You can use Numpy's random() function to generate random numbers between 0 and 1 and then apply the round() function to those numbers to get binary labels.
Here is an example to generate 10 binary labels.
train_labels = np.round(np.random.random(10))