Dataset - LDM

Winoground

The Winoground dataset consists of 400 items, each containing two image-caption pairs (I0, C0), (I1, C1).
- Dataset
- JSON
DataComp-10M

DataComp-10M is used as a pretraining dataset
- Dataset
- JSON
CC3M and CC12M

CC3M and CC12M are used as datasets for training and evaluation
- Dataset
- JSON

You can also access this registry using the API (see API Docs).

3 datasets found