Making Text Embedders Few-Shot Learners

TL;DR

This paper introduces a few-shot learning approach for text embedding generation using large language models' in-context learning capabilities, achieving state-of-the-art results on multiple benchmarks.

Contribution

The authors propose a novel method leveraging LLMs' in-context learning for high-quality text embeddings, demonstrating significant improvements over existing approaches.

Findings

Achieves new SOTA on MTEB and AIR-Bench benchmarks.

Simple retention of original framework yields best results.

Effectively utilizes task examples within the query for improved embeddings.

Abstract

Large language models (LLMs) with decoder-only architectures demonstrate remarkable in-context learning (ICL) capabilities. This feature enables them to effectively handle both familiar and novel tasks by utilizing examples provided within their input context. Recognizing the potential of this capability, we propose leveraging the ICL feature in LLMs to enhance the process of text embedding generation. To this end, we introduce a novel model bge-en-icl, which employs few-shot examples to produce high-quality text embeddings. Our approach integrates task-related examples directly into the query side, resulting in significant improvements across various tasks. Additionally, we have investigated how to effectively utilize LLMs as embedding models, including various attention mechanisms, pooling methods, etc. Our findings suggest that retaining the original framework often yields the best results, underscoring that simplicity is best. Experimental results on the MTEB and AIR-Bench benchmarks demonstrate that our approach sets new state-of-the-art (SOTA) performance. Our model, code and dataset are freely available at https://github.com/FlagOpen/FlagEmbedding .