device_target=args_opt.device_target) context.set_context(device_id=args_opt.device_id) device_num = int(os.environ.get("DEVICE_NUM", 1)) if device_num > 1: context.reset_auto_parallel_context() context.set_auto_parallel_context( device_num=device_num, parallel_mode=ParallelMode.DATA_PARALLEL, mirror_mean=True) init() dataset = create_dataset(args_opt.data_path, cfg.epoch_size) batch_num = dataset.get_dataset_size() net = vgg16(num_classes=cfg.num_classes) lr = lr_steps(0, lr_max=cfg.lr_init, total_epochs=cfg.epoch_size, steps_per_epoch=batch_num) opt = Momentum(filter(lambda x: x.requires_grad, net.get_parameters()), Tensor(lr), cfg.momentum, weight_decay=cfg.weight_decay) loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction='mean', is_grad=False) model = Model(net, loss_fn=loss, optimizer=opt, metrics={'acc'},
parser.add_argument('--data_path', type=str, default='./cifar', help='path where the dataset is saved') parser.add_argument('--device_id', type=int, default=None, help='device id of GPU or Ascend. (Default: None)') args_opt = parser.parse_args() context.set_context(mode=context.GRAPH_MODE, device_target=args_opt.device_target) context.set_context(device_id=args_opt.device_id) context.set_context(enable_mem_reuse=True, enable_hccl=False) net = vgg16(batch_size=cfg.batch_size, num_classes=cfg.num_classes) lr = lr_steps(0, lr_max=cfg.lr_init, total_epochs=cfg.epoch_size, steps_per_epoch=50000 // cfg.batch_size) opt = Momentum(filter(lambda x: x.requires_grad, net.get_parameters()), Tensor(lr), cfg.momentum, weight_decay=cfg.weight_decay) loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction='mean', is_grad=False) model = Model(net, loss_fn=loss, optimizer=opt, metrics={'acc'}) dataset = dataset.create_dataset(args_opt.data_path, cfg.epoch_size) batch_num = dataset.get_dataset_size()
def test_vgg16(): inputs = Tensor(np.random.rand(1, 3, 112, 112).astype(np.float32)) net = vgg16() with pytest.raises(ValueError): print(net.construct(inputs))