meta-evolution

MetaDE is a GPU-accelerated evolutionary framework that optimizes Differential Evolution (DE) strategies via meta-level evolution. Supporting both JAX and PyTorch, it dynamically adapts mutation and crossover strategies for efficient large-scale black-box optimization.

Official Implementation of VisualClaw: A Real-Time, Personalized Agent for the Physical World

The code for paper "AutoPSO: A Meta-Framework for Automated Particle Swarm Optimization"

Next evolution of grounded evolution: evolves full agent blueprints (system prompt + tools + memory + planning + self-evaluation) using Docker sandboxing, multi-objective fitness, meta-evolution, and task specialization. Built using DeepSeek V4.

Self-referential Darwin-style agent forge that evolves its own mutation operators, evaluators, and core components. Built with DeepSeek V4.

Evaluator-grounded autonomous prompt evolution. Genetic algorithms + runtime validation + meta-evolution to evolve prompts that generate better AI agent code. 150 generations, 862/1000 best score.