@misc{indiciae05da06f0bc28,
  title = {Stabilizing the Modality Gap: Null-Space Projected Continual Learning for Vision-Language Models},
  author = {Tiantian Peng and Yuyang Liu and Shuo Yang and Qiuhe Hong and YongHong Tian},
  year = {2026},
  url = {https://arxiv.org/abs/2507.19839},
  note = {Source identifier: 2507.19839}
}