Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

DDPS: Discrete Diffusion Posterior Sampling for Paths in Layered Graphs

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 New representation makes every diffusion sample a valid graph path.

desk verdict PALM is a genuinely useful path encoding that guarantees validity; the guidance is a plausible heuristic with an unsupported posterior-sampling claim. read the letter →

arxiv 2504.20754 v1 pith:KRMAT7AX submitted 2025-04-29 cs.LG

Hao Luan , See-Kiong Ng , Chun Kai Ling This is my paper · ORCID
classification cs.LG
keywords discretediffusionmodelslayeredgraphspathgenerationpaddedadjacency-listmatrix(PALM)classifierguidanceposteriorsamplinggraphvaliditybyconstruction
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper sets out to show that discrete diffusion models can generate paths in layered graphs with a hard validity guarantee: every sample is a path, not merely a subgraph that resembles one. Its device is the padded adjacency-list matrix (PALM), which stores a path as one one-hot vector per vertex, each selecting a single outgoing edge. Because any PALM decodes deterministically into exactly one path and every path has at least one PALM, validity is built into the representation rather than left to the model. The paper also claims that reward guidance can steer samples toward preferred edges without retraining, by approximating the discrete posterior log-likelihood ratio with the gradient of expected reward with respect to PALM logits. Experiments report 100% valid samples on the tested graphs and reward improvement that grows with the guidance scale.

What carries the argument

The central object is the padded adjacency-list matrix (PALM): for a layered graph with vertex set $V$, a path is encoded as a collection $\{x_v\}_{v \in V}$, where $x_v \in \mathbb{R}^{D_v}$ is a one-hot vector picking one outgoing edge of vertex $v$ (with padding when $D_v = 0$). Following the selected edges from the singleton first layer reconstructs the path, which makes the decoding deterministic. The guiding machinery is Eq. (6)-(7): the log-likelihood ratio $\log(p(y|x_t,x_{t-1})/p(y|x_t))$ is approximated by $\nabla_z R(z)$, the gradient of the expected total reward $R$ with respect to the PALM logits $z = \tilde{p}_\theta(\tilde{x}_0|x_t)$, and this gradient is added to the denoising logits with scale $\lambda$ to produce the posterior sample in Algorithm 2.

What would settle it

On a layered graph small enough to enumerate every path, compute the exact ratio $\log(p(y|x_t,x_{t-1})/p(y|x_t))$ by brute force for a preferred-edge reward, then run Algorithm 2 with the same reward. If the empirical distribution of decoded paths does not move toward the exact posterior as $\lambda$ increases, or if the correlation between $\nabla_z R(z)$ and the true likelihood ratio is weak or negative, the Eq. (6) approximation is falsified while the PALM validity claim remains untouched.

Watch

Extended reading notes

Core claim

The central claim is stated in Section 3: the PALM-to-path mapping is many-to-one and onto, so (i) one PALM instance represents exactly one path, and (ii) every path is represented by at least one PALM. Consequently, a discrete diffusion model trained to output PALM one-hot vectors can only emit samples that decode to valid layered-graph paths, and the validity guarantee holds regardless of guidance strength. In Section 5 the paper further claims that its DDPS guidance, built on the gradient approximation in Eq. (6), increases the reward of generated paths while maintaining reasonable adherence to the learned distribution, with validity rates that stay at 100% where an adjacency-matrix baseline (SwinGNN) degrades to zero under strong guidance.

Load-bearing premise

The load-bearing assumption is that the gradient of expected reward with respect to PALM logits is a faithful proxy for the log-likelihood ratio in Eq. (6); the paper offers no error bound, derivation, or directional check, so if that approximation is poor the guided samples need not follow any well-defined conditional distribution, even though the PALM validity guarantee still holds.

Editorial extensions

If this is right

  • Every sample drawn from the PALM-based discrete diffusion model decodes to a valid path with no post-processing.
  • Increasing the guidance scale $\lambda$ raises average reward and eventually plateaus near the maximum achievable reward under the tested reward configurations.
  • The valid rate remains 100% under guidance, whereas the SwinGNN adjacency-matrix baseline drops to 0% as $\lambda$ grows.
  • Distributional distance between guided samples and the target posterior first falls, then rises after a sweet spot, so guidance scale can be tuned to balance reward against fidelity.
  • PALM-based training uses only the loss on vertices that lie on the path, which is the natural objective when the data distribution is over paths.

