Pith. sign in

REVIEW 4 major objections 5 minor 20 references

Scalable Generation of Spatial Transcriptomics from Histology Images via Whole-Slide Flow Matching

T0 review · 4 major / 5 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 STFlow predicts a tissue slide's gene expression from its H&E image by modeling all spots jointly with flow matching, and reports over 18% relative improvement over pathology foundation models on two benchmarks.

desk verdict A solid, well-ablated empirical model for ST prediction, but the headline SOTA gains ride on curated benchmarks and a loose 'flow matching' label; referee it, but ask for a protocol-clean rerun. read the letter →

arxiv 2506.05361 v1 pith:WQKQNWDT submitted 2025-05-25 cs.CV q-bio.GN

Tinglin Huang , Tianyu Liu , Mehrtash Babadi , Wengong Jin , Rex Ying This is my paper
classification cs.CVq-bio.GN
keywords spatialtranscriptomicsflowmatchinggeneexpressionpredictionwhole-slidehistologycell-cellinteractionE(2)-invariantattentionframeaveragingzero-inflatednegativebinomialprior
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

STFlow aims to make spatial transcriptomics scalable by predicting a tissue slide's gene expression directly from its H&E-stained histology image, replacing a slow, specialized laboratory assay with computation. The paper argues that prior prediction methods have two weaknesses: they predict each spot's genes independently, ignoring that neighboring cells regulate each other's expression, and their whole-slide encoders cannot handle the tens of thousands of spots on a typical slide. STFlow instead models the entire slide's gene expression as one joint distribution learned by flow matching, in which a denoiser iteratively refines an initial expression guess for every spot under the guidance of the image and of local spatial context encoded by an $E(2)$-invariant attention mechanism. On seventeen benchmark datasets spanning two collections, the paper reports that this design outperforms all eight baselines, including an over-18% relative improvement in average Pearson correlation over pathology foundation models, and it gives the best predictions for the four clinical biomarker genes tested.

What carries the argument

The argument runs on three coupled components. (1) Flow matching with a gene-specific prior: the network $f_\theta$ minimizes MSE between ground truth $Y$ and its reconstruction from the noised input $Y_t$, and at inference an initial sample from the zero-inflated negative binomial distribution $Z(\mu, \phi, \pi)$ is refined over $S$ steps, so prediction is a denoising trajectory rather than a single forward pass. (2) E(2)-invariant frame-averaging spatial attention: for each spot, direction vectors to its $k$ nearest neighbors are projected through four PCA-derived frames (two principal components with sign flips), embedded, and averaged, making the representation invariant to rotation, translation, and reflection; the attention weight then uses an MLP over query, key, spatial relation, and pairwise expression difference $(Y_{t,i}-Y_{t,j})$, which couples the gene predictions of neighboring spots. (3) Local neighborhood sparsity: attention is computed only among $k$ nearest neighbors, giving $O(Nkd + Nkd^2)$ complexity instead of $O(N^2)$ global attention, which is what makes the whole-slide formulation practical for up to tens of thousands of spots. The ablation studies identify both flow matching and frame averaging as individually necessary for the reported accuracy.

What would settle it

Reproduce the comparison on the unmodified benchmarks — the published version of COAD in HEST-1k and all STImage-1K4M organs, including those whose baseline correlation sits below 0.1 — and measure STFlow's average relative improvement over the pathology foundation models; a margin materially below the reported 18% would overturn the headline claim. The cell-interaction claim can be tested directly: in held-out slides, check whether spot pairs receiving high spatial-attention weights coincide with known co-expressed or ligand-receptor gene pairs, and whether the one-step (no flow matching) variant does equally well on that pairing test; if it does, the iterative joint modeling is not the source of the biological signal.

