Pith Number

pith:FHXUMIEK

pith:2026:FHXUMIEKFCKMMRYQ5VB3H26VSE

not attested not anchored not stored refs resolved

From Table to Cell: Attention for Better Reasoning with TABALIGN

Chunhe Wang, Guang Cheng, Hanwei Wu, Lei Ding, Tung Sum Thomas Kwok, Xiaofeng Lin, Xinyu Wang, Zeyong Zhang, Zhijiang Guo

TABALIGN pairs a diffusion language model planner that emits binary cell masks with an attention verifier trained on human standards to enforce cell grounding in table reasoning.

arxiv:2605.14465 v1 · 2026-05-14 · 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{FHXUMIEKFCKMMRYQ5VB3H26VSE}

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

Across eight benchmarks covering table question answering and fact verification, TABALIGN improves average accuracy by 15.76 percentage points over the strongest open-source baseline at comparable 8B-class scale, with a matched-backbone ablation attributing 2.87 percentage points of this gain to the DLM planner over an AR planner on a fixed reasoner.

C2weakest assumption

The 1,600 human-verified attention standards used to train TABATTN are representative and stable across different table layouts, domains, and model scales; if they are not, the verifier's scoring may not reliably enforce the cell-grounding contract.

C3one line summary

TABALIGN pairs a diffusion language model planner emitting binary cell masks with a trained attention verifier, raising average accuracy 15.76 points over strong baselines on eight table benchmarks while speeding execution 44.64%.

References

87 extracted · 87 resolved · 15 Pith anchors

[1] Adapting autoregressive vision language models for parallel diffusion decoding, 2025 2025

[2] Johnson, Jonathan Ho, Daniel Tarlow, and Rianne van den Berg

[3] URLhttps://openreview.net/forum?id=h7-XixPCAL

[4] Qwen3-VL Technical Report 2025 · arXiv:2511.21631

[5] Qwen2.5-VL Technical Report 2025 · arXiv:2502.13923

Receipt and verification

First computed	2026-05-17T23:39:06.731364Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

29ef46208a2894c64710ed43b3ebd5913f1406d4c3ba7d4c75de612c9832ec84

Aliases

arxiv: 2605.14465 · arxiv_version: 2605.14465v1 · doi: 10.48550/arxiv.2605.14465 · pith_short_12: FHXUMIEKFCKM · pith_short_16: FHXUMIEKFCKMMRYQ · pith_short_8: FHXUMIEK

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/FHXUMIEKFCKMMRYQ5VB3H26VSE \
  | 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: 29ef46208a2894c64710ed43b3ebd5913f1406d4c3ba7d4c75de612c9832ec84

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "9e7f88313850a5eaee3ab810d62ea492776f67bf233a5c15e27493754773f730",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.AI",
    "submitted_at": "2026-05-14T07:00:26Z",
    "title_canon_sha256": "d621ba3c98dd9262c7ecb24095d95a674207be560406ebfead348bb859db2e4a"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.14465",
    "kind": "arxiv",
    "version": 1
  }
}