dingyuan bai

The training results on the self-built dataset are zero.

1

@Jingkang50 ，Hi, according to the work you have done, I have created a small dataset specifically for railway scenarios. It only includes two classes of objects: pedestrians and railways. It...

稀疏训练报错：ValueError: autodetected range of [nan, nan] is not finite

2

参数设置：sr：0.000001 问题：第二轮的时候所有的值都变成nan了。 猜想：之前train训练的时候迭代次数少，得到的权重文件是300多M，可以正常稀疏训练。新train迭代了300次，得到的权重跟官方预训练权重一致，都是170多M，稀疏训练猜想：之前train训练的时候迭代次数少，得到的权重文件是300多M，可以正常稀疏训练。新train迭代了300次，得到的权重跟官方预训练权重一致，都是170多M，稀疏训练nan报错。

### 🐛 Describe the bug ### Description I encountered an issue when using HeteroData objects with the PyTorch Geometric library. Specifically, while processing and saving HeteroData objects, keys such as...