-
Notifications
You must be signed in to change notification settings - Fork 343
/
search_hyperparams.py
91 lines (72 loc) · 3.41 KB
/
search_hyperparams.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
"""
Peform hyperparemeters search
A brief definition/clarification of 'params.json' files:
"model_version": "resnet18", # "base" models or "modelname"_distill models
"subset_percent": 1.0, # use full (1.0) train set or partial (<1.0) train set
"augmentation": "yes", # whether to use data augmentation in data_loader
"teacher": "densenet", # no need to specify this for "base" cnn/resnet18
"alpha": 0.0, # only used for experiments involving distillation
"temperature": 1, # only used for experiments involving distillation
"learning_rate": 1e-1, # as the name suggests
"batch_size": 128, # for both train/eval
"num_epochs": 200, # as the name suggests
"dropout_rate": 0.5, # only valid for "cnn"-related models, not in resnet18
"num_channels": 32, # only valid for "cnn"-related models, not in resnet18
"save_summary_steps": 100,
"num_workers": 4
"""
import argparse
import os
from subprocess import check_call
import sys
import utils
import logging
PYTHON = sys.executable
parser = argparse.ArgumentParser()
parser.add_argument('--parent_dir', default='experiments/learning_rate',
help='Directory containing params.json')
def launch_training_job(parent_dir, job_name, params):
"""Launch training of the model with a set of hyperparameters in parent_dir/job_name
Args:
model_dir: (string) directory containing config, weights and log
data_dir: (string) directory containing the dataset
params: (dict) containing hyperparameters
"""
# Create a new folder in parent_dir with unique_name "job_name"
model_dir = os.path.join(parent_dir, job_name)
if not os.path.exists(model_dir):
os.makedirs(model_dir)
# Write parameters in json file
json_path = os.path.join(model_dir, 'params.json')
params.save(json_path)
# Launch training with this config
cmd = "{python} train.py --model_dir={model_dir}".format(python=PYTHON,
model_dir=model_dir)
print(cmd)
check_call(cmd, shell=True)
if __name__ == "__main__":
# Load the "reference" parameters from parent_dir json file
args = parser.parse_args()
json_path = os.path.join(args.parent_dir, 'params.json')
assert os.path.isfile(json_path), "No json configuration file found at {}".format(json_path)
params = utils.Params(json_path)
# Set the logger
utils.set_logger(os.path.join(args.parent_dir, 'search_hyperparameters.log'))
'''
Temperature and alpha search for KD on CNN (teacher model picked in params.json)
Perform hypersearch (empirical grid): distilling 'temperature', loss weight 'alpha'
'''
# hyperparameters for KD
alphas = [0.99, 0.95, 0.5, 0.1, 0.05]
temperatures = [20., 10., 8., 6., 4.5, 3., 2., 1.5]
logging.info("Searching hyperparameters...")
logging.info("alphas: {}".format(alphas))
logging.info("temperatures: {}".format(temperatures))
for alpha in alphas:
for temperature in temperatures:
# [Modify] the relevant parameter in params (others remain unchanged)
params.alpha = alpha
params.temperature = temperature
# Launch job (name has to be unique)
job_name = "alpha_{}_Temp_{}".format(alpha, temperature)
launch_training_job(args.parent_dir, job_name, params)