-
Notifications
You must be signed in to change notification settings - Fork 0
/
plot.py
220 lines (185 loc) · 7.1 KB
/
plot.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
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
import argparse
import matplotlib
import numpy as np
import seaborn as sns
import json
import csv
from matplotlib import pyplot as plt
from pathlib import Path
from datastuff import get_distortion_tests_name, get_distortion_tests, get_test_loader2
from create_consolidated_result_json import get_arch_dict
def parse_args():
parser = argparse.ArgumentParser()
parser.add_argument('--log_dir', type=str, default='save_dir')
parser.add_argument('--num_trains', type=int, nargs='+', default=[5000, 10000])
parser.add_argument('--arch_type', type=str, default='densenet')
args = parser.parse_args()
return args
def get_num_train_acc_arch_dict_old(logdir='save_dir', net_type='resnet',
num_train=1000,
base_date='2019-06-11-03-37'):
""" Looks like:
arch_dict[configs]
arch_dict['accs'] = []
arch_dict['archs'] = []"""
if isinstance(logdir, str):
logdir = Path(logdir)
base_str = f'{net_type}-{base_date}-*-{num_train}-*'
acc_paths = logdir.glob(f'{base_str}/accuracy0.txt')
config_paths = logdir.glob(f'{base_str}/config.json')
arch_paths = logdir.glob(f'{base_str}/log.txt')
arch_dict = {}
accs = []
archs = []
for acc_p in acc_paths:
with open(acc_p, 'r') as acc_f:
lines = acc_f.readlines()
for line in lines:
accs.append(float(line))
for arch_p in arch_paths:
with open(arch_p, 'r') as arch_f:
lines = arch_f.readlines()
for line in lines:
idx = line.index(',,')
archs.append(line[idx+2:-1])
for cfg in config_paths:
with open(cfg, 'r') as c:
cfgdata = json.load(c)
break
arch_dict = cfgdata
arch_dict['accs'] = accs
arch_dict['num_train'] = num_train
arch_dict['avg_acc'] = sum(accs) / len(accs)
arch_dict['max_acc'] = max(accs)
arch_dict['archs'] = archs
return arch_dict
def get_num_train_acc_arch_dict(logdir='save_dir', net_type='resnet',
num_train=1000, acc_type='normal',
base_date='2019-07-25'):
""" Looks like:
arch_dict[configs]
arch_dict['accs'] = []
arch_dict['archs'] = []"""
if isinstance(logdir, str):
logdir = Path(logdir)
base_str = f'{net_type}-{base_date}-*-{num_train}-*'
acc_paths = logdir.glob(f'{base_str}/accuracies.json')
config_paths = logdir.glob(f'{base_str}/config.json')
arch_paths = logdir.glob(f'{base_str}/log-active.txt')
arch_dict = {}
acc_dict = {}
accs = []
archs = []
acc_ctr = 0
for acc_p in acc_paths:
with open(acc_p, 'r') as acc_f:
acc = json.load(acc_f)
for _, value in acc.items():
accs.append(float(value[acc_type]))
acc_dict[acc_ctr] = value
for arch_p in arch_paths:
with open(arch_p, 'r') as arch_f:
lines = arch_f.readlines()
for line in lines:
idx = line.index(',,')
archs.append(line[idx+2:-1])
for cfg in config_paths:
with open(cfg, 'r') as c:
cfgdata = json.load(c)
break
arch_dict = cfgdata
arch_dict['accs'] = accs
arch_dict['num_train'] = num_train
arch_dict['avg_acc'] = sum(accs) / len(accs)
arch_dict['max_acc'] = max(accs)
arch_dict['archs'] = archs
return arch_dict
def get_imgs(nr=0):
