search

cure-lab / SCINet

The GitHub repository for the paper: “Time Series is a Special Sequence: Forecasting with Sample Convolution and Interaction“. (NeurIPS 2022)
Apache License 2.0
633 stars 129 forks source link

stacks=2, bug #67

Open baicuya opened 1 year ago

baicuya commented 1 year ago

Unable to use stacks2,

For example: input_len = 48, horizon = 6, batch_size = 32 res1 shape: torch.randn(32, 48, 2) x shape: torch.randn(32, 6, 2) x = torch.cat((res1, x), dim=1), x shape: torch.randn(32, 54, 2)

issue: x = self.blocks2(x) d = x_odd.mul(torch.exp(self.phi(x_even)))

This tensor cannot match: RuntimeError: The size of tensor a (7) must match the size of tensor b (8) at non-singleton dimension 2

VEWOXIC commented 1 year ago

Looks like after the spliting, the x_odd and x_even have different length. Please check if the input satisfies the special requirement mentioned in the README.

Hoju0216 commented 7 months ago

您好，请问您这个问题已经解决了吗？

baicuya commented 7 months ago

没解决> 您好，请问您这个问题已经解决了吗？