Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

Generating particle physics Lagrangians with transformers

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A transformer trained on field lists can write gauge-invariant particle physics Lagrangians, with over 90 percent accuracy on held-out examples.

desk verdict A real first step in transformer-based Lagrangian generation, with a circularity problem that is honestly acknowledged and should not be fatal, but the physics claims need to be framed as generator imitation until independently validated. read the letter →

arxiv 2501.09729 v1 pith:BYTPIMFS submitted 2025-01-16 cs.LG cs.SChep-phhep-th

Yong Sheng Koay , Rikard Enberg , Stefano Moretti , Eliel Camargo-Molina This is my paper · ORCID
classification cs.LGcs.SChep-phhep-th
keywords transformerLagrangiangenerationgaugeinvarianceStandardModelsymbolicmathematicsembeddinganalysisout-of-distributiongeneralizationparticlephysics
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper asks whether a transformer, the architecture behind modern language models, can learn the rules for constructing particle physics Lagrangians: the equations that encode how fields interact while respecting the Standard Model gauge symmetries $\mathrm{SU}(3)\times\mathrm{SU}(2)\times\mathrm{U}(1)$. The authors train a 357-million-parameter transformer on roughly 280,000 sampled field-list/Lagrangian pairs (plus a uniform dataset for comparison) and report that on held-out lists with up to six matter fields it writes the complete Lagrangian correctly more than 90 percent of the time, including contraction structure. Beyond the training range, for seven to ten fields, more than 99 percent of its outputs remain structurally reasonable, though exact hypercharge conservation degrades. An analysis of the model's internal embeddings shows that it has organized fields by spin and gauge representation and encodes conjugation as a consistent vector direction. If these results hold, they suggest that symmetry-aware symbolic reasoning can be learned from examples, a step toward automated exploration of beyond-Standard-Model theories.

What carries the argument

The load-bearing machinery is a tokenization scheme that turns each field into a sequence of tokens for its spin, its $\mathrm{SU}(3)$ and $\mathrm{SU}(2)$ representations, its $\mathrm{U}(1)$ hypercharge, its helicity, and an identity token used to encode index contractions; derivatives carry the gauge groups they couple to, and each term is followed by explicit contraction information. This token stream is fed to a 357-million-parameter BART-style transformer, a bidirectional encoder paired with an autoregressive decoder, trained with cross-entropy loss on about 280,000 Lagrangians produced by an automated effective-field-theory operator generator plus custom code for kinetic and mass terms. The trained model is then probed by projecting encoder embeddings with t-SNE and by comparing vector offsets between conjugate fields.

What would settle it

Take a field content whose complete gauge-invariant operator basis is known independently from a group-theoretic calculation and check whether the generator used for training misses any allowed term; if it does, the transformer trained on that generator will never output the missing term, showing that its Lagrangian-writing ability is bounded by the generator.

Watch

Extended reading notes

Core claim

The paper's central claim is that a transformer, trained purely on input-output pairs of field lists and Lagrangians, can reproduce the gauge-invariance rules of the Standard Model group $\mathrm{SU}(3)\times\mathrm{SU}(2)\times\mathrm{U}(1)$ well enough to write the full Lagrangian for a given field content. On held-out test Lagrangians with up to six matter fields, more than 90 percent of predictions match the reference Lagrangian term-for-term under a score that includes correct contractions and penalizes extra terms; for field contents with seven to ten fields, beyond the training range, over 99 percent of outputs are structurally reasonable, though $\mathrm{U}(1)$ hypercharge conservation gradually degrades. The paper also claims the model has internalized physics concepts rather than memorized strings: embedding analysis shows fields clustered by spin and gauge representation, with conjugation represented as a nearly constant vector offset in embedding space.

Load-bearing premise

The evaluation assumes that the automated generator behind all training and in-distribution test labels produces the complete, correct set of gauge-invariant terms; if that generator has blind spots, the transformer's high scores show fidelity to the generator rather than mastery of particle physics.

