Pith Number

pith:KD37OJPH

pith:2026:KD37OJPHLG7ECMVMI4CUQI777Z

not attested not anchored not stored refs resolved

DiM\textsuperscript{3}: Bridging Multilingual and Multimodal Models via Direction- and Magnitude-Aware Merging

Daling Wang, Ercong Nie, Hinrich Sch\"utze, Mengjie Zhao, Mingyang Wang, Shi Feng, Xiaocui Yang, Yongkang Liu, Zijing Wang

Selective merging of direction- and magnitude-aware residual updates injects multilingual capability into multimodal models without training.

arxiv:2605.12960 v1 · 2026-05-13 · 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{KD37OJPHLG7ECMVMI4CUQI777Z}

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

Experiments on multilingual benchmarks in both text-only and vision-language settings, covering 57 languages across LLaVA- and Qwen-based backbones, show that DiM3 consistently outperforms existing merging baselines, substantially improves multilingual performance over the original multimodal model, and remains competitive with dedicated multilingual multimodal fine-tuning while largely retaining general multimodal ability.

C2weakest assumption

The assumption that multilingual and multimodal residual updates are heterogeneous in a way that can be selectively composed per parameter dimension using direction and magnitude awareness without unintended interference in the shared backbone.

C3one line summary

DiM3 merges multilingual and multimodal model updates in a direction- and magnitude-aware way to enhance multilingual performance in vision-language models while preserving original multimodal abilities.

References

71 extracted · 71 resolved · 10 Pith anchors

[1] OpenAI GPT-5 System Card 2025 · arXiv:2601.03267

[2] Gemini: A Family of Highly Capable Multimodal Models 2023 · arXiv:2312.11805

[3] InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency 2025 · arXiv:2508.18265

[4] Qwen3 Technical Report 2025 · arXiv:2505.09388

[5] xgqa: Cross-lingual visual question answering 2022

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-18T03:09:09.204085Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

50f7f725e759be4132ac47054823fffe45fae4028a59fa662ff78cdd107f01a5

Aliases

arxiv: 2605.12960 · arxiv_version: 2605.12960v1 · doi: 10.48550/arxiv.2605.12960 · pith_short_12: KD37OJPHLG7E · pith_short_16: KD37OJPHLG7ECMVM · pith_short_8: KD37OJPH

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/KD37OJPHLG7ECMVMI4CUQI777Z \
  | 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: 50f7f725e759be4132ac47054823fffe45fae4028a59fa662ff78cdd107f01a5

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "7fcbce155f5a653bc4d9cb8f097abf3f8237354f11cd20731ddaa486a6d9045c",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2026-05-13T03:50:54Z",
    "title_canon_sha256": "1373ba8a1da4cf3771021660241b7d85437ab61e60bb136f615a8630a94fcbb4"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.12960",
    "kind": "arxiv",
    "version": 1
  }
}