Coursera GAN第一周编程作业“Your First GAN“

import torch
from torch import nn
from tqdm.auto import tqdm
from torchvision import transforms
from torchvision.datasets import MNIST # Training dataset
from torchvision.utils import make_grid
from torch.utils.data import DataLoader
import matplotlib.pyplot as plt
torch.manual_seed(0) # Set for testing purposes, please do not change!

def show_tensor_images(image_tensor, num_images=25, size=(1, 28, 28)):
    '''
    Function for visualizing images: Given a tensor of images, number of images, and
    size per image, plots and prints the images in a uniform grid.
    '''
    image_unflat = image_tensor.detach().cpu().view(-1, *size)
    image_grid = make_grid(image_unflat[:num_images], nrow=5)
    plt.imshow(image_grid.permute(1, 2, 0).squeeze())
    plt.show()
    # UNQ_C1 (UNIQUE CELL IDENTIFIER, DO NOT EDIT)
# GRADED FUNCTION: get_generator_block
def get_generator_block(input_dim, output_dim):
    '''
    Function for returning a block of the generator's neural network
    given input and output dimensions.
    Parameters:
        input_dim: the dimension of the input vector, a scalar
        output_dim: the dimension of the output vector, a scalar
    Returns:
        a generator neural network layer, with a linear transformation 
          followed by a batch normalization and then a relu activation
    '''
    return nn.Sequential(
        # Hint: Replace all of the "None" with the appropriate dimensions.
        # The documentation may be useful if you're less familiar with PyTorch:
        # https://pytorch.org/docs/stable/nn.html.
        #### START CODE HERE ####
        nn.Linear(input_dim, output_dim),
        nn.BatchNorm1d(output_dim),
        nn.ReLU(inplace=True),
        #### END CODE HERE ####
    )
    # Verify the generator block function
def test_gen_block(in_features, out_features, num_test=1000):
    block = get_generator_block(in_features, out_features)

    # Check the three parts
    assert len(block) == 3
    assert type(block[0]) == nn.Linear
    assert type(block[1]) == nn.BatchNorm1d
    assert type(block[2]) == nn.ReLU
    
    # Check the output shape
    test_input = torch.randn(num_test, in_features)
    test_output = block(test_input)
    assert tuple(test_output.shape) == (num_test, out_features)
    assert test_output.std() > 0.55
    assert test_output.std() < 0.65

test_gen_block(25, 12)
test_gen_block(15, 28)
print("Success!")
# UNQ_C2 (UNIQUE CELL IDENTIFIER, DO NOT EDIT)
# GRADED FUNCTION: Generator
class Generator(nn.Module):
    '''
    Generator Class
    Values:
        z_dim: the dimension of the noise vector, a scalar
        im_dim: the dimension of the images, fitted for the dataset used, a scalar
          (MNIST images are 28 x 28 = 784 so that is your default)
        hidden_dim: the inner dimension, a scalar
    '''
    def __init__(self, z_dim=10, im_dim=784, hidden_dim=128):
        super(Generator, self).__init__()
        # Build the neural network
        self.gen = nn.Sequential(
            get_generator_block(z_dim, hidden_dim),
            get_generator_block(hidden_dim, hidden_dim * 2),
            get_generator_block(hidden_dim * 2, hidden_dim * 4),
            get_generator_block(hidden_dim * 4, hidden_dim * 8),
            # There is a dropdown with hints if you need them! 
            #### START CODE HERE ####
            nn.Linear(hidden_dim*8, im_dim),
            nn.Sigmoid(),
            #### END CODE HERE ####
        )
    def forward(self, noise):
        '''
        Function for completing a forward pass of the generator: Given a noise tensor, 
        returns generated images.
        Parameters:
            noise: a noise tensor with dimensions (n_samples, z_dim)
        '''
        return self.gen(noise)
    
    # Needed for grading
    def get_gen(self):
        '''
        Returns:
            the sequential model
        '''
        return self.gen
        # Verify the generator class
def test_generator(z_dim, im_dim, hidden_dim, num_test=10000):
    gen = Generator(z_dim, im_dim, hidden_dim).get_gen()
    
    # Check there are six modules in the sequential part
    assert len(gen) == 6
    test_input = torch.randn(num_test, z_dim)
    test_output = gen(test_input)

    # Check that the output shape is correct
    assert tuple(test_output.shape) == (num_test, im_dim)
    assert test_output.max() < 1, "Make sure to use a sigmoid"
    assert test_output.min() > 0