Pith. sign in

REVIEW 3 major objections 5 minor 26 references

Capturing Conditional Dependence via Auto-regressive Diffusion Models

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 Auto-regressive diffusion models provably capture the conditional dependencies that vanilla diffusion models miss.

desk verdict First AR-diffusion sampling theory, but the headline comparison rests on an assumption that excludes the paper's own Gaussian counterexample. read the letter →

arxiv 2504.21314 v1 pith:RLBPYJ43 submitted 2025-04-30 cs.LG stat.ML

Xunpeng Huang , Yujin Han , Difan Zou , Yian Ma , Tong Zhang This is my paper · ORCID
classification cs.LGstat.ML
keywords auto-regressivediffusionconditionaldependencescore-basedgenerativemodelsKLdivergenceOrnstein-Uhlenbeckprocesssamplingerrorboundsmodelingtheorydistributions
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

This paper argues that vanilla diffusion models underperform on data with high-level, cross-patch relationships because they do not adequately represent conditional dependence, and that auto-regressive diffusion fixes this by generating the data patch by patch, each new patch conditioned on the already generated ones. Its main theorem gives the first convergence bound for auto-regressive diffusion: the KL divergence between the true and generated joint distributions is controlled by the per-patch score error, the discretization step size, and the OU mixing time, with gradient complexity only a factor of K larger than vanilla DDPM. The paper also proves a contrasting negative result: a vanilla diffusion model can match the joint distribution arbitrarily well while its conditional distributions are arbitrarily wrong, so the AR advantage is structural rather than a matter of tuning. If correct, auto-regressive diffusion is the practical choice whenever data has sequential or hierarchical dependencies, at a modest inference-time cost.

What carries the argument

The load-bearing object is the stage-wise forward Ornstein-Uhlenbeck process: the data vector is split into K patches, and the forward process is run K times, each stage adding noise to one patch while freezing the already-noised earlier patches; Algorithm 1 reverses this stage-wise process patch by patch. The argument is carried by the chain-rule decomposition of KL divergence, $\mathrm{KL}(p_*\|\hat p_*)\le \mathrm{KL}(p_{*,1}\|\hat p_{*,1})+\sum_{k=1}^{K-1}\mathbb{E}_{x_{[1:k]}\sim p_{*,[1:k]}}\mathrm{KL}(p_{*,k+1|[1:k]}(\cdot|x_{[1:k]})\|\hat p_{*,k+1|[1:k]}(\cdot|x_{[1:k]}))$, which turns joint generation quality into per-patch conditional convergence. Each conditional transition is analyzed with a reverse OU step whose discretization error is controlled by a time schedule that is linear early in the process and exponentially decaying near T, removing the need for score smoothness along the entire SDE trajectory.

What would settle it

Fit an AR diffusion and a vanilla diffusion to a two-patch Gaussian with a known linear conditional $p_*(x_2|x_1)$, then measure the empirical conditional KL between the generated and true conditionals: if the AR conditional KL grows with $|x_1|$ or fails to shrink as T increases and the score error goes to zero, the claimed reduced-gap phenomenon fails outside Assumption A2. A simpler check is to verify whether the proof can be rerun with Assumption A2 replaced by only the second-moment bound, since Lemma 4.2 shows the Gaussian motivating example violates A2.

