NBFNet: Neural Bellman-Ford Networks

This is the official codebase of the paper

Neural Bellman-Ford Networks: A General Graph Neural Network Framework for Link Prediction

Zhaocheng Zhu,
Zuobai Zhang,
Louis-Pascal Xhonneux,
Jian Tang

NeurIPS 2021


NBFNet is a graph neural network framework inspired by traditional path-based
methods. It enjoys the advantages of both traditional path-based methods and modern
graph neural networks, including generalization in the inductive setting,
interpretability, high model capacity and scalability. NBFNet can be
applied to solve link prediction on both homogeneous graphs and knowledge graphs.


This codebase is based on PyTorch and TorchDrug. It supports training and inference
with multiple GPUs or multiple machines.


You may install the dependencies via either conda or pip. Generally, NBFNet works
with Python 3.7/3.8 and PyTorch version >= 1.8.0.

From Conda

conda install torchdrug pytorch=1.8.2 cudatoolkit=11.1 -c milagraph -c pytorch-lts -c pyg -c conda-forge
conda install ogb easydict pyyaml -c conda-forge

From Pip

pip install torch==1.8.2+cu111 -f https://download.pytorch.org/whl/lts/1.8/torch_lts.html
pip install torchdrug
pip install ogb easydict pyyaml


To reproduce the results of NBFNet, use the following command. All the datasets
will be automatically downloaded in the code.

python script/run.py -c config/inductive/wn18rr.yaml --gpus [0] --version v1

We provide the hyperparameters for each experiment in configuration files.
All the configuration files can be found in config/*/*.yaml.

For experiments on inductive relation prediction, you need to additionally specify
the split version with --version v1.

To run NBFNet with multiple GPUs or multiple machines, use the following commands

python -m torch.distributed.launch --nproc_per_node=4 script/run.py -c config/inductive/wn18rr.yaml --gpus [0,1,2,3]

python -m torch.distributed.launch --nnodes=4 --nproc_per_node=4 script/run.py -c config/inductive/wn18rr.yaml --gpus[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]

Visualize Interpretations on FB15k-237

Once you have models trained on FB15k237, you can visualize the path interpretations
with the following line. Please replace the checkpoint with your own path.

python script/visualize.py -c config/knowledge_graph/fb15k237_visualize.yaml --checkpoint /path/to/nbfnet/experiment/model_epoch_20.pth

Evaluate ogbl-biokg

Due to the large size of ogbl-biokg, we only evaluate on a small portion of the
validation set during training. The following line evaluates a model on the full
validation / test sets of ogbl-biokg. Please replace the checkpoint with your own

python script/run.py -c config/knowledge_graph/ogbl-biokg_test.yaml --checkpoint /path/to/nbfnet/experiment/model_epoch_10.pth


Here are the results of NBFNet on standard benchmark datasets. All the results are
obtained with 4 V100 GPUs (32GB). Note results may be slightly different if the
model is trained with 1 GPU and/or a smaller batch size.

Knowledge Graph Completion

Dataset MR MRR [email protected] [email protected] [email protected]
FB15k-237 114 0.415 0.321 0.454 0.599
WN18RR 636 0.551 0.497 0.573 0.666
ogbl-biokg 0.829 0.768 0.870 0.946

Homogeneous Graph Link Prediction

Dataset AUROC AP
Cora 0.956 0.962
CiteSeer 0.923 0.936
PubMed 0.983 0.982

Inductive Relation Prediction

Dataset [email protected] (50 sample)
v1 v2 v3 v4
FB15k-237 0.834 0.949 0.951 0.960
WN18RR 0.948 0.905 0.893 0.890

Frequently Asked Questions

  1. The code is stuck at the beginning of epoch 0.

    This is probably because the JIT cache is broken.
    Try rm -r ~/.cache/torch_extensions/* and run the code again.


If you find this codebase useful in your research, please cite the following paper.

  title={Neural Bellman-Ford Networks: A General Graph Neural Network Framework for Link Prediction},
  author={Zhu, Zhaocheng and Zhang, Zuobai and Xhonneux, Louis-Pascal and Tang, Jian},
  journal={arXiv preprint arXiv:2106.06935},


View Github