Pith. sign in

REVIEW 4 major objections 5 minor 8 cited by

On Statistical Rates of Conditional Diffusion Transformers: Approximation, Estimation and Minimax Optimality

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Pith's one-line read Conditional diffusion transformers can be minimax-optimal estimators of smooth densities.

desk verdict Solid approximation and estimation rates for conditional DiTs, but the minimax optimality claim rests on an unjustified constant-setting and should not survive as stated. read the letter →

arxiv 2411.17522 v1 pith:WZO5S3VF submitted 2024-11-26 stat.ML cs.AIcs.CVcs.LG

classification stat.MLcs.AIcs.CVcs.LG MSC 62G0562G2068T07
keywords diffusiontransformersconditionalmodelsclassifier-freeguidanceminimaxoptimalityscoreapproximationestimationtotalvariationdistanceHöldersmoothness
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 derives finite-sample statistical rates for conditional diffusion transformers (DiTs) trained with classifier-free guidance. Working under Hölder-smooth conditional densities, the authors prove bounds for score approximation, score estimation, and distribution estimation in total variation, for both the standard and the latent linear-subspace setting. The central payoff is that, under a stronger Hölder assumption, the upper bound for unconditional DiTs matches the known lower bound $n^{-\beta/(d_x+2\beta)}$, making the unconditional DiT a minimax-optimal distribution estimator. These rates matter because they say when transformer-based diffusion models are statistically efficient, and they make the dependence of sample complexity on data smoothness and intrinsic dimension explicit.

What carries the argument

The engine is a two-stage approximation: discretize the input and label domains into $N^{d_x+d_y}$ infinitesimal hypercubes and perform a term-by-term Taylor expansion of the Gaussian-smoothed conditional density $p_t(x|y)=\int p(x_0|y)\exp(-\|x-\alpha_t x_0\|^2/(2\sigma_t^2))\,dx_0$. This turns the score $\nabla p_t/p_t$ into polynomial pieces and algebraic operators (product, inverse, square root of the noise schedule) that single-layer self-attention transformers can approximate with controlled error. Under the stronger assumption that $p(x_0|y)=e^{-C_2\|x_0\|^2/2}f(x_0,y)$ with $f\ge C>0$, the score splits into a linear term plus a well-behaved ratio $\nabla h/h$, removing the threshold that causes extra log factors in the generic case.

What would settle it

Take a smooth density of the form $e^{-C_2\|x\|^2/2}f(x)$ with $f\in H^\beta$ but $f=0$ on a small ball, sample $n$ points, and measure the total-variation error of a trained unconditional DiT; if it still achieves $n^{-\beta/(d_x+2\beta)}$, the positivity lower bound is unnecessary, while a degraded error exponent would confirm that the minimax claim relies on it.

Watch

Extended reading notes

Core claim

Under the paper's assumptions, a conditional DiT score network can approximate the conditional score $\nabla\log p_t(x|y)$ with error scaling like $N^{-\beta/(d_x+d_y)}$ under generic Hölder smoothness and $N^{-2\beta/(d_x+d_y)}$ under the stronger factored-density assumption; converting the grid resolution $N$ into sample size $n$ gives score-estimation and total-variation distribution-estimation bounds. When the conditioning label is removed ($d_y=0$) and the constants are tuned, the total-variation rate becomes $n^{-\beta/(d_x+2\beta)}$, exactly matching the minimax lower bound for estimating a smooth density. The latent version, with intrinsic dimension $d_0 \le d_x$, replaces $d_x$ by $d_0$ in these rates, so the latent unconditional DiT is also minimax optimal under the same assumptions.

Load-bearing premise

The load-bearing premise is Assumption 3.2: after factoring out a Gaussian, the conditional density must stay bounded below by $C>0$ everywhere; if the data has empty or low-density holes, the score function can blow up and the tight rates and minimax matching no longer follow.

