-
Notifications
You must be signed in to change notification settings - Fork 3
/
dataset.py
49 lines (41 loc) · 1.87 KB
/
dataset.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
import copy
import torch
import torch.nn as nn
from torchvision import transforms, datasets
from torch.utils.data import DataLoader
from hook import BNStatisticsHook
transform = transforms.Compose([
transforms.Resize([224, 224]),
transforms.ToTensor()]
)
def get_imagenet_exp_dl(batch_size):
imagenet_dir = "./data/dataset/imagenet/"
train_ds = datasets.ImageNet(imagenet_dir, split="train", transform=transform)
val_ds = datasets.ImageNet(imagenet_dir, split="val", transform=transform)
train_dl = DataLoader(train_ds, batch_size, num_workers=8, pin_memory=True, shuffle=True)
val_dl = DataLoader(val_ds, batch_size, num_workers=8, pin_memory=True, shuffle=False)
return train_dl, val_dl
def get_dataloader(name, batch_size, shuffle=False, train=False):
if name == "imagenet":
split = "train" if train else "val"
dataset = datasets.ImageNet("./data/dataset/imagenet/", split=split, transform=transform)
elif name == "caltech256":
dataset = datasets.Caltech256("./data/dataset/", transform=transform, download=True)
elif name == "cifar100":
dataset = datasets.CIFAR100("./data/dataset/cifar100", transform=transform, download=True, train=train)
return DataLoader(dataset, batch_size, drop_last=True, shuffle=shuffle)
def get_grad_dl(model: nn.Module, dataloader: DataLoader, device):
model = copy.deepcopy(model).to(device)
criterion = nn.CrossEntropyLoss()
hook = BNStatisticsHook(model, train=False)
for x, y in dataloader:
model.zero_grad()
hook.clear()
x, y = x.to(device), y.to(device)
# y = torch.arange(len(x)).to(device)
y_pred = model(x)
loss = criterion(y_pred, y)
grad = torch.autograd.grad(loss, model.parameters())
grad = [g.detach() for g in grad]
mean_var_list = hook.mean_var_list
yield x, y, grad, mean_var_list