betaFeedback

WorksComputer Science

Forward-Value Discretization: Faster Stochastic Diffusion Sampling Without High-Order Approximation

Abstract

Stochastic diffusion samplers often require substantially more neural function evaluations (NFEs) than deterministic samplers to achieve comparable sample quality. Although higher-order discretization accelerates deterministic sampling, its benefits for stochastic sampling remains unclear. We establish matching upper and worst-case lower bounds showing that stochastic samplers with higher-order approximations remain first-order convergent, even with exact score functions. Thus, increasing the approximation order does not generally improve the convergence order. Motivated by this limitation, we introduce forward-value discretization, a training-free framework that evaluates the data/noise prediction model at a lookahead estimate constructed from previous predictions. For deterministic sampling, the resulting Forward DPMSolver (F-DPMSolver), improves sample quality at low NFE budgets through its DDIM-based variant, despite remaining first-order convergent. For stochastic sampling, we develop Forward DDPM (F-DDPM), which evaluates the model at a cleaner lookahead point adjusted for subsequent noise injection. This estimate reuses both the previous model prediction and the previously injected noise. Both variants admit first-order convergence guarantees with one model evaluation per iteration. Across CIFAR-10, ImageNet, FFHQ, and LSUN, F-DDPM consistently achieves lower FID than the evaluated stochastic baselines at matched NFE budgets. With 1.5 times as many NFEs as deterministic baselines, it achieves competitive sample quality, narrowing the practical efficiency gap between stochastic and deterministic sampling. Code is available at https://github.com/Na-Li66/F-DDPM.

Partial preliminary results of this work appeared in International Conference on Machine Learning 2026 [26].

Introduction

Diffusion probabilistic models (DPMs) have rapidly emerged as a leading paradigm in modern generative modeling, achieving state-of-the-art performance across a wide range applications in generative AI [62, 61, 59, 15] (Ho et al. 2020). Building on ideas from non-equilibrium thermodynamics [58], a DPM first defines a forward process that gradually injects noise into the clean data drawn from the true data distribution, and then generates new samples by reversing this process, turning pure noise into high-fidelity outputs. The reversal is driven by a pre-trained neural network that approximate the score functions of the noisy data distributions. DPMs now have shown remarkable success across various domains, including image synthesis [53, 52, 54], audio generation [32], language models [55] (Li et al. 2022; Li et al. 2026), and molecular design [22], underscoring their versatility and impact. See Yang et al. [65]; Croitoru et al. [13]; Lai et al. [34] for overviews of recent development.

Once a DPM is trained, sampling amounts to discretizing a differential equation that describes the reverse process: either the diffusion SDE [51, 2] (Ho et al. 2020) or the diffusion ODE [59] (Lu et al. 2022a). In practice, a DPM is typically trained as a noise prediction model: given a noisy sample and a time step, it predicts the noise added to the original clean data at that time in the forward process. Since subtracting this noise prediction from the noisy sample yields an estimate of the clean data, the same network equivalently acts as a clean-data predictor. As a result, sample quality relies on two factors: how the diffusion SDE/ODE is discretized, and how the pre-trained DPM is queried along the way.

Diffusion samplers can be broadly classified into two categories, depending on whether noise is injected beyond the initial step. Stochastic samplers (e.g., DDPM (Ho et al., 2020)) add Gaussian noise with during sampling iterations, while deterministic samplers (e.g., DDIM [59]) add none beyond initialization. Stochastic samplers tend to produce more diverse samples and give a more natural framework for classifier guidance, which often improves the quality of conditional samples. For example, Figure 1 compares Stable Diffusion samples generated with the prompts (a) “A hand pets a cat in a suitcase” and (b) “A rodeo area with men in cowboy hats, rodeo clowns and an audience in the stands”. For the first prompt, the stochastic samplers SDE-DPMSolver++1 and SDE-DPMSolver++2 (see (18) and (19) for their update rules) produce clearly recognizable cats, whereas the deterministic samplers DDIM and DPMSolver-3 (see (32) for the latter’s update rule) primarily depict the suitcase. For the second prompt, the stochastic samplers produce cowboy hats and audience in stands. The corresponding details are less clearly resolved in the deterministic samples. Nevertheless, these benefits come at a price. Empirically, stochastic samplers converge far more slowly than their deterministic counterparts. For instance, On CIFAR-10 with the EDM model, DDPM needs roughly 40 iterations to achieve lower than 3 in Fréchet inception distance (FID) [20], a metric used to assess the quality of images, whereas DDIM achieves comparable quality in fewer than 15 iterations. Since each iteration costs one neural function evaluation of the pre-trained DPM and sampling speed is often the primary bottleneck in deployment, this gap has driven growing interest in deterministic samplers.

