pith. machine review for the scientific record. sign in

arxiv: 2604.16044 · v1 · submitted 2026-04-17 · 💻 cs.CV

Recognition: unknown

Elucidating the SNR-t Bias of Diffusion Probabilistic Models

Jianhao Zeng, Kun Zhan, Lei Sun, Meng Yu, Xiangxiang Chu

Authors on Pith no claims yet

Pith reviewed 2026-05-10 09:00 UTC · model grok-4.3

classification 💻 cs.CV
keywords modelsbiasdiffusionduringsnr-tvariouscomponentscorrection
0
0 comments X

The pith

Diffusion models have an SNR-timestep mismatch during inference that the authors mitigate with per-frequency differential correction, raising generation quality across IDDPM, ADM, DDIM and others.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Diffusion models create images by starting from noise and gradually removing it step by step. In training, the amount of noise is locked to the step number, but during actual image creation this lock breaks, so the model sees the wrong noise level for the step it is on. The authors split each sample into low-frequency and high-frequency parts, then apply a separate correction to each part so the noise level better matches the step. They show this change improves output on several existing diffusion systems without much extra computation.

Core claim

during training, the SNR of a sample is strictly coupled with its timestep. However, this correspondence is disrupted during inference, leading to error accumulation and impairing the generation quality... propose a simple yet effective differential correction method to mitigate the SNR-t bias.

Load-bearing premise

That the observed SNR-t mismatch is the dominant source of error accumulation rather than a symptom of other training or sampling choices, and that applying independent corrections to frequency bands will not introduce new artifacts or instability.

Figures

Figures reproduced from arXiv: 2604.16044 by Jianhao Zeng, Kun Zhan, Lei Sun, Meng Yu, Xiangxiang Chu.

Figure 1
Figure 1. Figure 1: (a) During training, the SNR of perturbed sample [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The overall framework of Differential Correction in [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: Hyperparameter search experiments on CIFAR-10 (CS) [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Robust experimental results for Fig. 1c with varied random number seeds and sampling batch sizes. These figures show the network output ||ϵθ(·, t)||2 using forward samples xt via Eq. 2 and reverse predicted samples xˆt via Eq. 8, respectively. ||ϵθ(xˆt, t)||2 is always larger than ||ϵθ(xt, t)||2 in every figure. Based on the relationship between the score and the noise sθ(xt, t) = − ϵθ(xt,t) √ 1−α¯t , we f… view at source ↗
Figure 6
Figure 6. Figure 6: The expectation of ||x 0 θ(xt, t)||2 2, ||x 0 θ(xˆt, t)||2 2, together with the ground-truth norm of x0. This further indicates that the reconstruction operation in￾curs information loss. Notably, similar experiments are also reported in DPM-FR [64]. However, it focuses on the differ￾ences between the forward and reverse processes, whereas our work places greater emphasis on whether DPMs can fully reconstr… view at source ↗
Figure 7
Figure 7. Figure 7: Qualitative comparison between Qwen-Image (first row) and Qwen-Image-DCW (second row) using 10 steps, where the prompt is “A woman is walking on the beach by the sea”. FLUX, which is known for its high visual fidelity, to conduct extensive experiments. Given that our study focuses on the SNR-t bias, we conduct tests with a small number of steps to amplify the sampling errors of the baseline models as much … view at source ↗
Figure 8
Figure 8. Figure 8: Qualitative comparison between Qwen-Image (first row) and Qwen-Image-DCW (second row) using 10 steps, where the prompt is “There is a house and a path on a snowy mountain” [PITH_FULL_IMAGE:figures/full_fig_p016_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Qualitative comparison between Qwen-Image (first row) and Qwen-Image-DCW (second row) using 10 steps, where the prompt is “A balloon gently climbs into a serene blue sky” [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Qualitative comparison between FLUX (first row) and FLUX-DCW (second row) using 10 steps, where the prompt is “There is a house and a path on a snowy mountain” [PITH_FULL_IMAGE:figures/full_fig_p017_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Qualitative comparison between FLUX (first row) and FLUX-DCW (second row) using 10 steps, where the prompt is “A woman is walking on the beach by the sea” [PITH_FULL_IMAGE:figures/full_fig_p017_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Qualitative comparison between FLUX (first row) and FLUX-DCW (second row) using 10 steps, where the prompt is “A balloon gently climbs into a serene blue sky” [PITH_FULL_IMAGE:figures/full_fig_p018_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Qualitative comparison between Qwen-Image (first row) and Qwen-Image-DCW (second row) using 20 steps, where the prompt is “A woman is walking on the beach by the sea” [PITH_FULL_IMAGE:figures/full_fig_p018_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Qualitative comparison between FLUX (first row) and FLUX-DCW (second row) using 20 steps, where the prompt is “There is a house and a path on a snowy mountain” [PITH_FULL_IMAGE:figures/full_fig_p019_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Qualitative comparison between FLUX (first row) and FLUX-DCW (second row) using 20 steps, where the prompt is “A balloon gently climbs into a serene blue sky” [PITH_FULL_IMAGE:figures/full_fig_p019_15.png] view at source ↗
read the original abstract

