ARXIV:2604.18347 · VISION-LANGUAGE MODELS · SUBMITTED 21 APR · 04:18 UTC · FRESHNESS STALE

VerifiedSource: PDF linkedVerifiedPaperPack: citation fields availablePartialProof: unverified proof status

Multilingual Training and Evaluation Resources for Vision-Language Models

Daniela Baiamonte · Elena Fano · Matteo Gabburo · Stefano Simonazzi · Leonardo Rigutini · Andrea Zugarini · arXiv

This work introduces a comprehensive suite of multilingual resources for training and evaluating Vision-Language Models, addressing the current English-centric limitations.

Ship in 2-4 weeks›Score5.0Evidence unverified

Opportunity summary

Pain This work introduces a comprehensive suite of multilingual resources for training and evaluating Vision-Language Models, addressing the current English-centric limitations.

Evidence 0 refs | 3 sources | 50% coverage

Blocker Evidence unverified

Open Build Read PDF Signal Canvas Track

PROBLEM

This work introduces a comprehensive suite of multilingual resources for training and evaluating Vision-Language Models, addressing the current English-centric limitations. However, despite their growth, VLMs development is heavily grounded on English, leading to two…

METHOD

Full abstract

Vision Language Models (VLMs) achieved rapid progress in the recent years. However, despite their growth, VLMs development is heavily grounded on English, leading to two main limitations: (i) the lack of multilingual and multimodal datasets for training, and (ii) the scarcity of comprehensive evaluation benchmarks across languages. In this work, we address these gaps by introducing a new comprehensive suite of resources for VLMs training and evaluation spanning five European languages (English, French, German, Italian, and Spanish). We adopt a regeneration-translation paradigm that produces high-quality cross-lingual resources by combining curated synthetic generation and manual annotation. Specifically, we build Multi-PixMo, a training corpus obtained regenerating examples from Pixmo pre-existing datasets with permissively licensed models: PixMo-Cap, PixMo-AskModelAnything, and CoSyn-400k. On the evaluation side, we construct a set of multilingual benchmarks derived translating widely used English datasets (MMbench, ScienceQA, MME, POPE, AI2D). We assess the quality of these resources through qualitative and quantitative human analyses, measuring inter-annotator agreement. Additionally, we perform ablation studies to demonstrate the impact of multilingual data, with respect to English only, in VLMs training. Experiments, comprising 3 different models show that using multilingual, multimodal examples for training VLMs aids is consistently beneficial on non-English benchmarks, with positive transfer to English as well.

RESULT

ScienceToStartup currently rates this 5.0/10 on the public viability pass. Additionally, we perform ablation studies to demonstrate the impact of multilingual data, with respect to English only, in VLMs training. Code availability is flagged…

WHY NOW

Vision-Language Models moved forward this cycle; last verified April 2026. Public score 5.0/10. Production flags indicate code availability.

Continue into Read for claims, analysis, references, and neighboring papers.

Opportunity summary

Score5.0

PainThis work introduces a comprehensive suite of multilingual resources for training and evaluating Vision-Language Models, addressing the current English-centric limitations.

Evidence0 refs | 3 sources | 50% coverage

Blockerno shell-level blocker reported

Analysis summary

This work introduces a comprehensive suite of multilingual resources for training and evaluating Vision-Language Models, addressing the current English-centric limitations.

VerifiedSource: PDF linkedVerifiedPaperPack: citation fields availablePartialProof: unverified proof status

Competitive landscape

This work introduces a comprehensive suite of multilingual resources for training and evaluating Vision-Language Models, addressing the current English-centric limitations.

