Multimodal Dataset
Domain: Data & Datasets
🎯 Motivation
A dataset containing multiple data types: text + images, text + audio, text + video. For training multimodal models.
📋 Concrete Example
Multimodal: 100K items, each with: an image, a caption, and a question-answer pair. The model learns to understand relationships between images and text.
🔗 Analogy
A textbook with text and pictures - you learn from both together. A multimodal dataset has multiple data types for the model to learn from.