Beyond Imitation: Leveraging Fine-grained Quality Signals for Alignment

arXiv:2311.04072v2 Announce Type: replace
Abstract: Alignment with human preference is a desired property of large language models (LLMs). Currently, the main alignment approach is based on reinforcement learning from human feedback (RLHF). Despite the effectiveness of RLHF, it is intricate to implement and train, thus recent studies explore how to develop alternative alignment approaches based on supervised fine-tuning (SFT). A major limitation of SFT is that it essentially does imitation learning, which cannot fully understand what are the expected behaviors. …

abstract alignment arxiv beyond cs.cl explore feedback fine-grained human human feedback language language models large language large language models llms property quality reinforcement reinforcement learning rlhf studies train type