adaptively allocates the parameter budget among weight matrices according to their importance score
effective pruning of unimportant updates, which reduces their parameter budget while circumventing intensive exact SVD computations
AdaLoRa
Creator
Creator
Seonglae ChoCreated
Created
2023 Jul 9 5:43Editor
Editor
Seonglae ChoEdited
Edited
2023 Jul 9 13:8Refs
Refs