Editorial extensions

If this is right

  • A learned sequence model can reproduce gauge-invariance constraints from examples: held-out Lagrangians up to six matter fields are generated with over 90 percent exact Lagrangian score, and almost all mistakes are missing or extra terms rather than malformed ones.
  • The model's ability to write structurally reasonable Lagrangians survives beyond its training range: for seven to ten input fields more than 99 percent of outputs are reasonable, although U(1) conservation drops from about 91 to 87 percent.
  • Training-data design matters: a dataset enriched in short examples and trilinear interactions matches or beats a uniformly sampled dataset on complex terms, supporting the train-set-priming idea for symbolic tasks.
  • The embedding analysis indicates that symmetry structure and conjugation are emergent internal representations: fields cluster by spin and gauge representation, and conjugate pairs share a common vector offset.
  • Known failure modes point to concrete next steps: counting limitations in bidirectional encoders cause missed terms at high field numbers, and text-based number tokens cause fragile hypercharge arithmetic on out-of-distribution fractions.

Reading between the lines

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

  • Beyond the paper: if the embedding geometry truly tracks gauge representations, one could use the released model to probe which representation pairs it treats as equivalent and to detect where its notion of conjugation silently fails.
  • Beyond the paper: because all training and in-distribution labels come from a single automated generator, the reported accuracies are upper bounds on physical correctness; validating that generator on a few small field contents against an independent operator-basis computation would settle how much of the score is physics versus imitation.
  • Beyond the paper: the failure on non-minimal fractions suggests that replacing tokenized rationals with a continuous number encoding, as explored in other symbolic-mathematics transformers, could substantially improve out-of-distribution hypercharge conservation.
  • Beyond the paper: the tokenization is representation-agnostic enough that the same pipeline could in principle be pointed at larger gauge groups or higher-dimensional EFT operators, though the paper only argues this is easy and does not demonstrate it.
Share X Bluesky LinkedIn Reddit HN

Formalized claims in Lean

  1. Claim #1: The paper's central claim is that a transformer, trained purely on input-output pairs of field lists and Lagrangians, can reproduce the gauge-invariance rules of the Standard Model group $\mathrm{SU}(3)\times\mathrm{SU}(2)\times\mathrm{U}(1)$ well enough to write the full Lagrangian for a given field content. On held-out test Lagrangians with up to six matter fields, more than 90 percent of predic

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 trains a 357M-parameter BART transformer to translate a list of particle fields, with their Standard Model gauge quantum numbers, into a Lagrangian. The training and test data are generated by AutoEFT for interaction terms plus custom code for kinetic and mass terms, using a custom tokenization that encodes fields, derivatives, commutators, and index-contraction information. The authors report over 90% exact Lagrangian scores on held-out in-distribution Lagrangians, study out-of-distribution generalization to 7-10 fields and non-minimal U(1) charge representations, and analyze encoder embeddings for symmetry clusters and a conjugation axis. They release the model, datasets, and an interactive demo, and conclude in Section 8 that transformers are capable of writing the corresponding particle physics Lagrangian.

Significance. If the results hold, this is a useful benchmark for symbolic generation in theoretical physics, and the released model, datasets, and demo are valuable community assets. The paper is honest about the Standard Model shortfall (best score 0.77 in Table 13) and about the degradation of out-of-distribution performance, and it defines explicit evaluation metrics. However, the central validation is circular: accuracy is measured against the same generator that produced the labels, and the OOD "reasonable" check does not test full gauge invariance, so the physical interpretation of the high scores needs independent support before the Section 8 claim can be accepted at face value.

