ChunkKV: Semantic-Preserving KV Cache Compression for Efficient Long-Context LLM Inference

@misc{li2025chunkkvsemanticpreservingkv,
      title={ChunkKV: Semantic-Preserving KV Cache Compression for Efficient Long-Context LLM Inference}, 
      author={Bo Li and Xiang Liu and Zeyu Li and Zhenheng Tang and Xiaowen Chu and Xuming Hu and Peijie Dong},
      year={2025},
      eprint={2502.00299},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2502.00299}, 
}