CogDL: A Comprehensive Library for Graph Deep Learning (WWW 2023)
MIT License
Bot releases are visible (Hide)
The new v0.6 release updates the tutorials and adds more examples, such as GraphMAE, GraphMAE2, and BGRL.
Full Changelog: https://github.com/THUDM/cogdl/compare/v0.5.3...v0.6
Published by cenyk1230 over 2 years ago
The CogDL v0.5.3 release supports mixed-precision training by setting fp16=True and provides a basic example written by Jittor. It also updates the tutorial in the document, fixes downloading links of some datasets, and fixes potential bugs of operators.
Full Changelog: https://github.com/THUDM/cogdl/compare/v0.5.2...v0.5.3
Published by cenyk1230 almost 3 years ago
The CogDL 0.5.2 release adds a GNN example for ogbn-products and updates geom datasets. It also fixes some potential bugs including setting devices, using cpu for inference, etc.
Full Changelog: https://github.com/THUDM/cogdl/compare/v0.5.1...v0.5.2
Published by cenyk1230 almost 3 years ago
The CogDL 0.5.1 release adds fast operators including SpMM (cpu version) and scatter_max (cuda version). It also adds lots of datasets for node classification which can be found in this link.
Full Changelog: https://github.com/THUDM/cogdl/compare/v0.5.0...v0.5.1
Published by cenyk1230 almost 3 years ago
The CogDL 0.5.0 release focuses on modular design and ease of use. It designs and implements a unified training loop for GNN, which introduces DataWrapper
to help prepare the training/validation/test data and ModelWrapper
to define the training/validation/test steps.
Full Changelog: https://github.com/THUDM/cogdl/compare/0.4.1...v0.5.0
Published by cenyk1230 almost 3 years ago
The CogDL 0.5.0 release focuses on modular design and ease of use. It designs and implements a unified training loop for GNN, which introduces DataWrapper
to help prepare the training/validation/test data and ModelWrapper
to define the training/validation/test steps.
Published by cenyk1230 almost 3 years ago
The CogDL 0.5.0 release focuses on modular design and ease of use. It designs and implements a unified training loop for GNN, which introduces DataWrapper
to help prepare the training/validation/test data and ModelWrapper
to define the training/validation/test steps.
Published by THINK2TRY about 3 years ago
A new release! πππ
In the new v0.4.1 release, CogDL implements multiple deepgnn models and we also give a analysis of deepgnn in Chinese. Now CogDL. supports both reversible and actnn for memory efficiency to help build super deep GNNs. Come and have a try. BTW, we are glad to announce that we will give a tutorial on KDD 2021 in August. Please see this link for more details. π
revgcn
, revgat
, revgen
)l0fos
, aff30
, arxivvenue
.Published by cenyk1230 over 3 years ago
A new major release! πππ
The new v0.4.0 release refactors the data storage (from Data
to Graph
) and provides more fast operators to speed up GNN training. It also includes many self-supervised learning methods on graphs. BTW, we are glad to announce that we will give a tutorial on KDD 2021 in August. Please see this link for more details. π
Data
to Graph
), edge_index
from torch.Tensor
to tuple(Tensor, Tensor)
. The inputs of each GNN are unified as one parameter graph
.Published by cenyk1230 over 3 years ago
A new major release! πππ
It provides a fast spmm operator to speed up GNN training. We also release the first version of CogDL paper in arXiv. In the paper, we introduce the design, the characteristics, the features, and the reproducible leaderboards.
Welcome to join our slack!
Published by cenyk1230 almost 4 years ago
A new major release!! It includes easy-to-use experiment
and pipeline
APIs for all experiments and applications. It also provides oagbert
API. Thanks to all the contributors π
experiment
API (see examples/quick_start.py
for reference)automl
feature in experiment
API, the usage is in README
pipeline
API (see examples/pipeline.py
for reference)oagbert
API (see examples/oagbert.py
for reference)SGC
model (thanks to @KHTee)SGC-PN
model (thanks to @feng-y16)PPNP
model (thanks to @TiagoMAntunes)SAGPool
model (thanks to @frouioui)GDC_GCN
model (thanks to @kwyoke)JKNet
(thanks to @WXR1998)SIGN
model (thanks to @hmartelb)HGP-SL
model (thanks to @Sahandfer)DropEdge
model (thanks to @JiaYiLiJayee)Graph U-Net
modelPPRGo
modelnumba
transformers
Published by cenyk1230 almost 4 years ago
Trainer
API for flexible trainingSampler
API for training large-scale datasets and now supports GraphSAINT
sampler.STP-GNN
for pre-trainingGPT-GNN
for node classificationcomplex
, distmult
, rotate
, transe
)DeeperGCN
for node classificationGCNII
for node classificationJure's paper
.ogb
Published by cenyk1230 about 4 years ago
optuna
GCC
for graph classification: GCC
is a contrastive learning framework that implements unsupervised structural graph representation pre-training.GRAND
for node classification (thanks to @wzfhaha): GRAND
randomly drops node features in training process to implement data augmentatoin and achieves sota in benchmarks.DGI
for unsupervised node classification: DGI
applies local-global contrastive learning methods to train GNN and first achieves results comparable to semi-supervised methods in benchmarks.MVGRL
for unsupervised node classification: MVGRL
is a self-supervised approach based on contrastive multi-view learning to learn representations.ProNE++
for unsupervised node classification: ProNE++
employs graph filter and AutoML to help enhance node embeddings.GraphSAGE
for unsupervised node classification: unsupervised version of GraphSAGE.DisenGCN
for node classification: DisenGCN
disentangles node representations by separating different factors.CompGCN
/RGCN
for KG link prediction: RGCN
and CompGCN
are GNNs for knowledge graph embedding considering the type of edges.GCC
results for heterogeneous node classification taskoptuna
dgl.model_zoo
anymore.GCC
pre-trained model in saved/
Published by cenyk1230 about 4 years ago
The first open release includes basically everything in the repository.