Reading between the lines

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

  • Because the validity guarantee is representational rather than statistical, the same one-choice-per-item encoding should transfer to other constrained combinatorial objects (e.g., spanning trees or matchings) whenever a deterministic decoder exists; the paper does not test this.
  • The Eq. (6) approximation is the part most likely to fail generally: on small graphs one can compute the exact likelihood ratio and compare it with $\nabla_z R(z)$, which would isolate whether the reported reward gains come from correct posterior approximation or from a bias that happens to help.
  • The sweet-spot location in $\lambda$ varies with graph and reward sparsity, suggesting guidance scale should be tuned per task; extending DDPS to multi-objective or learned rewards would test whether the gradient signal remains aligned with the true likelihood ratio.
  • The paper's guidance is classifier-style and requires no retraining, so it could in principle be composed with classifier-free or RL-based discrete guidance, but the paper does not explore such combinations.
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 addresses constrained generation of paths in layered graphs with discrete diffusion. It introduces the padded adjacency-list matrix (PALM): one one-hot vector per vertex over its outgoing edges, so any assignment decodes to a unique valid path and every path has many PALM encodings. Training follows D3PM with uniform transition matrices, a cosine schedule, and a loss restricted to vertices on the path. For conditional generation, the paper proposes DDPS, which at each reverse step predicts clean logits z, computes the gradient of expected reward R(z), and adds a scaled gradient to the predicted x_{t-1} distribution. Experiments on three graphs compare validity with EDP-GNN and SwinGNN, show reward improvement as the guidance scale grows, and report distributional-distance curves as a function of lambda.

Significance. The PALM representation is a genuinely simple and correct way to enforce path validity: the many-to-one/onto argument in Section 3 is sound, and the 100% valid rate in Table 1 follows by construction. The reward curves in Figure 4 give preliminary evidence that the gradient-based guidance is a useful reward-shaping heuristic, and the paper is honest about the preliminary nature of the results. However, the paper's stronger claim, that DDPS is a form of posterior sampling built on the Bayes identity (5), is not established by the current derivation or experiments. The distinctive contribution beyond PALM is the guidance mechanism, and that mechanism currently rests on an unproved approximation and an underspecified algorithm.

major comments (3)
  1. [4.2, Eq. (6)] Equation (6) is the load-bearing step: it replaces log(p(y|x_t,x_{t-1})/p(y|x_t)) with nabla_z R(z), where z = tilde p_theta(x0|x_t). This cannot be a pointwise likelihood ratio because the left-hand side is a function of both x_t and x_{t-1}, while the right-hand side depends only on x_t (through z) and the reward. No derivation, error bound, or numerical check is provided, so Algorithm 2 is currently a reward-shaping heuristic rather than a posterior sampler. I recommend a small calibration study (e.g., on the Toy graph, compare DDPS samples against the exact reward-filtered posterior, or estimate both sides of Eq. (6) by Monte Carlo) and, if the approximation remains heuristic, stating that explicitly in the paper.
  2. [Algorithm 2, lines 7-9; Eq. (7)] Equation (7) states that guidance is added in log-probability space, but Algorithm 2 adds the gradient to a probability vector. In detail, line 7 defines tilde x_{t-1} = (prod_{tau=1}^t Q_tau) z, which as written is a distribution over x_t, not x_{t-1}; the product should run to t-1 if q(x_{t-1}|x_0) is intended. Line 8 then forms tilde x'_{t-1} = lambda g_t + tilde x_{t-1}, which can have negative or non-normalized entries before the categorical sampling in line 9; no clipping or renormalization is specified. The algorithm needs a consistent formulation (e.g., adding the gradient to logits and renormalizing) and a correction of the product index.
  3. [5.3, Figure 5] The target distribution in Section 5.3 is defined operationally as the lambda = 0 distribution filtered to maximum-reward samples, not as a posterior p(x0|y) under a stated likelihood. Since any guidance mechanism that shifts probability toward high-reward paths will produce distance curves that first decrease and then increase as lambda grows, the non-monotone 'sweet spot' in Figure 5 does not validate Eq. (6) or support the posterior-sampling interpretation. A direct calibration test on a small graph would be needed; otherwise the paper should present DDPS as empirical reward shaping.
minor comments (5)
  1. [2, footnote 2; Figure 3] Footnote 2 acknowledges that guidance scale technically cannot be compared across methods, but Figure 3 plots both methods against the same lambda and uses the comparison to conclude that SwinGNN is brittle; please remove the cross-method comparison or use a method-agnostic measure of guidance strength.
  2. [5.1, Table 1] Table 1 reports only validity, which is 100% by construction for PALM; reporting a distributional fidelity metric (e.g., distance to the validation path distribution) would make the comparison with EDP-GNN and SwinGNN more meaningful.
  3. [3, Definition 3] Definition 3 uses V for both the vertex set and its cardinality; use |V| for the cardinality to avoid ambiguity.
  4. [3, Eq. (4)] The loss masking in Eq. (4) ('only include losses incurred at vertices on the path') is not specified; please state how the mask is formed when random one-hot rows are assigned to off-path vertices.
  5. [5, experimental setup] The paper would benefit from a reproducibility statement: hyperparameters such as learning rate, batch size, UNet width/depth, and the exact sampling schedule are not given, and no code is released.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: PALM validity is by construction, DDPS guidance is an explicit external-reward heuristic, and the minor self-citations are not load-bearing.

