Skip to main content
Search

# ROSE

1 related articles

Subscribe to updates
  1. arXiv

    ROSE keeps multimodal models from forgetting as edits pile up: orthogonal subspace updates hold VLKEB reliability at 95.24 after 1,000 edits

    Multimodal large language models need continuous factual updates, yet sequential edits erode pretrained abilities and overwrite earlier changes. ROSE confines each weight update to the orthogonal complement of the existing knowledge subspace, uses a dynamic risk mask to protect parameters critical for general capability, and at inference decides by prototype similarity whether to activate the edited branch. On VLKEB, ROSE's reliability falls from 100.00 to 95.24 as edits grow from 1 to 1,000, while FT-L drops 26.15 points under the same setting.

You’re all caught up