This is the official codes for DSpar: Embarrassingly Simple Strategy for Efficient GNN training and inference via Degree-based Sparsification.

Install

This code is tested with Python 3.8 and CUDA 11.0. To reproduce the results in this paper, please follow the below configuration.

• Create and activate conda environment.

conda env create -f environment.yml
conda activate graph
pip install torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.0.0+cu117.html

• Build

cd src
pip install -v -e .

Reproduce results

Reproduce {Reddit, YELP} results.

cd mem_speed_bench
python ./non_ogb_datasets/train_full_batch.py --conf ./non_ogb_datasets/conf/$MODEL.yaml --$SPARSIFY --dataset $DATASET

MODEL must be chosen from {gcn, sage, gcn2}, SPARSIFY can be chosen from {spec_sparsify, random_sparsify}. spec_sparsity corresponds to Algorithm 1 in our paper. DATASET must be chosen from {reddit2, yelp}

If you do not want to apply any graph sparsification, just remove --$SPARSIFY

For mini-batch training,

cd mem_speed_bench
python ./non_ogbn_datasets/train_mini_batch.py --conf ./non_ogbn_datasets/conf/$MODEL.yaml --$SPARSIFY --grad_norm $GRAD_NORM

MODEL must be chosen from {saint_sage, cluster_gcn}. SPARSIFY can be chosen from {spec_sparsify, random_sparsify}. DATASET must be chosen from {reddit2, yelp}. For GRAD_NORM, it can found in appendix.

Reproduce ogbn-arxiv results.

cd mem_speed_bench
python ./arxiv/train_full_batch.py --conf ./arxiv/conf/$MODEL.yaml --$SPARSIFY

MODEL must be chosen from {gcn, sage, gcn2}, SPARSIFY can be chosen from {spec_sparsify, random_sparsify}. spec_sparsity corresponds to Algorithm 1 in our paper.

Reproduce ogbn-products results.

For full-batch training,

cd mem_speed_bench
python ./products/train_full_batch.py --conf ./products/conf/sage.yaml --$SPARSIFY

SPARSIFY can be chosen from {spec_sparsify, random_sparsify}.

For mini-batch training,

cd mem_speed_bench
python ./yaml/train_mini_batch.py --conf ./yaml/conf/$MODEL.yaml --$SPARSIFY

MODEL must be chosen from {cluster_sage, saint_sage}. SPARSIFY can be chosen from {spec_sparsify, random_sparsify}.

Get the occupied memory and training throughout.

Add the flag --deug_mem and --test_speed to the above commends. For example,

python ./arxiv/train_full_batch.py --conf ./arxiv/conf/$MODEL.yaml --n_bits $BIT_WIDTH $SPARSIFY --debug_mem --test_speed

Combining dspar and AMP

Add the flag --amp to the above commends.

Acknowledgment

Our code is based on the official code of GNNAutoScale.