A Robustly Optimized Long Text to Math Models for Numerical Reasoning On FinQA

TL;DR

This paper introduces a robust model for numerical reasoning in financial question answering, achieving top results in the FinQA challenge by combining specialized models for improved accuracy.

Contribution

The paper presents a novel ensemble approach with specialized models that significantly improves numerical reasoning performance in financial QA tasks.

Findings

Achieved 71.93% execution accuracy in FinQA

Achieved 67.03% program accuracy in FinQA

Outperformed previous methods in financial numerical reasoning

Abstract

Numerical reasoning is required when solving most problems in our life, but it has been neglected in previous artificial intelligence researches. FinQA challenge has been organized to strengthen the study on numerical reasoning where the participants are asked to predict the numerical reasoning program to solve financial question. The result of FinQA will be evaluated by both execution accuracy and program accuracy. In this paper, we present our approach to tackle the task objective by developing models with different specialized capabilities and fusing their strength. Overall, our approach achieves the 1st place in FinQA challenge, with 71.93% execution accuracy and 67.03% program accuracy.