A simple image text pair dataset I made by selecting image url - text pairs from various public datasets such as Laion-2b, cc12m, cc3m, Coyo-700m as well as several public synthetic datasets such as Leonardo and Playground-liked.
For training diffusion model pretraining