Pith. sign in

REVIEW 5 major objections 5 minor 53 references

FM4NPP: A Scaling Foundation Model for Nuclear and Particle Physics

T0 review · 5 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read A self-supervised foundation model, pretrained on 11 million simulated particle collisions and frozen, beats task-specific baselines in track finding, particle identification, and noise tagging with small adapters.

desk verdict A serious scaling-FM attempt for sparse detector data with fresh serialization and pretext, but the missing pretrain/test disjointness control and internal numeric inconsistencies undercut the headline claim. read the letter →

arxiv 2508.14087 v1 pith:JYLA3W7R submitted 2025-08-13 cs.LG cs.AIhep-ex

David Park , Shuhang Li , Yi Huang , Xihaier Luo , Haiwang Yu , Yeonju Go , Christopher Pinkenburg , Yuewei Lin
show 4 more authors
Shinjae Yoo Joseph Osborn Jin Huang Yihui Ren
This is my paper
classification cs.LGcs.AIhep-ex
keywords foundationmodelself-supervisedlearningparticletrackfindingidentificationnoisetaggingMamba2scalinglawssPHENIXTPC
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 a foundation model for experimental particle physics is possible: sparse, unordered 3D detector data can be serialized, pretrained self-supervised at scale, and then adapted to multiple physics tasks without retraining the backbone. It introduces a dataset of more than 11 million simulated sPHENIX proton-proton collisions and defines three downstream tasks: track finding, particle identification, and noise tagging. A frozen 188-million-parameter Mamba2 model, paired with adapters of under 3 million trainable parameters, outperforms GNN track finders, the 45M-parameter OneFormer3D segmentation model, and the sPHENIX Kalman-based reconstruction pipeline on the metrics reported. The paper further claims the learned representations are task-agnostic: raw embeddings show no track structure, but a single linear projection produces separable track clusters. A sympathetic reader would care because this suggests one pretrained model could serve many collider-reconstruction tasks and remain effective with little labeled data.

What carries the argument

The load-bearing machinery is the pair of a Hierarchical Raster Scan and a $k$-Next-Nearest-Neighbor prediction objective, running on a Mamba2 state-space backbone. The Hierarchical Raster Scan first partitions the TPC volume into 6 radial by 8 $\eta$ by 8 $\phi$ boxes, sorts spacepoints within each box by radius, then orders the boxes by the $(r,\phi,\eta)$ coordinates of their centers; this balances local track continuity with the global outward flow of particles. The self-supervised target predicts the coordinates of the $k$ nearest spacepoints at larger radius, making the objective geometric and independent of the serialization artifacts. Mamba2 provides linear-time sequence processing,

What would settle it

Train the same model and data with the $k$-Next-Nearest-Neighbor target retained but the Hierarchical Raster Scan replaced first by a random permutation and then by a simple radius-sorted ordering, keeping all hyperparameters fixed. If downstream track-finding ARI stays within about 1-2 points of the reported 0.9448 in either case, the serialization is not load-bearing. Then separately replace the target with $k$ random outer-radius spacepoints instead of the $k$ nearest ones; if downstream performance is again comparable, the nearest-neighbor geometry is not load-bearing.

Watch

Extended reading notes

Core claim

The central claim is that a sequence-model foundation model can learn generalizable detector physics from a purely geometric self-supervised objective. Spacepoints are mapped to cylindrical-polar coordinates $(r,\phi,\eta)$, partitioned into a $6\times 8\times 8$ grid, and serialized by a two-level Hierarchical Raster Scan before being processed by a Mamba2 backbone. The pretraining target is $k$-Next-Nearest-Neighbor prediction: for each spacepoint, predict the coordinates of the $k$ spatially nearest spacepoints that lie at larger radius, so the task tracks outward particle propagation and is not tied to the arbitrary 1D sequence order. After pretraining the backbone weights are frozen. Tr

Load-bearing premise

The load-bearing premise is that the Hierarchical Raster Scan ordering and the $k$-Next-Nearest-Neighbor objective preserve enough track structure for a Mamba2 sequence model to learn generalizable physics; the paper does not ablate either choice, so if a different serialization or pretext gives the same downstream results, the claim that the FM specifically learns task-agnostic physics is unsupported.

