Pith. sign in

REVIEW 1 major objections 43 references

GRAMformer: Any-Order Modality Interactions via Volumetric Multimodal Cross-Attention

T0 review · 1 major / 0 minor · reviewed 2026-06-28 · grok-4.3

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

Pith's one-line read VMA defines attention scores as the volume spanned by a query and multiple modality-specific keys to capture joint multimodal dependencies.

desk verdict VMA attention scores collapse to zero once modalities exceed embedding dimension, breaking the any-number claim. read the letter →

arxiv 2606.06249 v1 pith:AB3M2IAC submitted 2026-06-04 cs.CV cs.LG

Giordano Cicchetti , Eleonora Grassucci , Danilo Comminiello This is my paper
classification cs.CVcs.LG
keywords multimodaltransformercross-attentionvolumetricattentionany-orderinteractionslearningGRAMformerjointgeometry
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 introduces Volumetric Multimodal cross-Attention (VMA) inside the GRAMformer architecture. VMA replaces standard pairwise dot-product attention with a computation based on the geometric volume formed by one query vector and several modality-specific key vectors. This change is meant to let the model represent interactions that depend on the joint configuration of any number of modalities at once. Current approaches either pay quadratic cost in the number of modalities or lose explicit joint information through concatenation. The authors claim the new formulation yields both better effectiveness and lower complexity when the number of modalities grows.

What carries the argument

Volumetric Multimodal cross-Attention (VMA), which defines each attention score from the volume of the parallelepiped formed by a query vector and multiple modality-specific key vectors.

What would settle it

A controlled experiment on a three-or-more-modality task in which a GRAMformer equipped with VMA shows no accuracy or efficiency gain over a standard cross-attention baseline that uses pairwise scores or concatenation.

Watch

Extended reading notes

Core claim

VMA computes attention scores as a function of the joint geometry of a query and multiple modality-specific keys, specifically the volume spanned by those vectors. This captures multimodal dependencies that go beyond pairwise similarity and supports native modeling of any-order modality interactions inside the GRAMformer transformer, which is built to accept an arbitrary number of input modalities.

Load-bearing premise

That computing attention scores from the volume spanned by a query and multiple modality keys will capture meaningful joint dependencies better than pairwise dot products or concatenation.

Editorial extensions

If this is right

  • GRAMformer can accept any number of modalities without incurring quadratic cost in the modality count.
  • Attention scores now reflect the joint configuration of several modalities rather than only pairwise relations.
  • The architecture supports higher-order interactions directly instead of approximating them through successive pairwise steps.
  • Empirical results on multimodal tasks show gains in both effectiveness and computational efficiency.

Reading between the lines

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

  • The geometric-volume formulation could be swapped into other attention variants that currently rely on dot products.
  • Tasks that combine vision, audio, text and sensor streams might benefit from replacing concatenation layers with VMA blocks.
  • A natural next measurement would be the scaling curve of VMA versus pairwise attention as the modality count rises from two to ten.
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 manuscript introduces GRAMformer, a multimodal transformer architecture featuring Volumetric Multimodal cross-Attention (VMA). VMA defines attention scores as a function of the volume spanned by a query vector and multiple modality-specific key vectors, with the goal of capturing joint multimodal dependencies beyond pairwise interactions and enabling native modeling of any-order modality interactions. The paper claims this yields improved effectiveness and efficiency on multimodal tasks compared to existing pairwise or concatenation-based attention formulations.

Significance. If the central mechanism were non-degenerate and empirically validated, VMA could provide a geometrically motivated alternative to standard multimodal attention, potentially reducing reliance on quadratic pairwise computations while supporting arbitrary modality counts.

major comments (1)
  1. [Abstract] Abstract: The definition of VMA attention scores via the volume spanned by one query and multiple modality-specific keys is load-bearing for the central claim of handling 'any number of modalities.' In R^d the (hyper)volume of the parallelepiped formed by m > d vectors is identically zero (equivalently, sqrt(det(G)) = 0 for the Gram matrix G of the stacked vectors). With standard embedding dimensions (512–1024) this forces all attention scores to vanish for modality counts exceeding d, directly contradicting the 'any-order' and 'arbitrary modality counts' positioning. No alternative non-degenerate formulation (e.g., summed lower-order volumes, normalized exterior product, or learned projection) is indicated.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful review and the precise mathematical observation on the volumetric attention. We respond to the major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The definition of VMA attention scores via the volume spanned by one query and multiple modality-specific keys is load-bearing for the central claim of handling 'any number of modalities.' In R^d the (hyper)volume of the parallelepiped formed by m > d vectors is identically zero (equivalently, sqrt(det(G)) = 0 for the Gram matrix G of the stacked vectors). With standard embedding dimensions (512–1024) this forces all attention scores to vanish for modality counts exceeding d, directly contradicting the 'any-order' and 'arbitrary modality counts' positioning. No alternative non-degenerate formulation (e.g., summed lower-order volumes, normalized exterior product, or learned projection) is indicated.