Watch

Extended reading notes

Core claim

The paper's central claim is that whole-slide gene expression should be generated, not regressed: rather than predicting each spot's expression $Y_i$ from its own image patch, STFlow models the joint distribution $p(Y_0, \dots, Y_N \mid I_0, \dots, I_N)$ and samples from it through iterative denoising. The denoiser is trained with flow matching on linear interpolation paths $Y_t = tY + (1-t)Y_0$ between ground truth and a sample from a prior distribution, and the zero-inflated negative binomial distribution $Z(\mu, \phi, \pi)$ is chosen as the prior because real expression counts are sparse and overdispersed. Spatial context enters through frame-averaging attention restricted to each spot's $k$ nearest neighbors: the attention weight between two spots is a learned function of their image features, their relative direction vector encoded in an $E(2)$-invariant way, and the difference in their current expression, which is the mechanism the paper identifies with cell-cell interaction. The key reported finding is that this compound design works: average Pearson correlations of 0.415 on HEST-1k and 0.293 on STImage-1K4M place it ahead of every baseline, and ablations show that removing flow matching, removing frame averaging, or replacing the backbone with EGNN or E2CNN each costs accuracy.

Load-bearing premise

The performance claim rests on the benchmark comparison being fair and representative: the STImage benchmark used here keeps only organs where every method exceeds a Pearson correlation of 0.1, and the COAD dataset in the HEST benchmark differs from the version in the published collection, so gains measured on this curated comparison may not transfer to the unmodified benchmarks.

Editorial extensions

If this is right

  • If the reported gains hold, spatial transcriptomics becomes a computation on a histology scan: full slides with thousands of spots are processed in one pass, with the paper reporting orders-of-magnitude faster inference and lower memory usage than slide-level attention baselines.
  • The four biomarker genes tested (GATA3, ERBB2, UBE2C, VWF) are all predicted more accurately than by the best baseline or by the foundation model alone, which strengthens the case that histology-only prediction could support clinical biomarker screening.
  • The improvement is consistent across three pathology foundation models (Ciga, UNI, Gigapath), meaning the method stacks on top of image encoders and should inherit future encoder advances.
  • The reported step-count analysis shows that two refinement steps already beat one-step prediction and that five steps suffice, so the generative formulation buys accuracy at a small, predictable inference cost.

Reading between the lines

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

  • Editorial inference: the pairwise expression difference $(Y_{t,i}-Y_{t,j})$ inside the attention weight makes the model's attention maps a candidate readout of cell-cell signaling; one direct test would check whether high-attention spot pairs in held-out slides coincide with known ligand-receptor co-expressed gene pairs, which would turn the biological claim from an aggregate-accuracy argument int
  • Editorial inference: the $E(2)$-invariance property predicts that outputs are unchanged when the whole slide is rotated, translated, or reflected, and that the model transfers across coordinate conventions without retuning; this is testable with straightforward geometric perturbations and, if confirmed, would ease cross-institution and cross-array deployment.
  • Editorial inference: the largest relative gains in the reported tables occur on the hardest organs (HCC, READ, CCRCC), where baseline correlations are below 0.3; this suggests joint generative modeling may matter most when spot-level morphology alone carries the least signal, a pattern worth validating prospectively on new organs.
  • Editorial inference: the paper's stated limitation that the ZINB prior's hyperparameters are grid-searched rather than estimated suggests a concrete next step it does not take — estimating them from the training distribution, for instance with a small amortized model — which could both remove a free parameter and tighten the prior's match to each slide.
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

4 major / 5 minor

Summary. The paper introduces STFlow, a flow-matching-based generative model for predicting whole-slide spatial transcriptomics from H&E-stained histology images. The method models the joint distribution of gene expression across spots via an iterative denoising process, using a local spatial attention transformer with E(2)-invariance achieved through frame averaging. A zero-inflated negative binomial prior is used to capture the sparsity and overdispersion of gene expression counts. The authors evaluate STFlow on HEST-1k and STImage-1K4M, reporting large average improvements over spot-based and slide-based baselines, including pathology foundation models, along with better efficiency and lower memory usage. The paper also includes biomarker prediction experiments, ablations, and hyperparameter studies.

