-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathtrain.py
157 lines (130 loc) · 5.87 KB
/
train.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
from __future__ import print_function
from __future__ import division
import argparse
import torch
import torch.utils.data
import torch.nn as nn
import torch.optim as optim
from torch.autograd import Variable
import torchvision
from torchvision import models
from torchvision import datasets, transforms
from torchvision.datasets import ImageFolder
import matplotlib.pyplot as plt
import time
from tensorboardX import SummaryWriter
from glob import glob
from util import *
import numpy as np
from PIL import Image
import warnings
warnings.filterwarnings("ignore")
from vae import VAE, ShallowVAE
parser = argparse.ArgumentParser(description='PyTorch VAE')
parser.add_argument('--batch-size', type=int, default=128, metavar='N',
help='input batch size for training (default: 128)')
parser.add_argument('--epochs', type=int, default=20, metavar='N',
help='number of epochs to train (default: 20)')
parser.add_argument('--no-cuda', action='store_true', default=False,
help='enables CUDA training')
parser.add_argument('--seed', type=int, default=1, metavar='S',
help='random seed (default: 1)')
parser.add_argument('--log-interval', type=int, default=1, metavar='N',
help='how many batches to wait before logging training status')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
is_cuda = True
else:
is_cuda = False
BATCH_SIZE = args.batch_size
EPOCH = args.epochs
LOG_INTERVAL=args.log_interval
path = '../PetImages/'
kwargs = {'num_workers': 3, 'pin_memory': True} if is_cuda else {}
simple_transform = transforms.Compose([transforms.Resize((224,224))
,transforms.ToTensor(), transforms.Normalize([0.48829153, 0.45526633, 0.41688013],[0.25974154, 0.25308523, 0.25552085])])
train = ImageFolder(path+'train/',simple_transform)
valid = ImageFolder(path+'valid/',simple_transform)
train_data_gen = torch.utils.data.DataLoader(train,shuffle=True,batch_size=BATCH_SIZE,num_workers=kwargs['num_workers'])
valid_data_gen = torch.utils.data.DataLoader(valid,batch_size=BATCH_SIZE,num_workers=kwargs['num_workers'])
dataset_sizes = {'train':len(train_data_gen.dataset),'valid':len(valid_data_gen.dataset)}
dataloaders = {'train':train_data_gen,'valid':valid_data_gen}
model = ShallowVAE(latent_variable_size=500, nc=3, ngf=224, ndf=224, is_cuda=is_cuda)
#model = VAE(BasicBlock, [2, 2, 2, 2], latent_variable_size=500, nc=3, ngf=224, ndf=224, is_cuda=is_cuda)
if is_cuda:
model.cuda()
reconstruction_function = nn.MSELoss()
reconstruction_function.size_average = False
def loss_function(recon_x, x, mu, logvar):
MSE = reconstruction_function(recon_x, x)
# https://arxiv.org/abs/1312.6114 (Appendix B)
# 0.5 * sum(1 + log(sigma^2) - mu^2 - sigma^2)
KLD_element = mu.pow(2).add_(logvar.exp()).mul_(-1).add_(1).add_(logvar)
KLD = torch.sum(KLD_element).mul_(-0.5)
return MSE + KLD
optimizer = optim.Adam(model.parameters(), lr=1e-4)
def train(epoch):
model.train()
train_loss = 0
batch_idx = 1
for data in dataloaders['train']:
# get the inputs
inputs, _ = data
# wrap them in Variable
if torch.cuda.is_available():
inputs = Variable(inputs.cuda())
else:
inputs = Variable(inputs)
optimizer.zero_grad()
recon_batch, mu, logvar = model(inputs)
#print(inputs.data.size())
inputs.data = unnormalize(inputs.data,[0.48829153, 0.45526633, 0.41688013],[0.25974154, 0.25308523, 0.25552085])
#print("input max/min"+str(inputs.max())+" "+str(inputs.min()))
#print("recon input max/min"+str(recon_batch.max())+" "+str(recon_batch.min()))
loss = loss_function(recon_batch, inputs, mu, logvar)
loss.backward()
train_loss += loss.data[0]
optimizer.step()
if batch_idx % LOG_INTERVAL == 0:
print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
epoch, batch_idx * len(inputs), (len(dataloaders['train'])*128),
100. * batch_idx / len(dataloaders['train']),
loss.data[0] / len(inputs)))
batch_idx+=1
print('====> Epoch: {} Average loss: {:.4f}'.format(
epoch, train_loss / (len(dataloaders['train'])*BATCH_SIZE)))
return train_loss / (len(dataloaders['train'])*BATCH_SIZE)
def test(epoch):
model.eval()
test_loss = 0
for data in dataloaders['valid']:
# get the inputs
inputs, _ = data
# wrap them in Variable
if torch.cuda.is_available():
inputs = Variable(inputs.cuda())
else:
inputs = Variable(inputs)
recon_batch, mu, logvar = model(inputs)
inputs.data = unnormalize(inputs.data,[0.48829153, 0.45526633, 0.41688013],[0.25974154, 0.25308523, 0.25552085])
test_loss += loss_function(recon_batch, inputs, mu, logvar).data[0]
if((epoch+1)%10==0):
torchvision.utils.save_image(inputs.data, './imgs/Epoch_{}_data.jpg'.format(epoch), nrow=8, padding=2)
torchvision.utils.save_image(recon_batch.data, './imgs/Epoch_{}_recon.jpg'.format(epoch), nrow=8, padding=2)
test_loss /= (len(dataloaders['valid'])*128)
print('====> Test set loss: {:.4f}'.format(test_loss))
return test_loss
writer = SummaryWriter('runs/exp-1')
since = time.time()
for epoch in range(EPOCH):
train_loss = train(epoch)
test_loss = test(epoch)
writer.add_scalar('train_loss', train_loss, epoch)
writer.add_scalar('test_loss',test_loss, epoch)
torch.save(model.state_dict(), './models/Epoch_{}_Train_loss_{:.4f}_Test_loss_{:.4f}.pth'.format(epoch, train_loss, test_loss))
time_elapsed = time.time() - since
print('Training complete in {:.0f}m {:.0f}s'.format(
time_elapsed // 60, time_elapsed % 60))