| title | JLT: Clean-Latent Prediction in Latent Diffusion Transformers |
|---|---|
| authors | Funing Fu, Tenghui Wang, Junyong Cen, Qichao Zhu, Guanyu Zhou |
| date | 2026-05-26 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2605.27102 |
| affiliation | akatsuki-neo |
| upvotes | 29 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf code |
| abstract | Flow matching with clean-data prediction has shown that regressing the clean point can exploit low-dimensional structure more effectively than predicting an ambient noised quantity. We ask whether this principle remains useful after images are mapped into a learned latent space, where compression has already removed much of the raw pixel variability. We introduce JLT, a 130M latent diffusion Transformer over frozen FLUX.2 VAE codes, and compare clean-latent prediction with a matched velocity-prediction DiT under the same representation, backbone, and training settings. Although the three variables x, epsilon, and v are linearly convertible for a fixed corruption time, a local Gaussian analysis shows that velocity regression inherits an isotropic target-covariance floor and amplifies low-variance latent directions, while clean prediction damps them. On ImageNet 256 x 256, JLT-B/1 obtains FID-50K 2.50 with classifier-free guidance, with a large matched-target gap over velocity prediction. These results suggest that prediction targets in latent diffusion are representation-dependent geometric choices, rather than interchangeable algebraic parameterizations. |
cite
@misc{fu2026jlt,
title={JLT: Clean-Latent Prediction in Latent Diffusion Transformers},
author={Funing Fu and Tenghui Wang and Junyong Cen and Qichao Zhu and Guanyu Zhou},
year={2026},
eprint={2605.27102},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2605.27102}
}