-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathrun_distillation.py
executable file
·151 lines (130 loc) · 6.59 KB
/
run_distillation.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
import numpy as np
import matplotlib.pyplot as plt
import data
import random_features
import utils
import models
import torch_optimizer as optim
import torch
import torch.nn as nn
import training
from functools import partial
import math
import time
import os
import distillation
import argparse
import sys
import coresets
from torchvision import datasets,transforms
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--dataset', type=str)
parser.add_argument('--lr', type=float, default = 1e-3)
parser.add_argument('--jit', type=float, default = 5e-3)
parser.add_argument('--save_path', type = str)
parser.add_argument('--samples_per_class', type = int)
parser.add_argument('--init_strategy', type=str, default = 'random')
parser.add_argument('--learn_labels', action='store_true')
parser.add_argument('--platt', action='store_true')
parser.add_argument('--coreset', action='store_true')
parser.add_argument('--n_models', type=int, default = 8)
parser.add_argument('--ga_steps', type=int, default = 1)
parser.add_argument('--seed', type=int, default = 0)
parser.add_argument('--corruption', type=float, default = 0)
parser.add_argument('--n_batches', type = int, default = 4)
args = parser.parse_args()
transform_fn = None
from_loader = False
whitening_mat = None
if args.dataset == 'mnist':
im_size = 28
n_channels = 1
n_classes = 10
X_train, y_train, X_test, y_test = data.get_mnist(output_channels = 1, image_size = im_size)
elif args.dataset == 'fashion':
im_size = 28
n_channels = 1
n_classes = 10
X_train, y_train, X_test, y_test = data.get_fashion_mnist(output_channels = 1, image_size = im_size)
elif args.dataset == 'cifar10':
im_size = 32
n_channels = 3
n_classes = 10
X_train, y_train, X_test, y_test = data.get_cifar10(output_channels = n_channels, image_size = im_size)
whitening_mat = data.get_zca_matrix(X_train, reg_coef = 0.1)
X_train = data.transform_data(X_train, whitening_mat)
X_test = data.transform_data(X_test, whitening_mat)
elif args.dataset == 'cifar100':
im_size = 32
n_channels = 3
n_classes = 100
X_train, y_train, X_test, y_test = data.get_cifar100(output_channels = n_channels, image_size = im_size)
whitening_mat = data.get_zca_matrix(X_train, reg_coef = 0.1)
X_train = data.transform_data(X_train, whitening_mat)
X_test = data.transform_data(X_test, whitening_mat)
elif args.dataset == 'svhn':
im_size = 32
n_channels = 3
n_classes = 10
X_train, y_train, X_test, y_test = data.get_svhn(output_channels = n_channels, image_size = im_size)
X_train = data.layernorm_data(X_train)
X_test = data.layernorm_data(X_test)
whitening_mat = data.get_zca_matrix(X_train, reg_coef = 0.1)
X_train = data.transform_data(X_train, whitening_mat)
X_test = data.transform_data(X_test, whitening_mat)
elif args.dataset == 'split_mnist':
im_size = 28
n_channels = 1
n_classes = 2
X_train, y_train, X_test, y_test = data.get_mnist(output_channels = 1, image_size = im_size)
y_train = y_train//5
y_test = y_test//5
elif args.dataset == 'celeba':
im_size = 64
n_channels = 3
n_classes = 2
ds_train = datasets.CelebA('./data/', split = 'train', download = True, transform = transforms.Compose([transforms.ToTensor(), transforms.Resize([64, 64]), transforms.Normalize((0.5064, 0.4258, 0.3832), (0.3093, 0.2890, 0.2883))]), target_type = 'attr',
target_transform = transforms.Lambda(lambda x: x[20]))
train_loader = torch.utils.data.DataLoader(ds_train,
batch_size=1280,
shuffle=True,
num_workers=8)
ds_valid = datasets.CelebA('./data/', split = 'valid', download = True, transform = transforms.Compose([transforms.ToTensor(), transforms.Resize([64, 64]), transforms.Normalize((0.5064, 0.4258, 0.3832), (0.3093, 0.2890, 0.2883))]), target_type = 'attr',
target_transform = transforms.Lambda(lambda x: x[20]))
valid_set = next(iter(torch.utils.data.DataLoader(ds_valid,
batch_size=1000,
shuffle=True,
num_workers=8)))
X_valid = valid_set[0]
y_valid = valid_set[1]
X_train = train_loader
y_train = None
from_loader = True
else:
print("unrecognized dataset: {}".format(args.dataset))
sys.exit()
if args.dataset != 'celeba':
X_init = coresets.make_coreset(X_train, y_train, args.samples_per_class, n_classes, args.init_strategy, seed = args.seed)
else:
batch = next(iter(X_train))
X_init = coresets.make_coreset(batch[0], batch[1], args.samples_per_class, n_classes, args.init_strategy, seed = args.seed)
if args.dataset != 'celeba':
np.random.seed(args.seed)
valid_indices = []
for c in range(n_classes):
class_indices = np.where(y_train == c)[0]
valid_indices.append(class_indices[np.random.choice(len(class_indices), 500 if n_classes == 10 else 100)])
valid_indices = np.concatenate(valid_indices)
X_valid = X_train[valid_indices]
y_valid = y_train[valid_indices]
model_class = partial(models.ConvNet_wide, n_channels, net_norm = 'none', im_size=(im_size,im_size), k = 2, chopped_head = True)
scheduler = [(0, args.n_models, 1)]
n_iters = 100000 if not args.coreset else 1
distillation.distill_dataset(X_train, y_train,
model_class, args.lr, 8, args.n_batches, iters = n_iters,
ga_steps = args.ga_steps, platt = args.platt,
schedule = scheduler, save_location = args.save_path,
samples_per_class = args.samples_per_class, n_classes = n_classes, learn_labels = args.learn_labels,
batch_size = 1280, X_valid = X_valid, y_valid = y_valid,
n_channels = n_channels, im_size = im_size, X_init = X_init, jit = args.jit, seed = args.seed, corruption = args.corruption, whitening_mat = whitening_mat, from_loader = from_loader)