Figure 1. Stable Diffusion samples at different NFEs for the cat and rodeo prompts described in the text. In these examples, SDE-DPMSolver++1 and SDE-DPMSolver++2 depict the cat and human figures more clearly than DDIM and DPMSolver-3.

Higher-order discretization is a widely used training-free approach to accelerating deterministic diffusion sampling (Lu et al., 2022a,b; Zhao et al., 2023). For stochastic samplers, however, increasing the order of a deterministic approximation does not automatically improve the strong convergence order to the underlying diffusion SDE. Strong convergence measures the discrepancy between the numerical and continuous trajectories driven by the same Brownian motion, whereas distributional convergence measures the discrepancy between their laws or between the generated and target distributions. In particular, we show that SDE-DPM-Solver++2 remains first-order convergent despite its higher-order approximation.

This gap motivates the central question of this work:

Can we design a fast stochastic sampler that narrows the performance gap with deterministic samplers?

Our contributions

We begin by analyzing the convergence rates of existing deterministic and stochastic samplers, including DDIM, DPMSolver(Lu et al., 2022a), and SDE-DPMSolver++(Lu et al., 2022a). For deterministic samplers, prior works have established that a higher-order approximation achieves a corresponding higher-order convergence rate. For instance, DPMSolver-pp converges at a rate of O~(M−p)\widetilde{O}(M^{-p}), where MM denotes the number of iterations. This work reveals that this correspondence between approximation order and convergence order does not hold for the stochastic SDE-DPM-Solver++ schemes. In particular, SDE-DPM-Solver++2 remains first-order convergent in the strong sense, as established by matching upper and worst-case lower bounds. The reason is that the injected noise makes the generated samples vary too much at each step, for a higher-order expansion to help. This offers one explanation for the practical inefficiency of stochastic samplers.

Next, we explore a simple but under-explored idea: a forward-value discretization of the diffusion ODE. Given the current iterate and time, the idealized forward-value update would evaluate the DPM at the quantities corresponding the next time rather than the current one. We observe that such an update scheme can perform effectively even in the extremely low-NFE regime, which motivates a practical sampler that approximates it. The resulting sampler, named Forward DPMSolver (F-DPMSolver), first produces a rough estimate of the next iterate using a vanilla update (e.g., DDIM) with cached model predictions, and then combines the current iterate with a DPM evaluation at this estimate. Critically, it maintains one model evaluation per iteration by reusing the previous evaluation, so its number of model evaluations (NFE) is the same as that of the vanilla sampler. Empirical results show that F-DPMSolver improves FID substantially over higher-order samplers including DPMSolver-2, DPMSolver-3 (Lu et al., 2022b), and UniPC-3 (Zhao et al., [71]), even though its convergence rate still scales as Θ~(1/M)\widetilde{\Theta}(1/M). These results convey a surprising insight: a first-order sampler can also achieve high practical sampling efficiency.

Inspired by this insight, we extend the forward-value idea to stochastic samplers and propose a fast first-order stochastic sampler, termed Forward DDPM (F-DDPM). Whereas the deterministic sampler evaluates the DPM at a lookahead estimate for the next time step, the stochastic sampler uses an estimate at a carefully chosen cleaner intermediate time to account for subsequent noise injection. To construct this estimate, we form a tailored noise estimate that combines the available noise prediction with the injected noise from the previous step. We also establish a first-order strong error upper bound for F-DDPM. Empirically, F-DDPM consistently improves FID over the evaluated stochastic baselines at the same NFEs and substantially narrows the performance gap with deterministic samplers. With 1.5 times as many NFEs, it achieves sample quality comparable to or better than the evaluated deterministic baselines in many tested settings.

