Moved the input/output shift in the forward of the model.

[mygpt.git] / mygpt.py

diff --git a/mygpt.py b/mygpt.py
index 37fe6af..7f0c9e6 100755 (executable)
--- a/mygpt.py
+++ b/mygpt.py
@@ -127,10 +127,11 @@ class MyGPT(nn.Module):
         self.readout = nn.Linear(in_features = dim_model, out_features = vocabulary_size)
 
     def forward(self, x):
+        x = torch.cat((x.new_zeros(x.size(0), 1), x), 1)
         x = self.embedding(x)
         x = self.trunk(x)
         x = self.readout(x)
-        return x
+        return x[:, :-1]
 
 ######################################################################