leibai / agcrn Goto Github PK

In AVWGCN.init there are those lines:

        self.weights_pool = nn.Parameter(torch.FloatTensor(embed_dim, cheb_k, dim_in, dim_out))
        self.bias_pool = nn.Parameter(torch.FloatTensor(embed_dim, dim_out))

Like this, the FloatTensor constructor does not initialize the values in the tensor, so it's just some garbage. It might lead to nan values and break the whole training. It should be initialized with some values, zeros / ones / rand... anything.

2020-07-28 11:43: Experiment log path in: C:\Users\cyx\Desktop\Adaptive Graph Convolutional Recurrent Network for Traffic Forecasting\ts12_AGCRN-master\AGCRN-master\model\experiments\PEMSD4\20200728114350
Traceback (most recent call last):
so where is the log file?

Hi,
how to know the connection relationship of nodes from this model? or we can't get it ?
I will appreciate your reply so! thanks a lot

Hello,

Congrats for your good work, I added it to my repo, here:

https://github.com/Alro10/deep-learning-time-series

Regards,

what does the cheb_k mean?Thank you!

Hi,

I found that in BasicTrainer.py, the scale of the model's outputs and labels isn't consistent when setting args.real_value=True. You may need to inverse transform the model's outputs to make it right.

We integrated the model in PyTorch Geometric Temporal.

https://github.com/benedekrozemberczki/pytorch_geometric_temporal

KeyError: 'data' in run.py

I read the code carefully and wanted to run it on other datasets. But I didn't find any code for adjacency matrix. Could you indicate where the adjacency matrix data is introduced in this code? Thanks!