Update.

diff --git a/mygpt.py b/mygpt.py
index 954f4f0..7ff1035 100755 (executable)
--- a/mygpt.py
+++ b/mygpt.py
@@ -125,11 +125,10 @@ class MyGPT(nn.Module):
         self.readout = nn.Linear(in_features = dim_model, out_features = vocabulary_size)
 
     def forward(self, x):
-        x = F.pad(x, (1, 0))
+        x = F.pad(x, (1, -1))
         x = self.embedding(x)
         x = self.trunk(x)
         x = self.readout(x)
-        x = F.pad(x, (0, 0, 0, -1))
         return x
 
 ######################################################################