byMingda Chen, Jingfei Du, Ramakanth Pasunuru, Todor Mihaylov, Srini Iyer, Veselin Stoyanov and Zornitsa Kozareva This paper proposes to use self-supervision (MLM, NSP, CL, etc.) between pre-training and downstream usage to teach the LM to perform in-context learning. Analysis reveals that...
byCe Zhou, Qian Li, Chen Li, Jun Yu, Yixin Liu, Guangjing Wang, Kai Zhang, Cheng Ji et al. Complex QA and language models hybrid architectures, Survey, byXavier Daull, Patrice Bellot, Emmanuel Bruno, Vincent Martin and Elisabeth Murisasco Augmented Language Models: a Survey, byGr'egoire...