Pith. sign in

REVIEW 1 major objections 38 references

Graph Alignment Topology as an Inductive Bias for Grounding Detection

T0 review · 1 major / 0 minor · reviewed 2026-05-25 · grok-4.3

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

Pith's one-line read Training graph neural networks on bipartite alignment graphs between references and LLM outputs detects grounding more accurately than prior methods or GPT-4o.

desk verdict GNN on bipartite alignment graphs is a distinct inductive bias for grounding detection, but the SOTA claim on four datasets needs cross-dataset checks to be convincing. read the letter →

arxiv 2605.22963 v1 pith:22HI66ET submitted 2026-05-21 cs.CL cs.AI

Paul Landes , Pranav Herur , Adam Cross , Jimeng Sun This is my paper · ORCID
classification cs.CLcs.AI
keywords hallucinationdetectiongroundinggraphneuralnetworksbipartitegraphsLLMoutputsmessagepassingalignmenttopology
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

Large language models optimize for plausible text rather than explicit verification that each claim follows from a source document, which produces hallucinations in high-stakes settings such as clinical support. The paper constructs aligned bipartite graphs linking reference passages to generated statements and trains a graph neural network to perform message passing directly over the alignment edges. This structure supplies an inductive bias that previous retrieval, consistency, or verification pipelines do not use. The resulting detector reaches state-of-the-art accuracy on four separate hallucination and question-answering benchmarks while surpassing GPT-4o and other strong baselines.

What carries the argument

Aligned bipartite graphs between reference information and LLM outputs, modeled by graph neural network message passing over the alignment edges.

What would settle it

A fifth, previously unseen hallucination or question-answering dataset on which the trained graph detector scores below GPT-4o or the strongest baseline method.

Watch

Extended reading notes

Core claim

Constructing aligned bipartite graphs between reference information and LLM outputs and training a graph neural network to model alignment structure using message passing produces a detector whose performance exceeds all compared methods, including GPT-4o, on four diverse hallucination and question-answering datasets.

Load-bearing premise

That the alignment topology learned from the training graphs will capture grounding patterns that continue to hold for new references and new LLM outputs outside the four evaluated datasets.

Editorial extensions

If this is right

  • The detector outperforms all compared methods including GPT-4o across the four tested datasets.
  • Alignment topology supplies an inductive bias that retrieval augmentation and self-consistency alone do not provide.
  • The same graph construction can be applied in clinical decision support where strict factual grounding is required.
  • Message passing over alignment edges directly encodes whether generated propositions are entailed by source documents.

Reading between the lines

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

  • If the bipartite-graph construction proves robust, the same topology could be extracted from non-text modalities such as image captions or code snippets.
  • The approach could be inserted as an additional verification layer inside retrieval-augmented generation pipelines without retraining the base LLM.
  • Performance may degrade when reference documents contain internal contradictions that the current alignment procedure does not explicitly represent.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

1 major / 0 minor

Summary. The paper proposes constructing aligned bipartite graphs between reference documents and LLM-generated outputs, then training a graph neural network (GNN) via message passing over the alignment topology to serve as an inductive bias for detecting whether LLM responses are grounded. It reports that this approach yields state-of-the-art hallucination and grounding detection performance on four diverse datasets, outperforming prior methods as well as strong baselines including GPT-4o.

Significance. If the central claim holds and the learned alignment topology generalizes, the work would introduce a distinct inductive bias—graph topology over explicit alignments—into hallucination detection, distinct from retrieval augmentation or self-consistency techniques. This could be particularly relevant for high-stakes applications such as clinical decision support where explicit grounding verification is required.

major comments (1)
  1. [Abstract] Abstract: the central claim that message passing on aligned bipartite graphs supplies a transferable inductive bias for grounding detection is load-bearing, yet the manuscript reports results exclusively on four fixed hallucination/QA datasets with no cross-dataset evaluation, held-out domain testing, or ablation of the alignment-construction heuristics. Without such evidence it remains possible that the reported gains encode dataset-specific artifacts rather than general grounding structure.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the detailed and constructive review. The central concern regarding the generalizability of the alignment topology inductive bias is well-taken. We address it point-by-point below and commit to revisions that directly strengthen the evidence for transferability.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that message passing on aligned bipartite graphs supplies a transferable inductive bias for grounding detection is load-bearing, yet the manuscript reports results exclusively on four fixed hallucination/QA datasets with no cross-dataset evaluation, held-out domain testing, or ablation of the alignment-construction heuristics. Without such evidence it remains possible that the reported gains encode dataset-specific artifacts rather than general grounding structure.

