-
Notifications
You must be signed in to change notification settings - Fork 12
/
train_torch_model.py
128 lines (110 loc) · 4.08 KB
/
train_torch_model.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
import torch
import torch.nn as nn
import torch.nn.functional as F
import argparse
import torchvision
import time
from cnn.transforms import PIL2numpy, Normalize, ToTensor
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
self.conv1 = nn.Conv2d(1, 2, 3, 1)
self.conv2 = nn.Conv2d(2, 5, 2, 2, padding=1)
self.maxpool = nn.MaxPool2d(2, 2, padding=1)
self.fc1 = nn.Linear(320, 1000)
self.fc2 = nn.Linear(1000, 10)
def forward(self, x):
x = self.conv1(x)
x = F.relu(x)
x = self.maxpool(x)
x = self.conv2(x)
x = torch.sigmoid(x)
x = torch.flatten(x, 1)
x = self.fc1(x)
x = torch.sigmoid(x)
x = self.fc2(x)
return x # log_softmax is in CrossEntropyLoss
def get_train_loader():
transforms = torchvision.transforms.Compose([
PIL2numpy(),
Normalize(),
ToTensor()
])
train_dataset = torchvision.datasets.MNIST(
root='/workdir/data',
train=True,
download=True,
transform=transforms
)
test_dataset = torchvision.datasets.MNIST(
root='/workdir/data',
train=False,
download=True,
transform=transforms
)
train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=1)
test_loader = torch.utils.data.DataLoader(test_dataset, batch_size=1)
return train_loader, test_loader
def train_loop(dataset, model, criterion, optimizer, print_log_freq):
loss_log = []
acc_log = []
start_time = time.time()
model.train()
for idx, (image, target) in enumerate(dataset):
image = image.unsqueeze(0) # Add channel to make input 4D
optimizer.zero_grad()
pred = model(image)
loss = criterion(pred, target)
loss.backward()
optimizer.step()
loss_log.append(loss.item())
acc_log.append(pred.argmax().item() == target.item())
if idx % print_log_freq == 0:
loss_avg = sum(loss_log[-print_log_freq:])/print_log_freq
acc_avg = sum(acc_log[-print_log_freq:])/print_log_freq
loop_time = time.time() - start_time
start_time = time.time()
print(f'Train step {idx}, Loss: {loss_avg:.5f}, '
f'Acc: {acc_avg:.4f}, time: {loop_time:.1f}')
def val_loop(dataset, model, criterion):
loss_log = []
acc_log = []
start_time = time.time()
model.eval()
for idx, (image, target) in enumerate(dataset):
image = image.unsqueeze(0) # Add channel to make input 4D
with torch.no_grad():
pred = model(image)
loss = criterion(pred, target)
loss_log.append(loss.item())
acc_log.append(pred.argmax().item() == target.item())
loss_avg = sum(loss_log)/len(loss_log)
acc_avg = sum(acc_log)/len(acc_log)
loop_time = time.time() - start_time
print(f'Val step, Loss: {loss_avg:.5f}, '
f'Acc: {acc_avg:.4f}, time: {loop_time:.1f}')
def main(args):
train_loader, test_loader = get_train_loader()
model = Net()
if args.load_path:
states = torch.load(args.load_path)
model.load_state_dict(states)
print('Torch model weights were loaded')
optimizer = torch.optim.SGD(model.parameters(), lr=args.lr)
criterion = nn.CrossEntropyLoss()
for epoch in range(args.num_epochs):
train_loop(train_loader, model, criterion,
optimizer, args.print_log_freq)
val_loop(test_loader, model, criterion)
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--print_log_freq', type=int, default=1000,
help='Frequency of printing of training logs')
parser.add_argument('--load_path', type=str, default='',
help='Path to model weights to start training with')
parser.add_argument('--num_epochs', type=int, default=10,
help='Total number of epochs')
parser.add_argument('--lr', type=float, default=0.01,
help='learning rate')
args = parser.parse_args()
main(args)