Skip to content

Dataset · 5CD-AI

Vietnamese image descriptions and QA pairs

5CD-AI/Viet-LAION-Gemini-VQA

Over 5M descriptions and QA pairs for 843k images in Vietnamese to train multimodal models.

What's inside

Based on 843,529 images from the Vietnamese subset of LAION-2B, including photos of news, sports, travel, goods, clothing, transport, as well as sketches and charts. Using the Gemini-1.5-Flash-002 model, 5,061,174 detailed descriptions and QA pairs were generated in Vietnamese. The dataset suits visual question answering (VQA) tasks.

What you can do with it

  • Training multimodal models (VLMs) in Vietnamese language comprehension
  • Tuning chart, product, and sketch recognition systems from images
Tasks
Visual question answering
Languages
Vietnamese
Size
100K–1M rows
Downloads this month
67
Likes
48
Access
on request on Hugging Face

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("5CD-AI/Viet-LAION-Gemini-VQA")