    Authors: We agree that the absence of cross-dataset transfer experiments, held-out domain testing, and ablations on the alignment heuristics leaves the transferability claim under-supported in the current version. The four datasets are drawn from distinct hallucination and QA settings, but this does not substitute for explicit cross-evaluation. In the revision we will add: (1) cross-dataset experiments training on three datasets and testing on the held-out fourth, (2) domain-shift tests using an additional out-of-distribution corpus, and (3) systematic ablations that vary the alignment-construction rules (e.g., threshold, matching strategy) while keeping the GNN fixed. These results will be reported with the same metrics and baselines to isolate the contribution of the topology bias. revision: yes

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity; empirical GNN method with dataset-specific results

full rationale

The paper describes an empirical procedure: construct aligned bipartite graphs from reference and LLM output, then train a GNN via message passing to model alignment topology. Central claims consist of SOTA empirical results on four specific hallucination/QA datasets. No equations, fitted parameters renamed as predictions, self-citation load-bearing uniqueness theorems, or ansatz smuggling appear in the provided text. The derivation chain is self-contained as a standard supervised graph learning pipeline whose performance is measured externally on held-out test sets rather than reduced to its own inputs by construction.

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

Abstract-only review supplies no explicit free parameters, axioms, or invented entities; the ledger is therefore empty.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Graph Alignment Topology as an Inductive Bias for Grounding Detection." pith.science (2026). https://pith.science/paper/22HI66ET

@misc{pith2026260522963,
  author       = {Pith},
  title        = {Pith review of: Graph Alignment Topology as an Inductive Bias for Grounding Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/22HI66ET}},
  note         = {Machine review of arXiv:2605.22963}
}
read the original abstract

Large Language Models (LLMs) are optimized to produce distributionally plausible continuations rather than to explicitly verify whether generated propositions are entailed by source documents. This inductive bias enables generalization, but it does not encode whether responses are grounded with respect to a reference. These issues limit the use of LLMs in domains where strict factual correctness is crucial, such as clinical decision support. Existing hallucination detection approaches improve factuality through retrieval augmentation, self-consistency, or claim verification, but generally do not learn directly over alignment topology. To leverage alignment topology as an inductive bias, we construct aligned bipartite graphs between reference information and LLM outputs and train a graph neural network (GNN) to model alignment structure using message passing. The method achieves state-of-the-art results on four diverse hallucination and question-answering datasets, outperforming all compared methods, including foundational LLMs such as GPT-4o.

Figures

Figures reproduced from arXiv: 2605.22963 by the authors.

Figure 1
Figure 1. Overview of CALAMRFLOW on a MEDHALLU [27] dataset example. MEDHALLU denotes the reference evidence as knowledge and provides ground truth and hallucination responses. CALAMRFLOW constructs semantic graphs from the knowledge and candidate response, aligns the resulting reference–response graphs, weights the topology with flow-derived support features, and classifies the response as supported or unsupported. algorithm… view at source ↗
Figure 2
Figure 2. Graph-structure intervention analysis. Rows distinguish perturbation and ablation experi [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Per-dataset graph-structure intervention results. Rows distinguish perturbation and ablation [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Aligned cosine-distance distributions across the [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]
Figure 5
Figure 5. Figure 5: Three-dimensional t-SNE visualization of trained third-layer [PITH_FULL_IMAGE:figures/full_fig_p016_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages

  1. [1]

    What`s in a Summary? Laying the Groundwork for Advances in Hospital-Course Summarization

    Griffin Adams, Emily Alsentzer, Mert Ketenci, Jason Zucker, and Noémie Elhadad. What’s in a Summary? Laying the Groundwork for Advances in Hospital-Course Summarization. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4794–4811, June 2021. doi: 10.1865...

    work page doi:10.18653/v1/2021.naacl-main.382 2021

  2. [2]

    Abstract Meaning Representation for Sembanking

    Laura Banarescu, Claire Bonial, Shu Cai, Madalina Georgescu, Kira Griffitt, Ulf Hermjakob, Kevin Knight, Philipp Koehn, Martha Palmer, and Nathan Schneider. Abstract Meaning Representation for Sembanking. InProceedings of the 7th Linguistic Annotation Workshop and Interoperability with Discourse, pages 178–186, Sofia, Bulgaria, August 2013. Association fo...

    work page 2013

  3. [3]