Watch

Extended reading notes

Core claim

The paper establishes that auto-regressive diffusion---generating each data patch conditioned on previously generated patches---provably approximates the data's conditional distributions, and that this is a structural advantage over vanilla diffusion. Its main theorem shows that under Assumptions A1-A3, the generated joint distribution satisfies $\mathrm{KL}(p_*\|\hat p_*)\lesssim 2e^{-2T}L(m_0+d)+(L^2R\eta^2+T\eta)d+\eta m_0+\eta K R\epsilon_{\mathrm{score}}^2$, with gradient complexity $\tilde{O}(KL^2 d\epsilon^{-2})$. The proof routes through a stage-wise Ornstein-Uhlenbeck forward process and an adaptive reverse-time discretization, and it decomposes the joint error via the chain rule of KL divergence into a sum of expected conditional KL errors. The contrast is Lemma 4.2: for any tolerance, there exist Gaussian pairs whose joint KL divergence is as small as desired while the conditional KL divergence is arbitrarily large, so a vanilla diffusion model can match the joint distribution and still fail to capture dependencies. The paper also connects score estimation error to training loss: a K-patch AR model needs score error $\tilde{O}(K^{-1/2}\epsilon)$ and its inference complexity grows only by a factor K.

Load-bearing premise

The theorem's proof requires the data log-density to have a pointwise bounded gradient for all inputs (Assumption A2), which no Gaussian or other unbounded-support distribution satisfies.

Editorial extensions

If this is right

  • AR diffusion attains the bound $\mathrm{KL}(p_*\|\hat p_*)\lesssim 2e^{-2T}L(m_0+d)+(L^2R\eta^2+T\eta)d+\eta m_0+\eta K R\epsilon_{\mathrm{score}}^2$, with gradient complexity $\tilde{O}(KL^2d\epsilon^{-2})$---only a factor K worse than vanilla DDPM.
  • For any fixed conditioning tokens $x_{[1:k]}$, the generated next-patch distribution converges to $p_{*,k+1|[1:k]}(\cdot|x_{[1:k]})$ in KL divergence as T grows and the step size shrinks.
  • Vanilla diffusion has no such guarantee: Lemma 4.2 constructs Gaussian pairs with arbitrarily small joint KL but arbitrarily large conditional KL, so joint accuracy does not imply correct dependencies.
  • The global denoising objective (12) has the same gradient as the conditional score-matching objective (Lemma 3.1), so minimizing the practical training loss directly controls the score estimates required by Assumption A3.
  • In experiments with synthetic sun-shadow data and MNIST composites, AR diffusion satisfies inter-feature rules more often than DDPM when the patch order aligns with the dependence structure, and its training loss gap predicts inference-time performance.

Reading between the lines

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

  • If the theorem extends beyond Assumption A2, one testable prediction is that AR diffusion's advantage over vanilla diffusion grows with the strength of inter-patch dependence; a systematic sweep over correlation strength in synthetic Gaussian mixtures would quantify this.
  • The ablation study implies patch order is a design choice rather than a fixed hyperparameter, so a natural extension is to learn a patch ordering or condition graph that maximizes the expected reduction in conditional KL divergence.
  • The stage-wise argument suggests AR diffusion could serve as an evaluation protocol for world-model video generators: temporal patches must satisfy physical conditional laws, which is exactly the setting where the paper predicts a gap.
  • The factor-K inference cost and the $\sqrt{K}$ score-error allowance suggest a model-scale tradeoff: larger K improves conditional capture but demands more reverse steps, so a hierarchical coarse-to-fine patch schedule might preserve both benefits.
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 analyzes autoregressive (AR) diffusion models, where the data vector is split into K patches and each patch is generated by a separate reverse OU process conditioned on previously generated patches. The authors introduce a global score-matching objective, prove its equivalence to the practical denoising objective, and derive an upper bound on the KL divergence between the generated and true joint distributions (Theorem 4.3), with inference complexity of order \tilde{O}(K L^2 d \epsilon^{-2}). They also prove a negative result (Lemma 4.2) showing that closeness of joint distributions does not imply closeness of conditional distributions, which they use to argue that vanilla diffusion can fail to capture conditional dependence. Experiments on two synthetic image tasks and on concatenated MNIST digits support the claim that AR diffusion better captures dependencies when the patch order aligns with the underlying feature relationships.

