Skip to content

Dataset · Babelscape

REBEL dataset for relation extraction from text

Babelscape/rebel-dataset

A large English corpus for training entity relation extraction models.

What's inside

An English text dataset created for academic research on end-to-end Relation Extraction. Volume ranges from 1 to 10 million entries. Data is automatically labeled (silver dataset).

What you can do with it

  • Training models for factual extraction and knowledge graphs from text
  • Fine-tuning language models for text retrieval tasks
  • Building automated structuring systems for unstructured documents
Tasks
Text retrieval, Text generation
Languages
English
Size
1M–10M rows
Downloads this month
527
Likes
35

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("Babelscape/rebel-dataset")