# Evolutionary Ensemble of Agents

Source: [Evolutionary Ensemble of Agents](<https://www.sota2.com/research/paper/6a02b410dbc836b87d8246a2>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-07-08T05:00:28.404Z

Authors: Zongmin Yu, Liu Yang

Published: 2026-05-09T15:56:10.000Z

## 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.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Example-count generalization on 1D conservation law 2,000 steps k=5 examples standardized (train)](<https://www.sota2.com/research/sota/example-count-generalization-on-1d-conservation-law-2-000-steps-k-5-examples-standardized-train.md>): Example-count generalization; 1D conservation law 2,000 steps k=5 examples standardized (train); MAE: 0.108.

- [Example-count generalization on 1D conservation law 10,000 steps k=5 examples (standardized retraining)](<https://www.sota2.com/research/sota/example-count-generalization-on-1d-conservation-law-10-000-steps-k-5-examples-standardized-retraining.md>): Example-count generalization; 1D conservation law 10,000 steps k=5 examples (standardized retraining); MAE: 0.041.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.09018>)

[PDF](<https://arxiv.org/pdf/2605.09018.pdf>)