full rationale

I walked the derivation chain from the PALM representation through D3PM training to DDPS guidance. The structural guarantee is deliberate design, not a hidden identity with the input: Definition 3 defines a PALM as 'a stack of one-hot vectors recording a single outgoing edge of each vertex,' and the PALM-to-path conversion 'follow the edge e = (v_l, v_{l+1}) represented by x_l in the PALM and transit to v_{l+1}' then repeats until v_L. Because every PALM is an assignment of one outgoing edge to every vertex, any PALM decodes to a valid path by construction; no fitted parameter is later renamed as a prediction. The guidance signal in Eq. (6), log(p(y|x_t,x_{t-1})/p(y|x_t)) ≈ ∇_z R(z), is explicitly introduced as an approximation, and while the RHS has no x_{t-1} dependence and is not a pointwise likelihood ratio, that is a correctness/validity weakness rather than circularity: the reward R is externally specified, the guidance scale is swept rather than fitted, and the reward improvements in Fig. 4 are empirical measurements of a heuristic, not claims derived from the approximation. The self-citations are minor: Definition 1 is 'adapted from' Cerny et al. with an overlapping author, but it supplies a definition and benchmark construction, not an unverified uniqueness theorem, and the paper's validity and guidance claims do not rest on that citation. Section 5.3's 'target distribution' is the model's own lambda=0 output conditioned on maximum reward; this is self-referential as an evaluation baseline, but the paper uses it to measure a tradeoff, not to derive the method, so it does not make the central claim equivalent to its input. Overall, no load-bearing step reduces to its own inputs.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central claim rests on the layered-graph definition, the correctness of the D3PM backbone, and the validity of the gradient approximation in Eq. (6). No new physical or latent entities are introduced; PALM is a deterministic data representation, not a postulated object in need of independent evidence.