Other related work

Training-free/based acceleration. Existing acceleration strategies for diffusion sampling can be broadly divided into training-free and training-based approaches. Training-free methods reuse a fixed pre-trained DPM and speed up sampling solely by modifying the sampling update rule, making them broadly applicable to off-the-shelf DPMs. The methods discussed in the introduction belong to this category. In contrast, training-based strategies introduce an additional stage of training (e.g., distillation (Luhman and Luhman, [49]; Salmans and Ho, [56]; Meng et al., [50]; Zhang et al., [69]) and consistency models (Li et al., 2024b; Song et al., [60])). They aim to shorten sampling trajectories by adapting pre-trained models into related architectures, at the cost of additional training computation. Since our focus is training-free acceleration, we next summarize related work from both practical and theoretical viewpoints.

Training-free acceleration: practice. Most practical training-free accelerations are driven by higher-order numerical discretizations of the reverse-time SDE/ODE. For ODE-based samplers, a prominent line of work exploits the structure of the diffusion ODE to design higher-order solvers, including DPM-Solver++ (Lu et al., 2022a,b), exponential-integrator-based methods (Zhang and Chen, [70]), and predictor-corrector-based schemes such as UniPC (Zhao et al., [71]). For SDE-based samplers, acceleration is comparatively less explored due to the intrinsic difficulty of SDE discretization, but notable progress includes stochastic Improved Euler’s methods (Jolicoeur-Martineau et al., [28]), stochastic Adams methods (Xue et al., [64]), and stochastic Runge–Kutta methods (Wu et al., 2024).

Training-free acceleration: theory. Convergence theory has been established for a wide range of diffusion samplers [10, 35, 7, 9, 3, 27, 68] (Li et al. 2023; Huang et al. 2024a; Li and Yan 2024; Li et al. 2025a; Cai et al. 2026), mostly focusing on standard DDPM and DDIM. For provable acceleration, most existing results likewise concentrate on higher-order discretization and on controlling the resulting discretization error for the reverse-time dynamics (Li and Cai, 2024; Li and Jiao, 2024; Jiao and Li, 2024; Huang et al., 2024a,b; Li et al., 2024a; Yu and Yu, 2025; Li et al., 2025b; Gupta et al., 2024; Wu et al., 2024). Beyond designing improved solvers, another line of work accelerates generation via parallel sampling that implements multiple denoising steps in parallel [57, 8, 19]. In addition to continuous diffusion models, sampling acceleration has also been studied for discrete diffusion models [11, 16, 72, 17, 12] (Li and Cai 2025; Cai and Li 2026).

Notation. For integer n≥1n \ge1, we denote [n]:={1,2,…,n}[n] := \{1, 2, \ldots, n\}. For vector x\boldsymbol{x}, we denote by ∥x∥2\|\boldsymbol{x}\|_2 or ∥x∥\|\boldsymbol{x}\| its ℓ2\ell_2 norm. For matrix A\boldsymbol{A}, we denote by ∥A∥\|\boldsymbol{A}\| or ∥A∥op\|\boldsymbol{A}\|_{\mathrm{op}} its operator norm. In the rest of the paper, we use λ(t)\lambda(t) or λt\lambda_t interchangeably to denote the mapping from the parameter tt to λ\lambda. For random vector xt\boldsymbol{x}_t defined in (1), we let qtq_t denote its probability density function. For function f(M)f(M), we write f(M)=O(M−k)f(M) = O(M^{-k}) if there exists some constant C>0C > 0 such that ∣f(M)∣≤CM−k|f(M)| \le CM^{-k} for all M≥1M \ge1. Similarly, we denote f(M)=Ω(M−k)f(M) = \Omega(M^{-k}) if ∣f(M)∣≥cM−k|f(M)| \ge cM^{-k} holds for some constant c>0c > 0, and f(M)=Θ(M−k)f(M) = \Theta(M^{-k}) if both f(M)=O(M−k)f(M) = O(M^{-k}) and f(M)=Ω(M−k)f(M) = \Omega(M^{-k}) hold. Finally, we say f(M)=o(M−k)f(M) = o(M^{-k}) if lim⁡M→∞Mkf(M)=0\lim_{M \to\infty} M^k f(M) = 0.

Preliminaries

In this section, we review the basics of DPM sampling.

