Search papers, labs, and topics across Lattice.
INFIFORCE Intelligent Technology, Carnegie Mellon University, The University of Hong Kong, Princeton University Princeton University Carnegie Mellon University
CMU Machine Learning3
0
6
Extracting temporal geometry from generative models can boost reinforcement learning performance by over 2x without changing the optimal policy.
Diffusion models can efficiently sample lookahead action sequences for active search, outperforming traditional tree search while mitigating optimism bias.
LLMs can turn sparse rewards into dense training signals for RL agents, achieving comparable performance with significantly higher sample efficiency.
