@inproceedings{she-etal-2025-token,
    title = "Token Level Routing Inference System for Edge Devices",
    author = "She, Jianshu  and
      Zheng, Wenhao  and
      Liu, Zhengzhong  and
      Wang, Hongyi  and
      Xing, Eric P.  and
      Yao, Huaxiu  and
      Ho, Qirong",
    editor = "Mishra, Pushkar  and
      Muresan, Smaranda  and
      Yu, Tao",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingestion-acl-25/2025.acl-demo.16/",
    pages = "159--166",
    ISBN = "979-8-89176-253-4",
    abstract = "The computational complexity of large language model (LLM) inference significantly constrains their deployment efficiency on edge devices. In contrast, small language models offer faster decoding and lower resource consumption but often suffer from degraded response quality and heightened susceptibility to hallucinations. To address this trade-off, collaborative decoding, in which a large model assists in generating critical tokens, has emerged as a promising solution. This paradigm leverages the strengths of both model types by enabling high-quality inference through selective intervention of the large model, while maintaining the speed and efficiency of the smaller model. In this work, we present a novel collaborative decoding inference system that allows small models to perform on-device inference while selectively consulting a cloud-based large model for critical token generation. Remarkably, the system achieves a 60{\%} performance gain on CommonsenseQA using only a 0.5B model on an M1 MacBook, with under 7{\%} of tokens generation uploaded to the large model in the cloud."
}