Generate, Then Match: Lightweight Relevance Modeling for Generative Retrieval

Generative retrievers can generate relevant document identifiers yet rank the corresponding documents poorly, exposing a mismatch between generation likelihood and relevance. We introduce QIMatch, a lightweight adaptation method that learns a relevance matching function over a retriever's existing query and identifier representations. With the generator and identifier index fixed, two shared projections learn to match queries with complete identifiers through contrastive training followed by document ranking refinement, updating only 1.18M parameters without an additional neural relevance teacher. At inference, QIMatch combines matching scores with generation likelihood to rerank generated identifiers without encoding document text. On Natural Questions and MS MARCO, QIMatch achieves the highest Recall@1 and MRR@10 among the compared methods for both product-quantized and title-URL identifiers. Across these four settings, it improves MRR@10 over supervised fine-tuning baselines by 3.08-12.33 percentage points. Analyses show that the learned matching function transfers to identifiers excluded from its supervision in both families, while fusion can correct ranking errors that neither score resolves alone. These findings demonstrate how learning to match existing representations enables effective relevance adaptation with a fixed generator.