Merge branch 'master' into patch-1

Author: holly1238

beginner_source/blitz/autograd_tutorial.py

@@ -299,7 +299,7 @@
 # The only parameters that compute gradients are the weights and bias of ``model.fc``.
 
 # Optimize only the classifier
-optimizer = optim.SGD(model.fc.parameters(), lr=1e-2, momentum=0.9)
+optimizer = optim.SGD(model.parameters(), lr=1e-2, momentum=0.9)
 
 ##########################################################################
 # Notice although we register all the parameters in the optimizer,

beginner_source/blitz/cifar10_tutorial.py

@@ -70,14 +70,16 @@
     [transforms.ToTensor(),
      transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
 
+batch_size = 4
+
 trainset = torchvision.datasets.CIFAR10(root='./data', train=True,
                                         download=True, transform=transform)
-trainloader = torch.utils.data.DataLoader(trainset, batch_size=4,
+trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size,
                                           shuffle=True, num_workers=2)
 
 testset = torchvision.datasets.CIFAR10(root='./data', train=False,
                                        download=True, transform=transform)
-testloader = torch.utils.data.DataLoader(testset, batch_size=4,
+testloader = torch.utils.data.DataLoader(testset, batch_size=batch_size,
                                          shuffle=False, num_workers=2)
 
 classes = ('plane', 'car', 'bird', 'cat',
@@ -106,7 +108,7 @@ def imshow(img):
 # show images
 imshow(torchvision.utils.make_grid(images))
 # print labels
-print(' '.join('%5s' % classes[labels[j]] for j in range(4)))
+print(' '.join('%5s' % classes[labels[j]] for j in range(batch_size)))
 
 
 ########################################################################

intermediate_source/fx_conv_bn_fuser.py

@@ -51,7 +51,7 @@ def __init__(self):
             nn.BatchNorm2d(1),
             nn.Conv2d(1, 1, 1),
         )
-        self.wrapped = WrappedBatchnorm()
+        self.wrapped = WrappedBatchNorm()
 
     def forward(self, x):
         x = self.conv1(x)
@@ -259,4 +259,4 @@ def benchmark(model, iters=20):
 # feedback you have about using it. Please feel free to use the
 # PyTorch Forums (https://discuss.pytorch.org/) and the issue tracker
 # (https://github.com/pytorch/pytorch/issues) to provide any feedback
-# you might have.
+# you might have.

intermediate_source/torchvision_tutorial.rst

@@ -54,7 +54,7 @@ should return:
 
 If your model returns the above methods, they will make it work for both
 training and evaluation, and will use the evaluation scripts from
-``pycocotools``.
+``pycocotools`` which can be installed with ``pip install pycocotools``.
 
 .. note ::
   For Windows, please install ``pycocotools`` from `gautamchitnis <https://github.com/gautamchitnis/cocoapi>`__ with command 
@@ -317,8 +317,8 @@ Putting everything together
 In ``references/detection/``, we have a number of helper functions to
 simplify training and evaluating detection models. Here, we will use
 ``references/detection/engine.py``, ``references/detection/utils.py``
-and ``references/detection/transforms.py``. Just copy them to your
-folder and use them here.
+and ``references/detection/transforms.py``. Just copy everything under 
+``references/detection`` to your folder and use them here.
 
 Let’s write some helper functions for data augmentation /
 transformation:

prototype_source/fx_graph_mode_ptq_static.rst

@@ -311,6 +311,7 @@ The purpose for calibration is to run through some sample examples that is repre
 the statistics of the Tensors and we can later use this information to calculate quantization parameters. 
 
 .. code:: python
+
     def calibrate(model, data_loader):  
         model.eval()    
         with torch.no_grad():   
@@ -320,17 +321,19 @@ the statistics of the Tensors and we can later use this information to calculate
 
 7. Convert the Model to a Quantized Model 
 ----------------------------------------- 
-``convert_fx`` takes a calibrated model and produces a quantized model.   
+``convert_fx`` takes a calibrated model and produces a quantized model.
 
 .. code:: python
-    quantized_model = convert_fx(prepared_model)    
+
+    quantized_model = convert_fx(prepared_model)
     print(quantized_model)
-   
+
 8. Evaluation 
 ------------- 
 We can now print the size and accuracy of the quantized model.    
 
 .. code:: python
+
     print("Size of model before quantization")  
     print_size_of_model(float_model)    
     print("Size of model after quantization")   
@@ -372,6 +375,7 @@ we'll first call fuse explicitly to fuse the conv and bn in the model:
 Note that ``fuse_fx`` only works in eval mode.    
 
 .. code:: python
+
     fused = fuse_fx(float_model)    
 
     conv1_weight_after_fuse = fused.conv1[0].weight[0]  
@@ -383,6 +387,7 @@ Note that ``fuse_fx`` only works in eval mode.
 --------------------------------------------------------------------   
 
 .. code:: python
+
     scripted_float_model_file = "resnet18_scripted.pth" 
 
     print("Size of baseline model") 
@@ -397,6 +402,7 @@ quantized in eager mode. FX graph mode and eager mode produce very similar quant
 so the expectation is that the accuracy and speedup are similar as well.  
 
 .. code:: python
+
     print("Size of Fx graph mode quantized model")  
     print_size_of_model(quantized_model)    
     top1, top5 = evaluate(quantized_model, criterion, data_loader_test) 

recipes_source/recipes/loading_data_recipe.py

@@ -101,7 +101,7 @@
 
 # A data point in Yesno is a tuple (waveform, sample_rate, labels) where labels 
 # is a list of integers with 1 for yes and 0 for no.
-yesno_data_trainset = torchaudio.datasets.YESNO('./', download=True)
+yesno_data = torchaudio.datasets.YESNO('./', download=True)
 
 # Pick data point number 3 to see an example of the the yesno_data:
 n = 3

requirements.txt

@@ -2,6 +2,7 @@
 
 sphinx==1.8.2
 sphinx-gallery==0.3.1
+docutils==0.16
 sphinx-copybutton
 tqdm
 numpy