    XL-AMR: Enabling Cross-Lingual AMR Parsing with Transfer Learning Techniques

    Rexhina Blloshmi, Rocco Tripodi, and Roberto Navigli. XL-AMR: Enabling Cross-Lingual AMR Parsing with Transfer Learning Techniques. InProceedings of the 2020 Conference on Em- pirical Methods in Natural Language Processing (EMNLP), pages 2487–2500, Online, Novem- ber 2020. Association for Computational Linguistics. doi: 10.18653/v1/2020.emnlp-main.195. UR...

    work page doi:10.18653/v1/2020.emnlp-main.195 2020

  4. [4]

    Large-scale Simple Question Answering with Memory Networks

    Antoine Bordes, Nicolas Usunier, Sumit Chopra, and Jason Weston. Large-scale Simple Question Answering with Memory Networks.arXiv:1506.02075 [cs], June 2015. URL http://arxiv.org/abs/1506.02075

    work page Pith review arXiv 2015

  5. [5]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D. Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-V oss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chris Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott ...

    work page 1901

  6. [6]

    Di Luna and Giovanni Viglietta

    Li Chen, Rasmus Kyng, Yang P. Liu, Richard Peng, Maximilian Probst Gutenberg, and Sushant Sachdeva. Maximum Flow and Minimum-Cost Flow in Almost-Linear Time. In2022 IEEE 63rd Annual Symposium on F oundations of Computer Science (FOCS), pages 612–623, October 2022. doi: 10.1109/FOCS54457.2022.00064. URL https://ieeexplore.ieee.org/document/ 9996881

    work page doi:10.1109/focs54457.2022.00064 2022

  7. [7]

    Åström and Richard M

    Lester Randolph Ford and Delbert Ray Fulkerson. Flows in networks. InFlows in Net- works, Princeton Landmarks in Mathematics and Physics, page 212. Princeton University Press, 1962. ISBN 978-0-691-65184-2. URL https://press.princeton.edu/books/ hardcover/9780691651842/flows-in-networks

    work page arXiv 1962

  8. [8]

    Schoenholz, Patrick F

    Justin Gilmer, Samuel S. Schoenholz, Patrick F. Riley, Oriol Vinyals, and George E. Dahl. Neural Message Passing for Quantum Chemistry. InProceedings of the 34th International Conference on Machine Learning, pages 1263–1272. PMLR, July 2017. URL https:// proceedings.mlr.press/v70/gilmer17a.html

    work page 2017

Show all 38 references
  1. [9]

    The Curious Case of Neural Text Degeneration

    Ari Holtzman, Jan Buys, Li Du, Maxwell Forbes, and Yejin Choi. The Curious Case of Neural Text Degeneration. InInternational Conference on Learning Representations, 2020. URL https://openreview.net/forum?id=rygGQyrFvH

    2020

  2. [10]

    Knowledge-Centric Hallucination Detection

    Xiangkun Hu, Dongyu Ru, Lin Qiu, Qipeng Guo, Tianhang Zhang, Yang Xu, Yun Luo, Pengfei Liu, Yue Zhang, and Zheng Zhang. Knowledge-Centric Hallucination Detection. In Yaser Al-Onaizan, Mohit Bansal, and Yun-Nung Chen, editors,Proceedings of the 2024 Conference on Empirical Meth...

    2024 doi

  3. [11]

    RefChecker: Reference-based Fine-grained Hallucination Checker and Benchmark for Large Language Models, May 2024

    Xiangkun Hu, Dongyu Ru, Lin Qiu, Qipeng Guo, Tianhang Zhang, Yang Xu, Yun Luo, Pengfei Liu, Yue Zhang, and Zheng Zhang. RefChecker: Reference-based Fine-grained Hallucination Checker and Benchmark for Large Language Models, May 2024. URL http: //arxiv.org/abs/2405.14486

    2024

  4. [12]

    Towards Mitigating LLM Hallucination via Self Reflection

    Ziwei Ji, Tiezheng Yu, Yan Xu, Nayeon Lee, Etsuko Ishii, and Pascale Fung. Towards Mitigating LLM Hallucination via Self Reflection. In Houda Bouamor, Juan Pino, and Kalika Bali, editors, Findings of the Association for Computational Linguistics: EMNLP 2023, pages 1827–1843, S...

    2023 doi

  5. [13]

    PubMedQA: A Dataset for Biomedical Research Question Answering

    Qiao Jin, Bhuwan Dhingra, Zhengping Liu, William Cohen, and Xinghua Lu. PubMedQA: A Dataset for Biomedical Research Question Answering. In Kentaro Inui, Jing Jiang, Vin- cent Ng, and Xiaojun Wan, editors,Proceedings of the 2019 Conference on Empirical Meth- ods in Natural Lang...

