Fixed python3.5-related bug in models.py

facebookresearch · Jul 6, 2017 · 260bfd4 · 260bfd4
1 parent 59d2ea9
commit 260bfd4
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 5 deletions.
diff --git a/encoder/models.py b/encoder/models.py
@@ -38,7 +38,7 @@ def forward(self, sent_tuple):
         sent_len, idx_sort = np.sort(sent_len)[::-1], np.argsort(-sent_len)
         idx_unsort = np.argsort(idx_sort)
 
-        idx_sort = torch.cuda.LongTensor(idx_sort) if self.use_cuda else torch.LongTensor(idx_sort)
+        idx_sort = torch.from_numpy(idx_sort).cuda() if self.use_cuda else torch.from_numpy(idx_sort)
         sent = sent.index_select(1, Variable(idx_sort))
 
         # Handling padding in Recurrent Networks
@@ -47,7 +47,7 @@ def forward(self, sent_tuple):
         sent_output = nn.utils.rnn.pad_packed_sequence(sent_output)[0]
 
         # Un-sort by length
-        idx_unsort = torch.cuda.LongTensor(idx_unsort) if self.use_cuda else torch.LongTensor(idx_unsort)
+        idx_unsort = torch.from_numpy(idx_unsort).cuda() if self.use_cuda else torch.from_numpy(idx_sort)
         sent_output = sent_output.index_select(1, Variable(idx_unsort))
 
         # Pooling

diff --git a/models.py b/models.py
@@ -40,8 +40,8 @@ def forward(self, sent_tuple):
         # Sort by length (keep idx)
         sent_len, idx_sort = np.sort(sent_len)[::-1], np.argsort(-sent_len)
         idx_unsort = np.argsort(idx_sort)
-        
-        idx_sort = torch.cuda.LongTensor(idx_sort) if self.use_cuda else torch.LongTensor(idx_sort)
+
+        idx_sort = torch.from_numpy(idx_sort).cuda() if self.use_cuda else torch.from_numpy(idx_sort)
         sent = sent.index_select(1, Variable(idx_sort))
 
         # Handling padding in Recurrent Networks
@@ -50,7 +50,7 @@ def forward(self, sent_tuple):
         sent_output = nn.utils.rnn.pad_packed_sequence(sent_output)[0]
 
         # Un-sort by length
-        idx_unsort = torch.cuda.LongTensor(idx_unsort) if self.use_cuda else torch.LongTensor(idx_unsort)
+        idx_unsort = torch.from_numpy(idx_unsort).cuda() if self.use_cuda else torch.from_numpy(idx_sort)
         sent_output = sent_output.index_select(1, Variable(idx_unsort))
 
         # Pooling