Pith. sign in

REVIEW 1 major objections 1 minor 40 references

Scalable Non-Equivariant 3D Molecule Generation via Rotational Alignment

T0 review · 1 major / 1 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 This paper claims equivariant denoising is not required for high-quality 3D molecule generation: a learned per-molecule rotation in latent space lets a plain diffusion model match equivariant baselines while training and sampling faster.

desk verdict Genuinely new idea and strong results, but the ablation doesn't isolate the learned alignment; the missing identity-rotation control is the main weakness. read the letter →

arxiv 2506.10186 v2 pith:6APGZEPJ submitted 2025-06-11 cs.LG cs.AIq-bio.QM

Yuhui Ding , Thomas Hofmann This is my paper
classification cs.LGcs.AIq-bio.QM
keywords 3Dmoleculegenerationdiffusionmodelsrotationalalignmentequivariancerelaxationlatentnon-equivariantnetworkstransformerbenchmarks
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 challenges the dominant assumption of 3D molecule generation: that the denoising network must be $SE(3)$-equivariant, meaning its predictions rotate and translate in lockstep with the input molecule. It argues this constraint is a design choice, not a necessity, and proposes RADM, which first learns a per-molecule $SO(3)$ rotation with a small graph network and then trains a completely non-equivariant latent diffusion model on the aligned molecules. The rotation is learned without supervision, purely from reconstruction loss through an autoencoder whose decoder is deliberately non-equivariant, so that orientation determines the loss and the network is pushed to find simplifying alignments. On QM9 and GEOM-Drugs, RADM matches the best equivariant latent diffusion baseline on validity and atom stability and cuts sampling time by up to a factor of seven, with scaling from small to base transformer giving consistent gains. If the claim holds, molecule generation can adopt the scalable, well-optimized architectures of vision and language instead of waiting on specialized equivariant layers.

What carries the argument

The load-bearing component is the learned rotational alignment. A vanilla GNN maps a molecule to an unconstrained matrix $M \in \mathbb{R}^{3\times 3}$, which is projected onto the rotation group by the SVD-based map $R = \mathrm{SVD}_+(M) = U\,\mathrm{diag}(1,1,\det(UV^\top))V^\top$; this sample-dependent rotation is applied to the atomic coordinates before encoding. The rotation network is trained jointly with the autoencoder using only the reconstruction loss, and the non-equivariance of the decoder is what makes the gradient informative — if both encoder and decoder were equivariant, the $L^2$ reconstruction error would be constant under rotation and the rotation network would have nothing to learn. The same fixed autoencoder feeds a latent diffusion model whose denoiser is either a vanilla GNN or a DiT-style transformer with attention masking for variable atom counts, leaving permutation equivariance as the only inductive bias.

What would settle it

Train the identical pipeline but freeze the rotation network's output to a per-molecule random $SO(3)$ rotation sampled once at encoding time, leaving every other component unchanged, and compare QM9 molecule stability: if the random-rotation variant stays near RADM-GNN's 85.8% instead of sliding toward GraphLDM's 70.5%, the learned alignment is not what drives the gain. A complementary check is to apply the trained rotation network to held-out molecules and measure whether the orientation of shared substructures, such as ring-plane normals, is significantly more concentrated than under random rotation — a uniform orientation distribution would show the alignment itself is vacuous.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that the equivariance constraints built into state-of-the-art 3D molecule diffusion models can be relaxed without losing sample quality, provided the data is pre-aligned. A rotation network maps each molecule to an arbitrary $3\times 3$ matrix that is projected onto $SO(3)$ — the rotation group — by a singular value decomposition, and the resulting rotation is applied to the atomic coordinates before they enter the autoencoder. Because the decoder is deliberately non-equivariant, the reconstruction loss is sensitive to orientation, so the rotation network receives a gradient that pushes molecules toward orientations the encoder can represent compactly. The aligned latent space is then simple enough for a plain permutation-equivariant denoiser: on GEOM-Drugs, RADM-DiT-B reaches 85.0% atom stability against GeoLDM's 84.4% with equal 99.3% validity, and on QM9 it approaches GeoLDM's molecule stability (87.3% vs 89.4%) while exceeding its validity (94.1% vs 93.8%). The paper frames these results as evidence that a well-aligned representation, not equivariant layers, is what carries the performance.

