nlu_model.py

import logging
import pprint
from rasa.nlu.training_data import load_data
from rasa.nlu import config
from rasa.nlu.model import Trainer
from rasa.nlu.model import Interpreter
from rasa.nlu.test import run_evaluation


logfile = 'nlu_model.log'


def train_nlu(data_path, configs, model_path):
    logging.basicConfig(filename=logfile, level=logging.DEBUG)
    training_data = load_data(data_path)
    trainer = Trainer(config.load(configs))
    trainer.train(training_data)
    model_directory = trainer.persist(model_path, fixed_model_name='nlu')
    run_evaluation(data_path, model_directory)


def run_nlu(nlu_path):
    logging.basicConfig(filename=logfile, level=logging.DEBUG)
    interpreter = Interpreter.load(nlu_path)
    pprint.pprint(interpreter.parse("Share some latest news around the world?"))
    pprint.pprint(interpreter.parse("What is going on in technology?"))
    pprint.pprint(interpreter.parse("What is going on in education?"))


if __name__ == '__main__':
    # train_nlu('./data/nlu.md', 'nlu_config.yml', './models')
    run_nlu('./models/nlu')