Diffusion processes and parameterization

Forward and reverse processes. We consider a forward process (xt)0≤t≤T(\boldsymbol{x}_t)_{0 \le t \le T} in Rd\mathbb{R}^d whose marginal distribution at time t∈[0,T]t \in[0,T] satisfies

xt=dαtx0+σtz,(1)\boldsymbol{x}_t \mathrel{\mathop{=}^{\mathrm{d}}} \alpha_t \boldsymbol{x}_0 + \sigma_t \boldsymbol{z}, \tag*{(1)}

where x0∼q0\boldsymbol{x}_0 \sim q_0 is independent of the Gaussian noise z∼N(0,Id)\boldsymbol{z} \sim\mathcal{N}(\boldsymbol{0}, \boldsymbol{I}_d). We denote the distribution of xt\boldsymbol{x}_t, or its density, by qtq_t. The noise schedules αt\alpha_t and σt\sigma_t are positive for t∈(0,T)t \in(0,T) and are chosen so that the signal-to-noise (SNR) ratio αt/σt\alpha_t/\sigma_t decreases strictly with tt and the terminal distribution is approximately Gaussian: qT≈N(0,σ~2Id)q_T \approx\mathcal{N}(0,\widetilde{\sigma}^2\boldsymbol{I}_d).

These forward marginals can be obtained from the SDE

dxt=f(t)xt dt+g(t) dwt,t∈[0,T];x0∼q0,(2)\mathrm{d}\boldsymbol{x}_t = f(t)\boldsymbol{x}_t\,\mathrm{d}t + g(t)\,\mathrm{d}\boldsymbol{w}_t,\qquad t \in[0,T];\qquad\boldsymbol{x}_0 \sim q_0, \tag*{(2)}

where (wt)t≥0(\boldsymbol{w}_t)_{t \ge0} denotes a standard dd-dimensional Brownian motion. The coefficients are determined by the noise schedules as follows:

f(t)=dlog⁡αtdtandg2(t)=dσt2dt−2σt2dlog⁡αtdt.f(t) = \frac{\mathrm{d}\log\alpha_t}{\mathrm{d}t}\qquad\text{and}\qquad g^2(t) = \frac{\mathrm{d}\sigma_t^2}{\mathrm{d}t} - 2\sigma_t^2\frac{\mathrm{d}\log\alpha_t}{\mathrm{d}t}.

For further details on this formulation, we refer to Kingma et al. [31].

The time-reversal result of Anderson [1] gives the reverse SDE associated with (2), which is solved from TT to 00:

dxt=[f(t)xt−g2(t)st∗(xt)]dt+g(t) dw~t,t∈[0,T];xT∼qT.(3)\mathrm{d}\boldsymbol{x}_t = \left[f(t)\boldsymbol{x}_t - g^2(t)s_t^*(\boldsymbol{x}_t)\right]\mathrm{d}t + g(t)\,\mathrm{d}\widetilde{\boldsymbol{w}}_t,\qquad t \in[0,T];\qquad\boldsymbol{x}_T \sim q_T. \tag*{(3)}

where (w~t)t∈[0,T](\widetilde{\boldsymbol{w}}_t)_{t \in[0,T]} denotes standard Brownian motion running backward from TT to 00, and st∗(⋅):=∇log⁡qt(⋅)s_t^*(\cdot) := \nabla\log q_t(\cdot) is the score of the time-tt marginal. Sampling can alternatively be based on a deterministic process governed by the probability flow ODE:

x˙t=f(t)xt−12g2(t)st∗(xt),t∈[0,T];xT∼qT.(4)\dot{\boldsymbol{x}}_t = f(t)\boldsymbol{x}_t - \frac{1}{2}g^2(t)s_t^*(\boldsymbol{x}_t),\qquad t \in[0,T];\qquad\boldsymbol{x}_T \sim q_T. \tag*{(4)}

At every time tt, both reverse processes have marginal distribution qtq_t, matching that of the forward process; see Song et al. [62]. With the exact score st⋆(⋅)s_t^\star(\cdot) and initialization from qTq_T, both reverse processes recover q0q_0. In practice, sampling is approximate because the score is estimated and qTq_T is replaced by a Gaussian approximation.

