FinNLI: Novel Dataset for Multi-Genre Financial Natural Language Inference Benchmarking

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Magomere, Jabez, Kochkina, Elena, Mensah, Samuel, Kaur, Simerjot, Smiley, Charese H.
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909588833435648
author Magomere, Jabez
Kochkina, Elena
Mensah, Samuel
Kaur, Simerjot
Smiley, Charese H.
author_facet Magomere, Jabez
Kochkina, Elena
Mensah, Samuel
Kaur, Simerjot
Smiley, Charese H.
contents We introduce FinNLI, a benchmark dataset for Financial Natural Language Inference (FinNLI) across diverse financial texts like SEC Filings, Annual Reports, and Earnings Call transcripts. Our dataset framework ensures diverse premise-hypothesis pairs while minimizing spurious correlations. FinNLI comprises 21,304 pairs, including a high-quality test set of 3,304 instances annotated by finance experts. Evaluations show that domain shift significantly degrades general-domain NLI performance. The highest Macro F1 scores for pre-trained (PLMs) and large language models (LLMs) baselines are 74.57% and 78.62%, respectively, highlighting the dataset's difficulty. Surprisingly, instruction-tuned financial LLMs perform poorly, suggesting limited generalizability. FinNLI exposes weaknesses in current LLMs for financial reasoning, indicating room for improvement.
format Preprint
id arxiv_https___arxiv_org_abs_2504_16188
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle FinNLI: Novel Dataset for Multi-Genre Financial Natural Language Inference Benchmarking
Magomere, Jabez
Kochkina, Elena
Mensah, Samuel
Kaur, Simerjot
Smiley, Charese H.
Computation and Language
Artificial Intelligence
Machine Learning
We introduce FinNLI, a benchmark dataset for Financial Natural Language Inference (FinNLI) across diverse financial texts like SEC Filings, Annual Reports, and Earnings Call transcripts. Our dataset framework ensures diverse premise-hypothesis pairs while minimizing spurious correlations. FinNLI comprises 21,304 pairs, including a high-quality test set of 3,304 instances annotated by finance experts. Evaluations show that domain shift significantly degrades general-domain NLI performance. The highest Macro F1 scores for pre-trained (PLMs) and large language models (LLMs) baselines are 74.57% and 78.62%, respectively, highlighting the dataset's difficulty. Surprisingly, instruction-tuned financial LLMs perform poorly, suggesting limited generalizability. FinNLI exposes weaknesses in current LLMs for financial reasoning, indicating room for improvement.
title FinNLI: Novel Dataset for Multi-Genre Financial Natural Language Inference Benchmarking
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2504.16188