Customized FinGPT Search Agents Using Foundation Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Tian, Felix, Byadgi, Ajay, Kim, Daniel, Zha, Daochen, White, Matt, Xiao, Kairong, Yanglet, Xiao-Yang Liu
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909356804538368
author Tian, Felix
Byadgi, Ajay
Kim, Daniel
Zha, Daochen
White, Matt
Xiao, Kairong
Yanglet, Xiao-Yang Liu
author_facet Tian, Felix
Byadgi, Ajay
Kim, Daniel
Zha, Daochen
White, Matt
Xiao, Kairong
Yanglet, Xiao-Yang Liu
contents Current large language models (LLMs) have proven useful for analyzing financial data, but most existing models, such as BloombergGPT and FinGPT, lack customization for specific user needs. In this paper, we address this gap by developing FinGPT Search Agents tailored for two types of users: individuals and institutions. For individuals, we leverage Retrieval-Augmented Generation (RAG) to integrate local documents and user-specified data sources. For institutions, we employ dynamic vector databases and fine-tune models on proprietary data. There are several key issues to address, including data privacy, the time-sensitive nature of financial information, and the need for fast responses. Experiments show that FinGPT agents outperform existing models in accuracy, relevance, and response time, making them practical for real-world applications.
format Preprint
id arxiv_https___arxiv_org_abs_2410_15284
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Customized FinGPT Search Agents Using Foundation Models
Tian, Felix
Byadgi, Ajay
Kim, Daniel
Zha, Daochen
White, Matt
Xiao, Kairong
Yanglet, Xiao-Yang Liu
Computational Engineering, Finance, and Science
Human-Computer Interaction
Current large language models (LLMs) have proven useful for analyzing financial data, but most existing models, such as BloombergGPT and FinGPT, lack customization for specific user needs. In this paper, we address this gap by developing FinGPT Search Agents tailored for two types of users: individuals and institutions. For individuals, we leverage Retrieval-Augmented Generation (RAG) to integrate local documents and user-specified data sources. For institutions, we employ dynamic vector databases and fine-tune models on proprietary data. There are several key issues to address, including data privacy, the time-sensitive nature of financial information, and the need for fast responses. Experiments show that FinGPT agents outperform existing models in accuracy, relevance, and response time, making them practical for real-world applications.
title Customized FinGPT Search Agents Using Foundation Models
topic Computational Engineering, Finance, and Science
Human-Computer Interaction
url https://arxiv.org/abs/2410.15284