Noise and data prediction. A noise prediction model ϵθ(x,t)\epsilon_\theta(x,t), parameterized by θ\theta, approximates the scaled score function −σtst⋆(x)-\sigma_t s_t^\star(x). The ideal noise predictor satisfies

ϵθ⋆(x,t)=−σtst⋆(x)=E[z∣αtx0+σtz=x].(5)\epsilon_{\theta^\star}(x,t)=-\sigma_t s_t^\star(x)=\mathbb{E}\left[z\mid\alpha_t x_0+\sigma_t z=x\right]. \tag*{(5)}

For any noise predictor, the corresponding data predictor is defined by

μθ(x,t):=x−σtϵθ(x,t)αt.(6)\mu_\theta(x,t):=\frac{x-\sigma_t\epsilon_\theta(x,t)}{\alpha_t}. \tag*{(6)}

In particular, the ideal data predictor satisfies

μθ⋆(x,t)=E[x0∣αtx0+σtz=x].(7)\mu_{\theta^\star}(x,t)=\mathbb{E}\left[x_0\mid\alpha_t x_0+\sigma_t z=x\right]. \tag*{(7)}

These conditional expectations motivate the names: ϵθ\epsilon_\theta and μθ\mu_\theta predict the noise zz and clean data x0x_0, respectively, from the noisy input.

Time reparameterization and sampling grid. We use the time parameterization

λ(t)=λt:=log⁡(αt/σt),(8)\lambda(t)=\lambda_t:=\log(\alpha_t/\sigma_t), \tag*{(8)}

which is strictly decreasing on (0,T)(0,T). Its inverse is denoted by t(λ)t(\lambda). For notational convenience, we use the abbreviations

ϵt:=ϵθ(xt,t)andμt:=μθ(xt,t)\epsilon_t:=\epsilon_\theta(x_t,t) \qquad\text{and}\qquad\mu_t:=\mu_\theta(x_t,t)

whenever it is clear from context.

For numerical sampling, we discretize a truncated interval [tM,t0]⊂(0,T)[t_M,t_0]\subset(0,T) using M+1M+1 time points:

