ViTok-v2：将原生分辨率自编码器扩展到50亿参数
ViTok-v2: Scaling Native Resolution Auto-Encoders to 5 Billion Parameters
AI总结：
ViTok-v2通过NaFlex实现跨分辨率和宽高比泛化，引入新的DINOv3感知损失，提升重建性能并稳定训练，达到当前最佳水平。
摘要：
视觉变换器（ViT）自编码器已崭露头角，成为图像的有力分词器，提供了优于卷积分词器的重建性能。然而，现有ViT分词器无法探索这一领域，因为性能在训练分辨率之外会下降，且依赖对抗损失阻碍了稳定扩展。ViTok（Hansen-Estruch等，2025）发现压缩比r调解了重建-生成的权衡，较低的r意味着更好的重建但更难生成，因此提高分词器的重建性能是更帕累托最优分词器的关键。我们引入了ViTok-v2，通过NaFlex实现跨分辨率和宽高比的泛化，以及新的DINOv3感知损失，取代LPIPS和GAN目标，以在任何规模下实现稳定训练。ViTok-v2在约20亿张图像上训练并扩展到50亿参数，成为迄今为止最大的图像自编码器。ViTok-v2在256p分辨率上达到或超越了当前最佳重建水平，在512p及更高分辨率上超越了所有基线。在与流匹配生成器的联合扩展实验中，我们展示了同时扩展自编码器和生成器可以推进这一权衡的帕累托前沿。
英文摘要：
Vision Transformer (ViT) autoencoders have emerged as compelling tokenizers for images, offering improved reconstruction over convolutional tokenizers. However, existing ViT tokenizers cannot explore this landscape as performance degrades outside training resolutions, and reliance on adversarial losses prevents stable scaling. ViTok (Hansen-Estruch et al., 2025) found that the compression ratio r mediates a reconstruction-generation trade-off where lower r means better reconstructions but harder generations, so improving tokenizer reconstruction is key to more Pareto-optimal tokenizers. We introduce ViTok-v2, which addresses these limitations with native resolution support via NaFlex for generalization across resolutions and aspect ratios, and a novel DINOv3 perceptual loss that replaces both LPIPS and GAN objectives for stable training at any scale. ViTok-v2 is trained on about 2B images and scaled to 5B parameters, the largest image autoencoder to date. ViTok-v2 matches or exceeds state-of-the-art reconstruction at 256p and outperforms all baselines at 512p and above. In joint scaling experiments with flow matching generators, we show that scaling both the autoencoder and the generator advances the Pareto frontier of this trade-off.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
