Eliminate .data access for parameters as much as possible (#767)

Author: vishwakftw

dcgan/main.py

@@ -113,10 +113,10 @@
 def weights_init(m):
     classname = m.__class__.__name__
     if classname.find('Conv') != -1:
-        m.weight.data.normal_(0.0, 0.02)
+        torch.nn.init.normal_(m.weight, 0.0, 0.02)
     elif classname.find('BatchNorm') != -1:
-        m.weight.data.normal_(1.0, 0.02)
-        m.bias.data.fill_(0)
+        torch.nn.init.normal_(m.weight, 1.0, 0.02)
+        torch.nn.init.zeros_(m.bias)
 
 
 class Generator(nn.Module):

distributed/rpc/pipeline/main.py

@@ -134,8 +134,8 @@ def __init__(self, device, *args, **kwargs):
             if isinstance(m, nn.Conv2d):
                 nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')
             elif isinstance(m, nn.BatchNorm2d):
-                nn.init.constant_(m.weight, 1)
-                nn.init.constant_(m.bias, 0)
+                nn.init.ones_(m.weight)
+                nn.init.zeros_(m.bias)
 
     def forward(self, x_rref):
         x = x_rref.to_here().to(self.device)

distributed/rpc/rnn/rnn.py

@@ -42,7 +42,7 @@ def __init__(self, ntoken, ninp, dropout):
         super(EmbeddingTable, self).__init__()
         self.drop = nn.Dropout(dropout)
         self.encoder = nn.Embedding(ntoken, ninp).cuda()
-        self.encoder.weight.data.uniform_(-0.1, 0.1)
+        nn.init.uniform_(self.encoder.weight, -0.1, 0.1)
 
     def forward(self, input):
         return self.drop(self.encoder(input.cuda())).cpu()
@@ -56,8 +56,8 @@ def __init__(self, ntoken, nhid, dropout):
         super(Decoder, self).__init__()
         self.drop = nn.Dropout(dropout)
         self.decoder = nn.Linear(nhid, ntoken)
-        self.decoder.bias.data.zero_()
-        self.decoder.weight.data.uniform_(-0.1, 0.1)
+        nn.init.zeros_(self.decoder.bias)
+        nn.init.uniform_(self.decoder.weight, -0.1, 0.1)
 
     def forward(self, output):
         return self.decoder(self.drop(output))

regression/main.py

@@ -57,7 +57,7 @@ def get_batch(batch_size=32):
 
     # Apply gradients
     for param in fc.parameters():
-        param.data.add_(-0.1 * param.grad.data)
+        param.add_(-0.1 * param.grad)
 
     # Stop criterion
     if loss < 1e-3:

word_language_model/main.py

@@ -178,7 +178,7 @@ def train():
         # `clip_grad_norm` helps prevent the exploding gradient problem in RNNs / LSTMs.
         torch.nn.utils.clip_grad_norm_(model.parameters(), args.clip)
         for p in model.parameters():
-            p.data.add_(-lr, p.grad.data)
+            p.add_(-lr, p.grad)
 
         total_loss += loss.item()
 

word_language_model/model.py

@@ -41,9 +41,9 @@ def __init__(self, rnn_type, ntoken, ninp, nhid, nlayers, dropout=0.5, tie_weigh
 
     def init_weights(self):
         initrange = 0.1
-        self.encoder.weight.data.uniform_(-initrange, initrange)
-        self.decoder.bias.data.zero_()
-        self.decoder.weight.data.uniform_(-initrange, initrange)
+        nn.init.uniform_(self.encoder.weight, -initrange, initrange)
+        nn.init.zeros_(self.decoder)
+        nn.init.uniform_(self.decoder.weight, -initrange, initrange)
 
     def forward(self, input, hidden):
         emb = self.drop(self.encoder(input))
@@ -132,9 +132,9 @@ def _generate_square_subsequent_mask(self, sz):
 
     def init_weights(self):
         initrange = 0.1
-        self.encoder.weight.data.uniform_(-initrange, initrange)
-        self.decoder.bias.data.zero_()
-        self.decoder.weight.data.uniform_(-initrange, initrange)
+        nn.init.uniform_(self.encoder.weight, -initrange, initrange)
+        nn.init.zeros_(self.decoder)
+        nn.init.uniform_(self.decoder.weight, -initrange, initrange)
 
     def forward(self, src, has_mask=True):
         if has_mask: