Commit 57aaa363 authored by Khoa A Nguyen's avatar Khoa A Nguyen
Browse files

fixed folder

parent 48045172
MIT License
Copyright (c) 2020 Vijay Prakash Dwivedi, Chaitanya K. Joshi, Thomas Laurent, Yoshua Bengio, Xavier Bresson
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
# Benchmarking Graph Neural Networks
<br>
## Updates
**Nov 2, 2020**
* Project based on DGL 0.4.2. See the relevant dependencies defined in the environment yml files ([CPU](./environment_cpu.yml), [GPU](./environment_gpu.yml)).
+ Numerical experiments report faster training times with DGL 0.4.2 compared to DGL 0.5.2.
+ For the version of the project compatible with DGL 0.5.2 and relevant dependencies, please use this [branch](https://github.com/graphdeeplearning/benchmarking-gnns/tree/master-dgl-0.5.2).
* Added [ZINC-full](./data/script_download_molecules.sh) dataset (249K molecular graphs) with [scripts](./scripts/ZINC-full/).
**Jun 11, 2020**
* Second release of the project. Major updates :
+ Added experimental pipeline for Weisfeiler-Lehman-GNNs operating on dense rank-2 tensors.
+ Added a leaderboard for all datasets.
+ Updated PATTERN dataset.
+ Fixed bug for PATTERN and CLUSTER accuracy.
+ Moved first release to this [branch](https://github.com/graphdeeplearning/benchmarking-gnns/tree/arXivV1).
* New ArXiv's version of the [paper](https://arxiv.org/pdf/2003.00982.pdf).
**Mar 3, 2020**
* First release of the project.
<br>
<img src="./docs/gnns.jpg" align="right" width="350"/>
## 1. Benchmark installation
[Follow these instructions](./docs/01_benchmark_installation.md) to install the benchmark and setup the environment.
<br>
## 2. Download datasets
[Proceed as follows](./docs/02_download_datasets.md) to download the benchmark datasets.
<br>
## 3. Reproducibility
[Use this page](./docs/03_run_codes.md) to run the codes and reproduce the published results.
<br>
## 4. Adding a new dataset
[Instructions](./docs/04_add_dataset.md) to add a dataset to the benchmark.
<br>
## 5. Adding a Message-passing GCN
[Step-by-step directions](./docs/05_add_mpgcn.md) to add a MP-GCN to the benchmark.
<br>
## 6. Adding a Weisfeiler-Lehman GNN
[Step-by-step directions](./docs/06_add_wlgnn.md) to add a WL-GNN to the benchmark.
<br>
## 7. Leaderboards
[Leaderboards](./docs/07_leaderboards.md) of GNN models on each dataset. [Instructions](./docs/07_contribute_leaderboards.md) to contribute to leaderboards.
<br>
## 8. Reference
[ArXiv's paper](https://arxiv.org/pdf/2003.00982.pdf)
```
@article{dwivedi2020benchmarkgnns,
title={Benchmarking Graph Neural Networks},
author={Dwivedi, Vijay Prakash and Joshi, Chaitanya K and Laurent, Thomas and Bengio, Yoshua and Bresson, Xavier},
journal={arXiv preprint arXiv:2003.00982},
year={2020}
}
```
<br><br><br>
{
"gpu": {
"use": true,
"id": 1
},
"model": "GAT",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GAT/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 12
},
"net_params": {
"L": 3,
"hidden_dim": 19,
"out_dim": 76,
"residual": true,
"readout": "mean",
"n_heads": 4,
"in_feat_dropout": 0.0,
"dropout": 0.0,
"batch_norm": true,
"self_loop": false,
"layer_type": "dgl"
}
}
\ No newline at end of file
{
"gpu": {
"use": true,
"id": 1
},
"model": "GAT",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GAT/edgefeat/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 24
},
"net_params": {
"L": 3,
"hidden_dim": 20,
"out_dim": 60,
"residual": true,
"readout": "mean",
"n_heads": 3,
"in_feat_dropout": 0.0,
"dropout": 0.0,
"batch_norm": true,
"self_loop": false,
"edge_feat": false,
"layer_type": "edgefeat"
}
}
\ No newline at end of file
{
"gpu": {
"use": true,
"id": 1
},
"model": "GAT",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GAT/edgereprfeat/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 24
},
"net_params": {
"L": 3,
"hidden_dim": 13,
"out_dim": 39,
"residual": true,
"readout": "mean",
"n_heads": 3,
"in_feat_dropout": 0.0,
