Skip to content

Dataset · arsaporta

Symile-M3 multilingual audio-visual-text dataset

arsaporta/symile-m3

Multilingual dataset of audio, images, and text for evaluating multimodal models.

What's inside

Contains samples across three modalities (audio, images, text) in six languages (Arabic, Greek, English, Hindi, Japanese, Korean). Designed to test models' ability to capture data relationships where both text and audio are needed for image prediction.

What you can do with it

  • Training and evaluating multimodal language models
  • Zero-shot classification and zero-shot image classification tasks
Tasks
Zero shot classification, Zero shot image classification
Languages
Arabic, Greek, English, Hindi, Japanese, Korean, Telugu, Thai and 2 more
Size
10M–100M rows
Downloads this month
24.3K
Likes
8

How to load it

from datasets import load_dataset

ds = load_dataset("arsaporta/symile-m3")