Editorial extensions

If this is right

  • Score approximation error improves from $O((\log(1/\epsilon))^{d_x}/\sigma_t^4)$ under generic Hölder smoothness to $(\log(1/\epsilon))^{O(1)}/\sigma_t^2$ under the stronger factored-density assumption.
  • A tuned unconditional DiT is minimax-optimal in total variation for densities of the form $e^{-C_2\|x\|^2/2}f(x)$ with $f\in H^\beta$ and $f\ge C$, achieving the rate $n^{-\beta/(d_x+2\beta)}$.
  • Latent conditional DiTs inherit the same rates with the latent dimension $d_0$ in place of $d_x$, and the latent unconditional DiT matches the corresponding lower bound $n^{-\beta_0/(d_0+2\beta_0)}$.
  • The recommended early-stopping time $t_0=n^{-\beta/(d_x+d_y+\beta)}$ and terminal time $T\sim \log n$ are explicit enough to serve as practical guidance for choosing diffusion schedules.

Reading between the lines

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

  • The conditional model's own minimax optimality is not established; if the same lower-bound technique were extended to the conditional problem, the conditional DiT would likely be minimax as well, but that step is not in the paper.
  • The grid-and-Taylor technique should transfer to other transformer conditioning mechanisms, such as cross-attention, once a universal approximation theorem for those architectures is available; the paper explicitly leaves cross-attention and adaptive layer-norm variants unanalyzed.
  • A testable prediction is that datasets with low-density holes or rough conditional densities will need noticeably more samples than the Hölder-exponent formula suggests, since the tighter rates rely on $f$ staying bounded below by a positive constant.
  • The score decomposition that isolates the linear term also explains why Gaussian-like tails stabilize score matching: the remaining ratio term is bounded, so the network does not need to fit near-singular denominators.
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

4 major / 5 minor

Summary. The paper derives approximation, score-estimation, and distribution-estimation rates for conditional diffusion transformers (DiTs) in the in-context conditioning architecture, under generic and stronger Hölder smoothness assumptions, and extends the analysis to latent conditional DiTs under a linear subspace assumption. The main technical engine is a term-by-term Taylor expansion of the conditional score on a discretized grid, combined with a universal-approximation theorem for one-layer transformers. The headline claim is that, by reducing the conditional analysis to the unconditional setting, unconditional DiTs achieve the minimax optimal total-variation rate n^{-\beta/(dx+2\beta)}, with a parallel claim for latent unconditional DiTs.

Significance. If the central claims were fully established, the paper would make a substantial contribution: it would provide the first exact minimax optimality result for diffusion transformers, together with explicit parameter-norm bounds and sample-complexity rates for conditional and latent DiTs. The appendix contains an extensive proof apparatus, including transformer universal approximation lemmas with quantitative parameter bounds, covering-number arguments, and a careful treatment of score blow-up via truncation. These are genuine strengths. However, the minimax optimality step rests on an unsupported constant-setting assertion, and this is load-bearing for the paper's headline contribution. The estimation-rate theorems may still be of interest, but the exact-optimality claim is not established as written.

