Pith. sign in

REVIEW 3 major objections 5 minor 44 references

SUDO: Enhancing Text-to-Image Diffusion Models with Self-Supervised Direct Preference Optimization

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read SUDO lifts text-to-image diffusion models with self-made preference pairs, replacing costly human annotation.

desk verdict Useful self-supervised pair construction for diffusion DPO, but the 'downgrade' framing is misleading and the numbers lack error bars. read the letter →

arxiv 2504.14534 v1 pith:CWLYQQUH submitted 2025-04-20 cs.CV

Liang Peng , Boxi Wu , Haoran Cheng , Yibo Zhao , Xiaofei He This is my paper · ORCID
classification cs.CV
keywords text-to-imagegenerationdiffusionmodelsdirectpreferenceoptimizationself-supervisedlearningfine-tuningStableimage-levelpairs
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper proposes SUDO, a fine-tuning method for text-to-image diffusion models that replaces human-annotated preference pairs with self-generated ones. Taking the training image as the 'winning' sample and a randomly selected, mismatched image as the 'losing' sample, SUDO applies a direct preference optimization objective at the image level while keeping the standard pixel-level mean-squared-error loss. The claim is that this joint optimization improves both local detail and global image quality, and the experiments on Stable Diffusion 1.5 and SDXL across three evaluation sets show consistent gains over supervised fine-tuning, with most metrics at or above the human-annotated DPO baseline. If correct, SUDO offers a drop-in replacement for SFT that needs no extra data collection or annotation.

What carries the argument

The load-bearing object is the self-supervised preference pair created by the Downgrade operation. In each iteration the correct image $x^w$ is the winner and $x^{sl} = \mathrm{Downgrade}(x^w)$ is a randomly sampled image from the training set that does not correspond to the prompt. The loss is $$L_{\mathrm{SUDO}} = -\log\$\sigma$\!\left(C\left[\|\epsilon_\$\theta$(x^w_t,t)-\epsilon^w\|$_2^{2}$ - \|\epsilon_\$\theta$($x^{{sl}}$_t,t)-\$epsilon^{{sl}}$\|$_2^{2}$ - \left(\|\epsilon_{\mathrm{ref}}(x^w_t,t)-\epsilon^w\|$_2^{2}$ - \|\epsilon_{\mathrm{ref}}($x^{{sl}}$_t,t)-\$epsilon^{{sl}}$\|$_2^{2}$\right)\right]\right),$$ with $C$ a scale factor, and the total loss is $L = \lambda_1 L_{\mathrm{MSE}} + \lambda_2 L_{\mathrm{SUDO}}$. The objective widens the gap between the model's denoising error on the matching pair and its error on the mismatched pair, relative to the reference model, and this gap is what carries the image-level learning.

What would settle it

A controlled comparison could settle it: fine-tune with SUDO but replace the random-image loser with a same-prompt image generated by the base model, a true near-miss negative; if the gains disappear or reverse, the reported improvement comes from contrastive regularization against unrelated images rather than from preference optimization.

Watch

Extended reading notes

Core claim

The central discovery is that the expensive preference labels in DPO can be replaced by a free, self-supervised signal: any image that does not match the prompt is a valid losing sample. Formally, with $x^w$ the correct text-image pair and $x^{sl} = \mathrm{Downgrade}(x^w)$ obtained by randomly drawing a mismatched image from the training set, SUDO maximizes the log-likelihood of preferring $x^w$ over $x^{sl}$ in a DPO-style objective. The paper reports that this objective, combined with the MSE loss, consistently improves the base models on Pick-a-Pic V2, PartiPrompts, and HPDv2, measured by PickScore, Aesthetics, CLIP, HPS v2, and ImageReward. It also reports that the random-image downgrade is essential: blurring or randomly swapping grid patches makes performance worse than SFT.

Load-bearing premise

The load-bearing premise is that a randomly selected training image is a valid 'losing' sample for any prompt, so widening the model's denoising-error gap between the correct pair and the mismatched pair improves generation quality; the paper offers no theoretical argument for why this particular signal helps, and its own ablations show other downgrades are harmful.

