@inproceedings{rahman-etal-2026-compressing,
    title = "Compressing Long Context into Answer-Aligned Memory Embeddings for {LLM} Inference",
    author = "Rahman, Md Mostafizer  and
      Amin, Md Faizul Ibne  and
      Mia, Md. Shahajada  and
      Watanobe, Yutaka  and
      Liu, Fang",
    editor = "Lee, Hung-Yi  and
      Dev, Sunipa  and
      Li, Jessy  and
      Elliott, Desmond",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
    month = oct,
    year = "2026",
    address = "Budapest, Hungary",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-emnlp.1590/",
    pages = "33852--33871",
    ISBN = "979-8-89176-462-0"
}