Pith. sign in

REVIEW 3 major objections 6 minor 2 cited by

Designing a Conditional Prior Distribution for Flow-Based Generative Models

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper claims that placing a condition-specific Gaussian prior at each condition's average data point shortens the flow and cuts the sampling steps needed for high-quality conditional generation.

desk verdict A genuinely useful conditional prior for flow matching, with a solid empirical story and one acknowledged assumption that could use a robustness check. read the letter →

arxiv 2502.09611 v1 pith:35RS6SZM submitted 2025-02-13 cs.LG cs.CV

Noam Issachar , Mohammad Salama , Raanan Fattal , Sagie Benaim This is my paper · ORCID
classification cs.LGcs.CV
keywords flowmatchingconditionalgenerationpriordistributionGaussianmixturemodeltext-to-imageclass-conditionalsamplingefficiencylatentspace
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 claims that conditional flow-based generative models become more efficient when each condition gets its own starting distribution instead of a shared noise prior. For every class or text prompt, the method first computes an 'average' data point for that condition in a latent space, then trains the flow to transport samples from a Gaussian centered at that point to the conditional target distribution. Because prior samples are on average much closer to their targets, the learned trajectories are shorter, numerical integration errors stay smaller, and good samples can be produced with fewer function evaluations. On ImageNet-64 and MS-COCO the paper reports that this condition-aware prior improves FID, KID, and CLIP scores relative to conditional flow-matching and diffusion baselines, with the largest gains at low step counts.

What carries the argument

The load-bearing object is the conditional prior distribution (CPD), a Gaussian mixture model with one component per condition, placed in a latent space. Each component's mean is the condition's average data point, so the condition information enters before the flow starts; for discrete classes the mean and covariance are estimated from training samples, and for text prompts a learned mapper sends the prompt embedding to that mean. The matching training objective is Conditional Generation Joint Flow Matching (CGJFM), which generalizes flow matching to an arbitrary joint distribution $q(x_0, x_1, c)$ and uses the linear interpolant $\psi_t(x|x_1,c) = \sigma_t x + \mu_t$ with $\sigma_t = t\sigma_{\min}I + (1-t)\Sigma_c^{1/2}$ and $\mu_t = t x_1 + (1-t)\mu_c$. The network regresses the velocity field this interpolant induces, and the efficiency argument runs through the reduction in average path length $\Delta = |x_1 - x_0|$ that the condition-centered prior achieves.

What would settle it

Take a condition whose latent training examples split into two well-separated clusters and measure the average straight-line distance from the proposed Gaussian prior to the targets; if that distance is not smaller than the distance from a standard Gaussian prior, or if sampling at five to ten function evaluations shows no FID improvement, the central claim fails.

Watch

Extended reading notes

Core claim

Flow matching trains a neural network to reproduce a vector field that carries samples along interpolated paths between a source distribution and a target data distribution. The paper's discovery is that, in the conditional setting, the source should not be a fixed unimodal noise distribution shared by all conditions. It defines a conditional prior distribution $p_0 = \mathrm{GMM}(\mathcal{N}(\mu_i, \Sigma_i)_{i=1}^n, \pi)$, with $\mu_i = \mathbb{E}[x_1|c_i]$ and $\Sigma_i = \mathrm{cov}[x_1|c_i]$ in the latent space of a pretrained autoencoder; for text conditions, a trained mapper $P_\theta$ approximates $\mathbb{E}[x_1|c]$ from the condition embedding, and the covariance is a scalar hyperparameter. The model then learns the conditional vector field $\dot{x}_t = (\sigma_{\min}I - \Sigma_c^{1/2})x + x_1 - \mu_c$ that interpolates from that condition-specific Gaussian to the target. Because the prior is centered near the target mode, the average source-target distance drops, the velocity field has smaller Lipschitz constant, and the global truncation error bound permits larger step sizes. The paper claims this construction improves FID, KID, and CLIP scores at low numbers of function evaluations on ImageNet-64 and MS-COCO, and reaches convergence at roughly half the sampling steps of the strongest baselines.

