@inproceedings{yang-etal-2025-position,
    title = "Position Really Matters: Towards a Holistic Approach for Prompt Tuning",
    author = "Yang, Xianjun  and
      Cheng, Wei  and
      Zhao, Xujiang  and
      Yu, Wenchao  and
      Petzold, Linda Ruth  and
      Chen, Haifeng",
    editor = "Chiruzzo, Luis  and
      Ritter, Alan  and
      Wang, Lu",
    booktitle = "Findings of the Association for Computational Linguistics: NAACL 2025",
    month = apr,
    year = "2025",
    address = "Albuquerque, New Mexico",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/fix-sig-urls/2025.findings-naacl.474/",
    pages = "8501--8523",
    ISBN = "979-8-89176-195-7",
    abstract = "Prompt tuning is highly effective in efficiently extracting knowledge from foundation models, encompassing both language, vision, and vision-language models. However, the efficacy of employing fixed soft prompts with a \textit{predetermined position} for concatenation with inputs for all instances, irrespective of their inherent disparities, remains uncertain. Variables such as the position, length, and representations of prompts across diverse instances and tasks can substantially influence the performance of prompt tuning. We first provide a theoretical analysis, revealing that optimizing the position of the prompt to encompass the input can capture additional semantic information that traditional prefix or postfix prompt tuning methods fail to capture. Then, we present a holistic parametric prompt tuning strategy that dynamically determines different factors of prompts based on specific tasks or instances. Experimental results underscore the significant performance improvement achieved by dynamic prompt tuning across a wide range of tasks, including NLP, vision recognition, and vision-language tasks. Furthermore, we establish the universal applicability of our approach under full-data, few-shot, and multitask settings."
}