Ingest-And-Ground: Dispelling Hallucinations from Continually-Pretrained LLMs with RAG
This work aims to improve privacy process efficiency for organizations by reducing LLM hallucinations in privacy-related queries, offering a strong specific gain.
This paper addresses the problem of hallucinations in LLMs when handling privacy-related queries. By continually pre-training an LLM with a privacy-specific knowledge base and augmenting it with a semantic RAG layer, the authors doubled performance metrics compared to an out-of-box LLM.
This paper presents new methods that have the potential to improve privacy process efficiency with LLM and RAG. To reduce hallucination, we continually pre-train the base LLM model with a privacy-specific knowledge base and then augment it with a semantic RAG layer. Our evaluations demonstrate that this approach enhances the model performance (as much as doubled metrics compared to out-of-box LLM) in handling privacy-related queries, by grounding responses with factual information which reduces inaccuracies.