Evaluating LLMs' Mathematical Reasoning in Financial Document Question Answering

TL;DR

This paper evaluates large language models' ability to perform complex mathematical reasoning on financial documents with tables, introducing a new prompting method that enhances performance in semi-structured data tasks.

Contribution

It provides a comprehensive assessment of LLMs' mathematical reasoning in financial QA, and introduces a novel prompting technique for semi-structured documents that improves results.

Findings

LLMs' performance decreases with increased table complexity

New prompting method outperforms baseline techniques

Insights into LLMs' strengths and limitations in financial reasoning

Abstract

Large Language Models (LLMs), excel in natural language understanding, but their capability for complex mathematical reasoning with an amalgamation of structured tables and unstructured text is uncertain. This study explores LLMs' mathematical reasoning on four financial tabular question-answering datasets: TATQA, FinQA, ConvFinQA, and Multihiertt. Through extensive experiments with various models and prompting techniques, we assess how LLMs adapt to complex tables and mathematical tasks. We focus on sensitivity to table complexity and performance variations with an increasing number of arithmetic reasoning steps. The results provide insights into LLMs' capabilities and limitations in handling complex mathematical scenarios for semi-structured tables. Ultimately, we introduce a novel prompting technique tailored to semi-structured documents, matching or outperforming other baselines in performance while providing a nuanced understanding of LLMs abilities for such a task.