Skip to content

Dataset · mm-graph-org

Benchmark for multimodal graph neural networks

mm-graph-org/mm-graph

A collection of image and text datasets for evaluating models on multimodal graphs.

What's inside

The repository brings together all datasets used in the Multimodal Graph Benchmark research paper. It contains associated text and visual data (from 1 to 10 million entries) structured for graph machine learning tasks. The dataset is designed for reproducing research results and testing multimodal architectures.

What you can do with it

  • Evaluating and comparing multimodal graph neural networks (MM-GNN)
  • Training models for node classification and link prediction with text and images
  • Researching methods for combining visual and text embeddings in graph structures
Size
1M–10M rows
Downloads this month
932
Likes
4

How to load it

from datasets import load_dataset

ds = load_dataset("mm-graph-org/mm-graph")