    Authors: We acknowledge that the referee's geometric observation is correct: when the total number of vectors (one query plus one key per modality) exceeds the embedding dimension d, they are necessarily linearly dependent and the parallelepiped volume is identically zero. However, this does not undermine the practical utility of VMA. In multimodal learning the number of modalities m is invariably small (typically 2–5 and almost never >10), while standard embedding dimensions satisfy d ≥ 512. Consequently the volume remains non-zero and the attention scores are well-defined for every realistic setting. The phrases 'any-order' and 'arbitrary modality counts' are intended to convey that VMA natively supports joint interactions of any order without incurring quadratic pairwise costs or relying on concatenation; they do not claim support for m > d. We will revise the abstract and introduction to state explicitly that the mechanism handles any number of modalities up to the embedding dimension, thereby removing any ambiguity while preserving the core technical contribution. revision: yes

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: new geometric attention formulation is independent of its inputs

full rationale

The paper proposes VMA as a novel cross-attention mechanism whose scores are defined directly from the volume spanned by a query vector and multiple modality-specific key vectors. This construction is introduced as an alternative to pairwise dot-product or concatenation-based attention and does not reduce to any fitted parameter, prior self-citation, or renamed empirical pattern. No equations or claims in the abstract equate the output volume computation to its own inputs by definition, nor does the architecture rely on a uniqueness theorem or ansatz imported from the authors' prior work. The derivation chain therefore remains self-contained and non-circular.

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

Abstract provides no information on free parameters, background axioms, or invented entities beyond naming the new attention mechanism itself.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GRAMformer: Any-Order Modality Interactions via Volumetric Multimodal Cross-Attention." pith.science (2026). https://pith.science/paper/AB3M2IAC

@misc{pith2026260606249,
  author       = {Pith},
  title        = {Pith review of: GRAMformer: Any-Order Modality Interactions via Volumetric Multimodal Cross-Attention},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AB3M2IAC}},
  note         = {Machine review of arXiv:2606.06249}
}
read the original abstract

Transformer-based multimodal models rely on attention mechanisms to integrate information across heterogeneous modalities. Despite their success, existing multimodal attention formulations compute their scores through collections of pairwise dot-product interactions or by concatenating all the modalities into the keys, even when multiple modalities should be jointly involved. As a consequence, current approaches either incur quadratic complexity in the number of modalities or fail to explicitly model interactions that depend on the joint configuration of multiple representations. In this work, we introduce the Volumetric Multimodal cross-Attention (VMA), a novel cross-attention mechanism in which attention scores are defined as a function of the joint geometry of a query and multiple modality-specific keys. VMA computes the volume spanned by query and key vectors across multiple modalities, capturing joint multimodal dependencies beyond pairwise similarity, enabling native modeling of any-order modality interactions. We integrate VMA into our novel multimodal transformer architecture, named GRAMformer, explicitly designed to integrate any number of modalities. We evaluate the proposed model on multimodal learning tasks, demonstrating improved effectiveness and efficiency.

Figures

Figures reproduced from arXiv: 2606.06249 by the authors.

Figure 1
Figure 1. Acc-2 in the MOSI dataset under different en [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Comparison among conventional multimodal cross-attention mechanisms (concatenation, pairwise dot [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Pairwise, triplet and Volumetric Multimodal At [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Proposed Volumetric Multimodal cross-Attention [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Proposed architecture with GRAMformer for [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Number of parameters, memory usage, and accuracy scaling the number of modalities for pairwise, concate [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

43 extracted references · 2 canonical work pages

  1. [1]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. u. Kaiser, and I. Polosukhin, “Attention is all you need,” inAdvances in Neural Information Processing Systems, vol. 30, 2017

    2017

  2. [2]

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

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

    2021

  3. [3]

    Flamingo: a visual language model for few-shot learning,

    J.-B. Alayrac, J. Donahue, P. Luc, A. Miech, I. Barr, Y . Hasson, K. Lenc, A. Mensch, K. Millicah, M. Reynolds, R. Ring, E. Rutherford, S. Cabi, T. Han, Z. Gong, S. Samangooei, M. Monteiro, J. Menick, S. Borgeaud, A. Brock, A. Nematzadeh, S. Sharifzadeh, M. Binkowski, R. Barreira, O. Vinyals, A. Zisserman, and K. Simonyan, “Flamingo: a visual language mod...

