Halo: Estimation and Reduction of Hallucinations in Open-Source Weak Large Language Models
This addresses the issue of hallucinations in low-parameter LLMs for researchers and practitioners using open-source models, though it is incremental as it builds on existing methods like knowledge injection.
The paper tackled the problem of severe hallucinations in open-source weak large language models like BLOOM 7B, introducing HaloCheck to measure them and exploring techniques such as knowledge injection and teacher-student approaches to reduce hallucinations, with experiments showing effective reduction in challenging domains.
Large Language Models (LLMs) have revolutionized Natural Language Processing (NLP). Although convenient for research and practical applications, open-source LLMs with fewer parameters often suffer from severe hallucinations compared to their larger counterparts. This paper focuses on measuring and reducing hallucinations in BLOOM 7B, a representative of such weaker open-source LLMs that are publicly available for research and commercial applications. We introduce HaloCheck, a lightweight BlackBox knowledge-free framework designed to quantify the severity of hallucinations in LLMs. Additionally, we explore techniques like knowledge injection and teacher-student approaches to alleviate hallucinations in low-parameter LLMs. Our experiments effectively demonstrate the reduction of hallucinations in challenging domains for these LLMs.