datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
SeFi-Image-SemVAE-Demo
SeFi-Image SemVAE Demo
Eight small, standalone synthetic images for testing the public
SeFi-Image SemVAE
loader and feature-compression path. This is a smoke-test fixture, not a
training set or an evaluation benchmark.
Each sample is available twice:
as an ordinary JPEG under images/, for path-based command-line examples;
as embedded image bytes in the test Parquet split, for dataset-loader
examples.
No contact sheets, plots, or images from the SeFi code repository's assets/… See the full description on the dataset page: https://huggingface.co/datasets/SeFi-Image/SeFi-Image-SemVAE-Demo.SeFi-Image-DiT-Finetune-Demo
SeFi-Image DiT Fine-tuning Demo
This dataset contains 56 training and 8 validation image-text pairs for
fixed-1024 DiT fine-tuning smoke tests. It is derived byte-for-byte from the
ma-xu/fine-t2i
synthetic_enhanced_prompt_random_resolution subset at revision
28fdd5663ee202b5cafc01d6ed08a03f14957854. Only rows whose
image_generator is Z-Image-Turbo are included. FLUX.2-dev rows, the
existing SemVAE demo source ids, and rows rejected during prompt/image quality
review are excluded… See the full description on the dataset page: https://huggingface.co/datasets/SeFi-Image/SeFi-Image-DiT-Finetune-Demo.