-
Notifications
You must be signed in to change notification settings - Fork 0
/
train.py
142 lines (138 loc) · 6.33 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
import cv2
import os
import numpy as np
import random
from matplotlib import pyplot as plt
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.utils.data as data
import torch.optim as optim
from model import DAE,BVAE
import utils
def train_dae(DAEnet,optim_dae,train_data_generator,test_data_generator,criterion,check_point_dir,epoch_num,writer,output_file_path):
for epoch in range(epoch_num):
running_loss=0.0
for i,batch in enumerate(train_data_generator,0):
a,b=batch
a=a.type(torch.cuda.FloatTensor).cuda()
b=b.type(torch.cuda.FloatTensor).cuda()
decoded=DAEnet(b)
loss=criterion(decoded,a)
#loss=-(a * torch.log(decoded) + (1 - a) * torch.log(1 - decoded)).sum() / 32
writer.add_scalar("Loss/dae_train", loss, epoch)
optim_dae.zero_grad()
loss.backward()
optim_dae.step()
if(i%20==19):
with open(output_file_path,"a") as ofile:
ofile.write('[%d, %5d] loss: %.3f \n' %
(epoch + 1, i + 1, running_loss /20))
ofile.close()
running_loss = 0.0
if(epoch%100==99):
#print("saving!")
state = {
'checkpoint_num': epoch,
'state_dict': DAEnet.state_dict(),
'optimizer': optim_dae.state_dict(),
}
path=str(epoch+1)+".pt"
saves=os.path.join(check_point_dir,path)
torch.save(state,saves)
for i,batch in enumerate(test_data_generator,0):
a,b=batch
a=a.type(torch.cuda.FloatTensor).cuda()
b=b.type(torch.cuda.FloatTensor).cuda()
decoded=DAEnet(b)
loss=criterion(decoded,a)
writer.add_scalar("Loss/dae_test", loss, epoch)
writer.flush()
def train_bvae(BVAE,optim,train_data_generator,test_data_generator,check_point_dir,epoch_num,writer,output_file_path):
for epoch in range(epoch_num):
#running_loss=0.0
for i,batch in enumerate(train_data_generator,0):
batch=batch.type(torch.cuda.FloatTensor).cuda()
reconstruction,mean,std=BVAE(batch)
loss=BVAE.compute_loss(batch,reconstruction,mean,std)
writer.add_scalar("Loss/train", loss, epoch)
#running_loss+=loss.item()
optim.zero_grad()
loss.backward()
optim.step()
"""if(i%20==19):
with open(output_file_path,"a") as ofile:
ofile.write('[%d, %5d] loss: %.3f \n' %
(epoch + 1, i + 1, running_loss /20))
ofile.close()
running_loss = 0.0"""
if(epoch%100==99):
#print("saving!")
state = {
'checkpoint_num': epoch,
'state_dict': BVAE.state_dict(),
'optimizer': optim.state_dict(),
}
path=str(epoch+1)+".pt"
saves=os.path.join(check_point_dir,path)
torch.save(state,saves)
for i,batch in enumerate(test_data_generator,0):
batch=batch.type(torch.cuda.FloatTensor).cuda()
reconstruction,mean,std=BVAE(batch)
loss=BVAE.compute_loss(batch,reconstruction,mean,std)
writer.add_scalar("Loss/test", loss, epoch)
writer.flush()
def train_scan(SCAN_net,optim_scan,train_data_generator,test_data_generator,check_point_dir,epoch_num,writer,output_file_path):
losses=[]
for epoch in range(epoch_num):
for batchid,batch in enumerate(train_data_generator,0):
one_hot,image=batch
one_hot=one_hot.type(torch.cuda.FloatTensor).cuda()
image=image.type(torch.cuda.FloatTensor).cuda()
recon,m,s=SCAN_net(one_hot)
loss,kld=SCAN_net.compute_loss(image,one_hot,recon,m,s)
losses.append(kld)
writer.add_scalar("Loss/train/scan",loss,epoch)
optim_scan.zero_grad()
loss.backward()
#optim_scan_zero_grad()
optim_scan.step()
if(epoch%50==49):
#print("saving!")
state = {
'checkpoint_num': epoch,
'state_dict': SCAN_net.state_dict(),
'optimizer': optim_scan.state_dict(),
}
path_now=str(epoch+1)+".pt"
saves=os.path.join(check_point_dir,path_now)
torch.save(state,saves)
for i,batch in enumerate(test_data_generator,0):
one_hot,image=batch
one_hot=one_hot.type(torch.cuda.FloatTensor).cuda()
image=image.type(torch.cuda.FloatTensor).cuda()
recon,m,s=SCAN_net(one_hot)
loss,kld=SCAN_net.compute_loss(image,one_hot,recon,m,s)
writer.add_scalar("Loss/test/scan", loss, epoch)
writer.flush()
return losses
def train_recomb(SCAN_Recomb,optim_recomb,train_data_generator,check_point_dir):
for batch_id,batch in enumerate(train_data_generator,0):
op_type,y0,y1,y_ground_truth,images=batch
#print(op_type)
#print(one_hot_to_label(y0[0]))
#print(one_hot_to_label(y1[0]))
#print(one_hot_to_label(y_ground_truth[0]))
#img.append(images)
op_type=op_type.type(torch.cuda.FloatTensor)
y0=y0.type(torch.cuda.FloatTensor).cuda()
y1=y1.type(torch.cuda.FloatTensor).cuda()
y_ground_truth=y_ground_truth.type(torch.cuda.FloatTensor).cuda()
images=images.type(torch.FloatTensor).cuda()
out0,out1=SCAN_Recomb(y0,y1,op_type.unsqueeze(2))
loss=recomb.compute_loss(images,out0,out1,y_ground_truth)
optim_recomb.zero_grad()
#losses.append(loss.item())
loss.backward()
optim_recomb.step()
utils.save_model(SCAN_Recomb,optim_recomb,check_point_dir,"SCAN_Recomb.pt")