Skip to content

Dataset · mlfoundations

MINT-1T multimodal corpus chunk (PDF and web data)

mlfoundations/MINT-1T-PDF-CC-2023-23

Part of a giant open dataset with interleaved texts and images from PDFs and the web.

What's inside

This is a separate split (1 to 10 million records) of the large-scale English-language MINT-1T dataset created by the University of Washington team. The entire project combines 1 trillion text tokens and 3.4 billion images interspersed, including previously unused sources such as PDF documents and ArXiv articles. The set is intended for research in multimodal pre-training.

What you can do with it

  • Pre-training multimodal models (VLM) that understand coherent text and images
  • Training information extraction systems and complex PDF document analysis
Tasks
Image → text, Text generation
Languages
English
Size
1M–10M rows
Downloads this month
20.6K
Likes
10

How to load it

from datasets import load_dataset

ds = load_dataset("mlfoundations/MINT-1T-PDF-CC-2023-23")