About Plato RDS

Plato Research Dialogue System is a platform for building, training, and deploying conversational AI agents that allows us to conduct state of the art research in conversational AI, quickly create prototypes and demonstration systems, as well as facilitate conversational data collection. Read more

How does it work?


Clone the following repository: git@github.com:uber-research/

For MacOS:
brew install portaudio
brew install gmp
pip install pyaudio

For Ubuntu/Debian:
sudo apt-get install python3-pyaudio

For Windows:
Nothing needs to be pre-installed

After the steps above, install plato in editable mode: pip install -e .

Running Single Agents

To run a simulation:
plato run --config CamRest_user_simulator.yaml

To run a text based interaction:
plato run --config CamRest_text.yaml

To run a speech based interaction:
plato run --config CamRest_speech.yaml

To run a gui based interaction:
plato gui --config CamRest_GUI_speech.yaml

Training Multiple Agents

plato run --config MultiAgent_train.yaml

plato run --config MultiAgent_test.yaml

Training with Ludwig

Ludwig is a toolbox that allows to train and test
deep learning models without the need to write code. Briefly, you need to do the following:

1. Prepare your data in csv format.
2. Train your model with Ludwig
3. Update your Plato configuration file to include the path to the ludwig model
4. Run a Plato agent. For a walk-through, look at the User Guide