Skip to content

Dataset · mlfoundations

1-trillion token multimodal PDF and web corpus

mlfoundations/MINT-1T-PDF-CC-2023-06

A large-scale open corpus of interleaved English text and images for multimodal pre-training.

What's inside

Part of the massive open multimodal MINT-1T dataset (totaling 1 trillion text tokens and 3.4 billion images), gathered from English PDF documents and arXiv research papers. The specific split contains between 100 billion and 1 trillion records. Created by a team from the University of Washington.

What you can do with it

  • Pre-train large multimodal models (VLM)
  • Train systems to understand complex layouts of research papers and PDF documents
Tasks
Image → text, Text generation
Languages
English
Size
100B–1T rows
Downloads this month
21.5K
Likes
10

How to load it

from datasets import load_dataset

ds = load_dataset("mlfoundations/MINT-1T-PDF-CC-2023-06")