Skip to content

Latest commit

 

History

History
54 lines (43 loc) · 1.79 KB

README.md

File metadata and controls

54 lines (43 loc) · 1.79 KB
Raw

QIQE-KGC

source code and data for "Knowledge Graph Completion Method Based on Quantum Embedding and Quaternion Interaction Enhancement"

Because github limits the size of uploaded files, we put the code and data in the google cloud disk.[https://drive.google.com/file/d/1tueH04ZIBCzSuWiKTBMl5UAPuU3JRzFk/view?usp=sharing] A more detailed introduction is provided after the dissertation is accepted.

Our model is based on OpenKE and E2R.

Requirements

  • python>=3.7
  • torch>=1.8
  • tqdm

All experiments are run with 2 RTX 3090(24GB) GPUs.

How to Run

Each data set we have equipped with independent files and hyperparameters

1.If only to reproduce our results

Download the model from Google cloud disk and unzip it. Take the reproduction of FB15K dataset as an example.Go to the model file first.

cd QIQE-KGC
cd fb15k

Run the underlying C++ file of OpenKE, and run Test.py directly to reproduce the results.

bash bash make.sh
python Test.py

2.Train the model yourself

Training a model can be time consuming. Download the model from Google cloud disk and unzip it. Take the reproduction of FB15K dataset as an example.Go to the model file first.

cd QIQE-KGC
cd fb15k

Run the underlying C++ file of OpenKE, and run Test.py directly to reproduce the results.

bash make.sh
python Train.py
python Test.py

Troubleshooting

1.OSError: /data/KE/OpenKE/release/Base.so: invalid ELF header.

Please don't forget to run the underlying C++ file of OpenKE first.

2.RuntimeError: Attempting to deserialize object on CUDA device 1 but torch.cuda.device_count() is 1. Please use torch.load with map_location to map your storages to an existing device.

If you only have one GPU, modify the cuda number.