-
Notifications
You must be signed in to change notification settings - Fork 18
/
utils.py
309 lines (268 loc) · 11.8 KB
/
utils.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
309
"""
Author: c0ldstudy
2022-03-29 13:44:53
"""
import os
import torch
from torchvision import datasets, transforms
import numpy as np
import collections
import random
class TorchVisionDataset:
"""
- name: the dataset name
- subset: the subset of the main dataset. Dataset will be loaded as ``nlp.load_dataset(name, subset)``.
- label_map: Mapping if output labels should be re-mapped. Useful
if model was trained with a different label arrangement than
provided in the ``nlp`` version of the dataset.
- output_scale_factor (float): Factor to divide ground-truth outputs by.
Generally, TextAttack goal functions require model outputs
between 0 and 1. Some datasets test the model's correlation
with ground-truth output, instead of its accuracy, so these
outputs may be scaled arbitrarily.
- shuffle (bool): Whether to shuffle the dataset on load.
"""
def __init__(
self, name,data, split="train", shuffle=True,
):
self._name = name
self._split = split
self._dataset = data
# Input/output column order, like (('premise', 'hypothesis'), 'label')
self.input_columns, self.output_column = ("image", "label")
self._i = 0
self.examples = list(self._dataset)
if shuffle:
random.shuffle(self.examples)
def __len__(self):
return len(self._dataset)
def _format_raw_example(self, raw_example):
return raw_example
def __next__(self):
if self._i >= len(self.examples):
raise StopIteration
raw_example = self.examples[self._i]
self._i += 1
return self._format_raw_example(raw_example)
def __getitem__(self, i):
if isinstance(i, int):
return self._format_raw_example(self.examples[i])
else:
# `i` could be a slice or an integer. if it's a slice,
# return the formatted version of the proper slice of the list
return [self._format_raw_example(ex) for ex in self.examples[i]]
def get_json_data(self):
if self.examples:
new_data = []
for idx, instance in enumerate(self.examples):
new_instance = {}
new_instance["image"] = instance[0].numpy().tolist()
new_instance["label"] = instance[1]
new_instance["uid"] = idx
new_data.append(new_instance)
return new_data
def get_dataset(dataset_configs):
if dataset_configs['name'] == "MNIST":
return _read_mnist_dataset(dataset_configs, "MNIST")
elif dataset_configs['name'] == "CIFAR10":
return _read_cifar10_dataset(dataset_configs, "CIFAR10")
def _split_by_labels(num_classes, train_data, server_number_sampled, train_server_path):
subset_indices = []
for i in range(num_classes):
indices_xi = (torch.LongTensor(train_data.targets) == i).nonzero(as_tuple=True)[0]
sampled_indices = np.random.choice(
indices_xi, server_number_sampled, replace=False
)
subset_indices.extend(sampled_indices)
train_server_subset = torch.utils.data.Subset(train_data, subset_indices)
torch.save(train_server_subset, train_server_path)
return train_server_subset
def label_update(num_classes, train_data, server_number_sampled, train_server_path, custom=False, label_custom=None):
subset_indices = []
if custom == True:
for i in list(label_custom.keys()):
for ori_label in label_custom[i]:
single_number_sampled = int(server_number_sampled / len(label_custom[i]))
# print(single_number_sampled)
# break
indices_xi = (torch.LongTensor(train_data.targets) == ori_label).nonzero(as_tuple=True)[0]
sampled_indices = np.random.choice(indices_xi, single_number_sampled, replace=False)
subset_indices.extend(sampled_indices)
train_data.targets = torch.tensor(train_data.targets)
for i in list(label_custom.keys()):
mask = sum(train_data.targets==i for i in label_custom[i]).bool()
train_data.targets[mask] = i
# mask = sum(train_data.targets==i for i in label_binary[0]).bool()
# train_data.targets[mask] = 0
# mask = sum(train_data.targets==i for i in label_binary[1]).bool()
# train_data.targets[mask] = 1
# mask = sum(train_data.targets==i for i in label_binary[2]).bool()
# train_data.targets[mask] = 2
# mask = sum(train_data.targets==i for i in label_binary[3]).bool()
# train_data.targets[mask] = 3
else:
for i in range(num_classes):
indices_xi = (torch.LongTensor(train_data.targets) == i).nonzero(as_tuple=True)[0]
sampled_indices = np.random.choice(indices_xi, server_number_sampled, replace=False)
subset_indices.extend(sampled_indices)
train_server_subset = torch.utils.data.Subset(train_data, subset_indices)
# print(len(train_server_subset))
# print(np.unique(train_server_subset.targets, return_counts=True))
# exit()
torch.save(train_server_subset, train_server_path)
return train_server_subset
def _read_mnist_dataset(dataset_configs, dataset_name):
path = dataset_configs["dataset_path"]
# 1.1 Training Data
train_student_path = os.path.join(path, "train_split.pt")
if os.path.exists(train_student_path):
train_student_subset = torch.load(train_student_path)
else:
train_data = datasets.MNIST(
root=path,
train=True,
download=True,
transform=transforms.Compose([transforms.ToTensor(),]),
)
num_classes = len(train_data.classes)
student_number_sampled = dataset_configs["student_train_number"] // num_classes
train_student_subset = _split_by_labels(
