FinGPT: Large Generative Models for a Small Language

TL;DR

This paper develops and evaluates large Finnish language models, FinGPT and BLUUMI, addressing the challenge of limited data for small languages and introducing a Finnish benchmark for model assessment.

Contribution

It presents the creation of large monolingual and adapted multilingual Finnish language models, along with a new Finnish benchmark for evaluation.

Findings

FinGPT models perform well on Finnish tasks.

BLUUMI improves multilingual Finnish understanding.

Models are openly available for research.

Abstract

Large language models (LLMs) excel in many tasks in NLP and beyond, but most open models have very limited coverage of smaller languages and LLM work tends to focus on languages where nearly unlimited data is available for pretraining. In this work, we study the challenges of creating LLMs for Finnish, a language spoken by less than 0.1% of the world population. We compile an extensive dataset of Finnish combining web crawls, news, social media and eBooks. We pursue two approaches to pretrain models: 1) we train seven monolingual models from scratch (186M to 13B parameters) dubbed FinGPT, 2) we continue the pretraining of the multilingual BLOOM model on a mix of its original training data and Finnish, resulting in a 176 billion parameter model we call BLUUMI. For model evaluation, we introduce FIN-bench, a version of BIG-bench with Finnish tasks. We also assess other model qualities such as toxicity and bias. Our models and tools are openly available at https://turkunlp.org/gpt3-finnish.