Datasets:

Spawning

/

PD12M

like 145

Follow

Spawning 45

I've made a training code that utilizes multi-gpus. I first tried getting the images with the url but I realized this is a bottleneck for multi-gpu training. So I've made a script that downloads all the images and adds it to a new column, but it shows it would take 300 hours to download the whole thing. I would like to know what is the best way to utilize this kind of dataset, I don't want to spend almost two weeks to download the images if possible.

Thanks.