* This function generates an event that is compatible with standard * compliant browsers and IE9 - IE11 * * This will preve

Convert a list of integer into a human readable string Args: sequence (list<int>): the sentence to print clean (Bool):

Return the number of words present in the dataset Return: int: Number of word on the loader corpus

Get the id of the word (and add it to the dictionary if not existing). If the word does not exist and create is set to False, the function wil

Convert a list of integer into a human readable string. The difference between the previous function is that on a batch object, the values hav

Launch the training and/or the interactive mode

Called by Django only once during startup

Predict the sentence Args: question (str): the raw input sentence questionSeq (List<int>): output argument. If given

Create a single batch from the list of sample. The batch size is automatically defined by the number of samples given. The inputs shou

Return the list of the model files inside the model directory

Parse the argument to decide were to save/load the model This function is called at each checkpoint and the first time the model is load. If

Save the model parameters and the variables Args: sess: the current session

Return the dataset availables Return: list<string>: the supported corpus

Extract the words from a sample lines Args: line (str): a line containing the text to extract Return: list<lis

Return the size of the dataset Return: int: Number of training samples

Load samples from file Args: filename (str): pickle filename

Save samples to file Args: filename (str): pickle filename

Forward/training step operation. Does not perform run on itself but just return the operators to do so. Those have then to be run Arg

Return the name of the base prefix of the current dataset

Return the unique id for the client Args: client list<>: the client which send the message of the from [ip (str), port (int)] Return:

Parse the argument to decide were to save the summary, at the same place that the model The folder could already contain logs if we restore t

Create the computational graph

Use the previously instantiated bot to predict a response to the given sentence Args: sentence (str): the question to answer

Extract all data from the given vocabulary. Save the data on disk. Note that the entire corpus is pre-processed without restriction on

Return the answer to a given sentence (same as singlePredict() but with additional cleaning) Args: sentence (str): the raw input

Decode the output of the decoder and return a human friendly sentence decoderOutputs (list<np.array>):