major comments (4)
  1. [Section 3.4] The minimax matching is asserted rather than derived. The text states: 'By setting 2*nu3 = 1, we show that the unconditional DiT is the minimax optimal distribution estimator.' But nu3 is defined in Theorem 3.3 as nu3 = 4(12*beta*dx + 31*beta*d + 6*beta)/(d*(dx+dy)) + 12*(12*C_alpha*dx + 25*C_alpha*d + 6*C_alpha)/d + 72*C_sigma, and Theorem 3.4 has already fixed C_alpha = 2*beta/(dx+dy+2*beta) and C_sigma = beta/(dx+dy+beta). With dy = 0, nu3 is a concrete function of beta, dx, and d (with d = dx/L and L >= 2), and it is generally much larger than 1/2. The upper bound in Theorem 3.4 has exponent -(1/(2*nu3))*beta/(dx+dy+2*beta); matching the lower bound n^{-beta/(dx+2*beta)} requires nu3 = 1/2. No argument is given for this equality, and the paper's own definitions make it false in typical regimes. This is not a presentation issue: it is the only step that supports the abstract's and Section 3.4's minimax-optimality claim.
  2. [Remark 4.3 / Section 4] The same flaw appears in the latent result. Remark 4.3 states 'Setting 2*e_nu3 = 1, we obtain the minimax optimality of latent unconditional DiT.' The constant e_nu3 in Theorem 4.2 is again a fixed function of beta0, d0, ed, dy, C_alpha, and C_sigma, with C_alpha and C_sigma already chosen. The equality 2*e_nu3 = 1 is not derived from the preceding bounds; it is imposed to force the upper bound to meet the lower bound. Consequently, the latent minimax claim is unsupported for the same reason as the unconditional claim.
  3. [Section 3.4 and Corollary 3.4.2] Even if nu3 were treated as a free constant, imposing nu3 = 1/2 after the fact is circular. The quantity nu3 enters the proof through the choices t0 = N^{-C_sigma}, T = C_alpha log n, and through the covering-number and approximation bounds used to obtain Theorem 3.4. Changing nu3 to force the matching would change those choices and therefore change the very bound that is being matched. A valid minimax argument must either prove that nu3 = 1/2 under the stated assumptions or show that the exponents can be optimized independently; the paper does neither.
  4. [Corollary 3.4.2] The lower-bound class in Corollary 3.4.2 is stated as P = {p(x) = f(x) exp(-C2 ||x||^2/2) : f in H^beta(R^dx, B), f(x) >= C >= 0}. The upper bound in Theorem 3.4 is established under Assumption 3.2, which requires f(x,y) >= C for a positive constant C. Since the lower bound is stated for a potentially larger class (with f >= C >= 0, allowing C = 0), a lower bound on that larger class does not by itself transfer to a lower bound on the smaller class used by the upper bound. The relation between the two classes needs to be stated precisely; as written, the exact minimax conclusion does not follow from the cited lower bound.
minor comments (5)
  1. [Table 1] The 'Minimax Optimality' column contains the string '/reve' in the first and third rows; this appears to be a typo and should be fixed.
  2. [Section 3.4] The text says 'the distribution estimation error becomes eO(epsilon^{-1/(2*nu3)*beta/(dx+2*beta)})'; since the theorem is stated in terms of sample size n, this should likely read n, not epsilon.
  3. [Lemma D.3] The lemma statement begins 'Suppose Assumption Assumption D.3 holds'; the duplicated word should be removed.
  4. [Appendix G] After equation (G.21), the text begins with 'dk Applying the light tail property'; the stray 'dk' appears to be an editing artifact.
  5. [Theorem C.2] The theorem title reads 'under Generic Hölder Assumption' but the theorem is stated under Assumption 3.2 (the stronger Hölder assumption); the title appears mismatched.

Circularity Check

2 steps flagged · score 6.0 of 10

Minimax optimality rests on the bare assertion 'By setting 2ν3=1'; the non-minimax rate results are otherwise self-contained.

  1. fitted input called prediction [Section 3.4 (Minimax Optimal Estimation of Unconditional DiTs), after Theorem 3.4]
    "Then the distribution estimation error becomes eO(ϵ− 1 2ν3 β dx+2β ) under Assumption 3.2. Here eO(·) ignores the term aboutlog n. By setting 2ν3 = 1, we show that the unconditional DiT is the minimax optimal distribution estimator."

    The upper bound in Theorem 3.4 under Assumption 3.2 is O(n^{-(1/(2ν3))·β/(dx+2β)}) after removing the condition y (dy=0). The lower bound quoted in Corollary 3.4.2 is Ω(n^{-β/(dx+2β)}). These two bounds match only if 2ν3=1. But ν3 is not a free constant: it is defined in Theorem 3.3 as a fixed expression in β, dx, d, Cα, and Cσ, and Theorem 3.4 fixes Cα = 2β/(dx+dy+2β) and the early-stopping time. No derivation, parameter-selection argument, or feasibility check is given for the equality 2ν3=1; the paper simply imposes it after the bounds have been obtained. Thus the headline exact-minimax claim is the act of tuning a derived constant to meet the lower bound, rather than a consequence of the estimation analysis.

  2. fitted input called prediction [Section 4, Remark 4.3 (Minimax Optimal Estimation)]
    "Specifically, by applying Corollary 3.4.2 and substituting p(x|y) and dx by ph t (h|y) and d0 respectively in Assumption 3.2, we establish a distribution estimation lower bound of O(n−β0/(d0+2β0)). Setting 2eν3 = 1, we obtain the minimax optimality of latent unconditional DiT."

