Text Embeddings by Weakly-Supervised Contrastive Pre-training

TL;DR

E5 is a state-of-the-art text embedding model trained with weak supervision that excels in various tasks like retrieval and classification, outperforming traditional methods in zero-shot and fine-tuned scenarios.

Contribution

The paper introduces E5, a new contrastively trained text embedding model that achieves superior performance across multiple benchmarks with minimal supervision.

Findings

E5 outperforms BM25 in zero-shot retrieval on BEIR.

E5 achieves top results on MTEB with fewer parameters.

E5 demonstrates strong transferability across diverse tasks.

Abstract

This paper presents E5, a family of state-of-the-art text embeddings that transfer well to a wide range of tasks. The model is trained in a contrastive manner with weak supervision signals from our curated large-scale text pair dataset (called CCPairs). E5 can be readily used as a general-purpose embedding model for any tasks requiring a single-vector representation of texts such as retrieval, clustering, and classification, achieving strong performance in both zero-shot and fine-tuned settings. We conduct extensive evaluations on 56 datasets from the BEIR and MTEB benchmarks. For zero-shot settings, E5 is the first model that outperforms the strong BM25 baseline on the BEIR retrieval benchmark without using any labeled data. When fine-tuned, E5 obtains the best results on the MTEB benchmark, beating existing embedding models with 40x more parameters.