PyGOD 使用 成功解决 Runtimeerror: No such operator torch_sparse::hetero_neighbor_sample 报错

Posted 2022-07-13 叶庭云

问题描述：

按照教程一步步装好了 PyGOD，用到的命令如下所示（我的环境：PyTorch1.8.1 + CUDA11.1）：

pip list
nvidia-smi
nvcc --version


pip install networkx
pip install scikit-learn==0.23.1


pip install torch-scatter==2.0.8 -f https://pytorch-geometric.com/whl/torch-1.8.1+cu111.html
pip install torch-cluster==1.5.9 -f https://pytorch-geometric.com/whl/torch-1.8.1+cu111.html
pip install torch-sparse==0.6.10 -f https://pytorch-geometric.com/whl/torch-1.8.1+cu111.html
pip install torch-spline-conv==1.2.1 -f https://pytorch-geometric.com/whl/torch-1.8.1+cu111.html
pip install torch-geometric -f https://pytorch-geometric.com/whl/torch-1.8.1+cu111.html

pip install --upgrade pygod -i http://pypi.douban.com/simple --trusted-host pypi.douban.com

然而，运行代码测试图异常检测时，报错如下：

from pygod.utils import load_data

data = load_data('inj_cora')
data.y = data.y.bool()

from pygod.models import DOMINANT

model = DOMINANT()

model.fit(data)

......
Runtimeerror: No such operator torch_sparse::hetero_neighbor_sample
......

分析觉得是 torch_sparse 版本的问题，卸载后重新安装 0.6.12 版本解决了问题：

pip uninstall torch-sparse
pip install torch-sparse==0.6.12 -f https://pytorch-geometric.com/whl/torch-1.8.1+cu111.html

References：

• Github Issues Is PyTorch 1.7.1 and cuda 9.2 able to use neighbor_sample?#179
• PyGOD 0.3.0 document Installation
• PyGOD Github 页面
• pytorch-geometric 安装教程
• PyTorch与PyTorch Geometric的安装
• pytorch-geometric.com whl匹配版本