Significance. If the claims are validated, STFlow is a significant contribution: it addresses the memory bottleneck of whole-slide processing, explicitly incorporates spatial cell-cell interactions, and demonstrates that a generative formulation with a tailored prior can improve gene-expression prediction over regression-based baselines. The paper's strengths include a public implementation, experiments across 17 datasets and four biomarkers, multiple ablations, and clear disclosure of some benchmark modifications. However, the headline 'substantially outperforms state-of-the-art' claim is currently weakened by benchmark curation choices and by an internal inconsistency in the 'consistently outperforms' statement. These issues are fixable and do not invalidate the core methodology, but they must be addressed before the benchmark-level result can be accepted at face value.

major comments (4)
  1. [Section 4.1 and Appendix B (Tables 7-8)] The STImage benchmark used in this paper is not the official STImage-1K4M benchmark: Appendix B states that only organs where all models achieve Pearson correlation greater than 0.1 are included. This is a selection on the evaluation metric, and it can inflate the average improvement reported in Table 1 and the abstract. Since the abstract and Section 4.1 claim superiority on STImage-1K4M, the authors should either rerun the full protocol on the official STImage-1K4M (including all organs) or explicitly qualify the claim as applying only to the curated STImage-Bench subset. This is load-bearing for the central claim.
  2. [Footnote 2 and Appendix B] The COAD dataset used in the experiments is an older version than the one in the HEST-1k release, as disclosed in footnote 2. Because the paper compares against published HEST-1k results and other baselines, this version mismatch affects the comparability of the COAD row in Table 1. The authors should report the exact version used and, ideally, also provide results on the current HEST-1k COAD version so that readers can verify that the relative improvements are not an artifact of the dataset version.
  3. [Section 4.1 and Table 1] The statement that STFlow 'consistently outperforms all baselines' is contradicted by Table 1. For example, on HEST IDC, TRIPLEX achieves 0.606 while STFlow achieves 0.587; on STImage Breast, TRIPLEX achieves 0.418 vs STFlow 0.404; and on STImage Mouth, BLEEP achieves 0.191 vs STFlow 0.146. The paper should soften the claim to 'outperforms baselines on average' and discuss the datasets where STFlow does not lead, or provide an analysis of when spatial-context modeling helps and when it does not. As written, the overclaim undermines the paper's central narrative.
  4. [Section 3.2, Eq. (2)] The training objective in Eq. (2) is a plain mean-squared error between the predicted gene expression and the ground truth, not the standard conditional flow-matching loss on the vector field. For the linear interpolation Y_t = tY + (1-t)Y0, the conditional vector field is Y - Y0; minimizing MSE(Y, fθ) is equivalent to the flow-matching objective only up to a time-dependent weighting (specifically, dividing both sides by (1-t)). The text says fθ approximates the marginal vector field, which is imprecise as written. The authors should state this reparameterization explicitly and justify the unweighted denoiser objective, or use the standard flow-matching loss, so that the 'flow matching' claim is technically accurate.
minor comments (5)
  1. [Section 3.2, Algorithm 2] The term 'decay coefficient' is not formally defined. Algorithm 2 uses Y_t2 = Y_t1 + (Y_hat - Y_t1)/(1-t1) * (t2 - t1), which is an Euler step with velocity (Y_hat - Y_t1)/(1-t1). Please describe it as such and avoid the vague 'decay coefficient' terminology.
  2. [Figure 4 and text] There is a typo: 'UB2EC' should be 'UBE2C' in Figure 4 and the surrounding text.
  3. [Eq. (3)] The parameter φ is described as 'the number of failures until stopped,' which is non-standard for the negative binomial parametrization used here. It would be clearer to call φ the inverse dispersion parameter.
  4. [Section 4.1] The evaluation uses Pearson correlation on log1p-transformed counts while training uses MSE on raw counts. This train/evaluation mismatch should be discussed; in particular, a loss on raw counts may be dominated by highly expressed genes. The authors should justify this choice or include an ablation training on log1p-transformed targets.
  5. [Table 1 footnote] Footnote 2 notes that the COAD dataset was updated after the paper's release, but it does not specify which version was used to obtain the reported numbers. Please add the exact dataset version and, if possible, the download date.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: STFlow is an empirical model whose predictions are trained and evaluated on held-out splits; the minor self-citation for the S=5 default is not load-bearing, and the benchmark-curation concerns affect external validity, not derivation.

