MINT-1T
MINT-1T, a massive diverse dataset for training multi-modal models, enhances image and text understanding across various fields.
What is MINT-1T?
MINT-1T is a large scale multimodal dataset created by Salesforce AI. It contains over a trillion text tokens and 3.4 billion images sourced from various documents like HTML, PDFs, and ArXiv papers. This extensive dataset enhances diversity and supports advanced research in areas like multimodal learning and deep learning models.