Hypencoder: Hypernetworks for Information Retrieval

2025-02-07

AI Summary

Existing dense retrieval models rely on vector inner products to compute relevance, inherently limiting their representational capacity. This work proposes Hypencoder—a novel retrieval paradigm based on hypernetworks—that encodes each query into a lightweight, dynamic neural network; this network itself serves as a learnable, non-linear relevance function that directly maps document representations to scalar relevance scores, thereby overcoming the linearity constraint of inner-product matching. To our knowledge, this is the first application of hypernetworks to retrieval modeling, enabling query-driven, function-level matching. Experiments demonstrate that Hypencoder significantly outperforms strong baselines—including re-rankers and large-scale models—on in-domain retrieval, achieving end-to-end latency under 60 ms on an 8.8M-document corpus. Its advantages further widen on challenging tasks such as tip-of-the-tongue retrieval and instruction-following, as well as in cross-domain settings, confirming its strong generalization capability.

Technology Category

• Search and Optimization: Learning to Search
• Computer Vision: Image and Video Retrieval
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

The vast majority of retrieval models depend on vector inner products to produce a relevance score between a query and a document. This naturally limits the expressiveness of the relevance score that can be employed. We propose a new paradigm, instead of producing a vector to represent the query we produce a small neural network which acts as a learned relevance function. This small neural network takes in a representation of the document, in this paper we use a single vector, and produces a scalar relevance score. To produce the little neural network we use a hypernetwork, a network that produce the weights of other networks, as our query encoder or as we call it a Hypencoder. Experiments on in-domain search tasks show that Hypencoder is able to significantly outperform strong dense retrieval models and has higher metrics then reranking models and models an order of magnitude larger. Hypencoder is also shown to generalize well to out-of-domain search tasks. To assess the extent of Hypencoder's capabilities, we evaluate on a set of hard retrieval tasks including tip-of-the-tongue retrieval and instruction-following retrieval tasks and find that the performance gap widens substantially compared to standard retrieval tasks. Furthermore, to demonstrate the practicality of our method we implement an approximate search algorithm and show that our model is able to search 8.8M documents in under 60ms.

Problem

• Improves relevance score expressiveness in retrieval models
• Introduces hypernetworks for generating learned relevance functions
• Enhances performance in in-domain and out-of-domain search tasks

Innovation

• Hypernetwork generates neural network weights
• Hypencoder replaces vector with relevance function
• Approximate search handles 8.8M documents quickly