Home

Awesome

HousE: Knowledge Graph Embedding with Householder Parameterization

This is the code of the paper HousE: Knowledge Graph Embedding with Householder Parameterization for ICML 2022.

A more powerful and general framework for knowledge graph embedding.

Requirements

Data

Models

Usage

All training commands are listed in best_config.sh. For example, you can run the following commands to train HousE on WN18RR and FB15k-237 datasets.

# WN18RR
bash run.sh HousE wn18rr 0 0 0 1000 200 800 8 1 0.5 6.0 1.14940435933987 0.000575323908649059 60000 20000 8 0.0960737047401994

# FB15k-237
bash run.sh HousE FB15k-237 0 0 0 500 500 600 20 6 0.6 5.0 2.00378388680359 0.000794267891285676 100000 10000 16 0.00336727231946076

Acknowledgement

We refer to the code of RotatE. Thanks for their contributions.