Beyond Prediction: Steering VLM Agents with Retrospective World Modeling
cs.AI, cs.LG
Submitted: 2026-09-30
Updated: 2026-09-30
Code: https://github.com/mpSchrader/gym-sokoban
Terminology
Sources
- Mastering Diverse Domains through World Models
- Active Intelligence in Video Avatars via Closed-loop World Modeling
- Agent Learning via Early Experience
- Why Do LLM Agents Fail in Exploring New Environments? A World-Modeling Perspective
- Motus: A Unified Latent Action World Model
- World Action Models are Zero-shot Policies
- GigaWorld-0: World Models as Data Engine to Empower Embodied AI
- GigaWorld-Policy: An Efficient Action-Centered World--Action Model
- High-Dimensional Continuous Control Using Generalized Advantage Estimation
- Gymnasium: A Standard Interface for Reinforcement Learning Environments
- VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model
- GPT-4o System Card
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems