By nvidia

tao-train-nvdinov2

Build visual features without labeled data.

visual representation learning

Only the most recent Revisions are kept. Restoring one saves its content again as the newest.