Deficiency of Large Language Models in Finance: An Empirical Examination of Hallucination
Haoqiang Kang, Xiao-Yang Liu
TL;DR
This paper empirically examines hallucination in finance-focused LLMs across three tasks: acronym recognition, financial term explanations, and historical stock price queries. It evaluates multiple models (e.g., Llama2 variants, GPT-3.5/4, FinMA-7B) and assesses mitigation techniques including few-shot prompting, DoLa, RAG, and prompt-based tool learning. The findings show serious hallucinations in off-the-shelf FinLLMs, with RAG and tool-based approaches significantly improving factual accuracy, while multi-task finetuning can reduce instruction-following ability. The work highlights the need for grounding LLMs with external data sources and tooling to enable reliable, up-to-date financial reasoning in real-world use.
Abstract
The hallucination issue is recognized as a fundamental deficiency of large language models (LLMs), especially when applied to fields such as finance, education, and law. Despite the growing concerns, there has been a lack of empirical investigation. In this paper, we provide an empirical examination of LLMs' hallucination behaviors in financial tasks. First, we empirically investigate LLM model's ability of explaining financial concepts and terminologies. Second, we assess LLM models' capacity of querying historical stock prices. Third, to alleviate the hallucination issue, we evaluate the efficacy of four practical methods, including few-shot learning, Decoding by Contrasting Layers (DoLa), the Retrieval Augmentation Generation (RAG) method and the prompt-based tool learning method for a function to generate a query command. Finally, our major finding is that off-the-shelf LLMs experience serious hallucination behaviors in financial tasks. Therefore, there is an urgent need to call for research efforts in mitigating LLMs' hallucination.