pith:JSS5QPY6
ACWM-Phys: Investigating Generalized Physical Interaction in Action-Conditioned Video World Models
Out-of-distribution generalization in action-conditioned world models succeeds on simple rigid interactions but drops on deformable and high-dimensional cases.
arxiv:2605.08567 v2 · 2026-05-09 · cs.CV
Add to your LaTeX paper
\usepackage{pith}
\pithnumber{JSS5QPY6ZXK347QT2WRQMVMXRZ}
Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge
Record completeness
Claims
OoD generalization depends not only on the physical regime but also on effective task complexity: models generalize well on visually simple, low-dimensional interactions with clear geometric structure, but suffer larger drops on deformable contacts, high-dimensional control, and complex articulated motion.
That the fully controllable simulator accurately captures the rich physical interactions required for generalized world understanding and that the chosen action space and data protocols expose the true limits of current models rather than simulator artifacts.
ACWM-Phys benchmark shows action-conditioned world models generalize on simple geometric interactions but drop sharply on deformable contacts, high-dimensional control, and complex articulated motion, indicating reliance on visual appearance over learned physics.
Formal links
Receipt and verification
| First computed | 2026-05-20T00:02:12.733733Z |
|---|---|
| Builder | pith-number-builder-2026-05-17-v1 |
| Signature | Pith Ed25519
(pith-v1-2026-05) · public key |
| Schema | pith-number/v1.0 |
Canonical hash
4ca5d83f1ecdd5be7e13d5a30655978e74b1adf06925ca85b61951ff61071289
Aliases
· · · · ·Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/JSS5QPY6ZXK347QT2WRQMVMXRZ \
| jq -c '.canonical_record' \
| python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 4ca5d83f1ecdd5be7e13d5a30655978e74b1adf06925ca85b61951ff61071289
Canonical record JSON
{
"metadata": {
"abstract_canon_sha256": "a0915f1e26d8d276b03a02d4f7e3d2c4bb7a9e48a8775249cedac8d386349765",
"cross_cats_sorted": [],
"license": "http://creativecommons.org/licenses/by-sa/4.0/",
"primary_cat": "cs.CV",
"submitted_at": "2026-05-09T00:00:47Z",
"title_canon_sha256": "6213f41c9dc6708847fe33372b83d93063653fb3ec4ccedb82bc15255d6b4ebc"
},
"schema_version": "1.0",
"source": {
"id": "2605.08567",
"kind": "arxiv",
"version": 2
}
}