You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+45-16Lines changed: 45 additions & 16 deletions
Original file line number
Diff line number
Diff line change
@@ -241,22 +241,51 @@ And we get:
241
241
242
242
## Intuitively Build Complex Architectures
243
243
244
-
Now we will visit a more complex example that combines several of the above operations. You will notice that as we add
245
-
more and more complexity to our network, the Torch code becomes more and more verbose.
246
-
On the other hand, thanks to autograd, the complexity of our PyTorch code does not increase at all.
244
+
Now we will visit a more complex example that combines several of the above operations.
245
+
The graph below is a random network that I created using the Torch [nngraph](https://github.com/torch/nngraph) package,
246
+
and the Torch code definition using nngraph can be found [here](https://github.com/amdegroot/pytorch-containers/blob/master/complex_graph.lua) and a raw Torch implementation can be found [here](https://github.com/amdegroot/pytorch-containers/blob/master/complex_net.lua) for comparison to the PyTorch code that follows.
0 commit comments