tchaton/lightning-geometric

Integrate pytorch

TorchScripted Pytorch Geometric Examples with Pytorch Lightning and Hydra

Setup on MacOs. Please, adapt to others OS :)

brew install cmake
pyenv install 3.7.8
pyenv local 3.7.8
python -m venv
source .venv/bin/activate
poetry install

PRINCIPAL CMD

python train.py model={{MODEL}} dataset={{DATASET}} loggers={{LOGGERS}} log={{LOG}} notes={{NOTES}} name={{NAME}} jit={{JIT}}

• LOGGERS str: Configuration file to log to Wandb, currently using mine as thomas-chaton
• LOG bool: Wheter to log training to wandb
• NOTES str: A note associated to the training
• NAME str: Training name appearing on Wandb.
• LOG bool: Wheter to make model jittable.

Working Inference

Have a look at test/test_inference.py

SUPPORTED COMBINAISONS

{{DATASET}}	{{MODEL}}	DATASET DESCRIPTION	MODEL DESCRIPTION	WORKING
zinc	pna	The ZINC dataset from the "Grammar Variational Autoencoder" https://arxiv.org/abs/1703.01925	The Principal Neighbourhood Aggregation graph convolution operator from the "Principal Neighbourhood Aggregation for Graph Nets" https://arxiv.org/abs/2004.05718	True
faust	spline	The FAUST humans dataset from the "FAUST: Dataset and Evaluation for 3D Mesh Registration" http://files.is.tue.mpg.de/black/papers/FAUST2014.pdf	The spline-based convolutional operator from the "SplineCNN: Fast Geometric Deep Learning with Continuous B-Spline Kernels"https://arxiv.org/abs/1711.08920	In progress
ppi	gat	The protein-protein interaction networks from the "Predicting Multicellular Function through Multi-layer Tissue Networks" https://arxiv.org/abs/1707.04638	The graph attentional operator from the "Graph Attention Networks" https://arxiv.org/abs/1710.10903 True	True
cora	agnn	The citation network datasets "Cora", "CiteSeer" and "PubMed" from the "Revisiting Semi-Supervised Learning with Graph Embeddings" https://arxiv.org/abs/1603.08861	"Attention-based Graph Neural Network for Semi-Supervised Learning" https://arxiv.org/abs/1803.03735	True
cora	sage	""	The GraphSAGE operator from the "Inductive Representation Learning on Large Graphs" https://arxiv.org/abs/1706.02216	True
cora	sgc	""	The simple graph convolutional operator from the "Simplifying Graph Convolutional Networks" https://arxiv.org/abs/1902.07153	True
cora	tag	""	The topology adaptive graph convolutional networks operator from the "Topology Adaptive Graph Convolutional Networks" https://arxiv.org/abs/1710.10370	True
cora	dna	""	The dynamic neighborhood aggregation operator from the "Just Jump: Towards Dynamic Neighborhood Aggregation in Graph Neural Networks" https://arxiv.org/abs/1904.04849	True
reddit	sage	The Reddit dataset from the "Inductive Representation Learning on Large Graphs" https://arxiv.org/abs/1706.02216	""	True
reddit	agnn	""	""	True
icews18	renet	The Integrated Crisis Early Warning System (ICEWS) dataset used in the, e.g., "Recurrent Event Network for Reasoning over Temporal Knowledge Graphs" https://arxiv.org/abs/1904.05530	The Recurrent Event Network model from the "Recurrent Event Network for Reasoning over Temporal Knowledge Graphs" https://arxiv.org/abs/1904.05530	Waiting for support for TGCN
cora	argva	""	The Adversarially Regularized Variational Graph Auto-Encoder model from the "Adversarially Regularized Graph Autoencoder for Graph Embedding" https://arxiv.org/abs/1802.04407`	True
cora	arma	""	The ARMA graph convolutional operator from the "Graph Neural Networks with Convolutional ARMA Filters" https://arxiv.org/abs/1901.01343>	True
cora	gcn	""	The GCN graph convolutional operator from the "Semi Supervised Classification with Graph Convolution Networks" https://arxiv.org/pdf/1609.02907.pdf.01343>	True
cora	gcn2	""	The graph convolutional operator with initial residual connections and identity mapping (GCNII) from the "Simple and Deep Graph Convolutional Networks" https://arxiv.org/abs/2007.02133	True

DATASET SIZES

529M    ./Flickr
 74M    ./FAUST
 16M    ./cora
3.5G    ./Reddit
383M    ./ZINC
1.8G    ./MNISTSuperpixels
182M    ./OgbnArxiv
192M    ./PPI
156M    ./ICEWS18
6.8G    .