Load-bearing premise

The load-bearing premise, stated by the paper itself as an intuition (Section 3.1), is that reconstruction-driven alignment genuinely helps: rotating each molecule into a learned orientation is what lets a plain diffusion network match equivariant models, and the paper offers no guarantee that this alignment transfers beyond the datasets and molecule sizes tested.

Editorial extensions

If this is right

  • RADM-DiT-B matches the state-of-the-art equivariant GeoLDM on GEOM-Drugs atom stability (85.0% vs 84.4%) and validity (99.3% each), and exceeds it on QM9 validity (94.1% vs 93.8%), so non-equivariant generators can be practically competitive.
  • Sampling 100 molecules takes 7 seconds with RADM-DiT-S and 19 seconds with RADM-DiT-B, versus 49 seconds for GeoLDM and 55 seconds for EDM, transferring the speed of parallel transformer architectures to molecule diffusion.
  • Scaling the denoiser from DiT-S to DiT-B improves every quality metric on both datasets, indicating that non-equivariant molecule diffusion inherits the scaling behavior of standard deep learning models.
  • In conditional generation, RADM-DiT-B achieves lower mean absolute error than both EDM and GeoLDM on all six QM9 properties tested.
  • The learned rotations are visually meaningful: after alignment, shared substructures such as rings appear in similar orientations across molecules, evidence that the autoencoder discovers chemically sensible frames without supervision.

Reading between the lines

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

  • The same recipe — a learned transform that removes nuisance symmetries, followed by a symmetry-free generative model — should transfer to other structured generation problems such as point clouds, protein backbones, or crystal structures, though the paper does not test those settings.
  • Because the alignment objective is purely reconstructive, nothing prevents task-aware alignment: coupling the rotation network to downstream objectives such as property prediction could produce frames even more favorable than the reconstruction-only ones reported here.
  • A sharper ablation than the ones in the paper would train the autoencoder with fixed random rotations per molecule, isolating whether the gain comes from the learned consistency of the frames or merely from the autoencoder's denoising.
  • The approach never requires the aligned frame to be canonical or consistent across re-samplings, treating alignment only as nuisance reduction; this weaker goal is what lets it sidestep known impossibility results for continuous canonicalization.
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

1 major / 1 minor

Summary. The paper proposes RADM, a non-equivariant latent diffusion model for 3D molecule generation. A graph neural network predicts a per-molecule SO(3) rotation, the molecule is rotated accordingly, and an autoencoder with an equivariant encoder and a non-equivariant decoder maps the rotated molecule into a latent space. A non-equivariant GNN or DiT is then trained as the denoising network in this aligned latent space. Experiments on QM9 and GEOM-Drugs report atom stability, molecule stability, validity, and uniqueness numbers that are competitive with the equivariant baselines EDM and GeoLDM, along with improved training and sampling efficiency. Conditional property prediction on QM9 also shows lower MAE than the reported EDM and GeoLDM baselines.

Significance. If the central mechanism is confirmed, the paper would provide a practical alternative to SE(3)-equivariant denoising networks for 3D molecule generation, with the scalability and speed advantages of transformer-based non-equivariant architectures. The paper is clearly written, includes a code repository, and reports a same-backbone ablation that shows large gains over the non-equivariant baselines GraphLDM and GraphLDM-aug. The reported results, if reproducible, would be a meaningful advance in relaxing architectural equivariance constraints without losing sample quality. However, the attribution of these gains specifically to the learned rotational alignment is not yet cleanly established by the presented experiments, and the cross-paper baseline comparisons leave some uncertainty in the headline claims.

