stable-worldmodel: A Platform for Reproducible World Modeling Research and Evaluation

arXiv 2026 67.3 system

TLDR

An open-source platform for reproducible world modeling research with standardized data, baselines, and evaluation benchmarks.

Reasoning

The paper addresses fragmentation in world model research by providing a unified platform with high-performance data loading, clean baselines, and systematic evaluation benchmarks. Its strengths include reproducibility and scalability, but the evaluation is primarily in-silico, though real-world dataset support (LeRobot) is included.

Read-first score

Read-first score 67.3, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 51.

Methodology quality 18%

100

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=baseline,benchmark,dataset,evaluation,experiment,metric,result

Recency 6%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Citation impact 18%

80.8

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.80841558

Topical relevance 29%

72.9

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Reproducibility 18%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=code,dataset

Citation velocity 12%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

FoundationFrontierBridgeMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 393.

Keyword Scores

world model

world dynamics prediction

world simulator

video world model

model-based reinforcement learning world model

generative world model

interactive world model

Deep Analysis

Innovations

High-performance Lance-based data layer with native support and conversion tools for MP4, HDF5, and LeRobot datasets
Clean, well-tested implementations of modern world model baselines and planning solvers
Broad suite of environments and tasks extended with controllable visual, geometric, and physical factors of variation for systematic in-silico evaluation

Methodology

The platform provides a unified pipeline integrating a high-performance Lance-based data layer for efficient video data loading, clean implementations of world model baselines and planning solvers, and a suite of environments with controllable factors of variation for systematic evaluation of dynamics understanding, control performance, representation quality, and out-of-distribution generalization.

Key Results

No experimental results are reported in the abstract; the paper focuses on describing the platform's design and features.

Resources

Paper