发表机构
RICS; Daegu Gyeongbuk Institute of Science and Technology; IPAI; Samsung Advanced Institute of Technology, Samsung Electronics Co., Ltd; AIIS; Seoul National University（RICS; 大邱庆北科学技术院; IPAI; 三星电子三星综合技术院; AIIS; 首尔大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本综述提出统一框架，按任务信息编码位置（权重、提示、嵌入）分类大型语言模型的高效适配方法，分析各范式优劣与关系，并指出未来开放问题。
AI 中文摘要
随着大型语言模型在各类下游任务中的部署日益广泛，高效任务适配已成为一个核心挑战。为此，研究者提出了多种任务适配方法，涵盖参数高效微调、上下文学习和嵌入注入方法。然而，这些研究方向大多在各自范式内独立发展，导致跨范式的关系和权衡未被充分探索，尤其是对于近期出现的基于嵌入的适配方法。本综述提出了一个统一框架，根据任务信息的编码位置和方式对任务适配方法进行分类：模型权重、输入提示或注入的任务嵌入。我们提供了一个综合分类体系，整合了这些范式，分析了它们的主要优势和局限性，以解释不同适配范式如何演变，阐明了范式间的关系，并指出了未来研究中的开放问题。
英文摘要
As large language models are increasingly deployed across diverse downstream tasks, efficient task adaptation has emerged as a central challenge. In response, a wide range of task adaptation methods have been proposed, spanning parameter-efficient fine-tuning, in-context learning, and embedding-injection approaches. However, these lines of work have largely evolved within individual paradigms, leaving their cross-paradigm relationships and trade-offs underexplored, especially for recently emerging embedding-based adaptations. This survey presents a unified framework that categorizes task adaptation methods by where and how task information is encoded: model weights, input prompts, or injected task embeddings. We provide a comprehensive taxonomy that integrates these paradigms, analyze their key strengths and limitations to explain how different adaptation paradigms have evolved, clarify relationships across paradigms, and highlight open problems for future research.
CommentsAccepted by AACL-IJCNLP 2026 Main
