Zero-Shot Reinforcement Learning via Function Encoders

This repo is the official codebase for the paper "Zero-Shot Reinforcement Learning via Function Encoders".

The code consists of the following four examples, each of which is effectively a seperate project with its own dependencies:

1. Minimum working example
   1. Start here for the most basic example of training a function encoder on simple data.
2. Multi-task RL
3. Multi-agent RL
4. Hidden-parameter System Identification

See each project directory for its own README.

Warning:

Each project was created independently, so it is best to run them with the respective project as the working directly. Otherwise, there may be dependency issues.

The best approach would be to create a virtual env in each of the project directories and install the dependencies there.