Skip to content

Dataset · nyu-visionx

Multimodal model alignment dataset collection

nyu-visionx/Cambrian-Alignment

100K to 1M English image-text question-answer pairs for VLM alignment training.

What's inside

This English dataset combines visual question answering (VQA) data from popular open sources: LLaVA, Mini-Gemini, Allava, and ShareGPT4V. It contains between 100K and 1M image-text examples gathered for the Cambrian project.

What you can do with it

  • Perform alignment training for multimodal models (connecting visual encoder and language model)
  • Fine-tune VLMs to answer image-based questions
Tasks
Visual question answering, Question answering
Languages
English
Size
100K–1M rows
Downloads this month
6.6K
Likes
39

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("nyu-visionx/Cambrian-Alignment")