"dropout": 0.0,
"batch_norm": true,
"self_loop": false,
"edge_feat": false,
"layer_type": "edgereprfeat"
}
}
\ No newline at end of file
{
"gpu": {
"use": true,
"id": 1
},
"model": "GAT",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GAT/isotropic/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 24
},
"net_params": {
"L": 3,
"hidden_dim": 20,
"out_dim": 60,
"residual": true,
"readout": "mean",
"n_heads": 3,
"in_feat_dropout": 0.0,
"dropout": 0.0,
"batch_norm": true,
"self_loop": false,
"edge_feat": false,
"layer_type": "isotropic"
}
}
\ No newline at end of file
{
"gpu": {
"use": true,
"id": 1
},
"model": "GCN",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GCN/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 12
},
"net_params": {
"L": 3,
"hidden_dim": 74,
"out_dim": 74,
"residual": true,
"readout": "mean",
"in_feat_dropout": 0.0,
"dropout": 0.0,
"batch_norm": true,
"self_loop": false
}
}
\ No newline at end of file
{
"gpu": {
"use": true,
"id": 1
},
"model": "GIN",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GIN/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 12
},
"net_params": {
"L": 3,
"hidden_dim": 60,
"residual": true,
"readout": "sum",
"n_mlp_GIN": 2,
"learn_eps_GIN": true,
"neighbor_aggr_GIN": "sum",
"in_feat_dropout": 0.0,
"dropout": 0.0,
"batch_norm": true
}
}
\ No newline at end of file
{
"gpu": {
"use": true,
"id": 3
},
"model": "GatedGCN",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GatedGCN/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 12
},
"net_params": {
"L": 3,
"hidden_dim": 44,
"out_dim": 44,
"residual": true,
"edge_feat": false,
"readout": "mean",
"in_feat_dropout": 0.0,
"dropout": 0.0,
"batch_norm": true,
"layer_type": "edgereprfeat",
"pos_enc": false
}
}
\ No newline at end of file
{
"gpu": {
"use": true,
"id": 3
},
"model": "GatedGCN",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GatedGCN/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 12
},
"net_params": {
"L": 3,
"hidden_dim": 44,
"out_dim": 44,
"residual": true,
"edge_feat": false,
"readout": "mean",
"in_feat_dropout": 0.0,
"dropout": 0.0,
"batch_norm": true,
"layer_type": "edgereprfeat",
"pos_enc": true,
"pos_enc_dim": 20
}
}
\ No newline at end of file
{
"gpu": {
"use": true,
"id": 3
},
"model": "GatedGCN",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GatedGCN/edgefeat/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 24
},
"net_params": {
"L": 3,
"hidden_dim": 38,
"out_dim": 38,
"residual": true,
"edge_feat": false,
"readout": "mean",
"in_feat_dropout": 0.0,
"dropout": 0.0,
"batch_norm": true,
"layer_type": "edgefeat",
"pos_enc": false
}
}
\ No newline at end of file
{
"gpu": {
"use": true,
"id": 3
},
"model": "GatedGCN",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GatedGCN/edgereprfeat/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 24
},
"net_params": {
"L": 3,
"hidden_dim": 35,
"out_dim": 35,
"residual": true,
"edge_feat": false,
"readout": "mean",
"in_feat_dropout": 0.0,
"dropout": 0.0,
"batch_norm": true,
"layer_type": "edgereprfeat",
"pos_enc": false
}
}
\ No newline at end of file
{
"gpu": {
"use": true,
"id": 3
},
"model": "GatedGCN",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GatedGCN/isotropic/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 24
},
"net_params": {
"L": 3,
"hidden_dim": 48,
"out_dim": 48,
"residual": true,
"edge_feat": false,
"readout": "mean",
"in_feat_dropout": 0.0,
"dropout": 0.0,
"batch_norm": true,
"layer_type": "isotropic",
"pos_enc": false
}
}
\ No newline at end of file
{
"gpu": {
"use": true,
"id": 2
},
"model": "GraphSage",
"dataset": "OGBL-COLLAB",
"out_dir": "out/COLLAB_edge_classification/GraphSage/",
"params": {
"seed": 41,
"epochs": 500,
"batch_size": 32768,
"init_lr": 0.001,
"lr_reduce_factor": 0.5,
"lr_schedule_patience": 10,
"min_lr": 1e-5,
"weight_decay": 0.0,
"print_epoch_interval": 1,
"max_time": 12
},