@article{medadynamickvcacheallocationfor,
  title = {MEDA: Dynamic KV Cache Allocation for Efficient Multimodal Long-Context Inference},
  author = {Zhongwei Wan and Hui Shen and Xin Wang and Che Liu and Zheda Mai and Mi Zhang},
  year = {2025},
  eprint = {2502.17599},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2502.17599v1},
}