Editorial extensions

If this is right

  • One frozen pretrained backbone can serve multiple downstream tasks: track finding, particle identification, and noise tagging are all handled by training only small adapters, so the expensive pretraining is amortized over many applications.
  • Pretraining on unlabeled simulated detector data scales predictably: power-law improvements in model size, data, and compute mean the same recipe can be projected to larger budgets, with the observed plateau at 188M parameters marking a saturation point worth studying.
  • The FM matters most where labels are scarce: relative to adapter-only training it gives a 2.9x gain in track-finding ARI at low label counts versus 1.3x at high label counts, suggesting pretraining sharply reduces the need for labeled tracking data.
  • Because a single linear projection on frozen embeddings creates separable track clusters, simple and cheap adapter heads are sufficient in deployment, which is why the method can beat much larger task-specific models with a fraction of the trainable parameters.
  • The order-independent, geometry-defined pretraining objective is a template that could be transferred to other detectors or collider geometries without redesigning the self-supervised task for each one.

Reading between the lines

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

  • The paper does not ablate the serialization scheme or the pretraining objective. A direct test is whether replacing the Hierarchical Raster Scan with a random permutation or a space-filling curve, keeping all else fixed, preserves downstream performance; if so, the specific physical ordering is less central than the scale of pretraining itself.
  • The claim that a single linear mapping specializes the representations is demonstrated mainly through visual dimensionality-reduction plots for track finding. A quantitative linear-probe benchmark across all three tasks would be a sharper test of the task-agnostic claim.
  • The dataset covers proton-proton collisions at 200 GeV with a mean of about 856 spacepoints per event. The most consequential stress test is high-occupancy heavy-ion data, where sequence lengths and track densities are far higher and the raster-scan ordering may behave very differently.
  • If the representation is truly task-agnostic, the same frozen backbone may extend beyond TPC tracking to calorimeter or silicon-tracker data, as the authors hint in their future-work discussion, but that extension is not yet demonstrated.
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

5 major / 5 minor

Summary. The paper proposes FM4NPP, a Mamba2-based foundation model for sparse particle-detector data. Spacepoints from sPHENIX TPC simulations are serialized with a proposed Hierarchical Raster Scan, and the model is pretrained self-supervised with a k-Next-Nearest-Neighbor objective on 11M+ simulated p+p events. With frozen FM weights and lightweight adapters, the authors report performance on track finding, particle identification, and noise tagging, claim consistent gains over GNN baselines and OneFormer3D, and report scaling-law behavior in model size, data size, and compute. The paper also introduces labeled downstream tasks and argues that the learned representations are task-agnostic and can be specialized by a single linear projection.

Significance. If the empirical claims hold, this would be a valuable benchmark contribution: it provides a large simulated dataset, a practical frozen-FM transfer recipe, and the first demonstration of 188M-parameter pretraining for detector data, substantially larger than prior specialized track-reconstruction networks. The scaling-law and data-efficiency results are also of interest to the ML-for-physics community. However, the significance is currently undercut by unresolved data-leakage control and by internally inconsistent quantitative reporting in the main text versus appendices. The strengths are the scale of the pretraining corpus, the realistic simulation provenance, and the evaluation across three complementary downstream tasks; the weaknesses are experimental-control omissions and overstatement of the 'consistently outperforms' claim.

