-
Notifications
You must be signed in to change notification settings - Fork 25
/
Trainer.py
147 lines (120 loc) · 6.54 KB
/
Trainer.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
from model_network import STTrajSimEncoder
import yaml
import torch
import data_utils
from lossfun import LossFun
import test_method
import time
import random
class STsim_Trainer(object):
def __init__(self):
config = yaml.safe_load(open('config.yaml'))
self.feature_size = config["feature_size"]
self.embedding_size = config["embedding_size"]
self.date2vec_size = config["date2vec_size"]
self.hidden_size = config["hidden_size"]
self.num_layers = config["num_layers"]
self.dropout_rate = config["dropout_rate"]
self.concat = config["concat"]
self.device = "cuda:" + str(config["cuda"])
self.learning_rate = config["learning_rate"]
self.epochs = config["epochs"]
self.train_batch = config["train_batch"]
self.test_batch = config["test_batch"]
self.traj_file = str(config["traj_file"])
self.time_file = str(config["time_file"])
self.dataset = str(config["dataset"])
self.distance_type = str(config["distance_type"])
self.early_stop = config["early_stop"]
def ST_eval(self, load_model=None):
net = STTrajSimEncoder(feature_size=self.feature_size,
embedding_size=self.embedding_size,
date2vec_size=self.date2vec_size,
hidden_size=self.hidden_size,
num_layers=self.num_layers,
dropout_rate=self.dropout_rate,
concat=self.concat,
device=self.device)
if load_model != None:
net.load_state_dict(torch.load(load_model))
net.to(self.device)
dataload = data_utils.DataLoader()
road_network = data_utils.load_netowrk(self.dataset).to(self.device)
with torch.no_grad():
vali_node_list, vali_time_list, vali_d2vec_list = dataload.load(load_part='test')
embedding_vali = test_method.compute_embedding(road_network=road_network, net=net,
test_traj=list(vali_node_list),
test_time=list(vali_d2vec_list),
test_batch=self.test_batch)
acc = test_method.test_model(embedding_vali, isvali=False)
print(acc)
def ST_train(self, load_model=None, load_optimizer=None):
net = STTrajSimEncoder(feature_size=self.feature_size,
embedding_size=self.embedding_size,
date2vec_size=self.date2vec_size,
hidden_size=self.hidden_size,
num_layers=self.num_layers,
dropout_rate=self.dropout_rate,
concat=self.concat,
device=self.device)
dataload = data_utils.DataLoader()
dataload.get_triplets()
data_utils.triplet_groud_truth()
optimizer = torch.optim.Adam([p for p in net.parameters() if p.requires_grad], lr=self.learning_rate,
weight_decay=0.0001)
lossfunction = LossFun(self.train_batch, self.distance_type)
net.to(self.device)
lossfunction.to(self.device)
road_network = data_utils.load_netowrk(self.dataset).to(self.device)
bt_num = int(dataload.return_triplets_num() / self.train_batch)
batch_l = data_utils.batch_list(batch_size=self.train_batch)
best_epoch = 0
best_hr10 = 0
lastepoch = '0'
if load_model != None:
net.load_state_dict(torch.load(load_model))
optimizer.load_state_dict(torch.load(load_optimizer))
lastepoch = load_model.split('/')[-1].split('_')[3]
best_epoch = int(lastepoch)
for epoch in range(int(lastepoch), self.epochs):
net.train()
s1 = time.time()
for bt in range(bt_num):
a_node_batch, a_time_batch, p_node_batch, p_time_batch, n_node_batch, n_time_batch, batch_index = batch_l.getbatch_one()
a_embedding = net(road_network, a_node_batch, a_time_batch)
p_embedding = net(road_network, p_node_batch, p_time_batch)
n_embedding = net(road_network, n_node_batch, n_time_batch)
loss = lossfunction(a_embedding, p_embedding, n_embedding, batch_index)
optimizer.zero_grad()
loss.backward()
optimizer.step()
s5 = time.time()
print("train time: ", s5-s1)
if epoch%2 == 0:
net.eval()
with torch.no_grad():
s6 = time.time()
vali_node_list, vali_time_list, vali_d2vec_list = dataload.load(load_part='vali')
embedding_vali = test_method.compute_embedding(road_network=road_network, net=net,
test_traj=list(vali_node_list),
test_time=list(vali_d2vec_list),
test_batch=self.test_batch)
acc = test_method.test_model(embedding_vali, isvali=True)
s7 = time.time()
print("test time: ", s7-s6)
print('epoch:', epoch, acc[0], acc[1], acc[2], loss.item())
# save model
save_modelname = './model/{}_{}_2w_ST/{}_{}_epoch_{}_HR10_{}_HR50_{}_HR1050_{}_Loss_{}.pkl'.format(self.dataset, self.distance_type,
self.dataset, self.distance_type, str(epoch), acc[0], acc[1], acc[2], loss.item())
torch.save(net.state_dict(), save_modelname)
if acc[0] > best_hr10:
best_hr10 = acc[0]
best_epoch = epoch
if epoch - best_epoch >= self.early_stop:
break
'''
save_optname = './optimizer/{}/tdrive_TP_2w_ST/{}_{}_epoch_{}.pkl'.format(self.dataset, self.dataset,
self.distance_type,
str(epoch))
torch.save(optimizer.state_dict(), save_optname)
'''