This dataset repo aggregates the training subsets, evaluation assets, and result dumps published from the DiT4SR workspace under a single Hugging Face dataset namespace.
Replication/ stores the merged replication training subset.
CustomDataset/ stores the custom training subset.
NovelTextConditioned/ stores the novelty text-conditioned training subset.
Evaluation/ is reserved for benchmark inputs, prompts, and curated test assets.… See the full description on the dataset page:
https://huggingface.co/datasets/NisargUpadhyay/ImageSuperResolution.
