- Rating
- 2
- Confidence
- 3
- Informativeness
- 1
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper proposes to use polar coordinates instead of what the authors call “ambient coordinates”. The authors analyze the precision loss if the “ambient coordinates” are used and give formulae of polar operations. Based on these building blocks, the paper proposes a hyperbolic vision transformer called the polar-ViT, with fully connected layers, attention and positional encoding in polar coordinates. The authors use some image classification datasets and compare polar-ViT with some other vision transformer models.
Critical strengths
- This paper has a good and comprehensive survey of previous related works.
- This paper is well-organized and generally easy to follow.
- The paper focuses on vision transformer models and in the experiments compares polar-ViT with many other ViT models.
Critical weaknesses
- The contribution of the paper is overly incremental. Although using polar coordinates is not well-studied for hyperbolic neural networks before, the transform between “ambient coordinates” and polar coordinates is straightforward. The precision loss is an extension of (Mishne et al., 2023); the Lorentz fully connected layer and attention layer follow (Chen et al., 2022); the horosphere idea has been used in (Chen et al., 2026); the layer normalization is similar to batch normalization of (Chen et al., 2025).
- The experiments only consider “for 100 epochs with the HexFormer training recipe”. It is not a natural setting because in image classification one can select number of epochs, etc., using validation.
- The paper is not comparing polar-ViT with the state-of-the-art Euclidean ViT. It is unclear why a hyperbolic version of ViT would be expected to work better than a Euclidean counterpart for Euclidean image datasets.
Questions for authors
See "weaknesses".
4 comments
3. Comparison with state-of-the-art Euclidean ViTs, and why a hyperbolic model should help on image data
Polar-ViT modifies the plain ViT, so the matched comparison is the plain ViT at equal depth, width, heads, patch size and recipe. HexFormer makes the same comparison. We also follow LViT [10] and compare with a stronger fine-tuned Euclidean ViT (Flat-ViT, Table 3). For additional context, published from-scratch Euclidean ViTs of about 3M parameters at 100 epochs reach 73.81 (ViT in [6]), 74.64 [5], and 57.07 and 60.74 on Tiny-ImageNet. Published numbers above ours come from models three times larger (SL-CaiT [6]), from convolutional tokenizers trained for 300 to 5000 epochs [11], or from an added 200-epoch self-supervised stage [7]. These additions are independent of the geometry and can be applied to Polar-ViT in the same way. On ImageNet-1k, where we trained both models with one known published recipe, the margin is 3.5 points at the Tiny size and 2.2 at the Small size (Table 2), and it carries over to fine-tuning (Table 3).
As for why a hyperbolic model should help on image data, this is a well studied and discussed setting in the literature. The main consensus is that the geometry applies to the learned representation not to the pixels. A Euclidean ViT also maps patches into a feature space unrelated to pixel geometry, and the question is which geometry fits that space. Image tasks can carry hierarchical structure, for example, ImageNet classes are hierarchies of WordNet, and CIFAR-100 groups its classes into 20 superclasses, and CUB and Aircraft, where Polar-ViT gains 3.9 and 5.3 points over the Euclidean ViT after fine-tuning at the Tiny size (Table 3), come with label hierarchies used by H-CAST [9]. CNN features of image datasets are also moderately tree-like, and Khrulkov et al. study this using gromov hyperbolicity distortion measurements. They found a relative δ-hyperbolicity of 0.17 to 0.26 on CIFAR-10, CIFAR-100, CUB and MiniImageNet [12], where 0 is a tree and 1 is not hyperbolic at all. We do not claim that negative curvature helps on every image task but the hyperbolicity values hint at possible benefits from using that space. The gain of Polar-ViT comes mainly from components that depend on the curvature. The horospherical shift adds 2.1 points (Table 5), and under it the query-key distance grows logarithmically with the token gap (Eq. 11). In flat space the same construction is a translation, and the distance grows linearly.
References
- Mishne et al. The numerical stability of hyperbolic representation learning. ICML 2023.
