Search papers, labs, and topics across Lattice.
NVIDIA, The University of Texas at Austin
1
0
2
5
Scaling visuomotor context to 8K timesteps enables robots to master complex tasks and adapt in real-time, outperforming previous models by a staggering margin.