free parameters (3)
  • guidance scale lambda = not fitted; swept across scales, sweet spot near 100 on Toy
    Controls the weight of the reward gradient in Eq. (7) and Algorithm 2. The paper's central guidance tradeoff depends on this hand-chosen hyperparameter, and no principled selection rule is given.
  • diffusion timesteps T = 256
    Number of forward and reverse diffusion steps, chosen by hand and fixed for all experiments.
  • loss weighting gamma = not specified in the paper
    Weights the variational bound term against the denoising term in Eq. (4). The value is not reported, and it affects what the model optimizes.
assumptions (6)
  • domain assumption Layered graph conditions of Definition 1: vertex partition into L layers, edges only between adjacent layers, first layer is a singleton, and every zero-outdegree vertex before the last layer has zero indegree.
    The PALM-to-path decoding uses the unique first vertex and the adjacency-only edge structure. The paper adapts this definition from Cerny et al. (2024), which is cited as background.
  • domain assumption The D3PM discrete diffusion framework with uniform transition matrices correctly learns categorical distributions over PALM entries.
    The method inherits D3PM training and inference from Austin et al. (2021) without a new correctness proof, and the per-vertex transition matrices in Appendix B.2 are constructed to match variable vector lengths.
  • standard math Bayes' rule applied in Eq. (5) correctly expresses the discrete posterior log-probability.
    The posterior decomposition into a likelihood ratio and a prior term is a standard application of Bayes' rule and is used as the starting point for guidance.
  • ad hoc to paper The log-likelihood ratio in Eq. (5) can be approximated by the gradient of expected reward with respect to predicted logits, as written in Eq. (6).
    This is the central heuristic of DDPS. No derivation, error bound, or regime of validity is supplied, and the paper itself labels it an approximation.
  • domain assumption Algorithm 1 computes the expected total reward exactly from the one-step predicted PALM logits.
    The algorithm treats the predicted logits as defining a distribution over outgoing edges and assumes path reward decomposes as a sum of per-edge rewards, which is true for the binary edge-reward setup but is not formally derived in the paper.
  • ad hoc to paper Training loss only at vertices on the path is sufficient to learn the path distribution, despite random one-hot assignments for off-path vertices.
    Off-path entries are irrelevant for decoding, but the model never learns a principled distribution for them. This is acceptable only because the decoding procedure ignores off-path choices.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DDPS: Discrete Diffusion Posterior Sampling for Paths in Layered Graphs." pith.science (2026). https://pith.science/paper/KRMAT7AX

@misc{pith2026250420754,
  author       = {Pith},
  title        = {Pith review of: DDPS: Discrete Diffusion Posterior Sampling for Paths in Layered Graphs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KRMAT7AX}},
  note         = {Machine review of arXiv:2504.20754}
}
read the original abstract

Diffusion models form an important class of generative models today, accounting for much of the state of the art in cutting edge AI research. While numerous extensions beyond image and video generation exist, few of such approaches address the issue of explicit constraints in the samples generated. In this paper, we study the problem of generating paths in a layered graph (a variant of a directed acyclic graph) using discrete diffusion models, while guaranteeing that our generated samples are indeed paths. Our approach utilizes a simple yet effective representation for paths which we call the padded adjacency-list matrix (PALM). In addition, we show how to effectively perform classifier guidance, which helps steer the sampled paths to specific preferred edges without any retraining of the diffusion model. Our preliminary results show that empirically, our method outperforms alternatives which do not explicitly account for path constraints.

Figures

Figures reproduced from arXiv: 2504.20754 by the authors.

Figure 2
Figure 2. The PALM representation (transposed) for the path highlighted in figure 1. Columns cor￾respond to vertices: blue indicates on-path vertices (A, C, G, H) and their specific one-hot vectors; gray indicates off-path vertices. Each gray column vec￾tor is an arbitrary one-hot selection if the vertex has outgoing edges. Faded entries are padding zeros. that our proposed method is superior in generating paths compared to n… view at source ↗
Figure 3
Figure 3. Valid rates for baseline and our method under different guidance strengths. • Heights and Ridge are layered graphs converted from real-world city maps using the method of (Cern ˇ y et al. ´ , 2024). Heights is a middle-scale graph with 12 layers and 423 vertices while Ridge is larger, with 13 layers and 4293 total vertices. These graphs are relatively sparse, with out-degree of each vertex mostly being less than 6. … view at source ↗
Figure 4
Figure 4. Average reward values under various guidance scales for di [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: Metrics in path distributions for the small-size graph [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Additional experiments with guidance on Toy, Heights and Ridge. The first row contains the rewards obtained. The next 4 rows are based on FLGD and IS-L metrics. 15 [PITH_FULL_IMAGE:figures/full_fig_p015_6.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. Solving Inverse Problems via Diffusion-Based Priors: An Approximation-Free Ensemble Sampling Approach

    cs.LG 2025-06 conditional novelty 5.0 of 10

    A weighted-particle sampler evolves the posterior through the diffusion model's reverse dynamics, with theoretical error bounds and improved image reconstructions.