major comments (1)
  1. [Section 4.4, Table 2] Section 4.2 and 4.3, Table 1: Baseline numbers are copied from previous papers with different training setups, and no error bars are given for those baselines. The claim of comparability to state-of-the-art equivariant models would be substantially strengthened by re-evaluating EDM and GeoLDM with the same evaluation code and, if feasible, the same training budget. At minimum, the paper should state the training steps, batch sizes, and sampling protocols used by the baselines so that the comparison is transparent.
minor comments (1)
  1. [Section 4.5, Table 3] Section 4.2: The sentence 'RADM DiT-S improves the performance drastically compared with previous non-equivariant models' is informal; specify which metrics and which baselines are meant.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the learned rotation is trained by reconstruction loss and evaluated against external stability and validity metrics.

full rationale

The paper's derivation chain is not circular. The alignment network R_theta is trained jointly with the autoencoder to minimize reconstruction loss (Eq. 20, Algorithm 1), and the latent diffusion model is trained with the standard noise-prediction objective (Eq. 12). Neither objective contains the evaluation quantities (atom/molecule stability, RDKit validity, property MAE), so no fitted parameter is renamed as a prediction. Eq. 20 and Eq. 12 are independent objectives; no equation reduces to another by construction. Citations to GeoLDM, EDM, and DiT are code bases and baselines, not the present authors' prior work, and they supply architecture details rather than premises that assume the conclusion. The related learned-canonicalization literature (Kaba et al., Dym et al., Sareen et al.) is presented as background and is not invoked as a load-bearing uniqueness theorem. The PCA variant is an external ablation control, not a circular input. The absence of a theoretical guarantee that minimum-reconstruction alignments help diffusion is a limitation of the empirical argument, but it is not a circular reduction. The reviewer concern that Table 2 does not fully isolate the learned rotation from the new autoencoder is an experimental confounding issue, not circularity: it does not correspond to any quantity in the derivation being defined in terms of the target result. The central performance claims are validated by external metrics against independent baselines, so the derivation chain is self-contained.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical entities or fitted scientific constants. The rotation network is a learned function, not an ad hoc parameter. The key assumptions are the SE(3) invariance of molecular properties, the differentiability of the SVD+ parameterization, and the adequacy of the standard diffusion objective. The listed free parameters are training choices that are not fitted to the target metrics but could affect the results.

free parameters (3)
  • Fixed latent variance sigma^2 = not reported
    Eq (18) uses a fixed variance sigma^2 for the latent posterior q(z|x,h); this is a hand-chosen constant following VAE practice, and its value is not given in the paper.
  • Rotation network depth = 2 layers on both datasets
    Section 4.1 states the rotation network has 2 layers on both datasets, but no ablation or sensitivity analysis is provided for this architectural choice.
  • Autoencoder early stopping criterion = unspecified
    Section 3.1 says 'we adopt an early stopping training strategy as the regularization for the encoder' following Xu et al. (2023), but the stopping rule or epoch is not specified, making this a hidden training choice that could affect the latent space.
assumptions (4)
  • domain assumption Molecular properties are invariant under SE(3) transformations
    Section 2.1 states chemical properties remain unchanged under rotation and translation; this is the basis for the entire symmetry discussion and the choice to only align rotations after centering.
  • standard math The SVD+ projection is differentiable and trainable via gradient descent where det(M) != 0
    Section 3.1 Eq (16) uses SVD+ to map an arbitrary matrix to SO(3), citing Levinson et al. (2020) for smoothness. The paper assumes this parameterization gives usable gradients for the rotation network.
  • standard math The diffusion training objective in Eq (12) is a valid surrogate for the ELBO
    Section 2.2 derives the noise-prediction objective following Ho et al. (2020) and Kingma et al. (2021); the paper relies on this standard derivation without re-proving it.
  • domain assumption Early stopping on the autoencoder acts as sufficient regularization
    Section 3.1 adopts early stopping from Xu et al. (2023) as the only regularization for the encoder, but provides no analysis of whether this prevents overfitting or how it interacts with the alignment objective.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Scalable Non-Equivariant 3D Molecule Generation via Rotational Alignment." pith.science (2026). https://pith.science/paper/6APGZEPJ

