RetrievalRouter: Joint Modality and Architecture Selection for Document Retrieval
Emre Kuru · hf · 2026-08-27
RetrievalRouter introduces a method to improve document retrieval by adaptively selecting retrieval pipelines per query.
Key Features:
- Adaptive Selection: Dynamically chooses the best retrieval pipeline based on the specific query.
- Dual Optimization: Aims to enhance both accuracy and speed simultaneously.
- Broad Applicability: Demonstrates strong performance across diverse document benchmarks.
More from Research
- CoRL 2025: Hierarchical RL Framework Adapts Quadruped Skills to Complex Terrain — breadli428 · 2026-08-27
- Study finds most accurate models are not always the most preferred by users — soumitrashukla9 · 2026-08-27
- Modular Agent Outperforms End-to-End VLMs in CT Spatial Reasoning — Simon Vincent Abel · 2026-08-27
- Prefix Sliding: discarding stale reasoning tokens makes test-time scaling 3x faster — Bedrovelsen · 2026-08-27
- Code released: Localizing and intervening on harmful mechanisms in LLMs — boknilev · 2026-08-27
- Paper: Sparse RL can't find what a model never samples; dense rewards break the ceiling — iScienceLuvr · 2026-08-27