-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmain.py
140 lines (115 loc) · 5.38 KB
/
main.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
import os
import time
import argparse
from model import *
from utils import *
import pandas as pd
import pickle
parser = argparse.ArgumentParser()
parser.add_argument('--rl_type', type=str, default='SA2C', help='SA2C/SNQN')
parser.add_argument('--dataset', default = 'lastfm', type = str, help='Kaggle/RC15')
parser.add_argument('--batch_size', default=1024, type=int)
parser.add_argument('--epoch', default=1000, type=int)
parser.add_argument('--hidden_size', default=64, type=int)
parser.add_argument('--reward_c', default=1.0, type=float)
parser.add_argument('--reward_b', default=1.0, type=float)
parser.add_argument('--reward_n', default=0.0, type=float)
parser.add_argument('--discount', default=0.5, type=float)
parser.add_argument('--dropout_rate', default=0.1, type=float)
parser.add_argument('--lr', default=0.005, type=float)
parser.add_argument('--lr2', default=0.005, type=float)
parser.add_argument('--n_neg', type=int, default=10)
parser.add_argument('--weight_n', default=1.0, type=float)
parser.add_argument('--encoder', default='GRU', help='SASRec/NItNet/Caser')
parser.add_argument('--device', default='cuda', type=str, help='cpu/cuda')
parser.add_argument('--gpu_num', default=0, type=int)
parser.add_argument('--patience', default=10, type=int)
parser.add_argument('--model_save', type=bool, default=True, help='saving model')
parser.add_argument('--model_save_path', default=None)
parser.add_argument('--model_load_ep', default=None, type=int)
parser.add_argument('--use_feats',type=bool, default=False)
parser.add_argument('--use_bcq', type=bool, default=False)
parser.add_argument('--use_norm', type=bool, default=False)
parser.add_argument('--use_new_ns', type=bool, default=False)
parser.add_argument('--tau', default=0.3, type=float)
parser.add_argument('--load_model', default=False)
parser.add_argument('--load_model_name', default=None)
# for SA2C
parser.add_argument('--smooth', type=float, default=0.0)
parser.add_argument('--clip', type=float, default=0.0)
parser.add_argument('--weight', type=float, default=1.0)
opt = parser.parse_args()
print(opt)
if opt.device == 'cuda':
device = f'cuda:{str(opt.gpu_num)}'
else:
device = 'cpu'
# torch.cuda.set_device(opt.gpu_num)
if opt.model_save == True and opt.model_save_path == None:
model_save_path = f'save_models/{opt.dataset}'
os.makedirs(model_save_path, exist_ok=True)
elif opt.model_save == True and opt.model_save_path != None:
model_save_path = opt.model_save_path
os.makedirs(opt.model_save_path, exist_ok=True)
save_model_name = f'save_models/{opt.dataset}/'
option_name = ['feat', 'bcq', 'norm', 'ns']
for i, option in enumerate([opt.use_feats, opt.use_bcq, opt.use_norm, opt.use_new_ns]):
if option:
save_model_name += f'_{option_name[i]}'
save_model_name += f'_{str(opt.lr)}_{str(opt.lr2)}.pt'
def main():
data_info = pd.read_pickle(f'data/{opt.dataset}/data_statis.df')
state_size = data_info['state_size'].values[0]
item_num = data_info['item_num'].values[0]
item2attribute_file = f'data/{opt.dataset}/item2attributes.json'
item2attribute, attribute_num = get_item2attribute_json(item2attribute_file, item_num+1)
opt.item2attribute = item2attribute.to(device)
if opt.rl_type == 'SA2C':
from model_SA2C import SA2C, train_test, test
pop_dict = pickle.load(open(f'data/{opt.dataset}/pop_dict.pickle', 'rb'))
model = SA2C(opt, item_num+1, state_size, attribute_num+1, device).to(device)
if opt.load_model:
model.load_state_dict(torch.load(f"save_models/{opt.dataset}/{opt.load_model_name}", map_location=device))
else:
from model3 import SNQN, train_test, test
pop_dict = None
model = SNQN(opt, item_num+1, state_size, attribute_num+1, device).to(device)
dataset = Data(opt, item_num, state_size, device, pop_dict)
replay_buffer = pd.read_pickle(f'data/{opt.dataset}/replay_buffer.df')
valid_data = pd.read_pickle(f'data/{opt.dataset}/sampled_val.df')
valid_data = dataset.eval_data_load(valid_data, opt.batch_size)
# test data 추가
test_data = pd.read_pickle(f'data/{opt.dataset}/sampled_test.df')
test_data = dataset.eval_data_load(test_data, opt.batch_size)
# early_stopping = EarlyStopping(model_save_path, patience=opt.patience)
start = time.time()
best_results = [[0 for i in range(4)] for j in range(2)]
best_epochs = [[0 for i in range(4)] for j in range(2)]
bad_counter = 0
a2c = False
for i in range(opt.epoch):
print('-'*100)
print('Epoch: ', i)
if opt.rl_type == "SA2C":
if i < 2:
a2c = 0
elif i == 2:
a2c = 2
else:
a2c = 1
loss, eval10, eval20 = train_test(model, dataset, replay_buffer, valid_data, opt.batch_size, i, a2c)
# early_stopping(np.array(eval10[:2] + eval20[:2]), model, i)
# if early_stopping.early_stop:
# print("Early stopping")
# break
flag = get_best_result([eval10, eval20], i, best_results, best_epochs)
if flag:
torch.save(model.state_dict(), save_model_name)
print('-' * 100)
end = time.time()
print("Run time: %f s" % (end - start))
model.load_state_dict(torch.load(model_save_path))
# print("Loading best model at epoch", early_stopping.best_epoch)
test(model, dataset, test_data)
if __name__ == '__main__':
main()