NAMI：通过桥接渐进式整流流 Transformer 实现高效图像生成
NAMI: Efficient Image Generation via Bridged Progressive Rectified Flow Transformers
NAMI 通过按分辨率分段使用不同层数的 DiT 并用 BridgeFlow 模块对齐各阶段流，在 1024×1024 生成上减少 64% 推理时间，为视觉生成模型提供了一种可复用的高效多分辨率训练与推理范式。
原文依据 reduces inference time by 64% for generating 1024x1024 resolution images
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论基于流的 Transformer 模型在图像生成上已达到最优性能，但由于参数量大，常面临推理延迟高和计算成本高的问题。为了在不牺牲质量的前提下提升推理效率，我们提出了桥接渐进式整流流 Transformer（NAMI），从时间、空间和架构三个维度分解生成过程。我们根据分辨率将整流流划分为不同阶段，并使用 BridgeFlow 模块连接它们。在低分辨率阶段使用较少的 Transformer 层来生成图像布局和概念轮廓，随着分辨率提高逐步增加层数。实验表明，我们的方法在保证生成质量的同时实现了快速收敛并减少了推理时间。本文的主要贡献总结如下：（1）我们引入了桥接渐进式整流流 Transformer，支持多分辨率训练，加速模型收敛；（2）NAMI 利用分段流和扩散 Transformer（DiT）的空间级联来快速生成图像，生成 1024×1024 分辨率图像时推理时间减少 64%；（3）我们提出了 BridgeFlow 模块来对齐不同阶段之间的流；（4）我们提出了 NAMI-1K 基准来评估人类偏好性能，旨在减轻分布偏差并全面评估模型有效性。结果表明，我们的模型与最优模型相比具有竞争力。
英文原文摘要 1554 字符
Flow-based Transformer models have achieved state-of-the-art image generation performance, but often suffer from high inference latency and computational cost due to their large parameter sizes. To improve inference efficiency without compromising quality, we propose Bridged Progressive Rectified Flow Transformers (NAMI), which decompose the generation process across temporal, spatial, and architectural demensions. We divide the rectified flow into different stages according to resolution, and use a BridgeFlow module to connect them. Fewer Transformer layers are used at low-resolution stages to generate image layouts and concept contours, and more layers are progressively added as the resolution increases. Experiments demonstrate that our approach achieves fast convergence and reduces inference time while ensuring generation quality. The main contributions of this paper are summarized as follows: (1) We introduce Bridged Progressive Rectified Flow Transformers that enable multi-resolution training, accelerating model convergence; (2) NAMI leverages piecewise flow and spatial cascading of Diffusion Transformer (DiT) to rapidly generate images, reducing inference time by 64% for generating 1024x1024 resolution images; (3) We propose a BridgeFlow module to align flows between different stages; (4) We propose the NAMI-1K benchmark to evaluate human preference performance, aiming to mitigate distributional bias and comprehensively assess model effectiveness. The results show that our model is competitive with state-of-the-art models.