major comments (5)
  1. [Particle Detector Dataset / Downstream Tasks] The paper never states that the downstream evaluation events are disjoint from the 11M-event pretraining corpus. The pretraining objective predicts k-next-nearest-neighbor coordinates for every spacepoint in an event, so if any test event appeared during pretraining, the frozen FM has already been optimized on the exact geometry of that event. This would inflate downstream metrics relative to baselines such as EggNet, Exa.TrkX, SAGEConv, and OneFormer3D, which see only labeled training events. Please specify the event-level train/test split, state whether the labeled downstream events are a subset of the pretraining corpus, and confirm that no pretraining event is used for downstream evaluation. This is load-bearing for the central transfer-learning claim.
  2. [Table 2 vs. Tables 3-5] The same model is reported with different numbers in the main text and appendices. For FM4NPP(m6), PID accuracy is 0.9039 in Table 2 but 0.8547 in Table 4; noise-tagging accuracy is 0.9713 in Table 2 but 0.9662 in Table 3, with recall 0.9367 vs. 0.9122 and precision 0.9190 vs. 0.9114. Track-finding efficiency/purity in Table 2 (96.08%/93.08%) also differ from the 'overall spacepoint efficiency/purity' in Table 5 (97.56%/98.34%). The authors need to reconcile these numbers and precisely define each metric; as printed, the quantitative claims are not reliable.
  3. [Abstract / Introduction / Table 2] The claim that the FM 'consistently outperforms baseline models across all downstream tasks' is contradicted by the authors' own Table 2 on noise tagging: FM4NPP(m6) recall is 0.9367, lower than OneFormer3D's 0.9404. The main text later correctly says performance is 'similar' to OneFormer3D. The abstract and introduction should be revised to report per-task, per-metric superiority, or the comparison should be accompanied by statistical error bars/ tests.
  4. [Methodology (Self-supervised Pretraining Objectives)] The novelty of the method rests on the Hierarchical Raster Scan serialization and the k-Next-Nearest-Neighbor objective, but no ablation is provided for either component. For example, replacing the HRS ordering with a radial sort or a space-filling curve, or replacing the k-next-neighbor objective with a simpler next-token objective, could yield similar downstream performance. Without such ablations, the specific claim that the proposed serialization and pretext are what make the representations task-agnostic is unsupported. The AdapterOnly comparison shows pretraining helps, but not that these particular design choices are necessary or beneficial.
  5. [Insights about FM Adaptation (Q3, Figure 8)] The task-agnostic claim is supported only by qualitative PCA/t-SNE visualizations (Figures 8, 16-18). The linear projection used in the right panels is trained on downstream labels, so the emergence of track-separated clusters after projection is expected and does not by itself demonstrate that the raw FM features are task-agnostic. Please add a quantitative probe, e.g., linear-probe accuracy or a clustering metric, and compare with a randomly initialized encoder or a non-pretrained baseline to substantiate the 'single linear mapping' conclusion.
minor comments (5)
  1. [Appendix B (Loss Re-scaling by Event Difficulty)] The event-difficulty bins and weights w_g(i) are not specified. For reproducibility, please provide the number of bins, bin boundaries, and the reweighting formula. Also specify the number of Mamba layers L and the depth of the adapters, which are not given in Table 1.
  2. [Figure 5] The scaling-law figures show no fitted exponents, no confidence intervals, and no goodness-of-fit values. The statement that performance 'plateaus at m6' appears based on one or two points; please add fit parameters and uncertainties.
  3. [Appendix C (Adapt GNNs)] There are broken internal references: 'the same data pre-processing protocol as discussed in .' and similar dangling references. The Pyzer-Knapp reference is also duplicated (2025a and 2025b appear to be the same paper with different journal capitalization).
  4. [Experiments and Results] No test-set size, number of runs, or seed information is provided, so it is unclear whether the reported differences are statistically significant. Please report the number of test events and, ideally, error bars or multiple-seed results.
  5. [Table 5] In Table 5 the FM4NPP parameter entry is listed as '188M + 2.39M', while the main text reports 2.39M trainable parameters. Please clarify whether the table reports total frozen parameters or trainable parameters, and keep the convention consistent.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: pretraining is self-supervised on geometry only, downstream adapters are trained on separate labeled data, and no fitted parameter is repackaged as a prediction.

