Search papers, labs, and topics across Lattice.
1
0
4
You can shrink LLM motion planners by 5x without sacrificing performance by distilling knowledge on-policy with dense token-level feedback.