RASA English Conversational Chatbot

Setup Mbaza

Here are the steps on how to set Mbaza up on your local machine:

1. Clone the Mbaza repository:

git clone https://github.com/MBAZA-NLP/mbaza-covid-chatbot-english.git

cd mbaza-covid-chatbot-english

2. (OPTIONAL) Isolate your python project using a virtual environment

We recommend that you work within a virtual environment.

• Create a new virtual environment by choosing a Python interpreter and making a ./venv directory to hold it:

python3 -m venv ./venv

• Activate the virtual environment:

source ./venv/bin/activate

3. Install the necessary dependencies:

make install

This will install the bot and all of its requirements. Note that this bot should be used with python 3.7, 3.8 or 3.9.

Test Mbaza on your terminal:

1. Train the bot:

By default, you won't have a trained model, so to train a model, use:

rasa train

The above command will take a significant amount of memory to train, if you want to train it faster, try the training command with:

rasa train --augmentation 0

2. Start the action server:

Prerequisite:

Ensure the following endpoint url in your endpoints.yml. This parameter may be different based on the environment where the bot is running. For local testing, set/verify the following:

action_endpoint:
  url: "http://localhost:5055/webhook"

rasa run actions

3. Run the bot (on another terminal):

rasa shell

Test Sara on Rasa X:

Prerequisite: Ensure you install Rasa X first

Follow the above steps to start the rasa action server, then run the bot with the following (instead):

rasa x