Dataset · OpenGVLab
Large multimodal reasoning dataset for InternVL
OpenGVLab/MMPR
Millions of examples for visual logical reasoning training that improved InternVL2.5 and InternVL3 quality.
What's inside
A large, high-quality English dataset for visual question answering and multimodal reasoning tasks (ranging from 1 to 10 million records). This corpus was used to train the InternVL2.5 and InternVL3 model lines, which showed notable growth on the OpenCompass benchmark.
What you can do with it
- Training VLM models for complex visual reasoning and image question answering
- Improving the accuracy of multimodal models on logical and analytical benchmarks
- Tasks
- Visual question answering
- Languages
- English
- Size
- 1M–10M rows
- Downloads this month
- 883
- Likes
- 52
How to load it
from datasets import load_dataset
ds = load_dataset("OpenGVLab/MMPR")