SRA 2：面向高效扩散训练的变分自编码器自表征对齐
SRA 2: Variational Autoencoder Self-Representation Alignment for Efficient Diffusion Training
利用预训练 VAE 的重建特征作为内在监督信号，通过轻量投影层对齐扩散 Transformer 的中间潜变量，无需外部编码器即可加速训练收敛，为扩散模型训练提供了一种低成本的正则化手段。
原文依据 SRA 2 aligns the intermediate latent features of diffusion transformers with VAE features via a lightweight projection layer
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去噪的扩散 Transformer 虽然生成性能强大，但训练收敛效率低下。现有解决方法，如 REPA（依赖外部表征编码器）或 SRA（需要双模型设置），由于外部依赖，在训练过程中不可避免地带来沉重的计算开销。为解决这些挑战，本文提出 SRA 2，一种用于高效扩散训练的轻量级内在自表征对齐框架。SRA 2 利用现成的预训练变分自编码器（VAE）特征：其重建特性确保了对丰富纹理细节、结构模式和基本语义信息等视觉先验的固有编码。具体而言，SRA 2 通过轻量投影层将扩散 Transformer 的中间潜变量特征与 VAE 特征对齐，并由特征对齐损失进行监督。该设计无需额外的表征编码器或双模型维护即可加速训练，形成简单而有效的流程。大量实验表明，与原始扩散 Transformer 相比，SRA 2 同时提升了生成质量和训练收敛速度，达到或超越了最优的加速方法，并且仅增加 4% 的额外 GFLOPs，外部引导模型零额外成本。
英文原文摘要 1376 字符
Denoising-based diffusion transformers, despite their strong generation performance, suffer from inefficient training convergence. Existing methods addressing this issue, such as REPA (relying on external representation encoders) or SRA (requiring dual-model setups), inevitably incur heavy computational overhead during training due to external dependencies. To tackle these challenges, this paper proposes SRA 2, a lightweight intrinsic self-representation alignment framework for efficient diffusion training. SRA 2 leverages off-the-shelf pre-trained Variational Autoencoder (VAE) features: their reconstruction property ensures inherent encoding of visual priors like rich texture details, structural patterns, and basic semantic information. Specifically, SRA 2 aligns the intermediate latent features of diffusion transformers with VAE features via a lightweight projection layer, supervised by a feature alignment loss. This design accelerates training without extra representation encoders or dual-model maintenance, resulting in a simple yet effective pipeline. Extensive experiments demonstrate that SRA 2 improves both generation quality and training convergence speed compared to vanilla diffusion transformers, matches or outperforms state-of-the-art acceleration methods, and incurs merely 4% extra GFLOPs with zero additional cost for external guidance models.