full rationale

The paper's central claim is that a frozen, self-supervised FM pretrained on 11M+ simulated sPHENIX events yields transferable representations that outperform task-specific baselines after lightweight adaptation. Tracing the derivation chain: the pretraining objective (k-Next-Nearest-Neighbor prediction) is defined purely in terms of 3D spacepoint geometry and deposited energy, not in terms of any downstream label (track ID, particle species, or noise flag). The downstream adapters are trained on labeled datasets with their own loss functions (Hungarian matching with Dice/Focal/classification losses for track finding; cross-entropy for PID and noise tagging), and the FM weights are frozen during this stage. There is no equation in which a downstream metric is defined as the pretraining loss, and no fitted parameter from the pretraining stage is renamed as a downstream prediction. The scaling-law exponents are descriptive summaries of validation MSE versus model size/data/compute, not used to claim predictive power for downstream tasks. The comparison against sPHENIX's ACTS-based pipeline cites Osborn et al. (2021), which shares an author with this paper, but that citation is used only to describe the baseline algorithm, not to justify the FM's superiority; the efficiency numbers are measured independently. The only substantive concern is that the paper does not explicitly state that the downstream evaluation events are disjoint from the 11M-event pretraining corpus. If they overlapped, the frozen FM could have been exposed to test-event structure, inflating apparent transfer. However, this is a missing experimental control or potential data-leakage issue, not a circular derivation: the pretraining objective and downstream labels are not definitionally linked, and no quoted equation or construction makes the outcome equivalent to the input. Therefore, no circularity step meets the evidentiary bar.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

No new physical entities are introduced. The main extra assumptions are about the simulation being representative, the serialization being a sensible inductive bias, and the pretext loss being predictive of downstream quality. Several preprocessing choices (grid bins, k, difficulty weights, encoding frequencies) are free parameters not justified by sensitivity analysis.

free parameters (4)
  • Grid binning (6 radial, 8 eta, 8 phi bins) = 6,8,8
    Chosen by hand; radial bins aligned with detector layers, eta/phi bins use frequency-based binning on 50k events. Determines serialization structure.
  • k (number of next-nearest neighbors to predict) = 10
    Stated as k=10 in the architecture section; no sensitivity study provided.
  • Event-difficulty rescaling weights w_g = Not reported
    Events binned by mean k-neighborhood distance and loss re-weighted by a per-bin factor; the weights are data-derived and not reported.
  • NeRF positional encoding max frequency L = Not stated
    The frequency ladder of the positional encoding is not specified; an implementation detail that could affect results.
assumptions (5)
  • domain assumption Mamba2 selective SSM (structured state space duality) correctly models sequential dependencies in serialized spacepoints
    The paper adopts Mamba2 as backbone without comparative validation against Transformers or other sequence models on this data.
  • domain assumption The detector simulation chain (PYTHIA8 'Detroit' tune, GEANT4 with FTFP_BERT_HP, full electronics emulation) faithfully reproduces the spacepoint distribution and labels of real sPHENIX data
    All training and evaluation uses simulation; no real-data validation. The paper relies on the sPHENIX production simulation being representative.
  • domain assumption Downstream task definitions (double-majority track matching, 60 MeV/c noise threshold, 5-class PID grouping) are the right operationalizations for physics performance
    For example, noise is tagged by p < 60 MeV/c, an operational proxy; different thresholds may change results.
  • domain assumption Scaling laws (power-law decrease of validation MSE with model/data/compute) generalize from these few runs to the full regime
    The paper fits scaling trends visually without reporting fit parameters, uncertainties, or multiple seeds.
  • ad hoc to paper Validation MSE on the k-next-neighbor pretext is a proxy for downstream task performance
    The scaling analysis optimizes pretext loss; the paper does not show that pretext loss correlates with downstream ARI/accuracy.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FM4NPP: A Scaling Foundation Model for Nuclear and Particle Physics." pith.science (2026). https://pith.science/paper/JYLA3W7R

@misc{pith2026250814087,
  author       = {Pith},
  title        = {Pith review of: FM4NPP: A Scaling Foundation Model for Nuclear and Particle Physics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JYLA3W7R}},
  note         = {Machine review of arXiv:2508.14087}
}
read the original abstract

