MCPcopy Create free account
hub / github.com/IntelligentSystemsLab/ST-EVCDP / forward

Method forward

baselines.py:33–41  ·  view source on GitHub ↗
(self, occ, prc)

Source from the content-addressed store, hash-verified

31 self.decoder = nn.Linear(seq-n_fea+1, 1)
32
33 def forward(self, occ, prc): # occ.shape = [batch, node, seq]
34 x = torch.stack([occ, prc], dim=3)
35 x = self.encoder(x)
36 x = torch.transpose(x.squeeze(), 1, 2) # shape [batch, seq-n_fea+1, node]
37 x, _ = self.lstm(x)
38 x = torch.transpose(x, 1, 2) # shape [batch, node, seq-n_fea+1]
39 x = self.decoder(x)
40 x = torch.squeeze(x)
41 return x
42
43
44class GCN(nn.Module):

Callers

nothing calls this directly

Calls

no outgoing calls

Tested by

no test coverage detected