VERSES AI

Modelling Abstractions in the Context of Deep Reinforcement Learning and Rate-distortion Theory

project page