The Troubling Emergence of Hallucination in Large Language Models -- An Extensive Definition, Quantification, and Prescriptive Remediations

no code implementations • 8 Oct 2023 • Vipula Rawte, Swagata Chakraborty, Agnibh Pathak, Anubhav Sarkar, S. M Towhidul Islam Tonmoy, Aman Chadha, Amit P. Sheth, Amitava Das

Finally, to establish a method for quantifying and to offer a comparative spectrum that allows us to evaluate and rank LLMs based on their vulnerability to producing hallucinations, we propose Hallucination Vulnerability Index (HVI).

Hallucination