    2019 doi

  6. [14]

    From TreeBank to PropBank

    Paul Kingsbury and Martha Palmer. From TreeBank to PropBank. InProceedings of the Third International Conference on Language Resources and Evaluation (LREC’02), Las Palmas, Canary Islands - Spain, May 2002. European Language Resources Association (ELRA). URL http://www.lrec-co...

    2002

  7. [15]

    Kipf and Max Welling

    Thomas N. Kipf and Max Welling. Semi-Supervised Classification with Graph Convolutional Networks. In5th International Conference on Learning Representations, ICLR 2017, Toulon, France, February 2017. OpenReview.net. doi: 10.48550/arXiv.1609.02907. URL https: //openreview.net/f...

    2017 doi arXiv

  8. [16]

    CALAMR: Component ALignment for Abstract Meaning Representation

    Paul Landes and Barbara Di Eugenio. CALAMR: Component ALignment for Abstract Meaning Representation. InProceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation, pages 2622–2637, Torino, Italy, May 2024. ELRA and ICC...

    2024

  9. [17]

    Chase, and Barbara Di Eugenio

    Paul Landes, Sitara Rao, Aaron J. Chase, and Barbara Di Eugenio. Toward Complete Hospital Discharge Summarization with Abstract Meaning Representation. InProceedings of the 2026 IEEE 14th International Conference on Healthcare Informatics, Minneapolis, MN, USA, June

    2026

  10. [18]

    URLhttps://zhang-informatics.github.io/ICHI2026/index.html

    IEEE. URLhttps://zhang-informatics.github.io/ICHI2026/index.html

  11. [19]

    Retrieval-augmented generation for knowledge-intensive NLP tasks

    Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Na- man Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. Retrieval-augmented generation for knowledge-intensive NLP tasks. In H. Larochelle, M...

    2020

  12. [20]

    HaluEval: A Large-Scale Hallucination Evaluation Benchmark for Large Language Models

    Junyi Li, Xiaoxue Cheng, Xin Zhao, Jian-Yun Nie, and Ji-Rong Wen. HaluEval: A Large-Scale Hallucination Evaluation Benchmark for Large Language Models. In Houda Bouamor, Juan Pino, and Kalika Bali, editors,Proceedings of the 2023 Conference on Empirical Methods in Natural Lang...

    2023 doi

  13. [21]

    Inference- time intervention: Eliciting truthful answers from a language model

    Kenneth Li, Oam Patel, Fernanda Viégas, Hanspeter Pfister, and Martin Wattenberg. Inference- time intervention: Eliciting truthful answers from a language model. InAdvances in Neural Information Processing Systems, volume 36, New Orleans, Louisiana, USA, 2023. Curran Associate...

    2023 doi arXiv

  14. [22]

    Yujia Li, Daniel Tarlow, Marc Brockschmidt, and Richard S. Zemel. Gated Graph Sequence Neural Networks. In4th International Conference on Learning Representations, Iclr ’16, 11 San Juan, Puerto Rico, May 2016. ICLR. doi: 10.48550/arXiv.1511.05493. URL https: //arxiv.org/abs/1511.05493

    2016 doi arXiv

  15. [23]

    I Know What You Asked: Graph Path Learning using AMR for Commonsense Reasoning

    Jungwoo Lim, Dongsuk Oh, Yoonna Jang, Kisu Yang, and Heuiseok Lim. I Know What You Asked: Graph Path Learning using AMR for Commonsense Reasoning. InProceedings of the 28th International Conference on Computational Linguistics, pages 2459–2471, Barcelona, Spain (Online), Decem...

    2020

  16. [24]

    ROUGE: A Package for Automatic Evaluation of Summaries

    Chin-Yew Lin. ROUGE: A Package for Automatic Evaluation of Summaries. InText Summariza- tion Branches Out, pages 74–81, Barcelona, Spain, July 2004. Association for Computational Linguistics. URLhttps://aclanthology.org/W04-1013

    2004

  17. [25]

    Decoupled Weight Decay Regularization

    Ilya Loshchilov and Frank Hutter. Decoupled Weight Decay Regularization. In7th International Conference on Learning Representations, Iclr ’19, New Orleans, LA, USA, September 2018. ICLR. URLhttps://openreview.net/forum?id=Bkg6RiCqY7

    2018

  18. [26]

    SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models

    Potsawee Manakul, Adian Liusie, and Mark Gales. SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models. In Houda Bouamor, Juan Pino, and Kalika Bali, editors,Proceedings of the 2023 Conference on Empirical Methods in Natural Language...

