Fine-Grained Self-Endorsement Improves Factuality and Reasoning

arXiv:2402.15631v1 Announce Type: new
Abstract: This work studies improving large language model (LLM) generations at inference time by mitigating fact-conflicting hallucinations. Particularly, we propose a self-endorsement framework that leverages the fine-grained fact-level comparisons across multiple sampled responses. Compared with prior ensemble methods (Wang et al., 2022;Chen et al., 2023)) that perform response-level selection, our approach can better alleviate hallucinations, especially for longform generation tasks. Our approach can broadly benefit smaller and open-source LLMs as it mainly conducts simple content-based comparisons. …

abstract arxiv chen cs.ai cs.cl ensemble fine-grained framework hallucinations inference language language model large language large language model llm multiple prior reasoning responses studies type work