Load-bearing premise

The argument assumes that, in the latent space, the training images belonging to a single condition form roughly one blob, so a single Gaussian centered at their average is a faithful starting distribution.

Editorial extensions

If this is right

  • At low sampling budgets the gain is largest: on ImageNet-64 the method reaches FID 13.62 at 15 function evaluations, while the flow baselines are still above 16 at the same budget.
  • Shorter average paths also speed up training, since the regression target is easier to fit when each source sample starts near its target; the paper shows a lower FID per epoch throughout training on MS-COCO.
  • For text prompts, an unseen condition still receives a sensible prior center by passing its embedding through the trained mapper, so the benefit carries over to new prompts without retraining.
  • The change is confined to the prior and the interpolant, leaving the network architecture untouched, so existing conditional flow-matching pipelines can adopt it as a drop-in replacement.

Reading between the lines

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

  • A natural stress test the paper does not run: for conditions whose latent examples form several separated modes, a multi-component prior per condition should preserve the short-path benefit, while the single-Gaussian-per-condition version will see its advantage shrink; the toy experiment with intersecting classes already hints at this limit.
  • The reported transport-cost numbers (570 and 510 versus 630 and 632 for the main baselines) could be used as a cheap, training-free diagnostic for whether a condition-aware prior will help on a new dataset.
  • Because the prior center is produced by a deterministic mapping from text embedding to latent space, improving the text-image embedding itself should directly improve the prior centers and further reduce the required number of sampling steps.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper proposes to replace the unimodal Gaussian prior used in conditional flow matching with a condition-specific Gaussian prior. For discrete conditions, the prior mean and covariance are estimated from per-class training data; for text conditions, a learned mapper regresses CLIP embeddings to the conditional mean in a pretrained latent space, and the covariance is fixed to an isotropic hyperparameter. The authors introduce a joint conditional flow matching objective (Eq. 16), define an affine interpolant between the condition-specific Gaussian and target samples (Eqs. 21-23), and evaluate the method on toy problems, ImageNet-64, and MS-COCO, reporting improved FID/KID/CLIP scores at low NFE as well as faster training convergence.

Significance. If the empirical results hold, the proposed idea is a simple and potentially useful drop-in modification for conditional flow matching, especially in compute-limited regimes. The paper contains a clean generalization of flow matching to arbitrary conditional joint distributions in Sec. 4.1, and it demonstrates consistent gains over CondOT and BatchOT at low NFE on two real datasets, with ablations isolating the role of sigma and the CLIP mapper. The contribution is incremental but practically relevant. The main concerns are formal precision of the flow construction and the unvalidated unimodality assumption for text conditions; both need to be addressed before the central claims can be considered fully supported.

major comments (3)
  1. [Sec. 4.1, Eq. (15)] The equality as written, ∫ p(x0|x1,c) dx1 dc = p(x0), is false under the definition p(x0|x1,c)=ρ(x0,x1,c)/q(x1,c). The integral over x1,c of p(x0|x1,c) weights by 1/q(x1,c), not by 1. The correct boundary condition is ∫ p(x0|x1,c) q(x1,c) dx1 dc = p(x0). Please correct the displayed equation and the surrounding derivation.
  2. [Secs. 4.2-4.3, Eqs. (21)-(24)] The role of the variable x in the interpolant is not consistent with the definition of x0 in Eqs. (13)-(16). In Sec. 4.2 x0 is a sample from p(x0|c)=N(μc,Σc), but Eq. (21) with σ0=Σc^{1/2} and μ0=μc gives ψ0(x0)=Σc^{1/2}x0+μc, which does not equal x0. If x in Eq. (21) is a standard Gaussian auxiliary variable, then x0 should be defined as σ0 x+μ0 and Eqs. (16) and (24) should be restated in terms of that variable. This is not a cosmetic issue: the objective in Eq. (16) and the ODE in Eq. (24) are only well-defined once the relationship between x0 and x is fixed.
  3. [Sec. 4.2 and Fig. 4] The claimed efficiency benefit depends on the assumption that each conditional distribution is approximately unimodal in the chosen latent space. The paper explicitly acknowledges this ('If the latter are close to being unimodal...'), but the assumption is not tested on real data. For MS-COCO, prompts describing multiple objects or attributes correspond to multimodal latent distributions, and the conditional mean may fall in low-density regions. The toy example in Fig. 4 shows only a small MMD gap (0.072 vs 0.084) in an intersecting case. All real-world results are reported as aggregate FID/KID/CLIP values without error bars or stratification by prompt complexity. Please provide per-condition or stratified results, or an explicit quantification of when the assumption fails.
minor comments (6)
  1. [Sec. 4.1, after Eq. (16)] Typo: 'we thus purpose' should be 'we thus propose'.
  2. [Sec. 4.2.1] Typo: 'represntation' should be 'representation'.
  3. [Fig. 5 caption] The caption notes that DDPM converges to a superior result given more steps, but the main text states that the method obtains superior results across all scores. Please clarify the NFE regime and qualify the claim accordingly.
  4. [Fig. 6 caption] The phrase 'NFE per training epoch' is unclear. Please specify how the NFE is computed (e.g., adaptive solver on a fixed validation batch at each epoch) and why it is reported on the training curve.
  5. [Table 1] Please specify the protocol used to compute the transport cost for the MS-COCO row, including how prior samples are drawn when the mapper is trained on the same data, and the number of pairs used.
  6. [Eq. (24)] The notation ψ0(x)=x0 is nonstandard; it should be ψ0(x0)=x0 or an explicit flow map from the initial condition.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the conditional prior is a data-driven construction and all reported gains are measured by external metrics.

