发表机构
University of California, San Diego; Qualcomm AI Research（加利福尼亚大学圣迭戈分校; 高通人工智能研究中心）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出扩散元提示（DMP）模型，利用扩散模型对提示分布建模，通过测试时引导策略提升采样稳定性，可提升多任务泛化能力，降低存储与推理成本，在复合分类等任务上优于现有方法。
AI 中文摘要
提示学习是适配基础模型的流行方法，但学习到的提示通常是特定任务的，无法泛化到新类别、新领域或任务组合。本文提出扩散元提示（Diffusion Meta-Prompt, DMP）模型，该框架利用扩散模型对学习到的提示分布进行建模。给定一组预先学习到的提示库，DMP的训练和采样无需访问原始任务示例或任务损失，可根据自然语言任务描述合成新提示。为提升采样稳定性，本文为DMP引入测试时引导策略，该策略在扩散采样过程中使用库中经训练选出的最优提示作为潜在锚点，无需重新训练DMP或访问测试类别。DMP可提升分类、检索和文本到图像生成任务的泛化能力，支持概念组合和负提示且无需显式训练；与提示检索方法相比，其存储和推理成本降低90%以上。在复合分类任务中，DMP在55组数据集对上相比现有元学习方法实现最高2.0%的平均增益，在Eurosat与Flowers等特定数据集对上增益高达8.5%；在分层分类任务中，DMP还可实现约2%-9%的跨任务泛化提升。本文进一步提供了理论保证，界定了从DMP采样得到的提示的期望任务损失。代码可访问：this https URL
英文摘要
Prompt learning is a popular method for adapting foundation models, but learned prompts are typically task-specific and fail to generalize to new classes, domains, or compositions of tasks. In this paper, we introduce a Diffusion Meta-Prompt (DMP) model , a framework that models the distribution of learned prompts using diffusion models. Given a repository of previously learned prompts, DMP is trained and sampled without access to the original task examples or task losses, and synthesizes new prompts conditioned on natural language task descriptions. To improve the sampling stability, we introduce a test-time steering strategy for DMP, which uses the best training-selected prompt in the repository as a latent anchor during diffusion sampling, without retraining the DMP or accessing test classes. DMP improves generalization across classification, retrieval and text-to-image generation tasks, supports concept composition and negative prompting without explicit training. It reduces storage and inference costs by over 90% compared to prompt retrieval methods. For composite classification, DMP achieves upto 2.0% average gain over prior meta-learning methods across 55 pairs of datasets with gains as high as 8.5% on specific pairs such as Eurosat and Flowers. DMP also enhances cross-task generalization with ~2-9% improvement for hierarchical classification task. We further provide a theoretical guarantee bounding the expected task loss of prompts sampled from a DMP. Code is available: https://github.com/DeepakSridhar/dmp
CommentsAccepted to NeurIPS 2026. Project page: https://deepaksridhar.github.io/dmp.github.io/. Code: https://github.com/DeepakSridhar/dmp
