Exploiting Language Instructions for Interpretable and Compositional Reinforcement Learning

Attempt to interpret the latent space from an RL agent to identify its current objective in a complex language instruction. This work followed my master thesis at the UvA.

Link to the paper: here.