Significance. If the central claims hold, this is a useful first theoretical treatment of AR diffusion sampling error and provides a principled explanation for a commonly observed empirical phenomenon. The proof is a substantial adaptation of existing diffusion-analysis machinery to a stage-wise OU process, and the explicit gradient-complexity bound, the training-objective equivalence, and the careful experiments with ablation on patch order are strengths. The significance is currently tempered by two gaps: the main theorem relies on a pointwise score-norm assumption that excludes Gaussian targets, so it does not cover the motivating counterexample; and the comparison with vanilla diffusion mixes a pointwise lower bound with an averaged upper bound. These issues are correctable in revision, so the paper has clear potential but needs further work before the advertised conclusions are fully supported.

major comments (3)
  1. [Section 2 (Assumption A2) and Theorem 4.3] Assumption A2 requires the global pointwise bound \|\nabla \ln p^*\| \le \sqrt{L}. This excludes every Gaussian target, since \|\Sigma^{-1}(x-\mu)\| is unbounded as \|x\| grows. The proof of Lemma 3.3 (Appendix A.2, Eq. (19)) uses this pointwise bound to control the variance term Var(\partial_x f^*) and to conclude that the marginal Hessian is bounded by 2L; without the pointwise gradient bound, only the expectation bound E\|\nabla f^*\|^2 \le Ld from Lemma D.2 is available, which leads to a dimension-dependent or non-uniform conditional-score Lipschitz constant. Because Theorem 4.3 relies on Lemma 3.3 through Lemma B.2, the main convergence theorem does not cover the Gaussian example used in Lemma 4.2 to motivate the failure of vanilla diffusion. Moreover, the abstract's phrase "under (possibly) the mildest data assumption" is not supported: the cited analyses of Chen et al. (2023) and Benton et al. (2024) require only second-moment and Hessian control, not a global pointwise gradient bound. Please either weaken A2, for example to a second-moment control of the score, and re-derive the affected constants, or explicitly restrict the scope of Theorem 4.3 and remove the "mildest" wording.
  2. [Section 4.1 (Lemma 4.2) and Theorem 4.3] Lemma 4.2 establishes a pointwise conditional-KL blow-up: for the constructed pair of Gaussians, KL(p^*(y|x)\|\hat{p}^*(y|x)) > M^2\|x\|^2 for a given x, while the joint KL is at most \epsilon. However, in this construction the marginal law of x has variance of order \epsilon^2/(d^2 M^2), so the averaged conditional KL, E_{x\sim p^*}[KL(p^*(y|x)\|\hat{p}^*(y|x))], is of order \epsilon^2/d and is small rather than divergent. By contrast, Theorem 4.3 and the proof of Lemma 4.1 bound the averaged conditional KL after taking x_{[1:k]} \sim p^*. The central comparison therefore mixes two different notions: an averaged upper bound for AR diffusion and a pointwise lower bound for vanilla diffusion. Furthermore, Lemma 4.2 is a purely distributional construction; it does not show that a vanilla score-based diffusion model trained to small joint score error will realize this bad conditional distribution. Please restate the comparison in a single metric, or provide an averaged lower bound, or clearly state that the lower bound is a generic failure of joint closeness and not a property specific to vanilla diffusion.
  3. [Section 4.1 and Abstract (scope of the 'reduced gap' claim)] The abstract and the introduction claim that AR diffusion 'produces samples with a reduced gap in approximating the data conditional distribution' compared with typical diffusion models. As written, this is a comparative claim about the two algorithms under comparable assumptions. But the upper bound for AR diffusion is established under A1-A3, which include the pointwise score bound in A2, while the lower bound for vanilla diffusion is a counterexample stated in terms of arbitrary Gaussian joint distributions and does not involve the score-estimation error or the reverse diffusion dynamics. Thus the two results are not established within a common setting. To make the comparative claim precise, the authors should either prove a lower bound for vanilla diffusion under the same kind of score-error assumption used for AR diffusion, or substantially weaken the comparative statement, for example to 'AR diffusion admits a conditional-Divergence bound under the stated assumptions, while joint-only convergence does not imply pointwise conditional convergence in general.'
minor comments (5)
  1. [Section 2 (Assumption A2)] The paper explicitly acknowledges in Section 2 that an additional gradient norm upper bound is required compared with previous work, which is in tension with the abstract's 'mildest' phrasing; please harmonize these statements in the final version.
  2. [Lemma 3.3] The displayed inequality in Lemma 3.3 appears to be missing the minus sign between the two score terms; as printed it resembles a product rather than a difference, and should be corrected to \|\nabla \ln p_{*,k|[1:k-1]}(x|y) - \nabla \ln p_{*,k|[1:k-1]}(x'|y)\|.
  3. [Lemma 4.2] The parameter M is declared as any real number, but the bound involves M^2; please state M>0 and clarify the notation x(1:d_{k+1}), including the dimension of x and the sense in which the first d_{k+1} coordinates are used.
  4. [Appendix B (Theorem B.1)] The constant c defined near the end of Theorem B.1 appears garbled: 'c = ln sqrt(1/(4L^2)+4 + 1/(2L))^2 < 1' mixes logarithms, square roots, and squares ambiguously; the proof would benefit from a clean definition of c and of the relation between \delta and c.
  5. [Remark 3 and experiments] The term 'gradient complexity' in Remark 3 actually refers to the number of reverse-time score evaluations at inference, not to training gradients; please rename it to something like 'score-evaluation complexity' or clarify the terminology. Also, the experimental estimate of the constant C in Eq. (65) is heuristic; please describe its limitations more explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central bound is a conditional statement under a standard oracle score-error assumption, and the derivation does not reduce to its inputs.