Large language models have revolutionized artificial intelligence by enabling large, generalizable models trained through self-supervision. This paradigm has inspired the development of scientific foundation models (FMs). However, applying this capability to experimental particle physics is challenging due to the sparse, spatially distributed nature of detector data, which differs dramatically from natural language. This work addresses if an FM for particle physics can scale and generalize across diverse tasks. We introduce a new dataset with more than 11 million particle collision events and a suite of downstream tasks and labeled data for evaluation. We propose a novel self-supervised training method for detector data and demonstrate its neural scalability with models that feature up to 188 million parameters. With frozen weights and task-specific adapters, this FM consistently outperforms baseline models across all downstream tasks. The performance also exhibits robust data-efficient adaptation. Further analysis reveals that the representations extracted by the FM are task-agnostic but can be specialized via a single linear mapping for different downstream tasks.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

53 extracted references · 37 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Agostinelli, S.; et al. 2003. GEANT4 - A Simulation Toolkit . Nucl. Instrum. Meth. A, 506: 250--303

    work page 2003

  4. [4]

    R.; Chang, Z.; Elayavalli, R

    Aguilar, M. R.; Chang, Z.; Elayavalli, R. K.; Fatemi, R.; He, Y.; Ji, Y.; Kalinkin, D.; Kelsey, M.; Mooney, I.; and Verkest, V. 2022. pythia8 underlying event tune for RHIC energies . Phys. Rev. D, 105(1): 016011

    work page 2022

  5. [5]

    R.; Finnie, L.; Finnie, N.; Germain, C.; Gligorov, V

    Amrouche, S.; Basara, L.; Calafiura, P.; Estrade, V.; Farrell, S.; Ferreira, D. R.; Finnie, L.; Finnie, N.; Germain, C.; Gligorov, V. V.; Golling, T.; Gorbunov, S.; Gray, H.; Guyon, I.; Hushchyn, M.; Innocente, V.; Kiehn, M.; Moyse, E.; Puget, J.-F.; Reina, Y.; Rousseau, D.; Salzburger, A.; Ustyuzhanin, A.; Vlimant, J.-R.; Wind, J. S.; Xylouris, T.; and Y...

    work page 2020

  6. [6]

    Battaglia, P.; Pascanu, R.; Lai, M.; Jimenez Rezende, D.; et al. 2016. Interaction networks for learning about objects, relations and physics. Advances in neural information processing systems, 29

    work page 2016

  7. [7]

    Belmont, R.; et al. 2024. Predictions for the sPHENIX physics program . Nucl. Phys. A, 1043: 122821

    work page 2024

  8. [8]

    P.; Lucic, A.; Stanley, M.; Allen, A.; Brandstetter, J.; Garvan, P.; Riechert, M.; Weyn, J

    Bodnar, C.; Bruinsma, W. P.; Lucic, A.; Stanley, M.; Allen, A.; Brandstetter, J.; Garvan, P.; Riechert, M.; Weyn, J. A.; Dong, H.; et al. 2025. A foundation model for the Earth system. Nature, 1--8

    work page 2025

Show all 53 references
  1. [9]

    A.; Adeli, E.; Altman, R.; Arora, S.; von Arx, S.; Bernstein, M

    Bommasani, R.; Hudson, D. A.; Adeli, E.; Altman, R.; Arora, S.; von Arx, S.; Bernstein, M. S.; Bohg, J.; Bosselut, A.; Brunskill, E.; et al. 2021. On the opportunities and risks of foundation models. arXiv preprint arXiv:2108.07258

    2021 arXiv

  2. [10]

    Brookhaven National Laboratory . 2025. sPHENIX Detector at RHIC . https://www.bnl.gov/rhic/sphenix.php. Accessed: 2025-07-19

    2025

  3. [11]

    Busza, W.; Rajagopal, K.; and van der Schee, W. 2018. Heavy Ion Collisions: The Big Picture, and the Big Questions . Ann. Rev. Nucl. Part. Sci., 68: 339--376

    2018

  4. [12]

    Calafiura, P.; Chan, J.; Delabrouille, L.; and Wang, B. 2024. EggNet: An Evolving Graph-based Graph Attention Network for Particle Track Reconstruction. arXiv:2407.13925

    2024 arXiv

  5. [13]

    V.; Hushchyn, M.; and Ustyuzhanin, A

    Calafiura, P.; Farrell, S.; Gray, H.; Vlimant, J.-R.; Innocente, V.; Salzburger, A.; Amrouche, S.; Golling, T.; Kiehn, M.; Estrade, V.; Germaint, C.; Guyon, I.; Moyse, E.; Rousseau, D.; Yilmaz, Y.; Gligorov, V. V.; Hushchyn, M.; and Ustyuzhanin, A. 2018. TrackML: A High Energy...

