Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/jw9730/hot
[NeurIPS'21] Higher-order Transformers for sets, graphs, and hypergraphs, in PyTorch
https://github.com/jw9730/hot
equivariance gnn graph hypergraph pytorch self-attention transformer
Last synced: 14 days ago
JSON representation
[NeurIPS'21] Higher-order Transformers for sets, graphs, and hypergraphs, in PyTorch
- Host: GitHub
- URL: https://github.com/jw9730/hot
- Owner: jw9730
- Created: 2021-10-27T10:24:27.000Z (about 3 years ago)
- Default Branch: main
- Last Pushed: 2022-11-28T21:30:23.000Z (almost 2 years ago)
- Last Synced: 2024-08-01T16:54:09.149Z (3 months ago)
- Topics: equivariance, gnn, graph, hypergraph, pytorch, self-attention, transformer
- Language: Python
- Homepage:
- Size: 1.95 MB
- Stars: 60
- Watchers: 1
- Forks: 7
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Higher-order Transformer (PyTorch)
[**Transformers Generalize DeepSets and Can be Extended to Graphs and Hypergraphs**](https://arxiv.org/abs/2110.14416) \
[Jinwoo Kim](https://bit.ly/jinwoo-kim), [Saeyoon Oh](https://github.com/saeyoon17), [Seunghoon Hong](https://maga33.github.io/) @ [KAIST](https://www.kaist.ac.kr/en/) \
NeurIPS 2021
[](https://paperswithcode.com/sota/graph-regression-on-pcqm4m-lsc?p=transformers-generalize-deepsets-and-can-be)
***We present a generalization of Transformers to sets, graphs, and hypergraphs, and reduce its computational cost to linear to input size.***
- Powerful operations, involving both local and global interactions over input elements
- Translation between different-order graphs (e.g., set-to-graph, graph-to-set, graph-to-vector)
- Theoretically and empirically stronger than MPNNs, even with the same linear complexity
- Works well on large datasets!
In this repository, we provide the PyTorch implementation of:
- Up-to-second-order Transformers: {set, graph} input × {set, graph, vector} output
- A space of their variants: {dense, sparse} data × {softmax, kernel} attention
- A special extension for higher-order hyperedge prediction: set input, *k*-hyperedges output
- Corresponding equivariant linear layers as baselines
In practice, we recommend to use the ***sparse kernel variant*** as it scales linearly to data size while still powerful.
## How do I use Higher-order Transformers for my project?
Go to your repository and execute the following.
```bash
# this is enough
pip install torch>=1.8.1
# add this repository as a submodule
git submodule add https://github.com/jw9730/hot hot
git submodule init
# run basic tests
cd hot
python3 run_tests.py
cd ..
```
For basic use cases, please see `run_perf_tests.py` and `run_tests.py`.
To update the code, execute the following:
```bash
git submodule update --remote --merge
```
## Setting up experiments
Using Docker
```bash
git clone https://github.com/jw9730/hot.git hot
cd hot
docker build --no-cache --tag hot:latest .
docker run -it --gpus all --ipc=host --name=hot -v /home:/home hot:latest bash
# upon completion, you should be at /hot inside the container
```
Using pip
```bash
git clone https://github.com/jw9730/hot.git hot
cd hot
pip3 install torch==1.8.1+cu111 torchvision==0.9.1+cu111 torchaudio==0.8.1 -f https://download.pytorch.org/whl/torch_stable.html
pip install -r requirements.txt
pip install torch-sparse==0.6.9 torch-scatter==2.0.6 -f https://data.pyg.org/whl/torch-1.8.0+cu111.html
pip install torch-geometric==1.6.3
```
## Running experiments
Runtime and memory analysis
```bash
python3 run_perf_tests.py
```
PCQM4M-LSC graph regression
```bash
cd regression/examples/pcqm4m-lsc
# Second-order Transformers (sparse kernel)
bash enc.sh
bash enc-small.sh
# Comparison with more baselines
bash enc-short.sh
bash enc-small-short.sh
# Second-order MLP
bash mlp-short.sh
# Vanilla Transformer + Laplacian PE
bash laplacian-short.sh
```
Set-to-graph prediction
```bash
cd set-to-graph/main
python download_jets_data.py
cd ../examples
# Delaunay (50)
cd delaunay-a
bash enc-kernel.sh
bash enc-softmax.sh
bash s2g.sh
# visualize
bash visualize-kernel.sh
bash visualize-softmax.sh
# Delaunay (20-80)
cd ../delaunay-b
bash enc-kernel.sh
bash enc-softmax.sh
bash s2g.sh
# visualize
bash visualize-kernel.sh
bash visualize-softmax.sh
# Jets
cd ../jets
bash enc-kernel.sh
bash enc-softmax.sh
# test
bash test-enc-kernel.sh
bash test-enc-softmax.sh
```
*k*-uniform hyperedge prediction
```bash
cd k-uniform-hyperedge/examples
# GPS
cd gps
bash enc.sh
bash s2g+.sh
bash hyper-sagnn-e.sh
bash hyper-sagnn-w.sh
# MovieLens
cd ../movielens
bash enc.sh
bash s2g+.sh
bash hyper-sagnn-e.sh
bash hyper-sagnn-w.sh
# Drug
cd ../drug
bash enc.sh
bash s2g+.sh
bash hyper-sagnn-e.sh
bash hyper-sagnn-w.sh
```
## References
This implementation uses code from the following repositories:
- [Performer](https://github.com/lucidrains/performer-pytorch.git) for FAVOR+ attention kernel
- [Graph Transformer](https://github.com/graphdeeplearning/graphtransformer.git) for Laplacian PE of baseline model
- [Graphormer](https://github.com/microsoft/Graphormer.git) for graph regression experiment
- [Set2Graph](https://github.com/hadarser/SetToGraphPaper.git) for set-to-graph prediction experiment
- [Hyper-SAGNN](https://github.com/ma-compbio/Hyper-SAGNN.git) for k-uniform hyperedge prediction experiment
## Citation
If you find our work useful, please consider citing it:
```bib
@article{kim2021transformers,
author = {Jinwoo Kim and Saeyoon Oh and Seunghoon Hong},
title = {Transformers Generalize DeepSets and Can be Extended to Graphs and Hypergraphs},
booktitle = {2021 Conference on Neural Information Processing Systems (NeurIPS)},
year = {2021}
}
```