像素扩散新参数化 MFlow 与卷积精修模块 ConvRefiner
Rethinking the design space of Vanilla Pixel Diffusion Transformers
arXiv 上的新论文，像素扩散模型不再二选一，MFlow 让模型自己学该用哪种预测方式，ConvRefiner 还能低成本补上高频细节，做生成方向的可以看看。
一篇新论文提出 MFlow，一种自适应参数化，可在 x0-prediction 和 u-prediction 之间按数据主方向插值，并端到端随扩散训练学习。实验覆盖 ImageNet、UCF-101 视频和大规模文生图，结果显示最优参数化随模型容量系统性变化，容量越大越偏向 velocity prediction。论文还发现像素扩散 Transformer 普遍存在高频细节缺失，为此提出 ConvRefiner：冻结扩散主干，事后用对抗损失训练的轻量卷积模块，少量额外训练即可恢复高频统计且保持 FID。
Rethinking the design space of Vanilla Pixel Diffusion Transformers
Pixel-space diffusion models have recently re-emerged as a compelling alternative to latent diffusion, avoiding the lossy latent bottleneck and separately trained autoencoder. JiT shows that a vanilla diffusion transformer can generate directly in pixel space by replacing the commonly used velocity prediction ($\boldsymbol{u}$-prediction) with clean image prediction ($\boldsymbol{x_0}$-prediction). In this work, we revisit this choice and introduce MFlow, an adaptive parameterization that learns to interpolate between $\boldsymbol{x_0}$- and $\boldsymbol{u}$-prediction across principal directions of the data, learned end-to-end with diffusion training. We find that the preferred parameterization varies systematically with the data representation and model capacity -- notably, increasing model capacity shifts preference toward velocity prediction. We evaluate MFlow across ImageNet, UCF-101 video, and large-scale text-to-image generation, demonstrating competitive or improved generative performance across diverse settings. In addition, we find that vanilla pixel diffusion transformers, irrespective of parameterization, retain a persistent high-frequency deficit relative to real images. To address this, we introduce ConvRefiner, a lightweight convolutional module trained post-hoc with an adversarial loss while keeping the diffusion backbone frozen. With only a small amount of additional training, ConvRefiner restores the high-frequency statistics of generated images while preserving FID. Project webpage: https://soumik-kanad.github.io/mflow-diffusion/.
