1 paper
Yusaku Takama, Ning Ding, Tatsuya Yokota +1
Existing architectures of multi-domain learning have two types of adapters: shared LoRA for all domains and domain-specific LoRA for each particular domain. However, it remains unc…