major comments (3)
  1. [§3.1, §4, §8] The central claim in §8 that the model is capable of writing the corresponding particle physics Lagrangian rests on the unstated premise that the AutoEFT-plus-custom-code generator of §3.1 produces exactly the correct set of gauge-invariant, Lorentz-invariant, renormalizable terms for each field content. Since every training label and every in-distribution test label is that generator's output, the >90% scores in Table 3 quantify agreement with the generator, not with an independent standard of physical correctness. Please validate the generator on a sample of field contents against an independent enumeration (hand calculation or a second package such as Sarah or Sym2Int), and in Appendix D explicitly compare the generator's Standard Model Lagrangian with the known one-generation SM Lagrangian, including the six Yukawa terms the model is said to miss. The Section 1 acknowledgment of the circularity is not a substitute for this validation.
  2. [§5, Table 4] The OOD reasonable-Lagrangian criterion in §5 checks syntax, quantum-number plausibility, mass dimension, valid contracted indices, completeness, commutator pairing, and U(1) conservation, but it does not check full SU(3)×SU(2) gauge invariance. A predicted term with a forbidden SU(2) contraction or a color combination containing no singlet would still be counted as reasonable, so the >99% figures in Table 4 overstate the OOD generalization claim. Please re-score the OOD predictions with a full invariant check (for example by running each predicted term through AutoEFT or an invariant-theory routine) and report separately the fractions that are fully gauge invariant and those that only conserve U(1).
  3. [§6.2, Appendix F, Eq. (15)] The conjugation-axis analysis reduces the 1024-dimensional encoder embeddings to three dimensions with t-SNE before computing difference vectors and cosine similarities. t-SNE is stochastic and designed to preserve local neighborhood structure, not global vector offsets, so a preferred direction in the projected space does not establish a consistent conjugation direction in the model's internal representation. Please repeat the analysis on the raw embeddings (or with a linear projection such as PCA) or otherwise justify that the t-SNE projection preserves offset structure; as it stands, the evidence for the claim that the model has learned conjugation as a consistent transformation is not conclusive.
minor comments (5)
  1. [Table 4] The two-row layout of Table 4 is garbled in the manuscript: for n=2 and n=7 the numbers appear as four entries rather than two, making the reported percentages hard to read.
  2. [§3.2] The exact sampling algorithm for the uniform dataset is not given beyond the statement that U(1) hypercharges are positive; please provide the concrete distribution or a pointer to the released generation code.
  3. [§4, Eq. (14)] The Lagrangian score can be negative because of the length penalty, and Figure 3 shows negative values, but the text never explains how a negative score should be interpreted; a sentence on this would help.
  4. [§4 and Appendix D] The matching rule for terms in the evaluation metrics is not specified for duplicate or repeated predicted terms; the repeated-term failure mode noted for the 2HDM in Appendix D would be easier to interpret with an explicit statement of whether terms are treated as a multiset.
  5. [Appendix E] The OOD U(1) test set can contain copies of the same Lagrangian with different non-minimal charge representations, but the text does not report how many distinct underlying Lagrangians remain after de-duplication; please state the effective sample sizes for each entry in Tables 15-18.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper evaluates against a held-out generator and external model benchmarks; the generator-ground-truth caveat is disclosed, not a hidden reduction.

full rationale

The paper's derivation chain is a supervised sequence-modeling benchmark, not a first-principles physical derivation. Training and test labels are both produced by the AutoEFT-plus-custom-code pipeline (Section 3.1), but the test sets are newly generated and explicitly filtered to remove overlap with training data (Section 4). Accuracy therefore measures held-out generalization to the same generator distribution, which is a legitimate, non-circular evaluation: the model can fail, and it does fail on OOD field counts and on the Standard Model benchmark (Appendix D, best SM Lagrangian score 0.77). The paper's central claim, that transformers can write Lagrangians from a field list, is conditional on the AutoEFT generator enumerating correct gauge-invariant, Lorentz-invariant, renormalizable terms; this ground-truth assumption is a benchmark-validity caveat, not a circular reduction. The paper explicitly acknowledges this in Section 1: 'The code we use to generate our dataset can write Lagrangians from a list of particles automatically, which at first might seem like it defeats the purpose of a training a transformer model with data generated by it.' No load-bearing self-citation occurs: AutoEFT [43] is external, and no author-overlapping uniqueness theorem or ansatz is invoked. The embedding and conjugation analyses are interpretability claims, not derivations of the main result. There is no equation-level reduction of a predicted quantity to a fitted parameter or to the input definition, so no circularity step can be exhibited.

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

The paper's claims rest on the correctness of the AutoEFT-based generator and on the tokenization faithfully encoding gauge contractions. The free parameters are dataset-sampling choices that shape the measured accuracy. No new physical entities are introduced; the ID tokens are representational labels, not physical objects.

