triple-walk 0.0.4

A pytorch extension library to perform triple walks on knowledge graphs

A simple algorithm to learn embeddings of entities in knowledge graph.

What is it

TripleWalk is an algorithm to learn vector embeddings of entities in a knowledge graph by performing random walks on triples.

Installation

Please not this package is only available for python 3.8+ and Pytorch >= 1.9.0

Install from PyPI

pip install triple-walk

Install from Github

pip install git+https://github.com/udel-cbcb/triple_walk.git#egg=triple_walk

Triple Walk

Author : Sachin Gavali

Requirements

1. Pytorch >= 1.9.0
2. NVIDIA-GPU (Cuda Toolkit >= 11.4
3. AMD-GPU (ROCM == 4.0.1)
4. Python == 3.8

Examples

• SkipGram Triple Walk model : SkipGramTriple
• CBOW Triple Walk model : CBOWTriple