The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.
Description
120,000 Questions - Multi-discipline Multi-modal Understanding & Reasoning Data, covering six major academic fields such as art, engineering, and medicine, as well as various specialized sub-domains. It includes professional-level mixed text and visual questions (e.g., charts, engineering drawings, and art analysis) that have been quality-checked by specialists. This dataset enables deep understanding of textual and visual information and can enhance the logical reasoning and knowledge application capabilities of large models.
For more details, please refer to the link: https://www.nexdata.ai/datasets/llm/1751?source=Huggingface
Specifications
Data size
120,000 questions
Image resolution
short side resolution ≥ 500 pixels
Subject categories
arts, business, science, medicine,humanities and social sciences, engineering
QA length
question length ≥ 10 Chinese characters, and the answer and analysis length ≥ 40 characters
Collection equipment
mobile phone, scanner
Language
Chinese
Diversity
multiple disciplines, multiple image types, multiple types of questions
Data format
.jpg, .png, .josn
Licensing Information
Commercial License
- Downloads last month
- 9