def FMNIST(mode="train", download=True):
# 由于不同数据集区分训练或测试集的方式不同，因此建议数据集以统一的接口定义
from torchvision.datasets.mnist import FashionMNIST
from torchvision import transforms
weak = transforms.ToTensor()
if mode == "train":
return FashionMNIST(root=root, train=True, download=download, transform=weak)
else:
return FashionMNIST(root=root, train=False, transform=weak, download=download)
def __init__(self, mode, transform):
dataset = FashionMNIST(root=data_path, download=True, train=mode)
data = getattr(dataset, 'train_data' if mode else 'test_data')
labels = getattr(dataset, 'train_labels' if mode else 'test_labels')
self.transform = transform
self.input_images = np.array(data).astype(np.float)
self.input_labels = np.array(labels).astype(np.long)
def __init__(self, mode):
dataset = FashionMNIST(root='./data', download=True, train=mode)
data = getattr(dataset, 'train_data' if mode else 'test_data')
labels = getattr(dataset, 'train_labels' if mode else 'test_labels')
input_images = np.array(data).astype(np.float)
self.input_images = input_images
self.target_labels = np.array(labels).astype(np.long)
def get_data_loaders(train_batch_size, val_batch_size):
data_transform = Compose([ToTensor(), Normalize((0.1307, ), (0.3081, ))])
train_data = FashionMNIST(download=True,
root=".",
transform=data_transform,
train=True)
val_data = FashionMNIST(download=False,
root=".",
transform=data_transform,
train=False)
train_loader = DataLoader(Subset(train_data, range(N_TRAIN_EXAMPLES)),
batch_size=train_batch_size,
shuffle=True)
val_loader = DataLoader(Subset(val_data, range(N_VALID_EXAMPLES)),
batch_size=val_batch_size,
shuffle=False)
return train_loader, val_loader
def get_iterator(
mode
): ## Will be used in the training set, it is kind of a data loader
ds = FashionMNIST(root=opt['data'], download=True,
train=mode) ##Changed loader to
data = getattr(
ds, 'train_data' if mode else 'test_data') ##Fashion MNIST Loader such
labels = getattr(ds, 'train_labels' if mode else
'test_labels') ##that it directly downloads the data
tds = tnt.dataset.TensorDataset([data, labels])
return tds.parallel(batch_size=opt['b'],
num_workers=0,
shuffle=mode,
pin_memory=True)
"""
Load MNIST dataset from MNSIT and give a dictionary with it's training/validation dataloaders
"""
import torch
from torchvision.datasets.mnist import FashionMNIST
import torchvision.transforms as transforms
from torch.utils.data import DataLoader
data_train = FashionMNIST("./data/fashion_mnist",
download=True,
train=True,
transform=transforms.Compose([
transforms.Resize((28, 28)),
transforms.ToTensor()
]))
data_val = FashionMNIST("./data/fashion_mnist",
train=False,
download=True,
transform=transforms.Compose([
transforms.Resize((28, 28)),
transforms.ToTensor()
]))
dataloader_train = DataLoader(data_train,
batch_size=1000,
shuffle=True,
num_workers=8)
dataloader_val = DataLoader(data_val, batch_size=1000, num_workers=8)
dataloaders = {
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
import torchvision.transforms as transforms
from torch.optim.lr_scheduler import ExponentialLR
from torch.utils.data import DataLoader
from torchvision.datasets.mnist import MNIST, FashionMNIST
from test_models import *
from utils.scheduler import MinExponentialLR
# init dataset
transform = transforms.Compose(
[transforms.CenterCrop(28),
transforms.ToTensor()])
train_data = FashionMNIST('../data',
train=True,
download=True,
transform=transform)
test_data = FashionMNIST('../data',
train=False,
download=True,
transform=transform)
# init dataloader
train_data_loader = DataLoader(train_data,
batch_size=256,
shuffle=True,
num_workers=8)
test_data_loader = DataLoader(test_data, batch_size=1024, num_workers=8)
# init models and optimizers
models = [model_a(), model_b(), model_c(), model_d(), model_e()]
def create_datasets(data_root):
trainset = FashionMNIST(root = data_root, train = True, download = True, transform = transforms.ToTensor())
testset = FashionMNIST(root = data_root, train = False, download = True, transform = transforms.ToTensor())
return trainset, testset
def fashionmnist(train=True):
dataset = FashionMNIST(root=root, train=train)
xs = llist(Image.fromarray(img.numpy(), mode='L') for img in dataset.data)
ys = llist(int(i) for i in dataset.targets)
return xs, ys