Setting Up Your First Rasa Project

Rasa is an open source framework for creating conversational AI and chatbots. If you are a looking to configure your first project in Rasa, you’ve come to the right place. In this blog, Iwe will set up a Rasa project from the ground up, step by step.

What Is Rasa?

Before diving in, let’s clarify what Rasa is. Rasa consists of two primary components:

1. Rasa Open Source: A framework for building natural language understanding (NLU) and dialogue management.

2. Rasa X: A tool for improving and managing your assistant over time.

Rasa is written in Python and allows for flexible customisation, making it a popular choice among developers.

Prerequisites

To set up Rasa, you need:

• Python 3.8 or 3.9 (Rasa currently doesn’t support 3.10+)

• pip (Python package manager)

• Virtual Environment (Optional but recommended)

Familiarity with Python and basic command-line usage is helpful but not required.

Step 1: Installing Python and Creating Virtual Environment

1. Download Python:

• Head over to the Python website and download Python 3.8 or 3.9.

1. Create a Virtual Environment: Using a virtual environment keeps your Rasa project dependencies isolated from your global Python setup.

   python -m venv venv
   source venv/bin/activate

Step 2: Install Rasa

1. Install Rasa via pip:

   pip install rasa

   1. Verify the installation:

   rasa --version

You should see the Rasa version and Python version displayed.

Step 3: Create Your First Rasa Project

Now, let’s create your Rasa project:

1. Run the following command:

   rasa init

   1. Follow the prompts:

• Rasa will set up a new project directory with the following structure:

  my_project/
  ├── actions/
  ├── data/
  ├── models/
  ├── tests/
  ├── config.yml
  ├── credentials.yml
  ├── domain.yml
  └── endpoints.yml

• We will be prompted to train a model and test your assistant. Go ahead and try it!

Step 4: Understand the Key Files in Rasa

Here’s a breakdown of the key files in your project:

• domain.yml: Defines your bot’s personality, intents, responses, and entities.

• data/nlu.yml: Contains training examples for intent recognition.

• data/stories.yml: Defines conversation flows to train the dialogue model.

• config.yml: Specifies the machine learning pipeline for intent classification and entity recognition.

• endpoints.yml: Configures where to find external services (e.g., action server).

• credentials.yml: Configures third-party integrations like Slack or Telegram.

Step 5: Train Your Rasa Model

1. Use the following command to train your assistant if not already trained:

   rasa train

   1. This will generate a model file in the models/ directory, ready to power your chatbot!

Step 6: Test Your Assistant

1. Test your assistant locally:

   rasa shell

   1. Type in some messages to see how the assistant responds. For example:

> Hi Hello! 

Hey! How are you?

Step 7: Add Custom Actions

Want your bot to perform actions like fetching data from an API? Add custom actions!

1. Open the actions/ directory and create a Python file (e.g., actions.py).

2. Write your custom action:

   from rasa_sdk import Action
   from rasa_sdk.executor import CollectingDispatcher

   class ActionHelloWorld(Action):
   def name(self):
   return "action_hello_world"

   def run(self, dispatcher, tracker, domain):
       dispatcher.utter_message(text="Hello, world!")
       return []
   

   1. Update domain.yml to include your action:

   actions:

   • action_hello_world
     1. Start the action server:

   rasa run actions

References 🔗

• Rasa Official Documentation

• Python Download

• Virtualenv Documentation

Happy coding