MICE: Minimal Interaction Cross-Encoders for efficient Re-ranking
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_ | 1866908861053534208 |
|---|---|
| author | Vast, Mathias Morand, Victor van Cooten, Basile Soulier, Laure Mothe, Josiane Piwowarski, Benjamin |
| author_facet | Vast, Mathias Morand, Victor van Cooten, Basile Soulier, Laure Mothe, Josiane Piwowarski, Benjamin |
| contents | Cross-encoders deliver state-of-the-art ranking effectiveness in information retrieval, but have a high inference cost. This prevents them from being used as first-stage rankers, but also incurs a cost when re-ranking documents. Prior work has addressed this bottleneck from two largely separate directions: accelerating cross-encoder inference by sparsifying the attention process or improving first-stage retrieval effectiveness using more complex models, e.g. late-interaction ones. In this work, we propose to bridge these two approaches, based on an in-depth understanding of the internal mechanisms of cross-encoders. Starting from cross-encoders, we show that it is possible to derive a new late-interaction-like architecture by carefully removing detrimental or unnecessary interactions. We name this architecture MICE (Minimal Interaction Cross-Encoders). We extensively evaluate MICE across both in-domain (ID) and out-of-domain (OOD) datasets. MICE decreases fourfold the inference latency compared to standard cross-encoders, matching late-interaction models like ColBERT while retaining most of cross-encoder ID effectiveness and demonstrating superior generalization abilities in OOD. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_16299 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MICE: Minimal Interaction Cross-Encoders for efficient Re-ranking Vast, Mathias Morand, Victor van Cooten, Basile Soulier, Laure Mothe, Josiane Piwowarski, Benjamin Information Retrieval Cross-encoders deliver state-of-the-art ranking effectiveness in information retrieval, but have a high inference cost. This prevents them from being used as first-stage rankers, but also incurs a cost when re-ranking documents. Prior work has addressed this bottleneck from two largely separate directions: accelerating cross-encoder inference by sparsifying the attention process or improving first-stage retrieval effectiveness using more complex models, e.g. late-interaction ones. In this work, we propose to bridge these two approaches, based on an in-depth understanding of the internal mechanisms of cross-encoders. Starting from cross-encoders, we show that it is possible to derive a new late-interaction-like architecture by carefully removing detrimental or unnecessary interactions. We name this architecture MICE (Minimal Interaction Cross-Encoders). We extensively evaluate MICE across both in-domain (ID) and out-of-domain (OOD) datasets. MICE decreases fourfold the inference latency compared to standard cross-encoders, matching late-interaction models like ColBERT while retaining most of cross-encoder ID effectiveness and demonstrating superior generalization abilities in OOD. |
| title | MICE: Minimal Interaction Cross-Encoders for efficient Re-ranking |
| topic | Information Retrieval |
| url | https://arxiv.org/abs/2602.16299 |