Editorial extensions

If this is right

  • SUDO can be applied to any text-to-image diffusion model without collecting human preference annotations, making image-level alignment as cheap as ordinary supervised fine-tuning.
  • On SD1.5, SUDO improves all five metrics over the base model on all three evaluation sets, with ImageReward rising from -14.81 to 71.00 on Pick-a-Pic V2, far above SFT's 45.03 and DPO's 4.13.
  • On SDXL, where SFT degrades the base model, SUDO still improves most metrics and reaches HPS win rates above 92 percent on all three sets.
  • The random-image downgrade is not interchangeable: blur and random-grid downgrades fall below SFT, so the specific choice of losing sample matters.
  • Removing the MSE loss leaves SUDO's gains largely intact, indicating the image-level preference term alone carries most of the benefit.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A testable implication the authors do not pursue: if the benefit comes from the model learning to separate matched from mismatched pairs, then harder negatives, images that are semantically related to the prompt but wrong, should give larger gains than uniformly random images.
  • Since the paper notes the random-image strategy may fail on very small training sets, a natural extension is curriculum or retrieved negatives from a larger pool, which would also clarify whether the signal is pure contrastive regularization rather than preference optimization.
  • The paper's framing as 'preference' optimization is one interpretation; equally consistent with the data is that SUDO acts as a form of data augmentation or regularization that prevents fine-tuning from overfitting to the pixel-level MSE loss. A comparison against a standard contrastive loss on the same pairs would separate these explanations.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper introduces SUDO, a self-supervised direct preference optimization method for fine-tuning text-to-image diffusion models. In each training iteration, the input image is treated as the winning sample and a randomly selected image from the training set is used as the losing sample; the DPO-style loss in Eq. (8) is then added to the standard MSE loss. Experiments on Stable Diffusion 1.5 and SDXL, evaluated on Pick-a-Pic V2, PartiPrompts, and HPDv2 with five automatic metrics, are reported to show that SUDO improves over SFT and the base models and is competitive with Diffusion-DPO, without using preference annotations.

Significance. If the reported results are robust, SUDO offers a practical alternative to SFT for improving text-to-image models with no additional human preference labels, which is a valuable contribution. The paper provides a wide range of quantitative comparisons, ablations on the downgrade operation, and qualitative examples. However, the mechanism behind the method is not clearly established, the reported gains are not consistent across all metrics and models, and the statistical basis of the claims is weak because all numbers are point estimates without error bars or significance tests.

major comments (3)
  1. [Section 3.2, Eq. (8) and Table 3] The only 'Downgrade' operation that improves performance is random text-image mismatch, not an image-quality degradation; blur and random grid both perform worse than SFT. This directly contradicts the paper's assertion that the losing sample is obtained by 'degrading the winner' and undermines the interpretation of SUDO as a direct preference optimization method over image quality. The authors should either provide an explicit analysis of why this operation helps (for example, as a contrastive text-image alignment regularizer) or substantially reframe the method and its claims away from 'preference optimization' and 'global image-level learning.'
  2. [Table 2] The claim of 'consistently enhancing model performance' and of 'significant enhancements' compared to DPO is overstated. On SDXL, SUDO's PickScore is lower than DPO on all three evaluation sets (22.34 vs 22.57 on Pick-a-Pic V2, 22.79 vs 22.90 on PartiPrompts, 22.98 vs 23.18 on HPDv2), and SUDO is below the base model on Aesthetics on Pick-a-Pic (59.97 vs 60.01). Win rates also favor DPO on several metrics (e.g., Pick-a-Pic V2 PickScore win rate 60.80 vs 72.60). The text should be revised to describe SUDO as competitive with DPO on some metrics and superior on others, not uniformly better.
  3. [Section 4.2 and Table 2] No error bars, multiple seeds, or significance tests are reported for any of the quantitative results. The terms 'significant', 'substantially higher', and 'far exceeds' are used throughout without statistical quantification. The authors should report standard deviations over at least three seeds or paired significance tests for the main comparisons, especially for the win rates, before claiming consistent improvements.
minor comments (5)
  1. [Table 1 and Section 4.1] The description of SUDO's data requirement as 'standard image-text pairs' is imprecise because the training data is the Pick-a-Pic V2 dataset, which was collected through user preferences; only the preference annotations are unused. This should be clarified to avoid implying the dataset is arbitrary image-text pairs.
  2. [Section 4.3] The sentence 'We provide qualitative comparisons in Figure 4 and Figure 1 for SD1.5 and SDXL' is incorrect; the SDXL qualitative results appear in Figure 5 in the main text and Appendix Figures 1 and 2, not in Figure 1. Please correct the figure cross-references.
  3. [Section 3.2 and Eq. (8)] Equation (8) is typeset with broken line breaks and subscripts in the submitted manuscript, which makes the loss definition hard to read. In addition, there are typographical errors such as 'wining' instead of 'winning', 'encourages' instead of 'encourages', and 'demonstraing' instead of 'demonstrating'.
  4. [Table 2] The 'Win rate' rows are not defined in the caption or text; please specify how a win is determined for each metric, as this is important for interpreting the pairwise comparisons.
  5. [Section 5] The limitation that the random-image strategy 'may prove ineffective' for very small datasets is stated without supporting evidence; a small-scale experiment varying dataset size would make this claim more concrete.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: SUDO's self-supervised preference construction is a new training input, and the central claims are evaluated against external benchmarks.