@misc{pith2026250610186,
  author       = {Pith},
  title        = {Pith review of: Scalable Non-Equivariant 3D Molecule Generation via Rotational Alignment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6APGZEPJ}},
  note         = {Machine review of arXiv:2506.10186}
}
read the original abstract

Equivariant diffusion models have achieved impressive performance in 3D molecule generation. These models incorporate Euclidean symmetries of 3D molecules by utilizing an SE(3)-equivariant denoising network. However, specialized equivariant architectures limit the scalability and efficiency of diffusion models. In this paper, we propose an approach that relaxes such equivariance constraints. Specifically, our approach learns a sample-dependent SO(3) transformation for each molecule to construct an aligned latent space. A non-equivariant diffusion model is then trained over the aligned representations. Experimental results demonstrate that our approach performs significantly better than previously reported non-equivariant models. It yields sample quality comparable to state-of-the-art equivariant diffusion models and offers improved training and sampling efficiency. Our code is available at https://github.com/skeletondyh/RADM

Figures

Figures reproduced from arXiv: 2506.10186 by the authors.

Figure 1
Figure 1. Molecules generated by RADMDiT-B on QM9 (the three on the left) and GEOM-Drugs (the three on the right). non-equivariant models. RADMDiT-S outperforms EDM on molecule stability and outperforms both EDM and EDM￾bridge on validity. We find that scaling from DiT-S to DiT-B consistently boosts performance on all metrics. RADMDiT-B greatly reduces the gap with the best equivariant model. It performs slightly worse than G… view at source ↗
Figure 2
Figure 2. Molecule samples from the training set [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Molecule samples after applying the learned rotations. [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

40 extracted references · 29 canonical work pages

  1. [1]

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

    Austin, J., Johnson, D. D., Ho, J., Tarlow, D., and Van Den Berg, R. Structured denoising diffusion models in discrete state-spaces. In Advances in Neural Information Processing Systems, 2021

    work page 2021

  2. [2]

    and Gomez-Bombarelli, R

    Axelrod, S. and Gomez-Bombarelli, R. Geom, energy-annotated molecular conformations for property prediction and molecular generation. Scientific Data, 2022

    work page 2022

  3. [3]

    Deep regression on manifolds: a 3d rotation case study

    Br \'e gier, R. Deep regression on manifolds: a 3d rotation case study. In 2021 International Conference on 3D Vision (3DV), 2021

    work page 2021

  4. [4]

    X., Funkhouser, T., Guibas, L., Hanrahan, P., Huang, Q., Li, Z., Savarese, S., Savva, M., Song, S., Su, H., et al

    Chang, A. X., Funkhouser, T., Guibas, L., Hanrahan, P., Huang, Q., Li, Z., Savarese, S., Savva, M., Song, S., Su, H., et al. Shapenet: An information-rich 3d model repository. arXiv preprint arXiv:1512.03012, 2015

    arXiv 2015

  5. [5]

    Flashattention: Fast and memory-efficient exact attention with io-awareness

    Dao, T., Fu, D., Ermon, S., Rudra, A., and R \'e , C. Flashattention: Fast and memory-efficient exact attention with io-awareness. In Advances in Neural Information Processing Systems, 2022

    work page 2022

  6. [6]

    and Nichol, A

    Dhariwal, P. and Nichol, A. Diffusion models beat gans on image synthesis. In Advances in neural information processing systems, 2021

    work page 2021

  7. [7]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., and Houlsby, N. An image is worth 16x16 words: Transformers for image recognition at scale. In International Conference on Learning Representations, 2021

    2021

  8. [8]

    Dym, N., Lawrence, H., and Siegel, J. W. Equivariant frames and the impossibility of continuous canonicalization. In International Conference on Machine Learning, 2024

    work page 2024

Show all 40 references
  1. [9]

