👋 Introducing the Rasa Playground

Command line bot Rasa

post by JohannesKann on Jun 27, 2019

Hi

I’m running: Python 3.7.3, rasa 1.0.7, rasa-nlu 0.15.0 and rasa-sdk 1.0.0

I run Python and Rasa on a Dell Windows 7 desktop with 32GB RAM

I have been following a number of the on-line tutorials from various contributors, but noticed that many of the examples are 6 to 12 months old, and so when trying to implement them with the above versions of Python and Rasa I run into troubles, typically due to Rasa being updated several times since these tutorials were first written.

I have managed to amend my code to sort out most issues, but I can’t seem to complete the dialogue_management_model, particularly I can’t get the command line bot to work. I think the train_dialogue routine is fine as that seems to work.

All the examples I have seen for run_bot use commands that are now deprecated, and I can’t find the current alternative methods. What should I do to get the code to run correctly?

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from __future__ import unicode_literals

import logging
import os
import asyncio

from rasa.core.agent import Agent
from rasa.core.interpreter import RasaNLUInterpreter
from rasa.core.policies import KerasPolicy, MemoizationPolicy
from rasa.utils.endpoints import EndpointConfig
from bot_constants import NLU_MODELS_DIR, BOT_NAME
from rasa.core.train import online

logger = logging.getLogger(__name__)

async def train_dialogue(domain_file = './data/domain.yml',
               model_path = './models/dialogue',
               training_data_file = './data/stories.md'):

agent = Agent(domain_file, policies = [MemoizationPolicy(max_history = 5),\
                                       KerasPolicy(max_history = 3,\
                                                   epochs = 300,\
                                                   batch_size = 50,\
                                                   validation_split = 0.2,\
                                                   augmentation_factor=50)])

training_data = await agent.load_data(training_data_file)

agent.train(training_data)

agent.persist(model_path)

return agent

def run_bot(serve_forever = True):
    interpreter = RasaNLUInterpreter(os.path.join(NLU_MODELS_DIR + '/' + BOT_NAME))
    action_endpoint = EndpointConfig(url="http://localhost:5055/webhook")
    agent = Agent.load('./models/dialogue', interpreter=interpreter, action_endpoint=action_endpoint)
    if serve_forever:
        online.serve_agent(agent)
    return agent

if __name__ == '__main__':
    loop = asyncio.get_event_loop()
    loop.run_until_complete(train_dialogue())
    run_bot()

post by MetcalfeTom on Jul 5, 2019

Hi @JohannesKann,

Why do you want to run the bot using the Python API specifically? As far as I can tell, you should be able to achieve the same thing using rasa train and rasa run.

Anyway, to help with your queries you can consult the rasa_core migrations docs from the old repo, but I think await agent.handle_channels([ConsoleInputChannel()]) should work here.