search

chwang0721 / MST-OATD

[KDD'2024] Multi-Scale Detection of Anomalous Spatio-Temporal Trajectories in Evolving Trajectory Datasets
MIT License
1 stars 3 forks source link

H = self.Norm_A_N(adj, d_norm)报错？ #2

Closed yantianzi closed 2 days ago

yantianzi commented 2 weeks ago

存储为npz文件的稀疏矩阵A和D张量的大小分别是(25718x25718)和(1x1)，所以当在train.py里执行到D.mm(A).mm(D)这儿的时候，报错RuntimeError: mat1 and mat2 shapes cannot be multiplied (1x1 and 25718x25718)。

chwang0721 commented 2 weeks ago

你好，我在运行代码时并未出现这个问题，A和D的形状都为25718x25718