We-Math/README.md

metadata

license: cc-by-nc-4.0
dataset_info:
  features:
    - name: ID
      dtype: string
    - name: split
      dtype: string
    - name: knowledge concept
      dtype: string
    - name: question
      dtype: string
    - name: option
      dtype: string
    - name: answer
      dtype: string
    - name: image_path
      dtype: image
    - name: key
      dtype: string
    - name: question number
      dtype: int64
    - name: knowledge concept description
      dtype: string
  splits:
    - name: testmini
      num_bytes: 44509869
      num_examples: 1740
  download_size: 23075805
  dataset_size: 44509869
task_categories:
  - question-answering
  - text-generation
language:
  - en
tags:
  - LLM
  - NLP
  - CV
size_categories:
  - 1K<n<10K

Dataset Card for WE-MATH (ACL 2025)

GitHub | Paper | Website

Inspired by human-like mathematical reasoning, we introduce We-Math, the first benchmark specifically designed to explore the problem-solving principles beyond the end-to-end performance. We meticulously collect and categorize 6.5K visual math problems, spanning 67 hierarchical knowledge concepts and 5 layers of knowledge granularity.

Citation

If you find the content of this project helpful, please cite our paper as follows:

@article{qiao2024we,
  title={We-Math: Does Your Large Multimodal Model Achieve Human-like Mathematical Reasoning?},
  author={Qiao, Runqi and Tan, Qiuna and Dong, Guanting and Wu, Minhui and Sun, Chong and Song, Xiaoshuai and GongQue, Zhuoma and Lei, Shanglin and Wei, Zhe and Zhang, Miaoxuan and others},
  journal={arXiv preprint arXiv:2407.01284},
  year={2024}
}