    2022

  4. [4]

    mPLUG-Owl3: Towards long image-sequence understanding in multi-modal large language models,

    J. Ye, H. Xu, H. Liu, A. Hu, M. Yan, Q. Qian, J. Zhang, F. Huang, and J. Zhou, “mPLUG-Owl3: Towards long image-sequence understanding in multi-modal large language models,” inInternational Conference on Learning Representation, 2025

    2025

  5. [5]

    J. Lu, D. Batra, D. Parikh, and S. Lee,ViLBERT: pretraining task-agnostic visiolinguistic representations for vision-and-language tasks. 2019

    2019

  6. [6]

    Multimodal transformer for unaligned multimodal language sequences,

    Y .-H. H. Tsai, S. Bai, P. P. Liang, J. Z. Kolter, L.-P. Morency, and R. Salakhutdinov, “Multimodal transformer for unaligned multimodal language sequences,” inAnnual Meeting of the Association for Computational Linguistics, pp. 6558–6569, 2019. 10 GRAMformer

    2019

  7. [7]

    Scaling rectified flow transformers for high-resolution image synthesis,

    P. Esser, S. Kulal, A. Blattmann, R. Entezari, J. M ¨uller, H. Saini, Y . Levi, D. Lorenz, A. Sauer, F. Boesel, D. Podell, T. Dockhorn, Z. English, and R. Rombach, “Scaling rectified flow transformers for high-resolution image synthesis,” inInternational Conference on Machine Learning, 2024

    2024

  8. [8]

    TACA: Rethinking cross-modal interaction in multimodal diffusion transformers,

    Z. Lv, T. Pan, C. Si, Z. Chen, W. Zuo, Z. Liu, and K.-Y . K. Wong, “TACA: Rethinking cross-modal interaction in multimodal diffusion transformers,” 2025

    2025

Show all 43 references
  1. [9]

    Videobert: A joint model for video and language representation learning,

    C. Sun, A. Myers, C. V ondrick, K. Murphy, and C. Schmid, “Videobert: A joint model for video and language representation learning,” inIEEE/CVF International Conference on Computer Vision (ICCV), October 2019

    2019

  2. [10]

    Unveiling the power of audio-visual early fusion transformers with dense interactions through masked modeling,

    S. Mo and P. Morgado, “Unveiling the power of audio-visual early fusion transformers with dense interactions through masked modeling,” inIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024

    2024

  3. [11]

    Language is not all you need: Aligning perception with language models,

    S. Huang, L. Dong, W. Wang, Y . Hao, S. Singhal, S. Ma, T. Lv, L. Cui, O. K. Mohammed, B. Patra, Q. Liu, K. Aggarwal, Z. Chi, J. Bjorck, V . Chaudhary, S. Som, X. Song, and F. Wei, “Language is not all you need: Aligning perception with language models,” inNeural Information P...

    2023

  4. [12]

    Video-LLaMA: An instruction-tuned audio-visual language model for video understanding,

    H. Zhang, X. Li, and L. Bing, “Video-LLaMA: An instruction-tuned audio-visual language model for video understanding,” inConference on Empirical Methods in Natural Language Processing: System Demonstrations, pp. 543–553, 2023

    2023

  5. [13]

    BLIP-2: bootstrapping language-image pre-training with frozen image encoders and large language models,

    J. Li, D. Li, S. Savarese, and S. Hoi, “BLIP-2: bootstrapping language-image pre-training with frozen image encoders and large language models,” inInternational Conference on Machine Learning, 2023

    2023

  6. [14]

    Cross-modal gated feature enhancement for multimodal emotion recognition in conversations,

    S. Zhao, J. Ren, and X. Zhou, “Cross-modal gated feature enhancement for multimodal emotion recognition in conversations,”Scientific Reports, no. 30004, 2025

    2025

  7. [15]

    Triplet attention: Rethinking the similarity in transformers,

    H. Zhou, J. Li, J. Peng, S. Zhang, and S. Zhang, “Triplet attention: Rethinking the similarity in transformers,” in ACM SIGKDD Conference on Knowledge Discovery & Data Mining, p. 2378–2388, 2021

    2021

  8. [16]

    MMT: Multi-way multi-modal transformer for multimodal learning,

    J. Tang, K. Li, M. Hou, X. Jin, W. Kong, Y . Ding, and Q. Zhao, “MMT: Multi-way multi-modal transformer for multimodal learning,” inInternational Joint Conference on Artificial Intelligence, IJCAI-22, pp. 3458–3465, 7 2022