- Chen et al. Fully hyperbolic neural networks. ACL 2022.
- Chen et al. Hyperbolic Busemann neural networks. CVPR 2026.
- Chen et al. Gyrogroup batch normalization. ICLR 2025.
- Alyoussef et al. HexFormer: Hyperbolic vision transformer with exponential map aggregation. arXiv:2601.19849.
- Lee et al. Vision transformer for small-size datasets. arXiv:2112.13492.
- Gani et al. How to train vision transformer on small-scale datasets? BMVC 2022.
- Beyer et al. Better plain ViT baselines for ImageNet-1k. arXiv:2205.01580.
- Park et al. Visually consistent hierarchical image classification. ICLR 2025.
- He et al. HyperCore: The core framework for building hyperbolic foundation models with comprehensive modules. arXiv:2504.08912.
- Hassani et al. Escaping the big data paradigm with compact transformers. arXiv:2104.05704.
- Khrulkov et al. Hyperbolic image embeddings. CVPR 2020.
- van der Klis et al. Fast and geometrically grounded Lorentz neural networks. arXiv:2601.21529.
Thanks for the rebuttal
Thanks for the rebuttal.
- I think your rebuttal exactly shows the incremental nature of the work.
- Thanks for the clarification. I now agree that the setting is okay.
- It is unclear how "the geometry applies to the learned representation not to the pixels" is reflected in your model. It seems to me that your neural layers directly apply to the pixel space.
Thank you for the reply. We are glad the training setting is resolved.
3. Where the geometry acts
There are two main approaches to applying hyperbolic models to images. Hybrid models keep a Euclidean encoder and map its features to hyperbolic space before the head. Khrulkov et al. [1] do this for example, and Guo et al. [2] describe the same design on MNIST, CIFAR-10/100 and ImageNet. We also followed this example for other settings in the rebuttal experiments we conducted.
On the other hand, fully hyperbolic models place the image on the manifold before the first learned layer. This has been done in Poincaré ResNet [3] which learns "directly from the pixel-level", HCNN [4] treats an image as "an ordered set of color vectors", and HCNN+ [5], FGG-LNN [6] and ILNN [7] also train fully hyperbolic image classifiers. The Hyperbolic U-Net [8] does the same for segmentation, "operating entirely in the Poincaré ball". For transformers, HexFormer [9] and LViT [10] project images "into the Lorentz space" before a Lorentz convolutional patch embedding.
Our approach follows the second group and uses the same lift as HCNN where we keep the flattened patch as the space coordinates and calculate the time coordinate with no exponential map. We believe this preserves the features without assuming direct hyperbolicity and transforming them accordingly.
However, based on you suggestion we attempted to test the effect of the geometry in early stages of the model. We replaced the patch embedding and the first blocks with their Euclidean counterparts and kept the rest of the model hyperbolic (CIFAR-100, recipe of Table 1, same seed).
| Euclidean part | top-1 |
|---|---|
| none (Polar-ViT, same seed) | 80.04 |
| patch embedding | 80.05 |
| patch embedding + block 1 | 79.14 |
| patch embedding + blocks 1 to 3 | 79.15 |
| patch embedding + blocks 1 to 6 | 78.34 |
| patch embedding + blocks 1 to 9 (Euclidean encoder, hyperbolic pooling and classifier) | 76.18 |
Making the direct pixel embedding layer Euclidean changes nothing almost (80.05 against 80.04), so the gain does not come from treating pixels as hyperbolic points. It comes from the blocks that act on learned features. When all the nine encoder blocks are Euclidean the model loses 3.9 points, and the last three blocks account for 2.2 of them. Early blocks contribute less, but block 1 still contributes 0.9.
