LOLA -- An Open-Source Massively Multilingual Large Language Model
This work addresses the problem of computational efficiency and linguistic diversity in multilingual AI systems for researchers and practitioners, representing an incremental advancement with novel architectural implementation.
The paper tackles the challenge of building efficient multilingual large language models by introducing LOLA, a massively multilingual model trained on over 160 languages using a sparse Mixture-of-Experts Transformer architecture, achieving competitive performance in natural language generation and understanding tasks.
This paper presents LOLA, a massively multilingual large language model trained on more than 160 languages using a sparse Mixture-of-Experts Transformer architecture. Our architectural and implementation choices address the challenge of harnessing linguistic diversity while maintaining efficiency and avoiding the common pitfalls of multilinguality. Our analysis of the evaluation results shows competitive performance in natural language generation and understanding tasks. Additionally, we demonstrate how the learned expert-routing mechanism exploits implicit phylogenetic linguistic patterns to potentially alleviate the curse of multilinguality. We provide an in-depth look at the training process, an analysis of the datasets, and a balanced exploration of the model's strengths and limitations. As an open-source model, LOLA promotes reproducibility and serves as a robust foundation for future research. Our findings enable the development of compute-efficient multilingual models with strong, scalable performance across languages.