Fetching the paper…

ReWiND: Language-Guided Rewards Teach Robot Policies without New Demonstrations · Around