---
title: SPaRC
canonical_url: "https://www.modelscope.cn/datasets/evalscope/SPaRC"
md_url: "https://www.modelscope.cn/datasets/evalscope/SPaRC.md"
repository: evalscope/SPaRC
last_updated: 2026-10-15
license: cc-by-sa-4.0
storage_size: "40 MB"
downloads: 160
stars: 0
---

# SPaRC

> SPaRC - evalscope 在 ModelScope 开源的数据集。Paper • Website • Code • Spatial-Gym • Solver • Generator

evalscope/SPaRC 是 ModelScope 魔搭社区上的数据集，存储大小 40 MB，采用 cc-by-sa-4.0 许可。

- **Repository**: evalscope/SPaRC
- **License**: cc-by-sa-4.0
- **Storage size**: 40 MB
- **Downloads**: 160
- **Stars**: 0
- **Last updated**: 2026-10-15

Source: https://www.modelscope.cn/datasets/evalscope/SPaRC

---

<p align="center">
  <img src="logo.png" alt="SPaRC Logo" width="128"/>
</p>

# SPaRC Dataset

[Paper](https://aclanthology.org/2025.emnlp-main.526/) • [Website](https://sparc.gipplab.org/) • [Code](https://github.com/lkaesberg/SPaRC) • [Spatial-Gym](https://github.com/lkaesberg/Spatial-Gym) • [Solver](https://github.com/lkaesberg/SPaRCSolver) • [Generator](https://github.com/lkaesberg/SPaRCPuzzleGenerator)


A grid-based puzzle dataset for benchmarking LLMs spatial reasoning capabilities.

## Data Schema

Each record (JSON) includes:

* `id` (string): unique puzzle identifier
* `difficulty_level` (int) & `difficulty_score` (float)
* `grid_size`: `{ "height": H, "width": W }`
* `polyshapes`: JSON string mapping shape IDs to binary grids
* `puzzle_array`: 2D array with cell codes (e.g., `S`, `E`, `+`, `P-O-112`)
* `solution_count` (int) and `solutions` list (with `index`, `path`, `pathLength`)
* `text_visualization`: YAML-style summary

## Sample Entry

```json
{
  "id": "10e68dc3a6fbfcdf",
  "difficulty_level": 3,
  "difficulty_score": 2.3261,
  "grid_size": {"height":3,"width":4},
  "polyshapes":"{\"112\":[[0,1,...]]}",
  "puzzle_array":[
    ["+","+","+","E"],
    ["+","P-O-112","+","o-O"],
    ["S","N","+","G"]
  ],
  "solution_count": 10,
  "solutions": [ { "index":0, "pathLength":34, ... } ],
  "text_visualization": "..."
}
```

## Citation Information
If you use the dataset in any way, please cite the following paper. Preprint: https://arxiv.org/abs/2505.16686

```bib
@inproceedings{kaesberg-etal-2025-sparc,
  title     = "{SP}a{RC}: A Spatial Pathfinding Reasoning Challenge",
  author    = "Kaesberg, Lars Benedikt and Wahle, Jan Philip and Ruas, Terry and Gipp, Bela",
  booktitle = "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing",
  month     = nov,
  year      = "2025",
  address   = "Suzhou, China",
  publisher = "Association for Computational Linguistics",
  url       = "https://aclanthology.org/2025.emnlp-main.526/",
  doi       = "10.18653/v1/2025.emnlp-main.526",
  pages     = "10359--10390"
}
```

## Related Papers

Two follow-up papers use the SPaRC puzzles. BibTeX for both is on the [website](https://sparc.gipplab.org/#citation).

- [Mind the Gap Between Spatial Reasoning and Acting! Step-by-Step Evaluation of Agents With Spatial-Gym](https://arxiv.org/abs/2604.09338) turns SPaRC into a step-by-step environment for agents.
- [Is Visual Prompting All You Need? Studying VLM Spatial Reasoning under Progressive Visual Scaffolds](https://arxiv.org/abs/2608.21170) studies vision-language models on SPaRC with visual scaffolds. Findings of EMNLP 2026.