full rationale

The paper's central claim, Theorem 4.3, is a conditional statement under Assumptions A1-A3. Assumption A3 is an oracle upper bound on the average conditional score-matching error, which is the same type of hypothesis used in prior diffusion analyses (Chen et al. 2023; Benton et al. 2024); the theorem's conclusion is not obtained by definitionally plugging the target quantity into the bound. The proof of Theorem 4.3 decomposes the joint KL divergence via the standard chain rule and then bounds initialization error, score-estimation error, and discretization error separately; none of these error terms is, by construction, the left-hand side KL that the theorem claims to control. Lemma 4.2 is an explicit counterexample construction for vanilla diffusion, not a fitted prediction, and it does not enter the proof of the AR upper bound. The paper's self-citations, notably [12] for chain-rule lemmas (Appendix D.3/D.4) and [10] for the synthetic evaluation setup, are auxiliary and independently standard; they do not carry the load-bearing premise of the paper. The empirical estimation of the constant C in Appendix E is a heuristic comparison of training losses, not a prediction derived from the theorem, and it does not feed back into the theoretical derivation. The mismatch between Assumption A2's pointwise score bound and the Gaussian example used in Lemma 4.2 is a real assumption-coverage limitation and a correctness risk, but it is not circularity: the paper does not derive A2 from the conclusion or fit the conclusion into A2. Overall, no step in the claimed derivation chain is equivalent by construction to its inputs.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the standard diffusion-theory assumptions A1 and A3, plus the nonstandard A2 pointwise score bound. The paper introduces no new physical entities or hidden forces. The effective freedom lies in the hyperparameters T, η, R and the chosen patch partition, which the experiments tune to showcase the mechanism.

