AnyGoal: On World Model Generalization Beyond Expert Rollouts
To answer this, we introduce AnyGoal, a benchmark of goal-shifted task variants built from widely used world-model environments, together with an open-source library.
ProofPaper ↗
Key points
- The goal of a world model is to capture the dynamics of an environment.
- We study this relationship through the lens of compositional generalization: can a world model recombine familiar state and action factors into valid dynamics?
- Across multiple environments and world model families, we find that current models consistently exhibit failures under these recombinations.
- By examining dynamics and representation failures across model families, interaction regimes, and evaluation metrics, we study where current world models generalize and where challenges remain for goal-shifted prediction and planning.
Sources (2)
- [1]AnyGoal: On World Model Generalization Beyond Expert RolloutsHugging Face Daily Papers · Oct 19, 12:00 AM
To answer this, we introduce AnyGoal, a benchmark of goal-shifted task variants built from widely used world-model environments, together with an open-source library.
The goal of a world model is to capture the dynamics of an environment.
- [2]AnyGoal: On World Model Generalization Beyond Expert RolloutsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 19, 05:59 PM · same content
Extractive summary: sentences quoted from the sources.
Before this
- Oct 18, 2026CoEvo-RSI: When Harnesses Evolve, Which Models Should Follow?
- Oct 18, 2026StopLoss: Teaching Agents to Predict Failure and Stop Early
- Oct 18, 2026World Model Interpretability: From Decoding Representations to Understanding Latent Worlds
- Oct 16, 2026Execution Before Memory: Diagnosing Memory Failures in Video World Models
- Oct 16, 2026OR-License: Evaluating LLM Agents as Real-World Operations Research Practitioners
- Oct 16, 2026Xiaomi-Lens-0 Technical Report