EVEWorld: Physical Evolution Supervision for Embodied World Models

arXiv 2026 36.9 method, benchmark

TLDR

EVEWorld is a physical evolution-supervision framework for embodied world models, improving instance consistency and reducing model laziness via restoration and temporal alignment.

Reasoning

The paper addresses a clear limitation (model laziness) and introduces both a framework and a metric, with strong benchmark results. However, the abstract lacks methodological detail and explicit discussion of limitations, relying on benchmark leaderboards rather than real-world deployments.

Read-first score

Read-first score 36.9, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 49.

Recency 5%

100

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

Topical relevance 25%

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

Methodology quality 15%

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

Reproducibility 15%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Graph prestige 15%

5.7

PageRank-style prestige over the local citation graph built from available OpenAlex references. local_pagerank=0.0016

Citation impact 15%

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. cited_by_count=0

Citation velocity 10%

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

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 432.

Keyword Scores

world model

world simulator

world dynamics prediction

generative world model

interactive world model

video world model

model-based reinforcement learning world model

Deep Analysis

Innovations

Proposes EVEWorld, a physical evolution-supervision framework for physically consistent target evolution in embodied world models.
Introduces Instance-Guided Restoration (IGR) to promote instance consistency through restoration supervision.
Introduces Temporal Instance Alignment (TIA) to promote cross-frame consistency by aligning target instances across adjacent frames.
Introduces the Model Laziness Rate (MLR), a metric measuring persistent violations of instance consistency in generated trajectories.

Methodology

EVEWorld consists of Instance-Guided Restoration (IGR), which uses restoration supervision to promote instance consistency, and Temporal Instance Alignment (TIA), which aligns target instances across adjacent frames to promote cross-frame consistency. The framework is evaluated on DreamGenBench, EWMBench, and PBench, and introduces the Model Laziness Rate (MLR) to measure persistent violations of instance consistency, with comparisons against GigaWorld-0 and WorldArena 2.0 Track 1 leaderboard rankings.

Key Results

EVEWorld achieved an 87.5% reduction in Model Laziness Rate compared with GigaWorld-0, and ranked 6th in JEPA Similarity and 17th overall on the WorldArena 2.0 Track 1 leaderboard.

Resources

Paper