Meta-Reflective Evolution Effectively Improves Production AI Systems

cs.LG, cs.AI, cs.CL, cs.SE

Submitted: 2026-10-08

Updated: 2026-10-08

Code: https://github.com/codelion/openevolve

Project page: https://bauplanlabs.github.io/SAO-workshop/papers/73.pdf

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Sources

• optimize_anything: A Universal API for Optimizing any Text Parameter
• AdaEvolve: Adaptive LLM Driven Zeroth-Order Optimization
• DABstep: Data Agent Benchmark for Multi-step Reasoning
• ShinkaEvolve: Towards Open-Ended And Sample-Efficient Program Evolution
• Meta-Harness: End-to-End Optimization of Model Harnesses
• Cortex AISQL: A Production SQL Engine for Unstructured Data
• Illuminating search spaces by mapping elites
• Do Agent Optimizers Compound? A Continual-Learning Evaluation on Terminal-Bench 2.0
• Meta Context Engineering via Agentic Skill Evolution
• Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models
• AIChilles: Automatically Uncovering Hidden Weaknesses in AI-Evolved Systems

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks