PFGNNPlus/main.py at main · zfan20/PFGNNPlus · GitHub

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
# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
# SPDX-License-Identifier: Apache-2.0

#!/usr/bin/env python
import matplotlib.pyplot as plt
import numpy as np
import argparse
import importlib
import random
import os
import time
from FLAlgorithms.servers.serveravg import FedAvg
from FLAlgorithms.servers.serverpFedMe import pFedMe
from FLAlgorithms.servers.serverpFedMeStruct import pFedMeStruct
from FLAlgorithms.servers.indserver import IndServer
from FLAlgorithms.trainmodel.models import *
from utils.model_utils import get_mostfree_gpu
from utils.plot_utils import *
import torch
from utils.metrics import set_seed

def main(args):

    # Get device status: Check GPU or CPU
    #device = torch.device("cuda:{}".format(gpu) if torch.cuda.is_available() and gpu != -1 else "cpu")
    device = torch.device("cuda:"+str(get_mostfree_gpu())) if torch.cuda.is_available() else torch.device("cpu")
    #device = torch.device("cpu")

    set_seed(args.random_seed)

    for i in range(args.times):
        print("---------------Running time:------------",i)
        start_time = time.time()
        # Generate model

        if args.model == 'sgcn':
            model = SGCN(args.gnnlayers, 768, args.out_dim), args.model

        if args.model == 'fedvarsgcn':
            model = FedVariationalSGCNEncoder(args), args.model

        if args.model == 'featmlp':
            model = FeatMLP(768, args.out_dim), args.model

        if args.model in ['transformer', 'relbigram', 'popularity', 'intbigram', 'intpopularity']:
            #model define should be after the data load, will be done in user or server
            model = None, args.model

        # select algorithm
        if args.algorithm == "FedAvg":
            server = FedAvg(device, args.dataset, args.algorithm, model, args)

        if args.algorithm == "pFedMe":
            server = pFedMe(device, args.dataset, args.algorithm, model, args)

        if args.algorithm == "pFedMeStruct":
            server = pFedMeStruct(device, args.dataset, args.algorithm, model, args)

        if args.algorithm != 'pretrain_test_no_finetune':
            if args.frompretrain_fewsteps:
                server.kstepsTrain(args.Ksteps)
            else:
                server.train()
            print('After all training', '-'*20)
        server.test(verbose=True)
        if not (args.algorithm == "IndTrain" or args.algorithm == 'pretrain_test_no_finetune' or args.algorithm == 'pretrain_test_with_finetune'):
            server.test_personalized_model(verbose=True)
        if args.save_item_embs:
            server.save_embs()

        end_time = time.time()
        print("Running time ", i, " used ", (end_time - start_time), " seconds")


if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument("--dataset", type=str, default="Cifar10", help='Dataset')
    parser.add_argument("--category", type=str, default="Electronics", help='category when multi-markets')
    parser.add_argument("--model", type=str, default="sgcn", choices=["dnn", "mclr", "cnn", "sgcn", "featmlp", "transformer", "relbigram", "popularity", "intbigram", "intpopularity", "fedvarsgcn"])
    parser.add_argument("--batch_size", type=int, default=1024)
    parser.add_argument("--learning_rate", type=float, default=0.001, help="Local learning rate")
    parser.add_argument("--beta", type=float, default=1.0, help="Average moving parameter for pFedMe, or Second learning rate of Per-FedAvg")
    parser.add_argument("--lamda", type=float, default=15, help="Regularization term")
    parser.add_argument("--num_global_iters", type=int, default=100)
    parser.add_argument("--local_epochs", type=int, default=100)
    parser.add_argument("--optimizer", type=str, default="SGD", choices=["SGD", "Adam"])
    parser.add_argument("--algorithm", type=str, default="pFedMe",choices=["pFedMe", "PerAvg", "FedAvg", "IndTrain", "pretrain_test_no_finetune", "pretrain_test_with_finetune", "pFedMeStruct"])
    parser.add_argument("--numusers", type=int, default=20, help="Number of Users per round")
    parser.add_argument("--K", type=int, default=5, help="Computation steps")
    parser.add_argument("--personal_learning_rate", type=float, default=0.09, help="Persionalized learning rate to caculate theta aproximately using K steps")
    parser.add_argument("--times", type=int, default=5, help="running time")
    parser.add_argument("--gpu", type=int, default=0, help="Which GPU to run the experiments, -1 mean CPU, 0,1,2 for GPU")

