Octopus: History-Free Gradient Orthogonalization for Continual Learning in Multimodal Large Language Models

@inproceedings{liu2026octopus,
  title = {Octopus: History-Free Gradient Orthogonalization for Continual Learning in Multimodal Large Language Models},
  author = {Yuehao Liu and Shanyan Guan and Weijia Zhang and Xuanming Shang and Yanhao Ge and Wei Li and Chao Ma},
  year = {2026},
  booktitle = {CVPR},
  eprint = {2605.14938},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.14938},
}