Entity · paper

Think Big, Search Small: Where Capacity Matters in Hierarchical Search Agents?

paperactiveprovisionalthink-big-search-small-where-capacity-matters-in-hierarchical-search-agents--ed68213c·1 events·first seen 7h ago

Aliases: Think Big, Search Small: Where Capacity Matters in Hierarchical Search Agents?

Co-occurring entities

Qinnan Cai

More like this (12)

hybrid dense-sparse retrieval Hierarchical Relative Policy Optimization multi-turn agent benchmarks multi-level agent evaluation Enhancing Decision-Making with Large Language Models through Multi-Agent Fictitious Play How AI Agents Reshape Knowledge Work: Autonomy, Efficiency, and Scope Beyond Global Replanning: Hierarchical Recovery for Cross-Device Agent Systems Hierarchical Advantage Weighting for Online RL Fine-Tuning of VLAs from Sparse Episode Outcomes Data Intelligence Agents: Interpreting, Modeling, and Querying Enterprise Data via Autonomous Coding Agents Training LLMs to Enforce Multi-Level Instruction Hierarchies via Gravity-Weighted Direct Preference Optimization Hierarchical Reinforcement Learning Trace Only What You Need: Structure-Aware On-Demand Hypergraph Memory for Long-Document Question Answering

Recent events (1)

6arXiv · cs.CL·7h ago·source ↗

Role-factorized hierarchical search agents: delegation capacity matters more than execution

A new arXiv paper investigates how model capacity should be distributed across roles in multi-agent search systems, factorizing hierarchical search into delegation, execution, and answer generation roles. Controlled sweeps across five multi-hop QA benchmarks find that scaling the delegation backbone improves exact match by ~11 points while scaling execution sub-agents yields only ~2.6 points, identifying task decomposition as the primary bottleneck. A 1.7B-parameter executor trained via trajectory distillation matches frontier sub-agent accuracy while using 37% fewer tokens, advancing the efficiency Pareto frontier. The results offer a concrete design recipe: concentrate capacity at delegation and downsize execution.

Evaluation and Benchmarking Inference Economics Qinnan Cai Think Big, Search Small: Where Capacity Matters in Hierarchical Search Agents?+1 more