SusGen-GPT: A Data-Centric LLM for Financial NLP and Sustainability Report Generation

TL;DR

SusGen-GPT introduces a data-centric approach with a new balanced dataset and benchmark, enabling efficient financial and ESG report generation with performance close to GPT-4 using significantly fewer parameters.

Contribution

The paper presents SusGen-30K dataset, TCFD-Bench benchmark, and SusGen-GPT models, advancing open-source NLP tools for finance and ESG with state-of-the-art results.

Findings

SusGen-GPT achieves near GPT-4 performance with 7-8B parameters.

The SusGen-30K dataset covers seven financial NLP tasks and ESG report generation.

The SusGen system effectively supports sustainability report creation using RAG.

Abstract

The rapid growth of the financial sector and the rising focus on Environmental, Social, and Governance (ESG) considerations highlight the need for advanced NLP tools. However, open-source LLMs proficient in both finance and ESG domains remain scarce. To address this gap, we introduce SusGen-30K, a category-balanced dataset comprising seven financial NLP tasks and ESG report generation, and propose TCFD-Bench, a benchmark for evaluating sustainability report generation. Leveraging this dataset, we developed SusGen-GPT, a suite of models achieving state-of-the-art performance across six adapted and two off-the-shelf tasks, trailing GPT-4 by only 2% despite using 7-8B parameters compared to GPT-4's 1,700B. Based on this, we propose the SusGen system, integrated with Retrieval-Augmented Generation (RAG), to assist in sustainability report generation. This work demonstrates the efficiency of our approach, advancing research in finance and ESG.