Add image-6 benchmark dataset

#2
by jaepoong - opened

Adds the image-6 text removal split with source, mask, and target image assets, and updates the dataset card config list/schema.

Closing this superseded PR because its parquet shard referenced local filesystem paths instead of embedding image bytes. Replaced by discussion #3, which embeds source, mask, and target image assets.

Superseded by #3.

jaepoong changed pull request status to closed

Sign up or log in to comment