full rationale

STFlow is an empirical systems paper; its central claim is measured benchmark performance, and no predicted quantity is algebraically or statistically forced by a fitted input. The flow-matching objective (Eq. 2) is a standard supervised MSE trained on the training split, and the ZINB prior (Eq. 3) is grid-searched, but the Limitation section openly describes this as hyperparameter selection rather than a fitted prediction. Test correlations are reported under patient-stratified cross-validation for HEST-1k and slide-level splits for STImage-1K4M, so the evaluation metric is not identical to the tuning criterion. The S=5 sampling default cites AlphaFlow and RNAFlow, including coauthor W. Jin, but the paper independently sweeps S in {1,2,5,10,16} and observes a plateau at S=5, so the self-citation is not load-bearing. The benchmark-curation details in Appendix B and footnote 2 (keeping only STImage organs with Pearson correlation above 0.1 and using an older COAD version) are properly viewed as threats to the external validity of the SOTA comparison, not as circular derivation: the reported model outputs do not reduce to their inputs by construction, and all comparisons are internally consistent and disclosed. No circular step was found.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The central claim depends on standard supervised-learning assumptions plus the ZINB prior and local attention. The only fitted quantities specific to the method are the ZINB prior parameters and standard hyperparameters; no new physical or biological entities are postulated.

free parameters (5)
  • ZINB zero-inflation probability pi = 0.5 (fixed)
    Set by hand in the prior distribution for flow matching sampling (Section 3.2 and Appendix A).
  • ZINB mean mu = grid search over {0.1, 0.2, 0.4}
    Hyperparameter for the prior, selected via grid search on the benchmark data; performance varies slightly across datasets (Table 13).
  • ZINB dispersion phi = grid search over {1, 2, 4}
    Hyperparameter for the prior, selected via grid search, as admitted in the Limitation section and Appendix C.
  • Number of neighbors k = 8
    Chosen for local spatial attention; hyperparameter search for slide-based methods covers {4, 8, 25} and STFlow uses 8.
  • Number of flow steps S = 5
    Selected based on the step-number ablation (Table 5); gains from S=2 to S=5 are small.
assumptions (6)
  • domain assumption H&E-stained histology images contain sufficient morphological signal to predict spatial gene expression.
    Foundational premise of the whole task, stated in Section 1.
  • domain assumption Local k-nearest-neighbor spatial context is sufficient to capture the cell-cell interactions that determine gene expression.
    The denoiser restricts attention to the k=8 nearest neighbors (Section 3.3, Eqs. 7-8), so long-range interactions are only propagated through layers.
  • domain assumption Frozen pathology foundation model embeddings are strong enough feature extractors for this task.
    STFlow uses UNI, Gigapath, or Ciga embeddings without fine-tuning (Section 3.1, Appendix A).
  • domain assumption Gene expression counts follow a zero-inflated negative binomial distribution.
    Justified by Figure 2; used to sample the initial flow state (Section 3.2).
  • domain assumption PCA-based frame averaging for the local coordinate context is well-defined and yields strict E(2)-invariance.
    Assumes no degeneracy, such as collinear neighbor sets, that would make PCA frames ambiguous (Section 3.3, Eqs. 5-6).
  • domain assumption Minimizing MSE on raw counts optimizes the log1p-normalized Pearson correlation used for evaluation.
    Training objective (Eq. 2) and evaluation metric (Section 4.1) are not provably aligned.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Scalable Generation of Spatial Transcriptomics from Histology Images via Whole-Slide Flow Matching." pith.science (2026). https://pith.science/paper/WQKQNWDT

@misc{pith2026250605361,
  author       = {Pith},
  title        = {Pith review of: Scalable Generation of Spatial Transcriptomics from Histology Images via Whole-Slide Flow Matching},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WQKQNWDT}},
  note         = {Machine review of arXiv:2506.05361}
}
read the original abstract

Spatial transcriptomics (ST) has emerged as a powerful technology for bridging histology imaging with gene expression profiling. However, its application has been limited by low throughput and the need for specialized experimental facilities. Prior works sought to predict ST from whole-slide histology images to accelerate this process, but they suffer from two major limitations. First, they do not explicitly model cell-cell interaction as they factorize the joint distribution of whole-slide ST data and predict the gene expression of each spot independently. Second, their encoders struggle with memory constraints due to the large number of spots (often exceeding 10,000) in typical ST datasets. Herein, we propose STFlow, a flow matching generative model that considers cell-cell interaction by modeling the joint distribution of gene expression of an entire slide. It also employs an efficient slide-level encoder with local spatial attention, enabling whole-slide processing without excessive memory overhead. On the recently curated HEST-1k and STImage-1K4M benchmarks, STFlow substantially outperforms state-of-the-art baselines and achieves over 18% relative improvements over the pathology foundation models.

