"""
Shows a small example of how to use transformations (perhaps unecessarily many)
on CIFAR10 dataset and training on a small CNN toy network.

Video explanation: https://youtu.be/Zvd276j9sZ8
Got any questions leave a comment I'm pretty good at responding on youtube

Programmed by Aladdin Persson <aladdin.persson at hotmail dot com>
*    2020-04-09 Initial coding
"""

# Imports
import torch
import torch.nn as nn  # All neural network modules, nn.Linear, nn.Conv2d, BatchNorm, Loss functions
import torch.optim as optim  # For all Optimization algorithms, SGD, Adam, etc.
import torch.nn.functional as F  # All functions that don't have any parameters
from torch.utils.data import (
    DataLoader,
)  # Gives easier dataset managment and creates mini batches
import torchvision.datasets as datasets  # Has standard datasets we can import in a nice way
import torchvision.transforms as transforms  # Transformations we can perform on our dataset

# Simple CNN
class CNN(nn.Module):
    def __init__(self, in_channels, num_classes):
        super(CNN, self).__init__()
        self.conv1 = nn.Conv2d(
            in_channels=in_channels,
            out_channels=8,
            kernel_size=(3, 3),
            stride=(1, 1),
            padding=(1, 1),
        )
        self.pool = nn.MaxPool2d(kernel_size=(2, 2), stride=(2, 2))
        self.conv2 = nn.Conv2d(
            in_channels=8,
            out_channels=16,
            kernel_size=(3, 3),
            stride=(1, 1),
            padding=(1, 1),
        )
        self.fc1 = nn.Linear(16 * 8 * 8, num_classes)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        x = self.pool(x)
        x = F.relu(self.conv2(x))
        x = self.pool(x)
        x = x.reshape(x.shape[0], -1)
        x = self.fc1(x)

        return x


# Set device
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# Hyperparameters
learning_rate = 1e-4
batch_size = 64
num_epochs = 5


# Load pretrain model & modify it
model = CNN(in_channels=3, num_classes=10)
model.classifier = nn.Sequential(nn.Linear(512, 100), nn.ReLU(), nn.Linear(100, 10))
model.to(device)

# Load Data
my_transforms = transforms.Compose(
    [  # Compose makes it possible to have many transforms
        transforms.Resize((36, 36)),  # Resizes (32,32) to (36,36)
        transforms.RandomCrop((32, 32)),  # Takes a random (32,32) crop
        transforms.ColorJitter(brightness=0.5),  # Change brightness of image
        transforms.RandomRotation(
            degrees=45
        ),  # Perhaps a random rotation from -45 to 45 degrees
        transforms.RandomHorizontalFlip(
            p=0.5
        ),  # Flips the image horizontally with probability 0.5
        transforms.RandomVerticalFlip(
            p=0.05
        ),  # Flips image vertically with probability 0.05
        transforms.RandomGrayscale(p=0.2),  # Converts to grayscale with probability 0.2
        transforms.ToTensor(),  # Finally converts PIL image to tensor so we can train w. pytorch
        transforms.Normalize(
            mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5]
        ),  # Note: these values aren't optimal
    ]
)


train_dataset = datasets.CIFAR10(
    root="dataset/", train=True, transform=my_transforms, download=True
)
train_loader = DataLoader(dataset=train_dataset, batch_size=batch_size, shuffle=True)

# Loss and optimizer
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(model.parameters(), lr=learning_rate)

# Train Network
for epoch in range(num_epochs):
    losses = []

    for batch_idx, (data, targets) in enumerate(train_loader):
        # Get data to cuda if possible
        data = data.to(device=device)
        targets = targets.to(device=device)

        # forward
        scores = model(data)
        loss = criterion(scores, targets)

        losses.append(loss.item())
        # backward
        optimizer.zero_grad()
        loss.backward()

        # gradient descent or adam step
        optimizer.step()

    print(f"Cost at epoch {epoch} is {sum(losses)/len(losses):.5f}")

# Check accuracy on training & test to see how good our model


def check_accuracy(loader, model):
    if loader.dataset.train:
        print("Checking accuracy on training data")
    else:
        print("Checking accuracy on test data")

    num_correct = 0
    num_samples = 0
    model.eval()

    with torch.no_grad():
        for x, y in loader:
            x = x.to(device=device)
            y = y.to(device=device)

            scores = model(x)
            _, predictions = scores.max(1)
            num_correct += (predictions == y).sum()
            num_samples += predictions.size(0)

        print(
            f"Got {num_correct} / {num_samples} with accuracy {float(num_correct)/float(num_samples)*100:.2f}"
        )

    model.train()


check_accuracy(train_loader, model)