full rationale

The paper's derivation chain is not circular. SUDO takes the Diffusion-DPO objective from prior external work [37] and replaces human-annotated losing samples with a self-supervised 'Downgrade' operation, implemented as randomly selecting images from the training dataset (Eq. 8 and Section 3.2). This is a new input to an existing framework, not a fitted quantity or a conclusion derived from the same data it predicts. The reported improvements are measured on held-out prompts using independent external metrics (PickScore, Aesthetics, CLIP, HPSv2, ImageReward), so the headline claim is not forced by construction. Hyperparameters such as C, lambda1, and lambda2 are fixed training choices, not fitted to the evaluation sets. There are no self-citations, no imported uniqueness theorems, and no prediction that is equivalent to an input by definition. The observation in Table 3 that blur and random grid downgrades hurt performance is an empirical finding, not a circular reduction; the naming of a random mismatched image as a 'downgrade' is a heuristic framing issue rather than a logical circularity. The paper's limitation statement about small training datasets is similarly a practical caveat, not evidence that the method's output is presupposed by its input. Overall, the central claim has independent empirical content and no circular step is exhibited.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claim rests on a DPO formulation inherited from prior work and on one ad hoc assumption about the validity of random mismatched pairs as losing samples. The loss weights are hand-chosen, but the ablation in Table 4 suggests robustness to the MSE weight.

free parameters (3)
  • lambda1 = 0.5
    Weight for the MSE loss in the combined objective (Eq. 9). Chosen by hand, not justified by a principled criterion. Ablation in Table 4 shows SUDO without MSE performs similarly, so the exact value is not critical to the main claim.
  • lambda2 = 0.5
    Weight for the SUDO loss in Eq. 9. Chosen by hand, equal to lambda1. Its value is not optimized or derived.
  • C = -2500
    Scale factor inside the log-sigmoid in Eq. 8, inherited from Diffusion-DPO [37]. Not tuned for SUDO specifically, but it is a free scaling choice in the loss.
assumptions (3)
  • domain assumption The diffusion denoising process can be treated as a multi-step MDP, and the DPO objective from Diffusion-DPO applies to diffusion models with step-wise rewards.
    The paper builds directly on the derivation in [37] (Eqs. 4-7) without re-deriving or testing its validity for the self-supervised pair construction.
  • ad hoc to paper A randomly selected image from the training dataset is a valid 'losing' sample for any given prompt.
    This is the key assumption of SUDO. It is empirically validated only indirectly through final metrics; the ablation shows the specific choice of downgrade matters.
  • ad hoc to paper Increasing the gap between the model's denoising errors on aligned vs. mismatched pairs translates to better image quality and text alignment.
    The paper's hypothesis that this loss captures 'global-level' information is not directly tested; it is supported only by downstream metrics.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SUDO: Enhancing Text-to-Image Diffusion Models with Self-Supervised Direct Preference Optimization." pith.science (2026). https://pith.science/paper/CWLYQQUH

