CRE-T1 Preview Technical Report: Beyond Contrastive Learning for Reasoning-Intensive Retrieval

TL;DR

This paper introduces T1, a generative retrieval model that enhances reasoning-intensive retrieval by dynamically generating reasoning trajectories, outperforming contrastive learning methods especially in scenarios with implicit reasoning and vocabulary mismatch.

Contribution

The paper proposes a novel generative retrieval approach that internalizes dynamic reasoning into vector representations, surpassing static contrastive methods in reasoning-intensive retrieval tasks.

Findings

T1-4B outperforms larger contrastively trained models on BRIGHT benchmark.

Dynamic reasoning generation improves retrieval performance in implicit reasoning scenarios.

Model achieves comparable results to multi-stage retrieval pipelines.

Abstract

The central challenge of reasoning-intensive retrieval lies in identifying implicitreasoning relationships between queries and documents, rather than superficial se-mantic or lexical similarity. The contrastive learning paradigm is fundamentallya static representation consolidation technique: during training, it encodes hier-archical relevance concepts into fixed geometric structures in the vector space,and at inference time it cannot dynamically adjust relevance judgments accord-ing to the specific reasoning demands of each query. Consequently, performancedegrades noticeably when vocabulary mismatch exists between queries and doc-uments or when implicit reasoning is required to establish relevance. This pa-per proposes Thought 1 (T1), a generative retrieval model that shifts relevancemodeling from static alignment to dynamic reasoning. On the query side, T1 dy-namically generates intermediate reasoning trajectories for each query to bridgeimplicit reasoning relationships and uses <embtoken> as a semantic aggregationpoint for the reasoning output. On the document side, it employs an instruction+ text + <embtoken> encoding format to support high-throughput indexing. Tointernalize dynamic reasoning capabilities into vector representations, we adopt athree-stage training curriculum and introduce GRPO in the third stage, enablingthe model to learn optimal derivation strategies for different queries through trial-and-error reinforcement learning. On the BRIGHT benchmark, T1-4B exhibitsstrong performance under the original query setting, outperforming larger modelstrained with contrastive learning overall, and achieving performance comparableto multi-stage retrieval pipelines. The results demonstrate that replacing static rep-resentation alignment with dynamic reasoning generation can effectively improvereasoning-intensive retrieval performance.