在线持续学习中，仅用25%数据实现与全量训练相当效果。
OASIS: Online Sample Selection for Continual Visual Instruction Tuning
- 基于历史数据动态评估样本价值，避免固定采样数量缺陷。
- 通过迭代更新筛选得分，降低选中样本的冗余度。
- 适合实时性强、数据持续流入的视觉指令微调场景。
在持续指令微调（CIT）场景中，新指令数据以在线流式方式持续到达，大规模数据带来的训练延迟严重阻碍实时适应。数据选择可缓解此开销，但现有策略多依赖预训练参考模型，这在CIT设置中不切实际，因未来数据未知。近期无参考模型的在线样本选择方法虽解决此问题，但通常每批次固定选择样本数（如top-k），易受分布偏移影响，因各批次信息量差异大。为此，我们提出OASIS，一种面向CIT的自适应在线样本选择方法：（1）通过估计每个样本相对于所有已见数据的信息量来选择有信息量的样本，突破批次级约束；（2）通过迭代更新筛选得分，最小化所选样本的信息冗余。在多个大型基础模型上的实验表明，OASIS仅使用25%的数据，即达到全量数据训练的性能水平，并优于当前最优采样方法。
原文摘要 · Abstract (English)
In continual instruction tuning (CIT) scenarios, where new instruction tuning data continuously arrive in an online streaming manner, training delays from large-scale data significantly hinder real-time adaptation. Data selection can mitigate this overhead, but existing strategies often rely on pretrained reference models, which are impractical in CIT setups since future data are unknown. Recent reference model-free online sample selection methods address this, but typically select a fixed number of samples per batch (e.g., top-k), making them vulnerable to distribution shifts where informativeness varies across batches. To address these limitations, we propose OASIS, an adaptive online sample selection approach for CIT that (1) selects informative samples by estimating each sample's informativeness relative to all previously seen data, beyond batch-level constraints, and (2) minimizes informative redundancy of selected samples through iterative selection score updates. Experiments on various large foundation models show that OASIS, using only 25 percent of the data, achieves comparable performance to full-data training and outperforms the state-of-the-art sampling methods.
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。