@misc{indiciae6f3803af3a0b,
  title = {TRIO: Token Reduction via Inference-Objective Guidance for Efficient Vision-Language Models},
  author = {Haokui Zhang and Congyang Ou and Dawei Yan and Peng Wang and Qingsen Yan and Yu Zhang and Ying Li and Rong Xiao},
  year = {2026},
  url = {https://arxiv.org/abs/2602.04657},
  note = {Source identifier: 2602.04657}
}