    E (n) equivariant normalizing flows

    Garcia Satorras, V., Hoogeboom, E., Fuchs, F., Posner, I., and Welling, M. E (n) equivariant normalizing flows. In Advances in Neural Information Processing Systems, 2021

    2021

  2. [10]

    Symmetry-adapted generation of 3d point sets for the targeted discovery of molecules

    Gebauer, N., Gastegger, M., and Sch \"u tt, K. Symmetry-adapted generation of 3d point sets for the targeted discovery of molecules. In Advances in neural information processing systems, 2019

    2019

  3. [11]

    R., Frey, J., Zhobro, M., Levina, A., and Martius, G

    Geist, A. R., Frey, J., Zhobro, M., Levina, A., and Martius, G. Learning with 3d rotations, a hitchhiker's guide to SO (3). In Forty-first International Conference on Machine Learning, 2024

    2024

  4. [12]

    Denoising diffusion probabilistic models

    Ho, J., Jain, A., and Abbeel, P. Denoising diffusion probabilistic models. In Advances in neural information processing systems, 2020

    2020

  5. [13]

    G., Vignac, C., and Welling, M

    Hoogeboom, E., Satorras, V. G., Vignac, C., and Welling, M. Equivariant diffusion for molecule generation in 3d. In International conference on machine learning, 2022

    2022

  6. [14]

    Hierarchical generation of molecular graphs using structural motifs

    Jin, W., Barzilay, R., and Jaakkola, T. Hierarchical generation of molecular graphs using structural motifs. In International conference on machine learning, 2020

    2020

  7. [15]

    Jo, J., Lee, S., and Hwang, S. J. Score-based generative modeling of graphs via the system of stochastic differential equations. In International conference on machine learning, pp.\ 10362--10383. PMLR, 2022

    2022

  8. [16]

    K., Fu, X., Liao, Y.-L., Gharakhanyan, V., Miller, B

    Joshi, C. K., Fu, X., Liao, Y.-L., Gharakhanyan, V., Miller, B. K., Sriram, A., and Ulissi, Z. W. All-atom diffusion transformers: Unified generative modelling of molecules and materials. arXiv preprint arXiv:2503.03965, 2025

    2025 arXiv

  9. [17]

    K., Zhang, Y., Bengio, Y., and Ravanbakhsh, S

    Kaba, S.-O., Mondal, A. K., Zhang, Y., Bengio, Y., and Ravanbakhsh, S. Equivariance with learned canonicalization functions. In International Conference on Machine Learning, 2023

    2023

  10. [18]

    Variational diffusion models

    Kingma, D., Salimans, T., Poole, B., and Ho, J. Variational diffusion models. In Advances in neural information processing systems, 2021

    2021

  11. [19]

    Equivariant flows: exact likelihood generative learning for symmetric densities

    K \"o hler, J., Klein, L., and No \'e , F. Equivariant flows: exact likelihood generative learning for symmetric densities. In International conference on machine learning, 2020

    2020

  12. [20]

    Crafting papers on machine learning

    Langley, P. Crafting papers on machine learning. In Langley, P. (ed.), Proceedings of the 17th International Conference on Machine Learning (ICML 2000), pp.\ 1207--1216, Stanford, CA, 2000. Morgan Kaufmann

    2000

  13. [21]

    Le, T., Cremer, J., Noe, F., Clevert, D.-A., and Sch \"u tt, K. T. Navigating the design space of equivariant diffusion-based generative models for de novo 3d molecule generation. In The Twelfth International Conference on Learning Representations, 2024

    2024

  14. [22]

    An analysis of svd for deep rotation estimation

    Levinson, J., Esteves, C., Chen, K., Snavely, N., Kanazawa, A., Rostamizadeh, A., and Makadia, A. An analysis of svd for deep rotation estimation. In Advances in Neural Information Processing Systems, 2020

    2020

  15. [23]

    and Xie, S

    Peebles, W. and Xie, S. Scalable diffusion models with transformers. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023

    2023

  16. [24]