    2018

  6. [14]

    G.; Kirillov, A.; and Girdhar, R

    Cheng, B.; Misra, I.; Schwing, A. G.; Kirillov, A.; and Girdhar, R. 2022. Masked-attention Mask Transformer for Universal Image Segmentation. arXiv:2112.01527

    2022 arXiv

  7. [15]

    G.; and Kirillov, A

    Cheng, B.; Schwing, A. G.; and Kirillov, A. 2021. Per-Pixel Classification is Not All You Need for Semantic Segmentation. arXiv:2107.06278

    2021 arXiv

  8. [16]

    Collaboration, A.; Aad, G.; et al. 2012. Observation of a New Particle in the Search for the Standard Model Higgs Boson with the ATLAS Detector at the LHC. Physics Letters B, 716(1): 1--29

    2012

  9. [17]

    Dao, T.; and Gu, A. 2024. Transformers are SSM s: Generalized Models and Efficient Algorithms Through Structured State Space Duality. In Forty-first International Conference on Machine Learning

    2024

  10. [18]

    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. 2021. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. In International Confere...

    2021

  11. [19]

    Fedus, W.; Zoph, B.; and Shazeer, N. 2022. Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity. Journal of Machine Learning Research, 23(120): 1--39

    2022

  12. [20]

    Gu, A.; and Dao, T. 2023. Mamba: Linear-time sequence modeling with selective state spaces. arXiv preprint arXiv:2312.00752

    2023 arXiv

  13. [21]

    Gu, A.; and Dao, T. 2024. Mamba: Linear-Time Sequence Modeling with Selective State Spaces. In First Conference on Language Modeling

    2024

  14. [22]

    Hoffmann, J.; Borgeaud, S.; Mensch, A.; Buchatskaya, E.; Cai, T.; Rutherford, E.; Casas, D. d. L.; Hendricks, L. A.; Welbl, J.; Clark, A.; et al. 2022. Training compute-optimal large language models. arXiv preprint arXiv:2203.15556

    2022 arXiv

  15. [23]

    Hubert, L.; and Arabie, P. 1985. Comparing partitions. Journal of Classification, 2(1): 193--218

    1985

  16. [24]

    Jiang, C.; and Qian, S. 2025. Application of Structured State Space Models to High energy physics with locality sensitive hashing. In International Conference on Artificial Intelligence and Statistics, 3961--3969. PMLR

    2025

  17. [25]

    Ju, X.; Murnane, D.; Calafiura, P.; Choma, N.; Conlon, S.; Farrell, S.; Xu, Y.; Spiropulu, M.; Vlimant, J.-R.; Aurisano, A.; Hewes, J.; Cerati, G.; Gray, L.; Klijnsma, T.; Kowalkowski, J.; Atkinson, M.; Neubauer, M.; DeZoort, G.; Thais, S.; Chauhan, A.; Schuy, A.; Hsu, S.-C.; ...

    2021

  18. [26]

    Kalman, R. E. 1960. A New Approach to Linear Filtering and Prediction Problems. Journal of Basic Engineering, 82(1): 35--45

    1960

  19. [27]

    B.; Chess, B.; Child, R.; Gray, S.; Radford, A.; Wu, J.; and Amodei, D

    Kaplan, J.; McCandlish, S.; Henighan, T.; Brown, T. B.; Chess, B.; Child, R.; Gray, S.; Radford, A.; Wu, J.; and Amodei, D. 2020. Scaling laws for neural language models. arXiv preprint arXiv:2001.08361

    2020 arXiv

  20. [28]

    Klest, H. 2020. Overview and design of the sPHENIX TPC . J. Phys. Conf. Ser., 1498: 012025

    2020

  21. [29]

