Less Is More: Pre-train A Strong Text Encoder For Dense Retrieval Using A Weak Decoder

Abstract

Dense retrieval requires high-quality text sequence embeddings to support effective search in the representation space. Autoencoder-based language models are appealing in dense retrieval as they train the encoder to output high-quality embedding that can reconstruct the input texts. However, in this paper, we provide theoretical analyses and show empirically that an autoencoder language model with a low reconstruction loss may not provide good sequence representations because the decoder may take shortcuts by exploiting language patterns. To address this, we propose a new self-learning method that pre-trains the autoencoder using a \textit\{weak\} decoder, with restricted capacity and attention flexibility to push the encoder to provide better text representations. Our experiments on web search, news recommendation, and open domain question answering show that our pre-trained model significantly boosts the effectiveness and few-shot ability of dense retrieval models. Our code is availa

Stats

• citations28
• S2 citations—
• github stars45
• HF likes0
• heat score14.29
• arxiv keylu2021less

Related papers

• Pre-train A Discriminative Text Encoder For Dense Retrieval Via Contrastive Span Prediction (2022)10.21
• Challenging Decoder Helps In Masked Auto-encoder Pre-training For Dense Passage Retrieval (2023)0.00
• Lexmae: Lexicon-bottlenecked Pretraining For Large-scale Retrieval (2022)0.00
• Drop Your Decoder: Pre-training With Bag-of-word Prediction For Dense Passage Retrieval (2024)3.58
• Query Encoder Distillation Via Embedding Alignment Is A Strong Baseline Method To Boost Dense Retriever Online Efficiency (2023)0.00
• Dense Retrievers Can Fail On Simple Queries: Revealing The Granularity Dilemma Of Embeddings (2025)2.86
• Condenser: A Pre-training Architecture For Dense Retrieval (2021)14.90
• Back To Basics: A Simple Recipe For Improving Out-of-domain Retrieval In Dense Encoders (2023)0.00