Figures

Figures reproduced from arXiv: 2506.05361 by the authors.

Figure 1
Figure 1. An overview of gene expression prediction from histology image with STFlow. (a): The histology image is segmented into a set of spot images, each associated with a 2D coordinate and gene expression. Each spot image is then encoded using a pathology foundation model. (b): STFlow encodes the slide-level context by aggregating the neighboring spots through spatial attention and the interaction between cells is explicit… view at source ↗
Figure 2
Figure 2. Distribution of gene expression level on ST samples, with levels above 50 truncated for clarity. (Li et al., 2022; Biancalani et al., 2021; Cordell, 2009). However, the standard regression objective cannot model cell-cell interaction as it predicts gene expression in one go. To address this issue, we reformulate the gene expression regression model into a generative model, using samples from a prior distribution as … view at source ↗
Figure 3
Figure 3. The overall architecture of the proposed denoiser (a) and the E(2)-invariant spatial attention scheme (b). attention within the local neighbors of every spot. E(2)-Invariant Spatial Attention We introduce a spatial attention mechanism that generates spot representations in￾variant to E(2) operations, i.e., rotation, translation, and reflection, of the coordinates. To achieve this, we adapt frame averaging (FA), an E… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: STFlow for Biomarker Discovery in Breast Samples: (a) TENX141 and (b) TENX115. The top row of subfigures shows gene UB2EC, while the bottom row shows gene VWF. The Pearson correlation between ground truth and predictions is provided in each subfigure’s title. STFlow w/…
Figure 5
Figure 5. Figure 5: Efficiency comparison of slide-based models. et al., 2018). We utilize two datasets from HEST and report the average correlation for each gene across different cross￾validation folds. Specifically, the IDC dataset is used for GATA3 and ERBB2, while LUNG is used for UBE…
Figure 6
Figure 6. Figure 6: Two breast cancer examples of the refinement process performed by STFlow: (a) TENX141 and (b) TENX115. The top row of subfigures shows gene UB2EC, while the bottom row shows gene VWF [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: STFlow for Biomarker Discovery in Breast Samples (IDC dataset from HEST-1K). Hyperparameter study of ZINB distribution We set the zero-inflation probability π = 0.5, as higher dropout rates degrade ZINB to a near-zero distribution. We conducted an ablation over the mea…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

20 extracted references · 4 canonical work pages

  1. [1]

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

    arXiv

  2. [3]

    M., Bruna, J., Cohen, T., and Veli ˇckovi´c, P

    Bronstein, M. M., Bruna, J., Cohen, T., and Veli ˇckovi´c, P. Geometric deep learning: Grids, groups, graphs, geodesics, and gauges. arXiv preprint arXiv:2104.13478,

    arXiv

  3. [5]

    Se (3)-transformers: 3d roto-translation equivariant attention networks

    Fuchs, F., Worrall, D., Fischer, V ., and Welling, M. Se (3)-transformers: 3d roto-translation equivariant attention networks. Advances in neural information processing systems, 33:1970–1981,

    work page 1970

  4. [7]

    Dataset statistics of HEST-Bench. IDC PRAD PAAD SKCM COAD READ CCRCC HCC LUNG LYMPH Organ Breast Prostate Pancreas Skin Colon Rectum Kidney Liver Lung Axillary Lymph Nodes Technology Xenium Visium Xenium Xenium Visium Visium Visium Visium Xenium Visium #Patients 4 2 3 2 3 2 24 2 2 4 #Samples 4 23 3 2 6 4 24 2 2 4 #Splits 4 2 3 2 2 2 6 2 2 4 Avg. spots 492...

    work page 1909

  5. [8]

    Alphafold meets flow matching for generating protein ensembles

    10 Scalable Generation of Spatial Transcriptomics from Histology Images via Whole-Slide Flow Matching Jing, B., Berger, B., and Jaakkola, T. Alphafold meets flow matching for generating protein ensembles. arXiv preprint arXiv:2402.04845,

    arXiv

  6. [9]

