Pytorch 4 (#71)

* Update to pytorch4.1

* Refactor

* Fix dataloader

* Change to item

Author: dmitriy-serdyuk

README.md

@@ -8,7 +8,7 @@ This is an image captioning codebase in PyTorch. If you are familiar with neural
 
 ## Requirements
 Python 2.7 (because there is no [coco-caption](https://github.com/tylin/coco-caption) version for python 3)
-PyTorch 0.2 (along with torchvision)
+PyTorch 0.4.1 (along with torchvision)
 
 You need to download pretrained resnet model for both training and evaluation. The models can be downloaded from [here](https://drive.google.com/open?id=0B7fNdx_jAqhtbVYzOURMdDNHSGM), and should be placed in `data/imagenet_weights`.
 

dataloaderraw.py

@@ -108,9 +108,10 @@ def get_batch(self, split, batch_size=None):
                 img = np.concatenate((img, img, img), axis=2)
 
             img = img.astype('float32')/255.0
-            img = torch.from_numpy(img.transpose([2,0,1])).cuda()
-            img = Variable(preprocess(img), volatile=True)
-            tmp_fc, tmp_att = self.my_resnet(img)
+            img = torch.from_numpy(img.transpose([2, 0, 1])).cuda()
+            with torch.no_grad():
+                img = Variable(preprocess(img))
+                tmp_fc, tmp_att = self.my_resnet(img)
 
             fc_batch[i] = tmp_fc.data.cpu().float().numpy()
             att_batch[i] = tmp_att.data.cpu().float().numpy()
@@ -136,4 +137,3 @@ def get_vocab_size(self):
 
     def get_vocab(self):
         return self.ix_to_word
-        

train.py

@@ -114,7 +114,7 @@ def train(opt):
         loss.backward()
         utils.clip_gradient(optimizer, opt.grad_clip)
         optimizer.step()
-        train_loss = loss.data[0]
+        train_loss = loss.item()
         torch.cuda.synchronize()
         end = time.time()
         print("iter {} (epoch {}), train_loss = {:.3f}, time/batch = {:.3f}" \