让大模型在真实任务流中实时优化提示，提升多场景适应能力
EEVEE: Towards Test-time Prompt Learning in the Real World for Self-Improving Agents
- 用路由机制将不同任务输入分组，匹配最优提示配置
- 跨数据集测试时平均得分提升10.38至24.32点，领先现有方法37.2%以上
- 适合需要持续学习和自适应的智能代理系统
本文提出EEVEE，首个面向多数据集的测试时提示学习框架，使大语言模型代理能在真实世界任务流中实现测试时提示学习。现有方法多局限于单一数据集，而现实应用需处理来自多个数据集、领域和任务分布的异构输入流，限制了实际可用性。为缓解跨数据集干扰，EEVEE引入路由模块，将输入划分为任务簇并分配至合适的提示配置。该设计通过路由-提示协同进化策略优化，采用交替学习路由与提示的机制以解决二者依赖关系。在多个数据集上的实验表明，该框架在异构数据流下提升鲁棒性，同时保持单基准学习能力和效率。具体而言，相比Qwen3-4B-Instruct与DeepSeek-V3.2，EEVEE在多基准平均得分上分别提升10.38和24.32点，超越SOTA方法GEPA与ACE达37.2%和48.2%。
原文摘要 · Abstract (English)
In this paper, we propose EEVEE, the first multi-dataset test-time prompt learning framework for LLM agents, enabling test-time prompt learning under real-world task streams. Existing methods are largely designed for single-dataset settings, while real-world applications require models to handle heterogeneous input streams drawn from multiple datasets, domains, and task distributions, limiting their practical applicability. To mitigate cross-dataset interference, EEVEE introduces a router that partitions incoming inputs into task clusters and assigns them to suitable prompt configurations. This design is optimized via a router-prompt co-evolution strategy, which employs interleaved router and prompt learning phases to address their mutual dependency. Experiments across multiple datasets demonstrate that the framework improves robustness under heterogeneous data streams while maintaining single-benchmark learning capability and efficiency. Specifically, EEVEE improves average multi-benchmark scores by 10.38 and 24.32 points over Qwen3-4B-Instruct and DeepSeek-V3.2, surpassing SOTA methods GEPA and ACE by up to 37.2% and 48.2%.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。