AgentGarten: Agents Learn in Neural-Rendered Interactive Worlds

MirroS-Lab released AgentGarten, a framework where agents learn by exploring interactive worlds built from executable code for physics and a real-time neural renderer for visuals at 30fps. Agents reportedly learn in just 4 rounds what takes millions of RL steps.

2026-10-08 ~ 2026-10-09 · 2 related posts

• MirroS' AgentGarten pairs code physics with neural rendering to evolve agents at 30fps — 量子位 · 2026-10-08
• AgentGarten trains agents in rendered code worlds, learning in 4 rounds vs millions for RL — MirroS-Lab · 2026-10-09