Sora Explore Dataset is a corpus of 1,053,676 Sora-generated images for
synthetic-image detection research. The release preserves
206,325,340,988 bytes of original WebP image data without transcoding.
The dataset is stored in 105 Parquet shards, one for each original Sora export
ZIP. train-00000-of-00105.parquet corresponds to source shard 1 and
train-00104-of-00105.parquet corresponds to source shard 105. Each row embeds… See the full description on the dataset page:
https://huggingface.co/datasets/abrar71/sora-explore-dataset.
