metadata

dataset_info:
  features:
    - name: question
      dtype: string
    - name: options
      sequence: string
    - name: answer
      dtype: string
    - name: image
      dtype: image
  splits:
    - name: test
      num_bytes: 456758092.75
      num_examples: 2114
  download_size: 115580089
  dataset_size: 456758092.75
configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*
task_categories:
  - visual-question-answering
language:
  - en
tags:
  - flowchart
  - diagram
license: cc-by-nc-4.0

This dataset

This is the cleaned test set of Flowlearn.

Citation

Please cite the paper if you use this dataset.

@misc{pan2024flowlearnevaluatinglargevisionlanguage,
      title={FlowLearn: Evaluating Large Vision-Language Models on Flowchart Understanding}, 
      author={Huitong Pan and Qi Zhang and Cornelia Caragea and Eduard Dragut and Longin Jan Latecki},
      year={2024},
      eprint={2407.05183},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2407.05183}, 
}