- We will use Anaconda for python distribution management. You can follow the official documentation here to install it in your system.
- Download and install Microsoft Visual C++ from the link here.
Installation and Setup
- Open the Anaconda command prompt.
- Create a folder for the RASA project and cd into it.
- Create a new Python environment that will be used for RASA installation.
conda create --name rasaEnv python==3.7.6
Installation of Rasa Open Source using pip (requires Python 3.7, or 3.8).
- Activate the newly created environment rasaEnv
conda activate rasaEnv
- Any packages you install now will be only applied to rasaEnv.
- Let’s install all the dependencies. Both conda and pip can be used to install packages however, not all packages are available in conda.
conda install ujson
conda install tensorflow
pip install -U pip
pip install rasa==2.8.17
The command above will upgrade your pip to the latest version. We recommend using pip 21.3 or later due to performance improvements. Previous versions would take long time to resolve dependencies.
- Optional: You can install rasa X, a new addition to the stack that helps you test your application from browser
pip install -U rasa-x==1.0.0 --extra-index-url https://pypi.rasa.com/simple
- Initialize the RASA project
- select all default values and you are all set!
- Training and generating models
- If you have defined custom actions in actions.py, start the action server with the below command
rasa run actions
- Uncomment the below line in endpoints.yml to enable RASA HTTP endpoints for REST communication with your RASA server.
- Running RASA shell for testing
- Running RASA x
- Running RASA server
A complete list of CLI commands can be found here!