mohantesting/

funvid

A collection of image-to-video samples with quality-evaluation metadata and precomputed VAE/T5 latents for fast model training. The HF Dataset Viewer is disabled because the tarballs are plain tar.zst archives, not WebDataset shards.

	Contents




	

Asset Size Notes

First tarball (*.tar.zst) ~17 GB Initial sample set, ~12k clips

Extended tarball (*_merged.tar.zst) ~18 GB Initial set + newer samples

latent_caches/... ~94 GB Per-sample .pt… See the full description on the dataset page: https://huggingface.co/datasets/mohantesting/funvid.