full rationale

The paper's derivation chain is self-contained. The conditional prior p0 is defined in Eqs. 17-18 as a GMM whose per-class mean and covariance are the empirical statistics E[x1|ci] and cov[x1|ci]; for text, Eq. 20 trains a mapper Pθ to approximate E[x1|c] from CLIP embeddings. These are legitimate uses of training data to define a starting distribution, not fitted parameters that are then renamed as predictions. The flow-matching objective in Eq. 16 is the standard conditional flow-matching loss, regressing the velocity field to x1 - x0 for x0 drawn from the condition-specific prior; the target x1 is independent of the prior parameters, so no quantity is predicted from itself. The headline results (FID, KID, CLIP) are computed on held-out ImageNet-64 and MS-COCO validation sets and are external to the fitted prior and mapper. The lower transport cost reported in Table 1 is a mathematical consequence of centering the prior at the conditional mean, and the paper presents it as motivation rather than as an independent prediction; this is a design property, not circularity. The cited prior work on OT couplings and flow matching is external, and no load-bearing claim rests on a self-citation. The unimodality assumption in Sec. 4.2 is a substantive correctness risk for multimodal prompts, but it is an assumption about the data, not a circular derivation.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The central claim rests on two domain assumptions (unimodality of conditional modes and the path-length/error relation) plus one tunable hyperparameter (sigma). No new physical entities or ad hoc fitted constants beyond sigma are introduced; the per-class means/covariances are data statistics.

free parameters (1)
  • sigma (text-condition prior spread) = 0.7
    Standard deviation scalar for the isotropic Gaussian prior in the continuous-condition setting; ablated in Table 2 on MS-COCO, with FID ranging from 23.55 (sigma=0.2) to 7.55 (sigma=0.7). It is tuned on validation data and is not derived from the problem.
assumptions (3)
  • domain assumption Conditional data in the chosen latent space is approximately unimodal per condition
    Sec 4.2: 'we fit a dedicated Gaussian to data points with the same condition. If the latter are close to being unimodal, this approximation is expected to be tight.' The GMM prior quality depends on this; Fig. 4 shows degradation for intersecting multi-modal classes.
  • domain assumption Shorter average source-target distance reduces global truncation error of the learned vector field
    Sec 4.2 'Obtaining a Lower Global Truncation Error': the bound in Eq. 19 is used heuristically; the Lipschitz constant of the learned field v_theta is not shown to scale with path length, so this is an unproven premise for the claimed sampling efficiency gain.
  • standard math The interpolant in Eqs. 21-23 is a valid flow (affine, invertible for t in [0,1) with sigma_t > 0)
    Standard affine interpolation used in flow matching; requires sigma_min > 0 and symmetric positive semidefinite Sigma_c.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Designing a Conditional Prior Distribution for Flow-Based Generative Models." pith.science (2026). https://pith.science/paper/35RS6SZM

