llm-collaboration

This paper introduces scaling participation, a new paradigm for building modular AI systems through contributions from diverse stakeholders, where small models collaborate to outperform monolithic LLMs by up to 15.4% across various tasks, demonstrating emergent capabilities and improved diversity benefits.

This paper proposes TopoPrior, a framework that learns transferable topology priors from offline reference collaboration graphs to generate initial topologies for multi-agent LLM collaboration across domains, significantly reducing online search overhead and token consumption.

Introduces Capability Conditioned Scaffolding, a framework for LLM collaboration that adapts intervention based on user expertise domains to prevent Professional Domain Drift, with pilot evaluation on MMLU subsets.