Diffusion Probabilistic Models have demonstrated remarkable performance across a wide range of generative tasks. However, we have observed that these models often suffer from a Signal-to-Noise Ratio-timestep (SNR-t) bias. This bias refers to the misalignment between the SNR of the denoising sample and its corresponding timestep during the inference phase. Specifically, during training, the SNR of a sample is strictly coupled with its timestep. However, this correspondence is disrupted during inference, leading to error accumulation and impairing the generation quality. We provide comprehensive empirical evidence and theoretical analysis to substantiate this phenomenon and propose a simple yet effective differential correction method to mitigate the SNR-t bias. Recognizing that diffusion models typically reconstruct low-frequency components before focusing on high-frequency details during the reverse denoising process, we decompose samples into various frequency components and apply differential correction to each component individually. Extensive experiments show that our approach significantly improves the generation quality of various diffusion models (IDDPM, ADM, DDIM, A-DPM, EA-DPM, EDM, PFGM++, and FLUX) on datasets of various resolutions with negligible computational overhead. The code is at https://github.com/AMAP-ML/DCW.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The method rests on the assumption that frequency decomposition cleanly separates the components that should receive different corrections; no free parameters, new axioms, or invented entities are mentioned in the abstract.

axioms (1)
  • domain assumption Fourier or wavelet decomposition can be applied to intermediate denoising samples without destroying the diffusion process semantics
    Invoked when the authors state they decompose samples into frequency components and correct each individually.

pith-pipeline@v0.9.0 · 5507 in / 1228 out tokens · 38951 ms · 2026-05-10T09:00:23.077859+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

69 extracted references · 9 canonical work pages · 2 internal anchors

  1. [1]

    Estimating the optimal covariance with imperfect mean in diffusion probabilistic models

    Fan Bao, Chongxuan Li, Jiacheng Sun, Jun Zhu, and Bo Zhang. Estimating the optimal covariance with imperfect mean in diffusion probabilistic models. InICML, 2022. 6, 7

    2022

  2. [2]

    Analytic- DPM: an analytic estimate of the optimal reverse variance in diffusion probabilistic models

    Fan Bao, Chongxuan Li, Jun Zhu, and Bo Zhang. Analytic- DPM: an analytic estimate of the optimal reverse variance in diffusion probabilistic models. InICLR, 2022. 6, 7

    2022

  3. [3]

    Black Forest Labs. Flux. https://github.com/ black-forest-labs/flux, 2024. 2, 6, 7

    2024

  4. [4]

    Align your latents: High-resolution video synthesis with la- tent diffusion models

