Premier-TACO is a Few-Shot Policy Learner: Pretraining Multitask Representation via Temporal Action-Driven Contrastive Loss.
Ruijie ZhengYongyuan LiangXiyao WangShuang MaHal Daumé IIIHuazhe XuJohn LangfordPraveen PalanisamyKalyan Shankar BasuFurong HuangPublished in: CoRR (2024)