object that reads the

Sanitize the given tensor. This santize a given tensor by first applying l2 norm clipping and then adding Gaussian noise. It calls the privac

Return sensitivity at distane k. Args: counts: an array of scores noise_eps: noise parameter used l: moment whose sensitivity is being

Gets/sets the affix that is one character shorter.

Sigmoid with cutoff, e.g., 1.2sigmoid(x) - 0.1.

Return the target given the input for some tasks.

Returns a heterogeneous list of tensor_dim for the tensor. See tensor_dim for a more detailed explanation. Args: tensor: Input tensor. Ret

Helper function: unpickles a dictionary (used for loading CIFAR) :param file: filename of the pickle :return: tuple of (images, labels)

Fix the variable device to colocate its ops.

Splits a line of text into tokens.

Writes tensor to output_path as tsv

Writes vocabulary and marginal sum files.

Adds a cross entropy cost function.

Embeddings at the given index will be set to pretrained values.

This method updates the beam. For all elements of the beam, all allowed transitions are scored and insterted into a new beam. The beam size is capped

Batch compatible sparse to dense conversion. This is useful for one-hot coded target labels. Args: sparse_indices: [batch_size] tensor conta

Builds the model from the separate input/layers/output spec strings. Args: input_pattern: File pattern of the data in tfrecords of TF Examp

Builds a network with input prev_layer from a VGSLSpecs description. Args: prev_layer: The input tensor. model_str: Model definition

Computes the label of the relation between the focus token and its parent. Valid focus values range from -1 to sentence->size() - 1, inclusively.

Returns the first nested feature's computed value.

Converts a sentence to a key/value pair.

Counts the drops and adds between 2 bags of iterables. Simple bag of objects count returns the number of dropped and added elements, regardless o

Creates a vocabulary from the most frequent terms in the input file.

Lower-level versions of Get... that read directly from a character buffer without any bounds checking.

Decoding loop for long running process.

Returns the dimensionality of the embedding space.

Computes embeddings for each entry of sparse features sparse_features. Args: params: list of 2D tensors containing vector embeddings sparse

Number of predicates for the embedding at a given index (vocabulary size.)

Builds and evaluates a network.

Evaluates a network and checkpoints it to disk. Args: sess: tensorflow session to use parser: graph builder containing all ops references

Extend the hypothesis with result from latest step. Args: token: latest token from decoding. log_prob: log prob of the latest decoded

Convenience function for computing start/end byte offsets of a character StringPiece relative to original text.

Returns the feature name for a given feature value.

Returns the number of correct analogies after evaluation is complete.

Returns the offset (i.e. row number) of a particular beam at a particular step in the final concatenated score matrix.

Returns the gold tag for a given token.

Returns the gold tag for a given token.

Initializes the feature extractor. Must be called after Setup(). This does not take ownership over the context, which must outlive this.