mOSCAR: A Large-scale Multilingual and Multimodal Document-level Corpus
Futeral, Matthieu, Zebaze, Armel, Suarez, Pedro Ortiz, Abadji, Julien, Lacroix, Rémi, Schmid, Cordelia, Bawden, Rachel, Sagot, Benoît
–arXiv.org Artificial Intelligence
Multimodal Large Language Models (mLLMs) are trained on a large amount of text-image data. While most mLLMs are trained on caption-like data only, Alayrac et al. [2022] showed that additionally training them on interleaved sequences of text and images can lead to the emergence of in-context learning capabilities. However, the dataset they used, M3W, is not public and is only in English. There have been attempts to reproduce their results but the released datasets are English-only. In contrast, current multilingual and multimodal datasets are either composed of caption-like only or medium-scale or fully private data. This limits mLLM research for the 7,000 other languages spoken in the world. We therefore introduce mOSCAR, to the best of our knowledge the first large-scale multilingual and multimodal document corpus crawled from the web. It covers 163 languages, 315M documents, 214B tokens and 1.2B images. We carefully conduct a set of filtering and evaluation steps to make sure mOSCAR is sufficiently safe, diverse and of good quality. We additionally train two types of multilingual model to prove the benefits of mOSCAR: (1) a model trained on a subset of mOSCAR and captioning data and (2) a model train on captioning data only. The model additionally trained on mOSCAR shows a strong boost in few-shot learning performance across various multilingual image-text tasks and benchmarks, confirming previous findings for English-only mLLMs.
arXiv.org Artificial Intelligence
Jun-12-2024
- Country:
- South America
- Paraguay > Asunción
- Asunción (0.04)
- Colombia > Meta Department
- Villavicencio (0.04)
- Chile > Santiago Metropolitan Region
- Santiago Province > Santiago (0.04)
- Paraguay > Asunción
- Oceania > Australia
- North America
- Dominican Republic (0.04)
- United States
- New York (0.04)
- Maryland > Baltimore (0.04)
- Tennessee > Davidson County
- Nashville (0.04)
- Pennsylvania > Philadelphia County
- Philadelphia (0.04)
- Hawaii > Honolulu County
- Honolulu (0.04)
- California
- Santa Clara County > Palo Alto (0.04)
- San Diego County > San Diego (0.04)
- Canada
- Ontario > Toronto (0.04)
- British Columbia > Metro Vancouver Regional District
- Vancouver (0.04)
- Europe
- Poland (0.04)
- United Kingdom
- Wales > Cardiff (0.04)
- Scotland > City of Edinburgh
- Edinburgh (0.04)
- England > Greater London
- London (0.04)
- Switzerland > Zürich
- Zürich (0.14)
- Italy > Tuscany
- Florence (0.04)
- Ireland > Leinster
- County Dublin > Dublin (0.04)
- France
- Île-de-France > Paris
- Paris (0.04)
- Provence-Alpes-Côte d'Azur > Bouches-du-Rhône
- Marseille (0.04)
- Occitanie > Haute-Garonne
- Toulouse (0.04)
- Île-de-France > Paris
- Denmark > Capital Region
- Copenhagen (0.04)
- Belgium > Brussels-Capital Region
- Brussels (0.04)
- Asia
- Philippines > Luzon
- Ilocos Region > Province of Pangasinan (0.04)
- Middle East
- Jordan (0.04)
- UAE > Abu Dhabi Emirate
- Abu Dhabi (0.04)
- Philippines > Luzon
- Africa > Rwanda
- South America
- Genre:
- Research Report (0.40)
- Industry:
- Information Technology > Security & Privacy (0.48)
- Technology: