Large Language Models in Finance: A Survey

Yinheng Li,Shaofei Wang,Han Ding,Hang Chen

2024-07-09

Abstract:Recent advances in large language models (LLMs) have opened new possibilities for artificial intelligence applications in finance. In this paper, we provide a practical survey focused on two key aspects of utilizing LLMs for financial tasks: existing solutions and guidance for adoption. First, we review current approaches employing LLMs in finance, including leveraging pretrained models via zero-shot or few-shot learning, fine-tuning on domain-specific data, and training custom LLMs from scratch. We summarize key models and evaluate their performance improvements on financial natural language processing tasks. Second, we propose a decision framework to guide financial professionals in selecting the appropriate LLM solution based on their use case constraints around data, compute, and performance needs. The framework provides a pathway from lightweight experimentation to heavy investment in customized LLMs. Lastly, we discuss limitations and challenges around leveraging LLMs in financial applications. Overall, this survey aims to synthesize the state-of-the-art and provide a roadmap for responsibly applying LLMs to advance financial AI.

General Finance,Artificial Intelligence,Computation and Language

What problem does this paper attempt to address?

The paper primarily explores the application and practice of large language models (LLMs) in the financial sector and attempts to address the following key issues:

1. **Overview of current methods using LLMs for financial tasks**: The paper first reviews various methods currently employed to handle financial tasks using LLMs, including leveraging pre-trained models through zero-shot or few-shot learning, fine-tuning models on domain-specific data, and training custom LLMs from scratch. The authors summarize some important models and evaluate the performance improvements of these models on financial natural language processing tasks.

2. **Provide a guiding framework for selecting appropriate LLM solutions**: Secondly, the paper proposes a decision-making framework to guide financial professionals in choosing suitable LLM solutions based on their application scenario's data, computational resources, and performance requirements. This framework offers a pathway from lightweight experiments to significant investments in custom LLMs.

3. **Discuss the limitations and challenges of using LLMs**: Finally, the paper discusses the limitations and challenges encountered when utilizing LLMs in financial applications. Overall, this review aims to synthesize existing research findings and provide a roadmap for responsibly applying LLMs to financial AI.

In short, the goal of the paper is to provide financial professionals and researchers with a practical overview, focusing on how to effectively use large language models to solve financial tasks, and offering guidance from existing solutions to the application process.