LLMs Bridge Gap Between Spanish and English in Financial Transactions

The financial industry’s reliance on natural language processing (NLP) and large language models (LLMs) has created a significant gap in NLP studies for Spanish compared to English. To bridge this gap, researchers have developed the Toisón de Oro framework, which establishes instruction datasets, fine-tuned LLMs, and evaluation benchmarks for financial LLMs in Spanish and English. This bilingual framework aims to facilitate the development of LLMs that can handle bilingual financial applications, providing a comprehensive assessment of multilingual performance and highlighting the importance of bridging this gap for global finance.

Can Financial LLMs Bridge the Gap Between Spanish and English?

The financial industry is a global phenomenon, with languages like Spanish playing a crucial role in international transactions. However, despite its importance, there exists a significant gap in natural language processing (NLP) and application studies for Spanish compared to English, particularly in the era of large language models (LLMs). To bridge this gap, researchers have developed a bilingual framework called Toisón de Oro, which establishes instruction datasets, fine-tuned LLMs, and evaluation benchmarks for financial LLMs in Spanish and English.

The Toisón de Oro framework is built upon a rigorously curated bilingual instruction dataset comprising over 144K samples from 15 datasets covering seven tasks. This dataset is designed to facilitate the development of LLMs that can handle bilingual financial applications. The framework’s evaluation benchmark, FLAREES, consists of 21 datasets covering nine tasks and provides a comprehensive assessment of multilingual performance.

The results of the FLAREES benchmark reveal a significant gap in multilingual performance between existing LLMs and those designed specifically for bilingual financial applications. Furthermore, the study highlights a bias in existing LLMs towards English language processing, which can have far-reaching implications for global finance.

What is Toisón de Oro?

Toisón de Oro is a bilingual framework that aims to bridge the gap between Spanish and English in financial natural language processing (NLP) and application studies. The framework consists of three key components: instruction datasets, fine-tuned LLMs, and evaluation benchmarks.

The instruction datasets are designed to facilitate the development of LLMs that can handle bilingual financial applications. These datasets consist of over 144K samples from 15 datasets covering seven tasks, including text classification, sentiment analysis, and named entity recognition.

The fine-tuned LLMs are trained on these instruction datasets and are designed to perform well in bilingual financial applications. The evaluation benchmarks, FLAREES, provide a comprehensive assessment of multilingual performance and help identify areas where existing LLMs fall short.

How Does Toisón de Oro Work?

Toisón de Oro works by leveraging the power of large language models (LLMs) to develop bilingual financial applications. The framework consists of three key components: instruction datasets, fine-tuned LLMs, and evaluation benchmarks.

The instruction datasets are designed to facilitate the development of LLMs that can handle bilingual financial applications. These datasets consist of over 144K samples from 15 datasets covering seven tasks, including text classification, sentiment analysis, and named entity recognition.

The fine-tuned LLMs are trained on these instruction datasets and are designed to perform well in bilingual financial applications. The evaluation benchmarks, FLAREES, provide a comprehensive assessment of multilingual performance and help identify areas where existing LLMs fall short.

What are the Benefits of Toisón de Oro?

Toisón de Oro offers several benefits for the development of bilingual financial applications. Firstly, it provides a comprehensive framework for developing LLMs that can handle bilingual financial tasks. Secondly, it highlights the importance of multilingual performance in financial NLP and application studies.

The study also reveals a bias in existing LLMs towards English language processing, which can have far-reaching implications for global finance. By developing LLMs that can handle bilingual financial applications, Toisón de Oro aims to bridge this gap and provide a more comprehensive assessment of multilingual performance.

What are the Challenges of Toisón de Oro?

Toisón de Oro faces several challenges in its development and implementation. Firstly, it requires a large amount of high-quality data for training LLMs that can handle bilingual financial applications.

Secondly, it needs to address the issue of bias in existing LLMs towards English language processing. This bias can have far-reaching implications for global finance and highlights the importance of developing LLMs that can handle bilingual financial tasks.

Finally, Toisón de Oro requires a comprehensive evaluation benchmark, FLAREES, which provides a comprehensive assessment of multilingual performance.

Publication details: “Dólares or Dollars? Unraveling the Bilingual Prowess of Financial LLMs Between Spanish and English”
Publication Date: 2024-08-24
Authors: Xiao Zhang, Ruoyu Xiang, Chenhan Yuan, Duanyu Feng, et al.
Source:
DOI: https://doi.org/10.1145/3637528.3671554

Quantum News

Quantum News

As the Official Quantum Dog (or hound) by role is to dig out the latest nuggets of quantum goodness. There is so much happening right now in the field of technology, whether AI or the march of robots. But Quantum occupies a special space. Quite literally a special space. A Hilbert space infact, haha! Here I try to provide some of the news that might be considered breaking news in the Quantum Computing space.

Latest Posts by Quantum News:

IBM Remembers Lou Gerstner, CEO Who Reshaped Company in the 1990s

IBM Remembers Lou Gerstner, CEO Who Reshaped Company in the 1990s

December 29, 2025
Optical Tweezers Scale to 6,100 Qubits with 99.99% Imaging Survival

Optical Tweezers Scale to 6,100 Qubits with 99.99% Imaging Survival

December 28, 2025
Rosatom & Moscow State University Develop 72-Qubit Quantum Computer Prototype

Rosatom & Moscow State University Develop 72-Qubit Quantum Computer Prototype

December 27, 2025