@misc{pith2026250414534,
  author       = {Pith},
  title        = {Pith review of: SUDO: Enhancing Text-to-Image Diffusion Models with Self-Supervised Direct Preference Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CWLYQQUH}},
  note         = {Machine review of arXiv:2504.14534}
}
read the original abstract

Previous text-to-image diffusion models typically employ supervised fine-tuning (SFT) to enhance pre-trained base models. However, this approach primarily minimizes the loss of mean squared error (MSE) at the pixel level, neglecting the need for global optimization at the image level, which is crucial for achieving high perceptual quality and structural coherence. In this paper, we introduce Self-sUpervised Direct preference Optimization (SUDO), a novel paradigm that optimizes both fine-grained details at the pixel level and global image quality. By integrating direct preference optimization into the model, SUDO generates preference image pairs in a self-supervised manner, enabling the model to prioritize global-level learning while complementing the pixel-level MSE loss. As an effective alternative to supervised fine-tuning, SUDO can be seamlessly applied to any text-to-image diffusion model. Importantly, it eliminates the need for costly data collection and annotation efforts typically associated with traditional direct preference optimization methods. Through extensive experiments on widely-used models, including Stable Diffusion 1.5 and XL, we demonstrate that SUDO significantly enhances both global and local image quality. The codes are provided at \href{https://github.com/SPengLiang/SUDO}{this link}.

Figures

Figures reproduced from arXiv: 2504.14534 by the authors.

Figure 1
Figure 1. Pixel-level and image-level optimization. In the conven [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. We develop SUDO, a method for fine-tuning text-to-image diffusion models. It incorporates direct preference optimization in a [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Different training process of SFT, DPO, and SUDO. We [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Qualitative comparisons with the SD1.5 base model. All results are generated with the same random seed. Comparing with SFT [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Qualitative comparisons with the SDXL base model. All results are generated using the same random seed. Please note that [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 1
Figure 1. Figure 1: Extra qualitative comparisons with the SD1.5 base model. [PITH_FULL_IMAGE:figures/full_fig_p011_1.png]
Figure 2
Figure 2. Figure 2: Extra qualitative comparisons with the SDXL base model. [PITH_FULL_IMAGE:figures/full_fig_p012_2.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 13 canonical work pages

  1. [1]

    Con- stitutional ai: Harmlessness from ai feedback

    Yuntao Bai, Saurav Kadavath, Sandipan Kundu, Amanda Askell, Jackson Kernion, Andy Jones, Anna Chen, Anna Goldie, Azalia Mirhoseini, Cameron McKinnon, et al. Con- stitutional ai: Harmlessness from ai feedback. arXiv preprint arXiv:2212.08073, 2022. 3

    arXiv 2022

  2. [2]

    Improving image generation with better captions

    James Betker, Gabriel Goh, Li Jing, Tim Brooks, Jianfeng Wang, Linjie Li, Long Ouyang, Juntang Zhuang, Joyce Lee, Yufei Guo, et al. Improving image generation with better captions. Computer Science. https://cdn. openai. com/papers/dall-e-3. pdf, 2(3):8, 2023. 3

    2023

  3. [3]

    Training diffusion models with reinforce- ment learning

    Kevin Black, Michael Janner, Yilun Du, Ilya Kostrikov, and Sergey Levine. Training diffusion models with reinforce- ment learning. arXiv preprint arXiv:2305.13301, 2023. 3

    arXiv 2023

  4. [4]

    Rank analysis of incomplete block designs: I

    Ralph Allan Bradley and Milton E Terry. Rank analysis of incomplete block designs: I. the method of paired compar- isons. Biometrika, 39(3/4):324–345, 1952. 4

    1952

  5. [5]

    Pixart- alpha: Fast training of diffusion transformer for photorealistic text-to-image synthesis

