用漂移场提升表示自编码器，实现高效图像生成
Distilling Drifting Transformers with Representation Autoencoders
- 引入漂移场方法，直接利用语义丰富的潜在表示进行蒸馏
- 仅16轮蒸馏即达1.48 FID，超越现有最优方法
- 支持零成本的一步生成，适合高效图像生成场景
尽管训练加速显著且性能优异，表示自编码器（RAE）常因蒸馏效果不佳而受批评。本文认为RAE在高质量单步生成上具备潜力。我们在ImageNet 256数据集上仅用16轮蒸馏即实现1.48 FID，优于多种先进方法。通过定量分析不同数据空间的几何行为，我们发现传统蒸馏依赖于教师去噪轨迹的先验，而RAE因潜在空间各向异性导致复杂且性质差的轨迹。为此，我们引入最近提出的漂移场作为蒸馏方法，利用语义丰富的RAE潜在表示，提供无依赖的直接监督。结合已有生成范式，提出多项改进，包括首个基于外推的引导采样流程，实现几乎无成本的一步生成。代码将公开。
原文摘要 · Abstract (English)
Despite the significant training acceleration and promising performance, Representation Autoencoders (RAEs) are mainly criticized for poor distillation effectiveness. In this work, we argue that RAE is competent at high-quality one-step generation. We achieve 1.48 FID with only 16-epoch distillation on ImageNet 256 dataset, surpassing various state-of-the-art methods. To achieve this, we quantitatively study the geometrical behavior of different underlying data spaces. We conclude that conventional distillation methods heavily rely on priors of plain teacher denoising trajectories, while RAE incurs much more complex trajectories with poor properties due to ill anisotropical latent space. We introduce the recently proposed drifting field as the distillation methodology, which makes use of semantically rich RAE latents and provides direct supervision involving no dependency. Bridging our Drift-RAE with previous generative paradigms, we propose several insightful modifications, including the first extrapolation-based guided sampling pipeline for one-step generation with barely no cost. The code will be made publicly available.
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。