@misc{pith2026250209611,
  author       = {Pith},
  title        = {Pith review of: Designing a Conditional Prior Distribution for Flow-Based Generative Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/35RS6SZM}},
  note         = {Machine review of arXiv:2502.09611}
}
read the original abstract

Flow-based generative models have recently shown impressive performance for conditional generation tasks, such as text-to-image generation. However, current methods transform a general unimodal noise distribution to a specific mode of the target data distribution. As such, every point in the initial source distribution can be mapped to every point in the target distribution, resulting in long average paths. To this end, in this work, we tap into a non-utilized property of conditional flow-based models: the ability to design a non-trivial prior distribution. Given an input condition, such as a text prompt, we first map it to a point lying in data space, representing an ``average" data point with the minimal average distance to all data points of the same conditional mode (e.g., class). We then utilize the flow matching formulation to map samples from a parametric distribution centered around this point to the conditional target distribution. Experimentally, our method significantly improves training times and generation efficiency (FID, KID and CLIP alignment scores) compared to baselines, producing high quality samples using fewer sampling steps.

Figures

Figures reproduced from arXiv: 2502.09611 by the authors.

Figure 1
Figure 1. An illustration of our approach. The LHS illustrates the standard flow matching paradigm, where every sample in the source Gaussian distribution (shown as a circular point) can be mapped to every sample in the conditional target mode (shown as a cross point), where each class samples are shown in a different color. In contrast, our method, shown on the RHS, constructs a class-specific conditional distribution as a s… view at source ↗
Figure 2
Figure 2. Trajectory illustration. A toy example illustrating the trajectory from the source to the target distribution for our method and conditional flow matching using optimal transport (CondOT). NFE=2 NFE=3 NFE=4 NFE=6 NFE=8 NFE=10 NFE=15 NFE=400 GT Prior Samples GT CondOT Train Ours Test (a) (b) [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. (a) NFE convergence illustration. A toy example illustrating convergence to the target distribution at different NFEs, for our method, compared to CondOT. (b). Generalization illustration. A toy example illustrating the generalization capabilities. LHS: Source prior and target samples for training classes RHS: As for LHS, but for test classes. (a) (b) (c) [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Multi-modal classes. A toy example illustrating multi￾modal classes with intersections in the prior. Each color represents a class (class A or B), with samples as points and the prior dis￾tribution as contour lines. (a) shows a standard Gaussian prior (in black), while…
Figure 5
Figure 5. Figure 5: Numerical evaluation. (a) We compare our method to class conditional flow matching using optimal transport paths (CondOT) (Lipman et al., 2022), BatchOT (Pooladian et al., 2023), and DDPM (Ho et al., 2020a), on the ImageNet-64 dataset. We consider the FID score (LHS), …
Figure 6
Figure 6. Figure 6: Training time. For a text-conditional model trained on MS-COCO, we consider the NFE per training epoch. We compare our method with text conditional flow matching using optimal transport paths (CondOT) (Lipman et al., 2022), BatchOT (Pooladian et al., 2023), and DDPM (H…
Figure 7
Figure 7. Figure 7: A visualization of our results on MS-COCO. We show, for four different text prompts: (a). The sample corresponding to the text in the conditional source distribution, which is used as the center of Gaussian corresponding to the text prompt (LHS) (b). Six randomly gener…
Figure 8
Figure 8. Figure 8: A visualization of our results for different NFEs. We consider a model trained on MS-COCO, and two different vali￾dation prompts: Top: “There are yellow flowers inside a vase”, Bottom: “A bowl full of oranges”. tion, which is used as the center of Gaussian correspondin…
Figure 9
Figure 9. Figure 9: Visual comparison of randomly generated samples for prompts from the MS-COCO validation set using our method, in comparison to flow matching, for a model trained on MS-COCO. 12 [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Source-Lifted Flow Matching for Intervenable Multimodal Imitation

    cs.RO 2026-07 conditional novelty 6.5 of 10

    Orthogonal Source Lifting makes flow-matching imitation intervenable by selecting only the source endpoint, changing future routes in 91.1% of matched-prefix tests while keeping free-deployment performance strong.

