@article{gearanefficientkvcachecompression,
  title = {GEAR: An Efficient KV Cache Compression Recipe for Near-Lossless Generative Inference of LLM},
  author = {Hao Kang and Qingru Zhang and Souvik Kundu and Geonhwa Jeong and Zaoxing Liu and Tushar Krishna and Tuo Zhao},
  year = {2024},
  eprint = {2403.05527},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2403.05527v4},
}