free parameters (4)
  • mixing time T
    Inference horizon for each patch; the theorem requires T = Θ(ln((m0+d)/ε^2)), not fitted to data.
  • step size η
    Maximum reverse-process step size; the theorem sets η = O(ε^2/(L^2 d ln(...))).
  • number of reverse steps R
    R = Θ((T + ln(1/δ))/η), chosen to satisfy the step-size schedule in the theorem.
  • patch partition (patch size and order) = patch size 16 (Task 1), 8 (Task 2)
    The AR model's ability to capture dependencies depends on the chosen patch ordering; the experiments deliberately pick sizes that align with dependencies in Task 1 and misalign them in Task 2.
assumptions (4)
  • domain assumption A1: bounded second moment of p* (E_{x~p*}∥x∥^2 ≤ m0)
    Standard in diffusion sampling analysis; used throughout the proof of Theorem 4.3 and Lemma 3.2.
  • domain assumption A2: ∥∇^2 ln p*∥ ≤ L and ∥∇ ln p*∥ ≤ √L
    The Hessian bound is standard, but the pointwise gradient bound is an extra restriction required to prove the dimension-free smoothness of conditional distributions in Lemma 3.3. It fails for Gaussian data and is acknowledged as additional in Remark 2.
  • domain assumption A3: score training error averaged over patches and timesteps is ≤ ε^2_score
    Oracle-type assumption standard in diffusion theory; it converts training error into sampling error but is not derived.
  • standard math KL chain rule and standard Gaussian LSI (Lemmas D.3, D.4, D.1)
    Borrowed from the cited literature, including the authors' own paper [12]; these are standard results.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Capturing Conditional Dependence via Auto-regressive Diffusion Models." pith.science (2026). https://pith.science/paper/RLBPYJ43

@misc{pith2026250421314,
  author       = {Pith},
  title        = {Pith review of: Capturing Conditional Dependence via Auto-regressive Diffusion Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RLBPYJ43}},
  note         = {Machine review of arXiv:2504.21314}
}
read the original abstract

Diffusion models have demonstrated appealing performance in both image and video generation. However, many works discover that they struggle to capture important, high-level relationships that are present in the real world. For example, they fail to learn physical laws from data, and even fail to understand that the objects in the world exist in a stable fashion. This is due to the fact that important conditional dependence structures are not adequately captured in the vanilla diffusion models. In this work, we initiate an in-depth study on strengthening the diffusion model to capture the conditional dependence structures in the data. In particular, we examine the efficacy of the auto-regressive (AR) diffusion models for such purpose and develop the first theoretical results on the sampling error of AR diffusion models under (possibly) the mildest data assumption. Our theoretical findings indicate that, compared with typical diffusion models, the AR variant produces samples with a reduced gap in approximating the data conditional distribution. On the other hand, the overall inference time of the AR-diffusion models is only moderately larger than that for the vanilla diffusion models, making them still practical for large scale applications. We also provide empirical results showing that when there is clear conditional dependence structure in the data, the AR diffusion models captures such structure, whereas vanilla DDPM fails to do so. On the other hand, when there is no obvious conditional dependence across patches of the data, AR diffusion does not outperform DDPM.

Figures

Figures reproduced from arXiv: 2504.21314 by the authors.

Figure 1
Figure 1. Visualization of Task 1 and Task 2. In Task 1, a patch size of 16 ensures that correlated [PITH_FULL_IMAGE:figures/full_fig_p012_1.png] view at source ↗
Figure 2
Figure 2. Comparison of AR Diffusion and DDPM Performance on Task 1 [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 3
Figure 3. Comparison of AR Diffusion and DDPM Performance on Task 2 [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The training loss variance with a sliding window of [PITH_FULL_IMAGE:figures/full_fig_p038_4.png]
Figure 5
Figure 5. Figure 5: Comparison of AR Diffusion with U-Net and DDPM with U-Net Performance on Sun-Shadow Setting [PITH_FULL_IMAGE:figures/full_fig_p040_5.png]
Figure 6
Figure 6. Figure 6: Comparison of AR Diffusion with MLP and DDPM with MLP Performance on Sun-Shadow Setting [PITH_FULL_IMAGE:figures/full_fig_p041_6.png]
Figure 7
Figure 7. Figure 7: Comparison of AR Diffusion with U-Net and DDPM with U-Net Performance on Odd and Even Setting [PITH_FULL_IMAGE:figures/full_fig_p042_7.png]
Figure 8
Figure 8. Figure 8: Comparison of AR Diffusion with U-Net and DDPM with U-Net Performance on Arithmetic Sequence Setting [PITH_FULL_IMAGE:figures/full_fig_p042_8.png]
Figure 9
Figure 9. Figure 9: AR Diffusion with Parallel Order. In this section, we conduct an ablation study on Task 1 by modifying only the learning order of AR Diffusion, changing it from the raster scan order to the parallel order illustrated in [PITH_FULL_IMAGE:figures/full_fig_p043_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 13 canonical work pages

