ENCORE: Entropy-guided Reward Composition for Multi-head Safety Reward Models.

Xiaomin Li, Xupeng Chen, Jingxuan Fan, Eric Hanchen Jiang, Mingye Gao

Browse the full AAAI paper archive.