Pith Number

pith:CW7IWZ46

pith:2026:CW7IWZ46YM7VWQNJOBNGPBBPUG

not attested not anchored not stored refs pending

LangForce: Bayesian Decomposition of Vision Language Action Models via Latent Action Queries

Bin Yu, Changti Wu, Cong Huang, Kai Chen, Laurence T. Yang, Shijie Lian, Xiaopeng Lin, Yuzhuo Miao, Zhaolong Shen

LangForce restores language grounding in vision-language-action models by maximizing conditional pointwise mutual information between instructions and actions.

arxiv:2601.15197 v6 · 2026-01-21 · cs.AI · cs.CL · cs.CV · cs.RO

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{CW7IWZ46YM7VWQNJOBNGPBBPUG}

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

Without requiring new data, LangForce significantly improves generalization... including an 11.3% improvement on the challenging OOD SimplerEnv benchmark.

C2weakest assumption

That the identified information collapse is the dominant cause of poor OOD performance and that maximizing conditional PMI via the dual-branch architecture will reliably correct it without introducing new failure modes or requiring careful hyperparameter tuning.

C3one line summary

LangForce prevents vision-language-action models from ignoring language by maximizing conditional pointwise mutual information via Bayesian decomposition with latent action queries.

Formal links

1 machine-checked theorem link

Cited by

2 papers in Pith

VP-VLA: Visual Prompting as an Interface for Vision-Language-Action Models

FrameSkip: Learning from Fewer but More Informative Frames in VLA Training

Receipt and verification

First computed	2026-05-18T03:09:24.337228Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

15be8b679ec33f5b41a9705a67842fa1981380046c4fd4fb2995a6d2eef2fecf

Aliases

arxiv: 2601.15197 · arxiv_version: 2601.15197v6 · doi: 10.48550/arxiv.2601.15197 · pith_short_12: CW7IWZ46YM7V · pith_short_16: CW7IWZ46YM7VWQNJ · pith_short_8: CW7IWZ46

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/CW7IWZ46YM7VWQNJOBNGPBBPUG \
  | 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: 15be8b679ec33f5b41a9705a67842fa1981380046c4fd4fb2995a6d2eef2fecf

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "5d8f3b3c3e63e88ea90b00a8b588f3289b724b72ca3ba6584ba3dfe04fad01cd",
    "cross_cats_sorted": [
      "cs.CL",
      "cs.CV",
      "cs.RO"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.AI",
    "submitted_at": "2026-01-21T17:15:22Z",
    "title_canon_sha256": "a2f940e430f04da099d006def40320a211a8eaabe42cdfdf8b048208a6a7d653"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2601.15197",
    "kind": "arxiv",
    "version": 6
  }
}