-
Notifications
You must be signed in to change notification settings - Fork 2
/
run_generator.py
75 lines (63 loc) · 2.42 KB
/
run_generator.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
"""run_generator.py generates samples from a LAGAN generator
Usage:
run_generator.py <N> <weights> <output> [--latent-space=<Z>]
run_generator.py -h | --help
Arguments:
<N> Number of images to generate
<weights> Location of generator weights hdf5 file
<output> Location to save to
Options:
-h --help Show this screen
--latent-space=<Z> Latent space (noise) vector size [default: 200]
"""
from docopt import docopt
import h5py
import numpy as np
import os
import time
def main(n_jets, outfile, latent_space, gen_weights=None):
"""
:param n_jets: number of jet imags to generate
:param gen_weights: weights to use
:param outfile: hdf file to save to
:param latent_space: latent space size
:return: runtime
"""
t0 = time.time()
from models.networks.lagan import generator as build_generator
g = build_generator(latent_space, return_intermediate=False)
if gen_weights is not None:
g.load_weights(os.path.abspath(gen_weights))
noise = np.random.normal(0, 1, (n_jets, latent_space))
sampled_labels = np.random.randint(0, 2, n_jets)
generated_images = g.predict(
[noise, sampled_labels.reshape(-1, 1)], verbose=False, batch_size=64)
# Multiply to image scale and remove extraneous axis
generated_images *= 100
generated_images = np.squeeze(generated_images)
# Save generated images
with h5py.File(os.path.abspath(outfile), 'w') as f:
dset = f.create_dataset('image', data=generated_images)
sigs = f.create_dataset('signal', data=sampled_labels)
return time.time() - t0 # return runtime
def performance(args, logfile='g_speed.txt'):
"""
Helper function for running performance tests on generator
:param args: command line arguments dictionary
:param logfile: file to write times to
:return: Returns nothing
"""
# exponents = np.linspace(1, 5, 9)
for N in np.logspace(1, 5, 9):
# N = int(10**e)
print(N)
t = main(N, args['<weights>'], args['<output>'],
int(args['--latent-space']))
with open(logfile, 'a') as f:
f.write('{}\t{}\n'.format(N, t)) # record number of jets and runtime
if __name__ == '__main__':
arguments = docopt(__doc__, help=True)
print(arguments)
# performance(arguments)
main(int(arguments['<N>']), arguments['<weights>'], arguments['<output>'],
int(arguments['--latent-space']))