    Junsong Chen, Jincheng Yu, Chongjian Ge, Lewei Yao, Enze Xie, Yue Wu, Zhongdao Wang, James Kwok, Ping Luo, Huchuan Lu, et al. Pixart- alpha: Fast training of diffusion transformer for photorealistic text-to-image synthesis. arXiv preprint arXiv:2310.00426, 2023. 1, 3

    arXiv

  6. [6]

    A simple framework for contrastive learning of visual representations

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Ge- offrey Hinton. A simple framework for contrastive learning of visual representations. In International conference on ma- chine learning, pages 1597–1607. PmLR, 2020. 3

    2020

  7. [7]

    Directly fine-tuning diffusion models on differentiable re- wards

    Kevin Clark, Paul Vicol, Kevin Swersky, and David J Fleet. Directly fine-tuning diffusion models on differentiable re- wards. arXiv preprint arXiv:2309.17400, 2023. 3

    arXiv 2023

  8. [8]

    Emu: Enhanc- ing image generation models using photogenic needles in a haystack

    Xiaoliang Dai, Ji Hou, Chih-Yao Ma, Sam Tsai, Jialiang Wang, Rui Wang, Peizhao Zhang, Simon Vandenhende, Xi- aofang Wang, Abhimanyu Dubey, et al. Emu: Enhanc- ing image generation models using photogenic needles in a haystack. arXiv preprint arXiv:2309.15807, 2023. 1, 3

    arXiv 2023

Show all 44 references
  1. [9]

    Diffusion models beat gans on image synthesis

    Prafulla Dhariwal and Alexander Nichol. Diffusion models beat gans on image synthesis. Advances in neural informa- tion processing systems, 34:8780–8794, 2021. 3

    2021

  2. [10]

    Raft: Reward ranked finetuning for generative foundation model alignment

    Hanze Dong, Wei Xiong, Deepanshu Goyal, Yihan Zhang, Winnie Chow, Rui Pan, Shizhe Diao, Jipeng Zhang, Kashun Shum, and Tong Zhang. Raft: Reward ranked finetuning for generative foundation model alignment. arXiv preprint arXiv:2304.06767, 2023. 3

    2023 arXiv

  3. [11]

    Scaling recti- fied flow transformers for high-resolution image synthesis

    Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, et al. Scaling recti- fied flow transformers for high-resolution image synthesis. In Forty-first international conference on machin...

  4. [12]

    Re- inforcement learning for fine-tuning text-to-image diffusion models

    Ying Fan, Olivia Watkins, Yuqing Du, Hao Liu, Moonkyung Ryu, Craig Boutilier, Pieter Abbeel, Moham- mad Ghavamzadeh, Kangwook Lee, and Kimin Lee. Re- inforcement learning for fine-tuning text-to-image diffusion models. In Thirty-seventh Conference on Neural Informa- tion Proce...

    2023

  5. [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. Advances in neural information processing systems, 27, 2014. 3

    2014

  6. [14]

    Optimizing prompts for text-to-image generation

    Yaru Hao, Zewen Chi, Li Dong, and Furu Wei. Optimizing prompts for text-to-image generation. Advances in Neural Information Processing Systems, 36:66923–66939, 2023. 3

    2023

  7. [15]

    Momentum contrast for unsupervised visual rep- resentation learning

    Kaiming He, Haoqi Fan, Yuxin Wu, Saining Xie, and Ross Girshick. Momentum contrast for unsupervised visual rep- resentation learning. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition , pages 9729–9738, 2020. 3

    2020

  8. [16]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, and Ross Girshick. Masked autoencoders are scalable vision learners. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 16000– 16009, 2022. 3

    2022

  9. [17]

    Denoising dif- fusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising dif- fusion probabilistic models. Advances in neural information processing systems, 33:6840–6851, 2020. 3

    2020

  10. [18]

    Auto-encoding vari- ational bayes, 2013

    Diederik P Kingma, Max Welling, et al. Auto-encoding vari- ational bayes, 2013. 3

    2013

  11. [19]

    Pick-a-pic: An open dataset of user preferences for text-to-image generation

    Yuval Kirstain, Adam Polyak, Uriel Singer, Shahbuland Ma- tiana, Joe Penna, and Omer Levy. Pick-a-pic: An open dataset of user preferences for text-to-image generation. Advances in Neural Information Processing Systems , 36: 36652–36663, 2023. 6, 7, 8