    Kolodiazhnyi, M.; Vorontsova, A.; Konushin, A.; and Rukhovich, D. 2024. Oneformer3d: One transformer for unified point cloud segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 20943--20953

    2024

  22. [30]

    Li, Q.; Hu, Z.; Wang, Y.; Li, L.; Fan, Y.; King, I.; Song, L.; and Li, Y. 2024. Progress and Opportunities of Foundation Models in Bioinformatics. Briefings in Bioinformatics. Survey of foundation models in bioinformatics

    2024

  23. [31]

    Loshchilov, I.; and Hutter, F. 2017. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101

    2017 arXiv

  24. [32]

    Ma, Q.; Jiang, Y.; Cheng, H.; and Xu, D. 2024. Harnessing the deep learning power of foundation models in single-cell omics. Nature Reviews Molecular Cell Biology, 25(8): 593--594

    2024

  25. [33]

    P.; Tancik, M.; Barron, J

    Mildenhall, B.; Srinivasan, P. P.; Tancik, M.; Barron, J. T.; Ramamoorthi, R.; and Ng, R. 2021. Nerf: Representing scenes as neural radiance fields for view synthesis. Communications of the ACM, 65(1): 99--106

    2021

  26. [34]

    Moskowitz, C. 2023. Tiny Bubbles of Primordial Soup Re‑create Early Universe. Scientific American, 328(3). Accessed: 2025‑07‑19

    2023

  27. [35]

    K.; and Grover, A

    Nguyen, T.; Brandstetter, J.; Kapoor, A.; Gupta, J. K.; and Grover, A. 2023. ClimaX: a25 foundation model for weather and climate. In Proceedings of the 40th International Conference on Machine Learning, ICML'23. JMLR.org

    2023

  28. [36]

    A.; Pennington, J.; and Sohl-Dickstein, J

    Novak, R.; Bahri, Y.; Abolafia, D. A.; Pennington, J.; and Sohl-Dickstein, J. 2018. Sensitivity and Generalization in Neural Networks: an Empirical Study. In International Conference on Learning Representations

    2018

  29. [37]

    D.; Frawley, A

    Osborn, J. D.; Frawley, A. D.; Huang, J.; Lee, S.; Da Costa, H. P.; Peters, M.; Pinkenburg, C.; Roland, C.; and Yu, H. 2021. Implementation of ACTS into sPHENIX Track Reconstruction . Comput. Softw. Big Sci., 5(1): 23

    2021

  30. [38]

    Paganini, M. 2018. Machine Learning Algorithms for b -Jet Tagging at the ATLAS Experiment . J. Phys. Conf. Ser., 1085(4): 042031

    2018

  31. [39]

    O.; Manica, M.; Staar, P.; Morin, L.; Ruch, P.; Laino, T.; Smith, J

    Pyzer-Knapp, E. O.; Manica, M.; Staar, P.; Morin, L.; Ruch, P.; Laino, T.; Smith, J. R.; and Curioni, A. 2025 a . Foundation models for materials discovery--current state and future directions. Npj Computational Materials, 11(1): 61

    2025

  32. [40]

    O.; Manica, M.; Staar, P.; Morin, L.; Ruch, P.; Laino, T.; Smith, J

    Pyzer-Knapp, E. O.; Manica, M.; Staar, P.; Morin, L.; Ruch, P.; Laino, T.; Smith, J. R.; and Curioni, A. 2025 b . Foundation models for materials discovery – current state and future directions. npj Computational Materials, 11: 15

    2025

  33. [41]

    R.; Kieseler, J.; Iiyama, Y.; and Pierini, M

    Qasim, S. R.; Kieseler, J.; Iiyama, Y.; and Pierini, M. 2019. Learning representations of irregular particle-detector geometry with distance-weighted graph networks. The European Physical Journal C, 79(7): 1--11

    2019

  34. [42]

    K.; Bronstein, M

    Rusch, T. K.; Bronstein, M. M.; and Mishra, S. 2023. A survey on oversmoothing in graph neural networks. arXiv preprint arXiv:2303.10993

    2023 arXiv

  35. [43]

