World-Representation-Lab/World-Embedding-Regression

World Embedding Regression

World Embedding Regression is the physical-property regression dataset from the World Embedding Benchmark. It contains 5,000 simulation videos across 10 physics families, with 500 examples in each dataset configuration.

Usage

from datasets import load_dataset

dataset = load_dataset(
    "World-Representation-Lab/World-Embedding-Regression",
    "pendulum",
    split="test",
)

Fields

• family: physics family.
• id: unique example identifier.
• video: simulation video.
• regression_attribute: physical quantity to predict.
• regression_value: continuous target value.

The fixed test and scaling splits used by the benchmark are distributed with the evaluation code.

Citation

@misc{liu2026worldembeddingbenchmark,
  title         = {World Embedding Benchmark},
  author        = {Yiqi Liu and Ruifeng Yuan and Yang Wang and Long Li and Fengyu Cai and Hou Pong Chan and Jialin Yu and Hao Zhang and Chenghua Lin and Chenghao Xiao},
  year          = {2026},
  eprint        = {2610.03632},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2610.03632}
}