output = model(x)
loss = criterion(output, y)
predicted = output.argmax(1)
acc = (predicted == y).sum().item() / y.size(0)
accuracy_epoch += acc
loss_epoch += loss.item()
return loss_epoch, accuracy_epoch
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="SimCLR")
config = yaml_config_hook("./config/config.yaml")
for k, v in config.items():
parser.add_argument(f"--{k}", default=v, type=type(v))
args = parser.parse_args()
args.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
if args.dataset == "STL10":
train_dataset = torchvision.datasets.STL10(
args.dataset_dir,
split="train",
download=True,
transform=TransformsSimCLR(size=args.image_size).test_transform,
)
test_dataset = torchvision.datasets.STL10(
args.dataset_dir,
extracted_features.extend(h.cpu().detach().numpy())
feature_vector.extend(c.cpu().detach().numpy())
labels_vector.extend(y.numpy())
if step % 20 == 0:
print(f"Step [{step}/{len(loader)}]\t Computing features...")
feature_vector = np.array(feature_vector)
labels_vector = np.array(labels_vector)
extracted_features = np.array(extracted_features)
#print("Features shape {}".format(feature_vector.shape))
#print("feature extracted: ", extracted_features.shape)
return feature_vector, labels_vector, extracted_features
if __name__ == "__main__":
parser = argparse.ArgumentParser()
config = yaml_config_hook("config/config_fpi.yaml")
for k, v in config.items():
parser.add_argument(f"--{k}", default=v, type=type(v))
args = parser.parse_args()
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
if args.dataset == "CIFAR-10":
train_dataset = torchvision.datasets.CIFAR10(
root=args.dataset_dir,
train=True,
download=True,
transform=transform.Transforms(
size=args.image_size).test_transform,
)
test_dataset = torchvision.datasets.CIFAR10(
root=args.dataset_dir,
train=False,