SearchSwarm uses a harness to synthesize delegation trajectories for supervised fine-tuning, producing a 30B model that scores 68.1 on BrowseComp and 73.3 on BrowseComp-ZH.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
SearchSwarm: Towards Delegation Intelligence in Agentic LLMs for Long-Horizon Deep Research
SearchSwarm uses a harness to synthesize delegation trajectories for supervised fine-tuning, producing a 30B model that scores 68.1 on BrowseComp and 73.3 on BrowseComp-ZH.