    R.; Corke, R.; Desai, N.; Ilten, P.; Mrenna, S.; Prestel, S.; Rasmussen, C

    Sj\"ostrand, T.; Ask, S.; Christiansen, J. R.; Corke, R.; Desai, N.; Ilten, P.; Mrenna, S.; Prestel, S.; Rasmussen, C. O.; and Skands, P. Z. 2015. An introduction to PYTHIA 8.2 . Comput. Phys. Commun., 191: 159--177

    2015

  36. [44]

    sPHENIX Collaboration . 2025 a . acts : A common tracking software toolkit for sPHENIX, tagged at 33fc284f238a24405bcd6c2de3260f370d6f8403 . https://github.com/sPHENIX-Collaboration/acts/tree/33fc284f238a24405bcd6c2de3260f370d6f8403. Accessed August 3, 2025; part of the offici...

    2025

  37. [45]

    sPHENIX Collaboration . 2025 b . calibrations : Calibration code for sPHENIX, tagged at a3e66e69635514813ee3e20bf18b2bd59787b503 . https://github.com/sPHENIX-Collaboration/calibrations/tree/a3e66e69635514813ee3e20bf18b2bd59787b503. Accessed August 3, 2025; part of the official...

    2025

  38. [46]

    sPHENIX Collaboration . 2025 c . coresoftware : Core simulation and reconstruction software for the sPHENIX experiment, tagged at b849eba5c2cf8ada510d036aa9b9499cb31f0513 . https://github.com/sPHENIX-Collaboration/coresoftware/tree/b849eba5c2cf8ada510d036aa9b9499cb31f0513. Acc...

    2025

  39. [47]

    sPHENIX Collaboration . 2025 d . macros : Analysis and utility macros for sPHENIX, tagged at 661f781db23352a3fa72055ce7dbf1a0ee1c2167 . https://github.com/sPHENIX-Collaboration/macros/tree/661f781db23352a3fa72055ce7dbf1a0ee1c2167. Accessed August 3, 2025; part of the official ...

    2025

  40. [48]

    V.; Duckett, P.; Hart, M.; Pond, N.; Rettie, S.; Facini, G.; and Scanlon, T

    Stroud, S. V.; Duckett, P.; Hart, M.; Pond, N.; Rettie, S.; Facini, G.; and Scanlon, T. 2024. Transformers for Charged Particle Track Reconstruction in High Energy Physics. arXiv:2411.07149

    2024

  41. [49]

    C.; Xu, J.; Haas, M.; and Cevher, V

    Vankadara, L. C.; Xu, J.; Haas, M.; and Cevher, V. 2024. On feature learning in structured state space models. In The Thirty-eighth Annual Conference on Neural Information Processing Systems

    2024

  42. [50]

    N.; Kaiser, .; and Polosukhin, I

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, .; and Polosukhin, I. 2017. Attention is all you need. Advances in neural information processing systems, 30

    2017

  43. [51]

    Wang, W.; Dai, J.; Chen, Z.; Huang, Z.; Li, Z.; Zhu, X.; Hu, X.; and et al. 2023. InternImage: Exploring Large-Scale Vision Foundation Models With Deformable Convolutions. In CVPR 2023, 3591--3600

    2023

  44. [52]

    J.; Babuschkin, I.; Sidor, S.; Liu, X.; Farhi, D.; Ryder, N.; Pachocki, J.; Chen, W.; and Gao, J

    Yang, G.; Hu, E. J.; Babuschkin, I.; Sidor, S.; Liu, X.; Farhi, D.; Ryder, N.; Pachocki, J.; Chen, W.; and Gao, J. 2022. Tensor programs v: Tuning large neural networks via zero-shot hyperparameter transfer. arXiv preprint arXiv:2203.03466

    2022 arXiv

  45. [53]

    A.; Wagner, S

    Zhou, Y.; Chia, M. A.; Wagner, S. K.; Ayhan, M. S.; Williamson, D. J.; Struyven, R. R.; Liu, T.; Xu, M.; Lozano, M. G.; Woodward-Court, P.; et al. 2023. A foundation model for generalizable disease detection from retinal images. Nature, 622(7981): 156--163

    2023

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.