free parameters (4)
  • Dataset field-count sampling probabilities = {1: 25%, 2: 25%, 3: 25%, 4: 11%, 5: 7%, 6: 7%}
    Hand-chosen in Section 3.2 to oversample short Lagrangians; directly shapes the training distribution and the model's performance profile.
  • Trilinear enrichment fraction = ~50% of Lagrangians with >2 fields
    Hand-chosen enrichment to ensure trilinear terms appear; this changes the difficulty of the task and the model's accuracy on trilinear versus non-trilinear cases.
  • Context length = 2048 tokens
    Chosen in Appendix B to fit A100 GPU training within 1 week; filters long Lagrangians out of the dataset.
  • U(1) hypercharge sampling range = numerators -9..9, denominators 1..9, simplified fractions
    Hand-chosen sampling range in Section 3.2; defines the input distribution and the OOD charge tests.
assumptions (4)
  • domain assumption AutoEFT and the custom code produce the complete set of allowed gauge-invariant terms for a given field list.
    Section 3.1: the training labels and evaluation ground truth come from this pipeline; any error in the generator propagates to the accuracy numbers and to the validity of the OOD 'reasonable' outputs.
  • domain assumption Every term in the Lagrangian must have integer mass dimension and be renormalizable, limiting terms to up to 4 scalars, 2 fermions, or 2 fermions plus 1 scalar.
    Section 2.2: this constraint is baked into the dataset generation and into the definition of a 'reasonable' Lagrangian.
  • standard math SU(2) representations are pseudo-real, so a doublet and its conjugate transform equivalently.
    Section 2.2 states this and uses it to ignore SU(2) conjugation; standard group theory.
  • ad hoc to paper t-SNE projection to 3 dimensions preserves vector-offset structure well enough for cosine-similarity analysis of conjugation.
    Section 6.2: the conjugation vectors are computed in t-SNE space, an assumption the paper does not justify; t-SNE is nonlinear and stochastic.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generating particle physics Lagrangians with transformers." pith.science (2026). https://pith.science/paper/BYTPIMFS

@misc{pith2026250109729,
  author       = {Pith},
  title        = {Pith review of: Generating particle physics Lagrangians with transformers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BYTPIMFS}},
  note         = {Machine review of arXiv:2501.09729}
}
abstract

In physics, Lagrangians provide a systematic way to describe laws governing physical systems. In the context of particle physics, they encode the interactions and behavior of the fundamental building blocks of our universe. By treating Lagrangians as complex, rule-based constructs similar to linguistic expressions, we trained a transformer model -- proven to be effective in natural language tasks -- to predict the Lagrangian corresponding to a given list of particles. We report on the transformer's performance in constructing Lagrangians respecting the Standard Model $\mathrm{SU}(3)\times \mathrm{SU}(2)\times \mathrm{U}(1)$ gauge symmetries. The resulting model is shown to achieve high accuracies (over 90\%) with Lagrangians up to six matter fields, with the capacity to generalize beyond the training distribution, albeit within architectural constraints. We show through an analysis of input embeddings that the model has internalized concepts such as group representations and conjugation operations as it learned to generate Lagrangians. We make the model and training datasets available to the community. An interactive demonstration can be found at: \url{https://huggingface.co/spaces/JoseEliel/generate-lagrangians}.

Figures

Figures reproduced from arXiv: 2501.09729 by the authors.

Figure 1
Figure 1. More technical details on the architecture and training process can be found in Appendix B. [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Training Data Distribution of both sampled (left) and uniform (right) datasets. [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Distribution of the Lagrangian scores (Eq. 14) for the sampled model (left) and the uniform [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Cumulative distribution of the fraction of wrong terms in predicted Lagrangian with [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]
Figure 5
Figure 5. Figure 5: Cumulative distribution of the fraction of wrong terms in predicted Lagrangian with and [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: Lagrangian scores distribution (left) and mean values of different metrics (right) across different n-field scenarios. Dotted lines indicate the separation between InD dataset and OOD dataset. In the right plot, the metrics shown include the correct score(NCorrect/NTru…
Figure 7
Figure 7. Figure 7: Failure modes of sampled model in OOD n-field scenarios. Left: Fractions of number of predicted terms over number of expected terms. Middle: The number of unique fields found in the predicted Lagrangians. Right: The number of expected fields used in predicted Lagrangia…
Figure 8
Figure 8. Figure 8: t-SNE visualization of individual field embeddings. Each plot shows the representation of [PITH_FULL_IMAGE:figures/full_fig_p019_8.png]
Figure 9
Figure 9. Figure 9: Left: Vector offsets of word pairs illustrating the gender and royalty relation. Right: Vector offsets of field pairs illustrating the conjugation relation 19 [PITH_FULL_IMAGE:figures/full_fig_p019_9.png]
Figure 10
Figure 10. Figure 10: Normalized distribution of absolute values of cosine similarities between pairs of conjuga [PITH_FULL_IMAGE:figures/full_fig_p020_10.png]
Figure 11
Figure 11. Figure 11: Accuracy (Percentage of Lagrangians with perfect prediction) with varying OOD Fields [PITH_FULL_IMAGE:figures/full_fig_p029_11.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. Refining Integration-by-Parts Reduction of Feynman Integrals with Machine Learning

    hep-th 2025-02 conditional novelty 6.0 of 10

    Machine learning program search rediscovers state-of-the-art integration-by-parts seeding heuristics and finds a modestly smaller seed set for a single two-loop benchmark integral.

