CALIPER: Clean Scenes Cannot Rank Physical Inference in Pretrained Visual Representations

arXiv 2026 23.9 method

TLDR

Shows clean fixed-camera scenes cannot assess physical inference in pretrained visual encoders; proposes CALIPER test revealing differences only with varied camera/lighting/clutter.

Reasoning

The paper offers a strong, controlled empirical design with calibration and swap checks, convincingly showing that clean-scene benchmarks cannot distinguish physical inference. Its main limitation is reliance solely on simulated episodes and lack of real-world validation, and the abstract leaves the linear probe analysis incomplete.

Read-first score

Read-first score 23.9, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 22.

Recency 5%

100

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

Reproducibility 15%

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

Topical relevance 25%

31.4

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=benchmark

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: 131.

Keyword Scores

world dynamics prediction

world model

interactive world model

world simulator

video world model

model-based reinforcement learning world model

generative world model

Resources

Paper