-
Notifications
You must be signed in to change notification settings - Fork 2
/
main.py
181 lines (160 loc) · 6.27 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
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
import numpy as np
import argparse
import importlib
import random
import os
import tensorflow as tf
from fedsimul.utils.model_utils import read_data
import tempfile
# GLOBAL PARAMETERS
OPTIMIZERS = ['fedavg', 'fedprox', 'fedmom', 'fedmomprox']
DATASETS = ['mnist', 'nist', 'shakespeare']
MODEL_PARAMS = {
'mnist.mclr': (10,), # num_classes
'nist.mclr': (10,),
'shakespeare.stacked_lstm': (80, 80, 256),
}
def read_args():
''' Parse command line arguments or load defaults. '''
parser = argparse.ArgumentParser()
parser.add_argument('--optimizer',
help='name of optimizer;',
type=str,
choices=OPTIMIZERS,
default='fedavg')
parser.add_argument('--dataset',
help='name of dataset;',
type=str,
choices=DATASETS,
default='mnist')
parser.add_argument('--model',
help='name of model;',
type=str,
default='mclr')
parser.add_argument('--num_rounds',
help='number of rounds to simulate;',
type=int,
default=100)
parser.add_argument('--eval_every',
help='evaluate every ____ rounds;',
type=int,
default=1)
parser.add_argument('--batch_size',
help='batch size when clients train on data;',
type=int,
default=10)
parser.add_argument('--num_epochs',
help='number of epochs when clients train on data;',
type=int,
default=1)
parser.add_argument('--learning_rate',
help='learning rate for inner solver;',
type=float,
default=0.01)
parser.add_argument('--gamma',
help='constant for momentum',
type=float,
default=0.9)
parser.add_argument('--seed',
help='seed for randomness;',
type=int,
default=0)
parser.add_argument('--gpu_id',
help='gpu_id',
type=int,
default=0)
parser.add_argument('--verbose',
help='toggle the verbose output',
action='store_true')
# Async
parser.add_argument('--asyn', '-a',
help='toggle asynchronous simulation',
action='store_true')
parser.add_argument('--participate_rate', '-p',
help='probability of participating',
type=float,
default=0.02)
parser.add_argument('--refresh_rate', '-q',
help='probability of refreshing',
type=float,
default=1)
parser.add_argument('--adp_p',
help='toggle adaptive participate rate',
action='store_true')
parser.add_argument('--adp_q',
help='toggle adaptive refresh rate',
action='store_true')
parser.add_argument('--window_size', '-w',
help='moving window size',
type=int,
default=5)
parser.add_argument('--alpha',
help='exponent in discount function',
type=float,
default=0.5)
try:
args = vars(parser.parse_args())
except IOError as msg:
parser.error(str(msg))
if not args['asyn']:
# args['participate_rate'] = None
# args['refresh_rate'] = None
args['window_size'] = 1
args['alpha'] = 0.
# Set seeds
random.seed(1 + args['seed'])
np.random.seed(12 + args['seed'])
tf.set_random_seed(123 + args['seed'])
# load selected model
if args['dataset'].startswith("synthetic"): # all synthetic datasets use the same model
model_path = '%s.%s.%s.%s' % ('fedsimul', 'models', 'synthetic', args['model'])
else:
model_path = '%s.%s.%s.%s' % ('fedsimul', 'models', args['dataset'], args['model'])
mod = importlib.import_module(model_path)
model = getattr(mod, 'Model')
# load selected trainer
opt_path = 'fedsimul.trainers.%s' % args['optimizer']
mod = importlib.import_module(opt_path)
optimizer = getattr(mod, 'Server')
# add selected model parameter
args['model_params'] = MODEL_PARAMS['.'.join(model_path.split('.')[2:])]
# print and return
if args['verbose']:
maxLen = max([len(ii) for ii in args.keys()])
fmtString = '\t%' + str(maxLen) + 's : %s'
print('Arguments:')
for keyPair in sorted(args.items()):
print(fmtString % keyPair)
return args, model, optimizer
def main():
# parse command line arguments
args, model, optimizer = read_args()
tmpdir = tempfile.gettempdir()
# read data
train_path = os.path.join(tmpdir, 'data', args['dataset'], 'train')
test_path = os.path.join(tmpdir, 'data', args['dataset'], 'test')
dataset = read_data(train_path, test_path)
# REVIEW: distribution of dataset
# train_data = dataset[2]
# test_data = dataset[3]
# dist = []
# for k in train_data:
# dist.append(len(train_data[k]['x']) + len(test_data[k]['x']))
# # print(dist)
# import matplotlib.pyplot as plt
# fig = plt.figure(figsize=(4, 3))
# plt.hist(dist, bins=21)
# # plt.title('MNIST')
# plt.xlabel("number of samples")
# plt.ylabel("number of users")
# plt.tight_layout()
# plt.savefig('mnist_hist.pdf')
# exit()
# call appropriate trainer
t = optimizer(args, model, dataset)
t.train()
if __name__ == '__main__':
# suppress tf warnings
tf.compat.v1.logging.set_verbosity(tf.compat.v1.logging.ERROR)
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
main()