@article{improvingmodelrepresentationandreducingk,
  title = {Improving Model Representation and Reducing KV Cache via Skip Connections with First Value Heads},
  author = {Zhoutong Wu and Yuan Zhang and Yiming Dong and Chenheng Zhang and Cong Fang and Kun Yuan and Zhouchen Lin},
  year = {2025},
  eprint = {2510.16807},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2510.16807},
}