Data-Centric Financial Large Language Models

Abstract: LLMs show promise for natural language tasks but struggle when applied directly to complex domains like finance. LLMs have difficulty reasoning about and integrating all relevant information. We propose a data-centric approach to enable LLMs to better handle financial tasks. Our key insight is that rather than overloading the LLM with everything at once, it is more effective to preprocess and pre-understand the data. We create a financial LLM (FLLM) using multitask prompt-based finetuning to achieve data pre-processing and pre-understanding. However, labeled data is scarce for each task. To overcome manual annotation costs, we employ abductive augmentation reasoning (AAR) to automatically generate training data by modifying the pseudo labels from FLLM's own outputs. Experiments show our data-centric FLLM with AAR substantially outperforms baseline financial LLMs designed for raw text, achieving state-of-the-art on financial analysis and interpretation tasks. We also open source a new benchmark for financial analysis and interpretation. Our methodology provides a promising path to unlock LLMs' potential for complex real-world domains.