zhangdddong/SimRE

SimRE: Simple Contrastive Learning with Soft Logical Rule for Knowledge Graph Embedding

SimRE: Simple Contrastive Learning with Soft Logical Rule for Knowledge Graph Embedding

Official code repository "SimRE: Simple Contrastive Learning with Soft Logical Rule for Knowledge Graph Embedding".

Requirements

• python>=3.7
• torch>=1.6 (for mixed precision training)
• transformers>=4.15

How to Run

Step 1: preprocess the dataset

bash data/preprocess.sh

Step 2: mining rule by RNNLogic

https://github.com/DeepGraphLearning/RNNLogic

Step 3, training the model

python src/main.py

Step 4, evaluate a trained model

python src/evaluate.py