Hello everyone. I have a question. I’ll be grateful for any help
I have a network:
class Net(nn.Module): def __init__(self): super(Net,self).__init__(); self.Charts = nn.ModuleList([net() for i in range(10)]) #net is a module I defined before def forward(self,x): x = torch.cat([self.Charts[i](x) for i in range(10)],1) return x
But I want to evaluate all the modules at the same time instead of going trough the loop:
[self.Charts[i](x) for i in range(10)]. Otherwise running in the GPU doesn’t improve performance
Anyone know how to do that?
(Also I know 10 submodules is not enough to see an improvement by running it in GPU but this is only an example, I have way more submodules in my actuala code).