BWhat changed
Compared semantic and perceptual autoencoder spaces under mostly controlled flow-training settings, varying noise/sampling schedules and assessing reconstruction fidelity, compression and generative learnability.
Scientific scope
Compared semantic and perceptual autoencoder spaces under mostly controlled flow-training settings, varying noise/sampling schedules and assessing reconstruction fidelity, compression and generative learnability. Institutional empirical study, not proof of an optimal universal latent space or independent replay. DiT/SiT, flow matching, VAE/RAE and timestep shifts are inherited. Comparisons depend on chosen autoencoders,256pxImageNet, schedule search and evaluation budget; scores do not imply every full FLUX2 product capability. Individual study authors not resolved.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Institutional empirical study, not proof of an optimal universal latent space or independent replay. DiT/SiT, flow matching, VAE/RAE and timestep shifts are inherited. Comparisons depend on chosen autoencoders,256pxImageNet, schedule search and evaluation budget; scores do not imply every full FLUX2 product capability. Individual study authors not resolved.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction