device = torch.device('cuda' if args.cuda else 'cpu')
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
kwargs = {'num_workers': 1, 'pin_memory': True} if args.cuda else {}
normalize = transforms.Normalize(mean=[0.4914, 0.482, 0.4465],
std=[0.2023, 0.1994, 0.2010])
train_loader = torch.utils.data.DataLoader(
cifar.CIFAR100('./cifar-100', train=True, download=True,
transform=transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
normalize
])),
batch_size=args.batch_size, shuffle=True, **kwargs)
test_loader = torch.utils.data.DataLoader(
cifar.CIFAR100('./cifar-100', train=False, transform=transforms.Compose([
transforms.ToTensor(),
normalize
])),
batch_size=args.test_batch_size, shuffle=True, **kwargs)
model = archs[args.arch](num_classes=100)
bn_names = []
if args.l1_decay > 0:
if not args.all_bn:
import os
import sys
import time
import numpy
import torch
from torchvision import transforms
from torchvision.datasets import cifar
from openvino.inference_engine import IECore, IENetwork
model_name = sys.argv[1]
normalize = transforms.Normalize(mean=[0.4914, 0.482, 0.4465],
std=[0.2023, 0.1994, 0.2010])
test_loader = torch.utils.data.DataLoader(cifar.CIFAR100(
'./cifar-100',
train=False,
transform=transforms.Compose([transforms.ToTensor(), normalize])),
batch_size=1,
shuffle=True)
# Initialize inference engine core
ie = IECore()
# build network
net = IENetwork(model="{}.xml".format(model_name),
weights="{}.bin".format(model_name))
input_name = next(iter(net.inputs))
output_name = next(iter(net.outputs))
exec_net = ie.load_network(network=net, device_name="CPU")
correct = 0
def gen_val_datasets(self, transform=None, target_transform=None) -> Dataset:
ds = cifar.CIFAR100(root="~/.cache/torch/data", train=True, download=True,
transform=transform, target_transform=target_transform)
return Subset(OrderDataset(ds), list(range(40000, 50000)))
def train():
import ubelt as ub
root = ub.ensure_app_cache_dir('clab')
train_dset = cifar.CIFAR100(root=root, download=True, train=True)
test_dest = cifar.CIFAR100(root=root, download=True, train=False)
def gen_test_datasets(self, transform=None, target_transform=None) -> Dataset:
return cifar.CIFAR100(root="~/.cache/torch/data", train=False, download=True,
transform=transform, target_transform=target_transform)