    Andreas Blattmann, Robin Rombach, Huan Ling, Tim Dock- horn, Seung Wook Kim, Sanja Fidler, and Karsten Kreis. Align your latents: High-resolution video synthesis with la- tent diffusion models. InCVPR, 2023. 1

    2023

  5. [5]

    Stochastic self-guidance for training-free enhancement of diffusion models

    Chubin Chen, Jiashu Zhu, Xiaokun Feng, Nisha Huang, Chen Zhu, Meiqi Wu, Fangyuan Mao, Jiahong Wu, Xiangxiang Chu, and Xiu Li. Stochastic self-guidance for training-free enhancement of diffusion models. InICLR, 2026. 2

    2026

  6. [6]

    Wavegrad: Estimating gradients for waveform generation

    Nanxin Chen, Yu Zhang, Heiga Zen, Ron J Weiss, Moham- mad Norouzi, and William Chan. Wavegrad: Estimating gradients for waveform generation. InICLR, 2021. 1

    2021

  7. [7]

    Layer-wise instance binding for regional and occlusion control in text-to-image diffusion transformers.arXiv preprint arXiv:2603.05769, 2026

    Ruidong Chen, Yancheng Bai, Xuanpu Zhang, Jianhao Zeng, Lanjun Wang, Dan Song, Lei Sun, Xiangxiang Chu, and Anan Liu. Layer-wise instance binding for regional and occlusion control in text-to-image diffusion transformers.arXiv preprint arXiv:2603.05769, 2026. 2

    work page arXiv 2026

  8. [8]

    A Downsampled Variant of ImageNet as an Alternative to the CIFAR datasets

    Patryk Chrabaszcz, Ilya Loshchilov, and Frank Hutter. A downsampled variant of ImageNet as an alternative to the CIFAR datasets.arXiv preprint arXiv:1707.08819, 2017. 6

    work page Pith review arXiv 2017

  9. [9]

    Usp: Unified self-supervised pretraining for image generation and under- standing

    Xiangxiang Chu, Renda Li, and Yong Wang. Usp: Unified self-supervised pretraining for image generation and under- standing. InICCV, 2025. 2

    2025

  10. [10]

    DiffEdit: Diffusion-based semantic image editing with mask guidance

    Guillaume Couairon, Jakob Verbeek, Holger Schwenk, and Matthieu Cord. DiffEdit: Diffusion-based semantic image editing with mask guidance. InICLR, 2023. 2

    2023

  11. [11]

    Diffusion models beat GANs on image synthesis

    Prafulla Dhariwal and Alexander Nichol. Diffusion models beat GANs on image synthesis. InNeurIPS, 2021. 1, 2, 3, 6

    2021

  12. [12]

    Genie: Higher-order denoising diffusion solvers

    Tim Dockhorn, Arash Vahdat, and Karsten Kreis. Genie: Higher-order denoising diffusion solvers. InNeurIPS, 2022. 2

    2022

  13. [13]

    Generative adversarial nets

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial nets. InNeurIPS,

  14. [14]

    An introduction to wavelets.IEEE Computa- tional Science and Engineering, 1995

    Amara Graps. An introduction to wavelets.IEEE Computa- tional Science and Engineering, 1995. 5, 6

    1995

  15. [15]

    Texts-diff: Texts-aware diffusion model for real-world text image super-resolution.arXiv preprint arXiv:2601.17340, 2026

    Haodong He, Xin Zhan, Yancheng Bai, Rui Lan, Lei Sun, and Xiangxiang Chu. Texts-diff: Texts-aware diffusion model for real-world text image super-resolution.arXiv preprint arXiv:2601.17340, 2026. 2

    work page arXiv 2026

  16. [16]

    GANs trained by a two time-scale update rule converge to a local nash equilibrium

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bern- hard Nessler, and Sepp Hochreiter. GANs trained by a two time-scale update rule converge to a local nash equilibrium. InNeurIPS, 2017. 6, 5

