Token Bottleneck Scene Dynamics

Learn to compress entire scenes into a single bottleneck token that captures temporal dynamics. Enables efficient visual tracking and robotic manipulation by forcing reconstruction from minimal target hints, achieving superior performance with training costs comparable to standard autoencoders.

adu2021 Updated 7 repo stars

File contents

adu2021/skillxiv commit 5d1b36d727

Frequently asked questions

npx get-skill add adu2021/token-bottleneck-scene-dynamics