Summary
WorldEvolver is a self-evolving world model framework for long-horizon LLM agent planning introduced by Xuan Zhang, Wenxuan Zhang, and See-Kiong Ng (arXiv:2507.00003). World models give LLM agents foresight by predicting action consequences before execution, but unreliable predictions can degrade downstream decision-making. WorldEvolver addresses this by revising its deployment-time context while keeping the downstream agent and all model parameters frozen. It integrates three modules: Episodic Memory, which exploits real action transitions via retrieval-based simulation; Semantic Memory, which extracts persistent heuristic rules from prediction-observation mismatches; and Selective Foresight, which filters low-confidence predictions before integrating them into the agent's reasoning context. Evaluated on ALFWorld and ScienceWorld for downstream agent success, Word2World for prediction accuracy, and AgentBoard for agent performance, WorldEvolver achieves the highest prediction accuracy across three backbone models and outperforms other world model baselines in downstream agent success rate, demonstrating that test-time memory revision improves both prediction fidelity and planning performance.
Paper Overview
Research Area: Agent
Authors: Xuan Zhang, Wenxuan Zhang, See-Kiong Ng
Published: 2026-07-01
arXiv: 2507.00003
Abstract
World models offer a principled way to equip long-horizon LLM agents with foresight: predictions of action consequences before execution. However, unreliable foresight can be ignored, misused, or even degrade downstream decision-making.
In this paper, the authors introduce WorldEvolver, a self-evolving world model framework that revises its deployment-time context while keeping the downstream agent and all model parameters frozen.
Key Modules
1. Episodic Memory — exploits real action transitions through retrieval-based simulation.
2. Semantic Memory — extracts persistent heuristic rules from prediction-observation mismatches.
3. Selective Foresight — filters low-confidence predictions before integrating them into the agent's reasoning context.
Evaluation
- Benchmarks: ALFWorld and ScienceWorld (downstream agent success), Word2World (world model prediction accuracy), AgentBoard (downstream agent success rate).
- Results: Extensive experiments show WorldEvolver achieves the highest prediction accuracy across three backbone models and leads other world model baselines in downstream agent success rate.
Takeaway
Test-time memory revision simultaneously enhances both prediction fidelity and planning performance — without any parameter updates to the agent or the world model.
---
*Auto-collected on 2026-07-01*
This page is an English static mirror generated for search and AI citation.
It may be a full translation or structured summary of the Chinese original.
Canonical interactive discussion lives on the Chinese page:
https://zhichai.net/topic/178208338