发帖
雷达卡
Grounding Language to Entities and Dynamics
for Generalization in Reinforcement Learning
Austin W. Hanjie 1 Victor Zhong 2 Karthik Narasimhan 1
Abstract unseen entities or dynamics (Narasimhan et al., 2018; Zhong
We investigate the use of natural language to drive et al., 2020). While existing methods demonstrate success-
the generalization of control policies and intro- ful transfer to new settings, they assume a groun ...
京公网安备 11010802022788号
