-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathmain_baseline.py
197 lines (168 loc) · 8.71 KB
/
main_baseline.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
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
import os
import time
import torch
import argparse
from torch.utils.data import DataLoader
from tqdm import tqdm
import numpy as np
import json
from copy import deepcopy
from models.SASRec import SASRec
from utils.utils import evaluate
from data.MyDataset import MyDataset
def str2bool(s):
if s not in {'false', 'true'}:
raise ValueError('Not a valid boolean string')
return s == 'true'
parser = argparse.ArgumentParser()
parser.add_argument('--dataset', required=True)
parser.add_argument('--train_dir', required=True)
parser.add_argument('--model_name', default='SASRec', type=str)
parser.add_argument('--exp_name', default='base', type=str)
parser.add_argument('--batch_size', default=128, type=int)
parser.add_argument('--lr', default=0.001, type=float)
parser.add_argument('--maxlen', default=50, type=int)
parser.add_argument('--embed_dim', default=16, type=int)
parser.add_argument('--num_epochs', default=50, type=int)
parser.add_argument('--num_test_neg_item', default=100, type=int)
parser.add_argument('--dropout_rate', default=0.5, type=float)
parser.add_argument('--l2_emb', default=0.0, type=float)
parser.add_argument('--enable_feature_embedding_l2_norm', action='store_true', default=False)
parser.add_argument('--device', default='cpu', type=str)
parser.add_argument('--inference_only', default=False, type=str2bool)
parser.add_argument('--state_dict_path', default=None, type=str)
parser.add_argument('--pretrain_model_path', default=None, type=str)
parser.add_argument('--save_freq', default=10, type=int)
parser.add_argument('--val_freq', default=1, type=int)
args = parser.parse_args()
save_dir = os.path.join(args.dataset + '_' + args.train_dir, args.exp_name)
if not os.path.isdir(args.dataset + '_' + args.train_dir):
os.makedirs(args.dataset + '_' + args.train_dir)
if not os.path.isdir(save_dir):
os.makedirs(save_dir)
with open(os.path.join(save_dir, 'args.txt'), 'a') as f:
f.write(time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) + '\n')
f.write('\n'.join([str(k) + ',' + str(v) for k, v in sorted(vars(args).items(), key=lambda x: x[0])]))
f.close()
if __name__ == '__main__':
# dataset
dataset_train = MyDataset(data_dir='data/' + args.dataset,
max_length=args.maxlen, mode='train', device=args.device)
dataset_valid = MyDataset(data_dir='data/' + args.dataset,
max_length=args.maxlen, mode='val', neg_num=args.num_test_neg_item, device=args.device)
dataset_test = MyDataset(data_dir='data/' + args.dataset,
max_length=args.maxlen, mode='test', neg_num=args.num_test_neg_item, device=args.device)
usernum = dataset_train.user_num
itemnum = dataset_train.item_num
user_features_dim = dataset_train.user_features_dim
item_features_dim = dataset_train.item_features_dim
print('number of users: %d' % usernum, 'number of items: %d' % itemnum)
config = {'embed_dim': args.embed_dim,
'dim_config': {'item_id': itemnum+1, 'user_id': usernum+1,
'item_feature': item_features_dim, 'user_feature': user_features_dim},
'device': args.device,
'maxlen': args.maxlen}
dataset_meta_data = json.load(open(os.path.join('data', 'dataset_meta_data.json'), 'r'))
config['item_feature'] = dataset_meta_data[args.dataset]['item_feature']
config['user_feature'] = dataset_meta_data[args.dataset]['user_feature']
if args.model_name == "SASRec":
model = SASRec(config).to(args.device)
else:
raise ValueError("model name not supported")
f = open(os.path.join(save_dir, 'log.txt'), 'a')
f.write(time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) +' model: ' + args.model_name + '\n')
for name, param in model.named_parameters():
try:
