Language world model reference implementation (Qwen-AgentWorld style): one latent backbone predicts the next textual observation across 7 agent environments (MCP/Search/Terminal/SWE/Web/OS/Android) with cross-env transfer. Deterministic oracle + monotone benchmark. Pure Python, zero deps.
python zero-dependency tool-use environment-simulation world-model agent-environment agentworld language-world-model cross-environment-transfer reward-prediction
-
Updated
Nov 3, 2026 - Python