-
Notifications
You must be signed in to change notification settings - Fork 4
/
testing.py
242 lines (205 loc) · 11.2 KB
/
testing.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
from numpy.linalg import svd
from torch.optim import SGD, Adam
from collections import Counter
from itertools import chain
from utils.utils import *
import torch
import clip
import matplotlib.pyplot as plt
from torch.autograd import Variable
from numpy.random import multivariate_normal
from tqdm import tqdm
from utils.my_ipca import MyIPCA as IPCA
from sklearn.decomposition import PCA
from sklearn.metrics import roc_auc_score
def test(task_list, args, train_data, test_data, model):
# noise cannot be used without use_md
if args.noise: assert args.use_md
zeroshot = Zeroshot(args.model_clip, args)
cil_tracker = Tracker(args)
cal_cil_tracker = Tracker(args)
til_tracker = Tracker(args)
auc_softmax_tracker = AUCTracker(args)
openworld_softmax_tracker = OWTracker(args)
# cil_correct, til_correct are for cumulative accuracy throughout training
cil_correct, til_correct, total = 0, 0, 0
c_correct, c_total, p_correct, p_total = 0, 0, 0, 0
cum_acc_list, total_loss_list, iter_list, total_iter = [], [], [], 0
train_loaders, test_loaders, calibration_loaders = [], [], []
args.mean, args.cov, args.cov_inv = {}, {}, {}
args.mean_task, args.cov_noise, args.cov_inv_noise = {}, {}, {}
# calibration weight w and bias b
w, b = None, None
param_copy = None
combined_sigma = 0
if args.task_type == 'concept': if_shift = []
for task_id in range(args.load_task_id + 1):
task_loss_list = []
if args.validation is None:
t_train = train_data.make_dataset(task_id)
t_test = test_data.make_dataset(task_id)
else:
t_train, t_test = train_data.make_dataset(task_id)
if args.calibration:
assert args.cal_batch_size > 0
assert args.cal_epochs > 0
assert args.cal_size > 0
t_train, t_cal = calibration_dataset(args, t_train)
calibration_loaders.append(make_loader(t_cal, args, train='calibration'))
train_loaders.append(make_loader(t_train, args, train='train'))
test_loaders.append(make_loader(t_test, args, train='test'))
if hasattr(model, 'preprocess_task'):
model.preprocess_task(names=train_data.task_list[task_id][0],
labels=train_data.task_list[task_id][1],
task_id=task_id,
loader=train_loaders[-1])
# Load model
if args.test_model_name is None:
test_model_name = 'model_task_'
else:
test_model_name = args.test_model_name
if os.path.exists(args.load_dir + '/' + test_model_name + str(task_id)):
filename = args.load_dir + '/' + test_model_name + str(task_id)
args.logger.print("Load a trained model from:")
args.logger.print(filename)
state_dict = torch.load(filename)
model.net.load_state_dict(state_dict)
if args.train_clf_id is not None:
if task_id <= args.train_clf_id:
filename = args.load_dir + '/' + f'model_task_clf_epoch=10_{task_id}'
else:
filename = args.load_dir + '/' + f'model_task_clf_epoch=10_{args.train_clf_id}'
filename = torch.load(filename)
for n, p in model.net.named_parameters():
if 'head' in n:
if n in filename.keys():
args.logger.print("changed head:", n)
print('before', torch.sum(p), torch.sum(filename[n]))
p.data = filename[n].data
print('after', torch.sum(p), torch.sum(filename[n]))
else:
raise NotImplementedError(args.load_dir + '/' + test_model_name + str(task_id), "Load dir incorrect")
# Load statistics for MD
if os.path.exists(args.load_dir + f'/cov_task_{task_id}.npy'):
args.compute_md = True
args.logger.print("*** Load Statistics for MD ***")
cov = np.load(args.load_dir + f'/cov_task_{task_id}.npy')
args.cov[task_id] = cov
args.cov_inv[task_id] = np.linalg.inv(cov)
if args.noise:
mean = np.load(args.load_dir + f'/mean_task_{task_id}.npy')
args.mean_task[task_id] = mean
cov = np.load(args.load_dir + f'/cov_task_noise_{task_id}.npy')
args.cov_noise[task_id] = cov
args.cov_inv_noise[task_id] = np.linalg.inv(cov)
for y in range(task_id * args.num_cls_per_task, (task_id + 1) * args.num_cls_per_task):
mean = np.load(args.load_dir + f'/mean_label_{y}.npy')
args.mean[y] = mean
else:
args.logger.print("*** No MD ***")
if args.task_type == 'concept':
if 'shifted' in train_data.current_labels:
args.logger.print(train_data.current_labels)
if_shift.append(True)
init = int(train_data.current_labels.split('shifted: ')[-1].split(' -> ')[0])
test_loaders[init].dataset.update()
args.logger.print(len(test_loaders[init].dataset.targets))
else:
if_shift.append(False)
if args.modify_previous_ood and task_id > 0:
assert args.model == 'oe' or args.model == 'oe_fixed_minibatch'
param_copy = model.net.fc.weight.detach()
args.logger.print(param_copy.sum(1))
# Check/load if calibration is saved
if os.path.exists(args.load_dir + f'/w_b_task_{task_id}'):
w = torch.load(args.load_dir + f'/w_b_task_{task_id}')[0].to(args.device)
b = torch.load(args.load_dir + f'/w_b_task_{task_id}')[1].to(args.device)
if args.train_clf_id is not None:
if task_id <= args.train_clf_id:
continue
for x, y, _, _, _ in test_loaders[-1]: # THIS NEEDS TO BE CHANGE (FROM TEST TO TRAINLOADER) WHEN SAVING FEATURES
x, y = x.to(args.device), y.to(args.device)
with torch.no_grad():
if args.model_clip:
x = args.model_clip.encode_image(x).type(torch.FloatTensor).to(args.device)
if args.zero_shot:
text_inputs = torch.cat([clip.tokenize(f"a photo of a {c}") for c in train_data.seen_names]).to(args.device)
zeroshot.evaluate(x, text_inputs, y)
model.evaluate(x, y, task_id, w=w, b=b, total_learned_task_id=task_id, ensemble=args.pass_ensemble)
metrics = model.acc()
cil_tracker.update(metrics['cil_acc'], task_id, task_id)
til_tracker.update(metrics['til_acc'], task_id, task_id)
cal_cil_tracker.update(metrics['cal_cil_acc'], task_id, task_id)
if args.compute_auc:
in_scores = metrics['scores']
if args.compute_md: in_scores_md = metrics['scores_md']
auc_list, auc_list_md = [], []
auc_total_in_list, auc_total_out_list, out_id_list = [metrics['scores_total']], [], []
for task_out in range(args.n_tasks):
if task_out != task_id:
if args.validation is None:
t_test = test_data.make_dataset(task_out)
else:
_, t_test = train_data.make_dataset(task_out)
ood_loader = make_loader(t_test, args, train='test')
for x, y, _, _, _ in ood_loader:
x, y = x.to(args.device), y.to(args.device)
with torch.no_grad():
model.evaluate(x, y, task_id=task_id, total_learned_task_id=task_id, ensemble=args.pass_ensemble)
metrics = model.acc()
if task_out <= task_id:
cil_tracker.update(metrics['cil_acc'], task_id, task_out)
til_tracker.update(metrics['til_acc'], task_id, task_out)
out_scores = metrics['scores']
auc = compute_auc(in_scores, out_scores)
auc_list.append(auc * 100)
args.logger.print("in/out: {}/{} | Softmax AUC: {:.2f}".format(task_id, task_out, auc_list[-1]), end=' ')
auc_softmax_tracker.update(auc_list[-1], task_id, task_out)
if args.compute_md:
out_scores_md = metrics['scores_md']
auc_md = compute_auc(in_scores_md, out_scores_md)
auc_list_md.append(auc_md * 100)
args.logger.print("| MD AUC: {:.2f}".format(auc_list_md[-1]))
else:
args.logger.print('')
if task_out <= task_id:
auc_total_in_list.append(metrics['scores_total'])
else:
auc_total_out_list.append(metrics['scores_total'])
out_id_list.append(task_out)
args.logger.print("Average Softmax AUC: {:.2f}".format(np.array(auc_list).mean()))
if args.compute_md:
args.logger.print("Average MD AUC: {:.2f}".format(np.array(auc_list_md).mean()))
for task_out, out_scores in zip(out_id_list, auc_total_out_list):
auc = compute_auc(auc_total_in_list, out_scores)
args.logger.print("total in/out: {}/{} | AUC: {:.2f}".format(task_id, task_out, auc * 100))
openworld_softmax_tracker.update(auc * 100, task_id, task_out)
if len(auc_total_in_list) > 0 and len(auc_total_out_list) > 0:
auc = compute_auc(auc_total_in_list, auc_total_out_list)
args.logger.print("total in | AUC: {:.2f}".format(auc * 100))
# Report CIL, TIL, Cal_CIL. If compute_auc is True, they've already computed
args.logger.print("######################")
args.logger.print()
if args.compute_auc:
args.logger.print("Softmax AUC result")
auc_softmax_tracker.print_result(task_id, type='acc')
args.logger.print("Open World result")
openworld_softmax_tracker.print_result(task_id, type='acc')
args.logger.print("CIL result")
cil_tracker.print_result(task_id, type='acc')
cil_tracker.print_result(task_id, type='forget')
args.logger.print("TIL result")
til_tracker.print_result(task_id, type='acc')
til_tracker.print_result(task_id, type='forget')
args.logger.print()
if task_id == 0 and args.calibration:
cal_cil_tracker.mat = deepcopy(cil_tracker.mat)
if w is not None:
args.logger.print("Cal CIL result")
cal_cil_tracker.print_result(task_id, type='acc')
cal_cil_tracker.print_result(task_id, type='forget')
torch.save(cil_tracker.mat, args.logger.dir() + '/cil_tracker_train_clf_equal_test')
torch.save(til_tracker.mat, args.logger.dir() + '/til_tracker_train_clf_equal_test')
torch.save(auc_softmax_tracker.mat, args.logger.dir() + '/auc_softmax_tracker_train_clf_equal_test')
torch.save(openworld_softmax_tracker.mat, args.logger.dir() + '/openworld_softmax_tracker_train_clf_equal_test')
return cil_tracker.mat, til_tracker.mat, cal_cil_tracker.mat