摘要: 概述 Learning form the Void (LfVoid) 根据给定的language instruction对observation进行appearance-based and structure-based修改得到goal images,为RL提供奖励信号。提升了example-bas 阅读全文
posted @ 2023-11-28 15:53 Eirrac 阅读(93) 评论(0) 推荐(0)