The static inventory contains 3,198 YouTube IDs. Earlier collectors stored raw sources under videos/ and records under records/. V5 preserves those files and prepares model-ready frame packs under a versioned prepared/ namespace. New network acquisition samples video streams directly instead of retaining full video files. Existing local sources are reused. Prepared samples contain losslessly compressed RGB uint8 frame banks at 224×224, view indices for micro/mid/long/extralong/spatial, source frame positions, and provenance. Their loader materializes float32 [T,3,224,224] views in [0,1] for the bundled visual model, which applies its own CLIP normalization. Frame counts are 32/16/8/16/12.…
Independent publisher
Shah
gonnerthetooner
Models in Library0
Datasets in Library1
Models on Hugging Face3
Followers—
Compare the published evidence
This profile connects the 1 model and dataset record shown below to their upstream cards and sources. The library counts describe SAVRN's coverage, rather than this publisher's entire output.
Published tasks: Video classification.
Compare model specifications, source dates, and hardware estimates on the individual records. Publisher identity verification is separate from a model's quality, deployment readiness, or suitability for your work.