    parser.add_argument("--gnnlayers", type=int, default=3,
                        help="Number of GNN Layers")
    parser.add_argument("--out_dim", type=int, default=128,
                        help="The output dimension of items")
    parser.add_argument("--edge_type", type=str, default="also_buy",
                        help="The type of item-item edge")
    parser.add_argument("--weight_decay", type=float, default=1e-2,
                        help="weight regularization")
    parser.add_argument("--market", type=str, default='us',
                        help="market to run independently")
    parser.add_argument("--save_item_embs", action='store_true')

    #Transformer
    parser.add_argument("--hidden_size", type=int, default=128, help="hidden size of transformer model")
    parser.add_argument("--num_hidden_layers", type=int, default=1, help="number of layers")
    parser.add_argument('--num_attention_heads', default=2, type=int)
    parser.add_argument('--hidden_act', default="gelu", type=str) # gelu relu
    parser.add_argument("--attention_probs_dropout_prob", type=float, default=0.0, help="attention dropout p")
    parser.add_argument("--hidden_dropout_prob", type=float, default=0.3, help="hidden dropout p")
    parser.add_argument('--max_seq_length', default=100, type=int)


    #pretrained
    parser.add_argument('--from_pretrain', action='store_true')
    parser.add_argument('--pretrain_dataset', default='Electronics', type=str)
    parser.add_argument('--pretrain_market', default='ae', type=str)
    parser.add_argument('--pretrain_algorithm', default='IndTrain', type=str)


    #few-steps training
    parser.add_argument('--frompretrain_fewsteps', action='store_true')
    parser.add_argument('--Ksteps', default=1, type=int)

    #mempool params
    parser.add_argument('--num_clusters', default=6, type=int)
    parser.add_argument('--num_clusters_level1', default=24, type=int)
    parser.add_argument('--pool_kl_loss_weight', default=0.0, type=float)
    parser.add_argument('--pooldropout', default=0.1, type=float)
    parser.add_argument('--heads', default=2, type=int)
    parser.add_argument('--tau', default=1.0, type=float)
    parser.add_argument('--pool_lp_loss_weight', default=1.0, type=float)
    parser.add_argument('--personal_learning_rate_struct', default=0.01, type=float)
    parser.add_argument('--lamda_struct', default=1.0, type=float)

    parser.add_argument('--random_seed', default=0, type=int)

    args = parser.parse_args()

    print("=" * 80)
    print("Summary of training process:")
    print("Algorithm: {}".format(args.algorithm))
    print("Batch size: {}".format(args.batch_size))
    print("Learing rate       : {}".format(args.learning_rate))
    print("Average Moving       : {}".format(args.beta))
    print("Subset of users      : {}".format(args.numusers))
    print("Number of global rounds       : {}".format(args.num_global_iters))
    print("Number of local rounds       : {}".format(args.local_epochs))
    print("Dataset       : {}".format(args.dataset))
    print("Local Model       : {}".format(args.model))
    print("Market       : {}".format(args.market))
    if args.model == 'sgcn':
        print("#of GNN Layers       : {}".format(args.gnnlayers))
        print("Edge Type is       : {}".format(args.edge_type))
        print("The output dimension       : {}".format(args.out_dim))
        print("The weight decay       : {}".format(args.weight_decay))
    print("=" * 80)

    main(args)