References

  1. [1]Anderson, B. D. (1982). Reverse-time diffusion equation models. *Stochastic Processes and their Applications*, 12(3):313–326.DOI
  2. [2]Bao, F., Li, C., Zhu, J., and Zhang, B. (2022). Analytic-dpm: an analytic estimate of the optimal reverse variance in diffusion probabilistic models. *arXiv preprint arXiv:2201.06503*.arxiv.org/abs/2201.06503
  3. [3]Benton, J., De Bortoli, V., Doucet, A., and Deligiannidis, G. (2023). Nearly d-linear convergence bounds for diffusion models via stochastic localization. *arXiv preprint arXiv:2308.03686*.arxiv.org/abs/2308.03686
  4. [4]Blattmann, A., Rombach, R., Oktay, K., Müller, J., and Ommer, B. (2022). Retrieval-augmented diffusion models. *Advances in Neural Information Processing Systems*, 35:15309–15324.DOI
  5. [7]Chen, H., Lee, H., and Lu, J. (2023a). Improved analysis of score-based generative modeling: User-friendly bounds under minimal smoothness assumptions. In *International Conference on Machine Learning*, pages 4735–4763. PMLR.arxiv.org/abs/2211.01916
  6. [8]Chen, H., Ren, Y., Ying, L., and Rotskoff, G. M. (2024). Accelerating diffusion models with parallel sampling: Inference at sub-linear time complexity. *arXiv preprint arXiv:2405.15986*.arxiv.org/abs/2405.15986
  7. [9]Chen, S., Chewi, S., Lee, H., Li, Y., Lu, J., and Salim, A. (2023b). The probability flow ode is provably fast. *Advances in Neural Information Processing Systems*, 36:68552–68575.arxiv.org/abs/2305.11798
  8. [10]Chen, S., Chewi, S., Li, J., Li, Y., Salim, A., and Zhang, A. R. (2022). Sampling is as easy as learning the score: theory for diffusion models with minimal data assumptions. *arXiv preprint arXiv:2209.11215*.arxiv.org/abs/2209.11215
  9. [11]Chen, S., Cong, K., and Li, J. (2025). Optimal inference schedules for masked diffusion models. *arXiv preprint arXiv:2511.04647*.arxiv.org/abs/2511.04647
  10. [12]Chen, S. and Wang, L. (2026). Parallelism, critical windows, and separations among diffusion language models. arXiv preprint arXiv:2609.20539.arxiv.org/abs/2609.20539
  11. [13]Croitoru, F.-A., Hondru, V., Ionescu, R. T., and Shah, M. (2023). Diffusion models in vision: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence.DOI
  12. [14]Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., and Fei-Fei, L. (2009). Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee.
  13. [15]Dhariwal, P. and Nichol, A. (2021). Diffusion models beat gans on image synthesis. Advances in neural information processing systems, 34:8780–8794.
  14. [16]Dmitriev, D., Huang, Z., and Wei, Y. (2026a). Efficient sampling with discrete diffusion models: Sharp and adaptive guarantees. arXiv preprint arXiv:2602.15008.arxiv.org/abs/2602.15008
  15. [17]Dmitriev, D., Huang, Z., and Wei, Y. (2026b). Provably adaptive sampling with uniform and remasking discrete diffusion models. arXiv preprint arXiv:2608.23554.arxiv.org/abs/2608.23554
  16. [18]Gonzalez, M., Fernandez, N., Tran, T., Gherbi, E., Hajri, H., and Masmoudi, N. (2023). Seeds: Exponential sde solvers for fast high-quality sampling from diffusion models.arxiv.org/abs/2305.14267
  17. [19]Gupta, S., Cai, L., and Chen, S. (2024). Faster diffusion-based sampling with randomized midpoints: Sequential and parallel. arXiv preprint arXiv:2406.00924.arxiv.org/abs/2406.00924
  18. [20]Heusel, M., Ramsauer, H., Unterthiner, T., Nessler, B., and Hochreiter, S. (2017). Gans trained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information processing systems, 30.
  19. [22]Hoogeboom, E., Satorras, V. G., Vignac, C., and Welling, M. (2022). Equivariant diffusion for molecule generation in 3d. In International conference on machine learning, pages 8867–8887. PMLR.arxiv.org/abs/2203.17003
  20. [26]Jiao, Y., Li, N., Cai, C., and Li, G. (2026). Are first-order diffusion samplers really slower? a fast forward-value approach. In International Conference on Machine Learning, volume 2026, pages 1–9.arxiv.org/abs/2512.24927
  21. [27]Jiao, Y., Zhou, Y., and Li, G. (2025). Optimal convergence analysis of ddpm for general distributions. arXiv preprint arXiv:2510.27562.arxiv.org/abs/2510.27562
  22. [28]Jolicoeur-Martineau, A., Li, K., Piché-Taillefer, R., Kachman, T., and Mitliagkas, I. (2021). Gotta go fast when generating data with score-based models. arXiv preprint arXiv:2105.14080.
  23. [29]Karras, T., Aittala, M., Aila, T., and Laine, S. (2022). Elucidating the design space of diffusion-based generative models. In Proc. NeurIPS.arxiv.org/abs/2206.00364
  24. [30]Karras, T., Laine, S., and Aila, T. (2019). A style-based generator architecture for generative adversarial networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4401–4410.DOI
  25. [31]Kingma, D., Salimans, T., Poole, B., and Ho, J. (2021). Variational diffusion models. Advances in neural information processing systems, 34:21696–21707.arxiv.org/abs/2107.00630
  26. [32]Kong, Z., Ping, W., Huang, J., Zhao, K., and Catanzaro, B. (2021). DiffWave: A versatile diffusion model for audio synthesis. In International Conference on Learning Representations.arxiv.org/abs/2009.09761
