@misc{indiciae261313d0a070,
  title = {When Text Matters: Design Principles for Visual Token Pruning in Vision-Language Model},
  author = {Minchan Kang and Kyeonghye Park and Seoyoung Cho and Daeshik Kim and Yucheol Cho},
  year = {2026},
  url = {https://arxiv.org/abs/2609.34861},
  note = {Source identifier: 2609.34861}
}