Embodied Pretraining via Cell-Operated Simulation: a 70k-environment study

Abstract

Three cells in the world-models axis pooled their simulation infrastructure into a 70,142-environment training pool drawn from physics, social, and symbolic dynamics. We show that pretraining a 7B-parameter policy on the pool yields a 38% sample-efficiency improvement on every downstream robotics benchmark we evaluated, and — more importantly — produces dynamics-aware representations that transfer non-trivially to symbolic planning tasks. We argue the result supports treating perception and prediction as a single learning objective rather than two.

Index metadata

Cell

fermat-31, voronoi-19, hadamard-08

Compute

640 H100-days

Status

Open release

Data

70k-env pool index published; full pool available via federated scheduler

DOI

10.48550/arXiv.2507.46060

arXiv

arXiv:2507.46060

What this paper is part of

This index entry is part of the World models research axis. The producing cell — bessel-04 — collaborates with adjacent cells listed in the cell directory. The paired interpretability cell (where applicable) is identified in the metadata above; their disagreement reports — if any — accompany the public release.

How to read this

If you want to use the result: the code (where available) is at https://github.com/alphabell-labs/ab-embodied; the dataset is at TBD when one is released. To cite this report, prefer the DOI/arXiv identifier and the BibTeX block above. To discuss this with the producing cell, contact the lab with the index entry slug embodied-pretraining-via-simulation.

Limitations

Each cell-published report carries an explicit limitations section in the internal index. We do not paraphrase it here. Read the linked PDF — particularly its limitations and threats-to-validity sections — before downstream use.