    O., Rupp, M., and Von Lilienfeld, O

    Ramakrishnan, R., Dral, P. O., Rupp, M., and Von Lilienfeld, O. A. Quantum chemistry structures and properties of 134 kilo molecules. Scientific data, 2014

    2014

  17. [25]

    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, 2022

    2022

  18. [26]

    K., Kaba, S.-O., Akhound-Sadegh, T., and Ravanbakhsh, S

    Sareen, K., Levy, D., Mondal, A. K., Kaba, S.-O., Akhound-Sadegh, T., and Ravanbakhsh, S. Symmetry-aware generative modeling through learned canonicalization. arXiv preprint arXiv:2501.07773, 2025

    2025 arXiv

  19. [27]

    G., Hoogeboom, E., and Welling, M

    Satorras, V. G., Hoogeboom, E., and Welling, M. E (n) equivariant graph neural networks. In International conference on machine learning, 2021

    2021

  20. [28]

    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, 2015

    2015

  21. [29]

    and Ermon, S

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

    2019

  22. [30]

    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. In International Conference on Learning Representations, 2021

    2021

  23. [31]

    N., Kaiser, L

    Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L. u., and Polosukhin, I. Attention is all you need. In Advances in Neural Information Processing Systems, 2017

    2017

  24. [32]

    Digress: Discrete denoising diffusion for graph generation

    Vignac, C., Krawczuk, I., Siraudin, A., Wang, B., Cevher, V., and Frossard, P. Digress: Discrete denoising diffusion for graph generation. In The Eleventh International Conference on Learning Representations, 2023 a

    2023

  25. [33]

    Midi: Mixed graph and 3d denoising diffusion for molecule generation

    Vignac, C., Osman, N., Toni, L., and Frossard, P. Midi: Mixed graph and 3d denoising diffusion for molecule generation. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases, 2023 b

    2023

  26. [34]

    u tte, D., Fluri, J., Ding, Y., Orvieto, A., Sch \

    von R \"u tte, D., Fluri, J., Ding, Y., Orvieto, A., Sch \"o lkopf, B., and Hofmann, T. Generalized interpolating discrete diffusion. arXiv preprint arXiv:2503.04482, 2025

    2025 arXiv

  27. [35]

    A., Jaitly, N., Susskind, J

    Wang, Y., Elhag, A. A., Jaitly, N., Susskind, J. M., and Bautista, M. \'A . Swallowing the bitter pill: Simplified scalable conformer generation. In Forty-first International Conference on Machine Learning, 2024

    2024

  28. [36]

    Diffusion-based molecule generation with informative prior bridges

    Wu, L., Gong, C., Liu, X., Ye, M., and Liu, Q. Diffusion-based molecule generation with informative prior bridges. In Advances in Neural Information Processing Systems, 2022

    2022

  29. [37]

    Geodiff: A geometric diffusion model for molecular conformation generation

    Xu, M., Yu, L., Song, Y., Shi, C., Ermon, S., and Tang, J. Geodiff: A geometric diffusion model for molecular conformation generation. In International Conference on Learning Representations, 2022

    2022

  30. [38]

    S., Dror, R

    Xu, M., Powers, A. S., Dror, R. O., Ermon, S., and Leskovec, J. Geometric latent diffusion models for 3d molecule generation. In International Conference on Machine Learning, 2023

    2023

  31. [39]

    L., De Bortoli, V., Mathieu, E., Doucet, A., Barzilay, R., and Jaakkola, T

    Yim, J., Trippe, B. L., De Bortoli, V., Mathieu, E., Doucet, A., Barzilay, R., and Jaakkola, T. Se (3) diffusion model with application to protein backbone generation. In International Conference on Machine Learning, 2023

    2023

  32. [40]

    W., and Cornish, R

    Zhang, L., Ashouritaklimi, K., Teh, Y. W., and Cornish, R. Symdiff: Equivariant diffusion via stochastic symmetrisation. arXiv preprint arXiv:2410.06262, 2024

    2024 arXiv

Pith tools

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