CARL introduced as a reinforcement-learning agent for creating and steering self-organizing patterns
CARL's authors say the agent learns to create patterns in Lenia, steer their behavior and let people guide them in real time.
TLDR
Announced on September 10, CARL is described by its authors as a closed-loop reinforcement-learning agent for discovering and controlling self-organizing phenomena in Lenia, with real-time human guidance. A quote-post calls the work a milestone for actively controlling such patterns to achieve goals.
Combined views
—
2 Sources, first seen ago
— likes— comments— saves— reposts
