
RLCD proposes a method for aligning language models using reinforcement learning and contrastive distillation to improve model behavior without human feedback.


coremsg

rlcd-reinforcement-learning-from-contrastive-distillation-for-lm-alignment


RLCD: Reinforcement Learning from Contrastive Distillation for LM Alignment


title_rewrite


RLCD proposes a method using reinforcement learning to align language models by generating preference pairs with contrasting prompts.


rlcd-reinforcement-learning-from-contrastive-distillation-for-lm-alignment-at-iclr-2024


RLCD: Reinforcement Learning from Contrastive Distillation for LM Alignment at ICLR 2024
