LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning.

Kaihong Wang, Donghyun Kim, Margrit Betke

Browse the full ICCV paper archive.