    2017

  17. [17]

    Denoising diffu- sion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffu- sion probabilistic models. InNeurIPS, 2020. 1, 2, 6

    2020

  18. [18]

    Elucidating the design space of diffusion-based generative models

    Tero Karras, Miika Aittala, Timo Aila, and Samuli Laine. Elucidating the design space of diffusion-based generative models. InNeurIPS, 2022. 2, 4, 6, 7

    2022

  19. [19]

    Text2video-zero: Text- to-image diffusion models are zero-shot video generators

    Levon Khachatryan, Andranik Movsisyan, Vahram Tade- vosyan, Roberto Henschel, Zhangyang Wang, Shant Navasardyan, and Humphrey Shi. Text2video-zero: Text- to-image diffusion models are zero-shot video generators. In ICCV, 2023. 1

    2023

  20. [20]

    Refining generative process with discrim- inator guidance in score-based diffusion models

    Dongjun Kim, Yeongmin Kim, Se Jung Kwon, Wanmo Kang, and Il-Chul Moon. Refining generative process with discrim- inator guidance in score-based diffusion models. InICML,

  21. [21]

    DiffWave: A versatile diffusion model for audio synthesis

    Zhifeng Kong, Wei Ping, Jiaji Huang, Kexin Zhao, and Bryan Catanzaro. DiffWave: A versatile diffusion model for audio synthesis. InICLR, 2021. 1

    2021

  22. [22]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny images. 2009. 3, 6

    2009

  23. [23]

    Flux-text: A simple and advanced diffusion transformer baseline for scene text editing.arXiv preprint arXiv:2505.03329, 2025

    Rui Lan, Yancheng Bai, Xu Duan, Mingxing Li, Dongyang Jin, Ryan Xu, Lei Sun, and Xiangxiang Chu. Flux-text: A simple and advanced diffusion transformer baseline for scene text editing.arXiv preprint arXiv:2505.03329, 2025. 2

    work page arXiv 2025

  24. [24]

    There is no V AE: End- to-end pixel-space generative modeling via self-supervised pre-training

    Jiachen Lei, Keli Liu, Julius Berner, Y HoiM, Hongkai Zheng, Jiahong Wu, and Xiangxiang Chu. There is no V AE: End- to-end pixel-space generative modeling via self-supervised pre-training. InICLR, 2026. 2

    2026

  25. [25]

    Srdiff: Single image super-resolution with diffusion probabilistic models

    Haoying Li, Yifan Yang, Meng Chang, Shiqi Chen, Huajun Feng, Zhihai Xu, Qi Li, and Yueting Chen. Srdiff: Single image super-resolution with diffusion probabilistic models. Neurocomputing, 2022. 2

    2022

  26. [26]

    Alleviating exposure bias in diffusion mod- els through sampling with shifted time steps

    Mingxiao Li, Tingyu Qu, Ruicong Yao, Wei Sun, and Marie- Francine Moens. Alleviating exposure bias in diffusion mod- els through sampling with shifted time steps. InICLR, 2024. 2, 4, 3

    2024

  27. [27]

    On error propaga- tion of diffusion models

    Yangming Li and Mihaela van der Schaar. On error propaga- tion of diffusion models. InICLR, 2024. 2

    2024

  28. [28]

    Instaflow: One step is enough for high-quality diffusion-based text-to-image generation

    Xingchao Liu, Xiwen Zhang, Jianzhu Ma, Jian Peng, et al. Instaflow: One step is enough for high-quality diffusion-based text-to-image generation. InICLR, 2023. 2

    2023

  29. [29]

    Deep learning face attributes in the wild

    Ziwei Liu, Ping Luo, Xiaogang Wang, and Xiaoou Tang. Deep learning face attributes in the wild. InICCV, 2015. 6

    2015

  30. [30]

