modAL应用pytorch时TypeError: ＜class ‘torch.Tensor‘＞ datatype is not supported

mnist_data = MNIST('.', download=True, transform=ToTensor())
dataloader = DataLoader(mnist_data, shuffle=True, batch_size=60000)
X, y = next(iter(dataloader))

# the active learning loop
n_queries = 10
for idx in range(n_queries):
    print('Query no. %d' % (idx + 1))
    query_idx, query_instance = learner.query(X_pool, n_instances=100)
    learner.teach(
        X=X_pool[query_idx], y=y_pool[query_idx], only_new=True,
    )
    # remove queried instance from pool
    X_pool = np.delete(X_pool, query_idx, axis=0)
    y_pool = np.delete(y_pool, query_idx, axis=0)

在运行时会报错，TypeError: <class 'torch.Tensor'> datatype is not supported

这是因为modAL中的learner并不支持tensor类型的输入，所以这里x_pool和y_pool会出现错误

解决办法把tensor类型转化成numpy数据类型，即在X, y = next(iter(dataloader))后加两行

x=x.detach().cpu().numpy()
y=y.detach().cpu().numpy()

这里detach()作用是返回一个新的Tensor，只不过不再有梯度。

如果想把CUDA tensor格式的数据改成numpy时，需要先将其转换成cpu float-tensor随后再转到numpy格式。 numpy不能读取CUDA tensor 需要将它转化为 CPU tensor