    2023

  12. [20]

    Flux, 2024

    Black Forest Labs. Flux, 2024. Black Forest Labs. Flux,

    2024

  13. [21]

    Aligning text- to-image models using human feedback

    Kimin Lee, Hao Liu, Moonkyung Ryu, Olivia Watkins, Yuqing Du, Craig Boutilier, Pieter Abbeel, Mohammad Ghavamzadeh, and Shixiang Shane Gu. Aligning text- to-image models using human feedback. arXiv preprint arXiv:2302.12192, 2023. 3

    2023 arXiv

  14. [22]

    Aligning diffusion mod- els by optimizing human utility

    Shufan Li, Konstantinos Kallidromitis, Akash Gokul, Yusuke Kato, and Kazuki Kozuka. Aligning diffusion mod- els by optimizing human utility. Advances in Neural Infor- mation Processing Systems, 37:24897–24925, 2025. 3

    2025

  15. [23]

    Hunyuan-dit: A powerful multi-resolution diffusion transformer with fine-grained chi- nese understanding

    Zhimin Li, Jianwei Zhang, Qin Lin, Jiangfeng Xiong, Yanxin Long, Xinchi Deng, Yingfang Zhang, Xingchao Liu, Minbin Huang, Zedong Xiao, et al. Hunyuan-dit: A powerful multi-resolution diffusion transformer with fine-grained chi- nese understanding. arXiv preprint arXiv:2405.087...

    2024 arXiv

  16. [24]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017. 6

    2017 arXiv

  17. [25]

    Glide: Towards photorealistic image generation and editing with text-guided diffusion models.arXiv preprint arXiv:2112.10741, 2021

    Alex Nichol, Prafulla Dhariwal, Aditya Ramesh, Pranav Shyam, Pamela Mishkin, Bob McGrew, Ilya Sutskever, and Mark Chen. Glide: Towards photorealistic image generation and editing with text-guided diffusion models.arXiv preprint arXiv:2112.10741, 2021. 3

    2021 arXiv

  18. [26]

    Training language models to follow instructions with human feedback

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Car- roll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instructions with human feedback. Ad- vances in neural information processing systems, 35:...

    2022

  19. [27]

    Scalable diffusion models with transformers

    William Peebles and Saining Xie. Scalable diffusion models with transformers. In Proceedings of the IEEE/CVF inter- national conference on computer vision , pages 4195–4205,

  20. [28]

    Sdxl: Improving latent diffusion mod- els for high-resolution image synthesis

    Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna, and Robin Rombach. Sdxl: Improving latent diffusion mod- els for high-resolution image synthesis. arXiv preprint arXiv:2307.01952, 2023. 1, 2, 3, 6

    2023 arXiv

  21. [29]

    Aligning text-to-image diffusion models with reward backpropagation

    Mihir Prabhudesai, Anirudh Goyal, Deepak Pathak, and Ka- terina Fragkiadaki. Aligning text-to-image diffusion models with reward backpropagation. 2023. 3

    2023

  22. [30]

    Learning transferable visual models from natural language supervi- sion

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. In International conference on machine learning, ...

    2021

  23. [31]

    Direct preference optimization: Your language model is secretly a reward model

    Rafael Rafailov, Archit Sharma, Eric Mitchell, Christo- pher D Manning, Stefano Ermon, and Chelsea Finn. Direct preference optimization: Your language model is secretly a reward model. Advances in Neural Information Processing Systems, 36:53728–53741, 2023. 1, 3, 4

    2023

  24. [32]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695, 2022. 1, 2, 3, 6

    2022

  25. [33]

    Photorealistic text-to-image diffusion models with deep language understanding

    Chitwan Saharia, William Chan, Saurabh Saxena, Lala Li, Jay Whang, Emily L Denton, Kamyar Ghasemipour, Raphael Gontijo Lopes, Burcu Karagol Ayan, Tim Salimans, et al. Photorealistic text-to-image diffusion models with deep language understanding. Advances in Neural Information...