  2. Discrete Diffusion Models: A Unified Framework from Tokenization to Generation

    cs.LG 2026-07 unverdicted novelty 4.0 of 10

    Discrete diffusion models are re-framed as instances of a tokenization-centric, four-component design space (corruption, denoiser, objective, sampler) in a broad survey with no new experimental or theoretical results.

Reference graph

Works this paper leans on

51 extracted references · 26 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]

    Claude 3.5 Sonnet

    Anthropic. Claude 3.5 Sonnet . Large Language Model, 2025. URL https://claude.ai

    work page 2025

  3. [3]

    Structured denoising diffusion models in discrete state-spaces

    Jacob Austin, Daniel D Johnson, Jonathan Ho, Daniel Tarlow, and Rianne Van Den Berg. Structured denoising diffusion models in discrete state-spaces. Advances in Neural Information Processing Systems, 34: 0 17981--17993, 2021

    2021

  4. [4]

    D irichlet diffusion score model for biological sequence generation

    Pavel Avdeyev, Chenlai Shi, Yuhao Tan, Kseniia Dudnyk, and Jian Zhou. D irichlet diffusion score model for biological sequence generation. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, pp.\ 1276--1301. PMLR, 23--29 Jul 2023. URL https://proceedings.mlr.press/v202/avdeyev23a.html

    work page 2023

  5. [5]

    A continuous time framework for discrete denoising models

    Andrew Campbell, Joe Benton, Valentin De Bortoli, Thomas Rainforth, George Deligiannidis, and Arnaud Doucet. A continuous time framework for discrete denoising models. Advances in Neural Information Processing Systems, 35: 0 28266--28279, 2022

    2022

  6. [6]

    Generative flows on discrete state-spaces: Enabling multimodal flows with applications to protein co-design

    Andrew Campbell, Jason Yim, Regina Barzilay, Tom Rainforth, and Tommi Jaakkola. Generative flows on discrete state-spaces: Enabling multimodal flows with applications to protein co-design. arXiv preprint arXiv:2402.04997, 2024

    arXiv 2024

  7. [7]

    Contested logistics: A game-theoretic approach

    Jakub C ern \`y , Chun Kai Ling, Darshan Chakrabarti, Jingwen Zhang, Gabriele Farina, Christian Kroer, and Garud Iyengar. Contested logistics: A game-theoretic approach. In International Conference on Decision and Game Theory for Security, pp.\ 124--146. Springer, 2024

    work page 2024

  8. [8]

    Diffusion policy: Visuomotor policy learning via action diffusion

    Cheng Chi, Zhenjia Xu, Siyuan Feng, Eric Cousineau, Yilun Du, Benjamin Burchfiel, Russ Tedrake, and Shuran Song. Diffusion policy: Visuomotor policy learning via action diffusion. The International Journal of Robotics Research, pp.\ 02783649241273668, 2023

    2023

Show all 51 references
  1. [9]

    Diffusion models beat gans on image synthesis

    Prafulla Dhariwal and Alexander Nichol. Diffusion models beat gans on image synthesis. In M. Ranzato, A. Beygelzimer, Y. Dauphin, P.S. Liang, and J. Wortman Vaughan (eds.), Advances in Neural Information Processing Systems, volume 34, pp.\ 8780--8794, 2021. URL https://proceed...

    2021

  2. [10]

    Spearman's footrule as a measure of disarray

    Persi Diaconis and Ronald L Graham. Spearman's footrule as a measure of disarray. Journal of the Royal Statistical Society Series B: Statistical Methodology, 39 0 (2): 0 262--268, 1977

    1977

  3. [11]

    LTLDoG : Satisfying temporally-extended symbolic constraints for safe diffusion-based planning

    Zeyu Feng, Hao Luan, Pranav Goyal, and Harold Soh. LTLDoG : Satisfying temporally-extended symbolic constraints for safe diffusion-based planning. IEEE Robotics and Automation Letters, 9 0 (10): 0 8571--8578, 2024. doi:10.1109/LRA.2024.3443501

    2024

  4. [12]

    Diffusion meets options: Hierarchical generative skill composition for temporally-extended tasks

