Tag
The paper proposes Style-Debiased DPO (SD-DPO), a method for factuality-aware synthetic preference data to improve knowledge elicitation in large language models, addressing issues where standard DPO may incorrectly penalize correct responses due to stylistic differences.
The paper proposes a synthetic simulation framework and benchmark for evaluating and updating knowledge in large language models, demonstrating a 14.23% improvement over existing methods.
PASTA is a novel framework for knowledge updating in LLMs that combines data augmentation, question-answering generation, and self-learning DPO to integrate factual information from news articles, achieving accuracy improvement from 0.02 to 0.82 while preserving general capabilities.