    The same post-hoc normalization is repeated for the latent case. The latent distribution estimation upper bound in Theorem 4.2 has exponent -(1/(2eν3))·β0/(d0+2β0), and equality with the lower bound Ω(n^{-β0/(d0+2β0)}) is achieved only by the asserted condition 2eν3=1. The quantity eν3 is a fixed combination of β0, d0, ed, Cα, and Cσ defined in Theorem 4.2; the paper supplies no derivation that this combination equals 1/2. Therefore the latent minimax optimality claim also reduces to an imposed equality between a derived constant and the value needed to match the external lower bound, rather than being established by the preceding approximation and estimation bounds.

full rationale

The paper's core approximation and estimation theorems (Theorems 3.1, 3.2, 3.3, 4.1, 4.2 and their corollaries) are derived from a detailed score-decomposition, Taylor-expansion, and transformer-universal-approximation argument; those rate results are not circular and stand on their own. The circularity is confined to the minimax-optimality claims. In Section 3.4, the unconditional DiT upper bound is O(n^{-(1/(2ν3))·β/(dx+2β)}), while the lower bound quoted from Fu et al. [2024b] is Ω(n^{-β/(dx+2β)}); the paper's only bridge is the sentence 'By setting 2ν3 = 1'. However, ν3 is an explicit function of the already-chosen constants β, dx, d, Cα, Cσ, and Theorem 3.4 fixes Cα and t0. The equality 2ν3=1 is never derived and is generally inconsistent with the stated choices; it is instead an imposed normalization that makes the upper exponent equal to the lower exponent. Remark 4.3 repeats the same move for the latent model with 'Setting 2eν3 = 1'. Because the central 'exact minimax optimality' headline therefore reduces to a fitted constant rather than to the proved bounds, the paper is partially circular. The score approximation, score estimation, and distribution estimation rates (without the minimax identification) are independent and self-contained, so the score is 6 rather than higher.

Assumptions & free parameters 4 free parameters · 7 assumptions · 0 invented entities

The central claim depends on the stated data assumptions and on the transformer universal approximation theorem. There are 4 hand-chosen scale and time constants, 7 axioms, and no invented entities.

free parameters (4)
  • C_alpha = 2*beta/(dx+dy+2*beta) in Theorem 3.4
    Hand-chosen terminal-time constant. Controls T = C_alpha * log N and appears in nu3; the minimax claim requires changing it to make 2*nu3=1.
  • C_sigma = beta/(dx+dy+beta) in Theorem 3.4
    Hand-chosen early-stopping constant. Sets t0 = N^(-C_sigma) and enters nu1 and nu3; no independent empirical basis is given.
  • N = n^(1/nu1 * (dx+dy)/(beta+dx+dy)) in estimation theorems
    Discretization resolution; chosen as a power of n to balance Taylor error and transformer approximation error. Not fitted to data but selected to optimize the rates.
  • t0, T = N^(-C_sigma), C_alpha * log N
    Early-stopping and terminal times; later set to powers of n in Theorem 3.4. These choices determine the final exponents and are not derived from the data distribution.
assumptions (7)
  • standard math Taylor expansion with Holder remainder on discretized domains
    Used in Lemmas G.2, G.3, and G.4 to approximate the diffused score by local polynomials on N^(dx+dy) grids.
  • standard math Universal approximation theorem for one-layer single-head transformers (Kajitsuka and Sato 2024, Theorem F.2)
    All score approximation results rely on this to replace Taylor polynomials and algebraic operators by transformer subnetworks.
  • domain assumption Assumption 3.1: conditional density is Holder smooth and light-tailed
    Defines the data class for the generic-rate Theorems 3.1, 3.3, and 3.4.
  • domain assumption Assumption 3.2: density factors as exp(-C2*||x||^2/2) * f(x,y) with f >= C > 0
    Needed for the score decomposition in equation (3.2), boundedness of h, and the tighter rates and minimax claims in Section 3.4.
  • domain assumption Assumption 4.1: x = U h with orthonormal U and h in R^(d0)
    Underlies all latent DiT results and the score decomposition in equation (4.1).
  • domain assumption KL(P0(.|y) || N(0,I)) <= c and finite latent KL in distribution estimation
    Assumed in Theorems 3.4, 4.2, and D.3 to control the initial-to-Gaussian transition error.
  • ad hoc to paper In-context conditional DiT architecture with concatenated label and time embeddings and latent encoder-decoder (Figures 1 and 3)
    The rates are proven only for this specific architecture, not for adaptive layer norm or cross-attention DiTs.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On Statistical Rates of Conditional Diffusion Transformers: Approximation, Estimation and Minimax Optimality." pith.science (2026). https://pith.science/paper/WZO5S3VF

