Search papers, labs, and topics across Lattice.
University of Illinois Urbana-Champaign
1
0
3
7
Rollout design in LLM reinforcement learning is more than just sampling trajectories – it's a modular pipeline you can optimize for reliability, coverage, and cost.
