Pith Number

pith:SMUSK7IX

pith:2026:SMUSK7IXTVJTAMSNRIWZLWYFP6

not attested not anchored not stored refs resolved

Grounded or Guessing? LVLM Confidence Estimation via Blind-Image Contrastive Ranking

Charese H. Smiley, Erfan Miahi, Ivan Brugere, Kundan Thind, Mohammad M. Ghassemi, Reza Khanmohammadi, Simerjot Kaur

Training probes to prefer real images over blacked-out ones lets them detect when vision-language models actually use visual input for their answers.

arxiv:2605.10893 v2 · 2026-05-11 · cs.CL

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{SMUSK7IXTVJTAMSNRIWZLWYFP6}

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

BICR achieves the best cross-LVLM average on both calibration and discrimination simultaneously, with statistically significant discrimination gains robust to cluster-aware analysis at 4-18x fewer parameters than the strongest probing baseline.

C2weakest assumption

That penalizing higher confidence on the blacked-out image view via ranking loss on hidden states will cause the probe to reliably treat the presence of visual information as a signal of prediction reliability (abstract, paragraph describing the training objective).

C3one line summary

BICR trains a lightweight probe on contrastive hidden states from real versus blind images to detect visual grounding in LVLM predictions, outperforming baselines on calibration and discrimination with fewer parameters.

References

61 extracted · 61 resolved · 11 Pith anchors

[1] Mitigating hallucination in large vision- language models via modular attribution and intervention 2025

[2] Don’t miss the forest for the trees: Attentional vision calibration for large vision language models 2025

[3] doi: 10.18653/v1/ 2024.findings-acl.586 2025 · doi:10.18653/v1/

[4] Hidden in plain sight: VLMs overlook their visual representations 2025

[5] Reference- free hallucination detection for large vision-language models 2024 · doi:10.18653/v1/2024.findings-emnlp.262

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-20T00:00:42.569712Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

9329257d179d5330324d8a2d95db057fb2ac517f81a169b7324ac6e87e36a8c8

Aliases

arxiv: 2605.10893 · arxiv_version: 2605.10893v2 · doi: 10.48550/arxiv.2605.10893 · pith_short_12: SMUSK7IXTVJT · pith_short_16: SMUSK7IXTVJTAMSN · pith_short_8: SMUSK7IX

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/SMUSK7IXTVJTAMSNRIWZLWYFP6 \
  | 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: 9329257d179d5330324d8a2d95db057fb2ac517f81a169b7324ac6e87e36a8c8

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "a387993af524cff36faabb839a651d8c05bc86143c44b1cce94dc3d3dcf8da7a",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2026-05-11T17:35:10Z",
    "title_canon_sha256": "520b3966640d7ace8afa9a7ea8c97592ce5ac4545ecd0961391f3fd8adb4001e"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.10893",
    "kind": "arxiv",
    "version": 2
  }
}