Optimizing Retrieval Components for a Shared Backbone via Component-Wise Multi-Stage Training
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911605292269568 |
|---|---|
| author | Li, Yunhan Xie, Mingjie Gong, Zihan Shi, Zeyang Wu, Gengshen Yang, Min |
| author_facet | Li, Yunhan Xie, Mingjie Gong, Zihan Shi, Zeyang Wu, Gengshen Yang, Min |
| contents | Recent advances in embedding-based retrieval have enabled dense retrievers to serve as core infrastructure in many industrial systems, where a single retrieval backbone is often shared across multiple downstream applications. In such settings, retrieval quality directly constrains system performance and extensibility, while coupling model selection, deployment, and rollback decisions across applications.
In this paper, we present empirical findings and a system-level solution for optimizing retrieval components deployed as a shared backbone in production legal retrieval systems. We adopt a multi-stage optimization framework for dense retrievers and rerankers, and show that different retrieval components exhibit stage-dependent trade-offs. These observations motivate a component-wise, mixed-stage configuration rather than relying on a single uniformly optimal checkpoint. The resulting backbone is validated through end-to-end evaluation and deployed as a shared retrieval service supporting multiple industrial applications. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_00805 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Optimizing Retrieval Components for a Shared Backbone via Component-Wise Multi-Stage Training Li, Yunhan Xie, Mingjie Gong, Zihan Shi, Zeyang Wu, Gengshen Yang, Min Information Retrieval Recent advances in embedding-based retrieval have enabled dense retrievers to serve as core infrastructure in many industrial systems, where a single retrieval backbone is often shared across multiple downstream applications. In such settings, retrieval quality directly constrains system performance and extensibility, while coupling model selection, deployment, and rollback decisions across applications. In this paper, we present empirical findings and a system-level solution for optimizing retrieval components deployed as a shared backbone in production legal retrieval systems. We adopt a multi-stage optimization framework for dense retrievers and rerankers, and show that different retrieval components exhibit stage-dependent trade-offs. These observations motivate a component-wise, mixed-stage configuration rather than relying on a single uniformly optimal checkpoint. The resulting backbone is validated through end-to-end evaluation and deployed as a shared retrieval service supporting multiple industrial applications. |
| title | Optimizing Retrieval Components for a Shared Backbone via Component-Wise Multi-Stage Training |
| topic | Information Retrieval |
| url | https://arxiv.org/abs/2602.00805 |