    2023 doi

  19. [27]

    A Human Evaluation of AMR-to-English Generation Systems

    Emma Manning, Shira Wein, and Nathan Schneider. A Human Evaluation of AMR-to-English Generation Systems. InProceedings of the 28th International Conference on Computational Linguistics, pages 4773–4786, Barcelona, Spain (Online), December 2020. International Committee on Compu...

    2020 doi

  20. [28]

    MedHallu: A Comprehensive Benchmark for Detecting Medical Hallucinations in Large Language Models, February 2025

    Shrey Pandit, Jiawei Xu, Junyuan Hong, Zhangyang Wang, Tianlong Chen, Kaidi Xu, and Ying Ding. MedHallu: A Comprehensive Benchmark for Detecting Medical Hallucinations in Large Language Models, February 2025. URLhttp://arxiv.org/abs/2502.14302

    2025

  21. [29]

    HalluciNot: Hallucination Detection Through Context and Common Knowledge Verification, April 2025

    Bibek Paudel, Alexander Lyzhov, Preetam Joshi, and Puneet Anand. HalluciNot: Hallucination Detection Through Context and Common Knowledge Verification, April 2025. URL http: //arxiv.org/abs/2504.07069

    2025

  22. [30]

    Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. Exploring the limits of transfer learning with a unified text-to-text transformer.The Journal of Machine Learning Research, 21(1):140:5485–140:5551, Ja...

    2020

  23. [31]

    SQuAD: 100,000+ Questions for Machine Comprehension of Text

    Pranav Rajpurkar, Jian Zhang, Konstantin Lopyrev, and Percy Liang. SQuAD: 100,000+ Questions for Machine Comprehension of Text. InProceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 2383–2392, Austin, Texas, November 2016. Association ...

    2016 doi

  24. [32]

    Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks

    Nils Reimers and Iryna Gurevych. Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks. InProceedings of the 2019 Conference on Empirical Methods in Nat- ural Language Processing and the 9th International Joint Conference on Natural Lan- guage Processing (EMNLP-IJCNLP...

    2019

  25. [33]

    doi: 10.18653/v1/D19-1410

    Association for Computational Linguistics. doi: 10.18653/v1/D19-1410. URL https://aclanthology.org/D19-1410

    doi

  26. [34]

    GraphEval: A knowledge graph-based LLM hallucina- tion evaluation framework

    Benjamin Sansford, David Smith, et al. GraphEval: A knowledge graph-based LLM hallucina- tion evaluation framework. InProceedings of the KDD 2024 Workshop on Knowledge-Infused Learning (KiL). ACM, 2024. 12

    2024

  27. [35]

    The Graph Neural Network Model.IEEE Transactions on Neural Networks, 20 (1):61–80, January 2009

    Franco Scarselli, Marco Gori, Ah Chung Tsoi, Markus Hagenbuchner, and Gabriele Mon- fardini. The Graph Neural Network Model.IEEE Transactions on Neural Networks, 20 (1):61–80, January 2009. ISSN 1941-0093. doi: 10.1109/TNN.2008.2005605. URL https://ieeexplore.ieee.org/document/4700287

    2009 doi

  28. [36]

    Towards Clinical Encounter Summarization: Learning to Compose Discharge Summaries from Prior Notes.arXiv:2104.13498 [cs], April 2021

    Han-Chin Shing, Chaitanya Shivade, Nima Pourdamghani, Philip Resnik, Douglas Oard, and Parminder Bhatia. Towards Clinical Encounter Summarization: Learning to Compose Discharge Summaries from Prior Notes.arXiv:2104.13498 [cs], April 2021. URL http: //arxiv.org/abs/2104.13498

    2021

  29. [37]

    O’Donnell, Mario Giulianelli, and Ryan Cotterell

    Taiga Someya, Anej Svete, Brian DuSell, Timothy J. O’Donnell, Mario Giulianelli, and Ryan Cotterell. Information Locality as an Inductive Bias for Neural Language Models. In Wanxiang Che, Joyce Nabende, Ekaterina Shutova, and Mohammad Taher Pilehvar, editors,Proceedings of the...

    2025 doi

  30. [38]

    Manning, and Curtis Langlotz

    Yuhao Zhang, Derek Merck, Emily Tsai, Christopher D. Manning, and Curtis Langlotz. Op- timizing the Factual Correctness of a Summary: A Study of Summarizing Radiology Re- ports. InProceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 51...

    2020 doi

Pith tools

Reviewed May 25, 2026 · model on record in the stance chip above.