The World’s Largest Multimodal Dataset | Episode 1
The formula for AI datasets
Frederik, Head of Machine Learning at Encord, introduces the world's largest multimodal AI dataset. This dataset is designed to foster the development of models that fuse many modalities, similar to the well-known CLIP model - the ignition for new AI applications requiring more than just single modalities. In this episode, Frederik explains the formula they used to build the dataset, divided into three stages: pre-training, high-quality post-training, and a quality evaluation set.
Speakers
Frederik Hvilshøj
ML Lead @ Encord
The World's Largest Multimodal AI Dataset
The open-source E-MM1 dataset has 100+ million groups of images, videos, text, audio and 3D point clouds, giving AI teams more training data for their AI models.