    2022

  9. [17]

    Triplet attention transformer for spatiotemporal predictive learning,

    X. Nie, X. Chen, H. Jin, Z. Zhu, Y . Yan, and D. Qi, “Triplet attention transformer for spatiotemporal predictive learning,”IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), pp. 7021–7030, 2024

    2024

  10. [18]

    Gramian multimodal representation learning and alignment,

    G. Cicchetti, E. Grassucci, L. Sigillo, and D. Comminiello, “Gramian multimodal representation learning and alignment,” inInternational Conference on Learning Representations (ICLR), 2025

    2025

  11. [19]

    Learning transferable visual models from natural language supervision,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, G. Krueger, and I. Sutskever, “Learning transferable visual models from natural language supervision,” in International Conference on Machine Learning (ICML), 2021

    2021

  12. [20]

    CLAP: learning audio concepts from natural language supervision,

    B. Elizalde, S. Deshmukh, M. Al Ismail, and H. Wang, “CLAP: learning audio concepts from natural language supervision,” inIEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 1–5, IEEE, 2023

    2023

  13. [21]

    CLIP4Clip: An empirical study of clip for end to end video clip retrieval,

    H. Luo, L. Ji, M. Zhong, Y . Chen, W. Lei, N. Duan, and T. Li, “CLIP4Clip: An empirical study of clip for end to end video clip retrieval,”Neurocomputing, vol. 508, pp. 293–304, 2021

    2021

  14. [22]

    ImageBind one embedding space to bind them all,

    R. Girdhar, A. El-Nouby, Z. Liu, M. Singh, K. V . Alwala, A. Joulin, and I. Misra, “ImageBind one embedding space to bind them all,” inIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 15180– 15190, 2023

    2023

  15. [23]

    InternVideo2: Scaling video foundation models for multimodal video understanding,

    Y . Wang, K. Li, X. Li, J. Yu, Y . He, G. Chen, B. Pei, R. Zheng, J. Xu, Z. Wang, Y . Shi, T. Jiang, S. Li, H. Zhang, Y . Huang, Y . Qiao, Y . Wang, and L. Wang, “InternVideo2: Scaling video foundation models for multimodal video understanding,”ArXiv preprint: arXiv:2403.15377, 2024

    2024

  16. [24]

    Flowing from words to pixels: A noise-free framework for cross-modality evolution,

    Q. Liu, X. Yin, A. Yuille, A. Brown, and M. Singh, “Flowing from words to pixels: A noise-free framework for cross-modality evolution,” inComputer Vision and Pattern Recognition Conference, pp. 2755–2765, 2025

    2025

  17. [25]

    A triangle enables multimodal alignment beyond cosine similarity,

    G. Cicchetti, E. Grassucci, and D. Comminiello, “A triangle enables multimodal alignment beyond cosine similarity,” inNeural Information Processing Systems, 2025

    2025

  18. [26]

    Contrasting with symile: Simple model-agnostic representation learning for unlimited modalities,

    A. Saporta, A. M. Puli, M. Goldstein, and R. Ranganath, “Contrasting with symile: Simple model-agnostic representation learning for unlimited modalities,” inNeural Information Processing Systems (NeurIPS), 2024

    2024

  19. [27]

    Principled multimodal representation learning,

    X. Liu, X. Xia, S.-K. Ng, and T.-S. Chua, “Principled multimodal representation learning,”ArXiv preprint: arXiv:2507.17343, 2025

    2025

  20. [28]

    Quadruple attention in many-body systems for accurate molecular property predictions,

    J. Rao, D. Xu, W. Wei, Y . Chen, M. Yang, and Y . Yang, “Quadruple attention in many-body systems for accurate molecular property predictions,” inInternational Conference on Machine Learning, 2025. 11 GRAMformer

    2025

  21. [29]

    Matrix theory,

    F. R. Gantmacher, “Matrix theory,”Chelsea Publishing Company, 1959

    1959

  22. [30]

    M-SENA: An integrated platform for multimodal sentiment analysis,

    H. Mao, Z. Yuan, H. Xu, W. Yu, Y . Liu, and K. Gao, “M-SENA: An integrated platform for multimodal sentiment analysis,” inAssociation for Computational Linguistics: System Demonstrations, pp. 204–213, 2022

    2022

  23. [31]

    Gated attention for large language models: Non-linearity, sparsity, and attention-sink-free,