torch.nn.init.xavier_normal_(param.data)
except:
pass # just ignore those failed init layers
model.train() # enable model training
epoch_start_idx = 1
if args.state_dict_path is not None:
try:
model.load_state_dict(torch.load(args.state_dict_path, map_location=torch.device(args.device)))
tail = args.state_dict_path[args.state_dict_path.find('epoch=') + 6:]
epoch_start_idx = int(tail[:tail.find('.')]) + 1
except: # in case your pytorch version is not 1.6 etc., pls debug by pdb if load weights failed
print('failed loading state_dicts, pls check file path: ', end="")
print(args.state_dict_path)
print('pdb enabled for your quick check, pls type exit() if you do not need it')
import pdb
pdb.set_trace()
if args.inference_only:
model.eval()
t_test = evaluate(model, dataset_test, args)
print('test (NDCG@10: %.4f, HR@10: %.4f)' % (t_test[0], t_test[1]))
bce_criterion = torch.nn.BCEWithLogitsLoss() # torch.nn.BCELoss()
adam_optimizer = torch.optim.Adam(model.parameters(), lr=args.lr, betas=(0.9, 0.98))
T = 0.0
t0 = time.time()
best_val_HR = 0.0
best_val_NDCG = 0.0
best_HR = 0.0
best_NDCG = 0.0
best_epoch = -1
best_state_dict = None
for epoch in range(epoch_start_idx, args.num_epochs + 1):
if args.inference_only: break # just to decrease identition
dataloader = DataLoader(dataset_train, batch_size=args.batch_size, shuffle=True)
step = 0
epoch_loss = 0.0
train_loop = tqdm(dataloader, desc="Training Progress")
for data in train_loop:
step += 1
user_id, history_items, history_items_len, target_item_id, \
user_features, item_features, label, cold_item = data
logits = model(user_id, target_item_id, history_items, history_items_len, user_features, item_features)
if args.model_name == "CB2CF":
logits, loss_mse = logits
adam_optimizer.zero_grad()
loss = bce_criterion(logits, label)
if 'item_embedding' in model.state_dict().keys():
for param in model.item_embedding.parameters():
loss += args.l2_emb * torch.norm(param)
if 'user_embedding' in model.state_dict().keys():
for param in model.user_embedding.parameters():
loss += args.l2_emb * torch.norm(param)
if args.enable_feature_embedding_l2_norm:
for name in model.state_dict().keys():
if 'item_fm_2nd_order_sparse_emb' in name or 'user_fm_2nd_order_sparse_emb' in name:
loss += args.l2_emb * torch.norm(model.state_dict()[name])
if args.model_name == "CB2CF":
loss += loss_mse * args.CB2CF_alpha
loss.backward()
adam_optimizer.step()
epoch_loss += loss.item()
train_loop.set_description("Epoch {}/{}".format(epoch, args.num_epochs))
train_loop.set_postfix(loss=epoch_loss/step)
print("Epoch: {}, loss: {}".format(epoch, epoch_loss / step))
if epoch % args.val_freq == 0:
model.eval()
t1 = time.time() - t0
T += t1
print('Evaluating', end='')
t_test = evaluate(model, dataset_test, args)
t_valid = evaluate(model, dataset_valid, args)
print('epoch:%d, time: %f(s), valid (NDCG@10: %.4f, HR@10: %.4f), test (NDCG@10: %.4f, HR@10: %.4f)'
% (epoch, T, t_valid[0], t_valid[1], t_test[0], t_test[1]))
if t_valid[1] > best_val_HR:
best_val_HR = t_valid[1]
best_HR = t_test[1]
best_NDCG = t_test[0]
best_epoch = epoch
best_state_dict = deepcopy(model.state_dict())
f.write(str(t_valid) + ' ' + str(t_test) + '\n')
f.flush()
t0 = time.time()
model.train()
if epoch % args.save_freq == 0 or epoch == args.num_epochs:
folder = save_dir
fname = 'epoch={}.lr={}.embed_dim={}.maxlen={}.l2_emb={}.pth'
fname = fname.format(epoch, args.lr, args.embed_dim,
args.maxlen, args.l2_emb)
torch.save(model.state_dict(), os.path.join(folder, fname))
f.write("best epoch: {}, best NDCG@10: {}, best HR@10: {}".format(best_epoch, best_NDCG, best_HR) + '\n')
f.close()
print("best epoch: {}, best NDCG@10: {}, best HR@10: {}".format(best_epoch, best_NDCG, best_HR))
torch.save(best_state_dict, os.path.join(save_dir, 'best.pth'))
print("Done")