File size: 2,571 Bytes
2b4c19c 5ca84c3 6b3d6a7 96f48f3 2b4c19c 6b3d6a7 2b4c19c 8f8e335 05b94ed 8f8e335 6b3d6a7 8f8e335 05b94ed 38f3b7d 05b94ed 8f8e335 6b3d6a7 8f8e335 18dc923 5ca84c3 96f48f3 6b3d6a7 8f8e335 6b3d6a7 8f8e335 5ca84c3 8f8e335 6b3d6a7 8f8e335 05b94ed 7f9283e 6b3d6a7 7f9283e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 | ---
configs:
- config_name: phase_variation
data_files: phase_variation-*.parquet
default: true
- config_name: size_variation
data_files: size_variation-*.parquet
- config_name: contrastive_probing
data_files: contrastive_probing.parquet
dataset_info:
features:
- name: index
dtype: int64
- name: image
dtype: image
- name: question
dtype: string
- name: answer
dtype: string
---
# SpatialTunnel
SpatialTunnel is a Blender-rendered diagnostic dataset for studying how vision-language models represent spatial relations internally. It was introduced in **Why Far Looks Up: Probing Spatial Representation in Vision-Language Models** ([arXiv:2605.30161](https://arxiv.org/abs/2605.30161)).
## Resources
- [Project page](https://cheolhong0916.github.io/whyfarlooksup.github.io/)
- [Contrastive-probing code](https://github.com/cheolhong0916/contrastive-probing)
- [SpatialTunnel generation code](https://github.com/cube-c/spatialtunnel-dataset-gen)
## Dataset Configurations
| Config | File | Rows | Description |
|---|---|---|---|
| `phase_variation` | `phase_variation-*.parquet` | 12,288 | Binary depth-comparison questions with controlled angular-position variation. |
| `size_variation` | `size_variation-*.parquet` | 4,400 | Binary depth-comparison questions under controlled object-size variation. |
| `contrastive_probing` | `contrastive_probing.parquet` | 1,200 | Balanced spatial-relation questions for contrastive probing. |
## Format
All configurations use Parquet files with the following columns:
For `phase_variation` and `size_variation`, each rendered image is paired with four binary VQA questions that vary the object order and comparison direction (`closer`/`farther`).
| Column | Description |
|---|---|
| `index` | Row index within the selected config. |
| `image` | PNG image stored as a Hugging Face `Image` feature. |
| `question` | Spatial question to ask the model. |
| `answer` | Ground-truth answer for the row. |
`contrastive_probing.tsv` is a TSV file for compatibility with the [contrastive-probing code](https://github.com/cheolhong0916/contrastive-probing).
---
## Citation
If you use this dataset, please cite our paper.
```bibtex
@article{min2026whyfarlooksup,
title = {Why Far Looks Up: Probing Spatial Representation in Vision-Language Models},
author = {Min, Cheolhong and Jung, Jaeyun and Lee, Daeun and Jeon, Hyeonseong and
Su, Yu and Tremblay, Jonathan and Song, Chan Hee and Park, Jaesik},
journal = {arXiv preprint arXiv:2605.30161},
year = {2026},
}
```
|