Skip to content

Dataset · jasperai

103.8M MONET Image-Text Pairs

jasperai/monet

Large curated dataset of 103.8 million image-text pairs for text-to-image training.

What's inside

MONET (Massive, Open, Non-redundant and Enriched Text-to-image) is a curated image-text dataset for T2I. Collected from 2.9 billion raw pairs across nine open sources (6 real and 3 synthetic) after safety filtering, domain filtering, exact and near-duplicate removal, and re-captioning. The result is 103.8 million high-quality pairs; language is en.

What you can do with it

  • train a T2I diffusion model
  • fine-tune LoRA for T2I
  • evaluate zero-shot image classification and feature extraction
Tasks
Text → image, Image feature extraction, Zero shot image classification
Languages
English
Size
100M–1B rows
Downloads this month
83.3K
Likes
195

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("jasperai/monet")