1 paper · 1 filter
Huahui Yi, Wei Xu, Ziyuan Qin +4
Existing prompt-based approaches have demonstrated impressive performance in continual learning, leveraging pre-trained large-scale models for classification tasks; however, the ti…