-
Notifications
You must be signed in to change notification settings - Fork 2
/
tuner.py
62 lines (57 loc) · 2 KB
/
tuner.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
import subprocess
import os
import sys
logdir = 'dp_tuning'
data_size = 10_000
hidden_sizes = [128, 256, 512]
epsilons = [0.01, 0.1, 1.0, 10.0]
lambdas = [0.01, 0.1, 1.0, 10.0]
dropouts = [0.0, 0.2, 0.4]
learning_rate = [0.01, 0.001, 0.0001]
# for hidden_size in hidden_sizes:
# for lr in learning_rate:
# for dropout in dropouts:
# subprocess.call([sys.executable,
# 'main.py',
# "--log_dir",
# logdir,
# "--data_split",
# str(data_size),
# "--hidden_size",
# str(hidden_size),
# "--validate",
# "--dropout",
# str(dropout),
# "--tag",
# f"hidden_{hidden_size}_dropout_{dropout}_lr_{lr}"],
# env=os.environ.copy())
for epsilon in epsilons:
subprocess.call([sys.executable,
'main.py',
"--log_dir",
logdir,
"--data_split",
str(data_size),
"--epsilon",
str(epsilon),
"--dp",
"--adversarial",
"--validate",
"--tag",
f"epsilon_{epsilon}"],
env=os.environ.copy())
for lbd in lambdas:
subprocess.call([sys.executable,
'main.py',
"--log_dir",
logdir,
"--data_split",
str(data_size),
"--hplambda",
str(lbd),
"--dp",
"--adversarial",
"--validate",
"--tag",
f"lambda_{lbd}"],
env=os.environ.copy())