paths = get_distortion_tests('test-distortions/')
# imgs = np.zeros(len(paths))
imgs = []
for idx, path in enumerate(paths):
img = np.load(path)
# imgs[idx] = img[nr]
imgs.append(img[nr])
return imgs
# def get_img_index(img):
def to_subplot(axes, picture, title):
# axes.axis('off')
axes.imshow(picture, cmap='gray', interpolation='nearest')
axes.set_title(title, fontsize=30)
axes.set_yticklabels([])
axes.set_xticklabels([])
#axes.xticks([], [])
#axes.yticks([], [])
def plot_influence(arch_string, s_tests_id=0, title=None):
"""DOCS"""
fig, axes = plt.subplots(nrows=4, ncols=5)
# for id, ax in enumerate(axes[:, 0]):
# if id > 1:
# ax.set_ylabel('Harmful ', rotation=0, size='large')
# else:
# ax.set_ylabel('Helpful ', rotation=0, size='large')
imgs = get_imgs()
to_subplot(
axes[0, 0],
[[],[],[]],
# get_test_loader2().dataset.data[int(s_tests_id)],
# 'normal')
'')
# for i, ii, iii in [[0, 1, 0], [0, 2, 1], [0, 3, 2], [1, 0, 3], [1, 1, 4], [1, 2, 5], [1, 3, 6]]:
for i, ii, iii in [[0, 1, 0], [0, 2, 1], [0, 3, 2], [0, 4, 7], [1, 0, 3], [1, 1, 4], [1, 2, 5], [1, 3, 6], [1, 4, 8]]:
to_subplot(
axes[i, ii],
imgs[iii],
f'{get_distortion_tests_name()[iii]}')
for i, ii, iii in [[2, 0, 9], [2, 1, 10], [2, 2, 11], [2, 3, 12], [2, 4, 13], [3, 0, 14], [3, 1, 15], [3, 2, 16], [3, 3, 17], [3, 4, 18]]:
to_subplot(
axes[i, ii],
imgs[iii],
f'{get_distortion_tests_name()[iii]}')
if not title:
fig.suptitle(
f"CIFAR10 test dataset postprocessed with various distortions",
fontsize=45)
else:
# fig.suptitle(title, size='x-large')
fig.suptitle(title, fontsize=45)
fig.subplots_adjust(left=None, bottom=None, right=None, top=None,
wspace=-0.2, hspace=0.4)
fig.tight_layout()
inf_fig_folder = "figures/influences/inf_"
fig.set_figheight(13)
fig.set_figwidth(13)
# if recursion_depth and r_avg:
# fig_fn = f'{inf_fig_folder}{arch_string}_rec-dep{recursion_depth}_r-avg{r_avg}_{get_cifar10_class(class_nr)}_{s_tests_id}.svg'
# fig_fn = f'{inf_fig_folder}{arch_string}_rec-dep{recursion_depth}_r-avg{r_avg}_{get_cifar10_class(class_nr)}_{s_tests_id}.png'
# else:
# fig_fn = f'{inf_fig_folder}{arch_string}_{get_cifar10_class(class_nr)}_{s_tests_id}.svg'
# fig_fn = f'{inf_fig_folder}{arch_string}_{get_cifar10_class(class_nr)}_{s_tests_id}.png'
# print(f'Saved to: {fig_fn}')
# fig.savefig(fig_fn, figsize=(7, 7), dpi=150)
plt.show()
if __name__ == "__main__":
args = parse_args()
arch_dicts = []
# args.num_trains = [500, 1000, 5000, 10000, 25000]
accs = {}
network_type = 'densenet'
arch_d = get_arch_dict(net_type=network_type, base_date='2019')
types = get_distortion_tests_name()
for num_train in args.num_trains:
for typ in types:
d = get_num_train_acc_arch_dict(num_train=num_train, acc_type=typ)
accs[typ] = d['avg_acc']
arch_dicts.append(d)
# with open(f'vgg-per_type.csv', 'a') as fout:
# wrtr = csv.writer(fout)
# wrtr = csv.DictWriter(fout, fieldnames=types)
# wrtr.writeheader()
# wrtr.writerow(accs)
# wrtr.writerow([i for i in range(len(d['accs']))])
# wrtr.writerow(lay_num)
# plot_influence('hi')
with open(f'{network_type}.json', 'w+') as fout:
json.dump(arch_d, fout, indent=2)