-
Notifications
You must be signed in to change notification settings - Fork 3
/
param_search.py
158 lines (148 loc) · 8.92 KB
/
param_search.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
import argparse
import torch
import numpy as np
import random
import pickle
import os
import optuna
from util.loader import DataLoader
from config.model_param import model_specific_param
from model_imports import *
def objective(trial):
args.lr = trial.suggest_categorical("lr", [1e-4, 5e-3, 1e-3, 5e-2, 1e-2])
args.reg = trial.suggest_categorical("reg", [1e-5, 5e-4, 1e-4, 5e-3, 1e-3])
if args.model == 'KNN':
args.knn_num = trial.suggest_int("knn_num", 3, 12)
elif args.model == 'SimGCL':
args.cl_rate = trial.suggest_categorical("cl_rate", [0.0, 0.2, 0.4, 0.5, 0.6, 0.8, 1.0])
args.tau = trial.suggest_categorical("tau", [0.2, 0.4, 0.6, 0.8, 1.0])
args.eps = trial.suggest_categorical("eps", [0.1, 0.2, 0.3, 0.4, 0.5])
elif args.model == 'XSimGCL':
args.cl_rate = trial.suggest_categorical("cl_rate", [0.0, 0.2, 0.4, 0.5, 0.6, 0.8, 1.0])
args.tau = trial.suggest_categorical("tau", [0.2, 0.4, 0.6, 0.8, 1.0])
args.eps = trial.suggest_categorical("eps", [0.1, 0.2, 0.3, 0.4, 0.5])
elif args.model == 'NCL':
args.alpha = trial.suggest_categorical("alpha", [0.5, 1.0, 1.5, 2.0])
args.ssl_reg = trial.suggest_loguniform("ssl_reg", 1e-8, 1e-4)
args.proto_reg = trial.suggest_loguniform("proto_reg", 1e-8, 1e-4)
args.tau = trial.suggest_categorical("tau", [0.01, 0.05, 0.1, 0.2, 0.5])
args.hyper_layers = trial.suggest_int("hyper_layers", 1, 3)
args.num_clusters = trial.categorical("num_clusters", [20, 50, 80, 100, 200, 500])
elif args.model == 'ALDI':
args.alpha = trial.suggest_categorical("alpha", [0.5, 1.0, 1.5, 2.0])
args.beta = trial.suggest_categorical("beta", [0.05, 0.1, 0.2, 0.5])
args.gamma = trial.suggest_categorical("gamma", [0.1, 0.2, 0.5, 1.0])
args.tws = trial.suggest_categorical("tws", [0, 1])
args.freq_coef_M = trial.suggest_categorical("freq_coef_M", [2.0, 4.0, 6.0, 8.0])
elif args.model == 'GAR':
args.alpha = trial.suggest_categorical("alpha", [0.05, 0.1, 0.2, 0.5])
args.beta = trial.suggest_categorical("beta", [0.05, 0.1, 0.2, 0.5])
elif args.model == 'CLCRec':
args.num_neg = trial.suggest_categorical("alpha", [32, 64, 128, 256])
args.temp_value = trial.suggest_categorical("temp_value", [0.1, 0.2, 0.5, 1.0, 2.0])
args.lr_lambda = trial.suggest_categorical("lr_lambda", [0.1, 0.2, 0.5, 1.0])
args.num_sample = trial.suggest_categorical("num_sample", [0.1, 0.2, 0.5, 1.0])
elif args.model == 'CCFCRec':
args.positive_number = trial.suggest_categorical("positive_number", [2, 5, 10])
args.negative_number = trial.suggest_categorical("negative_number", [10, 20, 40, 80])
args.self_neg_number = trial.suggest_categorical("self_neg_number", [10, 20, 40, 80])
args.tau = trial.suggest_categorical("tau", [0.1, 0.2, 0.5, 1.0])
args.lambda1 = trial.suggest_categorical("lambda1", [0.2, 0.4, 0.6, 0.8, 1.0])
elif args.model == 'DropoutNet':
args.n_dropout = trial.suggest_categorical("n_dropout", [0.2, 0.5, 0.8])
elif args.model == 'Heater':
args.n_expert = trial.suggest_categorical("n_expert", [3, 5, 7])
args.n_dropout = trial.suggest_categorical("n_dropout", [0.2, 0.5, 0.8])
args.alpha = trial.suggest_categorical("alpha", [0.2, 0.5, 0.8])
elif args.model == 'MetaEmbedding':
args.alpha = trial.suggest_categorical("alpha", [0.2, 0.5, 0.8])
elif args.model == 'GoRec':
args.pre_cluster_num = trial.suggest_categorical("pre_cluster_num", [20, 50, 80, 100, 200, 500])
args.uni_coeff = trial.suggest_categorical("uni_coeff", [1, 5, 10, 15])
args.kl_coeff = trial.suggest_categorical("kl_coeff", [10, 50, 100, 500, 1000])
args.dropout = trial.suggest_categorical("dropout", [0.0, 0.2, 0.5])
# model register
available_models = ['MF', 'LightGCN', 'SimGCL', 'XSimGCL', 'NCL', 'KNN', 'DUIF', 'DeepMusic', 'MTPR',
'VBPR', 'AMR', 'GAR', 'ALDI', 'CLCRec', 'LARA', 'CCFCRec', 'DropoutNet', 'Heater',
'MetaEmbedding', 'GoRec']
if args.model in available_models:
if args.model == 'MF' or args.model == 'LightGCN' or args.model == 'SimGCL' \
or args.model == 'XSimGCL' or args.model == 'NCL':
# recommender backbone training
model = eval(args.model)(args, training_data, warm_valid_data, cold_valid_data, all_valid_data,
warm_test_data, cold_test_data, all_test_data, user_num, item_num,
warm_user_idx, warm_item_idx, cold_user_idx, cold_item_idx, device)
else:
# cold-start model training
model = eval(args.model)(args, training_data, warm_valid_data, cold_valid_data, all_valid_data,
warm_test_data, cold_test_data, all_test_data, user_num, item_num,
warm_user_idx, warm_item_idx, cold_user_idx, cold_item_idx, device,
user_content=user_content, item_content=item_content)
print(f"Registered model: {args.model}.")
