Symphony: A Decentralized Multi-Agent Framework for Scalable Collective Intelligence
This addresses scalability and cost issues for deploying multi-agent AI systems, though it appears incremental as it builds on existing decentralized concepts.
The paper tackled the problem of centralized orchestration in LLM-based agent frameworks by introducing Symphony, a decentralized multi-agent system that enables coordination on consumer-grade GPUs, resulting in substantial accuracy gains on reasoning benchmarks.
Most existing Large Language Model (LLM)-based agent frameworks rely on centralized orchestration, incurring high deployment costs, rigid communication topologies, and limited adaptability. To address these challenges, we introduce Symphony, a decentralized multi-agent system which enables lightweight LLMs on consumer-grade GPUs to coordinate. Symphony introduces three key mechanisms: (1) a decentralized ledger that records capabilities, (2) a Beacon-selection protocol for dynamic task allocation, and (3) weighted result voting based on CoTs. This design forms a privacy-saving, scalable, and fault-tolerant orchestration with low overhead. Empirically, Symphony outperforms existing baselines on reasoning benchmarks, achieving substantial accuracy gains and demonstrating robustness across models of varying capacities.