Evolutionary Ensemble of Agents
TLDR
EvE is a decentralized framework that co-evolves coding agents and their guidance to discover algorithms, demonstrating superior adaptation and performance.
Key contributions
- Introduces EvE, a decentralized framework for co-evolving coding agents and their guidance.
- Co-evolves agent guidance states and code solvers, evaluating agents via synchronous races and Elo ratings.
- Discovered a robust rescale-then-interpolate mechanism for In-Context Operator Networks (ICON).
- Shows stage-dependent agent adaptation is crucial for navigating complex codebases and breaking performance ceilings.
Why it matters
This paper introduces a novel approach to algorithmic discovery by evolving agent guidance, not the agents themselves. It demonstrates that dynamic adaptation within an agent ensemble is critical for overcoming performance plateaus in complex problem-solving, offering a new paradigm for leveraging existing LLM capabilities.
Original Abstract
We introduce Evolutionary Ensemble (EvE), a decentralized framework that organizes existing, highly capable coding agents into a live, co-evolving system for algorithmic discovery. Rather than reinventing the wheel within the "LLMs as optimizers" paradigm, EvE fixes the base agent substrate and focuses entirely on evolving the cumulative guidance and skills that dictate agent behaviors. By maintaining two co-evolving populations, namely functional code solvers and agent guidance states, the system evaluates agents through a synchronous race, updating their empirical Elo ratings based on the marginal gains they contribute to the current solver state. When applied to a research bottleneck in In-Context Operator Networks (ICON), EvE autonomously discovered a robust rescale-then-interpolate mechanism that enables reliable example-count generalization. Crucially, controlled ablations reveal the absolute necessity of stage-dependent agent adaptation to navigate the shifting search landscapes of complex codebases. Compared to variants driven by a fixed initial agent or even a frozen "best-evolved" agent, EvE uniquely avoids phase mismatch, demonstrating that organizing agents into a self-revising ensemble is the fundamental driver for breaking through static performance ceilings.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.