@misc{pith2026241117522,
  author       = {Pith},
  title        = {Pith review of: On Statistical Rates of Conditional Diffusion Transformers: Approximation, Estimation and Minimax Optimality},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WZO5S3VF}},
  note         = {Machine review of arXiv:2411.17522}
}
read the original abstract

We investigate the approximation and estimation rates of conditional diffusion transformers (DiTs) with classifier-free guidance. We present a comprehensive analysis for ``in-context'' conditional DiTs under four common data assumptions. We show that both conditional DiTs and their latent variants lead to the minimax optimality of unconditional DiTs under identified settings. Specifically, we discretize the input domains into infinitesimal grids and then perform a term-by-term Taylor expansion on the conditional diffusion score function under H\"older smooth data assumption. This enables fine-grained use of transformers' universal approximation through a more detailed piecewise constant approximation and hence obtains tighter bounds. Additionally, we extend our analysis to the latent setting under the linear latent subspace assumption. We not only show that latent conditional DiTs achieve lower bounds than conditional DiTs both in approximation and estimation, but also show the minimax optimality of latent unconditional DiTs. Our findings establish statistical limits for conditional and unconditional DiTs, and offer practical guidance toward developing more efficient and accurate DiT models.

Figures

Figures reproduced from arXiv: 2411.17522 by the authors.

Figure 1
Figure 1. Conditional DiT Network Architecture. The architecture consists of a reshape layer R(·), a reversed reshape layer R−1 (·), and the embedding layers for label y and timestep t. The embeddings of y and t are concatenated with input sequences and then processed by a transformer network fT ∈ T h,s,r . use Lb(sW ) := 1 n Xn i=1 ℓ(x0,i, yi ; sW ), (2.1) as the empirical loss. In addition, we denote population loss as L(sW… view at source ↗
Figure 2
Figure 2. Approximate Score Function with Transformer Tscore under Assumption 3.1. The construc￾tion consists of the transformers to approximate local polynomials f1 and f2, and the algebraic operators. We highlight the overall term-by-term approximations and their corresponding lemmas to ensemble the transformers. Please see Appendix G for a detailed proof. Remark 3.3 (Approximation Rate). Given a fixed resolution N, the app… view at source ↗
Figure 3
Figure 3. Network Architecture of Latent Conditional DiT. The overall architecture consists of linear layer of encoder and decoder W⊤ U and WU that transform input x ∈ R dx into linear latent space R d0 , reshaping layer Re(·) and Re−1 (·), embedding layer for label y and timestep t. The embedding concatenates with input sequences and processes by the adapted transformer network T h,s,r Re = Re−1 ◦ gT ◦ f (FF) ◦ Re. W⊤ U ∈ R … view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Trapezoid function. For any x ∈ [0, 1]d0 and y ∈ [0, 1]dy , we define a partition of unity based on a product of 41 [PITH_FULL_IMAGE:figures/full_fig_p042_4.png]
Figure 5
Figure 5. Figure 5: Approximate Score Function under Assumption 3.2 with Transformer Tscore. The construc￾tion of the final score function consists of the approximation of diffused local polynomials f1 and f2 with transformer and transformer-approximate operators. We highlight the overall…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 8 Pith papers

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

  1. Curse of Attention: A Kernel-Based Perspective for Why Transformers Fail to Generalize on Time Series Forecasting and Beyond

    cs.LG 2024-12 reject novelty 7.0 of 10

    A theoretical analysis claims attention networks fail to learn residual features when time series steps have opposite signs, giving a possible explanation for the known advantage of linear residual models.

