CFGPT: Chinese Financial Assistant with Large Language Model

Large language models (LLMs) have demonstrated great potential in natural language processing tasks within the financial domain. In this work, we present a Chinese Financial Generative Pre-trained Transformer framework, named CFGPT, which includes a dataset (CFData) for pre-training and supervised fine-tuning, a financial LLM (CFLLM) to adeptly manage financial texts, and a deployment framework (CFAPP) designed to navigate real-world financial applications.

arXiv preprint