  2. HEPTAPOD: Orchestrating High Energy Physics Workflows Towards Autonomous Agency

    hep-ph 2025-12 conditional novelty 4.0 of 10

    HEPTAPOD uses LLM agents to drive FeynRules, MadGraph, Pythia, and analysis tools through schema-validated tool calls and run-card templates, demonstrated on a leptoquark signal scan.

Reference graph

Works this paper leans on

53 extracted references · 8 canonical work pages · cited by 2 Pith papers

  1. [1]

    Vaswani, N

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin, Attention is all you need (2023), arXiv:1706.03762 [cs.CL]

    arXiv 2023

  2. [2]

    Mikuni and F

    V . Mikuni and F. Canelli, Eur.Phys.J.Plus135, 463 (2020), arXiv:2001.05311 [hep-ex]

    arXiv 2020

  3. [3]

    B. M. Dillon, R. Mastandrea, and B. Nachman, Physical Review D 106, 10.1103/phys- revd.106.056005 (2022)

    doi:10.1103/phys- 2022

  4. [4]

    Mikuni and F

    V . Mikuni and F. Canelli, Mach. Learn. Sci. Tech. 2, 035027 (2021), arXiv:2102.05073 [physics.data-an]

    arXiv 2021

  5. [5]

    H. Qu, C. Li, and S. Qian, in International Conference on Machine Learning (PMLR, 2022) pp. 18281–18292

    work page 2022

  6. [6]

    S. Qiu, S. Han, X. Ju, B. Nachman, and H. Wang, Phys. Rev. D 107, 114029 (2023), arXiv:2203.05687 [hep-ph] . 30

    work page Pith review arXiv 2023

  7. [7]

    Builtjes, S

    L. Builtjes, S. Caron, P. Moskvitina, C. Nellist, R. R. de Austri, R. Verheyen, and Z. Zhang, (2022), arXiv:2211.05143 [hep-ph]

    arXiv 2022

  8. [8]

    B. Käch, D. Krücker, and I. Melzer-Pellmann, (2022), arXiv:2211.13623 [hep-ex]

    arXiv 2022

Show all 53 references
  1. [9]

    F. A. Di Bello et al., Eur. Phys. J. C 83, 596 (2023), arXiv:2212.01328 [hep-ex]

    2023 arXiv

  2. [10]

    Finke, M

    T. Finke, M. Krämer, A. Mück, and J. Tönshoff, JHEP 06, 184, arXiv:2303.07364 [hep-ph]

    arXiv

  3. [11]

    Butter, N

    A. Butter, N. Huetsch, S. Palacios Schweitzer, T. Plehn, P. Sorrenson, and J. Spinner, (2023), arXiv:2305.10475 [hep-ph]

    2023 arXiv

  4. [12]

    J. A. Raine, M. Leigh, K. Zoch, and T. Golling, Phys. Rev. D 109, 012005 (2024), arXiv:2307.02405 [hep-ph]

    2024 arXiv

  5. [13]

    He and D

    M. He and D. Wang, Eur. Phys. J. C 83, 1116 (2023), arXiv:2307.04723 [hep-ph]

    2023 arXiv

  6. [14]

    A. Li, V . Krishnamohan, R. Kansal, R. Sen, S. Tsan, Z. Zhang, and J. Duarte, in37th Conference on Neural Information Processing Systems (2023) arXiv:2312.04757 [hep-ex]

    2023 arXiv

  7. [15]

    Hammad, S

    A. Hammad, S. Moretti, and M. Nojiri, JHEP 03, 144, arXiv:2401.00452 [hep-ph]

    arXiv

  8. [16]

    Spinner, V

    J. Spinner, V . Bresó, P. de Haan, T. Plehn, J. Thaler, and J. Brehmer, (2024), arXiv:2405.14806 [physics.data-an]

    2024 arXiv

  9. [17]