    Simplifying, stabilizing and scaling continuous-time consistency models

    Cheng Lu and Yang Song. Simplifying, stabilizing and scaling continuous-time consistency models. InICLR, 2025. 2

    2025

  31. [31]

    DPM-solver: A fast ODE solver for diffusion probabilistic model sampling in around 10 steps

    Cheng Lu, Yuhao Zhou, Fan Bao, Jianfei Chen, Chongxuan Li, and Jun Zhu. DPM-solver: A fast ODE solver for diffusion probabilistic model sampling in around 10 steps. InNeurIPS,

  32. [32]

    Luhman and T

    Eric Luhman and Troy Luhman. Knowledge distillation in it- erative generative models for improved sampling speed.arXiv preprint arXiv:2101.02388, 2021. 2

    work page arXiv 2021

  33. [33]

    SDEdit: Guided image synthesis and editing with stochastic differential equations

    Chenlin Meng, Yutong He, Yang Song, Jiaming Song, Jiajun Wu, Jun-Yan Zhu, and Stefano Ermon. SDEdit: Guided image synthesis and editing with stochastic differential equations. InICLR, 2022. 2

    2022

  34. [34]

    On distillation of guided diffusion models

    Chenlin Meng, Robin Rombach, Ruiqi Gao, Diederik Kingma, Stefano Ermon, Jonathan Ho, and Tim Salimans. On distillation of guided diffusion models. InCVPR, 2023. 2

    2023

  35. [35]

    Shining yourself: High-fidelity ornaments virtual try-on with diffusion model

    Yingmao Miao, Zhanpeng Huang, Rui Han, Zibin Wang, Chenhao Lin, and Chao Shen. Shining yourself: High-fidelity ornaments virtual try-on with diffusion model. InCVPR,

  36. [36]

    Analysing the spectral biases in generative models

    Amitoj Singh Miglani, Shweta Singh, and Vidit Aggarwal. Analysing the spectral biases in generative models. InThe Fourth Blogpost Track at ICLR 2025, 2025. 4

    2025

  37. [37]

    Improved denoising diffusion probabilistic models

    Alexander Quinn Nichol and Prafulla Dhariwal. Improved denoising diffusion probabilistic models. InICLR, 2021. 6

    2021

  38. [38]

    Input perturbation reduces exposure bias in diffusion models

    Mang Ning, Enver Sangineto, Angelo Porrello, Simone Calderara, and Rita Cucchiara. Input perturbation reduces exposure bias in diffusion models. InICML, 2023. 1, 2, 6

    2023

  39. [39]

    Elucidating the exposure bias in diffusion models

    Mang Ning, Mingxiao Li, Jianlin Su, Albert Ali Salah, and Itir Onal Ertugrul. Elucidating the exposure bias in diffusion models. InICLR, 2024. 2, 4, 6, 7, 3, 5

    2024

  40. [40]

    Zero-shot image-to-image translation

    Gaurav Parmar, Krishna Kumar Singh, Richard Zhang, Yijun Li, Jingwan Lu, and Jun-Yan Zhu. Zero-shot image-to-image translation. InACM SIGGRAPH, 2023. 2

    2023

  41. [41]

    Scalable diffusion models with transformers

    William Peebles and Saining Xie. Scalable diffusion models with transformers. InCVPR, 2023. 6, 7, 5

    2023

  42. [42]

    Boosting diffusion models with moving average sampling in frequency domain

    Yurui Qian, Qi Cai, Yingwei Pan, Yehao Li, Ting Yao, Qibin Sun, and Tao Mei. Boosting diffusion models with moving average sampling in frequency domain. InCVPR, 2024. 2

    2024

  43. [43]

    From scale to speed: Adaptive test-time scaling for image editing.arXiv preprint arXiv:2603.00141,

