Mining Intrinsic Rewards from LLM Hidden States for Efficient Best-of-N Sampling.

Jizhou Guo, Zhaomin Wu, Hanchen Yang, Philip S. Yu

Browse the full KDD paper archive.