@article{keydiffkeysimilaritybasedkvcache,
  title = {KeyDiff: Key Similarity-Based KV Cache Eviction for Long-Context LLM Inference in Resource-Constrained Environments},
  author = {Junyoung Park and Dalton Jones and Matthew J Morse and Raghavv Goel and Mingu Lee and Chris Lott},
  year = {2025},
  eprint = {2504.15364},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2504.15364v3},
}