import torch
from torch.autograd import Variable
import numpy as np

import torch.nn as nn
import torch.nn.functional as F
from torch.utils.data import DataLoader

import torchvision
import torchvision.transforms as transforms


EPOCH= 1000


class my_network(nn.Module):
    def __init__(self):
        super(my_network, self).__init__()

        self.conv1= nn.Conv2d(3,64,5)
        self.conv2= nn.Conv2d(64,30,5)
        self.fc1= nn.Linear(30*5*5, 128)
        self.fc2= nn.Linear(128, 10)

    def forward(self,x):
        x= F.relu( self.conv1(x), inplace=True ) #28
        x= F.max_pool2d(x,(2,2)) #14
        x= F.relu(self.conv2(x), inplace=True) # 10
        x= F.max_pool2d(x,(2,2)) #5
        x= x.view(x.shape[0], -1)
        x=F.relu(self.fc1(x), inplace=True)
        x=F.relu(self.fc2(x), inplace=True)

        return x


if __name__ == "__main__":

    #data load
    transform= transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.5,0.5,0.5),(0.5,0.5,0.5))])

    trainset= torchvision.datasets.CIFAR10(root='./cifar10_data', train=True, download=True, transform=transform)
    testset= torchvision.datasets.CIFAR10(root='./cifar10_data', train=False, download=True, transform=transform)

    trainloader= DataLoader(trainset, batch_size=64, shuffle=True, num_workers=2)
    testloader= DataLoader(testset, batch_size=64, shuffle=True, num_workers=2)

    #train
    my_net= my_network()
    my_net.to('cuda')

    optim= torch.optim.SGD(my_net.parameters(), lr=0.0003, momentum=0.9)
    loss_function= nn.CrossEntropyLoss()

    for epoch in range(EPOCH):
        for i, data in enumerate(trainloader,8):
            inputs, labels= data
            inputs, labels= Variable(inputs.cuda()), Variable(labels.cuda())

            optim.zero_grad()
            out= my_net(inputs)
            loss= loss_function(out, labels)
            loss.backward()
            optim.step()

            if i%100==0:
                print("{}-{} loss:{:.5f}".format(epoch, i, loss))

        #test
        total=0
        correct=0
        for data2 in testloader:
            images, labels= data2
            images, labels= Variable(images.cuda()), Variable(labels.cuda())
            outputs= my_net(images)
            #print(outputs.data)
            _, predicted= torch.max(outputs, 1)
            #print(predicted.data)
            #print(labels.data)
            #exit()
            total+= labels.size(0)
            correct += (predicted == labels).sum()
        print("[{}] acc:{:.5f}".format(epoch, correct.type(torch.float32)/total))