  2. CaloTrilogy: Toward a Breakthrough in One-Step, End-to-End, Physics-Guided Shower Generation for Modern Calorimeters

    hep-ex 2026-06 unverdicted novelty 6.0 of 10

    Presents CaloTrilogy, a unified one-step generative model for high-granularity calorimeter showers that combines velocity field integration, learned priors, and physics losses to match SOTA quality.

Reference graph

Works this paper leans on

46 extracted references · 14 canonical work pages · cited by 2 Pith papers

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  3. [3]

    Albergo, M. S. and Vanden-Eijnden, E. Building normalizing flows with stochastic interpolants. arXiv preprint arXiv:2209.15571, 2022

    arXiv 2022

  4. [4]

    J., Bengio, Y., Tong, A., and Neklyudov, K

    Atanackovic, L., Zhang, X., Amos, B., Blanchette, M., Lee, L. J., Bengio, Y., Tong, A., and Neklyudov, K. Meta flow matching: Integrating vector fields on the wasserstein manifold, 2024. URL https://arxiv.org/abs/2408.14608

    arXiv 2024

  5. [5]

    J., Arbel, M., and Gretton, A

    Bińkowski, M., Sutherland, D. J., Arbel, M., and Gretton, A. Demystifying mmd gans, 2021. URL https://arxiv.org/abs/1801.01401

    arXiv 2021

  6. [6]

    A., Gardner, P., Rogers, T

    Bull, L. A., Gardner, P., Rogers, T. J., Cross, E. J., Dervilis, N., and Worden, K. Probabilistic inference for structural health monitoring: New modes of learning from data. ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part A: Civil Engineering, 7 0 (1), March 2021. ISSN 2376-7642. doi:10.1061/ajrua6.0001106. URL http://dx.doi.org/10...

    work page doi:10.1061/ajrua6.0001106 2021

  7. [7]

    Chen, R. T. Q. torchdiffeq, 2018. URL https://github.com/rtqichen/torchdiffeq

    work page 2018

  8. [8]

    Chen, R. T. Q., Rubanova, Y., Bettencourt, J., and Duvenaud, D. Neural ordinary differential equations, 2019

    work page 2019

Show all 46 references
  1. [9]

    Flow matching in latent space, 2023

    Dao, Q., Phung, H., Nguyen, B., and Tran, A. Flow matching in latent space, 2023. URL https://arxiv.org/abs/2307.08698

    2023 arXiv

  2. [10]

    Imagenet: A large-scale hierarchical image database

    Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., and Fei-Fei, L. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pp.\ 248--255. Ieee, 2009

    2009

  3. [11]

    and Nichol, A

    Dhariwal, P. and Nichol, A. Diffusion models beat gans on image synthesis. Advances in neural information processing systems, 34: 0 8780--8794, 2021

    2021

  4. [12]

    A., Garnelo, M., Lee, M

    Dilokthanakul, N., Mediano, P. A., Garnelo, M., Lee, M. C., Salimbeni, H., Arulkumaran, K., and Shanahan, M. Deep unsupervised clustering with gaussian mixture variational autoencoders. arXiv preprint arXiv:1611.02648, 2016

    2016 arXiv

  5. [13]

    Scaling rectified flow transformers for high-resolution image synthesis

    Esser, P., Kulal, S., Blattmann, A., Entezari, R., M \"u ller, J., Saini, H., Levi, Y., Lorenz, D., Sauer, A., Boesel, F., et al. Scaling rectified flow transformers for high-resolution image synthesis. In Forty-first International Conference on Machine Learning, 2024

    2024

  6. [14]

    datasauRus: Datasets from the Datasaurus Dozen, 2025

    Gillespie, C., Locke, S., Davies, R., and D'Agostino McGowan , L. datasauRus: Datasets from the Datasaurus Dozen, 2025. URL https://github.com/jumpingrivers/datasauRus. R package version 0.1.9, https://jumpingrivers.github.io/datasauRus/

