Event
RoboTTT improves long-horizon robot memory with fast weights and context scaling
1 reports · 1 sources
RoboTTT addresses the difficulty mainstream robot policies, represented by VLA, have in remembering operations that occurred earlier, proposing fast weights and context scaling to improve long-horizon memory. This report is an excerpted introduction and does not provide complete experimental details.
Report timeline
- RoboTTT uses fast weights and context scaling to help robots remember operations from five minutes earlier
RoboTTT targets a structural flaw in robot foundation models: mainstream policies such as VLA struggle to remember things that happened earlier. It proposes using fast weights and context scaling to improve long-horizon memory. The original text is an excerpt from an introduction and does not provide full experimental details.
Guyuehome ·