To explain this trend we measured the relative δ-hyperbolicity of the token features at every depth, following Khrulkov et al. [1]. Lower means more tree-like.
| site | Euclidean ViT | Polar-ViT |
|---|---|---|
| patch embedding | 0.301 | 0.186 |
| block 1 | 0.279 | 0.149 |
| block 2 | 0.272 | 0.124 |
| block 3 | 0.267 | 0.105 |
| block 4 | 0.272 | 0.101 |
| block 5 | 0.269 | 0.092 |
| block 6 | 0.244 | 0.085 |
| block 7 | 0.239 | 0.072 |
| block 8 | 0.225 | 0.066 |
| block 9 | 0.202 | 0.077 |
The Euclidean ViT has no hyperbolic component, but its features become more tree-like with depth (0.30 to 0.20), and its last block falls in the range Khrulkov et al. report for CNN features (0.17 to 0.26). This agrees with the ablation and could explain why removing the hyperbolic blocks deeper in hurt performance more. The early features are less tree-like but not flat, which fits the smaller gain of the early blocks.
This deserves a wider study across models and datasets. The depth/hyperbolicity trend also suggests using a different curvature per block. This normally could lead to instability since the curvature greatly affects the maximum radius we can represent. Learning the curvature without warmup in Table B of our reply to Reviewer HKB5 turned non-finite in epoch 1.
A second, smaller source of gain reported in the literature is that hyperbolic space represents data better at small embedding dimensions [11, 12]. HCNN shows this for images, where the hyperbolic models outperform the Euclidean ones at low dimensions of the final ResNet block. This could explain why it worked better for other methods. We will add both tables to the appendix.
References
- Khrulkov et al. Hyperbolic image embeddings. CVPR 2020.
- Guo et al. Clipped hyperbolic classifiers are super-hyperbolic classifiers. CVPR 2022.
- van Spengler et al. Poincaré ResNet. ICCV 2023.
- Bdeir et al. Fully hyperbolic convolutional neural networks for computer vision. ICLR 2024.
- Bdeir et al. Robust hyperbolic learning with curvature-aware optimization. NeurIPS 2026.
- van der Klis et al. Fast and geometrically grounded Lorentz neural networks. arXiv:2601.21529.
- Shi et al. Intrinsic Lorentz neural network. ICLR 2026.
- Mishra et al. Hyperbolic U-Net for robust medical image segmentation. MIDL 2026.
- Alyoussef et al. arXiv:2601.19849.
- He et al. arXiv:2504.08912.
- Nickel and Kiela. Poincaré embeddings for learning hierarchical representations. NeurIPS 2017.
- Sala et al. Representation tradeoffs for hyperbolic embeddings. ICML 2018.
We thank the reviewer for their time. Below we highlight the differences between us and established literature and hopefully answer your other points. All of these improve clarity and will be added to the paper as well. In the rebuttals above we also try our polar forms and ablate in multiple other settings which we would appreciate you taking into consideration as well.
1. Incremental over Mishne et al. [1] and Chen et al. [2, 3, 4]
The change of coordinates is straightforward but causes gradient issues in backpropagation. The main contribution is the derivation of cancellation-free forms of the operations (Eqs. 5 to 7) and of their gradients (Appendix D.1), the handling of the origin, and the precision bounds (Propositions 1 and 2). At the origin the direction of a point is not defined and its gradient grows as 1/‖x‖, so Eq. 8 takes the Klein vector, which stays smooth there, and Appendix B gives the origin cases of the tangent inner product and the centroid. As for the other conributions:
We will make these four distinctions explicit in the related work.
2. Training for 100 epochs with the HexFormer recipe is not a natural setting
The HexFormer [5] recipe matches the small-dataset ViT protocol of [6], with depth 9, width 192, 100 epochs, batch 128 and 10 warm-up epochs. Gani et al. [7] call it the default training setting for small-scale datasets. We use it so that our rows share the protocol of the Euclidean ViT and HexFormer rows of Table 1. If you mean selecting on a validation split, we do this for ImageNet-1k, for the segmentation experiments of Appendix F, and for the graph and ResNet experiments in the rebuttals to other reviewers above.When we do report the last epoch it is to match the published baselines and compare fairly against them.
Additionally, the ImageNet experiments follow two further published recipes, [8] for ImageNet-1k and [9] for fine-tuning. These are standard published recipes for Euclidean ViT training.