Skip to content

Dataset · hiyouga

Geometric image-text VQA pairs

hiyouga/geometry3k

A VQA dataset with geometric diagram images and English text question-answer pairs, containing about 1 to 10k examples.

What's inside

A visual question answering dataset containing images and English text, ranging from 1k to 10k examples. The data was converted from the InterGPS project using a script published in the card. Format is a Hugging Face dataset with images and question-answer pairs.

What you can do with it

  • Fine-tune VLMs for geometric diagram VQA
  • Evaluate models on visual questions
  • Train LoRA for image-question-answer tasks
Tasks
Visual question answering
Languages
English
Size
1K–10K rows
Downloads this month
30.6K
Likes
87

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("hiyouga/geometry3k")