Segment

Vision-Language Models

Adoption evidence

No public code link in the paper record yet

Commercial read

5.0/10 public viability

Direct

not classified

Adjacent

not classified

Substitute

not classified

Unknown

not classified

{ "contract_version": "paper-r2", "paper_id": "382fc46e-9a50-42eb-9c47-cb47ef5a8d89", "arxiv_id": "2604.18347", "canonical_route": "/paper/multilingual-training-and-evaluation-resources-for-vision-language-models", "active_tab": "synced from current hash by the drawer client", "selected_artifact": "multilingual-training-and-evaluation-resources-for-vision-language-models", "endpoints": { "paper_pack": "/api/v1/paper/multilingual-training-and-evaluation-resources-for-vision-language-models/paper-pack", "build_passport": "/api/v1/paper/multilingual-training-and-evaluation-resources-for-vision-language-models/build-passport", "mcp_resource": "sciencetostartup://surfaces/paper-workspace" } }

{ "surface": "paper", "mode": "paper", "query": "Multilingual Training and Evaluation Resources for Vision-Language Models", "normalized_query": "2604.18347", "route": "/paper/multilingual-training-and-evaluation-resources-for-vision-language-models", "paper_ref": "multilingual-training-and-evaluation-resources-for-vision-language-models", "topic_slug": null, "benchmark_ref": null, "dataset_ref": null }

{ "@context": "https://schema.org", "@graph": [ { "@type": "WebPage", "@id": "https://sciencetostartup.com/paper/multilingual-training-and-evaluation-resources-for-vision-language-models#webpage", "url": "https://sciencetostartup.com/paper/multilingual-training-and-evaluation-resources-for-vision-language-models", "name": "Multilingual Training and Evaluation Resources for Vision-Language Models", "description": "This work introduces a comprehensive suite of multilingual resources for training and evaluating Vision-Language Models, addressing the current English-centric limitations.", "isPartOf": { "@id": "https://sciencetostartup.com/#website" } }, { "@type": "ScholarlyArticle", "@id": "https://sciencetostartup.com/paper/multilingual-training-and-evaluation-resources-for-vision-language-models#scholarlyArticle", "headline": "Multilingual Training and Evaluation Resources for Vision-Language Models", "description": "This work introduces a comprehensive suite of multilingual resources for training and evaluating Vision-Language Models, addressing the current English-centric limitations.", "url": "https://sciencetostartup.com/paper/multilingual-training-and-evaluation-resources-for-vision-language-models", "sameAs": "https://arxiv.org/abs/2604.18347", "identifier": { "@type": "PropertyValue", "propertyID": "arXiv", "value": "2604.18347" }, "isAccessibleForFree": true, "isPartOf": { "@id": "https://sciencetostartup.com/#website" }, "datePublished": "2026-04-20T14:42:47.000Z", "author": [ { "@type": "Person", "name": "Daniela Baiamonte" }, { "@type": "Person", "name": "Elena Fano" }, { "@type": "Person", "name": "Matteo Gabburo" }, { "@type": "Person", "name": "Stefano Simonazzi" }, { "@type": "Person", "name": "Leonardo Rigutini" }, { "@type": "Person", "name": "Andrea Zugarini" } ], "additionalProperty": [ { "@type": "PropertyValue", "propertyID": "viabilityScore", "value": 5 }, { "@type": "PropertyValue", "propertyID": "researchDomain", "value": "Vision-Language Models" }, { "@type": "PropertyValue", "propertyID": "commercialReadiness", "value": "code" } ] }, { "@type": "BreadcrumbList", "itemListElement": [ { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://sciencetostartup.com" }, { "@type": "ListItem", "position": 2, "name": "Vision-Language Models", "item": "https://sciencetostartup.com/topics" }, { "@type": "ListItem", "position": 3, "name": "Multilingual Training and Evaluation Resources for Vision-La", "item": "https://sciencetostartup.com/paper/multilingual-training-and-evaluation-resources-for-vision-language-models" } ] } ] }

Competitive landscape

This work introduces a comprehensive suite of multilingual resources for training and evaluating Vision-Language Models, addressing the current English-centric limitations.

Segment

Vision-Language Models

Adoption evidence

No public code link in the paper record yet

Commercial read

5.0/10 public viability

Direct

not classified

Adjacent

not classified

Substitute

not classified

Unknown

not classified

Multilingual Training and Evaluation Resources for Vision-Language Models

Multilingual Training and Evaluation Resources for Vision-Language Models

Claim map

Constellation map

Competitive landscape

Buzz

PDF

REFERENCES

Related Papers

Related Resources

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline

Claim map

Constellation map

Competitive landscape

Buzz

PDF

REFERENCES

Related Papers

Related Resources

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline