-
Notifications
You must be signed in to change notification settings - Fork 0
/
plot_results.py
308 lines (271 loc) · 12.8 KB
/
plot_results.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
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
"""
plot.results.py
Written by Sindre Stenen Blakseth, 2020.
Script for plotting ESRGAN training and validation data recorded in .txt-files.
"""
import matplotlib.pyplot as plt
import numpy as np
from os.path import exists, join
import pickle
from config import run_dir as data_dir
#----------------------------------------------------------------------------
# Loading data to plot.
def load_data(filename):
data_dict = {}
# Pickle data if not already pickled.
if not exists(join(data_dir, filename + ".pkl")):
# Open text file.
with open(join(data_dir, filename + ".txt"), "r") as data_file:
train_data_strings = data_file.readlines()
# Format loaded strings.
n_lines = len(train_data_strings)
for i in range(n_lines):
train_data_strings[i] = train_data_strings[i].strip() # Remove last newline character.
# Load first line, which contains the number of variables stored per line.
n_var = train_data_strings[0]
print("Number of variables to load is " + n_var)
# Load second line, which contains labels.
data_labels = train_data_strings[1].split(",")
for label in data_labels:
data_dict[label] = []
# Load all other lines, which contain recorded training data.
for i in range(2, n_lines):
data = train_data_strings[i].split(",")
for j, label in enumerate(data_labels):
data_dict[label].append(float(data[j]))
# Convert lists to numpy arrays.
for label in data_labels:
data_dict[label] = np.asarray(data_dict[label])
# Save dict as pickle for easy access later.
with open(join(data_dir, filename + ".pkl"), 'wb') as f:
pickle.dump(data_dict, f)
else:
with open(join(data_dir, filename + ".pkl"), 'rb') as f:
data_dict = pickle.load(f)
return data_dict
#----------------------------------------------------------------------------
# Plotting data.
def main():
train_data_filename = "training_data"
val_loss_filename = "val_loss_metrics"
val_grad_filename = "val_w_and_grads"
val_pred_filename = "val_D_predicts"
train_data_dict = load_data(train_data_filename)
val_loss_dict = load_data(val_loss_filename)
val_grad_dict = load_data(val_grad_filename)
val_data_dict = {**val_loss_dict, **val_grad_dict}
#------------------------------------------------------------------------
print("Begin plotting generator training loss.")
plt.figure()
plt.title("Generator Training Loss")
plt.xlabel("Training iterations")
plt.ylabel("Loss")
plt.plot(train_data_dict['it'], train_data_dict['G_total_loss'], label='G total')
plt.plot(train_data_dict['it'], train_data_dict['G_rel_avg_loss'], label='G RaGAN')
plt.plot(train_data_dict['it'], train_data_dict['G_pix_l1_loss'], label='G L1-pix')
plt.plot(train_data_dict['it'], train_data_dict['G_vgg19_l1_loss'], label='G L1-feat')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "generator_training_loss.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting discriminator training loss.")
plt.figure()
plt.title("Discriminator Training Loss")
plt.xlabel("Training iterations")
plt.ylabel("Loss")
plt.plot(train_data_dict['it'], train_data_dict['D_total_loss'], label='D total')
plt.plot(train_data_dict['it'], train_data_dict['D_rel_avg_loss'], label='D RaGAN')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "discriminator_training_loss.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting generator validation loss.")
plt.figure()
plt.title("Generator Validation Loss")
plt.xlabel("Training iterations")
plt.ylabel("Loss")
plt.plot(val_data_dict['it'], val_data_dict['G_total_loss'], label='G total')
plt.plot(val_data_dict['it'], val_data_dict['G_rel_avg_loss'], label='G RaGAN')
plt.plot(val_data_dict['it'], val_data_dict['G_pix_l1_loss'], label='G L1-pix')
plt.plot(val_data_dict['it'], val_data_dict['G_vgg19_l1_loss'], label='G L1-feat')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "generator_validation_loss.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting discriminator validation loss.")
plt.figure()
plt.title("Discriminator Validation Loss")
plt.xlabel("Training iterations")
plt.ylabel("Loss")
plt.plot(val_data_dict['it'], val_data_dict['D_total_loss'], label='D total')
plt.plot(val_data_dict['it'], val_data_dict['D_rel_avg_loss'], label='D RaGAN')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "discriminator_validation_loss.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting PSNR on validation set.")
plt.figure()
plt.title("Average PSNR Score over the Validation Set")
plt.xlabel("Training iterations")
plt.ylabel("PSNR")
plt.plot(val_data_dict['it'], val_data_dict['psnr'], label='PSNR')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "validation_psnr.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting LPIPS on validation set.")
plt.figure()
plt.title("Average LPIPS Score over the Validation Set")
plt.xlabel("Training iterations")
plt.ylabel("LPIPS")
plt.plot(val_data_dict['it'], val_data_dict['lpips'], label='LPIPS')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "validation_lpips.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting accuracies on validation set.")
plt.figure()
plt.title("Average Discriminator accuracies over the Validation Set")
plt.xlabel("Training iterations")
plt.ylabel("Accuracy")
plt.plot(val_data_dict['it'], val_data_dict['D_HR_acc'], label='HR')
plt.plot(val_data_dict['it'], val_data_dict['D_SR_acc'], label='SR')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "validation_accuracies.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting mean gradients in generator.")
plt.figure()
plt.title("Mean Gradients of Generator Layers")
plt.xlabel("Training iterations")
plt.ylabel(r"mean($\partial w / \partial L$)")
plt.plot(val_data_dict['it'], val_data_dict['G_grad_start_mean'], label='first layer')
plt.plot(val_data_dict['it'], val_data_dict['G_grad_mid_mean'], label='mid layer')
plt.plot(val_data_dict['it'], val_data_dict['G_grad_end_mean'], label='last layer')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "generator_mean_gradients.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting mean absolute gradients in generator.")
plt.figure()
plt.title("Mean Absolute Value of Gradients of Generator Layers")
plt.xlabel("Training iterations")
plt.ylabel(r"mean($|\partial w / \partial L|$)")
plt.plot(val_data_dict['it'], val_data_dict['G_grad_start_abs_mean'], label='first layer')
plt.plot(val_data_dict['it'], val_data_dict['G_grad_mid_abs_mean'], label='mid layer')
plt.plot(val_data_dict['it'], val_data_dict['G_grad_end_abs_mean'], label='last layer')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "generator_mean_abs_gradients.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting variance of gradients in generator.")
plt.figure()
plt.title("Variance of Gradients of Generator Layers")
plt.xlabel("Training iterations")
plt.ylabel(r"var($\partial w / \partial L$)")
plt.plot(val_data_dict['it'], val_data_dict['G_grad_start_variance'], label='first layer')
plt.plot(val_data_dict['it'], val_data_dict['G_grad_mid_variance'], label='mid layer')
plt.plot(val_data_dict['it'], val_data_dict['G_grad_end_variance'], label='last layer')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "generator_var_gradients.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting mean gradients in discriminator.")
plt.figure()
plt.title("Mean Gradients of Discriminator Layers")
plt.xlabel("Training iterations")
plt.ylabel(r"mean($\partial w / \partial L$)")
plt.plot(val_data_dict['it'], val_data_dict['D_grad_start_mean'], label='first layer')
plt.plot(val_data_dict['it'], val_data_dict['D_grad_end_mean'], label='last layer')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "discriminator_mean_gradients.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting mean absolute gradients in discriminator.")
plt.figure()
plt.title("Mean Absolute Value of Gradients of Discriminator Layers")
plt.xlabel("Training iterations")
plt.ylabel(r"mean($|\partial w / \partial L|$)")
plt.plot(val_data_dict['it'], val_data_dict['D_grad_start_abs_mean'], label='first layer')
plt.plot(val_data_dict['it'], val_data_dict['D_grad_end_abs_mean'], label='last layer')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "discriminator_mean_abs_gradients.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting variance of gradients in discriminator.")
plt.figure()
plt.title("Variance of Gradients of Discriminator Layers")
plt.xlabel("Training iterations")
plt.ylabel(r"var($\partial w / \partial L$)")
plt.plot(val_data_dict['it'], val_data_dict['D_grad_start_variance'], label='first layer')
plt.plot(val_data_dict['it'], val_data_dict['D_grad_end_variance'], label='last layer')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "discriminator_var_gradients.pdf"))
plt.close()
#------------------------------------------------------------------------
print("Begin plotting mean absolute weights in generator.")
plt.figure()
plt.title("Mean Absolute Value of Weights of Generator Layers")
plt.xlabel("Training iterations")
plt.ylabel(r"mean($|w|$)")
plt.plot(val_data_dict['it'], val_data_dict['G_weight_start_abs_mean'], label='first layer')
plt.plot(val_data_dict['it'], val_data_dict['G_weight_mid_abs_mean'], label='mid layer')
plt.plot(val_data_dict['it'], val_data_dict['G_weight_end_abs_mean'], label='last layer')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "generator_mean_abs_weights.pdf"))
plt.close()
# ------------------------------------------------------------------------
print("Begin plotting variance of weights in generator.")
plt.figure()
plt.title("Variance of Weights of Generator Layers")
plt.xlabel("Training iterations")
plt.ylabel(r"var(w)")
plt.plot(val_data_dict['it'], val_data_dict['G_weight_start_variance'], label='first layer')
plt.plot(val_data_dict['it'], val_data_dict['G_weight_mid_variance'], label='mid layer')
plt.plot(val_data_dict['it'], val_data_dict['G_weight_end_variance'], label='last layer')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "generator_var_weights.pdf"))
plt.close()
# ------------------------------------------------------------------------
print("Begin plotting mean absolute weights in discriminator.")
plt.figure()
plt.title("Mean Absolute Value of Weights of Discriminator Layers")
plt.xlabel("Training iterations")
plt.ylabel(r"mean($|w|$)")
plt.plot(val_data_dict['it'], val_data_dict['D_weight_start_abs_mean'], label='first layer')
plt.plot(val_data_dict['it'], val_data_dict['D_weight_end_abs_mean'], label='last layer')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "discriminator_mean_abs_weights.pdf"))
plt.close()
# ------------------------------------------------------------------------
print("Begin plotting variance of weights in discriminator.")
plt.figure()
plt.title("Variance of Weights of Discriminator Layers")
plt.xlabel("Training iterations")
plt.ylabel(r"var($w$)")
plt.plot(val_data_dict['it'], val_data_dict['D_weight_start_variance'], label='first layer')
plt.plot(val_data_dict['it'], val_data_dict['D_weight_end_variance'], label='last layer')
plt.legend()
plt.grid()
plt.savefig(join(data_dir, "discriminator_var_weights.pdf"))
plt.close()
#----------------------------------------------------------------------------
if __name__ == "__main__":
main()
#----------------------------------------------------------------------------