README.md

metadata

dataset_info:
  features:
    - name: index
      dtype: int64
    - name: question
      dtype: string
    - name: hint
      dtype: string
    - name: A
      dtype: string
    - name: B
      dtype: string
    - name: C
      dtype: string
    - name: D
      dtype: string
    - name: answer
      dtype: string
    - name: category
      dtype: string
    - name: image
      dtype: image
    - name: source
      dtype: string
    - name: 12-category
      dtype: 'null'
    - name: comment
      dtype: string
  splits:
    - name: test
      num_bytes: 103342891
      num_examples: 4324
  download_size: 101580105
  dataset_size: 103342891
configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*

MMBench_Greek

This dataset is a Greek translation of MMBench, a multiple-choice benchmark for evaluating multimodal and Vision-Language Models.

The goal of this version is to make MMBench usable as an evaluation benchmark for Greek-capable VLMs.

Dataset Description

The dataset keeps the same structure and column names as the original MMBench_dev dataset. The text fields have been translated into Greek, while metadata fields are kept unchanged.

Each example contains an image, a question, optional hint text, multiple-choice answer options, the correct answer label, and category/source metadata.

Columns

• index: example index
• question: the question
• hint: optional hint
• A: answer option A
• B: answer option B
• C: answer option C
• D: answer option D
• answer: correct answer
• image: the image associated with the question
• source: source dataset
• l2-category: higher-level category
• comment: optional comment

Intended Use

This dataset is intended for evaluating Greek Vision-Language Models on multimodal multiple-choice question answering.

A typical evaluation prompt can ask the model to inspect the image, read the Greek question and answer options, and output only the correct option label.

Citation

Please cite the original MMBench paper when using this dataset:

@article{MMBench,
  title={MMBench: Is Your Multi-modal Model an All-around Player?},
  author={Liu, Yuan and Duan, Haodong and Zhang, Yuanhan and Li, Bo and Zhang, Songyang and Zhao, Wangbo and Yuan, Yike and Wang, Jiaqi and He, Conghui and Liu, Ziwei and Chen, Kai and Lin, Dahua},
  journal={arXiv preprint arXiv:2307.06281},
  year={2023}
}