---
dataset_info:
  features:
  - name: year
    dtype: string
  - name: exam_format
    dtype: string
  - name: subject
    dtype: string
  - name: exam_code
    dtype: string
  - name: question_number
    dtype: string
  - name: question
    dtype: string
  - name: choices
    sequence: string
  - name: answer_key
    dtype: string
  - name: answer
    dtype: string
  splits:
  - name: test
    num_bytes: 1634687
    num_examples: 3722
  - name: train
    num_bytes: 15348
    num_examples: 35
  - name: val
    num_bytes: 339548
    num_examples: 765
  download_size: 1784571
  dataset_size: 1989583
configs:
- config_name: default
  data_files:
  - split: test
    path: data/test-*
  - split: train
    path: data/train-*
  - split: val
    path: data/val-*
---
> **Note**: I do **not** own this dataset. All credit goes to the original authors.  
> If you use this dataset, please cite the original paper:  
> DOI: [10.1145/3628797.3628837](https://doi.org/10.1145/3628797.3628837); arXiv: [arXiv:2310.12059](https://arxiv.org/abs/2310.12059)  
> The original dataset on Hugging Face: https://huggingface.co/datasets/uitnlp/ViGEText_17to23