Pith Number

pith:TQVTQAHA

pith:2026:TQVTQAHA4TZTBIG3AZNVKM2QFP

not attested not anchored not stored refs pending

Reinforcing VLAs in Task-Agnostic World Models

Fengming Zhang, Junjie Lu, Kaixin Wang, Li Zhao, Rui Yu, Tianxiang Zhang, Xinyao Qin, Yucen Wang

A task-agnostic world model pre-trained on diverse behaviors combined with an off-the-shelf VLM allows VLAs to be fine-tuned for new tasks entirely through zero-shot imagined rollouts.

arxiv:2605.12334 v2 · 2026-05-12 · cs.AI

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{TQVTQAHA4TZTBIG3AZNVKM2QFP}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

Because both components are task-agnostic, VLAs can be readily finetuned for any new task entirely within this zero-shot imagination. ... proving that generalized physical priors can effectively substitute for costly task-dependent data.

C2weakest assumption

That a world model pre-trained solely on diverse task-free behaviors will capture sufficiently accurate and transferable physical priors to support reliable zero-shot inference and reward generation via an off-the-shelf VLM on unseen tasks.

C3one line summary

RAW-Dream lets VLAs learn new tasks in zero-shot imagination by using a world model pre-trained only on task-free behaviors and an unmodified VLM to supply rewards, with dual-noise verification to limit hallucinations.

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-21T01:05:21.143258Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

9c2b3800e0e4f330a0db065b5533502be6a279846a5c6663d0e34e9018acc6d5

Aliases

arxiv: 2605.12334 · arxiv_version: 2605.12334v2 · doi: 10.48550/arxiv.2605.12334 · pith_short_12: TQVTQAHA4TZT · pith_short_16: TQVTQAHA4TZTBIG3 · pith_short_8: TQVTQAHA

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/TQVTQAHA4TZTBIG3AZNVKM2QFP \
  | 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: 9c2b3800e0e4f330a0db065b5533502be6a279846a5c6663d0e34e9018acc6d5

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "3598c8cdc6505cd98e33bc9e9469709671951f6be6b91282e5376e315a0c1aff",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.AI",
    "submitted_at": "2026-05-12T16:16:15Z",
    "title_canon_sha256": "8cf227e754c8f31afc33536280ebbe3a5c85859213b19fcd75c2979ea7e69660"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.12334",
    "kind": "arxiv",
    "version": 2
  }
}