  1. [1]

    D., Ho, J., Tarlow, D., and Van Den Berg, R

    Austin, J., Johnson, D. D., Ho, J., Tarlow, D., and Van Den Berg, R. (2021). Structured denoising diffusion models in discrete state-spaces.Advances in Neural Information Processing Systems, 34:17981–17993

    work page 2021

  2. [2]

    Benton, J., Bortoli, V., Doucet, A., and Deligiannidis, G. (2024). Nearly d-linear convergence bounds for diffusion models via stochastic localization

    work page 2024

  3. [3]

    Chen, H., Lee, H., and Lu, J. (2023a). Improved analysis of score-based generative modeling: User-friendly bounds under minimal smoothness assumptions. InInternational Conference on Machine Learning, pages 4735–4763. PMLR

    work page 2023

  4. [4]

    Chen, S., Chewi, S., Li, J., Li, Y., Salim, A., and Zhang, A. R. (2023b). Sampling is as easy as learning the score: theory for diffusion models with minimal data assumptions. InInternational Conference on Learning Representations. 14

    work page 2023

  5. [5]

    and Bartlett, P

    Cheng, X. and Bartlett, P. (2018). Convergence of langevin mcmc in kl-divergence. InAlgorithmic Learning Theory, pages 186–211. PMLR

    work page 2018

  6. [6]

    De Bortoli, V. (2022). Convergence of denoising diffusion models under the manifold hypothesis. arXiv preprint arXiv:2208.05314

    arXiv 2022

  7. [7]

    De Bortoli, V., Thornton, J., Heng, J., and Doucet, A. (2021). Diffusion schrödinger bridge with applications to score-based generative modeling.Advances in Neural Information Processing Systems, 34:17695–17709

    work page 2021

  8. [8]

    and Nichol, A

    Dhariwal, P. and Nichol, A. (2021). Diffusion models beat gans on image synthesis.Advances in neural information processing systems, 34:8780–8794

    work page 2021

Show all 26 references
  1. [9]

    Gupta, A., Yu, L., Sohn, K., Gu, X., Hahn, M., Li, F.-F., Essa, I., Jiang, L., and Lezama, J. (2024). Photorealistic video generation with diffusion models. InEuropean Conference on Computer Vision, pages 393–411. Springer

    2024

  2. [10]

    Han, Y., Han, A., Huang, W., Lu, C., and Zou, D. (2025). Can diffusion models learn hidden inter-feature rules behind images?arXiv preprint arXiv:2502.04725

    2025 arXiv

  3. [11]

    Ho, J., Jain, A., and Abbeel, P. (2020). Denoising diffusion probabilistic models.Advances in neural information processing systems, 33:6840–6851

    2020

  4. [12]

    Huang, X., Zou, D., Dong, H., Zhang, Y., Ma, Y.-A., and Zhang, T. (2024). Reverse transition kernel: A flexible framework to accelerate diffusion inference.arXiv preprint arXiv:2405.16387

    2024 arXiv

  5. [13]

    Kang, B., Yue, Y., Lu, R., Lin, Z., Zhao, Y., Wang, K., Huang, G., and Feng, J. (2024). How far is video generation from world model: A physical law perspective.arXiv preprint arXiv:2411.02385

    2024 arXiv

  6. [14]

    and Cai, C

    Li, G. and Cai, C. (2024). Provable acceleration for diffusion models under minimal assumptions. arXiv preprint arXiv:2410.23285

