Sep 2026
Towards Understanding Token Evolution in Continual Learning of MLLMs
This work revisits catastrophic forgetting from an interpretability perspective and introduces correctness-guided layer alignment (CLA), which allows the new model to inherit multi-layer semantic consistency from its predecessor while learning from ground-truth supervision.
Xu-Sheng Cao, Yi-Fan Meng, Jia-Ze Li et al.
· International Journal of Com... · 0 citations