# JLT: Clean-Latent Prediction in Latent Diffusion Transformers

Source: [JLT: Clean-Latent Prediction in Latent Diffusion Transformers](<https://www.sota2.com/research/paper/6a167acddbd7150506f8022e>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-30T06:00:01.927Z

Authors: Funing Fu, Tenghui Wang, Guanyu Zhou, Junyong Cen, Qichao Zhu

Published: 2026-05-26T14:41:58.000Z

## 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.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Image Generation on ImageNet 256x256](<https://www.sota2.com/research/sota/image-generation-on-imagenet-256x256.md>): Image Generation; ImageNet 256x256; IS: 232.51.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.27102>)

[PDF](<https://arxiv.org/pdf/2605.27102.pdf>)

[Implementation](<https://github.com/akatsuki-neo/JLT>)
