Skip to content

Dataset · Vision-CAIR

Vision-language alignment data for MiniGPT-4

Vision-CAIR/cc_sbu_align

A compact image dataset from the MiniGPT-4 project designed to align visual data with language models.

What's inside

Contains fewer than 1,000 image examples prepared by researchers at KAUST for the MiniGPT-4 project. The data was used during the alignment stage to train the model to hold conversations about images.

What you can do with it

  • Fine-tuning and adapting multimodal models (VLMs) for image-based dialogue
  • Reproducing the MiniGPT-4 training pipeline
Size
up to 1K rows
Downloads this month
748
Likes
45

How to load it

from datasets import load_dataset

ds = load_dataset("Vision-CAIR/cc_sbu_align")