@article{sparkvoverheadawarekvcacheloadingforeffi,
  title = {SparKV: Overhead-Aware KV Cache Loading for Efficient On-Device LLM Inference},
  author = {Hongyao Liu and Liuqun Zhai and Junyi Wang and Zhengru Fang},
  year = {2026},
  eprint = {2604.21231},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2604.21231},
}