else:
raise ValueError(f"Invalid model name: {args.model}!")
# model running
model.run()
all_ndcg_results = model.overall_test_results[0][3]
return all_ndcg_results
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--dataset', default='citeulike')
parser.add_argument('--model', default='MF')
parser.add_argument('--epochs', type=int, default=500)
parser.add_argument('--layers', type=int, default=2)
parser.add_argument('--topN', default='10,20')
parser.add_argument('--bs', type=int, default=2048, help='training batch size')
parser.add_argument('--emb_size', type=int, default=64)
parser.add_argument('--lr', type=float, default=0.001)
parser.add_argument('--reg', type=float, default=0.0001)
parser.add_argument('--runs', type=int, default=1, help='model runs')
parser.add_argument('--seed', type=int, default=2024)
parser.add_argument('--use_gpu', default=True, help='Whether to use CUDA')
parser.add_argument('--save_emb', default=True, help='Whether to save the user/item embeddings')
parser.add_argument('--gpu_id', type=int, default=0, help='CUDA id')
parser.add_argument('--cold_object', default='item', type=str, choices=['user', 'item'])
parser.add_argument('--backbone', default='MF')
args, _ = parser.parse_known_args()
parser = model_specific_param(args.model, parser)
args = parser.parse_args()
print(args)
device = torch.device("cuda:%d" % (args.gpu_id) if (torch.cuda.is_available() and args.use_gpu) else "cpu")
# data loader
training_data = DataLoader.load_data_set(f'./data/{args.dataset}/cold_{args.cold_object}/warm_train.csv')
# following the widely used setting in previous works, the 'all' set is used for validation.
all_valid_data = DataLoader.load_data_set(f'./data/{args.dataset}/cold_{args.cold_object}/overall_val.csv')
warm_valid_data = DataLoader.load_data_set(f'./data/{args.dataset}/cold_{args.cold_object}/warm_val.csv')
cold_valid_data = DataLoader.load_data_set(
f'./data/{args.dataset}/cold_{args.cold_object}/cold_{args.cold_object}_val.csv')
all_test_data = DataLoader.load_data_set(f'./data/{args.dataset}/cold_{args.cold_object}/overall_test.csv')
warm_test_data = DataLoader.load_data_set(f'./data/{args.dataset}/cold_{args.cold_object}/warm_test.csv')
cold_test_data = DataLoader.load_data_set(
f'./data/{args.dataset}/cold_{args.cold_object}/cold_{args.cold_object}_test.csv')
# dataset information
data_info_dict = pickle.load(open(f'./data/{args.dataset}/cold_{args.cold_object}/info_dict.pkl', 'rb'))
user_num = data_info_dict['user_num']
item_num = data_info_dict['item_num']
warm_user_idx = data_info_dict['warm_user']
warm_item_idx = data_info_dict['warm_item']
cold_user_idx = data_info_dict['cold_user']
cold_item_idx = data_info_dict['cold_item']
print(f"Dataset: {args.dataset}, User num: {user_num}, Item num: {item_num}.")
# content obtaining
user_content, item_content = None, None
if args.cold_object == 'user':
user_content = np.load(f'./data/{args.dataset}/{args.dataset}_{args.cold_object}_content.npy')
print(f'user content shape: {user_content.shape}')
if args.cold_object == 'item':
item_content = np.load(f'./data/{args.dataset}/{args.dataset}_{args.cold_object}_content.npy')
print(f'item content shape: {item_content.shape}')
study = optuna.create_study(
direction="maximize",
storage="sqlite:///optuna.db",
study_name=f"{args.model}_{args.dataset}_{args.cold_object}_cs",
load_if_exists=True,
)
study.optimize(objective, n_trials=100)
best_param = study.best_params
pickle.dump(best_param, open(f'./param/{args.model}_{args.dataset}_{args.cold_object}_cs.pkl', 'wb'))
print(best_param)