Preview - rendered architecture diagrams; full validated atlas + dataset coming.
Back to models

InternVideo2 (spatiotemporal ViT with tubelet embedding, attention-pooling head)

history internvideo2 spatiotemporal validation pending

Architecture diagram

Rendered with TorchLens / Graphviz

Open SVG
Unrolled - full
InternVideo2 (spatiotemporal ViT with tubelet embedding, attention-pooling head) architecture diagram