This is a report of a img2dataset run on 10 workers with 16 cores to download the 2.3B samples of laion2B english. It took 3 days including 12h of down time due to a necessary fix for s3 auth. Total Objects: 694048 Total Size: 84.8 TiB

Downloading https://huggingface.co/datasets/laion/laion-high-resolution 170M images in >= 1024x1024

comparison of before/after on laion400m ; hardware 2Gbps 16 cores ; dataset laion400m

Downloading 1.2B samples where the language has not been identified

img2dataset download of laion2B-multi