    Zeyu Feng, Hao Luan, Kevin Yuchen Ma, and Harold Soh. Diffusion meets options: Hierarchical generative skill composition for temporally-extended tasks. In 2025 International Conference on Robotics and Automation (ICRA), 2025. URL https://openreview.net/pdf?id=WjjoYyJjJW

    2025

  5. [13]

    Nathan C. Frey, Dan Berenberg, Karina Zadorozhny, Joseph Kleinhenz, Julien Lafrance-Vanasse, Isidro Hotzel, Yan Wu, Stephen Ra, Richard Bonneau, Kyunghyun Cho, Andreas Loukas, Vladimir Gligorijevic, and Saeed Saremi. Protein discovery with discrete walk-jump sampling. In The T...

    2024

  6. [14]

    Guided discrete diffusion for electronic health record generation

    Jun Han, Zixiang Chen, Yongqian Li, Yiwen Kou, Eran Halperin, Robert E Tillman, and Quanquan Gu. Guided discrete diffusion for electronic health record generation. arXiv preprint arXiv:2404.12314, 2024

    2024 arXiv

  7. [15]

    Classifier-free diffusion guidance

    Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance. In NeurIPS 2021 Workshop on Deep Generative Models and Downstream Applications, 2021. URL https://openreview.net/forum?id=qw8AKxfYbI

    2021

  8. [16]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33: 0 6840--6851, 2020

    2020

  9. [17]

    Argmax flows and multinomial diffusion: Learning categorical distributions

    Emiel Hoogeboom, Didrik Nielsen, Priyank Jaini, Patrick Forr \'e , and Max Welling. Argmax flows and multinomial diffusion: Learning categorical distributions. In A. Beygelzimer, Y. Dauphin, P. Liang, and J. Wortman Vaughan (eds.), Advances in Neural Information Processing Sys...

    2021

  10. [18]

    Diffusion- LM improves controllable text generation

    Xiang Li, John Thickstun, Ishaan Gulrajani, Percy S Liang, and Tatsunori B Hashimoto. Diffusion- LM improves controllable text generation. In Advances in neural information processing systems, volume 35, pp.\ 4328--4343, 2022

    2022

  11. [19]

    Derivative-free guidance in continuous and discrete diffusion models with soft value-based decoding

    Xiner Li, Yulai Zhao, Chenyu Wang, Gabriele Scalia, Gokcen Eraslan, Surag Nair, Tommaso Biancalani, Shuiwang Ji, Aviv Regev, Sergey Levine, et al. Derivative-free guidance in continuous and discrete diffusion models with soft value-based decoding. arXiv preprint arXiv:2408.082...

    2024 arXiv

  12. [20]

    Decoupled weight decay regularization

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

    2017 arXiv

  13. [21]

    Discrete diffusion modeling by estimating the ratios of the data distribution

    Aaron Lou, Chenlin Meng, and Stefano Ermon. Discrete diffusion modeling by estimating the ratios of the data distribution. In Proceedings of the 41st International Conference on Machine Learning, volume 235 of Proceedings of Machine Learning Research, pp.\ 32819--32848. PMLR, ...

    2024

  14. [22]

    Elucidating the design space of classifier-guided diffusion generation

    Jiajun Ma, Tianyang Hu, Wenjia Wang, and Jiacheng Sun. Elucidating the design space of classifier-guided diffusion generation. In The Twelfth International Conference on Learning Representations, 2024. URL https://openreview.net/forum?id=9DXXMXnIGm

    2024

  15. [23]

    Concrete score matching: Generalized score matching for discrete data

    Chenlin Meng, Kristy Choi, Jiaming Song, and Stefano Ermon. Concrete score matching: Generalized score matching for discrete data. In Alice H. Oh, Alekh Agarwal, Danielle Belgrave, and Kyunghyun Cho (eds.), Advances in Neural Information Processing Systems, 2022. URL https://o...