    Kingma, D. P. and Ba, J. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980,

    arXiv

  7. [10]

    and Smidt, T

    Liao, Y .-L. and Smidt, T. Equiformer: Equivariant graph attention transformer for 3d atomistic graphs. arXiv preprint arXiv:2206.11990,

    arXiv

  8. [12]

    Symmetry-informed geometric representation for molecules, proteins, and crystalline materials

    Liu, S., Du, W., Li, Y ., Li, Z., Zheng, Z., Duan, C., Ma, Z., Yaghi, O., Anandkumar, A., Borgs, C., et al. Symmetry-informed geometric representation for molecules, proteins, and crystalline materials. arXiv preprint arXiv:2306.09375,

    arXiv

Show all 20 references
  1. [15]

    Leveraging information in spatial transcriptomics to predict super-resolution gene expression from histology images in tumors

    Pang, M., Su, K., and Li, M. Leveraging information in spatial transcriptomics to predict super-resolution gene expression from histology images in tumors. BioRxiv, pp. 2021–11,

    2021

  2. [16]

    J., and Lipman, Y

    Puny, O., Atzmon, M., Ben-Hamu, H., Misra, I., Grover, A., Smith, E. J., and Lipman, Y . Frame averaging for invariant and equivariant network design. arXiv preprint arXiv:2110.03336,

    arXiv

  3. [17]

    Artificial intelli- gence for science in quantum, atomistic, and continuum systems

    Zhang, X., Wang, L., Helwig, J., Luo, Y ., Fu, C., Xie, Y ., Liu, M., Lin, Y ., Xu, Z., Yan, K., et al. Artificial intelli- gence for science in quantum, atomistic, and continuum systems. arXiv preprint arXiv:2307.08423,

    arXiv

  4. [18]

    Diffusion generative modeling for spatially resolved gene expression inference from histology images

    Zhu, S., Zhu, Y ., Tao, M., and Qiu, P. Diffusion generative modeling for spatially resolved gene expression inference from histology images. arXiv preprint arXiv:2501.15598,

    arXiv

  5. [19]

    Dataset Table 7 lists the statistics of HEST-Bench

    B. Dataset Table 7 lists the statistics of HEST-Bench. Further details about these datasets can be found in (Jaume et al., 2024). Note that the COAD dataset differs from the version in (Jaume et al., 2024), as it was updated two months after the paper’s release. Table

    2024

  6. [2016]

    Protein-nucleic acid complex modeling with frame averaging transformer

    Huang, T., Song, Z., Ying, R., and Jin, W. Protein-nucleic acid complex modeling with frame averaging transformer. arXiv preprint arXiv:2406.09586,

    arXiv

  7. [2018]

    and Jin, W

    Nori, D. and Jin, W. Rnaflow: Rna structure & sequence design via inverse folding-based flow matching. arXiv preprint arXiv:2405.18768,

    arXiv

  8. [2019]

    Stimage-1k4m: A histopathology image-gene expres- sion dataset for spatial transcriptomics

    Chen, J., Zhou, M., Wu, W., Zhang, J., Li, Y ., and Li, D. Stimage-1k4m: A histopathology image-gene expres- sion dataset for spatial transcriptomics. arXiv preprint arXiv:2406.06393, 2024a. Chen, R. J., Ding, T., Lu, M. Y ., Williamson, D. F., Jaume, G., Song, A. H., Chen, B....

    arXiv

  9. [2021]

    How attentive are graph attention networks? arXiv preprint arXiv:2105.14491,

    Brody, S., Alon, U., and Yahav, E. How attentive are graph attention networks? arXiv preprint arXiv:2105.14491,

    arXiv

  10. [2022]

    T., Ben-Hamu, H., Nickel, M., and Le, M

    Lipman, Y ., Chen, R. T., Ben-Hamu, H., Nickel, M., and Le, M. Flow matching for generative modeling. arXiv preprint arXiv:2210.02747,

    arXiv

  11. [2023]

    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,

    arXiv

  12. [2024]

    Jia, Y ., Liu, J., Chen, L., Zhao, T., and Wang, Y

    URL https: //arxiv.org/abs/2406.16192v1. Jia, Y ., Liu, J., Chen, L., Zhao, T., and Wang, Y . Thi- togene: a deep learning method for predicting spatial transcriptomics from histological images. Briefings in Bioinformatics, 25(1):bbad464,

    arXiv

Pith tools

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