OmniVerifier-C1: Decomposed Graded Verifier for Rule-Governed Generative World Modeling

arXiv 2026 22.8 method

TLDR

Proposes OmniVerifier-C1, a decomposed graded verifier for rule-governed generative world modeling, using decoupled reinforcement learning to provide fine-grained pointwise critiques beyond binary verifiers.

Reasoning

The paper introduces a novel verifier that decomposes complex rules into fine-grained components and provides continuous scoring, with theoretical justification for pairwise verification. Strengths include a clear training framework and empirical comparisons against binary verifiers, though the abstract omits specific benchmark details and explicit limitations.

Read-first score

Read-first score 22.8, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 18.

Recency 5%

100

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

Methodology quality 15%

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

Reproducibility 15%

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

Topical relevance 25%

25.7

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

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: 111.

Keyword Scores

generative world model

world model

world simulator

interactive world model

world dynamics prediction

model-based reinforcement learning world model

video world model

Resources

Paper