    2024 arXiv

  7. [15]

    and Yan, Y

    Li, G. and Yan, Y. (2024).o(d/t) convergence theory for diffusion probabilistic models under minimal assumptions. arXiv preprint arXiv:2409.18959

    2024 arXiv

  8. [16]

    Li, T., Tian, Y., Li, H., Deng, M., and He, K. (2024). Autoregressive image generation without vector quantization.arXiv preprint arXiv:2406.11838

    2024 arXiv

  9. [17]

    C., Liu, D., Kahatapitiya, K., Jia, M., et al

    Liu, H., Liu, S., Zhou, Z., Xu, M., Xie, Y., Han, X., Pérez, J. C., Liu, D., Kahatapitiya, K., Jia, M., et al. (2024). Mardini: Masked autoregressive diffusion for video generation at scale.arXiv preprint arXiv:2410.20280

    2024 arXiv

  10. [18]

    Luo, Z., Chen, D., Zhang, Y., Huang, Y., Wang, L., Shen, Y., Zhao, D., Zhou, J., and Tan, T. (2023). Videofusion: Decomposed diffusion models for high-quality video generation.arXiv preprint arXiv:2303.08320

    2023 arXiv

  11. [19]

    Meng, L., Zhou, L., Liu, S., Chen, S., Han, B., Hu, S., Liu, Y., Li, J., Zhao, S., Wu, X., et al. (2024). Autoregressive speech synthesis without vector quantization. arXiv preprint arXiv:2407.08551. 15

    2024 arXiv

  12. [20]

    Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., and Chen, M. (2022). Hierarchical text- conditional image generation with clip latents. arxiv 2022.arXiv preprint arXiv:2204.06125

    2022 arXiv

  13. [21]

    L., Ghasemipour, K., Gontijo Lopes, R., Karagol Ayan, B., Salimans, T., et al

    Saharia, C., Chan, W., Saxena, S., Li, L., Whang, J., Denton, E. L., Ghasemipour, K., Gontijo Lopes, R., Karagol Ayan, B., Salimans, T., et al. (2022). Photorealistic text-to-image diffusion models with deep language understanding.Advances in neural information processing syst...

    2022

  14. [22]

    and Ermon, S

    Song, Y. and Ermon, S. (2019). Generative modeling by estimating gradients of the data distribution. Advances in neural information processing systems, 32

    2019

  15. [23]

    and Wibisono, A

    Vempala, S. and Wibisono, A. (2019). Rapid convergence of the unadjusted langevin algorithm: Isoperimetry suffices. Advances in neural information processing systems, 32

    2019

  16. [24]

    Zhang, Q., Dai, X., Yang, N., An, X., Feng, Z., and Ren, X. (2024). Var-clip: Text-to-image generator with visual auto-regressive modeling.arXiv preprint arXiv:2408.01181. 16 A Notations in Appendix Remark 4. With the OU (Eq. 3) and reverse OU process (Eq. 4), standard Gaussia...

    2024 arXiv

  17. [25]

    For anya∈ Rd, the two processes satisfy the uniqueness and regularity condition stated in Lemma D.5, that is, 60 and 61 have unique solution and˜pt|t′ k (·|a), ˆqt|t′ k (·|a)∈C2(Rd) fort>t ′ k

  18. [26]

    (62) Lemma D.7(Lemma C.9 in [3])

    For a.e.a∈ Rd (with respect to the Lebesgue measure), we have lim t→t′ k + KL ˜pt|t′ k (·|a) ˆqt|t′ k (·|a) = 0. (62) Lemma D.7(Lemma C.9 in [3]). Suppose that Assumption 3 holds. Ifσ2 t≤ αt 2L, we have∇ logpt is 2Lα−1 t -Lipschitz on Rd. Lemma D.8(Lemma C.6 in [3]). For any0≤...

Pith tools

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