SPaRC Dataset

Website • Solver • Generator

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

{
  "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

@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 = "10370--10401"
}