    Blekman, F

    F. Blekman, F. Canelli, A. De Moor, K. Gautam, A. Ilg, A. Macchiolo, and E. Ploerer, (2024), arXiv:2406.08590 [hep-ex]

    2024 arXiv

  10. [18]

    Quétant, J

    G. Quétant, J. A. Raine, M. Leigh, D. Sengupta, and T. Golling, Phys. Rev. D 110, 076023 (2024), arXiv:2406.13074 [hep-ph]

    2024 arXiv

  11. [19]

    Caron, N

    S. Caron, N. Dobreva, A. F. Sánchez, J. D. Martín-Guerrero, U. Odyurt, R. R. Ruiz de Aus- tri Bazan, Z. Wolffs, and Y . Zhao, (2024), arXiv:2407.07179 [hep-ex]

    2024 arXiv

  12. [20]

    Y . Wu, K. Wang, C. Li, H. Qu, and J. Zhu, Chin. Phys. C49, 013110 (2025), arXiv:2407.08682 [hep-ph]

    2025 arXiv

  13. [21]

    Tagami, T

    R. Tagami, T. Suehara, and M. Ishino, EPJ Web Conf.315, 03011 (2024), arXiv:2410.11322 [hep-ex]

    2024 arXiv

  14. [22]

    Brehmer, V

    J. Brehmer, V . Bresó, P. de Haan, T. Plehn, H. Qu, J. Spinner, and J. Thaler, (2024), arXiv:2411.00446 [hep-ph]

    2024

  15. [23]

    Van Stroud, P

    S. Van Stroud, P. Duckett, M. Hart, N. Pond, S. Rettie, G. Facini, and T. Scanlon, (2024), arXiv:2411.07149 [hep-ex]

    2024

  16. [24]

    Tomiya and Y

    A. Tomiya and Y . Nagai, PoSLATTICE2023, 001 (2024), arXiv:2310.13222 [hep-lat]

    2024 arXiv

  17. [25]

    Gao, (2024), arXiv:2408.13280 [hep-lat]

    L. Gao, (2024), arXiv:2408.13280 [hep-lat]

    2024 arXiv

  18. [26]

    Cranmer, A

    M. Cranmer, A. Sanchez-Gonzalez, P. Battaglia, R. Xu, K. Cranmer, D. Spergel, and S. Ho, (2020), arXiv:2006.11287 [cs.LG]

    2020 arXiv

  19. [27]

    Lemos, N

    P. Lemos, N. Jeffrey, M. Cranmer, S. Ho, and P. Battaglia, Rediscovering orbital mechanics with machine learning (2022), arXiv:2202.02306 [astro-ph.EP]

    2022 arXiv

  20. [28]

    Charton, CoRR abs/2112.01898 (2021), 2112.01898

    F. Charton, CoRR abs/2112.01898 (2021), 2112.01898

    2021 arXiv

  21. [29]

    Alnuqaydan, S

    A. Alnuqaydan, S. Gleyzer, and H. Prosper, Mach. Learn. Sci. Tech. 4, 015007 (2023), arXiv:2206.08901 [hep-ph]

    2023 arXiv

  22. [30]

    T. Cai, G. W. Merz, F. Charton, N. Nolte, M. Wilhelm, K. Cranmer, and L. J. Dixon, Mach. Learn. Sci. Tech. 5, 035073 (2024), arXiv:2405.06107 [cs.LG]

    2024 arXiv

  23. [31]

    Cheung, A

    C. Cheung, A. Dersy, and M. D. Schwartz, (2024), arXiv:2408.04720 [hep-th] . 31

    2024 arXiv

  24. [32]

    Dersy, M

    A. Dersy, M. D. Schwartz, and X. Zhang, Int. J. Data Sci. Math. Sci. 1, 135 (2024), arXiv:2206.04115 [cs.LG]

    2024 arXiv

  25. [33]

    Charton, in The Twelfth International Conference on Learning Representations(2024)

    F. Charton, in The Twelfth International Conference on Learning Representations(2024)

    2024

  26. [34]

    Lample and F

    G. Lample and F. Charton, Deep learning for symbolic mathematics (2019), arXiv:1912.01412 [cs.SC]

    2019 arXiv

  27. [35]

    T. R. Harvey and A. Lukas, JHEP 08, 161, arXiv:2103.04759 [hep-th]

    arXiv

  28. [36]