  27. [33]Krizhevsky, A., Hinton, G., et al. (2009). Learning multiple layers of features from tiny images.
  28. [34]Lai, C.-H., Song, Y., Kim, D., Mitsufuji, Y., and Ermon, S. (2025). The principles of diffusion models. arXiv preprint arXiv:2510.21890.arxiv.org/abs/2510.21890
  29. [35]Lee, H., Lu, J., and Tan, Y. (2023). Convergence of score-based generative modeling for general data distributions. In International Conference on Algorithmic Learning Theory, pages 946–985. PMLR.arxiv.org/abs/2209.12381
  30. [49]Luhman, E. and Luhman, T. (2021). Knowledge distillation in iterative generative models for improved sampling speed. arXiv preprint arXiv:2101.02388.arxiv.org/abs/2101.02388
  31. [50]Meng, C., Rombach, R., Gao, R., Kingma, D., Ermon, S., Ho, J., and Salimans, T. (2023). On distillation of guided diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 14297–14306.arxiv.org/abs/2210.03142
  32. [51]Nichol, A. Q. and Dhariwal, P. (2021). Improved denoising diffusion probabilistic models. In International Conference on Machine Learning, pages 8162–8171.arxiv.org/abs/2102.09672
  33. [52]Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., and Chen, M. (2022). Hierarchical text-conditional image generation with clip latents. arXiv preprint arXiv:2204.06125, 1(2):3.arxiv.org/abs/2204.06125
  34. [53]Rombach, R., Blattmann, A., Lorenz, D., Esser, P., and Ommer, B. (2022). High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695.arxiv.org/abs/2112.10752
  35. [54]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 systems, 35:36479–36494.arxiv.org/abs/2205.11487
  36. [55]Sahoo, S., Arriola, M., Schiff, Y., Gokaslan, A., Marroquin, E., Chiu, J., Rush, A., and Kuleshov, V. (2024). Simple and effective masked diffusion language models. Advances in Neural Information Processing Systems, 37:130136–130184.arxiv.org/abs/2406.07524
  37. [56]Salimans, T. and Ho, J. (2022). Progressive distillation for fast sampling of diffusion models. arXiv preprint arXiv:2202.00512.arxiv.org/abs/2202.00512
  38. [57]Shih, A., Belkheile, S., Ermon, S., Sadigh, D., and Anari, N. (2023). Parallel sampling of diffusion models. Advances in Neural Information Processing Systems, 36:4263–4276.arxiv.org/abs/2305.16317
  39. [58]Sohl-Dickstein, J., Weiss, E., Maheswaranathan, N., and Ganguli, S. (2015). Deep unsupervised learning using nonequilibrium thermodynamics. In International conference on machine learning, pages 2256–2265. pmlr.
  40. [59]Song, J., Meng, C., and Ermon, S. (2020a). Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502.arxiv.org/abs/2010.02502
  41. [60]Song, Y., Dhariwal, P., Chen, M., and Sutskever, I. (2023). Consistency models. arXiv preprint arXiv:2303.01469.arxiv.org/abs/2303.01469
  42. [61]Song, Y. and Ermon, S. (2019). Generative modeling by estimating gradients of the data distribution. Advances in neural information processing systems, 32.arxiv.org/abs/1907.05600
  43. [62]Song, Y., Sohl-Dickstein, J., Kingma, D. P., Kumar, A., Ermon, S., and Poole, B. (2020b). Score-based generative modeling through stochastic differential equations. arXiv preprint arXiv:2011.13456.arxiv.org/abs/2011.13456
  44. [64]Xue, S., Yi, M., Luo, W., Zhang, S., Sun, J., Li, Z., and Ma, Z.-M. (2024). Sa-solver: Stochastic adams solver for fast sampling of diffusion models. Advances in Neural Information Processing Systems, 36.
  45. [65]Yang, L., Zhang, Z., Song, Y., Hong, S., Xu, R., Zhao, Y., Zhang, W., Cui, B., and Yang, M.-H. (2023). Diffusion models: A comprehensive survey of methods and applications. ACM Computing Surveys, 56(4):1–39.
  46. [68]Zhang, J., Liu, Z., Yan, L., Li, G., and Gu, Y. (2025). Improving diffusion-based inverse algorithms under few-step constraint via learnable linear extrapolation. arXiv preprint arXiv:2503.10103.
  47. [69]Zhang, J., Xia, M., Li, G., and Gu, Y. (2026). Distilling drifting transformers with representation autoencoders. arXiv preprint arXiv:2606.15553.arxiv.org/abs/2606.15553
  48. [70]Zhang, Q. and Chen, Y. (2022). Fast sampling of diffusion models with exponential integrator. arXiv preprint arXiv:2204.13902.arxiv.org/abs/2204.13902
  49. [71]Zhao, W., Bai, L., Rao, Y., Zhou, J., and Lu, J. (2023). Unipc: A unified predictor-corrector framework for fast sampling of diffusion models. Advances in Neural Information Processing Systems, 36:49842–49869.arxiv.org/abs/2302.04867
  50. [72]Zhao, Y. and Cai, C. (2026). Adaptation to intrinsic dependence in diffusion language models. arXiv preprint arXiv:2602.20126.arxiv.org/abs/2602.20126
  51. [73]Zheng, K., Lu, C., Chen, J., and Zhu, J. (2023). Dpm-solver-v3: Improved diffusion ode solver with empirical model statistics. Advances in Neural Information Processing Systems, 36:55502–55542.arxiv.org/abs/2310.13268

Paper details

vlnc:W_2p4mnbg9
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Not recorded for this paper

View statement
Looking for related papers…
Contents