1 paper · 1 filter
Sung June Kim, Sangpil Kim, Honglak Lee
On-policy exploration is a crucial component for training robust Vision-Language Navigation agents, as it exposes the policy to a broader state distribution. However, such explorat…