-
Notifications
You must be signed in to change notification settings - Fork 0
/
train_q1.py
171 lines (138 loc) · 6.53 KB
/
train_q1.py
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
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
import torch
import torch.nn as nn
import torchvision.datasets as datasets
import torchvision.transforms as transforms
from sklearn.metrics import accuracy_score
from torch.functional import F
import matplotlib.pyplot as plt
import pickle
def load_data(batch_size):
transform_train_1 = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010))
])
transform_train_2 = transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010))
])
transform_train_3 = transforms.Compose([
transforms.RandomRotation(10),
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010))
])
transform_train_4 = transforms.Compose([
transforms.RandomAffine(0, shear=10, scale=(0.8, 1.2)),
transforms.ColorJitter(brightness=0.2, contrast=0.2, saturation=0.2),
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010))
])
transform_test = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010))
])
train_dataset_1 = datasets.CIFAR10(root='./data', train=True, download=True, transform=transform_train_1)
train_dataset_2 = datasets.CIFAR10(root='./data', train=True, download=True, transform=transform_train_2)
train_dataset_3 = datasets.CIFAR10(root='./data', train=True, download=True, transform=transform_train_3)
train_dataset_4 = datasets.CIFAR10(root='./data', train=True, download=True, transform=transform_train_4)
train_dataset = torch.utils.data.ConcatDataset([train_dataset_1, train_dataset_2])
train_dataset = torch.utils.data.ConcatDataset([train_dataset, train_dataset_3])
train_dataset = torch.utils.data.ConcatDataset([train_dataset, train_dataset_4])
test_dataset = datasets.CIFAR10(root='./data', train=False, download=True, transform=transform_test)
train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=batch_size, shuffle=True, num_workers=8,
pin_memory=True)
test_loader = torch.utils.data.DataLoader(test_dataset, batch_size=batch_size, shuffle=False, num_workers=8,
pin_memory=True)
return train_loader, test_loader
class CIFAR10CNN(nn.Module):
def __init__(self):
super(CIFAR10CNN, self).__init__()
self.conv1 = nn.Conv2d(3, 16, 3, padding=1)
self.conv2 = nn.Conv2d(16, 32, 3, padding=1)
self.conv3 = nn.Conv2d(32, 64, 3, padding=1)
self.pool = nn.MaxPool2d(2, 2)
self.fc1 = nn.Linear(64 * 4 * 4, 25)
self.fc2 = nn.Linear(25, 10)
self.Dropout = nn.Dropout(0.25)
self.loss = nn.CrossEntropyLoss()
def forward(self, input, y):
x = self.pool(F.relu(self.conv1(input)))
x = self.pool(F.relu(self.conv2(x)))
x = self.pool(F.relu(self.conv3(x)))
x = self.Dropout(x)
x = x.view(-1, 64 * 4 * 4)
x = F.relu(self.fc1(x))
x = self.fc2(x)
return x, self.loss(x, y)
def plot_epochs_losses_and_errors(loss_dict, error_dict):
plt.plot(loss_dict['test'], label = "test", color = "blue")
plt.plot(loss_dict['train'], label = "train", color = "green")
plt.title("Loss of train and test", fontweight = "bold", fontsize = 12)
plt.xlabel("epochs")
plt.ylabel("loss")
plt.legend()
plt.show()
plt.plot(error_dict['test'], label = "test", color = "blue")
plt.plot(error_dict['train'], label = "train", color = "green")
plt.plot([0.2]*len(loss_dict['train']), label = "error threshold", color ="red")
plt.title(f"Error of train and test", fontweight = "bold", fontsize = 12)
plt.xlabel("epochs")
plt.ylabel("error")
plt.legend()
plt.show()
def learn_and_predict(model, data_loaders, optimizer, num_epochs):
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model.to(device)
errors = {'train': [], 'test': []}
losses = {'train': [], 'test': []}
for epoch in range(num_epochs):
print(f'Epoch {epoch + 1}/{num_epochs}')
print('-' * 30)
for phase in ['train', 'test']:
if phase == 'train':
model.train()
else:
model.eval()
labels_list, preds_list = [], []
running_loss = 0.0
counter_batches = 0
for i, (data_images, data_labels) in enumerate(data_loaders[phase]):
counter_batches += 1
if torch.cuda.is_available():
images = data_images.to(device)
labels = data_labels.to(device)
else:
images = data_images
labels = data_labels
if phase == 'train':
optimizer.zero_grad()
outputs, loss = model(images, labels)
loss.backward()
optimizer.step()
with torch.no_grad():
outputs, loss = model(images, labels)
running_loss += loss.item()
_, predicted = torch.max(outputs.data, 1)
labels_list += labels.cpu().view(-1).tolist()
preds_list += predicted.view(-1).tolist()
epoch_acc = accuracy_score(labels_list, preds_list)
errors[phase].append(1 - epoch_acc)
losses[phase].append(running_loss / counter_batches)
print(f'{phase.title()} Accuracy: {epoch_acc}')
print()
with open("model_q1.pkl", "wb") as f:
pickle.dump(model, f)
return losses, errors
def train_model_q1():
num_epochs = 36
batch_size = 100
learning_rate = 0.001
train_loader, test_loader = load_data(batch_size=batch_size)
data_loaders = {'train': train_loader, 'test': test_loader}
model = CIFAR10CNN()
# print number of model parameters
print(f'Number of model parameters: {sum(p.numel() for p in model.parameters())}')
optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)
losses, errors = learn_and_predict(model, data_loaders, optimizer, num_epochs)
plot_epochs_losses_and_errors(losses, errors)