    2025

  7. [15]

    L., and Choi, Y

    Hessel, J., Holtzman, A., Forbes, M., Bras, R. L., and Choi, Y. Clipscore: A reference-free evaluation metric for image captioning, 2022. URL https://arxiv.org/abs/2104.08718

    2022 arXiv

  8. [16]

    Gans trained by a two time-scale update rule converge to a local nash equilibrium, 2018

    Heusel, M., Ramsauer, H., Unterthiner, T., Nessler, B., and Hochreiter, S. Gans trained by a two time-scale update rule converge to a local nash equilibrium, 2018. URL https://arxiv.org/abs/1706.08500

    2018 arXiv

  9. [17]

    Hinton, G. E. and Salakhutdinov, R. R. Reducing the dimensionality of data with neural networks. science, 313 0 (5786): 0 504--507, 2006

    2006

  10. [18]

    and Salimans, T

    Ho, J. and Salimans, T. Classifier-free diffusion guidance. arXiv preprint arXiv:2207.12598, 2022

    2022 arXiv

  11. [19]

    Denoising diffusion probabilistic models

    Ho, J., Jain, A., and Abbeel, P. Denoising diffusion probabilistic models. In Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M., and Lin, H. (eds.), Advances in Neural Information Processing Systems, volume 33, pp.\ 6840--6851. Curran Associates, Inc., 2020 a . URL https://...

    2020

  12. [20]

    Denoising diffusion probabilistic models

    Ho, J., Jain, A., and Abbeel, P. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33: 0 6840--6851, 2020 b

    2020

  13. [21]

    Extended flow matching: a method of conditional generation with generalized continuity equation, 2024

    Isobe, N., Koyama, M., Zhang, J., Hayashi, K., and Fukumizu, K. Extended flow matching: a method of conditional generation with generalized continuity equation, 2024. URL https://arxiv.org/abs/2402.18839

    2024 arXiv

  14. [22]

    Izmailov, P., Kirichenko, P., Finzi, M., and Wilson, A. G. Semi-supervised learning with normalizing flows. In International conference on machine learning, pp.\ 4615--4630. PMLR, 2020

    2020

  15. [23]

    Variational deep embedding: An unsupervised and generative approach to clustering, 2017

    Jiang, Z., Zheng, Y., Tan, H., Tang, B., and Zhou, H. Variational deep embedding: An unsupervised and generative approach to clustering, 2017. URL https://arxiv.org/abs/1611.05148

    2017 arXiv

  16. [24]

    Kingma, D. P. and Ba, J. Adam: A method for stochastic optimization, 2017. URL https://arxiv.org/abs/1412.6980

    2017 arXiv

  17. [25]

    J., and Brubaker, M

    Kobyzev, I., Prince, S. J., and Brubaker, M. A. Normalizing flows: An introduction and review of current methods. IEEE transactions on pattern analysis and machine intelligence, 43 0 (11): 0 3964--3979, 2020

    2020

  18. [26]

    Priorgrad: Improving conditional denoising diffusion models with data-dependent adaptive prior

    Lee, S.-g., Kim, H., Shin, C., Tan, X., Liu, C., Meng, Q., Qin, T., Chen, W., Yoon, S., and Liu, T.-Y. Priorgrad: Improving conditional denoising diffusion models with data-dependent adaptive prior. arXiv preprint arXiv:2106.06406, 2021

    2021 arXiv

  19. [27]

    Lin, T.-Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Doll \'a r, P., and Zitnick, C. L. Microsoft coco: Common objects in context. European conference on computer vision, pp.\ 740--755, 2014

    2014

  20. [29]

    H., Le, M., Vyas, A., Shi, B., Tjandra, A., and Hsu, W.-N

    Liu, A. H., Le, M., Vyas, A., Shi, B., Tjandra, A., and Hsu, W.-N. Generative pre-training for speech with flow matching. In The Twelfth International Conference on Learning Representations, 2024. URL https://openreview.net/forum?id=KpoQSgxbKH

    2024

  21. [30]

    Flow straight and fast: Learning to generate and transfer data with rectified flow

