Skip to content

GitLab

Switch branch/tag
Find file Normal viewHistoryPermalink
data.py 6.62 KB
Newer Older
chicm-ms's avatar
AutoML for model compression (#2573)  
chicm-ms committed
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 
# Copyright (c) Microsoft Corporation.
# Licensed under the MIT license.

import torch
import torch.nn.parallel
import torch.optim
import torch.utils.data
import torchvision
import torchvision.transforms as transforms
import torchvision.datasets as datasets
from torch.utils.data.sampler import SubsetRandomSampler
import numpy as np

import os


def get_dataset(dset_name, batch_size, n_worker, data_root='../../data'):
    cifar_tran_train = [
        transforms.RandomCrop(32, padding=4),
        transforms.RandomHorizontalFlip(),
        transforms.ToTensor(),
        transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),
    ]
    cifar_tran_test = [
        transforms.ToTensor(),
        transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),
    ]
    print('=> Preparing data..')
    if dset_name == 'cifar10':
        transform_train = transforms.Compose(cifar_tran_train)
        transform_test = transforms.Compose(cifar_tran_test)
        trainset = torchvision.datasets.CIFAR10(root=data_root, train=True, download=True, transform=transform_train)
        train_loader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True,
                                                   num_workers=n_worker, pin_memory=True, sampler=None)
        testset = torchvision.datasets.CIFAR10(root=data_root, train=False, download=True, transform=transform_test)
        val_loader = torch.utils.data.DataLoader(testset, batch_size=batch_size, shuffle=False,
                                                 num_workers=n_worker, pin_memory=True)
        n_class = 10
    elif dset_name == 'imagenet':
        # get dir
        traindir = os.path.join(data_root, 'train')
        valdir = os.path.join(data_root, 'val')

        # preprocessing
        input_size = 224
        imagenet_tran_train = [
            transforms.RandomResizedCrop(input_size, scale=(0.2, 1.0)),
            transforms.RandomHorizontalFlip(),
            transforms.ToTensor(),
            transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
        ]
        imagenet_tran_test = [
            transforms.Resize(int(input_size / 0.875)),
            transforms.CenterCrop(input_size),
            transforms.ToTensor(),
            transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
        ]

        train_loader = torch.utils.data.DataLoader(
            datasets.ImageFolder(traindir, transforms.Compose(imagenet_tran_train)),
            batch_size=batch_size, shuffle=True,
            num_workers=n_worker, pin_memory=True, sampler=None)

        val_loader = torch.utils.data.DataLoader(
            datasets.ImageFolder(valdir, transforms.Compose(imagenet_tran_test)),
            batch_size=batch_size, shuffle=False,
            num_workers=n_worker, pin_memory=True)
        n_class = 1000

    else:
        raise NotImplementedError

    return train_loader, val_loader, n_class


def get_split_dataset(dset_name, batch_size, n_worker, val_size, data_root='../data', shuffle=True):
    '''
        split the train set into train / val for rl search
    '''
    if shuffle:
        index_sampler = SubsetRandomSampler
    else:  # every time we use the same order for the split subset
        class SubsetSequentialSampler(SubsetRandomSampler):
            def __iter__(self):
                return (self.indices[i] for i in torch.arange(len(self.indices)).int())
        index_sampler = SubsetSequentialSampler

    print('=> Preparing data: {}...'.format(dset_name))
    if dset_name == 'cifar10':
        transform_train = transforms.Compose([
            transforms.RandomCrop(32, padding=4),
            transforms.RandomHorizontalFlip(),
            transforms.ToTensor(),
            transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),
        ])
        transform_test = transforms.Compose([
            transforms.ToTensor(),
            transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),
        ])
        trainset = torchvision.datasets.CIFAR100(root=data_root, train=True, download=True, transform=transform_train)
        valset = torchvision.datasets.CIFAR10(root=data_root, train=True, download=True, transform=transform_test)
        n_train = len(trainset)
        indices = list(range(n_train))
        # now shuffle the indices
        #np.random.shuffle(indices)
        assert val_size < n_train
        train_idx, val_idx = indices[val_size:], indices[:val_size]

        train_sampler = index_sampler(train_idx)
        val_sampler = index_sampler(val_idx)

        train_loader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=False, sampler=train_sampler,
                                                   num_workers=n_worker, pin_memory=True)
        val_loader = torch.utils.data.DataLoader(valset, batch_size=batch_size, shuffle=False, sampler=val_sampler,
                                                 num_workers=n_worker, pin_memory=True)
        n_class = 10
    elif dset_name == 'imagenet':
        train_dir = os.path.join(data_root, 'train')
        val_dir = os.path.join(data_root, 'val')
        normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                         std=[0.229, 0.224, 0.225])
        input_size = 224
        train_transform = transforms.Compose([
                transforms.RandomResizedCrop(input_size),
                transforms.RandomHorizontalFlip(),
                transforms.ToTensor(),
                normalize,
            ])
        test_transform = transforms.Compose([
                transforms.Resize(int(input_size/0.875)),
                transforms.CenterCrop(input_size),
                transforms.ToTensor(),
                normalize,
            ])

        trainset = datasets.ImageFolder(train_dir, train_transform)
        valset = datasets.ImageFolder(train_dir, test_transform)
        n_train = len(trainset)
        indices = list(range(n_train))
        np.random.shuffle(indices)
        assert val_size < n_train
        train_idx, val_idx = indices[val_size:], indices[:val_size]

        train_sampler = index_sampler(train_idx)
        val_sampler = index_sampler(val_idx)

        train_loader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, sampler=train_sampler,
                                                   num_workers=n_worker, pin_memory=True)
        val_loader = torch.utils.data.DataLoader(valset, batch_size=batch_size, sampler=val_sampler,
                                                 num_workers=n_worker, pin_memory=True)

        n_class = 1000
    else:
        raise NotImplementedError

    return train_loader, val_loader, n_class