facebookresearch/PyTorch-BigGraph
A distributed system for learning graph embeddings from large-scale graphs with billions of entities and trillions of edges.

Velocity · 7d
+1.2
★ / day
Trend
→steady
star history
PyTorch-BigGraph (PBG) is a distributed system developed by Facebook Research for learning graph embeddings for large-scale graphs. It trains on edge lists, producing feature vectors for each entity where adjacent entities are placed closer together in vector space. It supports GPU training and is designed for graphs with up to billions of entities.