Encoder vs Decoder: Comparative Analysis of Encoder and Decoder Language Models on Multilingual NLU Tasks

TL;DR

This study compares encoder and decoder language models on multilingual NLU tasks, revealing encoder models outperform decoders despite fewer parameters, and offers new evaluation methods for decoder models in this context.

Contribution

It extends the ScandEval benchmark to include decoder models and introduces a novel evaluation method for their performance on NLU tasks across multiple languages.

Findings

Encoder models outperform decoder models in NLU tasks.

Decoder models require more parameters to achieve similar performance.

UMAP analysis highlights distinct capabilities of encoder and decoder models.

Abstract

This paper explores the performance of encoder and decoder language models on multilingual Natural Language Understanding (NLU) tasks, with a broad focus on Germanic languages. Building upon the ScandEval benchmark, initially restricted to evaluating encoder models, we extend the evaluation framework to include decoder models. We introduce a method for evaluating decoder models on NLU tasks and apply it to the languages Danish, Swedish, Norwegian, Icelandic, Faroese, German, Dutch, and English. Through a series of experiments and analyses, we also address research questions regarding the comparative performance of encoder and decoder models, the impact of NLU task types, and the variation across language resources. Our findings reveal that encoder models can achieve significantly better NLU performance than decoder models despite having orders of magnitude fewer parameters. Additionally, we investigate the correlation between decoders and task performance via a UMAP analysis, shedding light on the unique capabilities of decoder and encoder models. This study contributes to a deeper understanding of language model paradigms in NLU tasks and provides valuable insights for model selection and evaluation in multilingual settings.