Pith. sign in

REVIEW 5 cited by

Exploring Representation-Aligned Latent Space for Better Generation

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2502.00359 v1 pith:P6SSMKL6 submitted 2025-02-01 cs.LG

Wanghan Xu , Xiaoyu Yue , Zidong Wang , Yao Teng , Wenlong Zhang , Xihui Liu , Luping Zhou , Wanli Ouyang
show 1 more author
Lei Bai
This is my paper
classification cs.LG
keywords latentspacediffusionlatentsmodelsrealsemanticgeneration
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Generative models serve as powerful tools for modeling the real world, with mainstream diffusion models, particularly those based on the latent diffusion model paradigm, achieving remarkable progress across various tasks, such as image and video synthesis. Latent diffusion models are typically trained using Variational Autoencoders (VAEs), interacting with VAE latents rather than the real samples. While this generative paradigm speeds up training and inference, the quality of the generated outputs is limited by the latents' quality. Traditional VAE latents are often seen as spatial compression in pixel space and lack explicit semantic representations, which are essential for modeling the real world. In this paper, we introduce ReaLS (Representation-Aligned Latent Space), which integrates semantic priors to improve generation performance. Extensive experiments show that fundamental DiT and SiT trained on ReaLS can achieve a 15% improvement in FID metric. Furthermore, the enhanced semantic latent space enables more perceptual downstream tasks, such as segmentation and depth estimation.

Discussion (0). Sign in to comment.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. DiffusionBench: On Holistic Evaluation of Diffusion Transformers

    cs.CV 2026-06 conditional novelty 6.0 of 10

    NanoGen unifies DiT training on ImageNet and T2I, reveals negative Pearson correlations (-0.377 to -0.580) in method rankings across metrics from 21 models, and motivates DiffusionBench for holistic evaluation.

  2. Improved Baselines with Representation Autoencoders

    cs.CV 2026-05 conditional novelty 6.0 of 10

    RAE v2 reaches gFID 1.06 on ImageNet-256 in 80 epochs by combining multi-layer encoder sums, complementary REPA targets, and free guidance via output reparameterization.

  3. Align & Invert: Solving Inverse Problems with Diffusion and Flow-based Models via Representation Alignment

    cs.CV 2025-11 conditional novelty 6.0 of 10

    Adding DINOv2 representation alignment to diffusion/flow inverse-problem solvers, using corrupted measurements as proxies, improves LPIPS/FID and cuts required sampling steps.

  4. Missing Fine Details in Images: Last Seen in High Frequencies

    cs.CV 2025-09 conditional novelty 6.0 of 10

    A wavelet-based VAE that trains low- and high-frequency branches separately improves image reconstruction and diffusion generation.

  5. Transition Models: Rethinking the Generative Learning Objective

    cs.LG 2025-09 conditional novelty 6.0 of 10

    TiM trains a single diffusion-type model on arbitrary time-interval transitions, achieving strong one-step and multi-step text-to-image generation with 865M parameters.

Pith tools