1 paper · 1 filter
Ozan Arkan Can, Deniz Yuret
In this paper, we present a state-of-the-art model and introduce a new dataset for grounded language learning. Our goal is to develop a model that can learn to follow new instructi…