解决视觉生成中概念响应不稳定的难题，提出在线平衡方法。
Imbalance in Balance: Online Concept Balancing in Generation Models
- 设计在线概念等化损失，动态优化生成模型对复杂概念的响应。
- 在三个测试集上显著提升基线模型的概念生成能力，最高提升23.7%。
- 无需离线数据处理，代码改动极小，适合快速集成到现有模型中。
在视觉生成任务中，复杂概念的响应和组合常缺乏稳定性且易出错，这一问题尚未得到充分研究。本文通过精心设计的实验探索了导致概念响应不佳的因果因素，并提出一种基于概念级别的等化损失函数（IMBA loss）。所提方法为在线式，无需离线数据处理，代码改动极少。在新构建的复杂概念基准 Inert-CompBench 及两个公开测试集上，该方法显著提升了基线模型的概念响应能力，性能媲美先进方法。相关代码已开源：https://github.com/KwaiVGI/IMBA-Loss。
原文摘要 · Abstract (English)
In visual generation tasks, the responses and combinations of complex concepts often lack stability and are error-prone, which remains an under-explored area. In this paper, we attempt to explore the causal factors for poor concept responses through elaborately designed experiments. We also design a concept-wise equalization loss function (IMBA loss) to address this issue. Our proposed method is online, eliminating the need for offline dataset processing, and requires minimal code changes. In our newly proposed complex concept benchmark Inert-CompBench and two other public test sets, our method significantly enhances the concept response capability of baseline models and yields highly competitive results with only a few codes released at https://github.com/KwaiVGI/IMBA-Loss.
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。