Pith. sign in

REVIEW 9 cited by

Elucidating the Exposure Bias in Diffusion Models

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 2308.15321 v6 pith:UFZTFQ53 submitted 2023-08-29 cs.LG cs.AIcs.CV

Mang Ning , Mingxiao Li , Jianlin Su , Albert Ali Salah , Itir Onal Ertugrul This is my paper · ORCID
classification cs.LGcs.AIcs.CV
keywords biasexposuresamplingdiffusionmodelsscalingtrainingadm-es
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

Diffusion models have demonstrated impressive generative capabilities, but their \textit{exposure bias} problem, described as the input mismatch between training and sampling, lacks in-depth exploration. In this paper, we systematically investigate the exposure bias problem in diffusion models by first analytically modelling the sampling distribution, based on which we then attribute the prediction error at each sampling step as the root cause of the exposure bias issue. Furthermore, we discuss potential solutions to this issue and propose an intuitive metric for it. Along with the elucidation of exposure bias, we propose a simple, yet effective, training-free method called Epsilon Scaling to alleviate the exposure bias. We show that Epsilon Scaling explicitly moves the sampling trajectory closer to the vector field learned in the training phase by scaling down the network output, mitigating the input mismatch between training and sampling. Experiments on various diffusion frameworks (ADM, DDIM, EDM, LDM, DiT, PFGM++) verify the effectiveness of our method. Remarkably, our ADM-ES, as a state-of-the-art stochastic sampler, obtains 2.17 FID on CIFAR-10 under 100-step unconditional generation. The code is available at \url{https://github.com/forever208/ADM-ES} and \url{https://github.com/forever208/EDM-ES}.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 9 Pith papers

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

  1. Large Concept Models: Language Modeling in a Sentence Representation Space

    cs.CL 2024-12 conditional novelty 7.0 of 10

    A sentence-level language model trained to autoregressively predict SONAR sentence embeddings can summarize, expand, and generate text in unseen languages.

  2. Conditional Flow Matching for Visually-Guided Acoustic Highlighting

    eess.AS 2026-02 conditional novelty 6.0 of 10

    Conditional flow matching with a rollout loss and early audio-visual fusion achieves state-of-the-art results on visually-guided acoustic highlighting.

  3. End-to-End Training for Autoregressive Video Diffusion via Self-Resampling

    cs.CV 2025-12 conditional novelty 6.0 of 10

    Resampling Forcing trains autoregressive video diffusion models on self-resampled degraded histories with a causal mask, achieving stable long-horizon generation without a teacher or discriminator.

  4. STARCaster: Spatio-Temporal AutoRegressive Video Diffusion for Identity- and View-Aware Talking Portraits

    cs.CV 2025-12 conditional novelty 6.0 of 10

    STARCaster is a 2D spatio-temporal video diffusion model that unifies identity-conditioned audio-driven portrait animation and novel-view synthesis without explicit 3D reconstruction.

  5. DMQ: Dissecting Outliers of Diffusion Models for Post-Training Quantization

    cs.CV 2025-07 conditional novelty 6.0 of 10

    A post-training quantization method that combines learned channel scaling and power-of-two scaling keeps diffusion image quality high at 4-bit weight, 6-bit activation precision.

  6. FreeDNA: Endowing Domain Adaptation of Diffusion-Based Dense Prediction with Training-Free Domain Noise Alignment

    cs.CV 2025-06 conditional novelty 6.0 of 10

    Aligning the L2-norm statistics of noise predictions during diffusion sampling improves domain adaptation for dense prediction, with a source-free version guided by high-confidence regions.

  7. Explorative Modeling: Unlocking a Third Pretraining Axis and End-to-End Generation

    cs.LG 2026-07 conditional novelty 5.0 of 10

    Training on the best of K generated candidates improves image, video, and language generative models, with the reported gains growing with scale and enabling single-pass end-to-end generation.

  8. DiffIER: Optimizing Diffusion Models with Iterative Error Reduction

    cs.CV 2025-08 reject novelty 4.0 of 10

    DiffIER claims that iteratively minimizing the distance between a diffusion model's predicted noise and a random Gaussian sample at each inference step improves generation quality.

  9. Comparative Analysis of Diffusion Generative Models in Computational Pathology

    eess.IV 2024-11 conditional novelty 3.0 of 10

    A benchmark of DDPM and LDM on colon pathology patches shows DDPM slightly outperforms LDM, and a pretrained model can generate images at unseen patch sizes, with mixed classification gains.

Pith tools