Merge pull request #15 from elliottzheng/master

fix:In pytorch 0.4 accuracy always zero

Author: fxia22

train_classification.py

@@ -80,7 +80,7 @@
         optimizer.step()
         pred_choice = pred.data.max(1)[1]
         correct = pred_choice.eq(target.data).cpu().sum()
-        print('[%d: %d/%d] train loss: %f accuracy: %f' %(epoch, i, num_batch, loss.data[0], correct/float(opt.batchSize)))
+        print('[%d: %d/%d] train loss: %f accuracy: %f' %(epoch, i, num_batch, loss.item(),correct.item() / float(opt.batchSize)))
 
         if i % 10 == 0:
             j, data = next(enumerate(testdataloader, 0))
@@ -92,6 +92,6 @@
             loss = F.nll_loss(pred, target)
             pred_choice = pred.data.max(1)[1]
             correct = pred_choice.eq(target.data).cpu().sum()
-            print('[%d: %d/%d] %s loss: %f accuracy: %f' %(epoch, i, num_batch, blue('test'), loss.data[0], correct/float(opt.batchSize)))
+            print('[%d: %d/%d] %s loss: %f accuracy: %f' %(epoch, i, num_batch, blue('test'), loss.item(), correct.item()/float(opt.batchSize)))
 
     torch.save(classifier.state_dict(), '%s/cls_model_%d.pth' % (opt.outf, epoch))

train_segmentation.py

@@ -80,7 +80,7 @@
         optimizer.step()
         pred_choice = pred.data.max(1)[1]
         correct = pred_choice.eq(target.data).cpu().sum()
-        print('[%d: %d/%d] train loss: %f accuracy: %f' %(epoch, i, num_batch, loss.data[0], correct/float(opt.batchSize * 2500)))
+        print('[%d: %d/%d] train loss: %f accuracy: %f' %(epoch, i, num_batch, loss.item(), correct.item()/float(opt.batchSize * 2500)))
 
         if i % 10 == 0:
             j, data = next(enumerate(testdataloader, 0))
@@ -95,6 +95,6 @@
             loss = F.nll_loss(pred, target)
             pred_choice = pred.data.max(1)[1]
             correct = pred_choice.eq(target.data).cpu().sum()
-            print('[%d: %d/%d] %s loss: %f accuracy: %f' %(epoch, i, num_batch, blue('test'), loss.data[0], correct/float(opt.batchSize * 2500)))
+            print('[%d: %d/%d] %s loss: %f accuracy: %f' %(epoch, i, num_batch, blue('test'), loss.item(), correct.item()/float(opt.batchSize * 2500)))
 
     torch.save(classifier.state_dict(), '%s/seg_model_%d.pth' % (opt.outf, epoch))