  2. The Computational Limits of State-Space Models and Mamba via the Lens of Circuit Complexity

    cs.CC 2024-12 conditional novelty 5.0 of 10

    Mamba and state-space models with constant depth and poly(n) precision are shown to lie in DLOGTIME-uniform TC0, the same circuit class as Transformers, so their expressiveness is bounded unless TC0=NC1.

  3. Force Matching with Relativistic Constraints: A Physics-Inspired Approach to Stable and Efficient Generative Modeling

    cs.LG 2025-02 reject novelty 4.0 of 10

    Force Matching replaces velocity matching in flow-based generative models with a relativistic force objective, but the toy experiments are designed so the model class matches the data generator exactly.

  4. Universal Approximation of Visual Autoregressive Transformers

    cs.LG 2025-02 reject novelty 4.0 of 10

    The paper's headline claim that VAR transformers universally approximate all Lipschitz image maps is not supported, because the theorem restricts the target class and its key lemma has an invalid linearity step.

  5. High-Order Matching for One-Step Shortcut Diffusion Models

    cs.CV 2025-02 reject novelty 4.0 of 10

    HOMO extends shortcut diffusion with acceleration and jerk supervision, but the proof of superior approximation is not supported and experiments lack error bars.

  6. On Computational Limits and Provably Efficient Criteria of Visual Autoregressive Models: A Fine-Grained Complexity Analysis

    cs.LG 2025-01 reject novelty 4.0 of 10

    Under SETH, the paper claims VAR models cannot be approximated faster than O(n^4) when attention entries are Theta(sqrt(log n)), but can be approximated in O(n^{2+o(1)}) when entries are o(sqrt(log n)).

  7. Circuit Complexity Bounds for Visual Autoregressive Model

    stat.ML 2025-01 reject novelty 4.0 of 10

    The authors show that a simplified formalization of the VAR image generation model lies in DLOGTIME-uniform TC0, meaning it can be simulated by constant-depth threshold circuits with polynomial size and precision.

  8. RichSpace: Enriching Text-to-Video Prompt Space via Text Embedding Interpolation

    cs.CV 2025-01 reject novelty 3.0 of 10

    Picking an interpolated text embedding with cosine similarity to a perpendicular foot can make a text-to-video model generate composite-feature videos that direct prompts miss.

Reference graph

Works this paper leans on

16 extracted references · 4 canonical work pages · cited by 8 Pith papers

  1. [1]

    Sumformer: Universal approxima- tion for efficient transformers

    Silas Alberti, Niclas Dern, Laura Thesing, and Gitta Kutyniok. Sumformer: Universal approxima- tion for efficient transformers. In Topological, Algebraic and Geometric Learning Workshops 2023, pages 72–86. PMLR,

  2. [5]

    Theory of consistency diffusion models: Distribution estimation meets fast sampling

    Zehao Dou, Minshuo Chen, Mengdi Wang, and Zhuoran Yang. Theory of consistency diffusion models: Distribution estimation meets fast sampling. In Forty-first International Conference on Machine Learning, 2024a. Zehao Dou, Subhodh Kotekal, Zhehao Xu, and Harrison H Zhou. From optimal score matching to optimal sampling. arXiv preprint arXiv:2409.07032, 2024b....

  3. [6]

    Diffusion Transformer Captures Spatial-Temporal Dependencies: A Theory for Gaussian Process Data

    Hengyu Fu, Zehao Dou, Jiawei Guo, Mengdi Wang, and Minshuo Chen. Diffusion transformer captures spatial-temporal dependencies: A theory for gaussian process data. arXiv preprint arXiv:2407.16134, 2024a. Hengyu Fu, Zhuoran Yang, Mengdi Wang, and Minshuo Chen. Unveil conditional diffusion mod- els with classifier-free guidance: A sharp statistical theory. a...