    Xiangyan Qu, Zhenlong Yuan, Jing Tang, Rui Chen, Datao Tang, Meng Yu, Lei Sun, Yancheng Bai, Xiangxiang Chu, Gaopeng Gou, et al. From scale to speed: Adaptive test-time scaling for image editing.arXiv preprint arXiv:2603.00141,

    work page arXiv

  44. [44]

    Multi-step denoising scheduled sampling: Towards alleviating exposure bias for diffusion models

    Zhiyao Ren, Yibing Zhan, Liang Ding, Gaoang Wang, Chaoyue Wang, Zhongyi Fan, and Dacheng Tao. Multi-step denoising scheduled sampling: Towards alleviating exposure bias for diffusion models. InAAAI, 2024. 2

    2024

  45. [45]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image synthesis with latent diffusion models. InCVPR, 2022. 1, 2

    2022

  46. [46]

    Image super- resolution via iterative refinement.TPAMI, 2022

    Chitwan Saharia, Jonathan Ho, William Chan, Tim Sali- mans, David J Fleet, and Mohammad Norouzi. Image super- resolution via iterative refinement.TPAMI, 2022. 2

    2022

  47. [47]

    Progressive distillation for fast sampling of diffusion models

    Tim Salimans and Jonathan Ho. Progressive distillation for fast sampling of diffusion models. InICLR, 2022. 2

    2022

  48. [48]

    Deep unsupervised learning using nonequilibrium thermodynamics

    Jascha Sohl-Dickstein, Eric Weiss, Niru Maheswaranathan, and Surya Ganguli. Deep unsupervised learning using nonequilibrium thermodynamics. InICML, 2015. 1, 2

    2015

  49. [49]

    Denoising diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denoising diffusion implicit models. InICLR, 2021. 6

    2021

  50. [50]

    Improved techniques for training consistency models

    Yang Song and Prafulla Dhariwal. Improved techniques for training consistency models. InICLR, 2024. 2

    2024

  51. [51]

    Score-based generative modeling through stochastic differential equations

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Ab- hishek Kumar, Stefano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equations. InICLR, 2021. 1

    2021

  52. [52]

    Consistency models

    Yang Song, Prafulla Dhariwal, Mark Chen, and Ilya Sutskever. Consistency models. InICML, 2023. 2

    2023

  53. [53]

    Self-guided generation of minority samples using diffusion models

    Soobin Um and Jong Chul Ye. Self-guided generation of minority samples using diffusion models. InECCV, 2024. 6

    2024

  54. [54]

    Don’t play favorites: Minority guidance for diffusion models

    Soobin Um, Suhyeon Lee, and Jong Chul Ye. Don’t play favorites: Minority guidance for diffusion models. InICLR,

  55. [55]

    Score-based generative modeling in latent space

    Arash Vahdat, Karsten Kreis, and Jan Kautz. Score-based generative modeling in latent space. InNeurIPS, 2021. 5, 7

    2021

  56. [56]

    From editor to dense geometry estimator.arXiv preprint arXiv:2509.04338, 2025

    JiYuan Wang, Chunyu Lin, Lei Sun, Rongying Liu, Lang Nie, Mingxing Li, Kang Liao, Xiangxiang Chu, and Yao Zhao. From editor to dense geometry estimator.arXiv preprint arXiv:2509.04338, 2025. 2

    work page arXiv 2025

  57. [57]

    Improved diffusion-based generative model with better adversarial robustness

    Zekun Wang, Mingyang Yi, Shuchen Xue, Zhenguo Li, Ming Liu, Bing Qin, and Zhi-Ming Ma. Improved diffusion-based generative model with better adversarial robustness. InICLR,

  58. [58]

    Qwen-image technical report, 2025

    Chenfei Wu, Jiahao Li, Jingren Zhou, Junyang Lin, Kaiyuan Gao, Kun Yan, Sheng ming Yin, Shuai Bai, Xiao Xu, Yilei Chen, Yuxiang Chen, Zecheng Tang, Zekai Zhang, Zhengyi Wang, An Yang, Bowen Yu, Chen Cheng, Dayiheng Liu, De- qing Li, Hang Zhang, Hao Meng, Hu Wei, Jingyuan Ni, Kai Chen, Kuan Cao, Liang Peng, Lin Qu, Minggang Wu, Peng Wang, Shuting Yu, Tingk...