    2022

  26. [34]

    Laion-aesthetics

    Christoph Schuhmann. Laion-aesthetics. https : / / laion.ai/blog/laion- aesthetics/ , 2022. Ac- cessed: 2023-11-10. 6

    2022

  27. [35]

    Adafactor: Adaptive learning rates with sublinear memory cost

    Noam Shazeer and Mitchell Stern. Adafactor: Adaptive learning rates with sublinear memory cost. In International Conference on Machine Learning, pages 4596–4604. PMLR,

  28. [36]

    Neural discrete representation learning

    Aaron Van Den Oord, Oriol Vinyals, et al. Neural discrete representation learning. Advances in neural information pro- cessing systems, 30, 2017. 3

    2017

  29. [37]

    Diffusion model align- ment using direct preference optimization

    Bram Wallace, Meihua Dang, Rafael Rafailov, Linqi Zhou, Aaron Lou, Senthil Purushwalkam, Stefano Ermon, Caiming Xiong, Shafiq Joty, and Nikhil Naik. Diffusion model align- ment using direct preference optimization. In Proceedings of the IEEE/CVF Conference on Computer Vision a...

    2024

  30. [38]

    Human preference score v2: A solid benchmark for evaluating human preferences of text-to-image synthesis

    Xiaoshi Wu, Yiming Hao, Keqiang Sun, Yixiong Chen, Feng Zhu, Rui Zhao, and Hongsheng Li. Human preference score v2: A solid benchmark for evaluating human preferences of text-to-image synthesis. arXiv preprint arXiv:2306.09341 ,

    arXiv

  31. [39]

    Sana: Efficient high-resolution image syn- thesis with linear diffusion transformers

    Enze Xie, Junsong Chen, Junyu Chen, Han Cai, Haotian Tang, Yujun Lin, Zhekai Zhang, Muyang Li, Ligeng Zhu, Yao Lu, et al. Sana: Efficient high-resolution image syn- thesis with linear diffusion transformers. arXiv preprint arXiv:2410.10629, 2024. 1

    arXiv

  32. [40]

    Imagere- ward: Learning and evaluating human preferences for text- to-image generation

    Jiazheng Xu, Xiao Liu, Yuchen Wu, Yuxuan Tong, Qinkai Li, Ming Ding, Jie Tang, and Yuxiao Dong. Imagere- ward: Learning and evaluating human preferences for text- to-image generation. Advances in Neural Information Pro- cessing Systems, 36:15903–15935, 2023. 6

    2023

  33. [41]

    Using human feedback to fine-tune diffusion models without any reward model

    Kai Yang, Jian Tao, Jiafei Lyu, Chunjiang Ge, Jiaxin Chen, Weihan Shen, Xiaolong Zhu, and Xiu Li. Using human feedback to fine-tune diffusion models without any reward model. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 8941– 895...

    2024

  34. [42]

    Scaling autoregres- sive models for content-rich text-to-image generation

    Jiahui Yu, Yuanzhong Xu, Jing Yu Koh, Thang Luong, Gun- jan Baid, Zirui Wang, Vijay Vasudevan, Alexander Ku, Yin- fei Yang, Burcu Karagol Ayan, et al. Scaling autoregres- sive models for content-rich text-to-image generation. arXiv preprint arXiv:2206.10789, 2(3):5, 2022. 6, 8

    2022 arXiv

  35. [43]

    Self-play fine-tuning of diffusion models for text-to-image generation

    Huizhuo Yuan, Zixiang Chen, Kaixuan Ji, and Quanquan Gu. Self-play fine-tuning of diffusion models for text-to-image generation. Advances in Neural Information Processing Sys- tems, 37:73366–73398, 2025. 3

    2025

  36. [44]

    Dspo: Direct score preference optimization for diffusion model alignment

    Huaisheng Zhu, Teng Xiao, and Vasant G Honavar. Dspo: Direct score preference optimization for diffusion model alignment. In The Thirteenth International Conference on Learning Representations. 3, 4 Appendix for SUDO: Enhancing Text-to-Image Diffusion Models with Self-Supervis...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.