    Liu, X., Gong, C., and Liu, Q. Flow straight and fast: Learning to generate and transfer data with rectified flow. arXiv preprint arXiv:2209.03003, 2022

    2022 arXiv

  22. [31]

    McCann, R. J. A convexity principle for interacting gases. Advances in Mathematics, 128: 0 153--179, 1997. URL https://api.semanticscholar.org/CorpusID:123005604

    1997

  23. [32]

    J., Mohamed, S., and Lakshminarayanan, B

    Papamakarios, G., Nalisnick, E., Rezende, D. J., Mohamed, S., and Lakshminarayanan, B. Normalizing flows for probabilistic modeling and inference. Journal of Machine Learning Research, 22 0 (57): 0 1--64, 2021

    2021

  24. [33]

    Pooladian, A.-A., Ben-Hamu, H., Domingo-Enrich, C., Amos, B., Lipman, Y., and Chen, R. T. Multisample flow matching: Straightening flows with minibatch couplings. arXiv preprint arXiv:2304.14772, 2023

    2023 arXiv

  25. [34]

    W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al

    Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al. Learning transferable visual models from natural language supervision. In International conference on machine learning, pp.\ 8748--8763. PMLR, 2021

    2021

  26. [35]

    High-resolution image synthesis with latent diffusion models

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., and Ommer, B. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 10684--10695, 2022

    2022

  27. [36]

    U-net: Convolutional networks for biomedical image segmentation, 2015

    Ronneberger, O., Fischer, P., and Brox, T. U-net: Convolutional networks for biomedical image segmentation, 2015. URL https://arxiv.org/abs/1505.04597

    2015 arXiv

  28. [37]

    Deep unsupervised learning using nonequilibrium thermodynamics

    Sohl-Dickstein, J., Weiss, E., Maheswaranathan, N., and Ganguli, S. Deep unsupervised learning using nonequilibrium thermodynamics. In International conference on machine learning, pp.\ 2256--2265. PMLR, 2015

    2015

  29. [38]

    P., Kumar, A., Ermon, S., and Poole, B

    Song, Y., Sohl-Dickstein, J., Kingma, D. P., Kumar, A., Ermon, S., and Poole, B. Score-based generative modeling through stochastic differential equations. arXiv preprint arXiv:2011.13456, 2020

    2011 arXiv

  30. [39]

    and Mayers, D

    S \"u li, E. and Mayers, D. F. An introduction to numerical analysis. Cambridge university press, 2003

    2003

  31. [40]

    Improving and generalizing flow-based generative models with minibatch optimal transport

    Tong, A., Malkin, N., Huguet, G., Zhang, Y., Rector-Brooks, J., Fatras, K., Wolf, G., and Bengio, Y. Improving and generalizing flow-based generative models with minibatch optimal transport. arXiv preprint arXiv:2302.00482, 2023

    2023 arXiv

  32. [41]

    Neural discrete representation learning, 2018

    van den Oord, A., Vinyals, O., and Kavukcuoglu, K. Neural discrete representation learning, 2018. URL https://arxiv.org/abs/1711.00937

    2018 arXiv

  33. [42]

    N., Kaiser, L., and Polosukhin, I

    Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., and Polosukhin, I. Attention is all you need, 2023. URL https://arxiv.org/abs/1706.03762

    2023 arXiv

  34. [43]

    Diffusers: State-of-the-art diffusion models

    von Platen, P., Patil, S., Lozhkov, A., Cuenca, P., Lambert, N., Rasul, K., Davaadorj, M., Nair, D., Paul, S., Berman, W., Xu, Y., Liu, S., and Wolf, T. Diffusers: State-of-the-art diffusion models. https://github.com/huggingface/diffusers, 2022

    2022

  35. [44]

    Zheng, Q., Le, M., Shaul, N., Lipman, Y., Grover, A., and Chen, R. T. Q. Guided flows for generative modeling and decision making, 2023. URL https://arxiv.org/abs/2311.13443

    2023 arXiv

  36. [45]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should ...

  37. [46]

    \@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@firs...

  38. [47]

    Johnson, Justin and Alahi, Alexandre and Fei-Fei, Li

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibset...

    2025 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.