    Z. Qiu, Z. Wang, B. Zheng, Z. Huang, K. Wen, S. Yang, R. Men, L. Yu, F. Huang, S. Huang, D. Liu, J. Zhou, and J. Lin, “Gated attention for large language models: Non-linearity, sparsity, and attention-sink-free,” inNeural Information Processing Systems, 2025

    2025

  24. [32]

    LanguageBind: Extending video-language pretraining to n-modality by language-based semantic alignment,

    B. Zhu, B. Lin, M. Ning, Y . Yan, J. Cui, H. Wang, Y . Pang, W. Jiang, J. Zhang, Z. Li, W. Zhang, Z. Li, W. Liu, and L. Yuan, “LanguageBind: Extending video-language pretraining to n-modality by language-based semantic alignment,” inInternational Conference on Learning Represe...

    2024

  25. [33]

    What to align in multimodal contrastive learning?,

    B. Dufumier, J. C. Navarro, D. Tuia, and J.-P. Thiran, “What to align in multimodal contrastive learning?,” in International Conference on Learning Representations, 2025

    2025

  26. [34]

    Multimodal phased transformer for sentiment analysis,

    J. Cheng, I. Fostiropoulos, B. Boehm, and M. Soleymani, “Multimodal phased transformer for sentiment analysis,” inConference on Empirical Methods in Natural Language Processing, pp. 2447–2458, Nov. 2021

    2021

  27. [35]

    Joint fine-grained disentanglement and modal-agnostic fusion multi-task framework for multimodal sentiment analysis,

    Y . Zheng, C. Zhang, Y . Wen, X. Qi, and P. Zhang, “Joint fine-grained disentanglement and modal-agnostic fusion multi-task framework for multimodal sentiment analysis,” inACM International Conference on Multimedia in Asia, 2025

    2025

  28. [36]

    MultiBench: Multiscale benchmarks for multimodal representation learning,

    P. P. Liang, Y . Lyu, X. Fan, Z. Wu, Y . Cheng, J. Wu, L. Chen, P. Wu, M. A. Lee, Y . Zhu, R. Salakhutdinov, and L.-P. Morency, “MultiBench: Multiscale benchmarks for multimodal representation learning,” inNeural Information Processing Systems, 2021

    2021

  29. [37]

    Tensor fusion network for multimodal sentiment analysis,

    A. Zadeh, M. Chen, S. Poria, E. Cambria, and L. philippe Morency, “Tensor fusion network for multimodal sentiment analysis,” inConference on Empirical Methods in Natural Language Processing, 2017

    2017

  30. [38]

    Efficient low-rank multimodal fusion with modality-specific factors,

    Z. Liu, Y . Shen, V . B. Lakshminarasimhan, P. P. Liang, A. Zadeh, and L. philippe Morency, “Efficient low-rank multimodal fusion with modality-specific factors,” inAssociation for Computational Linguistics, 2018

    2018

  31. [39]

    MISA: Modality-invariant and -specific representations for multi- modal sentiment analysis,

    D. Hazarika, R. Zimmermann, and S. Poria, “MISA: Modality-invariant and -specific representations for multi- modal sentiment analysis,” inACM International Conference on Multimedia, 2020

    2020

  32. [40]

    Learning modality-specific representations with self-supervised multi-task learning for multimodal sentiment analysis,

    W. Yu, H. Xu, Z. Yuan, and J. Wu, “Learning modality-specific representations with self-supervised multi-task learning for multimodal sentiment analysis,” inAAAI Conference on Artificial Intelligence, 2021

    2021

  33. [41]

    Integrating multimodal information in large pretrained transformers,

    W. Rahman, M. K. Hasan, S. Lee, A. Bagher Zadeh, C. Mao, L.-P. Morency, and E. Hoque, “Integrating multimodal information in large pretrained transformers,” inAssociation for Computational Linguistics, 2020

    2020

  34. [42]

    Multimodal multi-loss fusion network for sentiment analysis,

    Z. Wu, Z. Gong, J. Koo, and J. Hirschberg, “Multimodal multi-loss fusion network for sentiment analysis,” in North American Chapter of the Association for Computational Linguistics, 2024

    2024

  35. [43]

    Enriching multimodal sentiment analysis through textual emotional descriptions of visual-audio content,

    S. Wu, D. He, X. Wang, L. Wang, and J. Dang, “Enriching multimodal sentiment analysis through textual emotional descriptions of visual-audio content,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 39, pp. 1601–1609, 2025. 12 GRAMformer Appendix A Table ...

    2025

Pith tools

Reviewed June 28, 2026 · model on record in the stance chip above.