@inproceedings{wang-etal-2025-fier,
    title = "{FIER}: Fine-Grained and Efficient {KV} Cache Retrieval for Long-context {LLM} Inference",
    author = "Wang, Dongwei  and
      Liu, Zijie  and
      Wang, Song  and
      Ren, Yuxin  and
      Deng, Jianing  and
      Hu, Jingtong  and
      Chen, Tianlong  and
      Yang, Huanrui",
    editor = "Christodoulopoulos, Christos  and
      Chakraborty, Tanmoy  and
      Rose, Carolyn  and
      Peng, Violet",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
    month = nov,
    year = "2025",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/author-page-yu-wang-polytechnic/2025.findings-emnlp.515/",
    doi = "10.18653/v1/2025.findings-emnlp.515",
    pages = "9702--9713",
    ISBN = "979-8-89176-335-7",
    abstract = "The Key-Value (KV) cache reading latency increases significantly with context lengths, hindering the efficiency of long-context LLM inference. To address this, previous works propose retaining a small fraction of KV cache based on token importance. For example, KV eviction uses static heuristics to retain tokens, while KV retrieval dynamically selects query-relevant tokens for more adaptive cache management. However, we observe that important tokens are often sparsely distributed across the long context. This sparsity makes existing page-level KV retrieval inaccurate, as each page may include irrelevant tokens and miss critical ones. In this work, we propose Fier, a **Fi**ne-Grained and **E**fficient KV cache **R**etrieval method. Fier uses 1-bit quantized keys to estimate the importance of each token, resulting in efficient and precise retrieval. Experiments show that Fier matches full KV performance using only 11{\%} of the cache budget across various long-context tasks, reducing decoding latency by 1.2$\times$ to 1.5$\times$."
}