1
2
3
4
5
6
7
8
9
10
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
import torchvision.datasets as datasets
import torchvision.transforms as transforms
import matplotlib.pyplot as plt
import random
from torch.utils.data import DataLoader
1
device = 'cuda' if torch.cuda.is_available() else 'cpu'
1
2
3
4
5
# seed 고정
random.seed(1)
torch.manual_seed(1)
if device == 'cuda':
torch.cuda.manual_seed_all(1)
60,000개의 이미지 데이터가 있는데 batch_size = 100으로 설정하였다. 따라서 600개의 batch가 생긴다. -> total_batch = 600
1
2
3
# parameters
train_epochs = 15
batch_size = 100
1
2
3
4
5
# MNIST Dataset
mnist_train = datasets.MNIST(root='./', train=True,
transform=transforms.ToTensor(), download=True)
mnist_test = datasets.MNIST(root='./', train=False,
transform=transforms.ToTensor(), download=True)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz
Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz to ./MNIST\raw\train-images-idx3-ubyte.gz
100%|██████████| 9912422/9912422 [00:01<00:00, 9066483.55it/s]
Extracting ./MNIST\raw\train-images-idx3-ubyte.gz to ./MNIST\raw
0%| | 0/28881 [00:00<?, ?it/s]
Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz
Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz to ./MNIST\raw\train-labels-idx1-ubyte.gz
100%|██████████| 28881/28881 [00:00<00:00, 14446713.63it/s]
0%| | 0/1648877 [00:00<?, ?it/s]
Extracting ./MNIST\raw\train-labels-idx1-ubyte.gz to ./MNIST\raw
Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz
Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz to ./MNIST\raw\t10k-images-idx3-ubyte.gz
100%|██████████| 1648877/1648877 [00:00<00:00, 4511992.19it/s]
Extracting ./MNIST\raw\t10k-images-idx3-ubyte.gz to ./MNIST\raw
Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz
Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz to ./MNIST\raw\t10k-labels-idx1-ubyte.gz
100%|██████████| 4542/4542 [00:00<00:00, 4548836.86it/s]
Extracting ./MNIST\raw\t10k-labels-idx1-ubyte.gz to ./MNIST\raw
1
2
3
# DataLoader
data_loader = torch.utils.data.DataLoader(dataset=mnist_train, batch_size=batch_size,
shuffle=True, drop_last=True)
- 28 * 28 이미지 이므로 28 x 28 = 784개의 입력
- 0 ~ 9까지의 출력 데이터이므로 10개의 출력
1
2
# MNIST data image of shape 28 * 28 = 784
linear = torch.nn.Linear(784, 10, bias=True).to(device)
torch.nn.CrossEntropyLoss()를 사용하면 내부적으로 Softmax가 계산된다.
1
2
criterion = torch.nn.CrossEntropyLoss().to(device)
optimizer = torch.optim.SGD(linear.parameters(), lr=0.1)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
from tqdm.notebook import tqdm
for epoch in tqdm(range(train_epochs)):
avg_cost = 0
total_batch = len(data_loader)
# X : Image / y : Label
for X, y in data_loader:
X = X.view(-1, 28 * 28).to(device)
y = y.to(device)
optimizer.zero_grad()
hypothesis = linear(X) # 분류 결과를 얻음
cost = criterion(hypothesis, y) # 분류 결과와 실제 정답을 비교하여 cost 계산
cost.backward() # cost를 이용해 gradient 계산
optimizer.step() # gradient를 이용해 업데이트
avg_cost += cost / total_batch
print('Epoch {:4d}/{}, Cost: {:.4f}'.format(epoch, train_epochs, avg_cost))
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
0%| | 0/15 [00:00<?, ?it/s]
Epoch 0/15, Cost: 0.3310
Epoch 1/15, Cost: 0.3160
Epoch 2/15, Cost: 0.3071
Epoch 3/15, Cost: 0.3001
Epoch 4/15, Cost: 0.2950
Epoch 5/15, Cost: 0.2907
Epoch 6/15, Cost: 0.2874
Epoch 7/15, Cost: 0.2844
Epoch 8/15, Cost: 0.2819
Epoch 9/15, Cost: 0.2795
Epoch 10/15, Cost: 0.2777
Epoch 11/15, Cost: 0.2759
Epoch 12/15, Cost: 0.2745
Epoch 13/15, Cost: 0.2729
Epoch 14/15, Cost: 0.2717
1
import matplotlib.pyplot as plt
with torch.no_grad() : Gradient Update를 하지 않는다.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
# 모델을 이용하여 test
with torch.no_grad():
X_test = mnist_test.test_data.view(-1, 28 * 28).float().to(device)
y_test = mnist_test.test_labels.to(device)
prediction = linear(X_test)
correct = torch.argmax(prediction, 1) == y_test # True / False
accuracy = correct.float().mean() # True / False -> 1.0 / 0.0 -> mean
print('Accuracy: ', accuracy.item())
# 하나의 예측
r = random.randint(0, len(mnist_test) -1)
X_single_data = mnist_test.test_data[r:r + 1].view(-1, 28 * 28).float().to(device)
y_single_data = mnist_test.test_labels[r:r + 1].to(device)
print('Label: ', y_single_data.item())
single_pred = linear(X_single_data)
print('Prediction: ', torch.argmax(single_pred, 1).item())
plt.imshow(mnist_test.test_data[r:r + 1].view(28, 28), cmap='gray', interpolation='nearest')
plt.show()
1
2
3
Accuracy: 0.8826000094413757
Label: 0
Prediction: 0