    2025

  59. [59]

    PFGM++: Unlocking the potential of physics-inspired generative models

    Yilun Xu, Ziming Liu, Yonglong Tian, Shangyuan Tong, Max Tegmark, and Tommi Jaakkola. PFGM++: Unlocking the potential of physics-inspired generative models. InICML,

  60. [60]

    Manifold constraint reduces exposure bias in accelerated diffusion sampling

    Yuzhe Y AO, Jun Chen, Zeyi Huang, Haonan Lin, Mengmeng Wang, Guang Dai, and Jingdong Wang. Manifold constraint reduces exposure bias in accelerated diffusion sampling. In ICLR, 2025. 2

    2025

  61. [61]

    Towards understanding the working mechanism of text-to-image diffu- sion model

    Mingyang Yi, Aoxue Li, Yi Xin, and Zhenguo Li. Towards understanding the working mechanism of text-to-image diffu- sion model. InNeurIPS, 2024. 2, 5

    2024

  62. [62]

    LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop

    Fisher Yu, Ari Seff, Yinda Zhang, Shuran Song, Thomas Funkhouser, and Jianxiong Xiao. Lsun: Construction of a large-scale image dataset using deep learning with humans in the loop.arXiv preprint arXiv:1506.03365, 2015. 6

    work page internal anchor Pith review arXiv 2015

  63. [63]

    Bias mitigation in graph diffusion models

    Meng Yu and Kun Zhan. Bias mitigation in graph diffusion models. InICLR, 2025. 2

    2025

  64. [64]

    Frequency regulation for exposure bias mitigation in diffusion models

    Meng Yu and Kun Zhan. Frequency regulation for exposure bias mitigation in diffusion models. InACM MM, 2025. 2, 4, 6, 7, 3

    2025

  65. [65]

    Looka- head diffusion probabilistic models for refining mean estima- tion

    Guoqiang Zhang, Kenta Niwa, and W Bastiaan Kleijn. Looka- head diffusion probabilistic models for refining mean estima- tion. InCVPR, 2023. 4, 5, 3

    2023

  66. [66]

    Anti-exposure bias in diffusion models

    Junyu Zhang, Daochang Liu, Eunbyung Park, Shichao Zhang, and Chang Xu. Anti-exposure bias in diffusion models. In ICLR, 2025. 2, 6

    2025

  67. [67]

    Unipc: A unified predictor-corrector framework for fast sampling of diffusion models

    Wenliang Zhao, Lujia Bai, Yongming Rao, Jie Zhou, and Jiwen Lu. Unipc: A unified predictor-corrector framework for fast sampling of diffusion models. InNeurIPS, 2024. 2

    2024

  68. [68]

    Open-Sora: Democratizing Efficient Video Production for All

    Zangwei Zheng, Xiangyu Peng, Tianji Yang, Chenhui Shen, Shenggui Li, Hongxin Liu, Yukun Zhou, Tianyi Li, and Yang You. Open-sora: Democratizing efficient video production for all.arXiv preprint arXiv:2412.20404, 2024. 1

    work page internal anchor Pith review arXiv 2024

  69. [69]

    A woman is walking on the beach by the sea

    Zhenyu Zhou, Defang Chen, Can Wang, and Chun Chen. Fast ODE-based sampling for diffusion models in around 5 steps. InCVPR, 2024. 2 Elucidating the SNR-t Bias of Diffusion Probabilistic Models Supplementary Material A. Difference from Prior Works In this section, we outline the differences between the second experiment (Fig. 1c) in Sec 4 of this paper and ...

    2024