Overview of the Temporal Graph Benchmark (TGB) pipeline:
- TGB includes large-scale and realistic datasets from five different domains with both dynamic link prediction and node property prediction tasks
- TGB automatically downloads datasets and processes them into
numpy
,PyTorch
andPyG compatible TemporalData
formats. - Novel TG models can be easily evaluated on TGB datasets via reproducible and realistic evaluation protocols.
- TGB provides public and online leaderboards to track recent developments in temporal graph learning domain
To submit to TGB leaderboard, please fill in this google form
See all version differences and update notes here
Please update to version 0.8.0
fixing metric computation issue in node property prediction task, tgbn
leaderboards results are updated to reflect the changes.
Please refer to examples/nodeproppred/
example folders to how to compute the metric correctly. No changes for linkproppred
datasets.
the negative samples for the tgbl-wiki
and tgbl-review
dataset has been updated and redownload of the dataset would be needed (will be prompted automatically in this version when you use the dataloader)
You can install TGB via pip
pip install py-tgb
The project website can be found here.
The API documentations can be found here.
all dataset download links can be found at info.py
TGB dataloader will also automatically download the dataset as well as the negative samples for the link property prediction datasets.
if website is unaccessible, please use this link instead.
- For the dynamic link property prediction task, see the
examples/linkproppred
folder for example scripts to run TGN, DyRep and EdgeBank on TGB datasets. - For the dynamic node property prediction task, see the
examples/nodeproppred
folder for example scripts to run TGN, DyRep and EdgeBank on TGB datasets. - For all other baselines, please see the TGB_Baselines repo.
Our implementation works with python >= 3.9 and can be installed as follows
- set up virtual environment (conda should work as well)
python -m venv ~/tgb_env/
source ~/tgb_env/bin/activate
- install external packages
pip install pandas==1.5.3
pip install matplotlib==3.7.1
pip install clint==0.5.1
install Pytorch and PyG dependencies (needed to run the examples)
pip install torch==2.0.0 --index-url https://download.pytorch.org/whl/cu117
pip install torch_geometric==2.3.0
pip install pyg_lib torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.0.0+cu117.html
- install local dependencies under root directory
/TGB
pip install -e .
- first run the mkdocs server locally in your terminal
mkdocs serve
- go to the local hosted web address similar to
[14:18:13] Browser connected: http://127.0.0.1:8000/
Example: to track documentation of a new hi.py file in tgb/edgeregression/hi.py
- create docs/api/tgb.hi.md and add the following
# `tgb.edgeregression`
::: tgb.edgeregression.hi
- edit mkdocs.yml
nav:
- Overview: index.md
- About: about.md
- API:
other *.md files
- tgb.edgeregression: api/tgb.hi.md
git fetch origin
git checkout -b test origin/test
pip install mkdocs
pip install mkdocs-material
pip install mkdocstrings-python
pip install mkdocs-jupyter
pip install notebook
Our implementation works with python >= 3.9 and has the following dependencies
pytorch == 2.0.0
torch-geometric == 2.3.0
torch-scatter==2.1.1
torch-sparse==0.6.17
torch-spline-conv==1.2.2
pandas==1.5.3
clint==0.5.1
We thank the OGB team for their support throughout this project and sharing their website code for the construction of TGB website.