pith:GHJDZU6C
Test-Time Training for Visual Foresight Vision-Language-Action Models
Test-time training on predicted future images and real observations lets visual foresight VLA models adapt to out-of-distribution shifts without any architecture changes.
arxiv:2605.08215 v2 · 2026-05-06 · cs.CV · cs.LG · cs.RO
Add to your LaTeX paper
\usepackage{pith}
\pithnumber{GHJDZU6CXRSJTIVU2SSMK4PHIV}
Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge
Record completeness
Claims
Empirically, T³VF mitigates the OOD vulnerability of VF-VLA at a modest additional inference cost, without requiring any architectural modification or auxiliary modules.
That the predicted future image and its subsequent real observation form a reliable, low-noise supervision signal even under OOD shifts, and that the adaptive update filter can consistently separate useful from harmful updates without additional labeled data or validation.
T³VF applies test-time training with adaptive filtering to reduce OOD failures in VF-VLA models by treating predicted future images and actual next observations as natural training pairs.
Receipt and verification
| First computed | 2026-06-05T01:14:40.292163Z |
|---|---|
| Builder | pith-number-builder-2026-05-17-v1 |
| Signature | Pith Ed25519
(pith-v1-2026-05) · public key |
| Schema | pith-number/v1.0 |
Canonical hash
31d23cd3c2bc6499a2b4d4a4c571e745624a15b761741a5359e1e03a0debbef2
Aliases
· · · · ·Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/GHJDZU6CXRSJTIVU2SSMK4PHIV \
| 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: 31d23cd3c2bc6499a2b4d4a4c571e745624a15b761741a5359e1e03a0debbef2
Canonical record JSON
{
"metadata": {
"abstract_canon_sha256": "896e3a1d2333f6cc575dac8d20f969d70690dd812406fef9a5abfbb5c8eff41c",
"cross_cats_sorted": [
"cs.LG",
"cs.RO"
],
"license": "http://creativecommons.org/licenses/by/4.0/",
"primary_cat": "cs.CV",
"submitted_at": "2026-05-06T11:21:25Z",
"title_canon_sha256": "bda610f8b6f7851fc2804eca9f7bbc1bf041355f6a11adbfdd6d8b483488e6cb"
},
"schema_version": "1.0",
"source": {
"id": "2605.08215",
"kind": "arxiv",
"version": 2
}
}