    Nishimura, C

    S. Nishimura, C. Miyao, and H. Otsuka, (2024), arXiv:2409.10023 [hep-ph]

    2024

  29. [37]

    G. N. Wojcik, S. T. Eu, and L. L. Everett, (2024), arXiv:2407.07184 [hep-ph]

    2024 arXiv

  30. [38]

    G. N. Wojcik, S. T. Eu, and L. L. Everett, (2024), arXiv:2407.07203 [hep-ph]

    2024 arXiv

  31. [39]

    Constantin, T

    A. Constantin, T. R. Harvey, and A. Lukas, Fortsch. Phys. 70, 2100186 (2022), arXiv:2108.07316 [hep-th]

    2022 arXiv

  32. [40]

    Nishimura, C

    S. Nishimura, C. Miyao, and H. Otsuka, JHEP 23, 021, arXiv:2304.14176 [hep-ph]

    arXiv

  33. [41]

    M. D. Schwartz, Quantum Field Theory and the Standard Model (Cambridge University Press, 2013)

    2013

  34. [42]

    Mastropietro, (2023), arXiv:2312.11400 [hep-th]

    V . Mastropietro, (2023), arXiv:2312.11400 [hep-th]

    2023 arXiv

  35. [43]

    R. V . Harlander and M. C. Schaaf, Computer Physics Communications300, 109198 (2024)

    2024

  36. [44]

    Lewis, arXiv preprint arXiv:1910.13461 (2019)

    M. Lewis, arXiv preprint arXiv:1910.13461 (2019)

    2019 arXiv

  37. [45]

    Charton, What is my math transformer doing? – three results on interpretability and general- ization (2022), arXiv:2211.00170 [cs.LG]

    F. Charton, What is my math transformer doing? – three results on interpretability and general- ization (2022), arXiv:2211.00170 [cs.LG]

    2022 arXiv

  38. [46]

    Jelassi, S

    S. Jelassi, S. d’Ascoli, C. Domingo-Enrich, Y . Wu, Y . Li, and F. Charton, arXiv preprint arXiv:2306.15400 (2023)

    2023 arXiv

  39. [47]

    Sennrich, arXiv preprint arXiv:1508.07909 (2015)

    R. Sennrich, arXiv preprint arXiv:1508.07909 (2015)

    2015 arXiv

  40. [48]

    Araabi, C

    A. Araabi, C. Monz, and V . Niculae, in Proceedings of the 15th biennial conference of the Association for Machine Translation in the Americas (Volume 1: Research Track), edited by K. Duh and F. Guzmán (Association for Machine Translation in the Americas, Orlando, USA,

  41. [49]

    Golkar, A

    S. Golkar, A. Bietti, M. Pettee, M. Eickenberg, M. Cranmer, K. Hirashima, G. Krawezik, N. Lourie, M. McCabe, R. Morel, et al., arXiv preprint arXiv:2406.02585 (2024)

    2024 arXiv

  42. [50]

    Mikolov, W.-t

    T. Mikolov, W.-t. Yih, and G. Zweig, inProceedings of the 2013 Conference of the North Amer- ican Chapter of the Association for Computational Linguistics: Human Language Technologies, edited by L. Vanderwende, H. Daumé III, and K. Kirchhoff (Association for Computational Ling...

    2013

  43. [51]

    T. Wolf, L. Debut, V . Sanh, J. Chaumond, C. Delangue, A. Moi, P. Cistac, T. Rault, R. Louf, M. Funtowicz, J. Davison, S. Shleifer, P. von Platen, C. Ma, Y . Jernite, J. Plu, C. Xu, T. L. Scao, S. Gugger, M. Drame, Q. Lhoest, and A. M. Rush, Huggingface’s transformers: State-o...

    2020 arXiv

  44. [52]

    Georgi and M

    H. Georgi and M. Machacek, Nucl. Phys. B 262, 463 (1985)

    1985

  45. [53]

    Golkar, M

    S. Golkar, M. Pettee, M. Eickenberg, A. Bietti, M. Cranmer, G. Krawezik, F. Lanusse, M. Mc- Cabe, R. Ohana, L. Parker, B. R.-S. Blancard, T. Tesileanu, K. Cho, and S. Ho, xval: A continuous numerical tokenization for scientific language models (2024), arXiv:2310.02989 [stat.ML] . 32

    2024 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.