Encode the given response texts as vectors. Args: responses: a list of N strings, to be encoded. Returns: an

Encode the given texts as vectors. Args: contexts: a list of N strings, to be encoded. Returns: an (N, d) nu

Compute the similarities between context and responses. Uses a learned mapping on the response side.

Load contexts and responses from the given conversational dataset.

Reddit IDs start with t1_, t2_, etc. which need to be stripped.

Whether a line should be skipped depending on the length.

Vectorize the given strings.

Vectorize the given strings.

Encode the given texts.

Encode the given texts.

Rank the responses for each context.

Fit the tf-idf transform and compute idf statistics.

Perform any training steps using the (context, response) examples. Args: examples: a list of `(context, response)` string tuples,

Build the graph that applies a learned mapping to the vectors.

Command line utilities for manipulating tfrecords files.

Create an example dictionary.

Get the vocab file and casing info from the Hub module.

Create a train and dev set of context and response vectors.

Create the train op.

Create a feed dict for feeding the texts as input. This uses dynamic padding so that the maximum sequence length is the smaller of `m

Perform a grid search of training hyper-parameters. The model that does the best on the dev set will be stored.

Parse command-line args.

Parse command-line args.

Parse command-line args.

Parse command line arguments.

Format and print an individual tensorflow example.

Print the extra context features.

Print the other features, which will depend on the dataset. For now, only support string features.

Shuffles the input pcollection.

Compute a value from 0 to 1 used to compute the split.

Compute a value from 0 to 1 used to compute the split.

Compute a value from 0 to 1 used to compute the split.

Creates examples with multi-line context The examples will include: file_id: the name of the file where these lines were obtained.

Encode the given texts as contexts.

Encode the given texts as contexts.

Encode the given texts as responses.

Encode the given texts as responses.

Gets all linear paths of comments and replies from the thread. Each linear path is guaranteed to have at least two comments in it.

Rank the responses for each context.

Rank the responses for each context. Args: contexts: a list of strings giving the contexts to use. responses: a list

Rank the responses for each context, using cosine similarity.

Rank the responses for each context.

Run the beam pipeline.

Run the beam pipeline.

Run the beam pipeline. Args: argv: (optional) the command line flags to parse. comments_collection: (optional) a list of comment

Convert the enum to an instance of `BaselineMethod`.

Fit the tf-idf transform and compute idf statistics.

Train on the contexts and responses.

Trims text to be at most `max_length`, without splitting apart words.

Create a new `BM25Method` object.

Create a new `TfHubEncoder` object.

Create a new `USEDualEncoder` object.

Create a new `ConveRTEncoder` object.

Create a new `BERTEncoder` object.

Create a new `VectorSimilarityMethod` object.

Create a new `VectorMappingMethod` object.

String representation to use in argparse help text.

Creates (product_id, question, answer) tuples.

Convert a string dict to a serialized TF example. The dictionary maps feature names (strings) to feature values (strings).

Convert a string dict to a serialized TF example. The dictionary maps feature names (strings) to feature values (strings).

Convert a string dict to a serialized TF example. The dictionary maps feature names (strings) to feature values (strings).

Define a positive integer ArgumentParser type.

Define a positive integer ArgumentParser type.

Define a positive integer ArgumentParser type.

Format and print the contents of the tfrecord file to stdout.

Compute the number of examples in the input tfrecord file.

Creates serialized tensorflow examples from a reddit thread.

Convert a string parsed from argparse to an enum instance.

Create a _Comment object from a row in the BigQuery table.