num_classes, train_data, student_number_sampled, train_student_path
)
train_student_data = TorchVisionDataset(
name=dataset_name, data=train_student_subset, split="train",
)
# 1.2 Test Data
test_student_path = os.path.join(path, "test_split.pt")
if os.path.exists(test_student_path):
test_student_subset = torch.load(test_student_path)
else:
test_data = datasets.MNIST(
root=path,
train=False,
download=True,
transform=transforms.Compose([transforms.ToTensor(),]),
)
num_classes = len(test_data.classes)
student_number_sampled = dataset_configs["student_test_number"] // num_classes
test_student_subset = _split_by_labels(
num_classes, test_data, student_number_sampled, test_student_path
)
test_student_data = TorchVisionDataset(
name=dataset_name, data=test_student_subset, split="test",
)
# 1.3 Val Data
val_student_path = os.path.join(path, "val_split.pt")
if os.path.exists(val_student_path):
val_student_subset = torch.load(val_student_path)
else:
val_data = datasets.MNIST(
root=path,
train=False,
download=True,
transform=transforms.Compose([transforms.ToTensor(),]),
)
num_classes = len(val_data.classes)
student_number_sampled = dataset_configs["student_val_number"] // num_classes
val_student_subset = _split_by_labels(
num_classes, val_data, student_number_sampled, val_student_path
)
val_student_data = TorchVisionDataset(
name=dataset_name, data=val_student_subset, split="test",
)
print(f"train_data length: {len(train_student_data)}, test_data length: {len(test_student_data)}, val_data length: {len(val_student_data)}")
return {
"train": train_student_data,
"test": test_student_data,
"val": val_student_data,
}
def _read_cifar10_dataset(dataset_configs, dataset_name):
path = dataset_configs["dataset_path"]
custom = dataset_configs["binary"]
# label_to_str = {0: 'airplane', 1: 'automobile', 2: 'bird', 3: 'cat', 4: 'deer', 5: 'dog', 6: 'frog', 7: 'horse', 8: 'ship', 9: 'truck'} Just for information
label_custom = {0: [0,8], 1: [1,9], 2:[2,6,4], 3:[3,5,7]}
label_student = {2:[2,6,4], 3:[3,5,7]}
label_server= {2:[2,6,4], 3:[3,5,7]}
# 1.2 Training Data for Student
train_student_path = os.path.join(path, "train_split.pt")
if os.path.exists(train_student_path):
train_student_subset = torch.load(train_student_path)
else:
train_data = datasets.CIFAR10(
root=path,
train=True,
download=True,
transform=transforms.Compose([transforms.ToTensor()]),
)
# num_classes = len(train_data.classes)
# student_number_sampled = dataset_configs["student_train_number"] // num_classes
# train_student_subset = _split_by_labels(
# num_classes, train_data, student_number_sampled, train_student_path
# )
num_classes = len(label_custom)
student_number_sampled = dataset_configs["student_train_number"] // num_classes
train_student_subset = label_update(
num_classes, train_data, student_number_sampled, train_student_path, custom, label_custom)
# print(len(train_student_subset))
# # print(train_data.max)
# # print(train_data.min)
# print(torch.max(((train_student_subset[0][0]))))
# print(torch.min(((train_student_subset[0][0]))))
# exit()
train_student_data = TorchVisionDataset(
name=dataset_name, data=train_student_subset, split="train",
)
# 1.3 Validation Data for Student
val_student_path = os.path.join(path, "val_split.pt")
if os.path.exists(val_student_path):
val_student_subset = torch.load(val_student_path)
else:
val_data = datasets.CIFAR10(
root=path,
train=True,
download=True,
transform=transforms.Compose([transforms.ToTensor()]),
)
# num_classes = len(val_data.classes)
# student_number_sampled = dataset_configs["student_val_number"] // num_classes
# val_student_subset = _split_by_labels(
# num_classes, val_data, student_number_sampled, val_student_path
# )
num_classes = len(label_custom)
student_number_sampled = dataset_configs["student_val_number"] // num_classes
val_student_subset = label_update(
num_classes, val_data, student_number_sampled, val_student_path, custom, label_custom)
val_student_data = TorchVisionDataset(
name=dataset_name, data=val_student_subset, split="val",
)
# 2.2 Test Data for Student
test_student_path = os.path.join(path, "test_split.pt")
if os.path.exists(test_student_path):
test_student_subset = torch.load(test_student_path)
else:
test_data = datasets.CIFAR10(
root=path,
train=True,
download=True,
transform=transforms.Compose([transforms.ToTensor()]),
)
# num_classes = len(test_data.classes)
# student_number_sampled = dataset_configs["student_test_number"] // num_classes
# test_student_subset = _split_by_labels(
# num_classes, test_data, student_number_sampled, test_student_path
# )
num_classes = len(label_student)
student_number_sampled = dataset_configs["student_test_number"] // num_classes
test_student_subset = label_update(
num_classes, test_data, student_number_sampled, test_student_path, custom, label_student)
test_student_data = TorchVisionDataset(
name=dataset_name, data=test_student_subset, split="test",
)
print(
f"train_student_data length: {len(train_student_data)}, val_student_data length: {len(val_student_data)}, test_student_data length: {len(test_student_data)}"
)
return {
"train": train_student_data,
"val": val_student_data,
"test": test_student_data,
}