    2022

  16. [24]

    G2D2 : Gradient-guided discrete diffusion for image inverse problem solving

    Naoki Murata, Chieh-Hsin Lai, Yuhta Takida, Toshimitsu Uesaka, Bac Nguyen, Stefano Ermon, and Yuki Mitsufuji. G2D2 : Gradient-guided discrete diffusion for image inverse problem solving. arXiv preprint arXiv:2410.14710, 2024. URL https://openreview.net/forum?id=mZfBRjMWq0

    2024

  17. [25]

    Improved denoising diffusion probabilistic models

    Alexander Quinn Nichol and Prafulla Dhariwal. Improved denoising diffusion probabilistic models. In International conference on machine learning, pp.\ 8162--8171. PMLR, 2021

    2021

  18. [26]

    GLIDE : Towards photorealistic image generation and editing with text-guided diffusion models

    Alexander Quinn Nichol, Prafulla Dhariwal, Aditya Ramesh, Pranav Shyam, Pamela Mishkin, Bob Mcgrew, Ilya Sutskever, and Mark Chen. GLIDE : Towards photorealistic image generation and editing with text-guided diffusion models. In Proceedings of the 39th International Conference...

    2022

  19. [27]

    Unlocking guidance for discrete state-space diffusion and flow models

    Hunter Nisonoff, Junhao Xiong, Stephan Allenspach, and Jennifer Listgarten. Unlocking guidance for discrete state-space diffusion and flow models. In The Thirteenth International Conference on Learning Representations, 2025. URL https://openreview.net/forum?id=XsgHl54yO7

    2025

  20. [28]

    Permutation invariant graph generation via score-based generative modeling

    Chenhao Niu, Yang Song, Jiaming Song, Shengjia Zhao, Aditya Grover, and Stefano Ermon. Permutation invariant graph generation via score-based generative modeling. In Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics, volume 108,...

    2020

  21. [29]

    Pytorch: An imperative style, high-performance deep learning library

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems, 32, 2019

    2019

  22. [30]

    Bronstein, Pranam Chatterjee, Alexander Tong, and Joey Bose

    Jarrid Rector-Brooks, Mohsin Hasan, Zhangzhi Peng, Cheng-Hao Liu, Sarthak Mittal, Nouha Dziri, Michael M. Bronstein, Pranam Chatterjee, Alexander Tong, and Joey Bose. Steering masked discrete diffusion models via discrete denoising posterior prediction. In The Thirteenth Inter...

    2025

  23. [31]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj \"o rn Ommer. High-resolution image synthesis with latent diffusion models. In IEEE/CVF Conf. Comput. Vis. Pattern Recognit., pp.\ 10684--10695, 2022

    2022

  24. [32]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In Medical image computing and computer-assisted intervention--MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, part...

    2015

  25. [33]

    Minimal implementation of a D3PM (structured denoising diffusion models in discrete state-spaces), in PyTorch

    Simo Ryu. Minimal implementation of a D3PM (structured denoising diffusion models in discrete state-spaces), in PyTorch . https://github.com/cloneofsimo/d3pm, 2024

    2024

  26. [34]

    Simple guidance mechanisms for discrete diffusion models

    Yair Schiff, Subham Sekhar Sahoo, Hao Phung, Guanghan Wang, Sam Boshar, Hugo Dalla-torre, Bernardo P de Almeida, Alexander M Rush, Thomas PIERROT, and Volodymyr Kuleshov. Simple guidance mechanisms for discrete diffusion models. In The Thirteenth International Conference on Le...

    2025

  27. [35]

    Graph-constrained diffusion for end-to-end path planning

    Dingyuan Shi, Yongxin Tong, Zimu Zhou, Ke Xu, Zheng Wang, and Jieping Ye. Graph-constrained diffusion for end-to-end path planning. In The Twelfth International Conference on Learning Representations, 2024

    2024

  28. [36]

    Deep unsupervised learning using nonequilibrium thermodynamics

    Jascha Sohl-Dickstein, Eric Weiss, Niru Maheswaranathan, and Surya Ganguli. Deep unsupervised learning using nonequilibrium thermodynamics. In International conference on machine learning, pp.\ 2256--2265. PMLR, 2015

    2015

  29. [37]

    Generative modeling by estimating gradients of the data distribution

    Yang Song and Stefano Ermon. Generative modeling by estimating gradients of the data distribution. Advances in neural information processing systems, 32, 2019

    2019

  30. [38]

    Score-based generative modeling through stochastic differential equations

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equations. In International Conference on Learning Representations, 2021. URL https://openreview.net/forum?id=PxTIG12RRHS

    2021

  31. [39]

    D irichlet flow matching with applications to DNA sequence design

    Hannes Stark, Bowen Jing, Chenyu Wang, Gabriele Corso, Bonnie Berger, Regina Barzilay, and Tommi Jaakkola. D irichlet flow matching with applications to DNA sequence design. In Proceedings of the 41st International Conference on Machine Learning, volume 235 of Proceedings of M...

