TEA-Prompt：用行为锚定实现无样本回放的持续学习
Anchor Behavior, Not Parameters: Rehearsal-Free Continual Learning with a Single Shared Prompt
持续学习圈子的新思路：不改参数、只锁住注意力行为，就能让单条 prompt 学新任务不忘旧类别。0.38M 参数就在 CIFAR-100、ImageNet-R、CUB-200 上拿到 SOTA，还不存旧数据，做 ViT 迁移学习的人可以直接看代码复现。
论文提出 TEA-Prompt，一种基于单共享提示（one-prompt）的持续学习方法，在冻结的预训练视觉 Transformer 上只需 0.
38M 参数。现有方法通过冻结任务提示、扩展组件或稀疏梯度门控来保护旧知识，参数上不动，但仍允许的漂移会破坏提示在旧类别上的功能。
TEA-Prompt 的思路是不锚定参数、锚定行为：一个 token 评估模块为每个提示 token 计算连续的、依赖输入的相关性分数并注入注意力 logits，让 token 在任务间软性分工；同时用注意力行为锚定损失，仅靠紧凑的高斯查询统计量（不存储旧图像）把每个 token 的注意力输出约束在类别相关查询区域附近。
在 CIFAR-100、ImageNet-R 和 CUB-200 上取得 SOTA 准确率且遗忘显著更低。代码已在 GitHub 开源。
Anchor Behavior, Not Parameters: Rehearsal-Free Continual Learning with a Single Shared Prompt
Prompt-based continual learning adapts a frozen pre-trained vision transformer by tuning a small set of prompt tokens. Recently, one-prompt methods which share one single prompt across all tasks make this paradigm maximally parameter efficient. The open question is how a shared prompt can keep the knowledge of old classes while being updated for new ones. Existing methods protect old knowledge in parameter space, where they freeze task-specific prompts, expand new components, or sparsely gate which tokens receive gradients. However, this method sacrifices plasticity, yet the drift it still permits in shared tokens can corrupt the prompt's function on past classes. What must be preserved is the prompt's behavior. We propose TEA-Prompt, which couples (i) a token evaluation module that assigns each prompt token a continuous, input-conditional relevance score injected into the attention logits, softly specializing tokens across tasks without discarding any of them, and (ii) an attention-behavior anchoring loss that constrains each token's attention output around class-specific query regions using only compact Gaussian query statistics without stored old images, weighted by the token's past relevance. Anchoring the attention value allows the parameters to continue updating freely, as long as their interaction remains unchanged. TEA-Prompt achieves state-of-the-art accuracy with markedly lower forgetting on CIFAR-100, ImageNet-R, and CUB-200 with only 0.38M parameters. Code is publicy available at https://github.com/cuonga1cvp/TEA-Prompt.git
This report is categorized under AITOP's "Papers" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include TEA-Prompt, ViT, 持续学习, 提示词工程.
Source attribution: curated from arXiv cs.LG (@Duy-Cuong Bui, Trung-Anh Dang, Vincent Nguyen, Duc V. Le, Ngoc-Son Vu).
