BLIP3o/

BLIP3o-Pretrain-JourneyDB

This collection contains 4 million JourneyDB images.

from huggingface_hub import snapshot_download

snapshot_download( repo_id="BLIP3o/BLIP3o-Pretrain-JourneyDB", repo_type="dataset" )

You don’t need to unpack the .tar archives, use WebDataset support in 🤗datasets instead: from datasets import load_dataset import glob

data_files = glob.glob("/your/data/path/*.tar")… See the full description on the dataset page: https://huggingface.co/datasets/BLIP3o/BLIP3o-Pretrain-JourneyDB.