2026
SplitLoRA: Balancing Stability and Plasticity in Continual Learning Through Gradient Space Splitting
ICLR 2026poster
Continual Learning (CL) requires a model to learn multiple tasks in sequence while maintaining both stability—preserving knowledge from previously learned tasks, and plasticity—effectively learning new tasks. Orthogonal projection has emerged as an effective and popular paradigm in CL, where it part…