    2024

  32. [40]

    Inference-time alignment in diffusion models with reward-guided generation: Tutorial and review

    Masatoshi Uehara, Yulai Zhao, Chenyu Wang, Xiner Li, Aviv Regev, Sergey Levine, and Tommaso Biancalani. Inference-time alignment in diffusion models with reward-guided generation: Tutorial and review. arXiv preprint arXiv:2501.09685, 2025

    2025 arXiv

  33. [41]

    Layered graph security games

    Jakub C ern\' y , Chun Kai Ling, Christian Kroer, and Garud Iyengar. Layered graph security games. In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, IJCAI '24, 2024. ISBN 978-1-956792-04-1. doi:10.24963/ijcai.2024/298. URL https://do...

    2024 doi

  34. [42]

    DiGress : Discrete denoising diffusion for graph generation

    Clement Vignac, Igor Krawczuk, Antoine Siraudin, Bohan Wang, Volkan Cevher, and Pascal Frossard. DiGress : Discrete denoising diffusion for graph generation. In The Eleventh International Conference on Learning Representations, 2023. URL https://openreview.net/forum?id=UaAD-Nu86WX

    2023

  35. [43]

    Fine-tuning discrete diffusion models via reward optimization with applications to DNA and protein design

    Chenyu Wang, Masatoshi Uehara, Yichun He, Amy Wang, Avantika Lal, Tommi Jaakkola, Sergey Levine, Aviv Regev, Hanchen, and Tommaso Biancalani. Fine-tuning discrete diffusion models via reward optimization with applications to DNA and protein design. In The Thirteenth Internatio...

    2025

  36. [44]

    Equivariant diffusion policy

    Dian Wang, Stephen Hart, David Surovik, Tarik Kelestemur, Haojie Huang, Haibo Zhao, Mark Yeatman, Jiuguang Wang, Robin Walters, and Robert Platt. Equivariant diffusion policy. In 8th Annual Conference on Robot Learning, 2024. URL https://openreview.net/forum?id=wD2kUVLT1g

    2024

  37. [45]

    Geodiff: A geometric diffusion model for molecular conformation generation

    Minkai Xu, Lantao Yu, Yang Song, Chence Shi, Stefano Ermon, and Jian Tang. Geodiff: A geometric diffusion model for molecular conformation generation. In International Conference on Learning Representations, 2022. URL https://openreview.net/forum?id=PzcvxEMzvQC

    2022

  38. [46]

    Swin GNN : Rethinking permutation invariance in diffusion models for graph generation

    Qi Yan, Zhengyang Liang, Yang Song, Renjie Liao, and Lele Wang. Swin GNN : Rethinking permutation invariance in diffusion models for graph generation. Transactions on Machine Learning Research, 2024. ISSN 2835-8856. URL https://openreview.net/forum?id=abfi5plvQ4

    2024

  39. [47]

    Trippe, Valentin De Bortoli, Emile Mathieu, Arnaud Doucet, Regina Barzilay, and Tommi Jaakkola

    Jason Yim, Brian L. Trippe, Valentin De Bortoli, Emile Mathieu, Arnaud Doucet, Regina Barzilay, and Tommi Jaakkola. SE (3) diffusion model with application to protein backbone generation. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Pr...

    2023

  40. [48]

    Optimal escape interdiction on transportation networks

    Youzhi Zhang, Bo An, Long Tran-Thanh, Zhen Wang, Jiarui Gan, and Nicholas R Jennings. Optimal escape interdiction on transportation networks. In Proceedings of the 26th International Joint Conference on Artificial Intelligence, pp.\ 3936--3944, 2017

    2017

  41. [49]

    @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 ...

  42. [50]

    \@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...

  43. [51]

    @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...

    2024 arXiv

Pith tools

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