Calibri：通过参数高效校准增强扩散变换器

摘要

本文揭示了扩散变换器（DiTs）在提升生成任务性能方面的潜在能力。通过对去噪过程的深入分析，我们证明仅需引入单个可学习的缩放参数即可显著增强DiT模块的性能。基于这一发现，我们提出Calibri——一种参数高效的方法，通过优化校准DiT组件来提升生成质量。该方法将DiT校准构建为黑盒奖励优化问题，采用进化算法高效求解，仅需调整约100个参数。实验结果表明，尽管采用轻量化设计，Calibri能在各类文生图模型中持续提升性能。值得注意的是，该方法在保持高质量输出的同时，还能减少图像生成所需的推理步数。

English

In this paper, we uncover the hidden potential of Diffusion Transformers (DiTs) to significantly enhance generative tasks. Through an in-depth analysis of the denoising process, we demonstrate that introducing a single learned scaling parameter can significantly improve the performance of DiT blocks. Building on this insight, we propose Calibri, a parameter-efficient approach that optimally calibrates DiT components to elevate generative quality. Calibri frames DiT calibration as a black-box reward optimization problem, which is efficiently solved using an evolutionary algorithm and modifies just ~100 parameters. Experimental results reveal that despite its lightweight design, Calibri consistently improves performance across various text-to-image models. Notably, Calibri also reduces the inference steps required for image generation, all while maintaining high-quality outputs.

Calibri：通过参数高效校准增强扩散变换器

Calibri: Enhancing Diffusion Transformers via Parameter-Efficient Calibration

摘要

Support