  4. [7]

    Exploring the frontiers of softmax: Provable optimization, applications in diffusion model, and beyond.arXiv preprint arXiv:2405.03251,

    Jiuxiang Gu, Chenyang Li, Yingyu Liang, Zhenmei Shi, and Zhao Song. Exploring the frontiers of softmax: Provable optimization, applications in diffusion model, and beyond.arXiv preprint arXiv:2405.03251,

  5. [8]

    Gradient guidance for diffusion models: An optimization perspective

    Yingqing Guo, Hui Yuan, Yukang Yang, Minshuo Chen, and Mengdi Wang. Gradient guidance for diffusion models: An optimization perspective. arXiv preprint arXiv:2404.14743,

  6. [10]

    Fun- damental limits of prompt tuning transformers: Universality, capacity and efficiency

    171 Jerry Yao-Chieh Hu, Wei-Po Wang, Ammar Gilani, Chenyang Li, Zhao Song, and Han Liu. Fun- damental limits of prompt tuning transformers: Universality, capacity and efficiency. arXiv preprint arXiv:2411.16525, 2024a. Jerry Yao-Chieh Hu, Weimin Wu, Zhuoru Li, Sophia Pi, , Zhao Song, and Han Liu. On statistical rates and provably efficient criteria of lat...

  7. [11]

    Accelerating con- vergence of score-based diffusion models, provably

    Gen Li, Yu Huang, Timofey Efimov, Yuting Wei, Yuejie Chi, and Yuxin Chen. Accelerating con- vergence of score-based diffusion models, provably. arXiv preprint arXiv:2403.03852, 2024a. Gen Li, Yuting Wei, Yuxin Chen, and Yuejie Chi. Towards non-asymptotic convergence for diffusion-based generative models. In The Twelfth International Conference on Learning...

  8. [12]

    Scalable diffusion models with transformers

    William S Peebles and Saining Xie. Scalable diffusion models with transformers. 2023 ieee. In CVF International Conference on Computer Vision (ICCV), volume 4172,

Show all 16 references
  1. [13]

    Linear convergence of diffusion models under the manifold hypothesis

    Peter Potaptchik, Iskander Azangulov, and George Deligiannidis. Linear convergence of diffusion models under the manifold hypothesis. arXiv preprint arXiv:2410.09046,

  2. [14]

    Nonparametric regression using deep neural networks with relu acti- vation function

    Johannes Schmidt-Hieber. Nonparametric regression using deep neural networks with relu acti- vation function. The Annals of Statistics, 2020,

  3. [16]

    Medsegdiff-v2: Diffusion- based medical image segmentation with transformer

    Junde Wu, Wei Ji, Huazhu Fu, Min Xu, Yueming Jin, and Yanwu Xu. Medsegdiff-v2: Diffusion- based medical image segmentation with transformer. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 6030–6038, 2024a. Yuchen Wu, Minshuo Chen, Zihao Li, ...

  4. [2020]

    doi: 10.1016/j.neunet.2019.12.014

    ISSN 0893-6080. doi: 10.1016/j.neunet.2019.12.014. Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance. arXiv preprint arXiv:2207.12598,

  5. [2021]

    Score-based diffusion models via stochastic differential equations–a technical tutorial

    173 Wenpin Tang and Hanyang Zhao. Score-based diffusion models via stochastic differential equations–a technical tutorial. arXiv preprint arXiv:2402.07487,

  6. [2022]

    PLOT: Prompt learning with optimal transport for vision-language models

    Guangyi Chen, Weiran Yao, Xiangchen Song, Xinyue Li, Yongming Rao, and Kun Zhang. PLOT: Prompt learning with optimal transport for vision-language models. In The Eleventh Interna- tional Conference on Learning Representations (ICLR), 2023a. Hongrui Chen, Holden Lee, and Jianfe...

  7. [2023]

    Conditional image generation with score-based diffusion models

    Georgios Batzolis, Jan Stanczuk, Carola-Bibiane Schönlieb, and Christian Etmann. Conditional image generation with score-based diffusion models. arXiv preprint arXiv:2111.13606,

  8. [2024]

    A short note on an inequality between kl and tv

    Clément L Canonne. A short note on an inequality between kl and tv. arXiv preprint arXiv:2202.07198,

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.