LUSIFER: Language Universal Space Integration for Enhanced Multilingual Embeddings with Large Language Models

TL;DR

LUSIFER is a zero-shot method that enhances multilingual embeddings by integrating a universal encoder with LLM-based models, improving performance across diverse languages without multilingual supervision.

Contribution

It introduces a novel architecture combining a multilingual encoder with LLM-based embeddings, enabling effective multilingual performance without requiring multilingual training data.

Findings

Significantly improves multilingual embedding performance.

Effective for medium and low-resource languages.

Introduces a comprehensive multilingual benchmark.

Abstract

Recent advancements in large language models (LLMs) based embedding models have established new state-of-the-art benchmarks for text embedding tasks, particularly in dense vector-based retrieval. However, these models predominantly focus on English, leaving multilingual embedding capabilities largely unexplored. To address this limitation, we present LUSIFER, a novel zero-shot approach that adapts LLM-based embedding models for multilingual tasks without requiring multilingual supervision. LUSIFER's architecture combines a multilingual encoder, serving as a language-universal learner, with an LLM-based embedding model optimized for embedding-specific tasks. These components are seamlessly integrated through a minimal set of trainable parameters that act as a connector, effectively transferring the multilingual encoder's language understanding capabilities to the specialized embedding model. Additionally, to comprehensively evaluate multilingual embedding performance, we introduce a new benchmark encompassing 5 primary embedding tasks, 123 diverse datasets, and coverage across 14 languages. Extensive experimental results demonstrate that LUSIFER significantly enhances the multilingual performance across various embedding tasks, particularly for medium and low-resource languages, without requiring explicit multilingual training data.