Implementing RAG in Rasa with Faiss and Gemini

Steps

1. Create a Codespace:

   • Click on the green "Code" button on this page, then scroll down to "Codespaces".
   • Click on "Create codespace on main".

2. Set Up Environment:

   • In the codespace, rename the .env.example file from this repo to .env and add your license key to that file.

     RASA_PRO_LICENSE='your_rasa_pro_license_key_here'
     GEMINI_API_KEY='your_rasa_pro_license_key_here'
     

   • Set this environment variables by running

     source .env
     

   • Activate your python environment by running

     source .venv/bin/activate
     

3. Install Requirements

   • In the terminal, run:

     uv pip install -r requirements.txt
     

4. Train the Model:

   • In the terminal, run:

     rasa train
     

5. Talk to your Bot:

   • In the terminal, run

     rasa inspect
     

     GitHub will show a notification, click on the green button to view the inspector where you can chat with your assistant.

6. Run Custom Actions: In Rasa 3.10 and later, custom actions are automatically run as part of your running assistant. To double-check that this is set up correctly, ensure that your endpoints.yml file contains the following configuration:

   action_endpoint:
      actions_module: "actions" # path to your actions package
   

   Then re-run your assistant via rasa inspect every time you make changes to your custom actions.