@misc{indiciae47f24a702f6a,
  title = {VISion On Request: Enhanced VLLM efficiency with sparse, dynamically selected, vision-language interactions},
  author = {Adrian Bulat and Alberto Baldrati and Ioannis Maniadis Metaxas and Yassine Ouali and Georgios Tzimiropoulos},
  year = {2026},
  url = {https://arxiv.org/abs/2603.23495},
  note = {Source identifier: 2603.23495}
}