A reinforcement-learned Conductor coordinates pools of LLMs to reach state-of-the-art results on LiveCodeBench and GPQA by discovering communication topologies and targeted prompts.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Learning to Orchestrate Agents in Natural Language with the Conductor
A reinforcement-learned Conductor coordinates pools of LLMs to reach state-of-the-art results on LiveCodeBench and GPQA by discovering communication topologies and targeted prompts.