pku-dair / sgl Goto Github PK

Describe the bug

code:
from sgl.dataset import Planetoid
from sgl.models.homo import SGC
from sgl.tasks import NodeClassification

dataset = Planetoid("pubmed", "./", "official")
model = SGC(prop_steps=3, feat_dim=dataset.num_features, num_classes=dataset.num_classes)

device = "cuda:0"
test_acc = NodeClassification(dataset, model, lr=0.1, weight_decay=5e-5, epochs=200, device=device).test_acc

bug:
Traceback (most recent call last):
File "quick_start.py", line 9, in
test_acc = NodeClassification(dataset, model, lr=0.1, weight_decay=5e-5, epochs=200, device=device).test_acc
File "/home/ubuntu/miniconda3/lib/python3.8/site-packages/sgl/tasks/node_classification.py", line 39, in init
self.__test_acc = self._execute()
File "/home/ubuntu/miniconda3/lib/python3.8/site-packages/sgl/tasks/node_classification.py", line 49, in _execute
self.__model.preprocess(self.__dataset.adj, self.__dataset.x)
File "/home/ubuntu/miniconda3/lib/python3.8/site-packages/sgl/models/base_model.py", line 25, in preprocess
self._processed_feat_list = self._pre_graph_op.propagate(
File "/home/ubuntu/miniconda3/lib/python3.8/site-packages/sgl/operators/base_op.py", line 32, in propagate
feat_temp = csr_sparse_dense_matmul(self._adj, prop_feat_list[-1])
File "/home/ubuntu/miniconda3/lib/python3.8/site-packages/sgl/operators/utils.py", line 10, in csr_sparse_dense_matmul
ctl_lib = ctl.load_library("./csrc/libmatmul.so", "./operators/")
File "/home/ubuntu/miniconda3/lib/python3.8/site-packages/numpy/ctypeslib.py", line 163, in load_library
raise OSError("no file with expected extension")
OSError: no file with expected extension

Environment

• PyG version:2.0.4
• PyTorch versn:1.11.0
• OS:ubuntu18.04
• Python version:3.8.5
• CUDA/cuDNN version:CUDA 11.4
• How we installed PyTorch and PyG (conda, pip, source):conda
• Any other relevant information (e.g., version of torch-scatter):
• pytorch-scatter:2.0.9
• pytorch-sparse:0.6.13
• pytorch-spline-conv:1.2.1
• pytorch-cluster:1.6.0
• torch-geometric:2.0.4
• torchaudio: 0.11.0
• torchvision:0.12.0
• numpy:1.22.3
• scipy:1.8.0

Question
I don't know why such a mistake happened. Please help me.

Hi, Your paper is really interesting, and I try to retype this code and find that PASCA-V1(V2, V3) are not included in this respority. From this code, we know that SGAP is a fundamental framework of this paper, and SGC is constructed based on this framework, am I right?